Project No. TREN-05-FP6TR-S07.61320-518404-DRUID

DRUID Driving under the Influence of Drugs, and Medicines

Integrated Project 1.6. Sustainable Development, Global Change and Ecosystem 1.6.2: Sustainable Surface Transport

6th Framework Programme Deliverable 1.2.2 Effects of medicinal drugs on actual and simulated driving

Start date of project: 15.10.2006 Duration: 48 months Organisation name of lead contractor for this deliverable: UMaas

1 DRUID 6th framework programme Deliverable 1.2.2 6th Framework Programme Deliverable D 1.2.2. Effects of medicinal drugs on actual and simulated driving Status: Final

Editor/author: Jan Ramaekers, Maastricht University, The Netherlands Workpagage Leader: Anja Knoche, BASt, Germany Project Co-ordinator: Horst Schulze, BASt, Germany

Partners

Tim Leufkens, Wendy Bosker. Kim Kuypers, Annemiek Vermeeren, Jan Ramaekers; Maastricht University, The Netherlands Markus Schumacher, Anja Knoche; BASt, Germany Marie Laure Bocca, Pierre Denise, University of Caen, France Catherine Berthelon: IFSTTAR/INRETS, France Pierre Angelo Sardi, Gian Marco Sardi, Claudio Signoretti; SIPSiVi, Italy. Katerina Touliou, CERT/HIT, Greece Gisela Skopp, University of Heidelberg.

Date: July 7, 2011

2 DRUID 6th framework programme Deliverable 1.2.2 EXECUTIVE SUMMARY ...... 5

BACKGROUND AND RATIONALE...... 5 DESIGN, DOSING AND STUDY PROCEDURES ...... 8 STANDARD DRIVING PARAMETERS ...... 10 DEFINITION OF CLINICALLY RELEVANT DRUG EFFECTS ...... 11 STANDARD TOXICOLOGICAL ANALYSES ...... 11 STANDARD STATISTICAL METHODS ...... 11 MAIN RESULTS AND DISCUSSION...... 12 CONCLUSIONS ...... 21 REFERENCES ...... 21 CHAPTER 1: RESIDUAL EFFECTS OF ZOPICLONE 7.5 MG ON HIGHWAY DRIVING PERFORMANCE IN INSOMNIA PATIENTS AND HEALTHY CONTROLS: A PLACEBO CONTROLLED CROSSOVER STUDY ...... 23

ABSTRACT...... 24 INTRODUCTION ...... 25 METHODS ...... 25 STATISTICAL ANALYSIS ...... 30 RESULTS...... 31 DISCUSSION...... 38 REFERENCES ...... 41 CHAPTER 2: DRIVING PERFORMANCE OF CHRONIC USERS OF HYPNOTICS AND UNMEDICATED INSOMNIA PATIENTS ...... 44

ABSTRACT...... 45 INTRODUCTION ...... 46 METHODS ...... 47 STATISTICAL ANALYSIS ...... 51 RESULTS...... 59 DISCUSSION...... 61 REFERENCES ...... 64 CHAPTER 3 : EFFECTS OF BENZODIAZEPINES ON DRIVING PERFORMANCE OF ANXIETY PATIENTS...... 68

ABSTRACT...... 69 INTRODUCTION ...... 70 METHODS ...... 72 GROUP ...... 72 STATISTICS...... 78 RESULTS...... 79 DISCUSSION...... 91 REFERENCES ...... 94 CHAPTER 4: DAYTIME DRIVING IN TREATED (CPAP) AND UNTREATED SLEEP APNOEA PATIENTS...... 96

ABSTRACT...... 97 INTRODUCTION ...... 98 METHOD ...... 100 STATISTICAL ANALYSIS ...... 104 RESULTS...... 104 DISCUSSION...... 112 REFERENCES ...... 116 CHAPTER 5 : EFFECTS OF CODOLIPRANE AND ZOLPIDEM, ALONE OR IN COMBINATION, ON ELDERLY DRIVERS’BEHAVIOR...... 118

ABSTRACT...... 119 INTRODUCTION ...... 120 METHOD ...... 120 RESULTS...... 122

3 DRUID 6th framework programme Deliverable 1.2.2 DISCUSSION...... 128 REFERENCES ...... 129 CHAPTER 6 : ACUTE EFFECTS OF 3 DOSES OF ON SIMULATED DRIVING PERFORMANCE IN HEALTHY VOLUNTEERS...... 131

ABSTRACT...... 132 INTRODUCTION ...... 133 METHOD ...... 134 ҏSTATISTICS...... 136 RESULTS...... 137 DISCUSSION...... 140 REFERENCES ...... 143 CHAPTER 7 : DOSE RELATED EFFECTS OF DRONABINOL ON ACTUAL DRIVING PERFORMANCE OF OCCASIONAL AND HEAVY CANNABIS USERS ...... 147

ABSTRACT...... 148 INTRODUCTION ...... 149 METHOD ...... 151 STATISTICAL ANALYSIS ...... 153 RESULTS...... 155 DISCUSSION...... 161 REFERENCES ...... 164 CHAPTER 8 : EFFECTS OF ANALGETIC MEDICATION ON ACTUAL DRIVING ...... 166

ABSTRACT...... 167 INTRODUCTION ...... 167 METHODS ...... 169 STATISTICAL ANALYSIS ...... 176 RESULTS...... 176 DISCUSSION...... 183 REFERENCES ...... 185 CHAPTER 9: EFFECTS OF ANALGESICS ON DRIVING RELATED SKILLS ...... 188

ABSTRACT...... 189 INTRODUCTION ...... 190 METHODS ...... 191 STATISTICAL ANALYSIS ...... 201 RESULTS...... 201 DISCUSSION...... 211 REFERENCES ...... 214 CHAPTER 10: EFFECTS ON REAL DRIVING PERFORMANCE COMPARED WITH THE EFFECTS OF ALCOHOL ...... 217

ABSTRACT...... 218 INTRODUCTION ...... 219 METHOD ...... 220 STATISTICAL ANALYSES ...... 223 DISCUSSION...... 225 REFERENCES ...... 226 CHAPTER 11: BLOOD TO SERUM RATIOS OF HYPNOTICS, OPIOID AND NON-OPIOID ANALGESICS AS WELL AS ANTIPSYCHOTICS AND AMPHETAMINE-LIKE DRUGS AND THEIR ANALYSIS IN DRIED BLOOD SPOTS...... 227

ABSTRACT...... 228 INTRODUCTION ...... 229 METHODS ...... 231 RESULTS AND DISCUSSION ...... 235 CONCLUSION...... 243 REFERENCES ...... 249

4 DRUID 6th framework programme Deliverable 1.2.2 Executive summary

Background and rationale

Recent statistics reveal that more than 40000 people die on European roads each year, and another 1.7 million are injured. About a quarter of these deaths, some 10000 per year, are estimated to be caused by drink driving. Although alcohol is by far the most prevalent psychoactive substance affecting drivers, concerns have been mounting about the role of medicinal drugs to motor vehicle crashes. In a recent review of drug driving research conducted by The European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) recently published a literature review of drug driving research (EMCDDA 2008). In general, drug driving research has followed to methodological approaches (i.e. experimental and epidemiological) that are mutually supportive. Experimental studies aim to assess medicinal drug effects after controlled administration in actual driving tests, driving simulator tests or laboratory measures of skills related to drivers. Epidemiological studies aim to measure accident involvement of drivers who are receiving medicinal drug treatment. In experimental studies, the primary outcome measure is (performance) impairment. In epidemiological studies, the primary outcome measure is crash risk. Experimental research on individual drug effects is more prevalent in the scientific literature as compared to epidemiological work. Ideally both experimental and epidemiological research should be conducted on individual drug effect on driver safety. Unfortunately however, the combination of epidemiological and experimental data on traffic safety is simply lacking for most medicinal drugs. To date, only 2 drug classes have been consistently shown to increase driver impairment and crash in both experimental and epidemiological studies respectively. These drug classes are: benzodiazepines and tricyclic antidepressants. For other medicinal drug classes only experimental data is available. In general, the most important conclusions from the EMCDDA review regarding medicinal drugs effects on driving performance were:

· Benzodiazepines generally have impairing effects, but some types (whether long-, medium- or short-acting) cause severe impairment, while others are unlikely to have residual effects the following day.

· First-generation antihistamines are generally more sedating than second-generation ones, though there are exceptions in both groups.

· Tricyclic antidepressants show more impairment than the more recent types, though the results of experimental tests after consuming second-generation selective serotonin reuptake inhibitors are not always consistent.

· In every therapeutic class, some substances are associated with little or no impairment. These

5 DRUID 6th framework programme Deliverable 1.2.2 therapeutic drugs should preferably be prescribed to those wishing to drive.

For many other CNS drugs, information on crash risk or driver impairment is still lacking. For such drugs, evaluation of their impairing potential primarily depends on the evaluation of its pharmacological profile. A drug that is known to induce side effects relevant to driving, e.g. drowsiness, sedation, lack of concentration, is likely to be classified as an impairing drug in categorization systems for medicinal drugs and driving. Such classification systems thus generally include more potentially impairing medicinal drugs than those commonly found in experimental and epidemiological literature. From this point of view it is very interesting that a French research group (Orriols et al 2010) recently published a registry based study in 72685 drivers on the association between medicinal drug use and crash risk group for 4 separate risk levels that are distinguished in the French medication categorization system. These warning/risk levels include: no risk of impairment (level 0); be careful, read leaflet before driving (level 1); be very careful, to active advise from physician before driving (level 2) and danger, do not drive (level 3). Level 2 and level 3 medicines included a large number of CNS medicinal drugs such as , antiepileptics, antipsychotics, antiparkinsonian drugs, antidepressant, anxiolytics and hypnotics (including benzodiazepines), antihistamines and drugs used in alcohol dependence. The fraction of road traffic crashes that were attributable to drugs classified in levels 2 and 3 was 3.3%. Users of level 2 (OR=1.31[1.24-1.4]) and level 3 (OR= 1.25 [1.12-1.4]) drugs were more likely to be responsible for their crash A within-person case-control analysis furthermore showed that drivers were more likely to be exposed to level 3 medications on the crash day as compared to a control day (OR=1.15 [1.05-1.27]). This study nicely demonstrates that the use of a large range of medicinal drugs is associated with a significant increase in road crashes. It offers the perfect rationale for conducting more experimental and epidemiological drug driving research in order to identify medicinal drugs that have “impairing potential”of individual drugs that significantly increase crash risk and of contributing factors that may further increase the risk of become involved in traffic accidents, such as concomitant use of drugs and alcohol, treatment duration or treatment dose. The studies presented in this deliverable were all designed from to assess medicinal drug effects, covering a range of drug classes, on actual or simulated driving performance. Individual drugs that were selected for research belonged to one of the following drug classes.

Hypnotics/ anxiolytics GABA is a major inhibitory and widely-distributed neurotransmitter in the mammalian CNS. It is released by a web of short-axon interneurons occupying some 40% of all synapses. Benzodiazepine (BZD) ligands affect inhibitory GABA neurotransmission by allosterically modulating the neurotransmitter’s ability to open chloride channels at the GABAA /BZD receptor complex. The classic BZD anxiolytics and hypnotics act as agonists and achieve their anxiolytic, anticonvulsant and sedative effects through potentiation of GABA stimulated chloride flux. Previous reviews of pharmacodynamic studies with healthy volunteers have generally shown that BZD agonists can cause severe impairment in tests designed to measure psychomotor and driving performance (Saletu et al 1987; Van Laar 1998; Vermeeren 2004; Verster et al 2004; Woods et al 1992). Among psychomotor

6 DRUID 6th framework programme Deliverable 1.2.2 tasks, measures of CFF, DSST, tracking, and RT were particularly sensitive to the sedative effects of BZDs. The data generally indicate that BZDs cause a reduction in their users’ overall speed of information processing and motor response. In addition, performance impairment may still be present to a certain degree the morning after drug ingestion. These longer-lasting side effects are generally referred to as the ‘hangover’ or ‘residual’ effects of benzodiazepines. The practical relevance of psychomotor impairment under the influence of BZDs has been amply demonstrated in a long series of driving studies employing a standardized test (Vermeeren 2004).To date, most experimental driving studies assessing effects of hypnotics and anxiolytics have been conducted in healthy volunteers. The present studies were designed to the effects of these drug classes both in healthy volunteers as well as in insomnia patients, and to evaluate differences in drug effects in both groups.

Opioid and non-opioid analgesics

Analgesics can basically be distinguished in opioid and non-opioid analgesicss. Opioids include compounds such as , , , , , which all exert similar influence of the cerebral system. There are three principal classes of opioid receptors, ȝ, ț, į (mu, kappa, and delta), although more classes have been reported. Opioids are very effective anelgesics but are also know to produce a range of unpleasant side effecs such as sedation, nausea and vomiting. High doses of opioids may produce opioid toxicity (confusion, respiratory depression and seizures), but tolerance to (high) dose effects also rapidly develops during chronic use. Non-opioid anelgesics cover a large range of drugs that relief pain through a variety of pharmacological mechanisms. These include amongst others non-steroidal anti-inflammatory drugs (NSAIDS, eg. aspirin), paracetamol, cyclooxygenase (COX) inhibitors and CB1 receptors antagonists (cannabiniods). Studies of opioid effects on human performance generally report no impairment of psychomotor abilities in opioid dependent or tolerant patients. In contrast, evidence of no impairment of cognitive function during is inconclusive (Fishbain et al 2003; Strand et al 2011; Zacny 1996; Zacny et al 1998). Studies with non-opioid anelgesics in general also report a lack of impairment psychomotor and cognitive abilities after single and repeated doses. However, some non-opioid analegesics can be considered as potentially impairing but experimental need to be undertaken to assess the exact nature and magnitude of their effects on human performance. For examples, cannabinoids such as dronabinol are known to stimulate CB1 receptors in the brain and are expected to impair driver performance as has been previously demonstrated with cannabis (Ramaekers et al 2004).

Antipsychotics

Antipsychotics all share a common affinity for D2 receptors. Phenothiazines, such as and were the first D2 receptor antagonists used in the treatment of schizophrenia. Most produce profound sedation by blocking dopamine neurotransmission required for sustaining arousal (McClelland et al 1990; Wylie et al 1993). Additional blockade of histaminergic, anticholinergic and - adrenergic neurotransmission also contributes to the sedative potential of phenothiazines and results in a high prevalence of concentration difficulties, fatigue, and daytime sleepiness among. Since their

7 DRUID 6th framework programme Deliverable 1.2.2 introduction in the fifties, more selective and potent dopaminergic drugs such as have largely replaced these drugs. Like any dopaminergic receptor antagonists in empirical studies employing patients or healthy volunteers, haloperidol also produced severe sedation responsible for psychomotor impairment (King and Henry 1992; Ramaekers et al 1999). Yet sedation produced by selective dopaminergic antipsychotics is less profound and less capable of affecting a variety of mental functions and dependent behaviors, as compared to antipsychotics that block postsynaptic receptors within other monoamine systems as well. More recently, a new generation of comparable antipsychotics has been developed that, besides affinity for dopaminergic receptors, possess multiple mechanisms of action. , risperidone, olanzepine and are potent antagonists of the 5HT2A, H1 and Į1 receptor, and, in the case of clozapine and , the muscarinic acetylcholine receptor as well. None of these antipsychotics have been extensively investigated in studies designed for showing their effects on psychomotor and cognitive function. Yet in theory all of them should produce deficits in performance comparable to those observed for the earlier phenothiazines.

Design, dosing and study procedures

The following experimental studies were designed to assess the effects of medicinal drugs on actual or simulated driving performance.

Hypnotics, anxiolytics, sleep disorders: 1) Residual effects of zopiclone 7.5 mg on highway driving performance in insomnia patients and healthy controls (Maastricht University, Netherlands) 2) Actual driving performance of chronic users of hypnotics and unmedicated insomnia patients (Maastricht University, Netherlands) 3) Effects of alprazolam on simulated driving performance of anxious patients (CERTH/ HIT, Greece) 4) Simulated daytime driving in treated (CPAP) and untreated sleep apnoea patients (CERTH/HIT, Greece)

Opioid and non-opioid analgesics 5) Effects of codiliprane (codeine/paracetamol) and zolpidem, alone or in combination, on elderly drivers’behavior (University of Caen/ INRETS, France) 6) Acute effects of 3 doses of codiliprane (codeine/paracetamol) on simulated driving performance of healthy volunteers (University of Caen/ INRETS, France) 7) Effects of dronabinol on actual driving performance of occasional and heavy cannabis users (Maastricht University, Netherlands) 8) Effects of opioid analgesics on actual driving performance of pain patients (BASt, Germany/ Maastricht University, Netherlands).

8 DRUID 6th framework programme Deliverable 1.2.2 Antipsychotics 9) The effects of risperidone on driving performance of ambulant schizophrenic patients diagnosed with a psychosis (SIPSiVi, Italy) What all studies have in common is their use of patient populations or regular users of a drug under study. The only exception is the dose effect study of codiliprane on simulated driving that was conducted in healthy volunteers. As such, results from the present studies should provide a high face validity relative to many previous studies that have been conducted in healthy volunteers, because of its undisputed relevance and generalization to user populations. In general, the experimental studies employed either placebo controlled, cross-over within subjects designs (studies 5 and 6) a between group design (studies 2, 4, 8 and 9) or mixed design (studies 1, 3 and 7). The studies furthermore proceeded from conventional laboratory testing of psychomotor skills and cognition to sophisticated driving simulators (i.e. University of Caen, INRETS, CERTH/HIT), actual driving tests on a closed course (SIPSiVi) and actual on-the-road driving tests (Maastricht University, BAST) for establishing the driving hazard potential of the respective drugs. Driving tests were conducted at Tmax, when drug concentrations were maximal or, in case of hypnotic drug studies in the morning after a nocturnal dose in order to assess residual effects on driving. More details on study designs, screening, subject characteristics and in- and exclusion criteria can be found in the separate study reports that are included as separate chapters in the present deliverable. All studies adhered to the following common set of pre-defined instructions with respect to study procedures: · Number of subjects: the minimum number of subjects was 16. The choice for a subjects’ sample-size was always corroborated by a statistical power analysis. · Drug screens: subjects were always tested for drugs in urine prior to administration of drugs or medicines. Urine was checked for 5 major drugs: i.e. THC, benzodiazepines, opiates, stimulants and cocaine. Subjects that tested positive for drugs were dismissed (sent home) and asked to return to the lab at another date and drug negative. · Alcohol screens: subjects were always tested for alcohol (by breathalysing) prior to drug or medicine administration. Subjects that tested positive for alcohol were dismissed (sent home) and asked to return to the lab at another date and alcohol negative. · Driving experience: subjects needed to have a driver’s license. In case of the stimulant studies there will be no demand regarding driving experience as the target population is expected to be very young. · Blood alcohol concentration (BAC): all partners employed a standard BAC unit: i.e. mg/mL · Training sessions: all subjects received training sessions of actual driving tests, simulator driving tests and/or laboratory performance tests in order minimize learning effects. Training was performed in all subjects to achieve a stable performance level prior to study entrance. · Subjective measures: all partners included a set of subjective measures on drug effects (e.g. alertness, mental effort etc). · Ethics: all partners obtained study approval from their local (and national) ethics review boards

9 DRUID 6th framework programme Deliverable 1.2.2 and conducted their study according the declaration of Helsinki and Good clinical practice.

Standard driving parameters

All partners adhered to a standard set of driving parameters to increase comparability between studies. These driving parameters basically covered 3 core levels of driving behaviours: · Automated behaviours – Well-learned (over-learned) skills · Controlled behaviours – Controlled manoeuvres in traffic · Executive, strategic behaviours - Interactive functions with ongoing traffic, planning, risk taking

All partners agreed on a minimum of 2 driving scenarios to be included in each and every study. These scenarios represent the behavioural levels above, and constituted the primary driving measures over all studies.

Road tracking scenario (automated behaviours) The road tracking scenario was based on the Road Tracking Tests that has been used in the Netherland in over 100 studies for measuring drug effects on driving (O'Hanlon et al 1982). Participants are required to drive a 100km course maintaining a constant speed of 95 km/h and a steady lateral position in traffic lanes. The primary driving measure is the standard deviation of lateral position or SDLP. SDLP is an index of road tracking error or weaving, swerving and overcorrecting. SDLP is measured using an electro-optical device mounted on the rear of the vehicle which continuously records lateral position relative to the traffic lane. An increase in SDLP, measured in centimeters, indicates driver impairment, as the driver’s ability to hold the car in a steady lateral position diminishes.

Car-Following scenario (controlled behaviours) The Car Following task was developed to measure attention and perception performance, as errors in these areas often lead to accident causation. In this task participants are required to match the speed of a lead vehicle and to maintain a constant distance from the vehicle as it executes a series of deceleration and acceleration manoeuvres. The primary dependant variable is reaction time to lead vehicle’s speed decelerations. This test assesses a driver’s ability to adapt to manoeuvres of other motorists . (Brookhuis and de Waard 1993; Ramaekers and O'Hanlon 1994).

Risk taking scenario (strategic behaviours) Risk taking scenarios were only embedded in studies using a driving simulator. Standard parameters that were used by respective partners were gap acceptance, number of crashes, number of red light crossings and number of crashes during sudden event scenarios.

In addition, all partners including a number of laboratory tests measuring skills related to driving.

10 DRUID 6th framework programme Deliverable 1.2.2 These test included tracking tasks, attention tasks, reaction tasks and cognitive tasks. Performance parameters associated with these laboratory tests were considered secondary driving parameters.

Definition of clinically relevant drug effects

All partners employed alcohol effects on driving parameters as a standard reference to quantify impairment for any other drug. Any drug induced performance change > performance change induced by BAC 0.5 mg/mL was qualified as a clinical relevant drug effect. Drug effects equal to those produced by a BAC of 0.5 mg/ml were also considered to define the “threshold”of impairment for an individual drug. Drug effects were tested for comparability to BAC 0.5 mg/ml effects by means of equivalence testing (see section statistics). All partners conducted a placebo-controlled alcohol study in order to calibrate their primary driving parameters for the effects of BAC 0.5 mg/ml.

Standard toxicological analyses All partners collected whole blood, serum and blood spots for determining concentrations of dexamphetamine and MDMA during driving tasks. Blood samples were analysed by dr Gisela Skopp at the University of Heidelberg. Analysis was performed on whole blood, plasma and DBS by LC/MS/MS following evaluation of the analytical method according to international guidelines. B/p ratios were derived from in vitro partition experiments (different hematocrit values) and from corresponding blood and plasma samples (ex vivo). Bland Altman analysis was used to test agreement of concentrations determined from whole blood and corresponding DBS. Toxicological analysis served two objectives: 1) determination of drug concentration in whole blood and corresponding plasma samples to estimate ex vivo blood to plasma (b/p) ratios, and 2) comparison of drug levels in whole blood and corresponding DBS. A detailed report of toxicological analysis in blood is given in Chapter 11.

Standard statistical methods All primary driving parameters were analysed according to the following predefined statistical procedures.

Superiority testing and equivalence testing The general statistical analyses consisted of 2 steps: 1) Assessment of overall treatment effect by means of superiority testing (e.g. ANOVA for within group comparisons). Superiority testing basically indicates whether drug effects differ from placebo. 2) Equivalence testing of drug effects was based on difference scores from placebo (within group) relative to the alcohol criterion (i.e. equivalence to a BAC of 0.5 mg/ml). Basically, equivalence testing assessed whether the alcohol criterion values falls within the 95% CI for the drug effect. If yes, than the drug effect was considered equivalent to a BAC of 0.05 mg/ml (and thus clinically relevant for traffic safety). If the 95% CI was below the alcohol

11 DRUID 6th framework programme Deliverable 1.2.2 criterion value than a drug effect was considered not relevant

Concentration effect relations. Concentration effect relations were conducted for those studies that administered multiple doses of a dronabinol (Maastricht University) and codiliprane (University of Caen/INRETS). Data sets were analyzed according to a 2 step procedure. Data collected during different doses of a drug were converted into difference scores from placebo for analyses of the association between drug concentration and performance (i.e.: difference score = performance during drug treatments - performance during placebo treatment). A linear regression analysis was conducted to establish linear relationships between changes (from placebo) in task performance during drug treatment and log-transformed drug concentrations in serum. The total number of data points included in these equations was defined by the number of subjects x maximal number test repetitions x the number of drug doses. Second, individual drug concentrations in serum prior to performance assessments in each of the drug dose conditions was divided over a number mutually exclusive categories covering the full range of drug concentrations in a particular study. Corresponding change scores of task performance were then classified either as showing “impairment”or “no impairment”for all individual cases within each of these categories. Impairment was defined as a positive change score from placebo in case of SDLP (road tracking) and RT to speed decelerations (Car Following). Binomial tests were applied to measure whether the proportion of observations showing impairment or no impairment significantly differed from the hypothesized proportion. It was hypothesized that in case of no effect of a drug on task performance the proportion of observations showing impairment or no impairment will be equal; i.e. 50 percent.

Main results and discussion

The present section summarizes the main results from all experimental studies. For full reports of the studies please see Chapters 1-11.

Studies on hypnotics, anxiolytics and sleep disorder

Two experimental studies were specifically designed to compare the effects of controlled administration of zopiclone 7.5 mg and alprazolam 0.5mg on driving performance in patients and healthy controls. Both drugs had previously been shown to produce significant driving impairment in healthy volunteers (Vermeeren et al, 2004; Leufkens et al, 2007; Verster et al, 2004) but the generalisation of such results to patient population has been debated. The present studies demonstrated that both zopiclone and alprazolam produced significant driving impairment in patients as well as in healthy controls. Nocturnal doses of zopiclone 7.5mg significantly increased SDLP by 2.6 cm in chronic users of hypnotics, 2.1 cm in infrequent users of

12 DRUID 6th framework programme Deliverable 1.2.2 hypnotics and 3.6 cm in healthy volunteers during morning driving at 10-11 hrs after drug intake. The increment in SDLP was significantly less in the chronic users as compared with the controls. However, equivalence testing amply demonstrated that effects of zopiclone on SDLP in all groups were bigger or comparable to the effect observed after a BAC of 0.5 mg/mL. Laboratory measures of skills related to driving confirmed findings obtained from the driving tests, and demonstrated impairments of psychomotor and cognitive functions during zopiclone in all groups. Alprozalam 0.5 mg increased SDLP by 5.8 cm in medicated anxious patients, 4 cm in unmedicated anxious patients and 6.8 cm in healthy controls during simulated road tracking conducted between 1-2 hrs post dosing. These effects largely exceeded effects observed after a BAC=0.5 mg/ml during road tracking in the same driving simulator which indicates high clinical relevance. Moreover, alprazolam also increased brake reaction time and increased the time driven a close proximity to a leading vehicle in a car-following test. In general, these results can be taken to show that both zopiclone and alprazolam produce significant driving impairment. Results from the present studies in patients, confirm previous findings of driving studies conducted in healthy volunteers. This suggests that healthy volunteer studies can serve as valid model to predict hypnotic and anxiolytic drug effects in patients populations. Chronic users of hypnotics or anxiolytics did however significantly differ from infrequent users and healthy controls in one respect. Chronic users did not experience any sedative effects of zopiclone and alprazolam, whereas infrequent users and healthy users reported feelings of reduced alertness and sleep. This lack of awareness of (residual) sedative effects of zopiclone and alprazolam may lead insomnia and anxious patients to belief that car driving is safe during treatment with these drugs. Objective results for car driving measures however clearly indicate that driving was significantly impaired after both drugs. These results furthermore stress the major importance of prescribing physicians to warn their patients about the impairing effects of zopiclone and alprazolam on driving performance. Two additional studies were designed to compare general driving performance of insomnia patients and sleep apnoea patients to that of healthy controls. Insomnia patients had to meet the inclusion criteria for primary insomnia according to DSM-IV. Inclusion criteria for sleep apnoea patients OSAS were based an Apnoea-Hypopnoea Index (AHI) equal or larger than 10. Both studies followed identical designs: driving performance was assessed in treated patients, untreated patients and controls. In the insomnia study treatment implied the use of hypnotics on a prescription base. In the sleep apnoea study, treatment implied the use of continuous positive airway pressure (CPAP) as a mean to prevent cessations and partial obstructions of breathing. Results from the insomnia study showed that driving performance and driving related psychomotor performance did not differ between medicated insomniacs, unmedicated insomniacs and normal sleepers (i.e controls). These results indicate that driving performance of insomniac patients does not differ from that of normal sleepers, even in insomniacs that had been prescribed hypnotic mediation. The lack of driving impairment in medicated insomniacs could however be predicted from the type of hypnotics that patients were using. About 2/3 of the patients received short acting hypnotics or low doses of hypnotics that peviously were shown not to procedure any residual impairment in driving test ( e.g. zolpidem, temazepam). It is however more important to note here that therapeutic treatment of insomniacs with hypnotics did not

13 DRUID 6th framework programme Deliverable 1.2.2 improve driving performance of insomniac patients either. This shows, that potential impairing effects of hypnotics of driving performance will not automatically be compensated by a (presumed) beneficial, therapeutic of hypnotics on sleep. Driving performance of sleep apnoea patients receiving CPAP or no treatment was markedly impaired as compared to controls. SDLP during road tracking was about 10 cm higher in both apnoea groups relative to the control. In sleep apnoea patients displayed longer break reaction times and drove at closed proximity to leading vehicles as compared to controls. Driving impairment was comparable or even worse than that of anxious patients after alprazolam administration (see above) and indicated severe risk of traffic injury in this patient group. The current finding that sleep apnoea, in contrast to insomnia, produces severe driving impairment corroborates a large body of previous research (for review: Fulda and Schultz, 2001). In general, studies in sleep apnoea patients have consistently shown major driving and cognitive dysfunction in this population. Such deficits are generally less pronounced or even absent in insomnia patients.

Studies on opioid and non-opioid analgesics

Two studies were conducted to assess the effects of codeine/paracetamol combinations on simulated driving performance. Combinations of codeine and paracetamol are available on the European market under the brand name Codiliprane®. Codeine is an alkaloid found in the poppy and currently one of the most widely used opiates in the world. Codeine is a prodrug that is metabolized into the primary active compounds morphine and codeine-6-glucuronide (C6G). Codeine is, in general, used in single doses up to 60 mg (and no more than 240 mg in 24 hours). Paracetamol is widely used over the counter that is commonly used for the relief of headaches. Codiliprane tablets contain codeine 20mg and paracetamol 400mg. The “codiliprane” studies were conducted in order 1) to establish the effects of codeine and paracetamol on simulated driving peformance as a function of dose and serum concentration; and 2) to asses the effects of codiliprane and zolpidem, an hypnotic often used in combination with codiliprane in elderly patients. The former study was conducted in young healthy volunteers according to a double blind, placebo control study design including 3 codeine/paracetamol combinations: i.e. codeine 20mg/ paracetamol 400mg; codeine 40mg/ paracetamol 800mg and codeine 60mg/ paracetamol 1200 mg. The latter study was conducted in elderly volunteers according to a placebo controlled, cross-over study design assessing simulated driving after single doses of zolpidem 10mg, codeine 20mg/paracetamol 400mg and their combination. Results from the 1st study did not show any effect of the 3 codeine/paracetamol combinations on simulated driving parameters or laboratory measures of skills related to driving. Likewise, analyses of concentration effect relations for morphine, codeine and paracetamol also did not shown any significant association between drug concentration and the primary driving measures although correlations between SDLP and morphine/codeine concentration approached significance. The 2nd study assess the effects of single doses of codeine 20mg/paracetamol 400mg alone or in combination with zolpidem 10mg in elderly volunteers however demonstrated driving impairment after single doses of these when given alone. Codeine/paracetamol increased SDLP in the road tracking test and increased the number of crashes during the driving simulation. Nocturnal doses of zolpidem also

14 DRUID 6th framework programme Deliverable 1.2.2 increased SDLP to similar degrees (i.e. about 3 cm). The combination of both products did not affect any of the primary measures of simulated driving. The combination did however produce performance impairments on secondary measures of simulated driving such as speed and number of line crossings during road tracking. In general, data from the coliliprane studies provides conflicting results. The combinations of codeine/paracetamol produced no effect on driving when administered to young, healhy volunteers, even when given in high doses. In contrast, a low dose of codeine and paracetamol did produce driving impairment when administered to elderly volunteers. The present data thus seems to indicate that the impairing potential of codeine/paracetamol varies with age. The second analgesic compound that was included to assess its effects on driving is dronabinol. Dronabinol (Marinol®) is a cannabinoid that is used for the treatment of chronic pain, anorexia in AIDS and other wasting diseases, and as an antiemetic medication in cancer patients undergoing chemotherapy. The active ingredient dronabinol is synthetic ¨9- (THC). THC has been extensively shown to produce driving impairment in experimental studies and has been associated with increased crash risk in epidemiological studies (Ramaekers et al, 2004). The effects of dronabinol on driving have never been assessed before but are expected to be very similar to the effects of smoked cannabis. The present study was designed to assess the effects of single doses of dronabinol (10 and 20mg) on actual driving performance in 12 occasional and 12 daily cannabis users according to a double-blind, randomized, 3-way cross-over design. The study was performed in cannabis users in order to assess the effects of acute and chronic use of THC on driving. Single dose effects of dronabinol in occasional users represented acute effects of THC, whereas single dose effects of dronabinol in daily users represented chronic effects of THC use on driving. Results demonstrated that single doses of dronabinol impaired road tracking performance of occasional cannabis users during on-the-road driving tests in a dose related manner. Relative to placebo, SDLP significantly increased by approximately 2.5 and 4 cm respectively, after dronabinol 10 and 20 mg. The upper limits of the 95%CI associated with change SDLP exceed the alcohol criterion limit indicating that dronabinol effects after both doses on SDLP were comparable to the effects of a BAC=0.5 mg/ml. The effects of dronabinol on driving performance of heavy cannabis users however were less pronounced or even absent. Overall, dronabinol did not affect any driving measure as compared to placebo. Equivalence testing however indicated that the 95%CI associated with change scores in SDLP after dronabinol 10 and 20mg contained both the alcohol criterion. The 95%CI for the low dose also included the value zero. This basically indicates large individual variation in change SDLP after both doses of dronabinol. This suggests that tolerance to the impairing effects of high doses of dronabinol is not complete and that a high dose of dronabinol may cause impairment on some regular users of cannabis but not in others. These data are in line with previous research on the effects of smoked cannabis on driving, showing that THC markedly impairs driving and psychomotor skills in occasional users and that such effects mitigate in regular cannabis users, due to development of tolerance. Finally, a study was conducted to assess driving performance of 26 pain patients who had been receiving chronic treatments with either transdermal , transdermal buprenorphine, retarded oxycodone (sometimes in combination with ), retarded or retarded

15 DRUID 6th framework programme Deliverable 1.2.2 morfine. They performed a standardized on-the-road “road tracking test”and “car-following test”in and conducted a number neuropsychological tests, including the Wiener Test Battery. Patient performance was compared to that of healthy controls who performed the same test both under placebo and alcohol (BAC=0.5 mg/ml) conditions. Results from the driving test revealed that driving performance of patients was comparable to that of healthy controls. Neuropsychological tests measuring skills related to driving revealed that pain patients performed less as compared to healthy control on a number of tests. However, these differences could also be contributed to differences in age between both groups. Patients were generally older than their healthy controls. This may be an important bias as age was shown to correlate positively with driver impairment on these tests. A single dose of alcohol (i.e BAC=0.5 mg/ml) on skills related to driving did not produce any change in performance as compared to placebo.

Study on antipsychotics

One partner scheduled a study to assess driver performance of schizophrenic patients with a history of psychotic episodes receiving risperidone or treatment. Their driving performance was compared to that of a group of healthy controls who were treated with placebo and a single dose of alcohol. The patient group consisted of 15 ambulant schizophrenic patients that were recruited in collaboration with the psychiatrist health department of Rome. The inclusion criteria were: patients treated with risperidone or paliperidone, alone or with mood stabilizing medicines; regular use of fixed doses for at least three months (3-4 mg/day); diagnosis of schizophrenia, paranoia or bipolar psychosis. Other inclusion criteria were the possession of a normal driving license, driving a car at least once a week. The driving test procedures that were employed in the current study markedly differed from the general description of standardized driving test procedures given above. The study was conducted at a closed driving course that posed a number of limitations on driving procedures and driving environment. Due to the sharp curves in the closed-course, driving tests were conducted at a maximum speed of only 30 km/h only. The closed course consisted of a single lane with many curves and few straight sections. Consequently, it was impossible to conduct a road tracking test according to standardized procedures (e.g. mean driving speed 95 km/h, 1 hour driving of straight primary highway) in the given circumstances. As an alternative, the road tracking test was conducted at a driving speed of 30km/h while measuring the lateral positions relative to the right side lane delineation. The drivers were instructed to maintain a constant distance of 30 cm from right line during road tracking. SDLP was calculated separately for straight section as well as curved sections of the driving course. The test was conducted twice: i.e. with and without feedback on the actual driving speed. Risk taking was measured by means of a sudden event scenario. Drivers had to respond to a sudden crossing of a “bobby car”in order to assess their brake reaction time. Results demonstrated that overall mean lateral position and mean brake reaction time to sudden events differed between the schizophrenic patient group and controls receiving alcohol or placebo. The overall effects on lateral position were primarily caused by alcohol relative to placebo.

16 DRUID 6th framework programme Deliverable 1.2.2 Direct comparisons between driving performance of patients receiving risperidone and healthy controls receiving placebo did not reveal any significant differences. However, evaluations of SDLP and reaction time to sudden events indicated the performance of the patient group was comparable or worse as compared to healthy controls driving under the influence of a BAC= 0.5 mg/ml. In addition, driving performance of 11 patients was also evaluated on the standardized Vienna Test Battery. In total, 9 patients passed this driving evaluation tests and 2 patients failed the test. Together these results suggest that most driving parameters in this study were not sensitive to the effects of alcohol and risperidone. However the standard deviation of lateral position and reaction time to sudden events were signifanctly increased and comparable or bigger than those observed after a blood alcohol concentration of 0.5 mg/ml. The present data thus seems to indicate that patients under the influence of risperidone do demonstrate some impairments that should be considered of clinical relevance. It is however impossible to determine from the current data whether these impairments in patients were actually caused by risperidone, the underlying medical disease (psychosis) or both. The general lack of differences in driving performance between patients and controls (during placebo) seems to indicate that driving abilities of these groups are comparable. However, the general lack of alcohol effects on many driving parameters also indicated that these measures may not have been very sensitive to pick up impairments. The present data therefore do not provide a definitive answers to the question whether patients on risperidone are fit to drive. The fact that at least two driving parameters demonstrated relevant impairments in patients however potentially indicates an elevated risk in these patients. Individual patients using risperidone should we warned accordingly by their prescribing doctors when discussing their driving capabillities.

17 DRUID 6th framework programme Deliverable 1.2.2 Table 1. Summary of treatment and sleep disorder effects on primary and secondary driving parameters as well as subjective measures of arousal or sleep (ZOP=zopiclone; PLA=placebo; BAS=baseline and ALP=alprazolam)

Study 1: Study 2: Study 3: Study 4: Residual effects of zopiclone 7.5 mg Insomnia patients Alprazolam 0.5 mg in anxious patients Sleep apnoea (Maastricht University) (Maastricht University) (CERTH/HIT) patients (CERT/HIT) Chronic hypnotic Infrequent hypnotic Healthy controls Medicated insomnia Medicated Unmedicated Healthy controls Patients with CPAP users users patients vs unmedicated anxious patients anxious patients vs patients with no ZOP vs PLA ZOP vs PLA ZOP vs PLA insomnia patients vs ALP vs BAS ALP vs BAS ALP vs BAS CPAP vs controls healthy controls

Road tracking Increased SDLP; Increased SDLP; Increased SDLP; No difference between Increased SDLP; Increased SDLP; Increased SDLP; Increased SDLP in Impairment > BAC Impairment > BAC Impairment > BAC groups Impairment > Impairment > Impairment > both patients groups 0.5 mg/ml 0.5 mg/ml 0.5 mg/ml BAC 0.5 mg/ml BAC 0.5 mg/ml BAC 0.5 mg/ml relative to controls No difference between patient groups

Car-Following No effect No effect No effect No difference between Increased brake Increased brake Increased brake Increased brake groups reaction time reaction time reaction time reaction time and and increased and increased and increased increased time driven time driven at time driven at time driven at at close distance to close distance to close distance to close distance to leading vehicle in both leading vehicle leading vehicle leading vehicle patient groups

Risk Taking Not assessed Not assessed Not assessed Not assessed Not assessed Not assessed Not assessed Not assessed

Laboratory measures of Impairment of Impairment of Impairment of No group differences in No effect No effect Decreased No difference skills related to driving memory, tracking, memory, tracking, memory, tracking, verbal memory, divided reaction time between groups divided attention, divided attention, divided attention, attention, vigilance and inhibitory control inhibitory control inhibitory control inhibitory control.

Subjective measures Increased next day No effect Decreased next No group differences in No effect Decreased Decreased Decreased alertness (sleepiness/alertness) alertness day alertness sleepiness and alertness alertness alertness in patient groups

18 DRUID 6th framework programme Deliverable 1.2.2 Table 2. Summary of treatment and pain disorder effects on primary and secondary driving parameters as well as subjective measures of arousal or sleep (COD= codiliprane; PLA=placebo; ZOL = zolpidem; DRO=dronabinol)

Study 5: Study 6: Study 7: Study 8: Codiliprane and zolpidem in elderly: alone and in Codiliprane in healthy Dronabinol in THC users Opioid patients combination volunteers (Maastricht University) (BASt//Maastricht (University of Caen/ INRETS) (University of University) Caen/INRETS) COD vs PLA ZOL vs PLA COD+ZOL vs COD vs PLA Occasional THC users Heavy THC users Patients treated with PLA DRO vs PLA DRO vs PLA opioids vs controls

Road tracking Increased SDLP Increased SDLP No effect No effect Increased SDLP; No overall superiority No difference between Impairment DRO 10mg > effect of DRO on SDLP groups BAC 0.5 mg/ml Big individual variation in Impairment DRO 20mg > change SDLP after DRO BAC 0.8 mg/ml 10 and 20mg : 95% CI includes zero as well as alcohol criterion value

Car-Following No effect No effect No effect Not assessed Decreased Time to Speed No effects No difference between Adaption groups

Risk Taking Increased No effect No effect Not assessed Not assessed Not assessed Not assessed number of crashes

Laboratory measures of Not assessed Not assessed Not assessed No effects No effect on Standard No effect on Standard Patient perform worse than skills related to driving Filed Sobriety Tests Filed Sobriety Tests controls on some isolated tests

Subjective measures Not assessed Not assessed Not assessed Small increase of Increased sedation Increased sedation No difference between (sleepiness/alertness) sleepiness groups

19 DRUID 6th framework programme Deliverable 1.2.2 Table 3. Summary of risperidone/paliperidone effects on primary and secondary driving parameters as well as subjective measures of arousal or sleep

Study 9: Risperidone/paliperidone (SIPSiVi) patients diagnosed with psychosis and receiving risperidone vs controls during placebo and alcohol

Road tracking Increase in SDLP of patients > increase in SDLP of controls during alcohol (BAC=0.5 mg/ml). Alcohol also affected lateral position. The latter was not affected in patients

Car-Following Not assessed

Risk Taking RT to sudden events increased in patients and controls during alcohol relative to controls during placebo. Increase in RT in patients was comparable/bigger than BAC=0.5 mg/ml

Laboratory measures of 9 out 11 patients passed the Vienna skills related to driving driving evaluation test

Subjective measures Not assessed (sleepiness/alertness)

Toxicology Toxicological analyses were conducted in order to 1) determine drug concentrations in whole blood and corresponding plasma samples to estimate ex vivo blood to plasma (b/p) ratios, and 2) to compare drug levels in whole blood and corresponding Dry Blood Spots (DBS). The mean obtained b/p ratios were: hydromorphone: 1.04, morphine: 1.03, fentanyl: 0.87, norfentanyl: 1.19, oxycodone: 1.48, noroxycodone: 1.73, alprazolam: 0.81, zopiclone: 0.89, temazepam: 0.71, risperidone: 0.65, 9-OH-risperidone: 0.73. Ratios were close to reliably established ratios published in the current literature as far as available. For all analytes except zopiclone, the mean ratio of the DBS and blood concentrations and their relative standard deviations indicated that DBS analysis is as reliable as analysis from whole blood. Bland-Altman difference plots for the several substances supported this thesis. Zopiclone is expected to undergo degradation even in DBS. It is concluded that dividing the concentrations of the analytes blood by the obtained b/p ratios may give a reasonably good estimate of the coexisting concentration in plasma. There is sound evidence that the DBS assay has potential as a precise and inexpensive option for the determination of the investigated analytes in small blood samples. However, stability of zopiclone in DBS has to be tested and compared to the stability in

20 DRUID 6th framework programme Deliverable 1.2.2 whole blood specimens using a stability indicating method.

Conclusions

· Zopiclone 7.5mg and alprazolam 0.5mg produced significant driving impairment in patients as well as in healthy controls. · Zolpidem 10mg produced significant driving impairment in elderly subjects. · Healthy volunteer studies can serve as valid model to predict hypnotic and anxiolytic drug effects in patients populations. · Chronic users do subjectively not experience any sedative effects of zopiclone and alprazolam, whereas infrequent users and healthy users reported feelings of reduced alertness and sleep. This lack of awareness of (residual) sedative effects of zopiclone and alprazolam may lead insomnia and anxious patients to belief that car driving is safe during treatment with these drugs. · Diagnosis of sleep apnoea, but not insomnia, is a strong predictor of driver impairment. · Combinations of codeine and paracetamol in general do not produce driving impairment when assessed in healthy volunteers even at higher doses. However, driving impairment became apparent after the lowest dose when administered to elderly subjects. · Dronabinol impaired driving performance in occasional and heavy users in a dose-dependent way. Equivalence tests demonstrated that dronabinol induced increments in SDLP were bigger than impairment associated with BAC of 0.5 mg/mL in occasional and heavy users, although the magnitude of driving impairment was generally less in heavy users. · Ambulant patients receiving risperidone/paliperidone is comparable to that of healthy controls. Patients that are stabilized on risperidone/paliperidone treatment should be allowed to operate a vehicle. · Patients using risperidone drove with a lateral position that was comparable to that observed in controls. However the standard deviation of lateral position and reaction time to sudden events were significantly increased in patients and comparable or bigger than those observed in controls with blood alcohol concentration of 0.5 mg/ml. The present data thus seems to indicate that patients under the influence of risperidone do demonstrate impairments that should be considered of clinical relevance · There is sound evidence that the DBS assay has potential as a precise and inexpensive option for the determination of the investigated analytes in small blood samples. However, stability of zopiclone in DBS has to be tested and compared to the stability in whole blood specimens using a stability indicating method.

References

Brookhuis KA, de Waard D (1993): The use of psychophysiology to assess driver status. Ergonomics 36:1099-1110. EMCDDA (2008): Drug use, impaired driving and traffic accidents: European Monitoring Centre for Drugs and Drug Addiction, Lisbon.

21 DRUID 6th framework programme Deliverable 1.2.2 Fishbain DA, Cutler RB, Rosomoff HL, Rosomoff RS (2003): Are opioid-dependent/tolerant patients impaired in driving-related skills? A structured evidence-based review. J Pain Symptom Manage 25:559-577. King DJ, Henry G (1992): The effect of neuroleptics on cognitive and psychomotor function. A preliminary study in healthy volunteers. Br J Psychiatry 160:647-653. McClelland GR, Cooper SM, Pilgrim AJ (1990): A comparison of the central nervous system effects of haloperidol, chlorpromazine and in normal volunteers. Br J Clin Pharmacol 30:795-803. O'Hanlon JF, Haak TW, Blaauw GJ, Riemersma JB (1982): Diazepam impairs lateral position control in highway driving. Science 217:79-81. Orriols L, Delorme B, Gadegbeku B, Tricotel A, Contrand B, Laumon B, et al (2010): Prescription medicines and the risk of road traffic crashes: a French registry-based study. PLoS Med 7:e1000366. Ramaekers JG, Berghaus G, van Laar M, Drummer OH (2004): Dose related risk of motor vehicle crashes after cannabis use. Drug Alcohol Depend 73:109-119. Ramaekers JG, Louwerens JW, Muntjewerff ND, Milius H, de Bie A, Rosenzweig P, et al (1999): Psychomotor, Cognitive, extrapyramidal, and affective functions of healthy volunteers during treatment with an atypical () and a classic (haloperidol) antipsychotic. J Clin Psychopharmacol 19:209-221. Ramaekers JG, O'Hanlon JF (1994): Acrivastine, terfenadine and diphenhydramine effects on driving performance as a function of dose and time after dosing. Eur J Clin Pharmacol 47:261-266. Saletu B, Grunberger J, Linzmayer L, Anderer P (1987): Comparative placebo-controlled pharmacodynamic studies with and clozapine utilizing pharmaco-EEG and psychometry. Pharmacopsychiatry 20:12-27. Strand M, Fjeld B, Arnestad M, Morland J (2011): Psychomotor relevant performance after administration of opioids, narcoanalgesics and hallucinogens: Norwegian Institute of Public Health, Oslo, DRUID Deliverable 1.1.2.c. Van Laar MVE (1998): Driving and benzodiazepine use: evidence that they don't mix. CNS drugs:383-396. Vermeeren A (2004): Residual effects of hypnotics: epidemiology and clinical implications. CNS Drugs 18:297-328. Verster JC, Veldhuijzen DS, Volkerts ER (2004): Residual effects of sleep medication on driving ability. Sleep Med Rev 8:309-325. Woods JH, Katz JL, Winger G (1992): Benzodiazepines: use, abuse, and consequences. Pharmacol Rev 44:151-347. Wylie KR, Thompson DJ, Wildgust HJ (1993): Effects of depot neuroleptics on driving performance in chronic schizophrenic patients. J Neurol Neurosurg Psychiatry 56:910-913. Zacny JP (1996): Should people taking opioids for medical reasons be allowed to work and drive? Addiction 91:1581- 1584. Zacny JP, Hill JL, Black ML, Sadeghi P (1998): Comparing the subjective, psychomotor and physiological effects of intravenous and morphine in normal volunteers. J Pharmacol Exp Ther 286:1197-1207.

22 DRUID 6th framework programme Deliverable 1.2.2 Chapter 1: Residual effects of zopiclone 7.5 mg on highway driving performance in insomnia patients and healthy controls: a placebo controlled crossover study

T.R.M. Leufkens1, J.G. Ramaekers1, A.W. de Weerd2, W.J. Riedel1 & A. Vermeeren1,*

1Department of Neuropsychology and Psychopharmacology, Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, The Netherlands 2Department of Clinical Neurophysiology and Sleep Centre SEIN, Zwolle, The Netherlands

Date: 7/7/2011

*Corresponding author: Dr. A. Vermeeren Department of Neuropsychology & Psychopharmacology Faculty of Psychology and Neuroscience Maastricht University PO Box 616 6200 MD Maastricht, The Netherlands Tel: +31 43 388 1952 Fax: +31 43 388 4560 E-mail: [email protected]

23 DRUID 6th framework programme Deliverable 1.2.2 Abstract

Residual effects of hypnotics on driving performance have been mainly determined in studies using a standardized driving test with healthy good sleepers. Responses to effects may differ, however, between insomniacs and healthy volunteers due to the underlying sleep disorder. In addition, a majority of insomniacs uses hypnotics chronically resulting in the development of tolerance to impairing effects. Impaired driving performance in healthy volunteers may then be an overestimation of the actual effects in insomniacs. The present study aims to compare the residual effects of zopiclone 7.5 mg on driving performance of 16 middle-aged insomniacs chronically using hypnotics (chronic users), 16 middle-aged insomniacs not or infrequently using hypnotics (infrequent users) and 16 healthy, age matched, good sleepers (controls). The study was conducted according to a 3x2 double-blind, placebo controlled crossover design, with three groups and two treatment conditions. Treatments were single oral doses of zopiclone 7.5 mg and placebo administered at bedtime (23:30 hours). Between 10 and 11 hours after administration subjects performed a standardized highway driving test. Results indicated that zopiclone 7.5 mg significantly impaired driving performance in both insomnia groups and healthy controls. The magnitude of impairment was significantly less in the chronic users as compared with the controls. Effects found in the infrequent users were in line with previous studies, suggesting that these studies are able to validly predict the residual effects of hypnotics in insomnia patients who do not or infrequently use hypnotics. Keywords: zopiclone, hypnotics, residual effects, insomnia, on-the-road driving

24 DRUID 6th framework programme Deliverable 1.2.2 Introduction

Residual daytime sedation is one of the main problems associated with hypnotic drug use. Experimental studies have demonstrated that the sedative actions of hypnotics impair psychomotor and cognitive functioning the morning after evening administration (Vermeeren, 2004). The related reduced alertness and slowed reactions are a particular problem for individuals who have to drive a car the morning following an evening dose. Epidemiological studies have shown that use of benzodiazepines, as well as zopiclone, is associated with an increased risk of car accidents (Hemmelgarn et al., 1997; Barbone et al., 1998; Neutel, 1998; Glass et al., 2005). The severity and duration of residual effects on actual driving performance of hypnotics have been determined in experimental studies using a standardized driving test (Vermeeren, 2004). Most of those studies have been conducted in healthy volunteers rather than in the target population, i.e. patients suffering from insomnia. Responses to the residual effects of hypnotics, however, may differ between insomnia patients and healthy good sleepers due to the underlying sleep disorder. In insomnia patients, hypnotics are expected to improve sleep and, as a consequence, they are expected to improve daytime performance as well. This improvement is supposed to attenuate or even compensate for the impairing effects of hypnotics. In addition, the majority of insomnia patients use hypnotics for prolonged periods (Curran et al., 2003), which may result in the development of tolerance to the impairing effects. Impaired driving performance found in healthy medication naïve volunteers may then be an overestimation of the actual effects in insomnia patients. To date, there is a lack of experimental studies that assess driving performance following hypnotic administration in insomnia patients. Moreover, the residual effects of hypnotics on driving have not yet been directly compared between insomnia patients and healthy good sleepers. Recently, an experimental study explored driving performance between pharmacologically treated and untreated insomnia patients and healthy, good sleepers (Leufkens et al., in prep.). Results showed that performance was not significantly different between insomnia patients and healthy controls. In addition, there were no significant differences between insomnia patients who chronically used hypnotics and patients who used hypnotics infrequently. A limitation of that study was, however, that the chronic users group used a variety of hypnotic drugs most of which were not expected to produce residual sedation at all. In addition, variability in dose and half-life may have added to the absence of any performance impairment. In order to determine residual effects of hypnotics on driving performance in insomnia patients, studies need to be conducted with hypnotics that have been shown to produce residual impairment in healthy volunteer studies, such as zopiclone (Vermeeren et al., 1998; Vermeeren et al., 2002; Leufkens et al., 2009; Leufkens and Vermeeren, 2009). Therefore, the present study aims to compare the residual effects of the frequently prescribed hypnotic zopiclone 7.5 mg on driving performance of 16 insomnia patients who chronically use hypnotics, 16 insomnia patients who do not or infrequently use hypnotics and 16 healthy, age matched, good sleepers.

Methods

Subjects All subjects in the present study participated in a previous study by Leufkens et al. (Leufkens et al., in prep.). They were asked upon completion of the former study to continue their participation in the present study. In

25 DRUID 6th framework programme Deliverable 1.2.2 the previous study, insomnia patients, in the age range of 52 to 73 years, were initially recruited through a network of local general practitioners in the region of Maastricht, The Netherlands (Regionaal Netwerk Huisartsen, RNH). Possible candidates were selected from a computerized database of the Center for Data and Information Management of Maastricht University (MEMIC) (Metsemakers et al., 1992). This recruitment procedure was subsequently backed up by advertisement in local newspapers. Healthy controls were recruited by advertisements in local newspapers. Three groups of 16 subjects, ranging from 52 to 71 years of age, participated in the present study. Groups were 16 individuals with insomnia who chronically used hypnotics (‘chronic users’; 7 female and 9 male), 16 individuals with insomnia who did not or infrequently used hypnotics (‘infrequent users’; 8 female and 8 male) and 16 self-defined good sleepers, matched for age and driving experience (‘controls’; 7 female and 9 male). Their mean (±SD) ages were 62.6 (4.5) for the chronic users, 62.3 (6.2) for the infrequent users and 62.9 (4.3) for the controls. Insomnia patients had to meet the inclusion criteria for primary insomnia according to DSM-IV (Association, 1994): (i) subjective complaints of insomnia, defined as difficulties initiating sleep (sleep latency >30 min) and/or maintaining sleep (awakenings >30 min); (ii) duration of more than 1 month; (iii) the sleep disturbance causes clinically significant distress or impairment; (iv) insomnia does not occur exclusively during the course of a mental disorder and (v) insomnia is not due to another medical or sleep disorder or effects of medication or drug abuse. Sleep complaints were measured using Dutch versions of the Pittsburgh Sleep Quality Index (PSQI) (Buysse et al., 1989), the Sleep Wake Experience List (SWEL) (Van Diest et al., 1989) and the general version of the Groningen Subjective Quality of Sleep questionnaire (GSQS-gen) (Mulder-Hajonides van der Meulen, 1981). Additionally, a daily journal and the specific version of the Groningen Subjective Quality of Sleep questionnaire (GSQS-spec) (Mulder-Hajonides van der Meulen, 1981) were completed upon arising each morning for two weeks providing subjective estimates of sleep quality. Insomnia patients were assigned to the ‘chronic users’group when they used a benzodiazepine, zopiclone or zolpidem as sleeping medication for at least four nights per week during the previous three months or more. The average (±SD) nightly use of hypnotics in the chronic users group was 6.6 (1.0) nights per week. The average (±SD) duration of their hypnotic use was 7.1 (5.0) years. Hypnotics used were zopiclone (1 patient: 7.5 mg; 2 patients: 3.75 mg), temazepam (1 patient: 20 mg; 2 patients: 10 mg), midazolam (3 patients: 7.5 mg), oxazepam (3 patients: 10 mg; 20 mg; 50 mg), lormetazepam (1 patients: 0.5 mg; 1 patient: 2 mg), nitrazepam (1 patients: 5 mg) and flurazepam (1 patients: 15 mg). Patients not using hypnotics or using hypnotics less than or equal to three days per week were assigned to the ‘infrequent users’group. Ten infrequent users reported ingesting hypnotics at an average (±SD) nightly use of 1.1 (0.8) night per week. Their average (±SD) duration of use was 7.8 (8.1) years. Hypnotics used were zopiclone (3 patients: 7.5 mg), temazepam (1 patient: 20 mg; 4 patients: 10 mg), nitrazepam (1 patient: 5 mg) and lorazepam (1 patient: 10 mg) The other six insomnia patients did not use hypnotics. Self-defined good sleepers did not meet any of the criteria for insomnia and did not use any hypnotics. All participants had to meet the following inclusion criteria: possession of a valid driving license for at least three years; average driving experience of at least 3000 km per year over the last three years; mentally and physically fit to drive; good health based on a pre-study physical examination, medical history, vital signs, electrocardiogram, blood biochemistry, haematology, serology and urinalysis; body mass index (BMI)

26 DRUID 6th framework programme Deliverable 1.2.2 between 19 and 30 kg/m2. Exclusion criteria were history of drug or alcohol abuse; presence of a significant medical, neurological, psychiatric disorder, or sleep disorder other than insomnia; chronic use of medication that affects driving performance, except hypnotics; drinking more than 6 cups of coffee per day; drinking more than 21 alcohol containing beverages per week; smoking more than 10 cigarettes per day. Participants were screened for major psychopathology by use of the Symptom Checklist 90 Revised (SCL-90-R) (Derogatis, 1983), the Beck Depression Inventory (BDI) (Beck et al., 1961), the State-Trait Anxiety Inventory (STAI) (Spielberger et al., 1983) and the Multidimensional Fatigue Inventory (MFI) (Smets et al., 1995). During participation use of caffeine was prohibited from 8 hours prior to arrival on test days, until discharge the next morning. Alcohol intake was not allowed from 24 hours prior to each dosing until discharge. Smoking was prohibited from 1 hour prior to bedtime until discharge. In order to minimize withdrawal symptoms during the placebo night, patients assigned to the chronic users group were instructed to discontinue their hypnotic intake three nights before each treatment period. Chronic users who expected difficulties during the three hypnotic-free nights were provided escape medication, consisting of zolpidem at a maximum of 1 dose of 10 mg per night, to be used only in case of intolerable withdrawal effects. Zolpidem 10 mg was selected to limit variability in hypnotic drugs used and because it is known to be free from residual effects when taken at bedtime before 8 hours of sleep (Vermeeren, 2004). The study was conducted in accordance with the code of ethics on human experimentation established by the World Medical Association’s Declaration of Helsinki (1964) and amended in Edinburgh (2000). The protocol was approved by the medical ethics committee of Maastricht University and University Hospital of Maastricht. Subjects were explained the aims, methods, and potential hazards of the study and they signed a written informed consent prior to any study-related assessments.

Design and treatments The study was conducted according to a 3x2 double-blind, placebo controlled crossover design, with three groups (16 insomnia patients chronically using hypnotics, ‘chronic users’; 16 insomnia patients not or infrequently using hypnotics, ‘infrequent users’; and 16 self-defined good sleepers, matched for age and driving experience, ‘controls’) and two treatment conditions. Treatments were single oral doses of zopiclone 7.5 mg and placebo administered in identical looking capsules and ingested immediately before retiring to bed at 23:30 hours. Treatments orders were balanced within groups (placebo – zopiclone or vice versa). Washout periods between treatments were at least one week.

Assessments Sleep Sleep during treatment nights was evaluated objectively by polysomnography using montage including electroencephalogram, electrooculogram and electromyogram. Sleep stages were visually assessed by qualified technicians according to standardized criteria (Iber et al., 2007). Sleep continuity parameters derived after analysis are sleep onset latency (in min); wake after sleep onset (in min); total sleep time (in min); sleep efficiency (in %); and number of awakenings. Sleep architecture parameters are percentages in

27 DRUID 6th framework programme Deliverable 1.2.2 Stage 1, 2, Slow Wave and REM of the total sleep time. Upon arising subjects completed the specific version of the Groningen Sleep Quality Scale (GSQS- spec) (Mulder-Hajonides van der Meulen, 1981). In addition, subjects estimated sleep onset latency (in min), total sleep time (in min), time awake before rising (in min), and number of awakenings.

Driving performance Driving performance was assessed using two on-the-road driving tests, a highway driving test and a car following test. The Highway Driving Test (O'Hanlon, 1984) measures road tracking performance that is mainly determined by the delay lag between sensory information, execution of motor reaction and the vehicle’s dynamic response. In this test, subjects operate a specially instrumented vehicle over a 100 km (61 mi) primary highway circuit, accompanied by a licensed driving instructor having access to dual controls. The subjects’task is to maintain a constant speed of 95 km/h (58 mi/h) and a steady lateral position between the delineated boundaries of the slower traffic lane. The vehicle speed and lateral position are continuously recorded. These signals are edited off line to remove data recorded during overtaking maneuvers or disturbances caused by roadway or traffic situations. The remaining data are then used to calculate means and standard deviations of lateral position and speed. Standard deviation of lateral position (SDLP in centimeters) is the primary outcome variable. SDLP is a measure of road tracking error or ‘weaving’. The test duration is approximately 1 hour. The Car-Following Test measures changes in controlled information processing such as selective attention, stimulus interpretation and decision making, and speed of an adaptive motor response to events which are common in driving (Brookhuis and De Waard, 1994; Ramaekers and O'Hanlon, 1994). In the test two vehicles travel in tandem over a 2-lane, undivided, secondary highway at 70 km/h (44 mi/h). An investigator drives the leading car and the subject, in the second car, is instructed to follow at a distance between 25 and 35 meter. Subjects are further instructed to constantly attend the leading car since it may slow down or speed up at unpredictable times. They are required to follow the leading car’s speed movements, i.e. maintain the initial headway by matching the velocity of the car to the other’s. During the test, the speed of the leading car is automatically controlled by a modified ‘cruise control’system. At the beginning it is set to maintain a constant speed of 70 km/h and, by activating a microprocessor the investigator can start sinusoidal speed changes reaching amplitude of -10 km/h and returning to the starting level within 50 sec. The maneuver is repeated 6 times. The leading car’s speed and signals indicating the beginning of the maneuver are transmitted via telemetry to be recorded in the following vehicle together with the following vehicle’s speed. Phase-delay converted to a measure of the subject’s average reaction time to the movement of the leading vehicle (RT, in s) is taken as the primary dependent variable in this test. Headway is continuously recorded by means of an optical distance sensor and serves as a control variable. Test duration is approximately 25 minutes.

Cognitive and psychomotor performance Cognitive and psychomotor performance was assessed by use of a battery of laboratory tests for word learning, digit span, critical tracking, divided attention, psychomotor vigilance and inhibitory control. Tests were previously proven to be sensitive to daytime sleepiness or sedation due to use of hypnotics (Vermeeren et al., 1995; Vermeeren et al., 1998; Vermeeren et al., 2002; Verster et al., 2002; Vermeeren,

28 DRUID 6th framework programme Deliverable 1.2.2 2004; Leufkens et al., 2009; Leufkens and Vermeeren, 2009) or insomnia (Fulda and Schulz, 2001). The Word Learning Test (Rey, 1964) is a verbal memory test for the assessment of immediate recall, delayed recall and recognition performance. Fifteen monosyllabic nouns are presented and at the end of the sequence the subject is asked to recall as many words as possible. This procedure is repeated five times and after a delay of at least 30 minutes the subject is again required to recall as many words as possible. At this trial the nouns are not presented. Finally, a sequence of 30 monosyllabic nouns is presented, containing 15 nouns from the original set and 15 new nouns in random order. The subject has to indicate whether a noun originates from the old set or it is from a new set of nouns. The Psychomotor Vigilance Task (Dinges and Powell, 1985) is based on a simple visual RT test. Subjects are required to respond to a visual stimulus presented at variable interval (2000 to 10000 msec) by pressing either the right or the left button with the dominant hand. The visual stimulus is a counter turning on and incrementing from 0 to 60 sec at 1-msec intervals. In response to the subject’s button press, the counter display stops incrementing, allowing the subject 1 sec to read the RT before the counter restarts. If a response has not been made in 60 sec, the clock resets and the counter restarts. The Critical Tracking Test measures the ability to control an unstable signal in a tracking task. The signal deviates horizontally from a midpoint and the subject has to compensate this signal deviation by moving a joystick in opposite direction. The test includes five trials of which the lowest and the highest score are discarded (Jex et al., 1966). The Divided Attention Task measures the ability to divide attention between two simultaneously performed tasks (Moskowitz, 1973). The first task is to perform the CTT at a constant level of difficulty set at 50% of his or her maximum capacity. In the other task the subject has to monitor 24 single digits that are presented in the four corners of the screen. The digits change asynchronously at 5-second intervals. The subjects are instructed to remove their foot from a pedal as rapidly as possible whenever the digit ‘2’ appears. This signal occurs twice at every location, in random order, at intervals of 5 to 25 sec. Task duration is fixed at 12 minutes. The main performance parameters are average tracking error and speed of target detection in the visual search task. In the Stop Signal Task the concept of inhibitory control is defined as the ability to stop a pending thought or action and to begin another (Logan et al., 1984). The paradigm consists of two concurrent tasks, i.e. a go task (primary task) and a stop task (secondary task). The go signals (primary task stimuli) are two letters (‘X’or ‘O’) presented one at a time in the center of a computer screen. Subjects are required to respond to each letter as quickly as possible by pressing one of two response buttons. Occasionally, a stop signal (secondary task stimulus) occurs during the test. The stop signal consists of an auditory cue, i.e. a 1000 Hz tone, that is presented for 100 ms. The interval at which the stop signal is presented is dependent from the subject’s own successful and unsuccessful inhibitions. By continuously monitoring the subject’s response the stop signal reaction time is calculated during the task. Subjects are required to withhold any response in case a stop signal is presented.

Subjective evaluation Subjective evaluations of driving quality, sedation and mood were assessed using a series of visual analogue scales (100 mm). Subjects rated the degree of effort they had to put in driving performance using the Rating Scale Mental Effort (Zijlstra, 1993). The scale is a visual analogue scale (150 mm) with additional

29 DRUID 6th framework programme Deliverable 1.2.2 verbal labels. In addition, subjects rated the extent of influence of the drug on their driving performance, prior to and upon completion of the Highway Driving Test, using a 100 mm visual analogue scale. The driving instructors rated each subject’s driving quality and apparent sedation at the conclusion of the Highway Driving Test, using two 100 mm visual analogue scales. The subjects were instructed to rate their subjective feelings of alertness and sleepiness before the start of cognitive testing using a 16-item mood scale which provides three factor analytically defined summary scores for ‘alertness’, ‘contentedness’, and ‘calmness’(Bond and Lader, 1974) and the Karolinska Sleepiness Scale with scores ranging from 1 (extremely alert) to 9 (very sleepy, fighting sleep) (Akerstedt and Gillberg, 1990).

Blood samples Blood samples were taken at 9:30 hours after ingestion of zopiclone 7.5 mg to determine serum concentrations of hypnotics before driving. Samples were centrifuged after a clotting period, and serum was frozen at -20°C until analyses for pharmacokinetic assessments.

Procedure Subjects were individually trained to perform the laboratory tests during two sessions of approximately 1.5 hours in a previous study (Leufkens et al., in prep.). In that study, they underwent two nights of sleep evaluation. Subjects were therefore sufficiently familiarized with the testing facilities and procedures. Treatment periods started in the evening of Day 1, when the subjects arrived at the site at approximately 20:00 hours, and lasted until Day 2, when they were transported home after the driving test, at approximately 11:45 hours. On arrival at the sleeping facility in each treatment period, subjects’eligibility was verified. They were questioned about adverse events and use of medication since their last visit. Hereafter, electrodes for polysomnographic recording were attached. Subjects ingested their medication and retired to bed at 23:30 hours. They were awakened at 07:30 hours and served a light standardized breakfast. At 08:00 hours (i.e. 8.5 hours post dose) they filled out the subjective rating scales for sleep, mood, and daytime sleepiness, and started the laboratory tests. At approximately 9:00 hours a blood sample was taken. Subjects were subsequently transported to the start of the highway driving test. Before driving they rated the anticipated effect of the drug on their driving performance, and performed the highway driving test between 09:30 and 10:30 hours (i.e. 10:00 – 11:00 hours post-dose). Upon completion subjects were asked to rate the mental effort it took to perform the driving test and to evaluate the influence of the drug on their driving performance. Next, subjects performed the car following test, after which they returned to the testing facilities for removal of the electrodes.

Statistical analysis

Sample size was based on a power calculation for detecting a clinically relevant effect of 2.4 cm in the primary measure of this study, the SDLP. This change corresponds to the effects of alcohol on SDLP, while blood alcohol concentrations (BACs) are 0.5 g/L as measured in a previous study (Louwerens et al., 1987). Given a test-retest reliability of SDLP of at least r=0.70, a group of 16 subjects should permit detection of a mean change in SDLP of 2.0 cm, with a power of at least 90% and an Į risk of 0.05. Overall effects were analyzed using a mixed model analysis of variance with Group as between

30 DRUID 6th framework programme Deliverable 1.2.2 subject factor with three levels (‘chronic users’, ‘infrequent users’, ‘controls’) and Treatment as within subjects factor with two levels (zopiclone, placebo). Significant (p<0.05) main effects or interactions were further analyzed using three univariate comparisons between groups for each treatment, and paired t-tests between placebo and zopiclone within each group. Secondly, change scores in SDLP from placebo were calculated and used for equivalence testing using an alcohol criterion of 2.4 cm while blood alcohol concentrations are 0.5 g/L. All statistical analyses were done by using the Statistical Package for the Social Sciences (SPSS) statistical program (version 15.0 for Windows; SPSS, Chicago, IL).

Results

Driving performance Out of 96 driving tests, one was terminated before scheduled completion because the driving instructor judged that it would be unsafe to continue. The subject was a female insomnia patient from the infrequent users group who had been administered zopiclone. Her Standard Deviation of Lateral Position (SDLP) score was calculated from the data collected until termination of the ride.

Figure 1 presents mean ± SE SDLP values recorded after placebo and zopiclone 7.5 mg for each group separately.

placebo * * * zopiclone 23

22 +3.6 +1.6 cm cm 21 +2.1 cm 20

19

18

17 Standard Deviation of Lateral Position (SDLP in cm) in (SDLP Position Lateral of Deviation Standard 16

15 chronicMedicated users infrequentUnmedicated users healthyControl controls

Figure 1. Mean (±SE) SDLP for each group separately (* = significant drug effect; p<.05)

Analysis showed a highly significant overall Treatment effect on SDLP (F1,45=33.86, p<0.001). Zopiclone significantly impaired driving in all groups. Compared to placebo the increase in SDLP was +1.6 cm (p=0.010) in the chronic users group, +2.1 cm (p=0.020) in the infrequent users group and +3.6 cm (p<0.001) in the healthy control group. T-tests for independent samples showed that the mean increase in

31 DRUID 6th framework programme Deliverable 1.2.2 SDLP from placebo to zopiclone was significantly lower in the chronic users group than the control group (p=0.045). There was no difference between the infrequent users and controls and between the two insomnia groups. Tests of equivalence showed that zopiclone impaired driving performance in all three groups. The 95% confidence intervals crossed the alcohol criterion of 2.4 cm and none crossed the null-level (figure 2).

4,8

3,6

2,4 Position (SDLP; incm)

1,2 Mean change (95% CI) from placebo in Standard Deviation of Lateral placebo Deviation in- Lateral Standard from of CI) change (95% Mean 0 chronic users infrequent users healthy controls

Figure 2. Mean change (95% CI) from placebo in Standard Deviation of Lateral Position for each group separately

Standard Deviation of Speed (table 1) showed a significant main effect of Treatment (F1,45=12.24, p=0.001) and a Treatment by Group interaction (F2,45=3.42, p=0.041).

Overall, zopiclone significantly impaired subjects’control over speed variability. Paired t-tests showed that zopiclone significantly increased SDSP in the control group (p<0.004) and the chronic users group (p=0.009), but not in the infrequent users group. There were no significant overall Group effects on SDSP. There were no overall effects on any of the Car Following Test parameters.

Subjective evaluations of driving performance Mean ± SE scores of the subjective evaluations of driving performance are presented in table 1. The driving instructors did not judge the subjects’driving quality and appearance of being sedated to be significantly different between zopiclone and placebo in all groups. Overall, subjects’ratings of anticipated and experienced driving quality were lower after zopiclone as compared to placebo (anticipated: F1,44=6.14, p=0.017; experienced: F1,44=5.79, p=0.020). Paired t-tests showed that these differences reached significance within the control group, but not in the patient groups. Healthy controls expected driving quality to be worse after zopiclone administration (p=0.006) and they confirmed this expectation after the driving test (p=0.013). Changes

32 DRUID 6th framework programme Deliverable 1.2.2 Table 1. Mean (±SE) scores of driving tests and subjective evaluations for each condition and group separately

Group Statistics (p-values)

Chronic Infrequent Variable Treatment Controls DrugxGroup Drug Group users users

Highway Driving Test

SDLP (cm) placebo 19.7 (1.0) 18.2 (0.9) 17.8 (0.6) NS <.001 NS

zopiclone 21.3 (1.2)a 20.3 (1.0)a 21.4 (1.0)a

SDSP (km/h) placebo 2.2 (0.1) 2.3 (0.2) 2.1 (0.1) .041 .001 NS

zopiclone 2.5 (0.2)a 2.4 (0.1) 2.5 (0.4)a

Car Following Test

Reaction Time (sec) placebo 4.3 (0.5) 4.7 (0.4) 4.2 (0.6) NS NS NS

zopiclone 4.5 (0.3) 3.8 (0.5) 4.4 (0.6)

Headway placebo 1.2 (0.05) 1.1 (0.03) 1.2 (0.06) NS NS NS

zopiclone 1.2 (0.06) 1.2 (0.07) 1.1 (0.03)

Subjective Evaluations by Driving Instructors

Driving Quality placebo 58.4 (5.8) 58.8 (4.6) 59.2 (5.0) NS NS NS

zopiclone 58.6 (4.7) 61.1 (4.5) 55.8 (3.8)

Apparent Sedation placebo 18.9 (5.4) 15.1 (5.0) 18.4 (6.0) NS NS NS

zopiclone 23.4 (4.3) 21.8 (3.2) 21.8 (4.4)

Subjective Evaluations by Participants

Anticipated Driving Quality placebo 71.7 (5.6) 73.6 (4.8) 86.2 (3.5) NS .017 NS

zopiclone 67.6 (6.2) 71.7 (5.9) 66.4 (5.5)a

Experienced Driving Quality placebo 67.9 (4.9) 65.0 (4.5) 78.3 (4.3) NS .020 NS

zopiclone 60.8 (5.4) 62.4 (6.1) 66.8 (5.8)a

Mental Effort placebo 31.4 (5.3) 35.4 (5.4) 24.8 (4.7) NS <.001 NS

zopiclone 45.1 (8.9) 53.3 (7.9)a 45.1 (6.8)a

Karolinska Sleepiness Scale placebo 5.1 (0.4)b 4.4 (0.5) 3.5 (0.3) NS NS .009

zopiclone 5.1 (0.4)b 4.6 (0.4) 3.7 (0.3)

Alertness placebo 58.8 (4.5)b 64.9 (4.1) 73.9 (3.4) .013 NS NS

zopiclone 69.2 (3.9)a 65.8 (2.8) 66.2 (4.7)a

Contentedness placebo 68.3 (3.8) 72.0 (3.4) 76.5 (4.2) .048 NS NS

zopiclone 75.3 (3.5) 73.7 (3.7) 71.3 (4.5)

Calmness placebo 72.6 (3.9) 75.9 (3.9) 75.2 (3.7) NS NS NS

zopiclone 71.2 (3.5) 74.1 (4.0) 76.0 (3.6) a = significant Drug effect (p<.05); b = significantly different from control group (p<.05); NS = no significant effect

33 DRUID 6th framework programme Deliverable 1.2.2 Table 2. Mean (±SE) scores of cognitive performance tests for each condition and group separately

Group Statistics (p-values)

Chronic Infrequent Variable Treatment Controls DrugxGroup Drug Group users users

Word Learning Test

Immediate Recall Score placebo 39.5 (1.9) 43.6 (2.2) 46.2 (1.9) NS .003 NS

zopiclone 35.4 (2.9) 39.1 (2.7)a 41.6 (2.5)

Delayed Recall placebo 5.3 (0.7) 6.8 (0.9) 7.3 (0.8) NS .024 NS

zopiclone 4.8 (0.8) 5.8 (0.8) 5.4 (0.7)

Recognition Score placebo 25.4 (1.1) 25.0 (1.3) 25.5 (1.3) NS .014 NS

zopiclone 24.4 (0.8) 23.4 (1.4) 23.5 (1.5)a

Recognition Reaction Time (msec) placebo 848 (32) 785 (36) 851 (33) NS .006 NS

zopiclone 889 (39) 863 (40) 893 (34)

Psychomotor Vigilance Task

Average Reaction Time (msec) placebo 290 (13) 290 (15) 281 (8) NS NS NS

zopiclone 288 (12) 303 (16) 297 (11)

Lapses (>500 msec) placebo 3.1 (1.0) 1.7 (0.4) 2.1 (0.6) NS NS NS

zopiclone 2.6 (0.8) 3.1 (1.0) 2.6 (0.6)

Critical Tracking Task

Average Lambda (rad/sec) placebo 3.1 (0.2) 2.9 (0.2) 2.8 (0.2) NS .007 NS

zopiclone 2.9 (0.2) 2.7 (0.2) 2.7 (0.1)

Divided Attention Task

Average Error (mm) placebo 16.5 (1.5) 18.4 (1.1) 19.8 (1.3) NS <.001 NS

zopiclone 18.9 (1.2)a 20.5 (1.3)a 21.4 (1.1)

Reaction Time (msec) placebo 2052 (87) 1971 (75) 1941 (81) NS .034 NS

zopiclone 2038 (65) 2086 (71)a 2140 (93)

Stop Signal Task

Hits placebo 241 (3) 246 (1) 247 (1) NS .016 NS

zopiclone 240 (3) 245 (1) 245 (2)

Go Reaction Time (msec) placebo 437 (19) 432 (19) 438 (13) NS .001 NS

zopiclone 442 (18) 452 (20)a 459 (15)a

Stop Reaction Time (msec) placebo 181 (7) 184 (8) 202 (10) .034 <.001 .016

zopiclone 184 (9)b 197 (8)a 229 (9)a a = significant Drug effect (p<.05); b = significantly different from control group (p<.05); NS = no significant effect

34 DRUID 6th framework programme Deliverable 1.2.2 Table 3. Mean (±SE) scores of subjective and objective sleep quality

Group Statistics (p-values)

Infrequent Variable Treatment Chronic users Controls DrugxGroup Drug Group users

Subjective

Groningen Sleep Quality Scale placebo 10.8 (0.8)b 8.9 (0.9)b 4.2 (1.0) .022 <.001 <.001

zopiclone 5.3 (1.0)a, b 3.3 (0.6)a 2.4 (0.5)

Sleep Onset Time (min) placebo 114 (22)b, c 53 (11) 36 (8) .049 .001 <.001

zopiclone 44 (11)a 27 (4) 24 (4)

Awakenings (#) placebo 3.5 (0.6) 3.8 (0.6)b 2.0 (0.4) NS <.001 .027

zopiclone 1.9 (0.4)a, b 1.4 (0.4)a 0.7 (0.2)a

Total Sleep Time (min) placebo 241 (21)b 291 (23)b 385 (13) .012 <.001 <.001

zopiclone 355 (18)a, b, c 403 (12)a 419 (8)a

Polysomnographic parameters

Sleep Onset Time (min) placebo 42.8 (7.6)c 21.2 (4.3) 25.9 (4.5) NS .031 .008

zopiclone 31.3 (4.8)c 16.6 (2.6) 22.5 (2.6)

Wake After Sleep Onset (min) placebo 94.1 (11.4) 73.5 (9.2) 73.6 (13.2) NS <.001 .029

zopiclone 78.2 (12.6)b, c 48.4 (5.2)a 35.9 (4.5)a

Awakenings (#) placebo 8.7 (1.2) 7.6 (0.8) 7.3 (1.1) NS .004 NS

zopiclone 7.5 (1.0) 5.9 (0.8) 4.4 (0.8)a

Total Sleep Time (min) placebo 343 (15) 386 (11) 381 (13) NS <.001 .002

zopiclone 372 (14)a, b, c 414 (6)a 423 (5)a

Sleep Efficiency (%) placebo 71.5 (3.1) 80.4 (2.4) 79.6 (2.7) NS <.001 .002

zopiclone 77.3 (2.9)a, b, c 86.9 (1.0)a 88.6 (0.7)a

Stage 1 Sleep (% of Total Sleep Time) placebo 7.9 (0.9) 6.7 (0.8) 7.2 (1.1) NS .007 NS

zopiclone 7.5 (1.0) 4.8 (0.6)a 5.1 (0.8)a

Stage 2 Sleep (% of Total Sleep Time) placebo 55.9 (2.1) 51.5 (1.6) 53.2 (2.3) NS .009 NS

zopiclone 59.9 (2.2) 54.1 (2.0) 55.9 (2.2)

Stage SWS Sleep (% of Total Sleep Time) placebo 15.1 (2.3) 20.6 (1.9) 20.7 (2.2) NS .001 NS

zopiclone 18.5 (2.4) 24.4 (2.1)a 21.9 (2.2)

Stage REM Sleep (% of Total Sleep Time) placebo 20.9 (1.2) 21.1 (1.3) 18.9 (1.0) .027 <.001 NS

zopiclone 14.1 (0.8)a 16.8 (1.3)a 17.0 (1.0) a = significant Drug effect (p<.05); b = significantly different from control group (p<.05); c = significantly different from infrequent users group (p<.05); NS = no significant effect

in the patient groups were in the same direction, but smaller. T-tests for independent samples revealed that both insomnia groups rated their driving quality in the placebo condition significantly lower than the control group (chronic users: p=0.040; infrequent users: p=0.045). Overall, subjects’perceived mental effort to perform the driving test was increased after zopiclone as compared to placebo (F1,45=16.15, p<0.001). Paired t-tests showed that this difference reached significance

35 DRUID 6th framework programme Deliverable 1.2.2 within the control group (p=0.009) and the infrequent users group (p=0.008), but not in the chronic users group. Differences between groups were not significant.

Subjective evaluations of sleepiness and feelings Mean ± SE scores of the subjective evaluations of sleepiness and feelings are presented in table 1. Subjects’ratings of sleepiness as measured by the Karolinska Sleepiness Scale were significantly different between groups (F2,44=5.28, p=0.009), but not between treatments. The chronic users group felt more sleepy than the healthy controls after placebo (0.003) and after zopiclone (p=0.023) administration. Subjective feelings of alertness and contentedness as measured by Bond and Lader’s mood scale showed a significant Treatment by Group interaction (F2,45=4.81, p=0.013 and F2,45=3.25, p=0.048, respectively). In the placebo condition, the insomnia groups felt significantly less alert than the control group. This difference was significant for the chronic users group (p=0.011). Use of zopiclone increased next day alertness in the chronic users (p=0.029), whereas it impaired alertness in the healthy controls (p=0.040). For feelings of contentedness, further analyses did not reveal differences between Treatments and Groups. There were no overall main effects for subjective feelings of calmness.

Cognitive and psychomotor assessment Table 2 summarizes the mean ± SE scores of the cognitive performance tests.

Overall, zopiclone impaired all parameters of the Word Learning Test, i.e. immediate recall (F1,45=9.78, p=0.003), delayed recall (F1,45=5.49, p=0.024), recognition score (F1,45=6.48, p=0.014), and recognition reaction time (F1,45=8.50, p=0.006). Paired t-tests revealed that these effects did not reach significance in each group separately, except for the effect on immediate recall in the infrequent users group (p=0.039), and the effects on the recognition score in the control group (p=0.045). T-tests for independent samples revealed that the chronic users scored significantly worse than the healthy controls on the immediate recall score in the placebo condition (p=0.020). Other significant group differences were not found. Performance in the Psychomotor Vigilance Task did not show significant overall differences between Treatments and Groups.

Overall, zopiclone affected psychomotor performance in the Critical Tracking Task (F 1,44=8.15, p=0.007). Analysis for groups separately did not reveal significant Treatment effects, however. In the Divided Attention Task there was an overall significant impairment by zopiclone in both the tracking subtask (F1,44=16.49, p<0.001) and the detection subtask (F1,44=4.81, p=0.034). Tracking, as reflected by the average error, was significantly worse after zopiclone administration in both insomnia groups (chronic users: p=0.024; infrequent users: p=0.008), but not in the control group. Detection, as reflected by reaction time, was significantly impaired following zopiclone in the infrequent users group only (p=0.048).

A significant overall Treatment effect was found on the number of hits (F1,43=6.31, p=0.016) and the go reaction time (F1,4312.63, p<0.001) in the Stop Signal Task. For the number of hits, paired t-tests revealed that this effect did not reach significance in each group separately, however. Paired t-tests did show that the go reaction time was significantly slower after zopiclone administration in the infrequent users (p=0.026) and the controls (p=0.033). Stop reaction time showed a significant main effect of Treatment (F1,43=15.84, p=0.001), a significant overall Group difference (F2,43=4.58, p=0.016) and a Treatment by Group interaction

36 DRUID 6th framework programme Deliverable 1.2.2 (F2,43=3.67, p=0.034). Overall, zopiclone significantly slowed subjects’response to a stop signal. Paired t-tests showed that zopiclone significantly increased stop reaction time in the infrequent users group (p=0.046) and the control group (p=0.007), but not in the chronic users group. Overall analysis for Group differences revealed a significant effect following zopiclone administration

(F2,44=5.27, p=0.009), but not following placebo administration. Post-hoc analysis showed that the chronic users had significantly less problems with responding to a stop signal after zopiclone than the controls had (p=0.002).

Sleep quality Subjective evaluation Table 3 summarizes the mean ± SE scores of both the subjective and objective sleep parameters for each group separately after administration of placebo and zopiclone.

Significant overall main effects of Treatment and Group were found for all parameters (number of complaints: F1,45=49.34, p<0.001; sleep onset time: F1,45=14.05, p=0.001; number of awakenings:

F1,45=33.56, p<0.001; total sleep time: F1,44=52.29, p<0.001). In addition, significant interactions showed that the effect of zopiclone differed between groups in number of complaints (F2,45=4.18, p=0.022), sleep onset time (F2,45=3.24, p=0.049) and total sleep time (F 2,44=4.90, p=0.012). Paired t-tests of Treatment effect for each group separately showed that in both insomnia groups number of sleep complaints was significantly reduced (both groups: p<0.001) and total sleep time was significantly increased (chronic users: p<0.001; infrequent users: p=0.001) after zopiclone administration. Number of awakenings was significantly reduced after zopiclone in all groups (chronic users: p=0.022; infrequent users: p<0.001; controls: p=0.008). Sleep onset time was significantly diminished after zopiclone in the chronic users group only (p=0.014). Significant differences between chronic users and controls were found in number of complaints (p<0.001), sleep onset time (p=0.006) and total sleep time (p<0.001) after placebo. Following zopiclone administration chronic users reported significantly more sleep complaints (p=0.007), shorter total sleep time (p=0.001) and more awakenings (p=0.015) than controls. Comparisons between the infrequent users and the controls revealed that there were significant differences in number of complaints (p=0.001), total sleep time (p=0.001) and number of awakenings (p=0.027) after administration of placebo. Differences between the groups disappeared after zopiclone administration. Comparisons between the insomnia groups showed that the chronic users had a significantly longer sleep onset time (p=0.006) and a significantly shorter total sleep time (p=0.001) than the infrequent users in the placebo condition. There were no differences between the insomnia groups after zopiclone administration. Polysomnographic parameters Significant overall main Treatment effects were found in all polysomnographic parameters for sleep continuity (sleep onset time: F1,42=4.97, p=0.031; wake after sleep onset: F1,41=18.71, p<0.001; number of awakenings: F1,42=9.36, p=0.004; total sleep time: F1,42=23.30, p<0.001; sleep efficiency: F1,41=23.26, p<0.001). Paired t-tests showed that following zopiclone administration all groups slept significantly

37 DRUID 6th framework programme Deliverable 1.2.2 longer (chronic users: p=0.033; infrequent users: p=0.014; controls: p=0.006) and improved their sleep efficiency (chronic users: p=0.034; infrequent users: p=0.012; controls: p=0.009). Furthermore after zopiclone, the infrequent users and controls were significantly less time awake after sleep onset, p=0.016 and p=0.017, respectively. Lastly, the controls had significantly fewer awakenings after zopiclone administration as compared to placebo (p=0.027). Except in the number of awakenings, significant overall Group effects were found in all objective evaluations (sleep onset time: F2,42=5.47, p=0.008; wake after sleep onset: F2,41=3.86, p=0.029; total sleep time: F2,42=7.11, p=0.002; sleep efficiency: F2,41=7.61, p=0.002). Following zopiclone administration and compared to both the controls and infrequent users, the chronic users had a significantly longer wake after sleep onset time (p=0.001 and p=0.033, respectively), significantly less total sleep time (p=0.001 and p=0.004, respectively) and significantly worse sleep efficiency (p<0.001 and p=0.001, respectively). Sleep onset time appeared to be significantly longer for the chronic users as compared to the infrequent users after both placebo (p=0.029) and zopiclone administration (p=0.010). No differences between infrequent users and controls were found on any of the parameters for sleep continuity after placebo or zopiclone administration. Overall main effects of Treatment were also found in all parameters for sleep architecture

(percentage of Stage 1 Sleep: F1,42=8.13, p=0.007; percentage of Stage 2 Sleep: F1,42=7.40, p=0.009; percentage of Slow Wave Sleep: F1,42=11.57, p=0.001; percentage of REM Sleep: F1,42=37.10, p<0.001). In the infrequent users, Stage 1 Sleep and REM Sleep were significantly reduced after zopiclone administration (p=0.028 and p=0.001, respectively) and Slow Wave Sleep was significantly increased (p=0.001). Zopiclone also significantly decreased the percentage of Stage 1 Sleep in the controls (p=0.042) and REM Sleep in the chronic users (p<0.001). Overall main Group differences were not found on any of the parameters. A significant Treatment by Group interaction was shown in the percentage of REM sleep. Yet, further analysis did not reveal any group differences.

Serum concentrations Mean (±SE) serum concentrations for zopiclone were 9.9 (1.0) ng/mL in the chronic users group, 11.3 (1.2) ng/mL in the infrequent users group and 10.7 (0.7) ng/mL in the controls group. Overall Group analysis revealed no significant differences between the three groups. Correlations between change in SDLP from placebo to zopiclone and serum concentrations were 0.06 (n.s.) for the chronic users, 0.48 (n.s.) for the infrequent users and 0.54 (p<0.05) for the controls.

Discussion

Results of the present study show that a single oral dose of zopiclone 7.5 mg significantly impairs driving performance in insomnia patients who chronically use hypnotics, in insomnia patients who not or infrequently use hypnotics and in healthy, good sleepers at 10 to 11 hours after bedtime administration. The impairing effect of zopiclone on driving, as reflected by the rise in SDLP compared with placebo, was significantly

38 DRUID 6th framework programme Deliverable 1.2.2 different between the chronic users and the healthy controls. As a result, interpretation of the severity of the effects differs between groups. The effect found in the chronic users group (+1.6 cm) is on average of lesser magnitude than that produced by alcohol in a previous study while subjects drove with blood alcohol concentrations (BAC) of 0.5 mg/mL (+2.4 cm) (Louwerens et al., 1987), which is the legal limit for driving a car in most countries. In contrast, the increase of 3.6 cm in SDLP from placebo after zopiclone 7.5 mg administration in the healthy control group is above this effect of alcohol. Zopiclone produced an effect of +2.1 cm in the infrequent users group, which was slightly less than that of alcohol while BACs are 0.5 mg/mL. However, there was no statistical difference in effect of zopiclone between the infrequent users and the controls. The significantly decreased magnitude of effect of zopiclone 7.5 mg on driving performance in the chronic users as compared to the healthy controls suggests that residual effects are attenuated by chronic use of hypnotics. This implicates that results from studies conducted with healthy volunteers appear to give an overestimation of the actual effects in insomnia patients chronically using hypnotics. It should be mentioned, however, that SDLP scores following zopiclone administration were still significantly increased in the chronic users group, indicating that residual effects do not completely disappear. In addition, the effect of zopiclone in the chronic users group may even be slightly larger than found in the present study. Inspection of the SDLP scores after placebo administration showed that, although not reaching statistical significance, scores in the chronic users group were elevated compared with the infrequent users and healthy controls. The results suggest that withdrawal symptoms were still present in the chronic users despite discontinuation of own hypnotic intake three days before each treatment period. The patients may therefore have experienced discomfort from the hypnotic free night which may have affected their driving performance. As a consequence, the elevated SDLP scores in the placebo condition may have reduced the effect of zopiclone on driving performance in this group. Yet, according to polysomnographic analyses, sleep in the chronic users appeared not to be significantly affected by possible withdrawal effects during the placebo night. Besides a significant difference of 20 minutes in sleep onset as compared to the infrequent users, the sleep profile of the chronic users was comparable with that of the other two groups. The only indication of reduced ability to perform in the chronic users may be the decreased feelings of alertness and increased feelings of sleepiness as compared to the controls following placebo administration. Even more, the chronic users reported feeling significantly more alert after the zopiclone night, whereas the controls felt the exact opposite. This may suggest that the chronic users experienced more discomfort after the placebo night than after the zopiclone night and may have influenced their driving performance. After all, this was their fourth hypnotic-free night in succession possibly causing physical problems (Pétursson, 1994). The magnitude of residual effects of zopiclone 7.5 mg found in the infrequent users group is in line with previous studies conducted in healthy younger drivers (Vermeeren et al., 1998; Vermeeren et al., 2002; Leufkens et al., 2009). In those studies, the mean increase in SDLP from placebo after zopiclone administration ranged from +2.5 cm to +4.9 cm. This suggests that the residual effects of hypnotics found in healthy volunteers can validly predict the effects in older patients suffering from insomnia. The impairing effects on driving after zopiclone appeared not to be noticed by the insomnia patients. Prior to the start of the driving test, they did not expect to drive differently after administration of zopiclone than after placebo. In addition, after completion of the test they did not rate their driving quality differently

39 DRUID 6th framework programme Deliverable 1.2.2 between zopiclone and placebo. In contrast, the healthy controls anticipated their driving quality to be significantly worse after zopiclone. Their expectations were confirmed by their performance as they evaluated their driving quality after the test significantly lower after zopiclone administration as compared with placebo. Whereas the residual adverse effects of zopiclone remained undetected by the insomnia patients according to their subjective evaluations, opposite results were found for the evaluations of its therapeutic effects. Both insomnia groups reported significantly improved subjective sleep quality on most parameters after zopiclone administration as compared with placebo. The healthy controls, however, appeared to benefit considerably less from the sleep inducing properties of zopiclone. According the objective measures of sleep, there were virtually no differences in sleep quality between the groups, however. Still, the chronic users group felt significantly more alert the morning after an evening dose of zopiclone than after placebo. The infrequent users did not report a difference in alertness, whereas the healthy controls reported feeling significantly less alert after zopiclone than after placebo. The lack of awareness of residual sedative effects of zopiclone 7.5 mg may cause insomnia patients to belief that car driving is safe the morning after evening administration. Even more, these beliefs may be strengthened by the experienced improvement of subjective sleep quality. These results stress, however, the importance of general physicians to warn their patients about the impairing effects of zopiclone 7.5 mg on driving performance. The impairing effects of zopiclone on driving performance could not be completely corroborated by the results of the cognitive performance tests. Although there were overall significant differences between placebo and zopiclone on a majority of the parameters, in about only half of the cases impairing effects of zopiclone were found in one or two specific groups. For instance, zopiclone only impaired immediate recall in the infrequent users group and recognition performance in the healthy controls. Evident impairing residual effects of zopiclone 7.5 mg on verbal learning have been found recently, however, in both healthy older and younger subjects (Leufkens and Vermeeren, 2009l; Leufkens et al., 2009). Average scores of the three groups in the present study following placebo administration appeared to be slightly lower already than the average scores of healthy older subjects in the previous study (Leufkens and Vermeeren, 2009). This may suggest that scores were close to a minimum, showing possible floor effects. Performance on the psychomotor vigilance task was not different between placebo and zopiclone administration. Although there is ample evidence that the PVT is highly sensitive to the effects of sleep deprivation (Lim and Dinges, 2008), there seem to be almost no studies assessing effects of sedating drugs on this test. To our knowledge, the sedative residual effects of zopiclone 7.5 mg have not yet been investigated with use of this task. There is a recent study, however, showing significant effects of blood alcohol concentrations of 0.03 mg/mL on PVT performance in 18 healthy men (Howard et al., 2007). This suggests that the task should have been sufficiently sensitive to detect the residual effects zopiclone, which were comparable in magnitude to those of blood alcohol concentrations of 0.05 mg/mL as measured by SDLP. The failure to find an effect on the PVT therefore indicates that performance on this test is less sensitive to residual effects of GABAergic hypnotic drugs than acute effects of alcohol. To summarize, results of the present study indicate that driving performance is moderately impaired in insomnia patients after evening administration of zopiclone 7.5 mg at least until 11 hours after intake. Chronic use of hypnotics seems to attenuate the severity of effects of zopiclone 7.5 mg. Nevertheless, this

40 DRUID 6th framework programme Deliverable 1.2.2 reduction does not result in an absence of impairing effects in insomnia patients chronically using hypnotics. The magnitude of effects found in the infrequent users group was in line with previous studies investigating residual effects of zopiclone 7.5 mg in healthy, younger volunteers. This suggests that these studies are able to validly predict the residual effects of hypnotics in insomnia patients who do not use hypnotics.

Acknowledgements

This study is financially supported by the integrated project Driving Under the Influence of Drug, Alcohol and Medicine (DRUID), which is part of the European Union’s 6th Framework Programme. The content reflects only the authors’ view. The European Commission is not liable for any use that may be made of the information contained therein. The authors would like to express their gratitude to Gwenda Engels, Nicky van Gennip, Jolien Gooijers, Liene Ketelslegers, Jasmijn Kromhout, Loes van Langen, Anita van Oers, Elmy Theuniszen, Natalie Valle Guzman, Floor van de Water and Tim Weysen for the assistance in data collection; Renilde van den Bossche for the polysomnographic analyses; Cees van Leeuwen for the medical supervision; Henk Brauers, Willy Jeurissen, Jo Gorissen and Hans Sleebe for ensuring the safety of the subjects during driving; and Irma Brauers for the logistic work.

References

Akerstedt T, Gillberg M (1990) Subjective and objective sleepiness in the active individual. Int J Neurosci 52: 29-37. Association A P (1994) DSM-IV: Diagnostic and Statistical Manual of Mental Disorders. American Psychiatric Press, Washington DC. Barbone F, McMahon A D, Davey P G, Morris A D, Reid I C, McDevitt D G, MacDonald T M (1998) Association of road-traffic accidents with benzodiazepine use. Lancet 352: 1331-1336. Beck A T, Ward C H, Mendelson M, Mock J, Erbaugh J (1961) An inventory for measuring depression. Archives of General Psychiatry 4: 561-571. Bond A, Lader M (1974) The use of analogue scales in rating subjective feelings. Br J Med Psychol 47: 211- 218. Brookhuis K A, De Waard D (1994) Measuring driving performance by car-following in traffic. Ergonomics 37: 427-434. Buysse D J, Reynolds C F, 3rd, Monk T H, Berman S R, Kupfer D J (1989) The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Research 28: 193-213. Curran H V, Collins R, Fletcher S, Kee S C, Woods B, Iliffe S (2003) Older adults and withdrawal from benzodiazepine hypnotics in general practice: effects on cognitive function, sleep, mood and quality of life. Psychological Medicine 33: 1223-1237. Derogatis L R (1983) The SCL-90-R: Administration, Scoring and Procedures Manual II. Baltimore Clinical Psychometric Research, Baltimore. Dinges D F, Powell J W (1985) Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods Instrum Comp 17: 652-655.

41 DRUID 6th framework programme Deliverable 1.2.2 Fulda S, Schulz H (2001) Cognitive dysfunction in sleep disorders. Sleep Medicine Review 5: 423-445. Glass J, Lanctot K L, Herrmann N, Sproule B A, Busto U E (2005) Sedative hypnotics in older people with insomnia: meta-analysis of risks and benefits. Br Med J 331: 1169-1175. Hemmelgarn B, Suissa S, Huang A, Boivin J F, Pinard G (1997) Benzodiazepine use and the risk of motor vehicle crash in the elderly. JAMA 278: 27-31. Howard M E, Jackson M L, Kennedy G A, Swann P, Barnes M, Pierce R J (2007) The interactive effects of extended wakefulness and low-dose alcohol on simulated driving and vigilance. Sleep 30: 1334- 1340. Iber C, Ancoli-Israel S, Chesson A, Quan S F (2007) The AASM manual for the scoring of sleep ans associated events: rules, terminology and technical specifications. Westchester, Ill, American Academy of Sleep Medicine. Jex H R, McDonnell J D, Phatak A V (1966) A "critical" tracking task for man-machine research related to the operator's effective delay time. I. Theory and experiments with a first-order divergent controlled element. NASA CR-616. NASA Contract Rep NASA CR. Leufkens T R M, Lund J S, Vermeeren A (2009) Highway driving performance and cognitive functioning the morning after bedtime and middle-of-the-night use of gaboxadol, zopiclone and zolpidem. Journal of Sleep Research 18: 387-396. Leufkens T R M, Ramaekers J G, De Weerd A W, Riedel W J, Vermeeren A (in prep.) On-the-road driving performance and driving related skills in untreated insomnia patients and chronic users of hypnotics. Leufkens T R M, Vermeeren A (2009) Highway driving in elderly the morning after bedtime use of hypnotics: a comparison between temazepam 20 mg, zopiclone 7.5 mg and placebo. J Clin Psychopharmacol 29: 432-438. Lim J, Dinges D F (2008) Sleep deprivation and vigilant attention. Ann N Y Acad Sci 1129: 305-322. Logan G D, Cowan W B, Davis K A (1984) On the ability to inhibit simple and choice reaction time responses: a model and a method. J Exp Psychol Hum Percept Perform 10: 276-291. Louwerens J W, Gloerich A B M, Vries d, G., Brookhuis K A, O'Hanlon J F (1987) The relationship between drivers' blood alcohol concentration (BAC) and actual driving performance during high speed travel. In Noordzij, P C & Roszbach, R (eds), International Congres on Alcohol, Drugs and Traffic Safety, T86. Exerpta Medica, Amsterdam. Metsemakers J F M, Höppener P, Knottnerus J A, Kocken R J J, Limonard C B G (1992) Computerized health information in the Netherlands: a registration network of family practices. British Journal of General Practice 42: 102-106. Moskowitz H (1973) Proceedings: Psychological tests and drugs. Pharmakopsychiatrie, Neuro Psychopharmakologie 6: 114-126. Mulder-Hajonides van der Meulen W R E H (1981) Measurement of subjective sleep quality. Elsevier, Amsterdam. Neutel I (1998) Benzodiazepine-related traffic accidents in young and elderly drivers. Human Psychopharmacology 13: S115-S123. O'Hanlon J F (1984) Driving performance under the influence of drugs: rationale for, and application of, a new test. Br J Clin Pharmacol 18: 121s-129s. Pétursson H (1994) The benzodiazepine withdrawal syndrome. Addiction 89: 1455-1459.

42 DRUID 6th framework programme Deliverable 1.2.2 Ramaekers J G, O'Hanlon J F (1994) Acrivastine, terfenadine and diphenhydramine effects on driving performance as a function of dose and time after dosing. Eur J Clin Pharmacol 47: 261-266. Rey A (1964) L'examen clinique en psychologie. Presses Universitaires de France, Paris. Smets E M, Garssen B, Bonke B, De Haes J C (1995) The Multidimensional Fatigue Inventory (MFI) psychometric qualities of an instrument to assess fatigue. Journal of Psychosomatic Research 39: 315-325. Spielberger C D, Gorsuch R L, Lushene R E, Vagg P R, Jacobs A G (1983) Manual for the State-Trait Anxiety Inventory (Form Y). Consulting Psychologists Press, Inc., Palo Alto. Van Diest R, Milius H, Markusse R, Snel J (1989) The Sleep-Wake Experience List. Tijdschrift voor Sociale Gezondheidszorg 67: 343-347. Vermeeren A (2004) Residual effects of hypnotics: epidemiology and clinical implications. CNS Drugs 18: 297-328. Vermeeren A, Danjou P E, O'Hanlon J F (1998) Residual effects of evening and middle-of-the-night administration of zaleplon 10 and 20 mg on memory and actual driving performance. Human Psychopharmacology 13: S98-S107. Vermeeren A, O'Hanlon J F, Declerck A C, Kho L (1995) Acute effects of zolpidem and flunitrazepam on sleep, memory and driving performance, compared to those of partial sleep deprivation and placebo. Acta Ther 21: 47-64. Vermeeren A, Riedel W J, van Boxtel M P, Darwish M, Paty I, Patat A (2002) Differential residual effects of zaleplon and zopiclone on actual driving: a comparison with a low dose of alcohol. Sleep 25: 224- 231. Verster J C, Volkerts E R, Schreuder A H, Eijken E J, van Heuckelum J H, Veldhuijzen D S, Verbaten M N, Paty I, Darwish M, Danjou P, Patat A (2002) Residual effects of middle-of-the-night administration of zaleplon and zolpidem on driving ability, memory functions, and psychomotor performance. J Clin Psychopharmacol 22: 576-583. Zijlstra F R H (1993) Efficiency in work behavior. A design approach for modern tools. Delft, Delft University of Technology.

43 DRUID 6th framework programme Deliverable 1.2.2 Chapter 2: Driving performance of chronic users of hypnotics and unmedicated insomnia patients

T.R.M. Leufkens1, J.G. Ramaekers1, A.W. de Weerd2, W.J. Riedel1 & A. Vermeeren1,*

1Department of Neuropsychology and Psychopharmacology, Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, The Netherlands 2Department of Clinical Neurophysiology and Sleep Centre SEIN, Zwolle, The Netherlands

Date: 07.07.2011

*Corresponding author: Dr. A. Vermeeren Department of Neuropsychology & Psychopharmacology Faculty of Psychology and Neuroscience Maastricht University PO Box 616 6200 MD Maastricht, The Netherlands Tel: +31 43 388 1952 Fax: +31 43 388 4560 E-mail: [email protected]

44 DRUID 6th framework programme Deliverable 1.2.2 Abstract

Objectives: Impairing effects of hypnotics on next day driving performance may be compensated by improved sleep in insomnia. In addition, chronic use of hypnotics may lead to tolerance and reduce these residual effects. The present study aims to compare driving performance of patients complaining of insomnia and chronic users of hypnotics with that of healthy age-matched controls. Design: A 3x2 parallel group design with three groups and two within subject conditions Setting: University research institute. Participants: Twenty-two elderly insomnia patients chronically using hypnotics, 20 unmedicated elderly insomniacs and 21 healthy, age-matched controls Interventions: Insomniacs chronically using hypnotics ingested their own prescribed hypnotics. Unmedicated insomniacs and healthy controls did not ingest hypnotics. Measurements and results: A standardized highway driving test was performed between 10 and 11 hours after bedtime. In the evening and morning before the driving test, cognitive and psychomotor performance was assessed using laboratory tests. Results showed that driving performance and driving related psychomotor performance did not differ between unmedicated insomniacs, chronic users of hypnotics and normal sleepers. In chronic users, working memory, as measured by Digit Span backward, was significantly worse compared to controls. No group differences were found in verbal memory, divided attention, psychomotor vigilance and inhibitory control. Conclusions: Insomnia patients appear to be able to successfully perform a one hour driving task that requires prolonged attentional demands. Chronic use of hypnotics does not seem to change driving performance.

Keywords: on-the-road driving, insomnia, hypnotics, cognition

45 DRUID 6th framework programme Deliverable 1.2.2 Introduction

Approximately one-third of the general adult population suffers from insomnia, reporting difficulties in initiating sleep or in maintaining sleep, and feelings of nonrestorative sleep.1, 2 Although non-pharmacological strategies, such as cognitive behavioral therapy, are increasingly being implemented in the treatment of insomnia, pharmacotherapy is still the most frequently used treatment for insomnia.3 The primary choice of sleep-enhancing medication is sedative hypnotics, such as benzodiazepines and the newer benzodiazepine receptor agonists (e.g. zopiclone, zolpidem and zaleplon).4 Ideally, a hypnotic should improve sleep and be free from residual sedative effects after arising. It is known, however, that a number of hypnotics that are currently prescribed can produce next-day residual sedation, depending on type of hypnotic, dose, time after administration and frequency of dosing.5 This may lead to the impairment of a wide range of cognitive abilities and can have serious consequences for daily activities, such as driving a car. In order to select the safest alternative among the available hypnotics, patients and prescribing physicians should be informed about the possible impairing effects of hypnotics. To date, information of the residual effects on driving performance is mainly derived from experimental studies conducted with young, healthy, hypnotic naïve volunteers after a single night of treatment. Investigating the residual effects in this population leaves two important issues unanswered, however. First, responses to the adverse effects of hypnotics may be different in patients than in healthy volunteers because of the underlying disorder. Untreated insomniacs report reduced performance in daily life routines as a consequence of their disturbed sleep.6, 7 When treated pharmacologically, it can be expected that the sleep improving effects of a hypnotic improve daytime performance in insomniacs. The net effects of hypnotics on daytime performance may therefore be less in patients than in healthy volunteers. However, despite complaints of impaired daytime functioning among insomnia patients, there is still little objective evidence supporting this. Reviews of the literature show that most experimental studies examining cognitive and psychomotor performance have not been able to objectively demonstrate impaired daytime functioning in untreated insomnia patients.8, 9 Only minor impairment was found in tasks measuring attention span and vigilance.9 The absence of objective impairment could be explained by the use of short and relatively simple tests, in which patients are able to maintain adequate performance by temporarily increasing their efforts.7, 10, 11 Longer tasks and especially tasks that are closer to real-life activities could be more sensitive to the effects of insomnia.10 The second issue that has not yet been clarified in experimental designs using single doses in healthy young volunteers is whether residual effects of hypnotics on driving are still present in insomniacs who chronically use hypnotics. Although it is recommended not to use hypnotics for periods longer than four weeks, a majority of insomnia patients are treated for prolonged periods, which may result in tolerance and reduced effects on performance.12, 13 There is only one published study that compared performance of chronic users of hypnotics to that of untreated insomnia patients and self defined good sleepers.10 Results of a neuropsychological test battery showed no significant differences between these groups, thus suggesting that patients may have become tolerant to the impairing effects of hypnotics on cognition. Nonetheless, epidemiological studies have shown that chronic use of hypnotics is still related to an increased risk of becoming involved in traffic and occupational accidents.14-18 In summary, it is unclear whether insomnia has significant impairing consequences on daily life

46 DRUID 6th framework programme Deliverable 1.2.2 routines, such as driving a car, which can be attenuated by the use of hypnotics. Furthermore, it remains to be investigated whether the residual effects on driving performance are absent in insomniacs chronically using hypnotics. The present study aimed to determine whether driving performance of chronic users of hypnotics and unmedicated insomnia patients differs from normal sleepers of the same age. Therefore, we compared sleep and performance of two groups of insomnia patients to that of a group of age-matched normal sleepers. One group of patients used hypnotics occasionally and was unmedicated before testing. The other group consisted of chronic users of hypnotics who used their own prescribed hypnotics the night before testing. Driving and cognitive performance were assessed by two on-the-road tests (i.e. a standardized highway driving test and a car-following test) and battery of computerized tests. Sleep was evaluated by polysomnography and subjective rating scales.

Methods

Subjects A total of 42 insomnia patients and 21 healthy controls, in the age range of 52 to 73 years, were recruited through a network of local general practitioners in the region of Maastricht, The Netherlands (Regionaal Netwerk Huisartsen, RNH)19 and by advertisement in local newspapers. Insomnia patients had to meet the inclusion criteria for primary insomnia according to DSM-IV20: (i) subjective complaints of insomnia, defined as difficulties initiating sleep (sleep latency >30 min) and/or maintaining sleep (awakenings >30 min); (ii) duration of more than 1 month; (iii) the sleep disturbance causes clinically significant distress or impairment; (iv) insomnia does not occur exclusively during the course of a mental disorder and (v) insomnia is not due to another medical or sleep disorder or effects of medication or drug abuse. Healthy controls were self-defined good sleepers. Sleep complaints of patients and healthy controls were measured as a part of the screening procedure using Dutch versions of the Pittsburgh Sleep Quality Index (PSQI),21 the Sleep Wake Experience List (SWEL)22 and the general version of the Groningen Sleep Quality Scale (GSQS-gen)23. The GSQS provides a score between 0 and 14 representing a number of sleep complaints. Additionally, a daily journal and the specific version of the Groningen Sleep Quality Scale (GSQS-spec)23 were completed upon arising each morning for two weeks providing subjective estimates of sleep quality. Insomnia patients were assigned to one of two groups depending on the frequency and duration of their use of hypnotic drugs (benzodiazepine, zopiclone or zolpidem). Patients were assigned to a ‘chronic users’group when they used a hypnotic for at least four nights per week and longer than three months (n=22). Patients not using hypnotics or using hypnotics less than three days per week were assigned to the ‘unmedicated insomniacs’group (n=20) as they did not ingest hypnotics during the test periods. The hypnotics used in the chronic users group were temazepam (n=4), zopiclone (n=4), midazolam (n=4), oxazepam (n=3), zolpidem (n=2), lormetazepam (n=2), clonazepam (n=1), flurazepam (n=1) and nitrazepam (n=1). Average (±SD) duration of hypnotic use was 7.7 (±6.8) years, with a mean (±SD) frequency of use of 6.4 (±1.2) nights a week (table 1).

{Insert Table 1 about here}

47 DRUID 6th framework programme Deliverable 1.2.2 In the unmedicated insomniacs group, 7 patients reported no history of hypnotic use. The remaining 13 patients used hypnotics on average (±SD) 1.0 (±0.7) night a week, since 7.8 (±7.9) years. The hypnotics used were temazepam (n=6), zopiclone (n=4), loprazolam (n=1), nitrazepam (n=1) and lorazepam (n=1). All participants had to meet the following inclusion criteria: possession of a valid driving license for at least three years; average driving experience of at least 3000 km per year over the last three years; mentally and physically fit to drive; good health based on a pre-study physical examination, medical history, vital signs, electrocardiogram, blood biochemistry, haematology, serology and urinalysis; body mass index (BMI) between 19 and 30 kg/m2. Exclusion criteria were history of drug or alcohol abuse; presence of a significant medical, neurological, psychiatric disorder, or sleep disorder other than insomnia; chronic use of medication that affects driving performance, except hypnotics; drinking more than 6 cups of coffee per day; drinking more than 21 alcohol containing beverages per week; smoking more than 10 cigarettes per day. Participants were screened for major psychopathology by use of the Symptom Checklist 90 Revised (SCL-90-R),24 the Beck Depression Inventory (BDI),25 the State-Trait Anxiety Inventory (STAI),26 and the Multidimensional Fatigue Inventory (MFI).27 During participation use of caffeine was prohibited from 8 hours prior to arrival on test days, until discharge the next morning. Alcohol intake was not allowed from 24 hours prior to each dosing until discharge. Smoking was prohibited from 1 hour prior to bedtime until discharge. The study was conducted in accordance with the code of ethics on human experimentation established by the World Medical Association’s Declaration of Helsinki (1964) and amended in Edinburgh (2000). The protocol was approved by the medical ethics committee of Maastricht University and University Hospital of Maastricht. Subjects were explained the aims, methods, and potential hazards of the study and they signed a written informed consent prior to any study-related assessments.

Assessments Sleep On nights before testing sleep quality and duration was measured by polysomnography using montage including electroencephalogram, electrooculogram and electromyogram. Sleep stages were visually assessed by qualified technicians according to standardized criteria.28 Parameters derived after analysis are sleep onset latency (in min); wake after sleep onset (in min); total sleep time (in min); sleep efficiency (in %); and number of awakenings. Upon arising subjects completed the specific version of the Groningen Sleep Quality Scale (GSQS- spec).23 In addition, subjects estimated sleep onset latency (in min), total sleep time (in min), time awake before rising (in min), and number of awakenings.

Driving performance Driving performance was assessed using two standardized driving tests developed to measure different aspects of driving performance. The primary test is the Highway Driving Test29 which measures road tracking performance. Performance in this test is mainly determined by the delay lag between sensory processing, execution of motor reaction and the vehicle’s dynamic response. In this test, subjects operate a specially

48 DRUID 6th framework programme Deliverable 1.2.2 instrumented vehicle over a 100 km (61 mi) primary highway circuit, accompanied by a licensed driving instructor having access to dual controls. The subjects’task is to maintain a constant speed of 95 km/h (58 mi/h) and a steady lateral position between the delineated boundaries of the slower traffic lane. The vehicle speed and lateral position are continuously recorded. These signals are edited off line to remove data recorded during overtaking maneuvers or disturbances caused by roadway or traffic situations. The remaining data are then used to calculate means and standard deviations of lateral position and speed. Standard deviation of lateral position (SDLP in centimeters) is the primary outcome variable. SDLP is a measure of road tracking error or ‘weaving’. The test duration is approximately 1 hour. The Car-Following Test30, 31 measures changes in controlled information processing such as selective attention, stimulus interpretation and decision making, and speed of an adaptive motor response to events which are common in driving. In the test two vehicles travel in tandem over a 2-lane, undivided, secondary highway at 70 km/h (44 mi/h). An investigator drives the leading car and the subject, in the second car, is instructed to follow at a distance between 25 and 35 meter. Subjects are further instructed to constantly attend the leading car since it may slow down or speed up at unpredictable times. They are required to follow the leading car’s speed movements, i.e. maintain the initial headway by matching the velocity of the car to the other’s. During the test, the speed of the leading car is automatically controlled by a modified ‘cruise control’system. At the beginning it is set to maintain a constant speed of 70 km/h and, by activating a microprocessor the investigator can start sinusoidal speed changes reaching amplitude of -10 km/h and returning to the starting level within 50 sec. The maneuver is repeated 6 times. The leading car’s speed and signals indicating the beginning of the maneuver are transmitted via telemetry to be recorded in the following vehicle together with the following vehicle’s speed. Phase-delay converted to a measure of the subject’s average reaction time to the movement of the leading vehicle (RT, in s) is taken as the primary dependent variable in this test. Headway is continuously recorded by means of an optical distance sensor and serves as a control variable. Test duration is approximately 25 minutes.

Cognitive and psychomotor performance Cognitive and psychomotor performance was assessed by tests for word learning, digit span, tracking, divided attention, sustained attention and inhibitory control. Tests were selected based on their sensitivity to residual sedating effects of hypnotics or sleep disturbances, and their relation to driving performance. 5, 9, 32-37 The Word Learning Test38 is a verbal memory test for the assessment of immediate recall, delayed recall and recognition performance. Fifteen monosyllabic nouns are presented and at the end of the sequence the subject is asked to recall as many words as possible. This procedure is repeated five times and after a delay of at least 30 minutes the subject is again required to recall as many words as possible. At this trial the nouns are not presented. Finally, a sequence of 30 monosyllabic nouns is presented, containing 15 nouns from the original set and 15 new nouns in random order. The subject has to indicate whether a noun originates from the old set or from the new set of nouns. The Digit Span Forward and Backward is a subtest of the Wechsler Adult Intelligence Scale-Revised (WAIS-R).39 In this test, subjects are asked to repeat orally presented digits with increasing sequence length, either in forward or reverse order. There are two trials at each series length, and the test continues until both trials of a series length are failed. One point is awarded for each correct trial. The Critical Tracking Test40 measures the ability to control an unstable signal in a tracking task. The

49 DRUID 6th framework programme Deliverable 1.2.2 signal deviates horizontally from a midpoint and the subject has to compensate this signal deviation by moving a joystick in opposite direction. The test includes five trials of which the lowest and the highest score are discarded. The Divided Attention Task41 measures the ability to divide attention between two simultaneously performed tasks. The first task is to perform the CTT at a constant level of difficulty set at 50% of his or her maximum capacity. In the other task the subject has to monitor 24 single digits that are presented in the four corners of the screen. The digits change asynchronously at 5-second intervals. The subjects are instructed to remove their foot from a pedal as rapidly as possible whenever the digit ‘2’appears. This signal occurs twice at every location, in random order, at intervals of 5 to 25 sec. Task duration is fixed at 12 minutes. The main performance parameters are average tracking error and speed of target detection in the visual search task. In the Stop Signal Task42 the concept of inhibitory control is defined as the ability to stop a pending thought or action and to begin another. The paradigm consists of two concurrent tasks, i.e. a go task (primary task) and a stop task (secondary task). The go signals (primary task stimuli) are two letters (‘X’or ‘O’) presented one at a time in the center of a computer screen. Subjects are required to respond to each letter as quickly as possible by pressing on of two response buttons. Occasionally, a stop signal (secondary task stimulus) occurs during the test. The stop signal consists of an auditory cue, i.e. a 1000 Hz tone, that is presented for 100 ms. The interval at which the stop signal is presented is dependent from the subject’s own successful and unsuccessful inhibitions. By continuously monitoring the subject’s response the stop signal reaction time is calculated during the task. Subjects are required to withhold any response in case a stop signal is presented. The Psychomotor Vigilance Task43 is based on a simple visual RT test. Subjects are required to respond to a visual stimulus presented at variable interval (2000 to 10000 msec) by pressing either the right or the left button with the dominant hand. The visual stimulus is a counter turning on and incrementing from 0 to 60 sec at 1-msec intervals. In response to the subject’s button press, the counter display stops incrementing, allowing the subject 1 sec to read the RT before the counter restarts. If a response has not been made in 60 sec, the clock resets and the counter restarts.

Subjective evaluation Subjective evaluations of mood, sedation and driving quality were assessed using a series of visual analogue scales (100 mm). The subjects were instructed to rate their subjective feelings of alertness and sleepiness before the start of cognitive testing using a 16-item mood scale44 which provides three factor analytically defined summary scores for ‘alertness’, ‘contentedness’, and ‘calmness’ and the Karolinska Sleepiness Scale with scores ranging from 1 (extremely alert) to 9 (very sleepy, fighting sleep).45 At the conclusion of the Highway Driving test subjects rated the degree of mental effort they had to put in driving performance with the Rating Scale Mental Effort.46 This is a visual analogue scale (150 mm) with additional verbal labels. Finally, the driving instructors rated each subject’s driving quality and apparent sedation, using two 100 mm visual analogue scales.

Blood samples Blood samples were taken from chronic users at 9:30 hours after ingestion of the drug to determine serum

50 DRUID 6th framework programme Deliverable 1.2.2 concentrations of hypnotics before driving. Samples were centrifuged after a clotting period, and serum was frozen at -20°C until analyses for pharmacokinetic assessments.

Procedure Subjects were individually trained to perform the laboratory tests during two sessions of approximately 1.5 hours within 10 days before their first night. After training the subjects underwent two nights of sleep evaluation. The first night was a habituation and practice condition to familiarize subjects with the sleeping facilities and polysomnographic and test procedures. The second night was considered as the actual test condition. A test condition started in the evening of Day 1, when the subjects arrived at the site at approximately 19:00 hours, and lasted until Day 2, when they were discharged at approximately 11:45 hours. On arrival at the sleeping facility, subjects rated their subjective feelings and subjective sleepiness. From 19:30 hours until 20:30 hours they performed the first session of laboratory tests, comprising the Word Learning Test immediate and first delayed recall, the Critical Tracking Task, the Divided Attention Task, the Psychomotor Vigilance Task, the Stop Signal Task, and the Digit Span forward and backward. Hereafter, electrodes for polysomnographic recording were attached. Subjects retired to bed at 23:30 hours. Immediately preceding retiring, subjects in the chronic users group ingested their own prescribed hypnotic, whereas subjects in the unmedicated insomniacs group and controls did not ingest medication. Subjects were awakened at 07:30 hours and after arising a light standardized breakfast was served. At 08:00 hours subjects evaluated sleep quality and duration, and feelings of daytime sleepiness and alertness. Subsequently, they started the second session of laboratory tests, comprising the Word Learning Test second delayed recall and recognition, the Critical Tracking Task, the Divided Attention Task, the Psychomotor Vigilance Task, and the Digit Span forward and backward. At 9:00 hours blood samples were taken from the chronic users to determine serum concentrations of hypnotics before driving. Thereafter subjects were transported to the Highway Driving Test which they performed between 09:30 and 10:30 hours. Upon completion subjects rated the mental effort it took to perform this driving test, and subsequently they conducted the Car-Following Test. Upon completion of this test subjects returned to the testing facilities for removal of the electrodes and were discharged.

Statistical analysis The primary parameter of the study was the Standard Deviation of Lateral Position (SDLP, in cm). Driving and sleep related parameters were compared between the three groups, using a one-way analysis of variance with Group as between subject factor with three levels (‘chronic users’, ‘unmedicated insomniacs’, ‘controls’). Parameters which were assessed in evening and morning session were analyzed using 2x3 Repeated Measures analysis with Time of Day as within subject factor with two levels (evening, morning) and Group as between subject factor. Significant (p<0.05) main effects or interactions, were further analyzed using three univariate comparisons between groups for each time of day separately and paired t-tests between sessions for each group separately. If the model assumptions were violated, a suitable transformation or nonparametric method was chosen for analysis. All statistical analyses were done by using the Statistical Package for the Social Sciences (SPSS) statistical program (version 15.0 for Windows; SPSS, Chicago, IL).

51 DRUID 6th framework programme Deliverable 1.2.2 Table 1. Overview of hypnotics and doses used by chronic users, listed in increasing order of expected residual effects. Indicated are age and gender of the user, duration and frequency of hypnotic use

Hypnotic Dose (mg) Gender Age Duration of use Nights per week (years) zolpidem 10 female 58 3 4 zolpidem 10 male 59 10 7 midazolam 7.5 male 68 1.5 7 midazolam 7.5 male 64 5 7 midazolam 7.5 male 64 4 7 midazolam 7.5 female 57 3 7 lormetazepam 0.5 male 55 7 6 temazepam 10 female 63 30 4 temazepam 10 male 65 4 4 temazepam 10 female 69 15 7 temazepam 20 female 56 1.5 7 zopiclone 3.75 female 58 3 7 zopiclone 3.75 female 63 15 7 zopiclone 3.75 female 60 2 7 nitrazepam 5 female 68 12 7 zopiclone 7.5 male 61 13 7 oxazepam 10 male 68 1.5 7 oxazepam 20 male 62 9 7 oxazepam 50 male 63 1 7 lormetazepam 2 male 63 10 4 flurazepam 15 female 56 11 7 clonazepam 0.5 female 67 7 7

52 DRUID 6th framework programme Deliverable 1.2.2 Table 2. Means (±SD) of pre-study group characteristics

Variable Chronic users Unmedicated Controls F† insomniacs (n=22) (n=20) (n=21) Sociodemographics Gender (male:female) 11:11 10:10 13:8 Age (years) 62.1 (4.4) 60.8 (5.9) 61.7 (5.0) .40 Education (years) 12.3 (3.3) 10.9 (2.7) 13.2 (3.3) 2.67 Average Annual Mileage (1000 km) 8.5 (6.3) 11.9 (11.5) 9.7 (4.2) .98 Driving License (years) 38.0 (8.3) 39.0 (8.4) 40.6 (5.9) .64 Sleep Pittsburgh Sleep Quality Index 12.6 (3.5)a 12.6 (2.6)a 2.4 (1.6) 100.20*** Groningen Sleep Quality Scale - general 9.4 (3.0)a 11.1 (1.7)a 1.0 (1.5) 126.08*** Sleep Wake Experience List‡ Sleep Initiation Problems 12a 10a 0 16.73*** Sleep Maintenance Problems 14a 13a 0 23.26*** Early Morning Awakenings 15a 5a 0 7.61* Difficulty Waking Up 2 2 0 2.12 Tiredness Upon Waking Up 3 2 0 2.86 Daytime Sleepiness 5 3 1 2.80 Psychological Symptom Checklist 90-R Sleeping Problems 9.6 (3.1)a 10.0 (2.5)a 3.3 (0.7) 52.57*** Depression 26.7 (9.7)a 27.1 (10.7)a 18.0 (3.0) 7.62*** Anxiety 15.8 (7.3)a 13.9 (4.8) 10.8 (1.3) 5.28** Phobic anxiety 8.9 (4.5) 7.4 (1.0) 7.3 (0.8) 2.42 Psychoneuroticism 141.9 (37.1)a 137.9 (39.6)a 103.0 (13.1) 9.37*** Somatization 20.5 (7.9)a 19.1 (5.9)a 14.2 (1.8) 6.80** Cognitive Insufficiency 16.1 (6.3)a 15.8 (8.1)a 11.0 (1.8) 4.78* Interpersonal Sensitivity 25.8 (6.5) 24.8 (8.4) 21.7 (5.6) 2.11 Hostility 7.4 (1.6) 7.9 (2.8) 6.8 (1.3) 1.41 Beck Depression Inventory 8.6 (5.0)a 8.6 (6.6)a 2.8 (3.0) 9.38*** State Trait Anxiety Inventory State Anxiety 35.3 (10.1)a 41.6 (11.0)a 28.1 (7.5) 10.03*** Trait Anxiety 38.4 (10.9)a 41.9 (13.1)a 27.6 (5.8) 9.30*** Multidimensional Fatigue Inventory General Fatigue 11.8 (3.7)a 11.9 (3.7)a 6.8 (2.5) 16.36*** Physical Fatigue 10.9 (3.5)a 10.6 (3.5)a 6.6 (2.3) 12.23*** Mental Fatigue 11.5 (3.9)a 10.8 (3.5)a 7.0 (2.4) 11.24*** Reduced Motivation 9.8 (3.6)a 9.6 (4.2)a 6.7 (2.5) 5.29** Reduced Activity 10.1 (4.2)a 10.0 (3.2) 7.4 (2.6) 4.03* † = degrees of freedom (df) are 2,62, except for State Trait Anxiety Inventory – State: 2,61 and Trait: 2,56; ‡ = values for each variable are the frequencies in each group. Statistical analysis was conducted with the Kruskal-Wallis test, depicted are the Chi-square values; * = p<0.05; ** = p<0.01; *** = p<0.001; a = significant difference with control group (p<0.05; post-hoc analysis with Bonferroni correction); b = significant difference between frequent and in frequent users (p<0.05; post-hoc analysis with Bonferroni correction)

53 DRUID 6th framework programme Deliverable 1.2.2 Table 3. Sleep diary

Variable Chronic users Unmedicated Controls F† p insomniacs (n = 22) (n = 20) (n = 21) Groningen Sleep Quality Scale 6.0 (2.6)a 6.8 (2.0)a 1.9 (1.1) 32.76 <.001 Time in Bed (min) 521.0 (61.1) 498.9 (52.7) 495.1 (36.8) 1.53 .226 Sleep Onset Time (min) 43.0 (36.9)a 43.6 (29.8)a 18.3 (4.1) 5.44 .007 Awakenings (#) 0.75 (0.69)b 1.38 (0.93)a 0.46 (0.44) 8.50 .001 Early Morning Awakening (min) 33.1 (42.6) 50.0 (41.5)a 17.8 (13.8) 4.07 .022 Total Sleep Time (min) 410.9 (73.1)b 348.5 (71.2)a 440.3 (36.8) 10.63 <.001 Sleep Efficiency (%) 79.3 (13.3)a,b 69.2 (11.8)a 89.2 (6.7) 16.09 <.001 † = degrees of freedom (df) for Groningen Sleep Quality Scale: 2,61, for Time in Bed: 2,60, for Sleep Onset Time: 2,59, for Number of Awakenings:

2,58, for Early Morning Awakening: 2,59, for Total Sleep Time: 2,59, and for Sleep Efficiency: 2,59; a = significant difference with control group

(p<0.05; post-hoc analysis with Bonferroni correction); b = significant difference between frequent and in frequent users (p<0.05; post-hoc analysis with Bonferroni correction)

54 DRUID 6th framework programme Deliverable 1.2.2 Table 4. Mean (±SD) of objective and subjective sleep parameters

Variable Chronic users Unmedicated Controls F† p insomniacs (n = 22) (n = 20) (n = 21) Subjective Groningen Sleep Quality Scale 6.7 (3.9)a 7.8 (3.8)a 3.9 (3.5) 6.24 .003 Sleep Onset Time (min) 47.6 (97.9) 67.9 (73.0) 33.0 (36.8) 1.19 .321 Awakenings (#) 2.1 (1.6) 3.0 (2.5) 1.8 (1.5) 1.97 .149 Early Morning Awakening (min) 49.1 (57.5) 67.4 (62.0)a 26.4 (34.7) 3.23 .046 Total Sleep Time (min) 351 (101) 302 (94) 377 (111) 2.92 .062

Polysomnographic parameters Continuity Sleep Onset Time (min) 26.4 (9.4) 19.4 (13.4) 19.2 (14.6) 2.20 .120 Wake After Sleep Onset (min) 76.5 (65.0) 73.1 (39.4) 55.2 (35.7) 1.16 .321 Awakenings (#) 7.9 (4.1) 10.2 (4.3) 7.5 (4.5) 2.37 .102 Total Sleep Time (min) 383 (35) 389 (46) 408 (40) 2.15 .125 Sleep Efficiency (%) 79.6 (8.8) 80.8 (9.5) 84.6 (8.1) 1.82 .170

Architecture Stage 1 Sleep (% of Total Sleep Time) 6.6 (2.9) 6.9 (2.4) 6.1 (3.0) 0.50 .612 Stage 2 Sleep (% of Total Sleep Time) 56.6 (6.1) 54.5 (6.5) 56.0 (9.0) 0.44 .644 Stage SWS Sleep (% of Total Sleep Time) 18.4 (8.5) 18.5 (5.9) 16.5 (5.5) 0.58 .566 Stage REM Sleep (% of Total Sleep Time) 18.4 (4.3) 20.1 (5.8) 21.4 (6.0) 1.61 .209 † = degrees of freedom (df) for all parameters: 2,61; a = significant difference with control group (p<0.05; post-hoc analysis with Bonferroni correction); b = significant difference between frequent and in frequent users (p<0.05; post-hoc analysis with Bonferroni correction)

55 DRUID 6th framework programme Deliverable 1.2.2 Table 5. Overview of hypnotics and doses used by chronic users, listed in increasing order of expected residual effects.

Indicated are recommended therapeutic dose, drug’s half-life, expected residual effects, each users’Standard Deviation of Lateral Position (SDLP) scores, associated drug serum concentrations age and referral concentrations a Hypnotic Dos Recommend t1/2 (hours) Residual SDLP Serum Concentration at 8-11 hrs e ed effects (in concentratio post- dose (ng/mL)d (mg) therapeutic 8-12 hrs cm) n (ng/mL) at dose for post 9.5 hrs post adults (mg) dosec dose (for elderly) zolpidem 10 10 (5) 1.9 ± 0.2 unlikely 12.5 10.2 5.8 (11 hrs; plasma; 20 mg) zolpidem 10 10 (5) 1.9 ± 0.2 unlikely 15.6 53.1 5.8 (11 hrs; plasma; 20 mg) midazolam 7.5 7.5 (7.5) 1.9 ± 0.6 unlikely 13.2 9.5 1.5 (8 hrs; plasma; 7.5 mg) midazolam 7.5 7.5 (7.5) 1.9 ± 0.6 unlikely 16.8 8.1 1.5 (8 hrs; plasma; 7.5 mg) midazolam 7.5 7.5 (7.5) 1.9 ± 0.6 unlikely 16.3 52.3 1.5 (8 hrs; plasma; 7.5 mg) midazolam 7.5 7.5 (7.5) 1.9 ± 0.6 unlikely 19.9 ND 1.5 (8 hrs; plasma; 7.5 mg) lormetazepa 0.5 1 (0.5) 10 ± 2.5 unlikely 20.4 2.6 4.3 (11 hrs; serum; 1 mg) m temazepam 10 20 (10) 11 unlikely 14.7 191.7 160 (12 hrs; serum; 20 mg) temazepam 10 20 (10) 11 unlikely 27.6 85.4 160 (12 hrs; serum; 20 mg) temazepam 10 20 (10) 11 unlikely 24.0 198.7 160 (12 hrs; serum; 20 mg) temazepam 20 20 (10) 11 unlikely 13.3 137.6 160 (12 hrs; serum; 20 mg) zopiclone 3.75 7.5 (3.75) 5 unlikely 12.3 4.6 6.2 (8 hrs; plasma; 3.75 mg) zopiclone 3.75 7.5 (3.75) 5 unlikely 12.8 6.5 6.2 (8 hrs; plasma; 3.75 mg) zopiclone 3.75 7.5 (3.75) 5 unlikely 18.9 9.3 6.2 (8 hrs; plasma; 3.75 mg) nitrazepam 5 5 (5) 26 ± 3 minor 21.6 58.0 34 (8 hrs; serum; 5 mg) zopiclone 7.5 7.5 (3.75) 5 moderate 17.0 7.8 14.1 (8 hrs; plasma; 7.5 mg) oxazepam 10 20 (10) 8 ± 2.4 NA 16.5 37.6 231 (12.15 hrs; serum; 50 mg) oxazepam 20 20 (10) 8 ± 2.4 NA 15.4 122.9 231 (12.15 hrs; serum; 50 mg) oxazepam 50 20 (10) 8 ± 2.4 moderate 25.2 360.1 231 (12.15 hrs; serum; 50 mg) lormetazepa 2 1 (0.5) 10 ± 2.5 moderate 16.0 7.3 7.1 (11 hrs; serum; 2 mg) m flurazepam 15 30 (15) 1-2 (74 ± moderate 14.1 0.0 0.0 (>2 hrs; whole blood; 30 mg) 24)b clonazepam 0.5 1 (1) 19-60 NA 18.5 ND 5.5 (8 hrs; plasma; 1 mg) a Source Vermeeren (2004)5, and Riss et al. (2008)59; b half-life of active metabolite between brackets; c Source Vermeeren (2004)5 following prescribed dose ; d Source for zolpidem Bensimon et al. (1990)60, for midazolam Bornemann et al. (1985)61, for lormetazepam Brookhuis et al.

(1990)62, for temazepam Tedeschi et al. (1985)63, for zopiclone Billiard et al. (1987)64, for nitrazepam Lahtinen et al. (1978)65, for oxazepam Volkerts et al. (1992)66, for flurazepam Kaplan et al. (1973)67, and for clonazepam Wildin et al. (1990)68; ND = blood sample not drawn; NA = information not available

56 DRUID 6th framework programme Deliverable 1.2.2 Table 6. Mean (±SD) of driving performance parameters

Variable Chronic users Unmedicated Controls F† p (n = 22) insomniacs (n = 21) (n = 20) Highway Driving Test Standard Deviation of Lateral Position (cm) 17.4 (4.3) 17.7 (2.9) 16.8 (2.7) 0.38 .688 Standard Deviation of Speed (km/h) 2.11 (0.5) 2.34 (0.7) 2.18 (0.7) 0.76 .471 Car Following Test Reaction time (sec) 3.55 (1.57) 3.32 (1.40) 3.06 (1.01) 0.67 .518 Headway 1.14 (0.13) 1.21 (0.25) 1.15 (0.12) 0.94 .397 Subjective Evaluation of Driving Test Subjective Driving Quality (mm) 67.0 (11.2) 67.9 (9.7) 65.0 (13.9) 0.34 .713 Apparent Sedation (mm) 11.9 (11.4) 12.2 (14.9) 9.1 (10.2) 0.42 .657 Mental Effort (mm) 30.3 (22.6) 35.7 (24.9) 20.3 (14.1) 2.77 .071 † = degrees of freedom (df) for Highway Driving Test: 2,61, for Car Following test: 2,59, and for Subjective Evaluations: 2,62

57 DRUID 6th framework programme Deliverable 1.2.2 Table 7. Mean (±SD) of psychomotor and cognitive performance parameters and analysis of main effect of Group

Variable Chronic Unmedicated Controls F†,a p users insomniacs (n = 21) (n = 22) (n = 20) Critical Tracking Task Average lambda (in rad/sec) evening 3.34 (0.72) 3.23 (0.60) 3.03 (0.49) 0.06 .945 Average lambda (in rad/sec) morning 3.29 (0.70) 3.21 (0.72) 3.02 (0.49) Divided Attention Task Tracking subtask: Average Error (in mm) evening 14.1 (5.2) 14.4 (4.4) 17.4 (5.2) 0.75 .478 Tracking subtask: Average Error (in mm) morning 15.5 (5.9)b 14.9 (4.2) 18.7 (5.2) Detection subtask: Reaction Time (in msec) evening 1924 (328) 1974 (299) 1973 (335) 2.56 .086 Detection subtask: Reaction Time (in msec) morning 2030 (344) 1898 (288) 1920 (338) Stop Signal Task Go Reaction Tim e (in msec) 423 (63) 427 (77) 422 (50) 0.03 .972 Stop Signal Reaction Time (in msec) 179 (35) 178 (31) 181 (35) 0.05 .953 Psychomotor Vigilance Task Average Reaction Time (in msec) evening 269 (39) 257 (32) 264 (24) 1.33 .273 Average Reaction Time (in msec) morning 264 (39) 254 (34) 251 (22)b Median Reaction Time (in msec) evening 254 (34) 244 (31) 247 (23) 0.50 .610 Median Reaction Time (in msec) morning 252 (35) 242 (30) 241 (21) Lapses (>500 msec) evening 1.2 (2.0) 1.3 (1.1) 1.0 (1.3) 0.73 .486 Lapses (>500 msec) morning 1.3 (2.1) 0.7 (1.0) 0.7 (1.2) Word Learning Task Immediate Total Recall Score 46.3 (12.3) 46.8 (9.4) 49.4 (9.0) 0.53 .590 Delayed Recall Score evening 7.8 (3.5) 8.4 (2.9) 8.9 (3.4) 0.14 .870 Delayed Recall Score morning 6.5 (2.5)b 7.0 (3.4)b 7.3 (3.2)b Recognition Score 24.8 (4.4) 24.6 (3.0) 25.7 (3.4) 0.49 .615 Recognition Reaction Time (in msec) 924 (189) 892 (192) 883 (141) 0.33 .722 Digit Span Forward Score evening 3.6 (1.3) 3.8 (1.1) 3.9 (1.2) 1.43 .248 Forward Score morning 3.5 (0.9) 4.0 (1.1) 4.4 (1.1) Backward Score evening 3.5 (1.3) 3.9 (1.4) 4.3 (1.1) 0.04 .965 Backward Score morning 3.3 (1.0)c 3.8 (1.4) 4.2 (1.1) Subjective Evaluations of Feelings Alertness evening 65.9 (16.7)c 73.8 (13.2) 81.4 (13.4) 1.05 .356 Alertness morning 64.3 (18.6) 69.4 (14.0) 74.4 (14.6)b Contentedness evening 72.5 (21.2) 78.3 (13.0) 82.7 (12.6) 0.56 .576 Contentedness morning 74.8 (19.0) 76.3 (12.1) 81.1 (13.6) Calmness evening 69.8 (22.5) 74.9 (14.0) 82.4 (13.9) 0.62 .542 Calmness morning 72.7 (17.3) 73.0 (13.9) 79.8 (14.4) Karolinska Sleepiness Scale evening 3.9 (1.7) 3.9 (1.7) 3.0 (0.8) 0.26 .770 Karolinska Sleepiness Scale morning 4.6 (1.5) 4.3 (1.7) 3.8 (1.3)b †= degrees of freedom (df) for all parameters: 2,60, except for Psychomotor Vigilance Task: 2,59 and for Stop Signal Task: 2,62; a = test for Session by Group interaction, except for the Stop Signal Task which was only assessed in the evening session; b = significant Session effect p<0.05; c = significant difference from control group p<0.05

58 DRUID 6th framework programme Deliverable 1.2.2 Results

Pre-study group characteristics Table 2 illustrates the descriptive variables of sociodemographic, sleep and psychological data. One-way ANOVA showed that there were no differences between the groups in years of age, years of education, average annual mileage and years of possession of a driving license. Evaluation of sleep at home differed significantly between groups. Sleep quality was poorer in both insomnia groups as compared to controls, as indicated by significantly higher scores on the PSQI, GSQS- general and the sleep subscale of the SCL90-R (p<0.001). There were no differences in mean PSQI, GSQS and SCL90-R sleep scores between the insomnia groups. Sleep complaints were most frequently classified as sleep initiation and sleep maintenance problems, using the SWEL, with similar prevalences in both groups of insomniacs. Early morning awakening was more frequent in chronic users compared to unmedicated insomniacs, but the difference was not significant. None of the controls reported problems with sleep onset, sleep maintenance or early awakening. The two-week sleep diary showed that complaints of disturbed sleep as measured by the GSQS- specific were significantly increased in both insomnia groups as compared to controls (table 3). Subjective estimates of sleep times, averaged over two weeks, showed that sleep was worst in the unmedicated insomniacs. Compared with the chronic users and the healthy, good sleepers, the unmedicated insomniacs reported significantly more nocturnal awakenings (p<0.001), shorter total sleep time (p<0.001), and reduced sleep efficiency (p<0.001). Compared to controls they also reported significantly longer sleep onset times (p<0.018) and earlier awakening in the morning (p <0.018). Chronic users also reported significantly longer sleep onset times than controls (p<0.018), and reduced sleep efficiency (p<0.017). The difference from controls in total sleep time and early morning awakenings did not reach significance. Both insomnia groups scored significantly higher than controls on rating scales of depression (BDI: p<0.001, SCL90-R: p<0.004), without significant differences between the chronic users and unmedicated insomniacs. Anxiety was also increased in insomniacs as compared to controls, as shown by significantly higher scores on both subscales of the STAI in both groups (STAI state: chronic users p=0.044 and unmedicated insomniacs p<0.001; STAI trait: chronic users p=0.005 and unmedicated insomniacs p<0.001). A significant difference on the SCL90-R anxiety scale was found between the unmedicated insomniacs and controls (p=0.006), but not between the chronic users and controls. There were no significant differences in anxiety ratings between the insomnia groups. In addition to sleeping problems, depression and anxiety, the SCL90-R showed that insomnia patients scored higher than controls on the subscales for somatization (chronic users p=0.002 and unmedicated insomniacs p=0.027), cognitive insufficiency (chronic users p=0.021 and unmedicated insomniacs p=0.041), and psychoneuroticism (chronic users p=0.001 and unmedicated insomniacs p=0.003). The Fatigue Inventory showed that both insomnia groups reported suffering significantly more from general fatigue (both p<0.001), physical fatigue (both p<0.001) and mental fatigue (chronic users p<0.001 and unmedicated insomniacs p=0.002) when compared to the control group. Both groups reported significant reductions in motivation (chronic users p=0.015 and unmedicated insomniacs p=0.027), and the chronic users also in activity (p=0.040). No differences between the insomnia groups were found on any of

59 DRUID 6th framework programme Deliverable 1.2.2 these scales.

Sleep before driving Polysomnography showed no differences between the three groups on any of the sleep parameters recorded during sleep in the laboratory the night before driving (table 4). In contrast, subjective evaluations of sleep that night differed significantly between groups as measured by number of complaints in the GSQS-specific (p<0.003) and estimates of early morning awakening (p<0.046). Both insomnia groups reported significantly more sleep complaints than the control group (chronic users: p<0.043; unmedicated insomniacs: p<0.004), and the unmedicated insomniacs reported to awake significantly earlier than the controls (p<0.042). There were no significant group differences in subjective estimates of sleep onset time, number of awakenings and total sleep time.

Driving, cognitive and psychomotor performance Table 5 shows individual serum concentrations of hypnotics used by chronic users and their corresponding SDLP scores. Analysis showed that driving performance as measured by the highway driving test and the car-following test did not reveal significant differences between the three groups (table 6).

29

27

25

23

21

19

17

Standard DeviationStandard of Lateral Position (SDLP, in cm) 15

13

11 Chronic Unmedicated Controls users insomniacs Figure 1. Mean (±SD) and individual SDLP scores for each group separately. Open circles (ż) indicate SDLP scores of subjects using no hypnotic or a hypnotic that produces no or minor residual effects; closed circles (Ɣ) indicate SDLP scores of subjects using hypnotics likely to produce moderate or severe residual effects

The primary performance parameter, SDLP in the highway driving test, was on average normal and similar for unmedicated insomniacs (17.7 ±2.9 cm), chronic users (17.4 ±4.3 cm) and controls (16.8 ±2.7 cm) (figure 1)

Performance in the critical tracking test, divided attention task and the stop signal task showed no significant

60 DRUID 6th framework programme Deliverable 1.2.2 differences between the groups. There was, however, a significant overall group difference in the Digit Span

Backward (F2,60=3.93, p<0.025). Pairwise comparison revealed significantly lower scores in the chronic users as compared to the controls (p<0.020). There were no significant differences between the unmedicated insomniacs and the controls, and between the insomnia groups (table 7). Comparisons between performance in the evening and morning sessions showed significant Time of Day differences in the reaction times in the PVT and delayed recall in the word learning test. The latter was as expected because the interval between learning and delayed recall increased from evening to morning sessions. Average reaction time in the PVT was significantly faster in the morning as compared to the evening in controls (F1,59=8.57, p<0.005), but not in the insomnia groups. Performance in the critical tracking test, divided attention task and the stop signal task showed no significant differences between times of day (evening vs. morning).

Subjective rating scales Analysis of the subjective evaluation of mood revealed an overall significant group difference in feelings of alertness (F2,60=4.44, p<0.016). Post-hoc comparisons showed that the chronic users felt significantly less alert in the morning than the controls (p<0.013). In addition, the controls felt more alert in the evening when compared to the morning (p<0.004). Scores on the Karolinska Sleepiness Scale were not different between the three groups. There was, however, a significant between the evening and morning evaluation (F1,60=7.25, p<0.009). The control group reported significantly more feelings of sleepiness in the morning as compared to the evening (p<0.006).

Discussion

The present experimental study is the first directly comparing driving performance between insomnia patients who chronically use hypnotics, insomnia patients who do not or infrequently use hypnotics and healthy, good sleepers. Results show that driving, as measured by a standardized highway driving test and a car-following test, is not impaired in insomniacs, irrespective of the use of hypnotics. In addition, the present study shows that driving related psychomotor and cognitive performance is virtually similar between insomnia patients and healthy, good sleepers. Only working memory, as measured by the Digit Span backward, appeared to be significantly worse in the chronic users compared to the healthy controls. No group differences were found in verbal memory, divided attention, psychomotor function and inhibitory control. Results of the study corroborate previous findings by Vignola et al. showing an absence of neuropsychological deficits in both chronic users of hypnotics and unmedicated insomniacs.10 These investigators compared sleep and performance between 20 chronic users of benzodiazepine hypnotics, 20 unmedicated insomniacs and 20 good sleepers, using a battery of neuropsychological tests, polysomnography and subjective rating scales. Aside from lower scores in the Digit Span forward test, the study revealed no significant differences in performance between treated insomniacs, untreated insomniacs and healthy, good sleepers. Vignola et al. suggested that the absence of cognitive impairment in their study may have been due to methodological limitations. The tests used were of short duration and demanded low effort (e.g. digit symbol substitution test and purdue pegboard test). Consequently, insomnia patients may

61 DRUID 6th framework programme Deliverable 1.2.2 have been able to exert enough effort for a short period to complete the tests successfully. In the present study, subjects performed a 1-hour standardized highway driving test and a 25-minutes car following test. The prolonged attentional demands of these tasks were expected to reveal possible performance deficits in insomnia patients, which were not found with tasks of short duration. Nevertheless, there were no indications of deterioration in driving performance in insomnia patients, irrespective of use of hypnotics. These results can be interpreted in two ways. First, it can be argued that daytime performance was unaffected because sleep in the insomnia groups was not significantly different from controls as shown by polysomnographic data. This may be due to sleeping in a different environment, as home-based polysomnographic recordings have shown that insomnia patients’sleep appears more disturbed when they sleep at home than when they sleep at the laboratory.47 Indeed, in the present study, subjective sleep evaluations showed that sleep quality at the laboratory improved in both insomnia groups compared to sleep at home. In addition, subjective sleep quality for the healthy controls was worse during laboratory sleep than at home. These changes in sleep quality may have diminished differences between the groups. On the other hand, the failure to find objective sleep disturbances in the insomnia groups was expected. It is known that discrepancies between subjective and objective sleep parameters are characteristic in a majority of primary insomnia patients.e.g. 10, 48, 49 Assuming that patients’sleep is somehow disturbed, it is suggested that the current standards of sleep analysis may not be adequate for distinguishing insomnia from healthy, undisturbed sleep.50 A possible solution may be found in spectral analysis of the sleep microstructure, dissociating characteristic electroencephalographic activities. A second explanation for the absence of impairment in driving performance may be that the driving tests were not very demanding and patients were therefore able to complete the tasks normally by investing a little more effort. Driving is a well practiced and highly automated skill51 and may not require such high demands in particularly experienced drivers. All participants in the present study had ample driving experience and may not have had any difficulties in performing the test. Judging from the low scores on the mental effort scale this assumption seems to be confirmed. Scores on the scale can range from 0 to 150. The average scores in the present study were 30.3 for the chronic users, 35.7 for the unmedicated insomnia patients and 20.3 for the healthy controls, which corresponds to low or little effort. In a study comparing cognitive performance between patients with seasonal allergic rhinitis and healthy controls, subjects evaluated the mental effort they had to put in a 45- minutes Mackworth clock vigilance test considerably higher.52 Mental effort scores were around 90 for both groups, indicating substantially higher demands of that test as compared with the highway driving test. Still, the insomniacs, in particular the chronic users group, evaluated the degree of mental effort they had to put in the driving test tentatively higher than the healthy controls. Although not reaching statistical significance, this supports the idea that the insomnia patients compensated possible performance difficulties by increasing their effort. Establishing performance impairment in insomnia may require more demanding tasks.53 In addition to the findings that driving is not affected in insomnia patients, results of the present study provided no evidence that driving is impaired in patients chronically using hypnotics. One explanation may be related to tolerance. The absence of impairment, combined with the still present subjective sleep complaints suggests the development of, at least partial, tolerance to both therapeutic and residual effects of hypnotics. With respect to the residual effects, the results are partly supported by epidemiological data,17 showing that prolonged use of hypnotics is associated with a lowered risk of becoming involved in a car

62 DRUID 6th framework programme Deliverable 1.2.2 accident when compared with initial use of hypnotics. Nevertheless, the risk of injurious traffic accidents after chronic use of hypnotics remained twice as high in long-term hypnotic users in comparison to healthy, unmedicated drivers. An alternative explanation for the absence of residual effects on driving in the present study may be related to the wide variety of hypnotic drugs and doses used by the chronic users. Most importantly, the majority of hypnotics used are considered unlikely to produce residual effects.5 Consequently, the differences in degree of residual effects may explain the relatively high variability of performance in this group as compared to controls. Averaging may have masked any detectable impairment which is only associated with hypnotics belonging to category II or III, i.e. drugs and doses judged likely to produce moderate or severe impairment.54, 55 This was confirmed by post hoc inspection of the average SDLP scores from the category I users and the category II users, showing that performance of the former group was better than that of the latter group. Mean (±SD) SDLP scores were 16.8 (4.7) cm and 18.3 (3.5) cm respectively. Future research in patients chronically using the same hypnotic is needed to shed more light on this issue. The present study aimed, however, to evaluate driving performance in a representative, non-selective study sample of insomnia patients chronically using hypnotics. In contrast to driving, chronic use of hypnotics did seem to impair memory performance. Chronic users performed significantly worse on the Digit Span Backward as compared to the controls. This was supported by other parameters in memory tests showing a tentative non-significant pattern of cognitive decline in the chronic users. On average, performance in the Word Learning Test and the Digit Span Forward was worse in chronic users than in unmedicated insomniacs and controls. This does not seem to be the result from general sedation following administration of a hypnotic as no impairment was found in the psychomotor tests. Rather, the pattern suggests a general cognitive deterioration due to long-term use, confirming previous findings as reviewed by Barker et al.56 Finally, it may be questioned whether the study had sufficient power to detect differences in driving performance between insomnia patients and controls. We do not believe this is the case. First of all, the mean difference in SDLP between unmedicated patients and controls was small (i.e. less than 1 cm), and not considered clinically relevant. The minimum clinically relevant difference in SDLP is 2.4 cm which corresponds to the effect of alcohol when blood alcohol concentrations are at the legal limit for driving in most countries. Secondly, previous studies assessing driving performance in other patients groups, using the same standardized driving test, have shown that the method is sufficiently sensitive to detect significant impairment in small samples of patients with chronic nonmalignant pain57 and depressed patients receiving long-term antidepressant treatment.58 It seems therefore that the effects of insomnia are relatively small and less debilitating than the effects of pain and depression. To conclude, results of the present study indicate that driving performance is not impaired in patients suffering from insomnia, irrespective of use of hypnotics. Chronic users of hypnotics perform worse than healthy controls on the Digit Span backward. However, other driving related psychomotor and cognitive performance appears not to be affected in medicated and unmedicated insomnia patients. Therefore, studies investigating residual effects of hypnotics on driving performance in healthy volunteers are not expected to yield different results than in insomnia patients.

Acknowledgements

63 DRUID 6th framework programme Deliverable 1.2.2 This study is financially supported by the integrated project Driving Under the Influence of Drug, Alcohol and Medicine (DRUID), which is part of the European Union’s 6th Framework Programme. The article reflects only the authors’ view. The European Community is not liable for any use that may be made of the information contained therein. The authors like to express their gratitude to Gwenda Engels, Nicky van Gennip, Jolien Gooijers, Liene Ketelslegers, Jasmijn Kromhout, Loes van Langen, Anita van Oers, Elmy Theuniszen, Natalie Valle Guzman, Floor van de Water and Tim Weysen for the assistance in data collection; Renilde van den Bossche for the polysomnographic analyses; Cees van Leeuwen for the medical supervision; Henk Brauers, Willy Jeurissen and Jo Gorissen for ensuring the safety of the subjects during driving; and Irma Brauers for the logistic work.

References

1. Ohayon MM. Epidemiology of insomnia: what we know and what we still need to learn. Sleep Med Rev. 2002;6:97-111. 2. Morin CM, Leblanc M, Daley M, et al. Epidemiology of insomnia: Prevalence, self-help treatments, consultations, and determinants of help-seeking behaviors. Sleep Med. 2006;7:123-30. 3. Morin AK, Jarvis CI, Lynch AM. Therapeutic options for sleep-maintenance and sleep-onset insomnia. Pharmacother. 2007;27:89-110. 4. Verster JC, Volkerts ER, Olivier B, et al. Zolpidem and traffic safety - the importance of treatment compliance. Curr Drug Safe. 2007;2:220-6. 5. Vermeeren A. Residual effects of hypnotics: epidemiology and clinical implications. CNS Drugs. 2004;18:297-328. 6. Chambers MJ, Keller B. Alert insomniacs: Are they really sleep deprived? Clin Psychol Rev. 1993;13:649-66. 7. Varkevisser M, Kerkhof GA. Chronic insomnia and performance in a 24-h constant routine study. J Sleep Res. 2005;14:49-59. 8. Riedel BW, Lichstein KL. Insomnia and daytime functioning. Sleep Med Rev. 2000;4:277-98. 9. Fulda S, Schulz H. Cognitive dysfunction in sleep disorders. Sleep Med Review. 2001;5:423-45. 10. Vignola A, Lamoureux C, Bastien CH, et al. Effects of chronic insomnia and use of benzodiazepines on daytime performance in older adults. J Gerontol. 2000;55:P54-62. 11. Drummond SPA, Smith MT, Orff HJ, et al. Functional imaging of the sleeping brain: review of findings and implications for the study of insomnia. Sleep Med Rev. 2004;8:227-42. 12. Ashton H. The diagnosis and management of benzodiazepine dependence. Curr Opin Psychiat. 2005;18:249-55. 13. Paterniti S, Dufouil C, Alperovitch A. Long-term benzodiazepine use and cognitive decline in the elderly: the Epidemiology of Vascular Aging Study. J Clin Psychopharmacol. 2002;22:285-93. 14. Ray WA, Fought RL, Decker MD. Psychoactive drugs and the risk of injurious motor vehicle crashes in elderly drivers. Am J Epidemiol. 1992;136:873-83. 15. Hemmelgarn B, Suissa S, Huang A, et al. Benzodiazepine use and the risk of motor vehicle crash in the elderly. JAMA. 1997;278:27-31.

64 DRUID 6th framework programme Deliverable 1.2.2 16. Barbone F, McMahon AD, Davey PG, et al. Association of road-traffic accidents with benzodiazepine use. Lancet. 1998;352:1331-6. 17. Neutel I. Benzodiazepine-related traffic accidents in young and elderly drivers. Hum Psychopharmacol. 1998;13:S115-S23. 18. Dubois S, Bedard M, Weaver B. The impact of benzodiazepines on safe driving. Traff Inj Prev. 2008;9:404-13. 19. Metsemakers JFM, Höppener P, Knottnerus JA, et al. Computerized health information in the Netherlands: a registration network of family practices. Br J Gen Pract. 1992;42:102-6. 20. American Psychiatric Association. DSM-IV: Diagnostic and Statistical Manual of Mental Disorders. 4 ed. Washington DC: American Psychiatric Press; 1994. 21. Buysse DJ, Reynolds CF, 3rd, Monk TH, et al. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiat Res. 1989;28:193-213. 22. Van Diest R, Milius H, Markusse R, et al. The Sleep-Wake Experience List. Tijdschr Soc Gezondheidsz. 1989;67:343-7. 23. Mulder-Hajonides van der Meulen WREH. Measurement of subjective sleep quality. Amsterdam: Elsevier; 1981. 24. Derogatis LR. The SCL-90-R: Administration, Scoring and Procedures Manual II. Baltimore: Baltimore Clinical Psychometric Research; 1983. 25. Beck AT, Ward CH, Mendelson M, et al. An inventory for measuring depression. Arch Gen Psychiat. 1961;4:561-71. 26. Spielberger CD, Gorsuch RL, Lushene RE, et al. Manual for the State-Trait Anxiety Inventory (Form Y). Palo Alto: Consulting Psychologists Press, Inc.; 1983. 27. Smets EM, Garssen B, Bonke B, et al. The Multidimensional Fatigue Inventory (MFI) psychometric qualities of an instrument to assess fatigue. J Psychosom Res. 1995;39:315-25. 28. Iber C, Ancoli-Israel S, Chesson A, et al. The AASM manual for the scoring of sleep ans associated events: rules, terminology and technical specifications. Westchester, Ill 2007. 29. O'Hanlon JF. Driving performance under the influence of drugs: rationale for, and application of, a new test. Br J Clin Pharmacol. 1984;18:121s-9s. 30. Brookhuis KA, De Waard D, Mulder LJ. Measuring driving performance by car-following in traffic. Ergonomics. 1994;37:427-34. 31. Ramaekers JG, O'Hanlon JF. Acrivastine, terfenadine and diphenhydramine effects on driving performance as a function of dose and time after dosing. Eur J Clin Pharmacol. 1994;47:261-6. 32. Vermeeren A, O'Hanlon JF, Declerck AC, et al. Acute effects of zolpidem and flunitrazepam on sleep, memory and driving performance, compared to those of partial sleep deprivation and placebo. Acta Ther. 1995;21:47-64. 33. Vermeeren A, Danjou PE, O'Hanlon JF. Residual effects of evening and middle-of-the-night administration of zaleplon 10 and 20 mg on memory and actual driving performance. Hum Psychopharmacol. 1998;13:S98-S107. 34. Vermeeren A, Riedel WJ, van Boxtel MP, et al. Differential residual effects of zaleplon and zopiclone on actual driving: a comparison with a low dose of alcohol. Sleep. 2002;25:224-31. 35. Verster JC, Volkerts ER, Schreuder AH, et al. Residual effects of middle-of-the-night administration

65 DRUID 6th framework programme Deliverable 1.2.2 of zaleplon and zolpidem on driving ability, memory functions, and psychomotor performance. J Clin Psychopharmacol. 2002;22:576-83. 36. Leufkens TRM, Lund JS, Vermeeren A. Highway driving performance and cognitive functioning the morning after bedtime and middle-of-the-night use of gaboxadol, zopiclone and zolpidem. J Sleep Res. 2009;18:387-96. 37. Leufkens TRM, Vermeeren A. Highway driving in elderly the morning after bedtime use of hypnotics: a comparison between temazepam 20 mg, zopiclone 7.5 mg and placebo. J Clin Psychopharmacol. 2009;29:432-8. 38. Rey A. L'examen clinique en psychologie. Paris: Presses Universitaires de France; 1964. 39. Wechsler D. Wechsler Adult Intelligence Scale - Revised manual. New York: Psychological Corporation; 1981. 40. Jex HR, McDonnell JD, Phatak AV. A "critical" tracking task for man-machine research related to the operator's effective delay time. I. Theory and experiments with a first-order divergent controlled element. NASA CR-616 1966. 41. Moskowitz H. Proceedings: Psychological tests and drugs. Pharmakopsychiat Neuro Psychopharmakol. 1973;6:114-26. 42. Logan GD, Cowan WB, Davis KA. On the ability to inhibit simple and choice reaction time responses: a model and a method. J Exp Psychol Hum Percept Perform. Apr 1984;10:276-91. 43. Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods Instrum Comp. 1985;17:652-5. 44. Bond A, Lader M. The use of analogue scales in rating subjective feelings. Br J Med Psychol. 1974;47:211-8. 45. Akerstedt T, Gillberg M. Subjective and objective sleepiness in the active individual. Int J Neurosci. 1990;52:29-37. 46. Zijlstra FRH. Efficiency in work behavior. A design approach for modern tools. [PhD thesis]. Delft, Delft University of Technology; 1993. 47. Edinger JD, Fins AI, Sullivan RJ, Jr., et al. Do our methods lead to insomniacs' madness?: Daytime testing after laboratory and home-based polysomnographic studies. Sleep. 1997;20:1127-34. 48. Carskadon MA, Dement WC, Mitler MM, et al. Sel-reports versus sleep laboratory findings in 122 drug-free subjects with complaints of chronic insomnia. Am J Psychiat. 1976;133:1382-8. 49. Orff HJ, Drummond SPA, Nowakowski S, et al. Discrepancy between subjective symptomatology and objectiv neuropsychological performance in insomnia. Sleep. 2007;30:1205-11. 50. Bastien CH, LeBlanc M, Carrier J, et al. Sleep EEG power spectra, insomnia, and chronic use of benzodiazepines. Sleep. 2003;26:313-7. 51. Brouwer WH. Attention and driving: a cognitive neuropsychological approach. In: Leclercq M, Zimmermann P, eds. Applied Neuropsychology of Attention. Hove, UK: Routledge; 2002:230 - 54. 52. Hartgerink-Lutgens I, Vermeeren A, Vuurman E, et al. Disturbed cognitive functions after nasal provocation in patients with seasonal allergic rhinitis. Clin Exp Allergy. 2009;39:500-8. 53. Altena E, Van der Werf YD, Strijers RLM, et al. Sleep loss affects vigilance: Effects of chronic insomnia and sleep therapy. J Sleep Res. 2008;17:335-43. 54. Wolschrijn H, De Gier JJ, De Smet PAGM. Drugs and driving: a new categorization system for drugs

66 DRUID 6th framework programme Deliverable 1.2.2 affecting psychomotor performance. Maastricht: Institute for drugs, safety and behaviour, University of Limburg; 1991. IVGV 91. 55. De Gier JJ, Alvarez FJ, Mercier-Guyon C, et al. Prescribing and dispensing guidelines for medicinal drugs affecting driving performance. In: Verster JC, Pandi-Perumal SR, Ramaekers JG, et al., eds. Drugs, Driving and Traffic Safety. Basel, Switzerland: Birkhäuser Verlag; 2009:121-34. 56. Barker MJ, Greenwood KM, Jackson M, et al. Cognitive effects of long-term benzodiazepine use: a meta-analysis. CNS Drugs. 2004;18:37-48. 57. Veldhuijzen DS, Van Wijck AJM, Wille F, et al. Effect of chronic nonmalignant pain on highway driving performance. Pain. 2006;122:28-35. 58. Wingen M, Ramaekers JG, Schmitt JA. Driving impairment in depressed patients receiving long-term antidepressant treatment. Psychopharmacol. 2006;188:84-91. 59. Riss J, Cloyd J, Gates J, et al. Benzodiazepines in epilepsy: pharmacology and pharmacokinetics. Acta Neurol Scand. 2008;118:69-86. 60. Bensimon G, Foret J, Warot D, et al. Daytime wakefulness following a bedtime oral dose of zolpidem 20 mg, flunitrazepam 2 mg and placebo. Br J Clin Pharmacol. 1990;30:463-9. 61. Bornemann LD, Min BH, Crews T, et al. Dose dependent pharmacokinetics of midazolam. Eur J Clin Pharmacol. 1985;29:91-5. 62. Brookhuis KA, Volkerts ER, O'Hanlon JF. Repeated dose effects of lormetazepam and flurazepam upon driving performance. Eur J Clin Pharmacol. 1990;39:83-7. 63. Tedeschi G, Griffiths AN, Smith AT, et al. The effect of repeated doses of temazepam and nitrazepam on human psychomotor performance. Br J Clin Pharmacol. 1985;20:361-7. 64. Billiard M, Besset A, de Lustrac C, et al. Dose-response effects of zopiclone on night sleep and on nighttime and daytime functioning. Sleep. 1987;10:27-34. 65. Lahtinen U, Lahtinen A, Pekkola P. The effect of nitrazepam on manual skill, grip strength, and reaction time with special reference to subjective evaluation of effects on sleep. Acta Pharmacol Toxicol. 1978;42:130-4. 66. Volkerts ER, Van Laar MW, Van Willigenburg APPP, et al. A comparative study of on-the-road and simulated driving performances after nocturnal treatment with lormetazepam 1 mg and oxazepam 50 mg. Hum Psychopharmacol. 1992;7:297-309. 67. Kaplan S, deSilva JAF, Jack ML, et al. Blood level profile in man following chronic oral administration of flurazepam hydrochloride. J Pharm Sc. 1973;62:1932-5. 68. Wildin JD, Pleuvry BJ, Mawer GE, et al. Respiratory and sedative effects of clobazam and clonazepam in volunteers. Br J Clin Pharmacol. 1990;29:169-77.

67 DRUID 6th framework programme Deliverable 1.2.2 Chapter 3 : Effects of benzodiazepines on driving performance of anxiety patients

Katerina Touliou, CERT/HIT, Greece

68 DRUID 6th framework programme Deliverable 1.2.2 Abstract

Alprazolam is a widely prescribed anxiolytic for the treatment of anxiety, panic disorder, and depression. Reported adverse effects after alprazolam intake include sleepiness, sedation, drowsiness, and reduced alertness. Studies based on laboratory tests have found that alprazolam affects memory, attention and tracking by decreasing performance in most cases (Seppala et al., 1986; de Gier et al., 1986). In this study, the alprazolam effect (0.5 mg) was investigated in three groups: a) treated anxiety patients, b) untreated anxiety patients, and c) control group. 51 participants matched for age, gender and driving experience completed two driving tasks; a lane tracking and a car following scenario in a simulated environment. Driving variables, cognitive measures, blood samples and subjective assessments were gathered in a baseline and an oral administration of alprazolam (0.5 mg) condition. Alprazolam administration impaired weaving control (SDLP) in lane tracking scenario for all groups. It appears that moderate concentrations of alprazolam in blood serum (ng/mL) are associated with impairment in road tracking for the control group. For the untreated anxiety patients only two concentration levels were detected and no improvement in lateral position keeping was observed. Impairment (i.e. less weaving) was found in the treated anxiety group for the serum concentration groups of 5-10 and >15 ng/mL. Delayed brake reaction times (sec) were observed in treated and untreated anxiety patients in the car following scenario. Healthy participants showed riskier behaviour after alprazolam administration compared to treated and untreated anxiety patients (p<.001) who showed increased percentage of time spent with Time-to- Collision (TTC) values between 2 to 4 seconds. Alertness in attentional performance tests was significantly decreased only in healthy participants (p=.015). Only untreated patients and healthy participants reported decreased vigilance. This study clearly showed an alprazolam effect (0.5mg) in driving performance in treated, untreated anxiety patients and healthy participants. Future studies might focus on dose-related relationships between the same groups. Additionally, further comparisons with other medicinal drugs would be desirable to be investigating in the future reflecting the real medicinal drug users and, consequently, the real driving performance. Conclusively, the main findings of this study are in agreement with current research that alprazolam has a detrimental effect on driving behaviour. Thus, people under alprazolam medication should be informed by general practitioners about the impairing effects of alprazolam administration to their everyday activities and driving.

69 DRUID 6th framework programme Deliverable 1.2.2 Introduction

Alprazolam (Generic Xanax®) is a benzodiazepine derivative mainly prescribed for the treatment of generalised anxiety, panic disorder, and depression. Alprazolam is the most often prescribed psychoactive substance (Verster et al., 2002). The usual clinical dosages of alprazolam, administrated in divided doses, range from 0.5 mg to 4 mg/day for the treatment of anxiety disorder and from 6 to 10 mg/day for the treatment of panic disorder. Anxiety is often accompanied by depression; alprazolam is being prescribed for these cases as it has antidepressant properties (Petty et al., 1995). Reported adverse effects after alprazolam intake include sleepiness, sedation, drowsiness, and reduced alertness. Alprazolam is a widely prescribed medicine mainly prescribed to outpatients and it is the most used benzodiazepine for recreational purposes. It affects all aspects of everyday living activities including driving and operating machinery. Many studies have investigated its effect in cognitive and driving performance. Relevant studies have focussed on cognitive impairments that both directly and indirectly investigate the effect of alprazolam and other benzodiazepines on driving performance in close tracks or driving simulators. Studies based on laboratory tests have found that alprazolam affects memory, attention and tracking by decreasing performance in most cases (Seppala et al., 1986; de Gier et al., 1986). However, not all benzodiazepines deteriorate cognitive performance (Busto et al., 2000). It is questionable whether the disparities spring from variations in properties and effects of different benzodiazepines or because the tests do highlight diverse aspects of cognitive functioning with the application of various tests as different measures may lead to different results. Moreover, studies have been conducted in order to investigate if the clinical suggestion of tolerance after a few days of administration holds true. Current literature is in line with this proposition; it has been found that chronic use does not deteriorate performance. Impairment may be limited to the early stages of benzodiazepine intake, with the general clinical suggestion that tolerance develops within a few days of benzodiazepine use (Busto et al., 2000). Benzodiazepines’effect on driving performance (e.g. steering, reaction time, lane keeping) has been supported by many studies performed either on road or in simulated environment (Mortimor, 1986; Smiley, 1985). On the other hand, epidemiological and laboratory studies provide inconclusive findings regarding risk assessment and potentially impairing effects of benzodiazepines on driving behaviour (EMCDDA, 1999; Longo et al., 2001; deGier, 1986). On the whole, a handful of studies have investigated the effect of alprazolam in cognitive and psychomotor skills which are a related to driving and to actual driving measures. Snyder and colleagues (2005) conducted a study in order to investigate the degree and nature of cognitive impairment by the administration of two doses of alprazolam (0.5 and 1 mg). Several recordings of psychomotor function, working memory, visual attention, planning, and learning were obtained. Significant findings indicated that attention speed was affected by lower alprazolam doses, and higher dose affected psychomotor functions. The authors suggest the combination of these two led to the mediocre deterioration findings in memory and learning which are related to executive functions. Thus, lower doses affect simple choices and higher doses affect more complex functions. The participants in this study were healthy participants and not alprazolam users. Therefore, these findings might be important for developing baseline assessments for further comparisons aiming at dose-related relationships in relevant patient groups. Current research suggests that alprazolam impairs cognitive performance after acute dosages (e.g.

70 DRUID 6th framework programme Deliverable 1.2.2 Ellinwood et al., 1995; Kroboth et al., 1998; Vermeeren et al., 1995) in skills related to driving such as tracking, reaction and alertness. As noted above, the application of diverse measure types in order to investigate the effect of different types of benzodiazepines on driving performance it makes it even harder and complex to compare the results from different studies. The investigation of potential effects of alprazolam on driving with the application of cognitive and laboratory tests might allow erroneous transfer of inferences as related validity is questionable. Volkerts and colleagues (1992) advocated that there is low sensitivity in the inferences and findings derived by laboratory and cognitive tests when compared to real traffic and on the road tests. Verster and colleagues (2002) investigated the acute effects of alprazolam (1mg) on driving performance during real traffic in conjunction with laboratory tests related to driving skills. Participants had to complete a standardised driving task on a highway one hour after alprazolam administration. A cognitive tests’battery was administered two and a half hours after the driving task. Statistically significant differences were found between the alprazolam and placebo groups for Standard Deviation of Lateral Position (SDLP) and Standard Deviation of Speed (SDS) accompanied by impairment in laboratory tests. Vermeeren and colleagues (2009) reviewed related literature on anxiolytics’effect on driving. Their discussion on over-the-road driving points out that the mean increase of SDLP in this study was comparable to BAC=1.5mg/ml (Louwerens et al., 1987). Subjective scales showed, also, impairment in driving quality, decrease in alertness, lower mental activation, and increased mental effort. The comparability level of anxiolytics consumption to alcohol concentrations that are far above the legal limit in most countries (i.e. some countries have “zero tolerance” towards alcohol consumption) brings to attention the increased risk of patients to be involved in road accidents. The authors concluded that warnings should be given to users who are driving a car. Likewise, Leufkens and colleagues (2007) compared the effects of two types of release (extended release (ER) and immediate release (IR)) of alprazolam in both real driving settings and laboratory tests. The driving condition was, also, a highway driving task four hours after oral administration. Laboratory tests comprised cognitive and psychomotor tests that were administered one hour, 2 and a half hours, and five hours post alprazolam administration. Impairment was found for both formulations; however almost double was observed for the immediate release (XR) between four and five hours after administration. The findings were supported by the laboratory tests, which, also, showed impairment in memory functioning. Similar to the Verster and colleagues (2002) findings, the authors emphasised the increased risk of alprazolam users to be involved in traffic accidents. Verster and colleagues (2005) carried out a literature review on 14 placebo controlled and double blind studies which investigated the effects of anxiolytic drugs on on-the road tests. Standard Deviation of Lateral Position (SDLP) was the main driving parameter. They concluded that, among other types of benzodiazepines, a single dose of alprazolam might impair driving performance (i.e. increased weaving was recorded). The authors suggested patients treated with alprazolam should be cautioned when driving a car and that it might not be safe to drive while under alprazolam therapy. In general, relevant studies have shown detrimental impairment due to alprazolam administration on driving performance, controlled laboratory settings and subjective scales.

71 DRUID 6th framework programme Deliverable 1.2.2 Objectives

In the present study the major objective was to investigate the effect of alprazolam in treated and untreated anxiety patients compared to a healthy control group after oral alprazolam administration (0.5 mg) (acute phase) in a simulated environment. Primary variables were the vehicle variables (car simulator: driving performance measures). The secondary objective was to compare multiple cognitive and subjective measures, collected for each participant, in order to establish the whole range of driving impairment. It was hypothesised that alprazolam will affect driving performance of all groups. In addition, it was of interest to compare potentially additive effects (treated patients) to acute effects (untreated and healthy participants).

Methods

This section describes the participants’demographics and the sample selection process.

Participants

In total, 51 participants were recruited in the experiment. The following table presents age and gender distribution for each group.

Table 1: Group characteristics Group Age Gender (Mean±SD) (M/F) Treated 42.4±13.9 8/7 (Group A; N=15)

Untreated 36.9±8.9 9/9 (Group B; N=18) Control 35.4±8.8 8/10 (Group C; N=18)

All participants were screened prior participation and were medically examined by two collaborating doctors. Correct medical diagnosis was ensured by the collaborating doctors. In groups A and B (patients), the participants were diagnosed with anxiety and, specifically, they should have a Hamilton Anxiety Rating Scale score equal or greater than 20 (HAM-A: mild to moderate severity) (Hamilton, 1959). Patients in Group A were systematically using alprazolam for at least 2 months before the testing day. Patients in Group B did not receive any kind of treatment for at least two months before the testing day. Participants in Group C had no medical history of anxiety or alcohol abuse and were free of medication. Participants in Groups A, B and C were matched for age, gender and driving experience. Participants were

72 DRUID 6th framework programme Deliverable 1.2.2 experienced drivers (at least 3 years since they obtained their driving license) and were currently active drivers. Volunteers received reimbursement for their participation. In case of alcohol or drug screening tests were positive, participants were excluded from the study (with the exception of Group A, where participants were positive for benzodiazepines). Participants with other medical conditions (psychiatric, neurological conditions and/or cognitive impairments) were excluded from the study. Participants were asked to refrain from smoking prior testing and not to have any meal during the last four hours before the experiment. Before each phase (Tables 2 and 3), participants were medically examined. The collaborating doctor present was responsible for the acquisition and administration of alprazolam to the participants. The collaborating doctor closely monitored the participant during the experiment. If the collaborating doctor judged the participant was unable to perform any of the required tasks, the task was interrupted and the experiment stopped. Gathered data (subjective scales, vehicle logfiles, etc.) were anonymised and kept in safe place. The following steps were taken in order to ensure anonymity, confidentiality and protection of personal data. Only the chief administrator had access to raw data. Upon detailed experimental briefing, participants signed a written consent. Participants were assured that they were able to withdraw at any point without any consequences. If the participants were diagnosed with a medical problem during medical examination this was made known to them. The study was reviewed and approved by the Bioethics Committee of CERTH.1

Study design

The present study utilises a within repeated- and between-participants’design for the comparison of patient and control experimental groups in question. The independent variable is the alprazolam administration with two levels: (baseline and alprazolam 0.5 mg intake). Dependent variables are: a) the driving performance (simulated environment), b) attentional performance (winTAP), and c) subjective assessments. The aim was to investigate the acute effect of alprazolam administration (0.5mg) and the possibility of additive effects (treated) in two driving tasks.

Procedure

Participants were asked to arrive at the premises at the same time for each phase of the experiment (14:00- 17:00). At arrival they completed an introductory (background) questionnaire and written consent was obtained after detailed briefing. It was made sure that participants understood the objectives and the experimental procedure before they consented to participate Both alcohol screening (with breathanalyser) and urine drug screening were performed and if found negative, participants had a familiarisation drive in order to get used to the driving simulator. Participants had to complete the following tasks at the driving simulator: a) perform a lane tracking scenario for about 20 minutes in a highway environment maintaining a constant speed of 90 km/h and b) perform a car following scenario for about 20 minutes in a highway environment maintaining a safe distance from the

1 CERTH: Hellenic institute of Transport (HIT) is an institute of the non-profit organisation Centre for Research and Technology, Hellas . 73 DRUID 6th framework programme Deliverable 1.2.2 lead vehicle that was moving with a steady speed of 90 km/h. Four instances of abrupt breaking (leading vehicle) occurred randomly. Participants received instructions for each driving scenario. For the lane tracking scenario participants were instructed to maintain a constant speed of 90 km/h and steady lateral position. For the car following scenario they were instructed to maintain a safety distance from the lead vehicle. Scenarios were counter-balanced between the two phases and among participants. Following the simulator driving scenarios, participants completed the neuropsychological tests (winTAP, Zimmerman and Fimm, 1993). Participants had breaks between the driving scenarios and if required they could get more breaks. Participants were constantly monitored for simulator sickness symptoms. They were, also, screened during familiarisation driving phase by completing a simulator sickness questionnaire. Participants were asked not to have coffee or any alcoholic beverages 24 hours prior testing took place. Subjective questionnaires were filled in before and after each scenario and before and after the neuropsychological tests rating their sleepiness, mental effort, and driving quality. Detailed accounts of the driving scenarios, subjective scales, and the neuropsychological battery are given in the respective sections. The experimental study was carried out in three phases. The respective timelines are shown in tables 2 and 3. Baseline and treatment phases were counterbalanced in order to control for order effects and potentially related confounders. Blood collection lasted approximately 10-15 minutes and usually participants wanted to relax and take a small break before the driving tasks. Participants started the driving tasks almost 15 minutes after blood collection and about an hour after alprazolam intake (Table 3). The time difference between administration and driving is adequate for testing procedure and similar to relevant studies (e.g. Leufkens et al., 2007). Table 2: Phases I(familiarisation)-II(baseline assessment) timelines

Time Task -30 min Breathanalysing Urine screening

0 Scenario 1 (lane tracking) (KSS pre-post) or

Scenario 2 (car following) (KSS pre-post)

The order of scenarios is counterbalanced

+30 min Scenario 1 (lane tracking) (KSS pre-post) or

Scenario 2 (car following) (KSS pre-post)

The order of scenarios is counterbalanced

+60 min Attentional performance (WinTAP) (KSS pre-post) +90min Subjective assessments (RSME, DQS)

Grey parts of the table were parts of the experiments performed in phase II only to avoid overfamiliarisation. Phases II and III were separated by a wash out period of 7 days.

74 DRUID 6th framework programme Deliverable 1.2.2 Table 3: Phase III timeline (experimental measures)

Time Task -75 min Medical check Breathanalysing Urine screening -60 min Subjects are administered 0.5 mg of alprazolam -30 min Blood collection [serum/blood spots/whole blood (10ml)] The process of blood collection lasts 10-15 minutes (e.g. preparations, etc.)

0 Scenario 1 (lane tracking) (KSS pre-post) or

Scenario 2 (car following) (KSS pre-post)

The order of performance of scenarios is rotated among subjects.

+30 min Scenario 1 (lane tracking) (KSS pre-post) or

Scenario 2 (car following) (KSS pre-post)

The order of scenarios is counterbalanced

+60 min Attentional performance (WinTAP) (KSS pre-post) +90min Subjective assessments (RSME, DQS)

Blood samples were collected by a specialist (microbiologist) in 10 ml tubes and frozen after (both whole and serum) according to the proposed guidelines. Whole blood, serum and blood spot specimens were transported to the Institute of Legal Medicine and Traffic Medicine, University of Heidelberg, Germany for further analysis. The following detailed description of analysis and extraction was provided by the colleagues from the University of Heidelberg. Five calibration standards were prepared in each batch containing 2.5, 5, 10, 20, 50 ng alprazolam/mL and 0.125, 0.25, 0.5, 1.0, 2.5 ng Į-hydroxy-alprazolam/mL for each blood, serum and dried blood spots (DBS), respectively. DBS calibrators were prepared by spotting 100 µL of fortified blood onto filter paper which was dried at room temperature over night. DBS were punched out and transferred into plastic tubes. Extraction of DBS was performed in the same way as for whole blood, serum specimens and corresponding calibrators. 1.0 mL borate buffer (pH 8.5) and the internal standards (alprazolam-d5 20 ng/mL, Į-hydroxy-alprazolam-d5 2 ng/mL) were added to 100 µL serum, whole blood or DBS, respectively. Samples were extracted using 1.0 mL /iso-amylalcohol (95:5, v/v) and were shaken for about 10 min. Subsequently, samples were centrifuged (10 min at 4300g); the organic layer was transferred into a silanized vial and evaporated to dryness at 40°C. The residue was reconstituted in 50 µL of the mobile phase (4mM ammonium acetate buffer pH 3.2/methanol/acetonitrile 45:11:55, v/v/v). Analysis was performed on an API 4000 tandem mass spectrometer with a Turbo Ion ionization source operated in the positive-ion mode (Applied Biosystems, Darmstadt, Germany). It was interfaced to a HPLC pump equipped with an autosampler (1100 series, Agilent, Waldbronn, Germany). The samples (5 µL aliquots) were eluted from a Luna 5µ C18 (2) column (2.00 x 150 mm, 5 µm particle size,

75 DRUID 6th framework programme Deliverable 1.2.2 Phenomenex, Aschaffenburg, Germany) at a flow rate of 250 µL/min. Data were monitored in positive ionisation mode with the following transitions: alprazolam m/z 309à281* and 309à205; Į-hydroxy-alprazolam m/z 325.1à297.1* and 325.1à325.1; alprazolam-d5 m/z 314à286; Į-hydroxy-alprazolam-d5 m/z 330.1à302.1. Transitions marked with an asterisk were used for quantitation. Calibration lines were constructed with linear least squares regression using the ratio of the target analyte peak area to the corresponding internal standard peak area. Each specimen was tested twice, and the respective arithmetic mean is given. The assay has been evaluated according to international standards for the validation of bio analytical methods.

Driving and psychomotor tests

The CERTH/HIT driving simulator (Figure 1) is built around a Smart cabin equipped with sensors. The actuation of all control levers, windshield wipers, blinker, ignition key and light switch is electronically transmitted to the driving computer. All operational elements such as steering wheel, accelerator pedal, brake pedal, gearshift lever and handbrake lever, provide nature-true force reactions. The visual system includes five large-screens, each having a width of 2 meters. There visual system works with on-screen projection with video projectors (2500 ANSI-lumen). The sound system generates original sounds according to the situation (starter, engine noise, horn, screeching of tires, drive wind, rain, etc.). The vibration device creates nature true vibrations of the car according to the revolvation of the simulated engine.

Figure 1: Driving simulator

The simulator is equipped with special software which allows the development of specific driving scenarios aiming at creating a monotonous driving environment specifically for the studies conducted within the framework of DRUID. The simulated environment is capable of incorporating surrounding traffic with maximum 30 road users with artificial intelligence agents, comprising passenger cars, trucks, pedestrians and cyclists. The primary driving variable was the Standard Deviation of Lateral Position (SDLP) for the lane tracking scenario which is a very reliable index of weaving and overall lateral control of the vehicle. The car following

76 DRUID 6th framework programme Deliverable 1.2.2 scenario was designed with stable leading vehicle speed, therefore it was decided to calculate percentages of time (%) driven with certain Time-to-Collision (TTC) values. falling into five categories. The selected five categories were the following:

TTC 0-1: percentage of time driven with TTC between 0 and 1 sec TTC 1-1.5: percentage of time driven with TTC between 1 and 1.5 sec TTC 1.5-2: percentage of time driven with TTC between 1.5 and 2 sec TTC 2-4: percentage of time driven with TTC between 2 and 4 sec TTC >4: percentage of time driven with TTC above 4 sec

As there is no previous literature on recommended stratification, the decision was based on consortiums’ recommendations for intervals of interest that would be of greater potential for more in depth data collection of critical TTCs (i.e. 1-2). Obviously, very low TTC values suggest higher risk for the driver to be involved in an accident. In addition, more time spent in the category TTC 2-4, implies that participants spent more time in following the leading vehicle with safety distance keeping; thus complying with the given instructions. More time spent with TTC >4 suggests that participants were less concentrated and not attentive to the driving task and, consequently, did not follow the driving instructions. The winTAP (Zimmermann and Fimm, 1993) is a standardised neuropsychological battery of attentional performance. The selected individual tests applied were the following: a) alertness: choice between two stimuli, b) Go/NoGo: simple choice reaction task, c) divided attention audio: reaction to two consecutive same sounds (two low or two high), d) divided attention: visual: reaction to the formation of a small rectangular by small stars, and e) divided attention with both visual and auditory stimuli presented simultaneously. Reaction times (msec), omissions and errors were recorded.

Secondary measures

Several variables are derived from the driving simulator logfiles. Another parameter of interest was the time it takes to react to an abrupt breaking of the lead vehicle. Participants’reaction times to the lead vehicle’s abrupt breaking were recorded for the car following scenario.

Subjective measures

The Karolinska Sleepiness Scale (KSS) is a universally accepted, validated and standardised scale (Åkerstedt and Gillberg, 1990). Participants were asked to subjectively evaluate their vigilance state before and after the task completion across all conditions. The 9-point KSS was used: 1=very alert, 3=alert, 5=neither alert nor sleepy, 7=sleepy (but not fighting sleep), 9=very sleepy (fighting sleep). Sleepiness was subjectively rated before and after each driving scenario and before and after the neuropsychological tests were carried out. Higher scoring meant less vigilance and subsequently increased sleepiness. In addition, participants rated how much they had to try in order to complete the driving tasks [Rating Scale of Mental Effort (0-150); RSME] (Zijlstra, 1993) and how well they thought they performed in the driving tasks compared to their everyday driving experience (Driving Quality Scale, Brookhuis et al., 1985). In the present

77 DRUID 6th framework programme Deliverable 1.2.2 study, the original subjective scales were translated into Greek and back-translated by an independent professional and verified by a native English speaker.

Statistics

Within and between participants comparisons (superiority tests) were carried out with repeated measures of GLM and one-way ANOVAs. In case of violation of homogeneity and homoscedacity assumptions, non- parametric equivalents were administered (Friedman and Wilcoxon rank test, respectively). Within comparisons were carried out in order to investigate the effect of alprazolam (0.5mg) administration. Between comparisons were carried out in order to investigate difference between probable additive effects and acute effects. Moreover, the relation between alprazolam and alcohol effect was investigated. The alcohol effect was based on the alcohol criterion data collected in the CPAP study (i.e. Chapter 4). The Į level was set at .05. Statistical analyses were performed with the statistical programme Statistical Package for the Social Sciences (SPSS) (version 18.0 for Windows; SPSS, Chicago, IL).

78 DRUID 6th framework programme Deliverable 1.2.2 Results

This section is divided in 4 parts according to the types of collected data. Firstly, the main primary driving parameters are presented per driving scenario (i.e. lane tracking and car following). Secondly, the analyses of additional neuropsychological measures are presented. Finally, subjective scales’results are provided. Tables 4 to 6 present within participants’comparisons of the vehicle parameters for both driving scenarios and subjective assessments. The findings are discussed in the respective sections.

Table 4: Mean (±SE) scores and 95% confidence intervals of driving measures and subjective assessments for the treated anxiety patients TREATED ANXIETY PATIENTS Baseline Alprazolam Driving Significance Mean±SE [95% CI] Mean±SE [95% CI] measures Lane Tracking SDLP(m) 0.30±0.01 (0.29 – 0.32) 0.36±.02 (0.33 – 0.39) * Car Following TTC0-1 (sec) 6.46±0.21 (6.02 – 6.90) 7.05±0.20 (6.61 – 7.48) * TTC1-1.5 (sec) 8.09±0.24 (7.57 – 8.61) 10.03±0.21 (9.59 – 10.48) *

TTC1.5-2 (sec) 13.05±0.52 (11.93 – 14.17) 16.50±0.52 (15.40 – 17.61) *

TTC2-4 (sec) 20.08±0.65 (18.69 – 21.47) 24.09±0.69 (22.62 – 25.57) *

TTC>4 (sec) 52.32±0.94 (50.31 – 54.34) 42.32±0.89 (40.41 – 44.24) * BRT (msec) 0.92±0.43 (0.83 – 1.01) 0.97±0.05 (0.87 – 1.08) * Subjective assessments Mental Effort 34.37±5.13 (23.37 – 45.37) 34.7±4.37 (25.33 – 44.07) NS Driving quality 6.51±6.91 (-8.30 – 21.32) 4.83±6.06 (-8.18 – 17.83) NS Scale KSS 3.44±0.33 (2.75 – 4.14) 3.99±0.35 (3.24 – 4.74) NS

* Significant (p<.05); NS: Non-significant (p>.05).

Table 5: Mean (±SE) scores and 95% confidence intervals of driving measures and subjective assessments for the untreated anxiety patients UNTREATED ANXIETY PATIENTS Baseline Alprazolam Driving Significance Mean±SE [95% CI] Mean±SE [95% CI] measures Lane Tracking SDLP(m) 0.27±.03 (0.29 –0.30) 0.31±.01 (0.29 –0.33) * Car Following TTC0-1 (sec) 7.43±0.21 (7.00 – 7.87) 8.5±0.33 (7.81 – 9.20) * TTC1-1.5 (sec) 12.06±0.44 (11.89–12.94) 13.51±0.49 (12.47 –14.55) * TTC1.5-2 (sec) 14.65±0.49 (13.63–15.67) 16.41±0.46 (15.43 – 17.39) * TTC2-4 (sec) 24.06±0.99 (21.98–26.14) 26.64±0.69 (25.19 –28.09) * TTC>4 (sec) 41.80±1.56 (38.51–45.09) 34.94±1.54 (31.70 – 38.18) * BRT (msec) 0.85±0.02 (0.81 –0.90) 0.95±0.02 (0.91 –1.0) * Subjective assessments Mental Effort 33.30±2.99 (27.04 –39.58) 49.50±6.69 (35.38 –63.62) NS Driving -2.43±5.02 (-13.02 – 8.15) -19.13±5.96 (31.7– (-6.56)) * quality Scale KSS 3.44±0.24 (2.94 –3.94) 5.68±0.48 (4.67 –6.68) *

* Significant (p<.05); NS: Non-significant (p>.05).

79 DRUID 6th framework programme Deliverable 1.2.2 Table 6: Mean (±SE) scores and 95% confidence intervals of driving measures and subjective assessments for the control group.

CONTROL Baseline Alprazolam Driving Significance Mean±SE [95% CI] Mean±SE [95% CI] measures Lane Tracking SDLP(m) 0.26±.01 (0.24 –0.28) 0.33±.01 (0.31 –0.35) * Car Following TTC0-1 (sec) 4.49±0.22 (4.03 – 4.94) 7.51±0.30 (6.88 – 8.13) * TTC1-1.5 (sec) 7.06±0.22 (6.60 –7.52) 9.50±0.37 (8.72 –10.28) * TTC1.5-2 (sec) 8.41±0.39 (7.61 –9.20) 13.54±0.33 (12.85 –14.22) * TTC2-4 (sec) 38.09±1.09 (35.78 – 40.40) 34.05±0.89 (32.18 – 35.92) * TTC >4 (sec) 41.96±1.17 (39.48 – 44.44) 35.41±1.21 (32.86 – 37.95) * BRT (msec) 0.81±0.03 (0.76 –0.87) 0.84±0.03 (0.77 –0.91) NS Subjective assessments Mental Effort 46.67±3.80 (38.65– 54.70) 55.31±4.68 (45.43 – 65.20) NS Driving -6.50±3.17 (-13.19– 0.19) -17.54±7.09 (-32.50 – 2.59) NS quality KSS 3.90±0.34 (3.18–4.61) 6.07±0.29 (5.45–6.68) * * Significant (p<.05); NS: Non-significant (p>.05).

Driving parameters (primary variables)

Lane tracking It is evident from tables 4-6 that alprazolam intake impaired driving performance in all groups. In particular, increased weaving (SDLP= 5.8 cm) has been found in treated anxiety patients when driving an hour after alprazolam intake (F (1, 14) =11.31, p=.005). Similarly, untreated anxiety patients showed a significant increase of 4 cm after alprazolam intake (F (1, 17) =5.28, p=.035). On the same track, healthy participants showed the greatest increase in weaving (ǻSDLP=6.8 cm; F (1, 17) =36.34, p<.001). The following graph (figure 2) presents the percentages (%) of impaired/improved driving performance as a function of alprazolam serum concentration levels (ng/mL). It appears that alprazolam is associated with impairment in lateral position keeping even in low concentrations.

Figure 2: Percentage of impairment/improvement as a function of serum concentration

80 DRUID 6th framework programme Deliverable 1.2.2 (ng/mL) for the control group

For the untreated anxiety patients only two concentration levels were detected and impairment in lateral position keeping was observed.

Figure 3: Percentage of impairment/improvement as a function of serum concentration (ng/mL) for the untreated anxiety patients’group

Higher impairment was found for the treated anxiety group for serum concentration groups of 5-10 and >15 ng/mL. Almost in all cases impairment was observed for the other two groups.

Figure 4: Percentage of impairment/improvement as a function of serum concentration (ng/mL) for the treated anxiety patients’group

The following graph depicts the comparisons among the three experimental groups in lateral position change

81 DRUID 6th framework programme Deliverable 1.2.2 ǻSDLP). The change (i.e. simple effect size) in lateral position was compared with alcohol consumption effect in weaving measured in the alcohol calibration study conducted within the framework of DRUID project (see Chapter 4). Pairwise comparisons were conducted and a significant mean difference was found between the mean change in ǻSDLP in the control group after alprazolam intake (0.068±0.0112 m) and the control group after alcohol consumption (0.05%) (0.024±0.007) (p=.002). Moreover, a trend appears for the difference between the increase in weaving due to the acute effect of alprazolam in treated anxiety patients (0.058±0.017) and the deterioration in weaving because of alcohol consumption (0.024±0.007) (p=.061). This probably explains the inclusion of the alcohol criterion in the lower confidence limit (95% CI) (Figure 5) Equivalence is shown for untreated patients and alcohol. Additionally, equivalence testing showed that there is borderline equivalence for alcohol and alprazolam intake for treated patients (additive effect) (Figure 5).

Figure 5: Mean change of ǻSDLP (m) across groups including the alcohol criterion (0.05%)

Car following

Percentage (%) of time spent driving within Time-To-Collision (TTC) categories

The findings from TTC comparisons among conditions are described separately for each chosen TTC category. Within and between comparisons were carried out for each TTC category.

· Percentage (%) of time with TTC 0-1

As shown in tables 4-6, the percentage of time (%) spent driving with TTC values between 0 and 1,

82 DRUID 6th framework programme Deliverable 1.2.2 increased significantly for all groups (F (3,65)=20.65, p<.05). Pairwise comparisons (Bonferroni adjusted) revealed significantly greater percentage of time spent between 0 and 1 TTC values for untreated compared to treated patients (ǻ=1.46%, p=.003) in the alprazolam condition. As shown in the graph below, control participants were more affected by the alprazolam intake when compared to the alcohol effect (p<.001). Equivalence is shown for the treated and untreated anxiety patients.

Figure 6: Mean change (ǻ) in the percentage (%) of time driven with TTC 0-1 (sec) across group including the alcohol criterion (0.05%)

· Percentage (%) of time with TTC1-1.5

Likewise, percentage of time driven with TTC values between 1 and 1.5 increased significantly after alprazolam intake for all groups with greater increase observed in the control group (ǻ=2.45%, p<.001). Following between groups’pairwise comparisons revealed that untreated patients spent significantly more time driving with TTC values between 1 and 1.5 compared to both treated (ǻ=3.47%, p<.001) and healthy participants (ǻ=4%, p<.001) after alprazolam intake. No significant difference was found between treated patients and the control group (p>.05). Both control groups after alprazolam and after alcohol consumption showed increase in dangerous close following behaviour which is of significant equivalence as shown in Figure 7. However, untreated patients’ increase in time driven with TTC between 1 and 1.5 sec because of alprazolam intake is significantly less than the time increase because of alcohol consumption in the control group (p=.043).

83 DRUID 6th framework programme Deliverable 1.2.2 Figure 7: Mean change (ǻ) in the % of time driven with TTC 1-1.5 (sec) across group including the alcohol criterion (0.05%)

· Percentage of time with TTC 1.5-2

The percentage of time spent driving with TTC values between 1.5 and 2 seconds increased significantly for all groups as shown in tables 4 to 6. Pairwise comparisons revealed that healthy participants spent significantly less time with TTC values between 1.5 and 2 compared to treated (ǻ=2.97%, p<.001) and untreated patients (ǻ=2.87%, p<.001). The change in percentage of time (%) spent driving with TTC values between 1.5 and 2 seconds was significantly less after alcohol consumption compared to the change in treated, untreated patients and control participants (p<.001, p=.003, p<.001, respectively).

· Percentage of time with TTC 2-4

Significant increase in time spent driving with TTC values between 2 and 4 seconds (safety distance) was observed in the treated and untreated groups. On the other hand, significant decrease was observed for the control group. Healthy participants spent significantly more time with TTC values between 2 and 4 seconds than treated (ǻ=9.96%, p<.001) and untreated patients (ǻ=7.41%, p<.001) after alprazolam intake. Overall following behaviour was impaired as alprazolam serum concentration increased (Figure 8).

84 DRUID 6th framework programme Deliverable 1.2.2 Figure 8: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Control group -TTC 2-4)

On the other hand, untreated patients improved significantly in distance keeping behaviour as shown in figure 9.

Figure 9: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Untreated group -TTC 2-4)

85 DRUID 6th framework programme Deliverable 1.2.2 Figure 10: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Treated group -TTC 2-4)

Treated patients distance keeping behaviour showed only improvement after alprazolam intake (Figure 10). Participants spent significantly less time driving with TTC values between 2 and 4 seconds when they had consumed alcohol compared to the change due to alprazolam intake in both treated and untreated anxiety patients (p<.001). No difference was found for the comparison with the control group (alprazolam) but, also, no equivalence was shown.

· Percentage (%) of time with TTC>4

The time spent with TTC values higher than 4 seconds decreased significantly for all groups after alprazolam intake as shown in tables 4 to 6. Further pairwise comparisons revealed significantly greater decrease in percentage of time spent with TTC values greater than 4 seconds in treated patients (ǻ=7.38%, p<.001) compared to untreated and control groups (ǻ=6.96%, p<.001). However, variations were extremely large in this last TTC category; therefore results may not be reliable. Alprazolam intake significantly decreased the time spent with TTC values above 4 seconds in the treated, untreated, and control groups when compared to the alcohol consumption condition (p<.001). The following graph presents the distribution of percentages (%) within and between categories in order to get an overview of distance keeping behaviour. Participants from the control group (baseline condition) spent the most time with safe distance keeping and overall spent less time with dangerously low TTCs.

86 DRUID 6th framework programme Deliverable 1.2.2 TB: Treated Baseline TA: Treated Alprazolam UB: Untreated Baseline UA: Untreated Alprazolam CB: Control Baseline CA: Control Alprazolam

Figure 11: Percentages (%) of driven time per TTC category as a function of the experimental conditions

Brake reaction time (sec)

As shown in tables 4 to 6 significant increase in brake reaction time is observed in treated and untreated patient groups but not in control group. Pairwise comparisons showed trends towards significant differences between treated, untreated and control group (p=0.028 and p=0.056, respectively).2 The treated group showed the greatest deterioration in brake reaction time as a function of serum concentration as it is evident from graph 14. With increasing concentration, impairment increased as well. Impairment in brake reaction time is evident in most groups.

2 Bonferroni adjustment leads to lower p-values 87 DRUID 6th framework programme Deliverable 1.2.2 Figure 12: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Control group -BRT)

Figure 13: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Untreated group -BRT)

88 DRUID 6th framework programme Deliverable 1.2.2 Figure 14: Percentage of improvement/impairment as a function of serum concentration (ng/mL) (Treated group -BRT)

Following comparisons between the decrease in BRT in the experimental and alcohol group (p>.05) were of no statistical significance. In spite of non significant results, no equivalence to alcohol was found. This may be the result of great variation in braking reaction time for the alcohol consumption group (0.222±0.501).

Secondary variables (Neuropsychological measures)

The secondary measurements included in this section are the results from the neuropsychological tests of the WinTAP battery. Alertness was significantly decreased in healthy participants (p=.015). No other significant difference was found in both within and between comparisons (p>.05).

89 DRUID 6th framework programme Deliverable 1.2.2 Figure 15: Mean reaction time (msec) in alertness test

Subjective scales

As shown in Tables 4 to 6 non significant differences in sleepiness, mental effort and driving quality were found for the treated patients (p>.05). Untreated patients felt significantly less vigilant and believed they drove really badly in alprazolam condition (p=.001 and p=.021, respectively). Similarly, healthy participants felt significantly less vigilant after alprazolam intake (p=.018).

90 DRUID 6th framework programme Deliverable 1.2.2 Discussion

The findings of this study support the main hypothesis that alprazolam will impair driving performance in all three groups. Indeed significant increase in weaving (SDLP) was found in all groups after alprazolam administration. Specifically, alprazolam’s detrimental effects were evident in weaving in all groups with higher lateral deviation in the control group by more than 6 cm. Therefore vehicle lateral control is affected by alprazolam intake. Relevant literature is in agreement with these findings. Alprazolam intake deteriorates lateral control but acute effect (Curran, 1986; Leufjkens et al., 2007) is greater than chronic administration due to tolerance to the sedating effects because of repeated use, as it seems to be the case with lateral deviation in the treated anxiety patients’group in this study. Likewise, most alprazolam studies have found high increaments of SDLP. Verster and colleagues (2002) observed increaments of SDLP of approximately 9 cm which they compared to driving impairment with a blood alcohol concentration of 0.15% and stated that BAC above 0.15% corresponds to a 25 times increase of accident risk (Louwerens et al., 1986; Borkenstein et al, 1964). Similarly, in a review on effect of antidepressants on driving tests’changes it was found that changes in SDLP after acute doses of sedating effects of antidepressants were comparable to blood alcohol concentrations of 0.8 mg/mL (Ramaekers, 2003). In the present study, we found supportive evidence for alprazolam effect; however, not of such increment. Deterioration in weaving because of alcohol consumption was found to be equivalent to alprazolam effect in treated and untreated patients and significantly less compared to alprazolam effect to the control group. Therefore, the acute effect of alprazolam in anxiety patients may be comparable to alcohol BAC=.05% effect. However, the detrimental effects to healthy individuals could be of great interest for future research efforts on recreational/occasional use of alprazolam by drivers. The findings indicate probability for higher accident risk but there are no simulator studies to directly compare the findings. Moreover, lane deviations in real traffic deviate from corresponding measurements in a simulated environment. Thus ramification and extrapolation of results should be made with this difference taken into consideration. Further simulator studies could add upon these findings in order to provide a base of comparability for these increments.

Respective between groups’comparisons showed no significant differences among groups after alprazolam oral administration (p<.05). Therefore impairment may be comparable among groups as the difference among groups was approximately around 2-2.5 cm. However, according to figure 4, only in treated patients’ group a percentage of improvement was observed. The other two groups had almost solely impaired weaving. This improvement may depict (overall §20% improvement) tolerance. Non-sedative antidepressants were found not to affect SDLP values (Ramaekers, 2003). Therefore a bottom up thinking process could lead researchers to focus on the same sedating effect in all groups in an attempt to understand the non-significant differences among groups. This research idea could be investigated in another experiment as inferences from studies on antidepressants to experimental findings of benzodiazepines have less validity. Similarly, statistically significant differences were found in percentage of time spent across TTC categories for all groups. The time spent within each category increased after alprazolam intake, except for really high

91 DRUID 6th framework programme Deliverable 1.2.2 values for which following the leading vehicle ceased. In general, safety keeping behaviour was affected. For treated and untreated anxiety patients the time spent in risky following behaviour increased but the time of safe following was increased, as well. On the contrary, the control became riskier overall, i.e. they spent more time dangerously following the lead vehicle and less time in safety distance keeping. These findings are difficult to interpret especially under the prism of clear and specific guidelines given to all participants to keep a safety distance from the leading vehicle. Regarding extremely dangerous close following the picture is similar for all participants; they became riskier. However, with regards to safety distance only the control group showed decrease after alprazolam intake. These findings cannot lead to generalisations about protective or impairing role of alprazolam in safety distance keeping behaviour. The findings derived by the car following scenario are inconclusive. Figures 8 to 10 reflect the aforementioned findings as they show that alprazolam improves following behaviour for anxiety patients (regardless if treated or not) and clearly impairs distance keeping in healthy participants. Furthermore, extremely risky close following (% of time spent driving with TTC values between 0 and 1 second) of a leading vehicle has been shown to be equivalent, again, for anxiety patients and alcohol consumption participants. It seems that alprazolam effect was more detrimental for the healthy controls as they spent ever more time driving with very low TTC values. It seems that alprazolam intake “made”controls more prone to riskier close following and less safe in distance keeping from the leading vehicle. This effect disappeared for higher TTC values but driving still remained very risky for the control group and it was transformed into equivalent to the alcohol consumption group. In other words, healthy participants who consumed alcohol or alprazolam intake increased the time spent driving with risky TTC values. Regarding safe keeping behaviour, alcohol was shown to make healthy controls significantly less “conforming”when compared to anxiety patients with alprazolam intake. However, differences in safe distance keeping were not found for the control group who received alprazolam when compared to alcohol. Hence, the effect of alcohol might be more deteriorating in safety keeping behaviour than alprazolam in patients, probably because of inhibition alleviation. Furthermore, alprazolam affected brake reaction time in treated and untreated anxiety patients. As treated group showed the greatest deterioration in reaction time (sec), additive effect of alprazolam intake may be greater than acute for reaction time. It appears that additive effects are more powerful and treated anxiety patients do not show tolerance effects in psychomotor tests or psychomotor related driving skills, without of course, isolating them from the rest of parameters (i.e. tolerance might not be revealed for other variables). Subjective assessments of treated anxiety patients are in favour of the tolerance proposition as there was no significant difference in their evaluation of sleepiness and driving quality during driving scenarios in the simulated tasks. On the contrary, untreated and healthy participants reported increased sleepiness after alprazolam administration which reflects their driving performance. In addition, untreated patients reported that they thought they drove really badly, although greater deviations were recorded for the control group. It is important to keep in mind, though, that anxiety patients are overly self-conscious, pay high self-attention after the activity, or have high performance standards for themselves. It should be borne in mind that anxiety patients’ subjective assessments of their driving behaviour may be influenced, also, by their symptomatology. An overall depiction of the car following pattern for all groups and conditions is given in figure 11. The less “safety distance keeping”behaviour is observed for treated anxiety patients after alprazolam intake. For that

92 DRUID 6th framework programme Deliverable 1.2.2 reason this behaviour may result from additive effect were tolerance does not show. It was expected healthy participants to show safer keeping distance behaviour and they indeed kept safety distance for the longer period of time during the driving task. These findings suggest that alprazolam may not only affect alertness and psychomotor functions but, in addition, decision making, similar to inhibitions’alleviation as alcohol does in small amounts. Alprazolam significantly affected alertness in the control group but no other significant impairment was observed for the neuropsychological tests. Relevant studies have found differences. The greater effect was the acute effect in healthy individuals. It is important to note that this battery is standardised to driving behaviour but has not been applied in drug related research before. Subjective scales confirmed the effect perceived by participants. Treated patients did not perceive any difference in vigilance, mental effort, and driving quality. On the contrary, the other two groups-not used in alprazolam medication-reported that they felt significantly less vigilant (untreated and control) and that they drove badly (untreated). It is alarming that treated patients did not report any difference or change in the way they drive which might imply that their everyday driving performance is affected and they may not be aware of it and their risk of accidents due to lack of awareness may be increased. The effect of alprazolam in healthy participants was stronger than in treated and untreated patients. Alprazolam intake (0/5 mg) might improve the driving performance of anxiety patients but might have deteriorative effect in healthy controls’driving performance. This study clearly showed an alprazolam effect (0.5mg) in driving performance in treated, untreated anxiety patients and healthy participants. Future studies might focus on dose-related relationships for the same groups. Additionally, comparisons with other medicinal drugs would be desirable reflecting the real medicinal drug users and, consequently, the real driving performance. Conclusively, the main findings of this study are in agreement with current research that people under alprazolam medication should be informed about the potential detrimental effects of alprazolam administration to their everyday activities and driving. Likewise, physicians and medical practitioners should be educated and trained on how the adverse effects of alprazolam prescriptions may affect driving performance.

Acknowledgements

This work was conducted with the assistance of two teams of specialists led by Dr. Chrysoula Papadeli and Prof. Pavlidis. We would like to thank them for being attentive and patient in the recruitment, conduction process and debriefing. We are grateful to Dr. Gisela Skopp and Dr. Ricarda Jantos for the analysis and extraction of alprazolam samples and the respective description of the procedure included in this report.

93 DRUID 6th framework programme Deliverable 1.2.2 References

Åkerstedt, T. and Gillberg, M. (1990). Subjective and objective sleepiness in the active individual. International Journal of Neuroscience, 52, 29–37. Borkenstein, R.F. (1974) The role of the drinking driver in traffic accidents, the Grand Rapids study. Blutalkohol 11. Brookhuis, K.A., DeVries, G., Prins Van Wijngaarden, P., Veenstra, G., Hommes, M., Louwerens, J.W., O’Hanlon, J.F. (1985). The effects of increasing doses of Meptazisol (100, 200, 400 mg) and Glaferine (200mg) on driving performance. Report VK 85-16. Haren, The Nethrlands: Traffic Research Center, University of Groningen. Busto, U.E., Kaplan, H.L., Wright, C.E., Gomez-Mancilla, B., Zawertailo, L., Greenblatt, D.J., Seller, E.M. (2000) A comparative pharmacokinetic and dynamic evaluation of alprazolam sustained-release, bromazepam, and lorazepam. Journal of Clinical Psychopharmacology, 20, 628–635. Curran, H.V. (1986) Tranquillising memories: a review of the effects of benzodiazepines on human memory. Biological Psychology, 23, 179–213. de Gier, J.J., Hart, B.J., Nelemans, F.A. The effects of lorazepam and bromazepam on actual driving and psychomotor performance of patients. In: de Geir J.J., ed. Drugs and driving. London: Taylor and Francis, 1986:137 – 52. Ellinwood, E.H., Jr, Heatherly, D.G., Nikaido, A.M., Bjornsson, T.D., Kilts, C. (1985). Comparative pharmacokinetics and pharmacodynamics of lorazepam, alprazolam and diazepam. Psychopharmacology, 86, 392–399. European Monitoring Centre for Drugs and Drug Addiction. Literature review on the relation between drug use, impaired driving and traffic accidents (CT.97.EP.14). Lisbon: EMCDDA, 1999. Hamilton M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50–55. Kroboth, P.D., Folan, M.M., Bauer, K.S., Tullock, W., Wright, C.E., Sweeney, J.A. (1998) Do alprazolam- induced changes in saccadic eye movement and psychomotor function follow the same time course? Journal of Clinical Pharmacology, 38, 337–346. Leufkens, T.R.M., Vermeeren, A., Smink, B.E., van Ruitenbeek, P., Ramaekers, J.G. (2007). Cognitive, Psychomotor and actual driving performance in healthy volunteers after immediate and extended release formulations of alprazolam 1 mg. Psychopharmacology, 191, 951-959. Longo, M.C., Lokan, R.J., White, J.M.(2001). The relationship between blood benzodiazepine concentration and vehicle crash culpability. Journal of Traffic Medicine, 29, 36 – 43. Mortimer, R.G., Howat, P.A. (1986). Effects of alcohol and diazepam, slightly and in combination, on some aspects of driving performance. In: de Geir J.J., ed. Drugs and driving. London: Taylor and Francis. Petty, P., Madhukar, H., Trivedi, Fulton, A. M., Rush, J. (1995). Benzodiazepines as antidepressants: Does GABA play a role in depression? Biological Psychiatry, 38, (9), 578-591. Ramaekers, J.G. (2003). Antidepressants and driver impairment: empirical evidence from a standard on-the- road test, Journal of Clinical Psychiatry, 64, 20–29. Seppala, T, Mattila, M.J., Palva, E.S., Aranko, K. (1986) Combined effects of anxiolytics and alcohol on psychomotor performance in young and middle-aged subjects. In: de Geir J.J., ed. Drugs and driving. London: Taylor and Francis, 179 – 90. Smiley, A., Moskowitz, H.M., Ziedman, K. (1985). Effects of drugs on driving: driving simulator tests of

94 DRUID 6th framework programme Deliverable 1.2.2 secobarbital, diazepam, marijuana, and alcohol. Los Angeles: National Institute on Drug Abuse. Snyder, P.J., Werth, J., Giordani, B., Caveney, A.F., Feltner, D., Maruff, P. (2005). A method for determining the magnitude of change across different cognitive functions in clinical trials: the effects of acute administration of two different doses alprazolam. Human Psychopharmacology: Clinical and Experimental, 20, (4), 263-273. Vermeeren, A., Jackson, J.L., Muntjewerff, N.D., Quint, .PJ., Harrison, E.M., O’Hanlon, J.F. (1995) Comparison of acute alprazolam (0.25, 0.50 and 1.0 mg) effects versus those of lorazepam 2 mg and placebo on memory in healthy volunteers using laboratory and telephone tests. Psychopharmacology, 118,1–9. Verster, J.C., Volkerts, E.R. (2004) Clinical pharmacology, clinical efficacy, and behavioral toxicity of alprazolam: a review of the literature. CNS Drug Review, 10, 45–76. Verster, J.C., Volkerts, E.R., Verbaten, M.N. (2002). Effects of alprazolam on driving ability, memory functioning and psychomotor performance: a randomized, placebo-controlled study. Neuropsychopharmacology, 27, 260–269. Verster, J.C., Veldhuijzen, D.S., Volkerts, E.R. (2005) Is it safe to drive a car when treated with anxiolytics? Evidence from on-the-road driving studies during normal traffic. Current Psychiatry Review, 1, 215- 225. Zijlstra, F.R.H. (1993). Efficiency in work behavior. A design approach for modern tools. PhD thesis, Delft University of Technology. Delft, The Netherlands: Delft University Press. Zimmermann, P. & Fimm, B. (1993). Testbatterie zur Erfassung von Aufmerksamkeitsstörungen. Version 1.02. Freiburg: Psytest.

95 DRUID 6th framework programme Deliverable 1.2.2 Chapter 4: Daytime driving in treated (CPAP) and untreated sleep apnoea patients

Katerina Touliou, CERTH/HIT, Greece

96 DRUID 6th framework programme Deliverable 1.2.2 Abstract

Obstructive Sleep Apnoea Syndrome (OSAS) has been shown to be an increased risk factor for traffic accidents. Studies performed in driving simulators have reported improvements in driving performance with the application of Continuous Positive Airway Pressure (CPAP) treatment. In the present study, 18 OSAS patients were treated with CPAP treatment for 7 consecutive days. The experiment involved both baseline and after CPAP condition. A second group of healthy participants (N=18) was included in the study with a baseline and an alcohol consumption (BAC=0.05%) condition. All participants drove two scenarios in a simulated environment; a lane tracking and a car following scenario. No improvement due to CPAP treatment was found (p>.05) in weaving control (SDLP) for the lane tracking scenario. On the contrary, statistically significant impairment was found in SDLP due to alcohol (p=.027). Percentage (%) of time spent driven in clustered TTC categories (0-1,1-1.5,1.5-2,2-4,>4) was calculated. Treated OSAS patients spent significantly more time with safe keeping distance than untreated patients (p<.001). Likewise, intoxicated participants spent significantly less time with safe distance from lead vehicle in car following scenario (p=.008). Equivalence in impairment level was found in BRT (sec) for the car following scenario between OSAS and alcohol. Similarly, alcohol consumption significantly decreased alertness in healthy participants. Intoxicated participants were statistically significant less alert than sober (p=.007) and showed increased reaction times (msec) compared to OSAS untreated patients in a divided attention task. OSAS untreated patients reported higher sleepiness but, also, better quality of driving in general (self-assessment scales). Consequently, OSAS patients probably showed less awareness of impairment in driving performance which makes driving even more risky. In conclusion, the effect of sleep apnoea appears to be detrimental compared to the alcohol effect at 0.05%. The application of intermediate alcohol BAC levels (i.e. 0.02, 0.08., 0.1) could provide insight in finding comparable levels of impairment. Probably higher levels of alcohol levels are necessary to be included in a future research effort for the chosen types of driving parameters in order to perform comparisons to effects in driving fitness due to Obstructive Sleep Apnoea Syndrome.

97 DRUID 6th framework programme Deliverable 1.2.2 Introduction

A large number of subjective (i.e. self-report) and objective (e.g., insurance or police records) studies have looked at the prevalence of MVAs for patients suffering from Obstructive Sleep Apnoea Syndrome (OSAS) as compared to the general population (e.g., George, 1995; Maycock, 1996; Wu & Yan-Go, 1996). The majority of these studies have suggested that OSAS presents an increased risk factor for Motor Vehicle Accidents (MVAs). However, the results obtained from these studies have been criticised due to the various confounding variables that were not controlled for (e.g. gender, age, driving experience, etc.) and because of other possible biases that could have affected the results (i.e. informational, recall, and selection bias; cf., George, 2004). In order to obtain a more accurate measure regarding the risk of MVAs for OSAS patients as compared to healthy participants, researchers have turned to experimental testing using various types of off-road driving simulators. Results from these studies have shown that patients with OSAS have an increased accident rate in driving simulation tests (e.g. Findley et al., 1989, 1995; George, Boudreau, & Smiley, 1996; Juniper et al., 2000; Mazza et al., 2005), estimated to be around two to seven-times higher compared to healthy participants (e.g., Horne & Reyner, 1999; George, 2004; George et al., 1987; Findley et al., 1989). It has also been reported that OSAS patients exhibit slower reaction times than controls in road obstacle avoidance, resulting in four times more object collisions than normals (Findley et al., 1989). In addition, it has been demonstrated that OSAS patients perform poorer than controls in steering ability (referred to as “tracking error”; George et al., 1996; Juniper et al., 2000), with half of the patients being worse than any one control participant, and with some patients showing worse performance than healthy controls under the influence of alcohol (George et al., 1996). Finally, research conducted to-date have concluded that the OSAS patients face an increased difficulty in sustaining attention while driving, thus exhibiting poorer performance and lower vigilance during experimental testing when driving on a monotonous highway route (e.g. George, Boudreau, & Smiley, 1996; Juniper et al., 2000; Turkington et al., 2001). Generally, research findings strongly suggest that patients suffering from OSAS have a higher risk of having MVAs as compared to their healthy counterparts. If sleep apnoea patients are prone to traffic accidents, how fit are they to drive? Given that driving is an essential part of everyday life for the majority of people, a series of treatments have been developed, in order to assist OSAS patients in driving and other daily activities. Continuous Positive Airway Pressure (CPAP) represents the most commonly used treatment and it is considered to be the most effective one (Cassel et al., 1996; Yamamoto et al., 2000). Studies have shown that CPAP treatment can reduce the number of accidents in patients with OSAS, both in simulated driving (Findley et al., 1989) and in real-life situations (e.g. Cassel et al., 1996; Findley et al., 2000; George, Boudreau, & Smiley, 1997; Yamamoto et al., 2000). Specifically, studies have shown that regular use of CPAP improves self-reported (Cassel et al., 1996; Yamamoto et al., 2000) and objective MVA rates (Findley et al. 2000; George, 2001). Relevant studies have shown that CPAP treatment may effectively reduce the MVA risk of OSAS patients in experimental tests conducted in a simulated driving setting (e.g. Engleman et al., 1994; Findley et al., 1989; Note, however, that the task utilised in some of these studies was actually a choice reaction task that required sustained vigilance rather than a simulated driving task). If OSAS is a significant contributing factor to MVAs, then it affects both individual and public safety.

98 DRUID 6th framework programme Deliverable 1.2.2 Treatment of OSAS using CPAP can improve patient’s driving performance and, thus, reduce the risk of MVAs. The findings of previous studies, however, are inconclusive due to the several limitations that can be identified in their experimental design. For instance, the majority of these studies have predominantly used male patients as their participants (except in the case of Turkington et al., 2004, where gender was not reported), thus generalisability of findings to general OSAS population is difficult to be attained. Male patients are over-represented in OSAS population compared to female. One should also note that the majority of these studies have utilised only a single, brief practice session with a simulated driving task, which may have not allowed patients to fully understand the task and get familiarised with driving in a simulated environment. Another important limitation that can be identified in previous research is the issue of circadian rhythms influence on participants’performance. For example, in Orth and colleagues (2005) study, it was reported that testing was completed over different time periods (i.e. some tests took place in the morning and other at night) with a number of patients and healthy participants performing similarly in the driving task. Similalry, Turkinston and colleagues (2004) did not include a control group in their study, thus it was not possible to compare patients’performance to control participants in order to investigate whether it was impaired or not. Finally, a major limitation noted in past research concerns the choice of experimental methodology that may inadvertently have led to inconclusive findings. For instance, Orth and colleagues (2005) reported null results in participants’ vigilance testing, while the opposite was reported in the Cassel et al. Study. Therefore, the choice of the attentional measure used could have been not suitable or not comparable. Similarly, Turkington et al.’s choice of very brief post- and pre- CPAP treatment time measurements could have failed to provide representative results of the CPAP effectiveness and therapeutic time course. Last but not least, Orth and colleagues (2005) utilised a simulator setting that required manual scoring by a technician, which could have led to possible recording errors, while the majority of simulated driving studies have only focussed on the participants tracking error measurements, while ignoring the possible wide range of impairments that could be present in OSAS patients. On the whole, research to-date identifies OSAS as a significant risk factor to MVAs and supports that treatment of OSAS using CPAP can reduce this risk. As yet, however, there are no generally accepted regulations within the Europe Union concerning driving licensing and OSAS. Therefore, it seems pertinent that these regulations are established both for the benefit of the OSAS patients and the safety of all road users. Alcohol remains the greatest documented risk factor in driving performance and the literature is vast on alcohol effects on fitness to drive. Alcohol is the only substance affecting driving behaviour that legal limits apply. Alcohol-impaired driving is a major cause of serious and fatal car accidents. Drunk drivers are clearly over-represented in road traffic crashes. In Germany, for example, the severity of drink-drive crashes (expressed as fatalities per 1,000 injury crashes) is nearly twice as high as that of crashes in general (Swendler et al., 2004). The relative crash rate for a driver with a BAC of 1.5 g/l is about 22, but drivers’ relative crash rate for fatal crashes with that amount of alcohol in their blood is about 200 (Simpson & Mayhew, 1991). Individual differences play a sizeable role in the elimination of alcohol from the human organism. Difference in accident risks is, also, an outcome of causation. The driving behaviour theoretical perspectives may be clustered into major theoretical grounds; external and internal. Internal retribution is the research focus of this paper. The next level is a hierarchical distinction

99 DRUID 6th framework programme Deliverable 1.2.2 about the complexity of involved processes; ranging from low-level (e.g. steering) to high level processes (e.g. speed and route choice). Most theoretical perspectives focus on levels of control and information processing. Driving is a demanding, complex and multi-aspect task encompassing various skills and their interactions, such as mental alertness, dexterity, eye-hand coordination, automotised and not , and perceptual skills (from visual and auditory to sharp decision making) (Weiler et al., 2000). Drink related driving impairment is well documented in literature and quantified in thresholds or legal limits. The legal limit in Greece is Blood Alcohol Concentration of 0.05%. A discussion on zero tolerance is beyond the scope of this paper, but interest is shifted towards types of skills impaired and/or affected with taking into account assumptions of inference (i.e. tasks that would manifest themselves as automatic or more complex). Alcohol is classified as a depressant, due to its effects to the central nervous system (CNS). Existing diversity in findings across studies leads to no consensus on the effects on driving impairment in performance by a given amount of alcohol (34% of studies report impairment by .05%). Current research techniques have revealed deterioration in driving performance at lower BAC levels. However, Moskowitz and Robinson (1987) reported in their review that impairment was recorded in psychomotor tasks at a level of 0.07%. In addition, simple reaction time score (RT) was found to be an unreliable and insensitive measure. On the contrary, tracking and divided attention tasks were shown to be impaired at much lower levels (0.01-0.02%). In most studies deterioration is present above 0.08%.

In the present study, the research team aimed to investigate the level of risk for MVAs for OSAS patients, while controlling for some of the previous limitations, and to measure whether or not CPAP treatment can significantly reduce this risk. In order to accomplish this, the researchers compared the performance levels of CPAP-treated OSAS patients, untreated OSAS patients, and healthy control participants in a series of simulated driving and laboratory cognitive tasks. Participants were required to complete a series of practice sessions in order to ensure familiarity with the experimental setting; all practice and experimental sessions were conducted at the same time of day, hence, avoiding circadian influence on participant’s performance. Multiple simulated driving measures were recorded for each participant in order to establish the whole range of OSAS driving impairment. In the present study, researchers were also interested in comparing the level of driving impairment of OSAS patients to healthy participants that are intoxicated. This association would allow to directly defining the level of severity of OSAS with comparing it to alcohol consumption while driving.

Overall the main aims of the study were the following: · Investigate OSAS patients’driving performance · Measure efficacy of CPAP treatment to improve driving performance of OSAS patients · Compare impairment levels caused by OSAS and alcohol consumption

Method

Participants

Eighteen OSAS patients (17 male/1 female; 51.9±11.54 years old) and 18 healthy controls (14 male/4

100 DRUID 6th framework programme Deliverable 1.2.2 female; 45.5±16.4 years old) with Body Mass Index (BMI) 33.4±7.13 kg/m2 26.26±3.25 kg/m2, respectively matched for driving experience volunteered in this study. Participants received reimbursement for their participation. The OSAS patients were selected by the collaborating research doctor from a list of patients from the Centre of Air Medicine IASI. OSAS patients were selected after their participation in a polysomnographic study. Healthy participants were individuals who responded to an advertisement placed by the research team. All patients and control participants underwent medical examination to ensure that no other medical condition was present (except sleep apnoea for patients) and were free of medication. In addition, psychiatric and cognitive evaluation was performed in both groups in order to ensure that participants had no mental disorders and/or cognitive impairments. Moreover, discrete inclusion and exclusion criteria applied. Exclusion criteria for both groups were positive testing for alcohol (using the LifeLoc Technologies FC20 portable breath tester) or positive testing for cocaine, amphetamines, methamphetamines, opiates, marijuana/hashish and benzodiazepines (using the Medimpex United, Inc. XALEX 6 Panel Multi Drug Urine Testing Kit). Participants were tested each time they visited HIT premises. Inclusion criteria for sleep apnoea patients OSAS was based on a diagnosis of an Apnoea-Hypopnoea Index (AHI) •10 (after the polysomnographic study). The healthy group was matched as much as possible to the OSAS group in terms of age and gender. Healthy participants were screened by the affiliated medical professionals. The healthy group had no medical history in general and, specifically, did not suffer from any respiratory conditions. Participants in both groups were active and experienced drivers defined as having a driving license for at least 3 years and driving an annual distance of at least 10000 km. Written informed consent was obtained from all participants before enrolment and after briefing.

Study design

A mixed design was applied. A repeated within participants’for both OSAS and alcohol consumption groups of the experiment. In addition, a between participants’design was applied for comparisons between the control and the patient (OSAS) groups. Independent variables were: a) the CPAP treatment (two conditions: treated and untreated) and b) alcohol 0.05% consumption (two conditions: intoxicated and sober). Scenarios’order was counterbalanced to minimise order effects. The experimental protocol was reviewed and accepted by the Biethics Committee of CERTH3.

Procedure

Untreated OSAS patients were tested and then re-tested after having used the CPAP treatment continuously for at least 7 days. The healthy group was tested in two conditions, with BAC 0% and with BAC 0.05%, which is the legal limit for driving in Greece. Tests were performed at the same time of day for each participant, in the afternoon between 14:00 and 17:00. Control participants had no history of alcohol abuse. In order to calculate the BAC of the participants the equation described by the National Highway Traffic Safety Administration (NHTSA, 1994)4 was applied ([23.36 * (SD * 0.5)] / [W * GC * 1000]) * 0.806 * 100- [T *

3 CERTH: Centre for Research and Technology is a non-profit organisation and Hellenic Institute of Transport (HIT) is part of this large research centre 4 National Highway Traffic Safety Administration (Herbert Moskowitz, Ph.D.) http://www.nhtsa.dot.gov/people/injury/alcohol/bacreport.html 101 DRUID 6th framework programme Deliverable 1.2.2 0.017], where W = the weight of the participant in kg; GC = 0.58 for males/0.49 for females; SD = the number of standard drinks consumed; and T = time in hours since the start of the first sip. This formula is an updated Widmark’s formula and was utilised because it takes in to account the following variables: total body water (from weight and gender); alcohol dose (from number of standard drinks), which is used to calculate the amount of alcohol in body water and in the bloodstream; and rate of elimination (using averaged metabolism rates and hours since start of drinking event). Adequate time (approximately 15 minutes) had to pass in order to check the alcohol level with the breathanalyser. Time had to pass in order to avoid measuring residuals in the mouth cavity. In case participants did not reach the required level, additional shots of alcohol were administered (3 ml) till 0.05% was reached. The level of alcohol had to be sustained throughout the experiment; hence breathanalysis was performed prior and post driving scenarios and in-between neuropsychological tasks. Participants were asked to arrive at HIT at the same time of day for both of their tests as described above. At their first appointment they completed a background questionnaire and written consent was obtained after briefing about the main objectives of the study. After both alcohol screening with breathanalyser and urine drug screening were found negative, participants had a familiarisation drive in order to get used to the driving simulator. Participants had to complete the following tasks at the driving simulator: a) a lane tracking scenario for about 20 minutes on a highway environment maintaining a constant speed of 90 km/h and b) a car following scenario for about 20 minutes on a highway environment maintaining a safe distance from the lead vehicle that was moving with a steady speed of 90 km/h. The vehicle ahead would brake abruptly suddenly and unexpectatly during the scenario. Participants were instructed to maintain a constant speed of 90 km/h and steady lateral position (lane tracking scenario) and to maintain a safe distance to the lead vehicle (car following scenario). The order of the two scenarios was counter-balanced between the two tests and among participants. Participants were driving alone with no radio on and researchers were controlling for environmental noise for both driving and laboratory tests. After the simulator drives the participants had to complete a test battery on attentional performance (winTAP) (Zimmermann and Fimm, 1993). Participants were asked not to have coffee or alcohol at least 24 hours prior testing. EOG recordings were obtained whilst driving and during the neuropsychological tests. Participants had breaks in between scenarios and if they required more time they had extra breaks. Participants were closely observed in order to check for any simulator sickness symptoms (e.g. sweating, yawning). Subjective scales were filled in before and after each driving scenario and before and after each attentional test. Detailed accounts of the driving scenarios, subjective scales, attentional battery, and physiological measures are provided in the respective sections.

Driving and psychomotor tests

The CERTH/HIT driving simulator (Figure 1 in Chapter 3) is built around a Smart cabin equipped with sensors. The actuation of all control levers, windshield wipers, blinker, ignition key and light switch is electronically transmitted to the driving computer. All operational elements such as steering wheel,

102 DRUID 6th framework programme Deliverable 1.2.2 accelerator pedal, brake pedal, gearshift lever and handbrake lever, provide nature-true force reactions. The visual system includes five large-screens, each having a width of 2 m. There is on-screen projection with video projectors (2500 ANSI-lumen). The sound system generates original sounds according to the situation (starter, engine noise, horn, screeching of tires, drive wind, rain, etc.). The vibration device creates nature true vibrations of the car according to the revolvation of the simulated engine. The simulator is equipped with special software which allows the development of specific driving scenarios aiming at creating a monotonous driving environment. There is surrounding traffic with maximum 30 road users with embedded artificial intelligence elements including passenger cars, trucks, pedestrians and cyclists. The primary driving variables were the standard deviation of lateral position (SDLP) for the lane tracking scenario and the percentage (%) of time driven with certain Time-To-Collision (TTC) values. SDLP is a very reliable index of weaving and overall lateral control of the vehicle. The car following scenario was designed with stable leading vehicle speed. In this scenario the percentage of time (%) spent with Time-to-Collision (TTC) falling into five categories was calculated. The selected five categories were the following. As there is no previous literature on recommended stratification, the decision was based on the interest to investigate more clusters for the riskier TTCs, one cluster reflecting following behaviour, and one cluster for investigating non-conforming to given instructions behaviour.

TTC 0-1: percentage of time driven with TTC between 0 and 1 sec

TTC 1-1.5: percentage of time driven with TTC between 1 and 1.5 sec

TTC 1.5-2: percentage of time driven with TTC between 1.5 and 2 sec

TTC 2-4: percentage of time driven with TTC between 2 and 4 sec

TTC >4: percentage of time driven with TTC above 4 sec

Obviously, very low TTC values suggest there is higher risk to be involved in an accident in the car following scenario. In addition, more time spent in the category TTC 2-4 implies that participants spent more time in following the leading vehicle with keeping a safety distance, hence complying with given instructions. More time spent with TTC >4 suggests that participants were less concentrated and not attentive to the driving task and, consequently, did not follow the driving instructions all the time. Failure to follow instructions might be a sign of fatigue and lowered concentration levels. The winTAP (Zimmermann and Fimm, 1993) is a standardised neuropsychological battery of attentional performance. The selected individual tests applied were the following: a) alertness: choice between two stimuli, b) Go/NoGo: simple choice reaction task, c) divided attention audio: reaction to two consecutive same sounds (either low or high), d) divided attention visual: reaction to the formation of a small rectangular by small stars, and e) divided attention with both visual and auditory stimuli: presented simultaneously. Reaction times (msec), omissions and errors were recorded.

Secondary measures

Another parameter of interest was the time it takes to react to an abrupt breaking of the lead vehicle. Reaction times (sec) to abrupt breaking of the lead vehicle was recorded for the car following scenario. In

103 DRUID 6th framework programme Deliverable 1.2.2 addition EOG recordings were gathered.

Subjective measures

Participants were asked to subjectively evaluate their vigilance state. The Karolinska Sleepiness Scale (KSS) was applied. KSS is a universally accepted, validated and standardised scale which was administered before and after the task completion across all conditions. The 9-point KSS (Åkerstedt and Gillberg, 1990) was used: 1=very alert, 3=alert, 5=neither alert nor sleepy, 7=sleepy (but not fighting sleep), 9=very sleepy (fighting sleep). Sleepiness was subjectively rated before and after each driving scenario and before and after the neuropsychological tests were performed. Higher scoring meant less vigilance and subsequently increased sleepiness. In general, participants were less vigilant just before the neuropsychological tests and more vigilant at the beginning. In addition, participants rated how much they had to try in order to complete the driving tasks [Rating Scale of Mental Effort (0-150); RSME] (Zijlstra, 1993) and how well they thought they performed in the driving tasks compared to their everyday driving experience (Driving Quality scale, Brookhuis et al., 1985). tThe original subjective scales were translated into Greek and back-translated by an independent professional and then verified by a native English speaker.

Statistical analysis

Within and between participants comparisons (superiority tests) were carried out with repeated measures ANOVAs and one-way ANOVAs. In case of violation of homogeneity and homoscedacity assumptions, non- parametric equivalents were administered (Friedman and Wilcoxon rank test, respectively). Within comparisons were carried out in order to investigate the effect of CPAP treatment and alcohol level (0.05%). Between comparisons were carried out in order to investigate the relationship between OSAS and alcohol in driving impairment. The Į level was set at .05. Statistical analyses were carried out with SPSS 18.0 for Windows (SPSS 18.0, Chicago, IL.).

Results

This section is divided further in 4 subsections according to the types of collected data. Firstly, the main primary driving parameters are presented per driving scenario (i.e. lane tracking and car following). Secondly, additional neuropsychological measures are presented. Thirdly, EOG findings are presented and finally subjective scales results are given.

Driving parameters (primary variables)

Lane tracking OSAS patients showed more deteriorated performance in lane keeping (SDLP) than any other condition. CPAP treatment did not seem to significantly improve lane keeping behaviour, although decreased swerving is recorded. The following graph depicts mean SDLP (m) values per condition.

104 DRUID 6th framework programme Deliverable 1.2.2 *p = .027 **p <.001 ***p <.001 ****p <.001

Figure 1: Mean SDLP (m) values per condition in lane tracking scenario

As shown above, the OSAS patients had less control over the lateral position of the vehicle before CPAP treatment. After treatment, lateral control increases but not significantly (p>.05). However, the tracking control is still impaired compared to suggested thresholds (Brookhuis et al., 2003) and when compared to the control group before (F (1,34) = 10.57, p<.001)** and after alcohol consumption (F (1,34) = 8.23, p<.001)***. In addition, intoxicated control group weaving was greater compared to the no alcohol condition (F (1,17) = 5.9, p=.027)*. Alcohol had an effect on participants’lateral control but the magnitude was not as great as the sleep apnoea’s effect on OSAS patients****.

Car following

Percentage of time (%) spent with Time-to-Collision (TTC)

The results for TTC comparisons among conditions are described separately for each chosen percentage of time (%) driven TTC category and are summarised in the following two tables:

105 DRUID 6th framework programme Deliverable 1.2.2 CONTROL GROUP

Baseline Alcohol Significance Driving Mean±SE [95% CI] Mean±SE [95% CI] measures

Car Following

TTC0-1 (sec) 6.35±0.44 (5.41 – 7.28) 5.7±0.36 (4.93 – 6.46) *

TTC1-1.5 (sec) 7.17±0.49 (6.14 – 8.21) 10.05±0.77 (8.43 – 11.66) *

TTC1.5-2 (sec) 13.52±1.03 (11.35 – 15.68) 13.53±1.28 (10.82 – 16.23) NS

TTC2-4 (sec) 35.97±1.26 (33.32 – 38.62) 33.44±1.33 (30.63 – 36.24) *

TTC>4 (sec) 37.64±1.95 (33.53 – 41.75) 36.64±2.11 (32.18 – 41.1) NS

Table 1: Mean (±SE) scores and 95% confidence intervals of percentage of time spent (%) driven with TTC values according to the TTC clusters for the control group (alcohol consumption). * Significant (p<.05); NS: Non-significant (p>.05).

OSAS PATIENTS

No treatment CPAP treatment Significance Driving Mean±SE [95% CI] Mean±SE [95% CI] measures

Car Following

TTC0-1 (sec) 6.81±0.57 (5.67 – 7.94) 6.37±0.54 (5.16 – 7.58) *

TTC1-1.5 (sec) 11.83±1.33 (9.02 – 14.47) 11.95±1.39 (9.01 – 14.89) NS

TTC1.5-2 (sec) 16.94±1.36 (13.33 – 19.06) 14.58±1.39 (11.65 – 17.5) *

TTC2-4 (sec) 29.94±1.27 (27.26 – 32.62) 32.37±1.15 (29.95 – 34.79) *

TTC>4 (sec) 35.23±3.21 (28.46 – 41.99) 34.73±3.27 (27.82 – 41.64) NS

Table 2: Mean (±SE) scores and 95% confidence intervals for the time spent (%) driven with TTC values according to the TTC clusters for the OSAS patients. * Significant (p<.05); NS: Non-significant (p>.05).

· Percentage (%) of time with TTC0-1

Within and between comparisons were carried out for each TTC category. Untreated (6.81±0.57% of time) OSAS patients spent significantly more time with TTC values between 0 and 1 sec which is extremely risky (F (1,17) = 5.46, p = .032) compared to the CPAP treated condition (6.37±0.54% of time). Similarly,

106 DRUID 6th framework programme Deliverable 1.2.2 participants from the control group (6.35±0.44% of time) spent almost significantly more time with TTC values between 0 and 1 second when compared to the alcohol consumption condition (5.7±0.36% of time) (F(1,17) = 4.27, p = .054). Between groups’comparisons were not significant (p>.05).

· Percentage (%) of time with TTC 1-1.5

Intoxicated participants (10.05±0.77% of time) spent approximately three percent more time with TTC between the values 1-1.5 seconds compared to the no alcohol condition (7.18±0.49% of time) (F(1,17) = 34.68, p<.001). Similarly, untreated (11.83±1.33% of time) and treated OSAS patients (11.95±1.39% of time) spent statistically significant more time (around 4%) with TTC values between 0 and 1 second compared to the control group (7.17±0.49% of time) (F(1,34) = 10.73, p=.002 and F(1,34) = 10.46, p=.003, respectively). All other comparisons were of no statistical significance (p>.05).

· Percentage (%) of time with TTC 1.5-2

The only statistically significant difference was found between OSAS treated (14.58±1.39% of time) and untreated patients (16.94±1.36% of time) (F(1,17) = 8.63, p =.009). No other statistically significant difference was found within and between groups. Nevertheless, untreated OSAS patients and intoxicated participants spent more time than the other two conditions within the TTC values 1.5 and 2.

· Percentage (%) of time with TTC 2-4

Almost all comparisons were statistically significant (Figure 2) except the comparison between the treated OSAS group and the alcohol group (p>.05). As shown in the graph below, treated OSAS patients spent significantly more time (3%) compared to the time they spend driven with TTC 2-4 before treatment ( F(1,17)

= 18.05, p = .001). Likewise, participants in the control group spent more time (approx. 2.5%) with TTC 2-4 (F(1,17) = 8.93, p = .008) than the intoxicated group. Untreated OSAS patients spent more time than the control group in TTC 2-4 (F(1,34) = 11.42, p = .002). This TTC category is strongly connected to safety distance keeping; hence the significant difference among groups could reflect risk taking behaviour. The control group spent significantly more time (3%) than the treated OSAS patients in car following (F(1,34) = 4.49, p =.041). On the other hand, no statistically significant differences between the OSAS treated group and the alcohol group (p>.05) were found. Similarly, no significant difference were found between the alcohol and the untreated OSAS patients groups (p = .066; trend).

107 DRUID 6th framework programme Deliverable 1.2.2 Figure 2: Mean percentage (%) of time driven with TTC 2-4

· Percentage (%) of time with TTC >4

No statistically significant differences were found for both within and between group comparisons (p>.05). However, the control group spent more time with TTC values greater than four and treated OSAS group spent less time than the rest of groups/conditions with TTC values greater than four; the range of differences, though, was not greater than 3%.

Braking reaction time (sec)

Greater reaction times (sec) were recorded for the untreated OSAS patients and the lowest reaction times (sec) were recorded for the CPAP treated patients (F (1,17)=12.37, p =.003). Moreover, it seems that alcohol did not have an impact in braking reaction time for the alcohol group (p>.05) at the legal limit. On the contrary, participants from the control group reacted (brake) much faster when they were sober compared to the untreated OSAS patients (F (1,17)=6.55, p=.015).

Secondary variables

This section includes other measurements supporting the primary variables. The secondary measurements included are the results from the neuropsychological tests performed (WinTAP battery) and the EOG measurements.

108 DRUID 6th framework programme Deliverable 1.2.2 Neuropsychological measures

Control participants were statistically significant less alert when intoxicated (F (1,17) = 9.39, p=.007). Intoxicated participants showed statistically significant higher Reaction times (RTs) (ms) in the divided visual attention task compared to OSAS untreated patients (F (1,34) = 5.04, p<.032).

Figure 3: Mean reaction time (msec) in neuropsychological tests

Participants after alcohol consumption (0.05%) had significantly more omissions than CPAP treated OSAS patients (F(1,158) = 8.55, p=.004) when they had to react to both visual and auditory stimuli. No other statistically significant differences were found (either in errors and/or omissions).

Electrooculograph (EOG)

The following parameters were statistically analysed in both car following and lane tracking scenarios: · Blink duration · Number of blinks · Closing time (time from baseline to peak) · Reopening time (time from peak to return to baseline)

109 DRUID 6th framework programme Deliverable 1.2.2 The parameters were calculated in continuous time windows of 1 minute. During lane tracking task, none of the parameters were significantly different between CPAP and no CPAP runs. However, during car following task the number of blinks was significantly (p<.001) higher in patients with no CPAP (figure 4). It appears that OSAS patients before CPAP treatment had more eye-blinks than after CPAP treatment.

Figure 4: Number of blinks

Moreover, the closing time was significantly (p<.001) lower in patients with no CPAP (figure 5). It is evident that patients with CPAP had longer closing time than those without it.

Figure 5: Closing time (sec)

No other statistically significant differences were found regarding EOG measurements (p>.05).

Subjective scales

Higher scores are observed for OSAS patients in almost all conditions. Higher KSS scores for neuropsychological tests may be due to the fact that attention tests were carried out at the end of the experiment. Therefore, participants could have been tired near the end of the experiment (i.e. added tiredness from all conditions) and not only because of the effort required to perform the attentional tests.

110 DRUID 6th framework programme Deliverable 1.2.2 Figure 6: Mean RSME and median KSS scores Figure 7: Median KSS scores

Regarding subjective feeling of the quality of their driving experience (DQS) in the simulated environment, in comparison to how they normally drive, only 3 participants with OSAS mentioned that they drove worse than usual. Similarly, the same participants thought that they drove worse than usual even after CPAP treatment. One participant stated that he drove worse than usual in the control condition. The number of participants that they thought they drove worse than usual after alcohol consumption increased to 5.

111 DRUID 6th framework programme Deliverable 1.2.2 Discussion

The main findings demonstrated that sleep apnoea affects driving performance. Continuous Positive Airway Pressure (CPAP) treatment did not seem to improve significantly driving performance. Furthermore, the present experiment suggests that alcohol consumption at BAC=.05% impairs driving performance; however deterioration is less when compared to sleep apnoea. Equivalence might be present for brake reaction time (sec) between sleep apnoea and alcohol effect. The present study showed that sleepiness and hypo-vigilance induced by sleep apnoea may be an increasingly contributing factor in road accidents which are sometimes overlooked in driving research focusing mainly on alcohol and illicit drugs. In the present experiment, OSAS patients showed almost one third more weaving than the control group participants (31.46%). Many studies have shown increased accident risk for OSAS patients because of micro-sleeps, though a considerable amount of research is based on questionnaires and self-reporting assessments. According to self reports, OSAS patients experience sleepiness while driving because of lack of sleep or lack of quality sleep and, thus, report higher rate of driving related accidents. Treatment with nasal Continuous Positive Airway Pressure (nasal CPAP) has been found to decrease self-reported automobile crashes in patients with sleep apnoea. These studies have been carried out with knowledge of subjectivity error and biases involved. The step towards objectification of this type of research was conducted by Findley and colleagues (2000) who based their hypothesis about CPAP treatment decreasing automobile accidents based on accidents reports from the Department of Motor Vehicle. These data were supported by telephone interviews on the effect of the nasal CPAP treatment on their fitness to drive, description of sleeping hours, miles driven per week etc. It is essential to note that compliance to treatment was based on self-reports and although objective measures were taken into account –as far as reporting is concerned- the absence of a control group is important. In this study, OSAS patients reported higher sleepiness rates compared to the other groups which is in line with findings from self-reported based studies discussed in the previous paragraph. OSAS patients showed deteriorated driving performance in monotonous simulated environment which is in agreement with similar studies. For instance, Orth and colleagues (2005) found improvement due to nasal CPAP application. The analysis was based on accident crashes and concentration faults which are different parameters from the ones recorded in this study. Although accident risk is of fundamental importance in driving safety research and driving scenarios included random elements (e.g. obstacles’occurrence such as a deer or a vehicle) there was no control group in order to compare crashes across conditions. Furthermore, this study targeted different types of vehicle parameters and assessment variables than the one performed by CERTH/HIT. Moreover, several studies have shown significant improvement in driving behaviour as a result of using the nasal CPAP for different periods of time. For instance, Loredo and colleagues (1999) found differences in CPAP treatment after 7 days. The same treatment period was used in this study based on Loredo et al.’s (1999) findings. The differences were not found on the quality of sleep (sleep architecture) and, thus, changes in this type of sleep may be influential in order to find improvement in driving variables such as the ones measured in this experiment (SDLP, %TTC, and BRT). In other words, the period of treatment time

112 DRUID 6th framework programme Deliverable 1.2.2 may not suffice in order to reveal improvement in driving behaviour when the driving task is performed in a monotonous simulated environment with these specifically set and derived parameters. In addition, the monotonous environment may be the most “dangerous”choice or the most “accident-evoking”but it is far from a real traffic situation and induces sleepiness beyond control which reflects probably the worse-case scenario and not necessarily the most frequent situation. Turkington and colleagues (2004) focused on similar driving parameters. They aimed to find the most effective period of time for CPAP application and found that the threshold is seven consecutive days (similar to Loredo and colleagues findings but for different reasons). While they found significant tracking error (p=.004), reaction time (p=.036), and number of off road events per hour (p=.032) -which are parameters related to the ones measured in this experiment-it is important to note that the cases of OSAS patients included were much more severe cases than the general population and the ones included in this study. Therefore significant findings in 7 days treatment may be generalised only to a certain percentage of severe patients and not to other groups of mild and less severe OSAS. As a result, it is inconclusive if the significant improvement in tracking error –which is the most relevant measure- in 7 days treatment with nasal CPAP, would be found if the cases included were milder. The size of effect might have been inflated due to the initially measured deterioration. The percentage (%) of time spent driving within certain categories of TTC values has not been investigated in the literature before, thus, it is difficult to examine the TTC results under the prism of research-to-date and to step forward towards any generalisable inferences. Decrease in crashes, as mentioned above, might be associated with the significant less time spent with TTC values less than 2 seconds in treated OSAS patients (p=.032). Therefore, the accident risk might be reduced. This logical induction should serve simply to show probable relations in driving variables between the two studies and not actually compare effects and findings. Overall significant changes in percentage of time spent with TTC values less than two seconds were found in both treated and control group, meaning that CPAP treatment reduced risky behaviour and that alcohol even at legal limit (0.05%) increases risky behaviour. Likewise, safety distance keeping from lead vehicle temporally increases for both control participants and CPAP treated patients. Safer behaviour is observed in treated patients and the opposite seems to hold true for untreated patients and intoxicated participants. Driving impairment due to alcohol consumption, even for the legal national limit (0.05%), is evident in both lane and car following scenarios. One of the main objectives was to investigate the relation of sleep apnoea effect and alcohol in specific driving variables. Inferences should be conservative and differences were not found for brake reaction time (sec). This finding suggests that braking delay might be similar for sleep apnoea patients and intoxicated participants (alcohol consumption at legal limit). Brake reaction time (BRT) is the time to respond to sudden changes in the driving environment by fully depressing a brake pedal. Previous research has identified possible risk factors associated with delayed brake reaction time, such as alcohol use (Kuypers et al., 2006) and medications causing sedation such as antihistamines or psychotropic agents (Vuurman et al., 2004). Previous research has identified two primary components of Brake Reaction Time (BRT). The first component is the Initial Reaction Time (IRT, sometimes referred to as perception-reaction time), and the second is Physical Reaction Time (PRT, also called brake-movement time). IRT is the time from when the stimulus first appears to the start of foot movement off the accelerator; the PRT is the time from the first physical movement off the accelerator to the depression of the brake pedal. Research suggests these components could be influenced by different

113 DRUID 6th framework programme Deliverable 1.2.2 factors, but they may not be entirely independent like gender and age (Warshawsky-Livne and Shiner, 2002). Therefore, by controlling these factors, it might be possible to reveal effects. A future research effort could focus on investigating the effect of different levels of alcohol (BAC) specifically in the two aforementioned brake reaction components. Such inferences could even support the refining of existing rehabilitation programmes. Regarding neuropsychological assessment the findings are not in accordance with relevant studies of sleep apnoea as CPAP treatment did not seem to improve patients’alertness. In general, no major differences were found in alertness, divided attention, and simple choice task. Alcohol impairing effect was found in alertness. Intoxicated participants felt significantly less alert after alcohol consumption. However, the effect of awareness of being intoxicated was not controlled. A placebo-alcohol group would accommodate for this possibility. It seems that the effect in visual divided attention was significantly higher in alcohol group compared to OSAS patients. The findings suggest that alcohol at 0.05% affects more participants’alertness and visual attention after alcohol consumption than obstructive sleep apnoea does. A great deal of research has been carried out for determining the effects of alcohol on the capacity to divide attention (e.g. Patel, 1988; Moskowitz and Burns, 1990). As most of these studies investigated the effects of high or moderate blood-alcohol levels, the deterioration revealed was detrimental. Investigation of potential effects of low levels of alcohol (i.e. legal limit) could yield interesting results and shift research interest towards the processes underlying the distribution of attentional resources in intoxicated drivers (i.e. their scores in divided attention tasks). No significant differences were found between control and intoxicated participants. The alcohol level was either low for revealing such deterioration or participants tried too hard to compensate for it. Moreover, a future investigation of alcohol effect in distribution of attention would be focussing on how this “mobility”of attention –as it is often called- affects attentional shift in driving tasks. The latter could move research beyond increased or decreased measures of effect and provide valuable input for the development of alcohol related guidelines. Also, the definition of the qualitative characteristics of attentional mobility could prove valuable for further development of rehabilitation measures. Subjective assessments showed that participants under alcohol consumption thought that they drove worse than the rest of the participants. OSAS patients did not report any difference in driving quality which is in agreement with CPAP treatment results. As expected, OSAS patients reported higher sleepiness and increased mental effort while driving. Indeed, OSAS participants showed greater weaving and, consequently, less control than the rest of participants. However, not great differences were found for mental effort within each group. Overall, it seems that the effect of sleep apnoea is detrimental compared to the alcohol effect at 0.05%. The application of intermediate alcohol BAC levels (i.e. 0.02, 0.08., 0.1) could provide insight in finding comparable levels of impairment. Probably higher levels of alcohol are required in order to reveal any equivalence to OSAS for the chosen types of driving parameters. Following the need for further stratification of alcohol BAC levels within the same study, research focus could also be shifted towards the quality of sleep affected in OSAS patients and the appropriate and/or associated variables. For example, automatic processes might be more difficult to control while driving, hence easier to investigate improvements in less treatment time (e.g. in seven consecutive days) than other driving skills (e.g. safety distance may need 30 days).

114 DRUID 6th framework programme Deliverable 1.2.2 Participants said they used the nasal CPAP machine on a daily basis. The research team had only this information about the quality and frequency about the application of the CPAP machine during the night as they were all outpatients. Last, equivalence testing could be oriented to the study of the two components of brake reaction time for alcohol and other medicines.

Acknowledgements This work was conducted with the assistance of a specialised team of medical experts, led by Dr. Chrysoula Papadeli, experienced in diagnosis and treatment of OSAS patients. Medical professionals were responsible for the patients’recruitment, EOG measurements and analyses of respective results.

115 DRUID 6th framework programme Deliverable 1.2.2 References

Åkerstedt, T. and Gillberg, M. (1990). Subjective and objective sleepiness in the active individual. International Journal of Neuroscience, 52, 29–37. Brookhuis, K.A., De Waard, D., Fairclough, S.H. (2003). Criteria for driver impairment. Ergonomics, 46, 433- 445. Brookhuis, K.A., DeVries, G., Prins Van Wijngaarden, P., Veenstra, G., Hommes, M., Louwerens, J.W., O’Hanlon, J.F. (1985). The effects of increasing doses of Meptazisol (100, 200, 400 mg) and Glaferine (200mg) on driving performance. Report VK 85-16. Haren, The Nethrlands: Traffic Research Center, University of Groningen. Cassel, W., Ploch, T., Becker, C., Dugnus, D., Peter, J. H., von Wichert, P. (1996). Risk of traffic accidents in patients with sleepdisordered breathing: reduction with nasal CPAP. European Respiratory Journal, 9, 2606-2611. Engleman, H. M., & Wild, M. R. (2003). Improving CPAP use by patients with the sleep apnoea/hypopnea syndrome (SAHS). Sleep Medicine Review, 7, 81-99. Findley, L. J., Fabrizio, M. J., Knight, H., Norcross, B. B., Laforte, A. J., & Suratt, P. M. (1989). Driving simulator performance in patients with sleep apnea. American Review of Respiratory Disease, 140, 529-530. Findley, L., & Suratt, P. M. (2001). Serious motor vehicle crashes: the cost of untreated sleep apnea. Thorax, 56, 505. Findley, L. J., Unverzagt, M., Guchu, R., Fabrizio, M., Buckner, J., & Suratt, P. (1995). Vigilance and automobile accidents in patients with sleep apnea or narcolepsy. Chest, 108, 619-624. George, C. F. P. (2001). Motor vehicle collisions are reduced when sleep apnoea is treated with nasal CPAP. Thorax, 56, 508-512. George, C. F. P. (2004). Driving and automobile crashes in patients with obstructive sleep apnoea/syndrome. Thorax, 59, 804-807. George, C. F. P., Boudreau, A. C., & Smiley, A. (1996). Simulated driving performance in patients with obstructive sleep apnea. American Journal of Respiratory Critical Care Medicine, 154, 175-181. George, C., Nickerson, P., Hanley, P., Millar, T., & Kryger, M. (1987). Sleep apnea patients have more automobile accidents. Lancet, 2, A447. Hack, M. A., Choi, S. J., Vijayapalan, P., Davies, R. J. O., & Stranling, J. R. (2001). Comparison of the effects of sleep deprivation, alcohol, and obstructive sleep apnoea (OSA) on simulated steering performance. Respiratory Medicine, 95, 594-601. Horne, J. A., & Reyner, L. A. (1995). Driver sleepiness. Journal of Sleep Research, 4, 23-29. Horne, J. A., & Reyner, L. A. (1999). Vehicle accidents related to sleep: a review. Occupational Environmental Medicine, 56, 289-294. Juniper, M., Hack, M. A., George, C. F. et al. (2000). Steering simulation performance in patients with obstructive sleep apnoea and matched control subjects. European Respiratory Journal, 15, 590-595. Kuypers, K.P., Samyn, N., Ramaekers, J.G. (2006). MDMA and alcohol effects, combined and alone, on objective and subjective measures of actual driving performance and psychomotor function. Psychopharmacology. 187,467–475. Maycock, G. (1996). Sleepiness and driving: the experience of UK car drivers. Journal of Sleep Research, 5, 229-37. Moskowitz, H., Robinson, C.D. (1987). Effects of low doses of alcohol on driving–related skills: a review of

116 DRUID 6th framework programme Deliverable 1.2.2 the evidence. In: Technical Report, United States Department of Transportation, Washington, D.C., USA. Moskowitz, H., Burns, M. (1990). Effects of alcohol on driving performance. Alcohol Health & Research World, 14, 12-14. Naegele, B., Thouvard, V., Pepin, J. L., et al. (1995). Deficits of cognitive executive functions in patients with sleep apnea syndrome. Sleep, 18, 43-52. Orth, M., Duchna, H.-W., Leidag, M., Widdig, W., Rasche, K., Bauer, T. T., Walther, J. W., de Zeeuw, J., Malin, J.-P., Schultze-Werninghaus, G., & Kotterba, S. (2005). Driving simulator and neuropsychological testing in OSAS before and under CPAP therapy. European Respiratory Journal, 26, 898-903. Patel, R.M. (1988). Ethanol’s effect on human vigilance during a simple task in the presence of an auditory stressor. Psychological Reports, 63, 363-366. Simpson, H.M. & Mayhew, D.R. (1991). The hard core drinking driver. Traffic Injury Research Foundation of Canada, Ottawa. Sullivan, C. E., Issa, F. G., Berthon-Jones, N., et al. (1981). Reversal of obstructive sleep apnoea by continuous positive airway pressure applied through the nars. Lancet, 1, 862-865. Swedler, B.M, Biecheler, M.B., Laurell, H., Kroj, G., Lerner, M., Mathijssen, M.P.M., Mayhew, D. & Tunbridge, R.J. (2004). Worldwide trends in alcohol and drug impaired driving. In: Traffic Injury Prevention Volume 5, Issue 3. p. 175-184. Teran-Santos, J., Jimenez-Gomez, A., Cordero-Guevara, J., et al. (1999). The association between sleep apnea and the risk of traffic accidents. New England Journal of Medicine, 340, 847-851. Turkington, P. M., Sircar, M., Saralaya, D., & Elliott, M. W. (2004). Time course of changes in driving simulator performance with and without treatment in patients with sleep apnoea hypopnoea syncrome. Thorax, 59, 56-59. Vuurman EF, Rikken GH, Muntjewerff ND, de Halleux F, Ramaekers JG. Effects of desloratadine, diphenydramine and placebo on driving performance and psychomotor performance measurements. Eur J Clin Pharmacol. 2004;60:307–313. Warshawsky-Livne L, Shiner D. Effects of uncertainty, transmission type, driver age and gender on brake reaction and movement time. J Safety Res. 2002;33:117–128. Weiler, J. M., Bloomfield, J. R., Woodworth, G. G., Grant, A. R., Layton, T. A., Brown, T. L., McKenzie, D. R., and Baker, D. W. (2000). Effects of Fexofenadine, Diphenidramine, and Alcohol on driving performance. Annals of Internal Medicine, 132, (5), 354-363. Wu, H., & Yan-Go F. (1996). Self-reported automobile accidents involving patients with obstructive sleep apnea. Neurology, 46, 1254-1257. Yamamoto, H., Akashiba, T., Kosaka, N., Ito, D., Horie, T. (2000). Longterm effects of nasal continuous positive airway pressure on daytime sleepiness, mood and traffic accidents in patients with obstructive sleep apnoea. Respiratory Medicine, 94, 87-90. Zijlstra, F.R.H. (1993). Efficiency in work behavior. A design approach for modern tools. PhD thesis, Delft University of Technology. Delft, The Netherlands: Delft University Press. Zimmermann, P. & Fimm, B. (1993). Testbatterie zur Erfassung von Aufmerksamkeitsstörungen. Version 1.02. Freiburg: Psytest.

117 DRUID 6th framework programme Deliverable 1.2.2 Chapter 5 : effects of codoliprane and zolpidem, alone or in combination, on elderly drivers’ behavior

ML Bocca (UNICAEN), P. Denise (UNICAEN), C. Berthelon (IFSTTAR).

118 DRUID 6th framework programme Deliverable 1.2.2 Abstract

The majority of older adults regularly use several medications, in particular due to an increase of insomnia with age, but also due to an increase of pain. Paradoxically, most of experimental studies on drugs effects are conducted on healthy young subjects. The aim of the present study was to evaluate the effects of the combination of zolpidem 10 mg taken at bedtime and codoliprane taken at awakening (codeine 20 mg - paracetamol 400 mg) in aged subjects. It was a cross-over, double-blind, placebo-controlled study. Four combinations were tested: placebo-placebo, zolpidem-placebo, zolpidem-codoliprane, placebo-codoliprane. Subjects performed three driving test at 9.00 am : monotonous, urban and car following. Sixteen healthy subjects aged 55 to 65 years participated in this experiment. Standart Deviation of Lateral Position (SDLP) was significantly increased both with zolpidem and with codeine-paracetamol alone. The combination of these two drugs are less clear but tendencies were found. Results are not significant with urban driving test. Globally, results confirm that zolpidem or codoliprane in aged subjects impair driving performance, a result no found in young subjects. Although unexpected, the combination of two drugs did not clearly impair the driving behaviour in the highway driving task.

119 DRUID 6th framework programme Deliverable 1.2.2 Introduction

Epidemiological data clearly indicate that hypnotic drugs were present in a significant percentage of drivers involved in accidents (Barbone et al. 1998, Neutel 1998, Orriols et al., 2010), whatever the driver’s age (Movig et al., 2004), thus suggesting that using these drugs increases the risk of accident . This risk has also been demonstrated in many experimental studies (for a review, see Vermeeren 2004) when taken at therapeutic or at higher doses (Walsh et al. 2004). Although the elderly, and notably women, account for a growing share of the driving population (Maycock 2000) and that the use of psychotropic drugs is common in this population (Legrain 2005), information on the effects of these drugs in this population are scarce. Recent experimental studies however showed that some hypnotics, without residual effects on young drivers’behaviour, do have residual effects on elderly drivers’ behaviour (Meskali, Berthelon et al., 2009; Bocca, Marie et al. under press). Older people may thus be differently and more often affected by drugs than younger people, due to age-related changes in pharmacokinetics and pharmacodynamics, such as reduced clearance and increased sensitivity (Glass, Lanctot et al., 2005) and physiological changes that affect how older people metabolize the medications. In another hand, a large part of elderly regularly use antalgics and this type of drug could also increase the risk of accident (McGwin, Sims et al. 2000, Orriols et al. 2010). In particular, epidemiological studies find low relative risk for opioid users (Fishbain et al. 2002; Kelly et al. 2004), but some recent studies show that natural opium alkaloids could increase the risk of being involved in an accident (Bachs et al. 2009; Dubois et al. 2010; Engeland et al. 2007; Kelly et al. 2004). Our own study does not show acute effects of antalgics on young healthy drivers’behavior (Amato, Marie et al., submitted). Whatever the case, all data are in agreement with the fact that using multiple medications or polypharmacy increases with higher age and is linked with an increased risk of accident (Gjerde, Normann et al. in press). The main objective of the present study is to evaluate residual effects of a hypnotics (zolpidem 10mg), an acute effect of an analgesic (codeine / paracetamol 20mg/400mg) and of the combination of these two molecules in healthy elderly subjects (55 to 65 years old) to assess a potential increase effect of this combination on driving performance.

Method

Subjects The study was carried out on 16 healthy volunteers (8 men and 8 women, mean age = 59.6 + 2.2 years). Inclusion and non-inclusion criteria were verified at a pre-selection visit with a sleep-neurophysiologist clinician. Subjects had a medical check-up to confirm their good physical condition, the absence of sleep, alertness, neurological, cardiovascular, respiratory, hepatic, renal, or metabolic disorders, a poor hygiene or abnormal usual sleep patterns, such as night workers or shift-workers, and substance abusers (caffeine, drug, or alcohol). They had no treatment at the time of their inclusion and during the previous two months. Subjects for whom taking one of the drugs studied presented a health risk, were also excluded: hypersensitivity to one of the drugs studied, current or past dependence on alcohol, opiates, benzodiazepines or any illicit drug. Subjects smoking more than 5 cigarettes per day, drinking more than 28 units of alcohol a week or consuming more than 150 mg of caffeine per day were also excluded. All participants had normal or corrected to normal vision (visual acuity greater than or equal to 7/10). They had

120 DRUID 6th framework programme Deliverable 1.2.2 driven regularly for at least 30 years. The participants signed a consent form and an informed commitment form. The experimental protocol was approved by a French local ethics committee (CCPPRB – Consultative Committee for the Protection of People in Biomedical Research – of the Basse-Normandie region).

Study design There was four experimental sessions conducted according to a double-blind, balanced, cross-over design. For each subject, a washout period of two weeks was respected between each session. Before each session, subjects took two pills administered in identical capsules (see table 1). at home, 11 pm at lab, 8 am Placebo placebo Zolpidem (10mg) placebo Placebo Codeine/paracetamol (20mg/400mg) Zolpidem (10mg) Codeine/paracetamol (20mg/400mg)

Table 1. Place and time of the different treatments. Each line of the table corresponds to one session. Urine was collected at 7 am, when subjects awoke, to verify the absence of BZD, ethanol, cannabis, cocaïne, amphetamines. Experimental session started at 9 am.

Driving tests The driving experiment was carried out on the FAROS fixed-base driving simulator equipped with an ARCHISIM object database. This simulator comprised a steering cab made up of a quarter of a vehicle (Fig. 1). The images, generated at a frequency of approximately 30 Hz, were projected by a video projector onto a screen (H: 60°; V: 49°) located 1.90 m from the driver’s eyes. The acquisition frequency for the different signals (position, speed, acceleration, etc.) was 30 Hz.

There were three driving tests: 1) 60 min along a motorway in monotonous conditions (no traffic, repetitive landscape). Subjects were instructed to drive as straight as possible within the right (slower) traffic lane while maintaining a constant speed of 110 km/h. 2) 15 min along a simulated urban route into which prototypical scenarios of accident were introduced. Each participant drove the route twice. Six scenario of accident were introduced in a counterbalanced order during the two trips on the circuit. Subjects had to drive at a speed of approximately 50 km/h, and to follow the road signs indicating the town centre. "Pedestrian crossing”scenario (pedestrian). A pedestrian, initially hidden from the driver’s vision by a bus, starts to cross when the driver arrives. The pedestrian appears in the driver’s field of vision 2.4 seconds before the driver reaches his level, which corresponds to a time-to-obstacle of 2.4 s. This configuration only remains effective if the driver does not perform any manoeuvre to avoid the pedestrian (braking or avoidance by swerving). "Sudden stop of the vehicle ahead at a traffic light" scenario (traffic light). The driver is driving in a row of

121 DRUID 6th framework programme Deliverable 1.2.2 traffic on a straight, two-lane carriageway with two-way traffic. Vehicles ahead block his view of an upcoming “T”intersection and the traffic light at the intersection. The light turns red and the vehicles ahead brake over a distance of approximately 25 m to stop at the light. "Vehicle overtaking and merging back into the lane" scenario (overtaking). The driver is driving in the right- hand lane of a straight, major 3-lane urban road with two lanes going in his direction. A vehicle overtakes the driver at a speed 10 km/h faster than him. When it is positioned 20 m ahead of the driver, it merges back into the right-hand lane ahead of him and brakes to a speed of 30 km/h. "Vehicle pulling out from a parking space" scenario (parking). A vehicle parked on the right-hand side of the carriageway pulls out of its parking space when the driver is at a distance of 20 m. "Left-hand turn by the vehicle ahead" scenario (left hand turn). The driver is driving in a straight carriageway with 2 lanes separated by axial markings. Ahead of him (quite far), another vehicle is preparing to turn to the left. The speed of the vehicle driving ahead of the driver at first depends on the subject’s, so that the distance between the two vehicles is 25 m. Suddenly, this coupling is broken and the obstacle vehicle slows down to a speed of 10 km/h after 1.9 s. “Opposite vehicle crossing” scenario (opposite vehicle). The subject is driving in a straight line on a carriageway with two lanes separated by central marking and carrying two-way traffic. Another vehicle at some distance in front of the driver gets ready to turn left into a petrol station. This vehicle starts to cross the opposite lane when the subject’s vehicle is at a distance of 25 m from the potential point of impact. It executes its manoeuvre at low speed, 8 km/h. Note that to avoid any learning effect during the experimental session, participants were largely trained to the urban circuit and to the different scenarios. 3) 15 min of following a car which can accelerate and deccelerate (+/- 10 or 20 km/h) and can adopt three speeds, 70, 80 or 90 km/h during 20, 30 or 40 s. Subjects’task was to maintain a safe distance between this car and their own vehicle.

Subjective tests After each driving session subjects were asked to rate the Karolinska Sleepiness Scale (KSS) and their self- reported quality of driving performance. Visual Analog Scale (VAS) was filled in four times. Statistics Dependent variables varied in function of the test (see below). Analysis of data was performed with ANOVA to evaluate effects of treatment. For each significant effect of the treatment, a Dunnett’s post hoc test, a Tuckey’s post hoc test or a Wilcoxon test was applied. P-values < .05 were considered statistically significant. For monotonous situation, we analysed the standard deviation of the lateral position (SDLP, in meters), standard deviation of speed (SDS, in km/h), mean lateral position (mLP), mean speed (mS) and number of road exit. For urban situation with accident scenarios, number of collision, response time and mean speed at the start point of the scenarios were analysed. For car following situation, SDLP and response time to the changing speed of the followed vehicle were analysed. Results

Monotonous driving test SDLP significantly varied as a function of treatment [F(16, 3)=3.28, p<.05]. It was higher - with zolpidem at night and placebo at morning (zolp/pla, m = 0.496 m) and - with placebo at night and codeine-paracetamol at

122 DRUID 6th framework programme Deliverable 1.2.2 morning (pla/cod-para, m = 0.503 m) than with placebo at night and at morning (pla/pla, m = 0.451 m) (respectively: p<.05 and p<.05). SDLP did not vary significantly after the intake of the two active molecules relatively to placebo treatment (p=.13, m = 0.486 m) (Figure 1).

*

0.7 *

0.6

0.5

0.4 Amplitude des SDLP (m) SDLP des Amplitude SDLP (cm) SDLP 0.3

0.2 Pla/Pla Zolpi/Pla Pla/cod-para Zolpi/cod-para

TypeTreatment de traitements condition administrés * : p<.05

Figure 1: Monotonous test. Mean Standard Deviation of Lateral Position (SDLP) in each treatment condition.

The number of “road exits” was also significantly affected by the treatment (F (3, 16)=8.66, p<.05). This number was significantly higher after each treatment than after placebo (respectively Pla/Pla versus Zolpi/Pla T=5,5, Z=2,63, p=.0085 ; Pla/Pla versus Pla/cod-para T=5, Z=2,67, p=.008 ; Pla/Pla versus Zolpi/cod-para T=10,5, Z=2, p=.045) (Figure 2).

*

*

* 60

50

40

30

20 Number of Number road exit

10

0 Pla/Pla Zolpi/Pla Pla/cod-para Zolpi/cod-para

traitmentTreatment condition * : p<.05

123 DRUID 6th framework programme Deliverable 1.2.2 Figure 2: Monotonous test. Mean number of road exit in each treatment condition.

Mean speed and standard deviation of speed did not significantly vary between treatment (respectively F(3, 42)=.83, p=.48; F(3, 39)=2.09, p=.11) (Figure 3).

113 10

112 8

111 6

Amplitude SDLP

110 4 Average speed (km/h)

109 2 Standard Deviation of (SDSP) Speed (km/h) 108 0 Pla/Pla Zolpi/Pla Pla/cod-para Zolpi/cod-para TraitementTreatment condition Average speed SDSP

Figure 3: Monotonous test. Mean average speed and Mean SDSP as a function of treatment.

Urban driving test A total of 13 crashes were noted across all scenarios and the combination of placebo and codéine/paracétamol produced the higher number of collisions (Table 1). These collisions are largely due to seven females, only three males collided. Number of Zolp/cod- Pla/Pla Zolp/Pla Pla/cod-para Total collisions para 2 2 6 3 13

Table 1: Urban test. Number of collisions as a function of treatment.

ANOVA ([G2]*treatment) (G2 corresponded to the sex of the participant) did not show any effect of treatment on response time (RT) (F(3,42)=.73, p=.54) but males always react more rapidly than females (F(1,14)=6.02, p=.028) (Figure 4).

124 DRUID 6th framework programme Deliverable 1.2.2 2

1.8 1.6

1.4 1.2 fema le 1 male 0.8 0.6 Response time (s) time Response 0.4

0.2

0 pla/pla zolp /p la p la/cod-para zolp/cod-para Treatment

Figure 4. Urban test. Mean response time as a function of treatment and sex.

ANOVA [G2]*scenario*treatment showed a scenario effect (F (5,70)=30.63, p<.001) and an interaction treatment and sex (F (3,42)=4.67, p<.05) on mean speed at the moment origin of the scenarios. The overtaking scenario produced the highest speed, lowest speed were produce by parking and traffic light scenario (p<.01). Mean speed adopted by males was significantly higher with zolp/cod-para treatment than with pla/pla treatment, conversely mean speed adopted by females was lower with zolp/cod-para treatment than with pla/pla treatment (Figure 5).

60

50

40

male 30 female

20 Mean speed (km/h)

10

0 pla/pla zolp/pla pla/cod-para zolp/cod-para Treatment

Figure 5. Urban test. Mean response time as a function of treatment and sex.

Car following test Response time to the variation of speed of the followed car did not show any effect of sex or treatment (respectively ANOVA ([G2]*treatment; F(3,42)=0.54, p<.66 et F(1,14)=.69, p<.42) (Figure 6).

125 DRUID 6th framework programme Deliverable 1.2.2 7

6

5

4 female 3 male

2 Response (s) time 1

0 pla/pla pla/zolp pla/cod-para zolp/cod-para Treatment

Figure 6. Car following test. Mean response time as a function of sex and treatment.

SDLP only tended to vary as a function of treatment as showed by ANOVA ([G2]*treatment; F(3,42)=2.39, p<.08). A contrast analysis showed that Zolp/cod-para involved higher SDLP than pla/pla (F(1,14)=4.77, p<.046, figure 7).

60

50

40

female 30 male

SDLP (cm) 20

10

0 p la/p la pla/zolp pla/cod-para zolp/cod-para Treatment

Figure 7. Car following test. Mean SDLP as a function of sex and treatment.

KSS subjective test Results showed that drug/pharmacological condition (Zolpi/Pla, Pla/cod-para, Zolpi/cod-para) did not differ from the Pla/Pla condition (F(16,3)=0,094, p=.99; figure 8).

126 DRUID 6th framework programme Deliverable 1.2.2 8

6

4 KSS evaluationKSS

2

Pla/Pla Zolpi/Pla Pla/cod-para Zolpi/cod-para Treatment Figure 8. Average ratings on the KSS after one hour of monotonous driving. Visual Analog Scale (VAS)

Average ratings of drug/pharmacological condition did not significantly differ from the Pla/Pla condition (F(3,45)=.53, p=.66) but differed as a function of the hour (F(3,45)=4.74, p<.005; figure 9). Tuckey’s post hoc test showed higher score value after driving test than before driving test (VAS 1 vs VAS 2; p=.01) and higher value before driving test than three hours after driving test (VAS 1 vs. VAS 4; p=.001). There was no interaction between treatment and time on VAS (F(9,135)=.35, p=.95).

** 6 **

5

4

3 VAS ScoreVAS 2 Driving simulator task

1

0 VAS 1 - 9h VAS 2 - 11h VAS 3 - 12h VAS 4 - 13h Repeating VAS - time ** : p<.005

Figure 9. Average value on the VAS scores as a function of the time. Self-reported quality of driving performance.

Results did not show any significant effect of the treatment on self-reported quality scale performed one hour after monotonous driving test (F(16,3)=. 093, p=.99; Figure 10).

127 DRUID 6th framework programme Deliverable 1.2.2 4

2

0

-2

-4

Scores on self-reported quality of driving performance driving of quality self-reported on Scores normally" drove "I 0= Pla/Pla Zolpi/Pla Pla/cod-para Zolpi/cod-para Treatment Figure 10. Average scores of self-reported quality of driving performance as a function of treatment.

Discussion

The main objective of the present study was to evaluate, in healthy elderly subjects (55 to 65 years old), residual effects on driving performance of an hypnotic zolpidem (10mg), acute effect of an analgesic codeine/paracetamol (20mg/400mg), and effects of an association of these two treatments to assess a potential increased effect of this combination. Three driving tests were used: monotonous, urban and car following. In monotonous test, results showed that SDLP, a measure known to be sensible to the residual and acute effects of licit and illicit drugs, increased with a nighttime intake of zolpidem or a morning time intake of codeine/paracetamol but did not significantly vary with an intake of these two drugs. Conversely the number or road exits increased whatever the treatment and in car following test SDLP only increased after the intake of the two active molecules. In light of the results obtained with monotonous driving, zolpidem alone does thus not appear to be completely devoid of residual effects as it is usually described with highway driving, but studies were done with healthy young subjects or on middle-aged insomniac patients. Our previous studies, conducted with older middle-age drivers under zolpidem, already showed driving performance impairment (SDLPs significantly increased) in highway situations (Bocca et al., 2011). These results are also in line with some recent epidemiological data suggesting that zolpidem could not be devoid of residual effects (Gustaven et al. 2008). Speed and variation of speed, measured during one hour of monotonous driving, were not affected by treatment. Response times, measured in urban and in car following tests, was not affected by treatment but we noted that men reacted more quickly than women when confronted to sudden events in urban test. Crashes were sparse during urban tests, and statistical analysis was not possible due to the size of the sample. Note also that participants had a good knowledge of the different scenarios, acquired during the training runs, which can have activated a process of anticipation of their appearance and reduced the potential number of collisions. In another hand, in urban test, a nighttime intake of zolpidem with a morning intake of codeine-paracetamol produced differential effects on mean speed adopted by men and women: men increased their speed, conversely women diminished their speed. Finally, treatment seemed to be devoid of significant effect on scores of subjective tests. Only visual analog

128 DRUID 6th framework programme Deliverable 1.2.2 scores varied as a function of time. While these results cannot be straightforwardly interpreted they show a residual effect of zolpidem and also a combined effect of hypnotic and antalgic on driving performance. These effects would be more effective in monotonous and car following test than in urban test.

Acknowledgements This work was conducted as part of the Driving under the influence of drugs, alcohol, and medicines (DRUID) research consortium funded by European Union grant TREN-05- FP6TRS07.61320-518404- DRUID. This report reflects only the authors’view. The EuropeanCommunity is not liable for any use of the information contained herein. None of the authors have any financial or other relationships that might lead to a conflict of interest.

References Amato JN, Marie S, Berthelon C, Lelong-Boulouard V, Denise P, Bocca ML (submitted) Effects of Analgesics doses: Up to 3 doses of the association of codein and paracetamol did not impair driving performance in healthy young volunteers. Barbone F, Mac Mahon A, Davey P, Morris A, Reid I, MacDevitt D, MacDonald T (1998) Association of road- traffic accidents with benzodiazepine use. Lancet 352 (9137): 1331-1336 Bachs LC, Engeland A, Morland JG, Skurtveit S (2009) The risk of motor vehicle accidents involving drivers with prescriptions for codeine or tramadol. Clin Pharmacol Ther 85: 596-9 Bocca ML, Marie S, Lelong-Boulouard V, Bertran F, Couque C, Desfemmes T, Berthelon C, Amato JN, Moessinger M, Paillet-Loilier M, Coquerel A, Denise P (2011) Zolpidem and zopiclone impair similarly monotonous driving performance after a single nighttime intake in aged subjects. Psychopharmacology 214: 699–706 Dubois S, Bedard M, Weaver B (2010) The association between opioid analgesics and unsafe driving actions preceding fatal crashes. Accid Anal Prev 42: 30-7 Engeland A, Skurtveit S, Morland J (2007) Risk of road traffic accidents associated with the prescription of drugs: a registry-based cohort study. Ann Epidemiol 17: 597-602 Fishbain DA, Cutler RB, Rosomoff HL, Rosomoff RS (2002) Can patients taking opioids drive safely? A structured evidence-based review. J Pain Palliat Care Pharmacother 16: 9-28 Gjerde H, Normann PT, Christophersen AS, Samuelsen SO, Mørland J (in press) Alcohol, psychoactive drugs and fatal road traffic accidents in Norway: A case–control study. Accident Analysis & Prevention, In Press, Available online 21 January 2011 Glass J, Lanctot KL, Herrmann N, Sproule BA, Busto UE. (2005) Sedative hypnotics in older people with insomnia: meta-analysis of risks and benefits. Bmj. 19;331(7526):1169. Gustavsen I, Bramness JG, Skurtveit S, Engeland A, Neutel I, Mørland J. (2008) Road traffic accident risk related to prescriptions of the hypnotics zopiclone, zolpidem, flunitrazepam and nitrazepam. Sleep Med 9(8):818-22 Kelly E, Darke S, Ross J (2004) A review of drug use and driving: epidemiology, impairment, risk factors and risk perceptions. Drug Alcohol Rev 23: 319-44 McGwin G, Sims RV, Pulley L, Roseman JM (2000) Relations among chronic medical conditions,

129 DRUID 6th framework programme Deliverable 1.2.2 medication, and automobile crashes in the elderly: a population-based case-control study. American Journal of Epidemiology 152: 424-431 Maycock G, (2000) Forecasting older-driver accidents and casualties. Road Safety Research Report N°23, DLTR Movig KLL, Mathijssen MPM, Nagel PHA, van Egmondd T, de Gier JJ, Leufkens HGM, Egberts ACG (2004) Psychoactive substance use and the risk of motor vehicle accidents. Accident Analysis and Prevention 36: 631–636 Neutel I (1998) Benzodiazepine-related traffic accidents in young and elderly drivers. Human Psychopharmacology 13 (suppl. 2) : S115-S123 Legrain S (2005) Prescription médicamenteuse du sujet âgé. EMC-Médecine 2 : 127–136 Meskali M, Berthelon C, Marie S, Denise P, Bocca ML (2009) Residual effects of hypnotics drugs in aging drivers submitted to simulated scenarios of accidents: an exploratory study. Psychopharmacology 207(3): 461-467 Orriols L, Delorme B, Gadegbeku B, Tricotel A, Contrand B, Laumon B, Salmi LR, Lagarde E; CESIR research group (2010) Prescription medicines and the risk of road traffic crashes: a French registry-based study. PLoS Med. November 7(11):e1000366. Vermeeren A (2004) Residual Effects of Hypnotics: Epidemiology and Clinical Implications. CNS Drugs 18(5): 297-328. Walsh JM, Gier JJ de, Christopherson AS, Verstraete A G (2004) 'Drugs and Driving', Traffic Injury Prevention 5(3): 241 – 253

130 DRUID 6th framework programme Deliverable 1.2.2 Chapter 6 : Acute effects of 3 doses of analgesics on simulated driving performance in healthy volunteers

ML Bocca (UNICAEN), P Denise (UNICAEN), C Berthelon (INRETS)

131 DRUID 6th framework programme Deliverable 1.2.2 Abstract

The objective of this study was to evaluate the dose-effect relationship of the usual therapeutic doses of codeine/paracetamol (20/400, 40/800 and 60/1200 mg) on driving ability, psychomotor performance, subjective alertness, in link with blood concentrations in 16 healthy young volunteers. Driving performance, psychomotor vigilance tests and scales reflecting alertness were evaluated during the morning after drug intake in a double-blind, randomized, placebo-controlled study. Two blood samples were collected, 1 hour and 4 hours after drug intake. None of the classical parameters used to quantify the effects of drugs on driving and on psychomotor performance were affected by any of the three codeine/paracetamol doses. However, significant correlations were observed between the driving parameters and both morphine and codeine blood concentrations. The relationships between blood concentration and behavioral measures indicated that the inter-subject variability in drug effects, possibly linked to genetic factors, is hidden when analysis by treatment is used.

132 DRUID 6th framework programme Deliverable 1.2.2 Introduction Among factors suspected to increase the risk of road accidents, drug consumption is regularly implicated (Bachs et al. 2009; Chesher 1985; Engeland et al. 2007; Kelly et al. 2004; Leveille et al. 1994). Narcotic analgesics are among the most prescribed drugs and they appear to be involved in between 3 % of the accidents in Europe (De Gier 1995; Gaulier et al. 2003) and 10 % in the USA (Dubois et al. 2010). Several epidemiological studies have found low relative risk for opioid users (Fishbain et al. 2002; Kelly et al. 2004), but some recent studies have shown that drivers using natural opium alkaloids have an increased risk of being involved in an accident (Bachs et al. 2009; Dubois et al. 2010; Engeland et al. 2007; Kelly et al. 2004). Among analgesics, codeine is a light opioid of level II which is often associated with paracetamol, an analgesic of level I. These two drugs have synergetic activity in reducing moderate to strong pain (Quiding et al. 1982). In some studies, codeine is associated with an elevated risk of accidents (Bachs et al. 2009; Engeland et al. 2007). As codeine reduces vigilance, these accidents could be linked to the sedative effects of codeine (Leveille et al. 1994). Indeed, reduced vigilance and falling asleep while driving are the most common and severe causes of accidents (Schmidt et al. 2009). Pharmacological properties of opioids, particularly codeine, are well-known (Caraco et al. 1999; Kim et al. 2002; Lotsch et al. 2006; Somogyi et al. 2007). Codeine interacts with mu, delta and kappa opioid receptors in the CNS, which play an important role in the control of pain, particularly in the spinothalamocortical pathways (Rainville 2002). The effects of codeine on the CNS depend on the dose (Quiding et al. 1982), the route of administration and previous exposure (Huestis and Smith 2009). Codeine, like morphine but to a lesser degree, also has a depressive effect on the respiratory center and alertness. Experimental studies are complementary to the epidemiological studies to evaluate the risks of accidents following drug intake. With performance tests, several studies have been performed to evaluate the effects of codeine on driving-related skills and have shown that some impairments occurred with high doses (60-100 mg) (for review see (Zacny 1995)). Only one study has evaluated the effects of a single dose of 50 mg of codeine on driving performance with a simulator in young healthy subjects (Linnoila and Hakkinen 1974). The authors showed that 50 mg of codeine increased the risk of collisions in two driving tests, i.e. in emergency situations and in a monotonous driving test. The 50 mg dose, used in this study, is a current dose used in patients for reducing pain. However, the common therapeutic dose in one intake can vary between 20 and 60 mg, which, in France, represents 1 to 3 tablets of codeine/paracetamol. The primary objective of this study was therefore to evaluate the effects of the most common 3 therapeutic doses of codeine/paracetamol on driving performance in healthy young subjects to assess a potential dose-effect relationship of this drug. As driving a motor vehicle requires the possession of sufficient cognitive, visual and motor skills and involves managing attention to perform various driving- and non-driving-related tasks (Michon 1985; Salvucci et al. 2007), the secondary objective was to assess the effects of the 3 doses on cognitive and psychomotor functioning related to driving and on subjective feelings. For this purpose we used the Psychomotor Vigilance Task (PVT) (Dinges and Powell 1985), based on a simple visual reaction time, which is a test of vigilance most often used in sleep deprivation studies. Assessment of subjective feelings in change of mood and driving performance estimation were evaluated with several Visual Analog Scales (VAS) (Akerstedt and Gillberg 1990; Bond and Lader 1974; Brookhuis et al. 1985). Blood concentration of codeine and paracetamol was also assessed 1 hour and 4 hours after taking an oral dose.

133 DRUID 6th framework programme Deliverable 1.2.2 Method

ҏSubjects

The study was carried out on 16 healthy volunteers (8 men and 8 women). Their average age was 22.37 years + 2.7, weight was 64.15 kg + 8.53, and height was 171.80 cm + 5.27. All participants were graduates from at least the first cycle of University. Exclusion criteria were sleep, alertness, neurological, cardiovascular, respiratory, hepatic, renal, or metabolic disorders, chronic or transitory use of oral medication except contraceptives during the 2 years prior to the experiment, cigarette consumption > 5 cigarettes per day, and alcohol consumption > 28 units per week. The study was approved by the Caen Northwest III ethics committee and French Health Products Safety Agency (number 060702), and each subject gave consent in accordance with the requirements of the committee. Our 16 subjects completed the Horne and Ostberg morningness-eveningness questionnaire (Horne and Ostberg 1976). Subjects did not have extreme scores when typed for fluctuations in morning or evening alertness (10 Intermediate types, 4 Morning types and 2 Evening types).

Study design Before the study, subjects were trained to drive the simulator during 30 mn. The study was conducted according to a balanced, double-blind, cross-over design. Each volunteer participated in four sessions held at intervals of at least 2 weeks, and received the following treatments during these sessions : 20 mg of codeine and 400 mg of paracetamol, 40 mg of codeine and 800 mg of paracetamol, or 60 mg of codeine and 1 200 mg of paracetamol, or a placebo. All drugs were administered in identical gelatin capsules and the subjects received the same number of capsules at each session. The experiments took place from Tuesday to Friday. Subjects performed their four sessions on the same day of the week and at the same time of the day in order to avoid any possible interference with changes in biological rhythms from one day of the week to the next. The subjects were asked not to drink coffee in the morning of the medication intake and to abstain from alcohol and sport for 24 h prior to testing. They took their usual breakfast in the morning of the test but abstained from smoking. There were two subjects per day. The medication was administered to the first subject at 8:00 AM and to the second at 9:00 AM in the laboratory under the supervision of an experimenter. The starting time of the experiment was 9:00 AM for the first subject and 10:00 AM for the second (1 hour after oral administration of the drugs, at the theoretical maximal concentration (Cmax) of codeine). Subjects completed several analogue subjective scales during the morning, before and after the driving and PVT test. After the driving test and the PVT, subjects remained in the laboratory for one hour before leaving. Consumption of benzodiazepines, opiates, cannabis, cocaine, amphetamines was screened at each session by urine control. No subjects were positive.

Driving and psychomotor tests The driving task took place with the FAROS driving simulator that reproduces several aspects of a current

134 DRUID 6th framework programme Deliverable 1.2.2 medium-sized car (dashboard, five-speed gearbox, pedals and steering wheel). The simulator produces engine and tire-squeal sound effects that correlate with the speed of the vehicle but has no system that enables simulation of the movements of the car. The cabin was associated with an interactive display unit which reproduced the motorway scenery with computer-generated pictures on a 1.3x1.7 meter screen located in front of the cabin at 2.3 meters from the subject (Meskali et al. 2009). The test involved driving for 60 minutes during the day along a motorway in monotonous conditions (no traffic, occasional long, wide curves, and repetitive landscape). Subjects were instructed to drive on the right and to respect the speed limit set at 110 km/h during the complete test session. They had to drive as much as possible in the center of the right lane and at constant speed. The driving task used for this study was a monotonous motorway, yet it was realistic in order to maintain ecological validity. Driving parameters were collected with a 30 Hz frequency.

After completion of the driving test, a program calculated the mean speed (km/h) and the standard deviation of speed (SD speed, km/h), the standard deviation of the lateral position (SDLP, m) and the number of line crossings (n). The SDLP was described as the most sensitive measure for evaluating drug effects in real driving and is an index of vehicle control (O'Hanlon et al. 1982). We previously used the FAROS simulator for evaluating monotonous driving after medication intake and have shown that the simulator was sensitive enough to reveal drug impairments in monotonous conditions (Bocca et al. online).

Psychomotor vigilance task This is a simple RT test to assess sustained attention with a cue that occurs at random inter-stimulus intervals (ISI).The ISI varied randomly from 2 to 10 s. The standard test is 10 minutes in duration. During this time, subjects were seated comfortably and instructed to respond by pushing a button when a small circular area on a dark screen appeared. They were told to respond as rapidly as they could whenever they perceived the circular area. After their response, the reaction time appeared in a bright millisecond counter inside this circular area and remained on the screen for the duration of 1 second, which served as feedback. The percentage of lapses (number of responses greater than 500 ms/number of total stimuli), which is a parameter often used as the primary dependent variable in the test, was analyzed. The mean RT (RT, ms) and the variability in RT (SD RT, ms) were also analyzed to study the sedative effects of the compounds.

Subjective measures: This study employed multi- objective and subjective measures to characterize variations of driver fatigue and sedative effects of analgesics. After the driving task, all subjects rated their sleepiness on the Karolinska Sleepiness Scale (KSS). This scale is a 9-point verbally-anchored scale with the following steps: (1) very alert, (3) alert, (5) neither alert nor sleepy, (7) sleepy, and (9) very sleepy, fighting sleep (Akerstedt and Gillberg 1990). The other points represent intermediate stages between the two neighboring points without definitions. Each driver’s feeling parameters were recorded with “subjective driving quality”, using a continuous scale for self-rating their driving quality (Brookhuis et al. 1985). This scale consisted of a 100 mm vertical visual analog scale at three levels: (1) “I drove exceptionally poorly”at 0 mm, (2) “I drove normally”at 50 mm and (3) “I drove exceptionally well”at 100 mm. Responses were given on a visual analog scale of 100 mm with

135 DRUID 6th framework programme Deliverable 1.2.2 an “x” at the appropriate level. If the response was near ±50 mm the subjects estimated that they drove normally and safety. The subjects rated their subjective feelings in change of mood with a VAS (Bond and Lader 1974). We used the 9-visual analog scale items grouped into factor 1, reflecting “alertness”(Bond and Lader 1974) since we focused on the sedative effects of codeine/paracetamol drug. The nine 100 mm scales were given to each subject every hour during the experiment: when they arrived at the laboratory (baseline level: 8:00 or 9:00 AM), 1 hour after the medication intake (+1 hour: 9:00 or 10:00 AM), after the driving test (+2 hours: 10:00 or 11:00 AM) and at the end of the morning test (+3 hours: 12:00 or 1:00 PM). The subjects placed a mark on a horizontal line equivalent to the strength of a particular feeling at that time. The scales were scored by measuring from the end of the line to the subject’s mark in centimeters. The mean of the 9 values was calculated to reflect the “alertness factor”.

Pharmacokinetic assessment Concentrations of paracetamol, codeine and morphine which is one of the main metabolites of codeine (O'Neal et al. 1999), were measured in serum. The blood sample collected was 7,5 mL. Morphine was quantified because it is an active metabolite which is at least in part responsible for the analgesic activity of codeine. Moreover, this quantification is an indication of the presence of possible poor metabolizers in our selected subjects (Alvan et al. 1990). Two blood samples were collected; the first (T1) 1 hour after the oral administration (at the theoretical Cmax) and the second (T2) at the end of the experiment (i.e. 4 hours after the oral administration). Codeine and morphine were determined in serum after simple liquid-liquid extraction and were quantified by high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (HPLC-ESI- MS/MS) using a Shimadzu (Columbia, MD) Prominence HPLC system interfaced on an AB SCIEX 3200 QTRAP (Concorde, Ontario, Canada), with a TurboIonSpray™ source in positive ionization mode (adapted to (Coles et al. 2007) method). The cut-off was 1 ȝg/L, for both drugs. Paracetamol was quantified by fluorescence polarization immunoassay (Abbott, France) and the cut-off was 1 mg/L.

Statisticsҏ The statistical analyses were performed with STATISTICA software version 7.1 (StatSoft®). For SDLP, mean speed, SD speed, subjective scales (KSS, driving quality) a one-way ANOVA for repeated measure was performed in order to compare the 4 treatments. Post hoc comparisons to placebo were assessed with Dunnett’s test to evaluate each treatment effect. The effect of the treatments on the number of line crossings was tested with a non parametric Friedman ANOVA. A two-way repeated measures ANOVA with treatment (4 levels) by time (3 times) was used for the subjective alertness factor (only post ingestion values were included in this analysis). In the case of significant time effect, a post hoc Fisher’s LSD was performed. Linear regressions between performance task parameters and serum levels of codeine, morphine and paracetamol were calculated. P-values < 0.05 were considered statistically significant.

136 DRUID 6th framework programme Deliverable 1.2.2 Results

Plasma concentration of drugs after oral administration of codeine and paracetamol (table 1). There are no significant differences between T1 and T2 for each drug.

Driving tests Overall treatment effect The repeated measures ANOVA did not reveal any significant effect of codeine/paracetamol on the SDLP (F(3, 45)=0.60, p=0.61), on the mean speed (F(3, 45)=0.49, p=0.68) or on the SD speed (F(3,45)=1.72, p=0.17). The only significant result was found with the mean LP (F(3,45)=3.55, p=0.02), with a significant difference between the 20 mg dose and placebo (p < 0.006). Non parametric ANOVA performed on the number of road exits did not reveal any significant effect (F(3,45)=2.77, p=0.42), however, a significant increase was observed with the 40 mg dose compared to placebo (p < 0.04) if separate Wilcoxon tests were performed.

137 DRUID 6th framework programme Deliverable 1.2.2 Psychomotor Vigilance Test (PVT): (Table 2) The ANOVA revealed no significant influence of codeine/paracetamol on mean RT (F(3, 45)=0.88, p=0.45), on the SD RT (F(3, 45)=0.41, p=.74). The percentage of lapses did not differ statistically between doses of codeine/paracetamol or placebo (F(3,45)=3.48, p=0.32). Subjective scales (Table 2 and figure 2) The ANOVA revealed a significant effect of codeine/paracetamol on the KSS score (F(3,45)=10.50, p=0.01). Dunnett’s test did not reveal any significant treatment effect for any of the 3 doses of codeine/paracetamol, Fisher’s LSD test revealed a significant difference between 20 mg and 40 mg (p=0.01). No significant effect was found for the self-rated quality of driving performance (F(3, 45)=5.11, p=0.16). It appears that subjects judged their own driving quality, on average, as positive and nearly normal. For the alertness factor, the ANOVA did not show any treatment effect (F(3, 45)=1.86, p=0.14), but did show a time effect (F(2, 30)=11.05, p=0.00025). No treatment*time interaction was found (F(6, 90)=1.08, p=0.37). The significant time effect was seen in the significant difference in the alertness factor values between the +1 hour (just before the driving test) and both the + 2 hours (just after the driving test) and + 3 hours (at the end of the experiment) (P < 0.005) (Figure 2).

138 DRUID 6th framework programme Deliverable 1.2.2 Correlations between behavioral measures and blood concentrations Multiple correlations were performed between the different variables and the blood concentrations of codeine, morphine and paracetamol. As the driving test followed the T1 blood samples, the driving parameters were correlated with blood sample concentrations collected at T1. As the PVT was performed at the end of the morning and just before the T2 blood samples, the PVT parameters were correlated with blood sample concentrations collected at T2. Interestingly, a trend was found between SDLP changes and codeine T1 (r=0.27; p=0.07) and morphine T1 (r=0.33, p = 0.08) (Figures 3AB). The mean speed changes was also significantly and negatively correlated with morphine T1 (r=0.51, p<0.005) (Figure 3C).

139 DRUID 6th framework programme Deliverable 1.2.2 140

8 120 A

7 100 B T1 r = 0.27 6 80 p = 0.07

5 60 4

[Codeine] T1 40

[Morphine] T1 3 20 2 T1 r = 0.33 p = 0.07 0 1

0 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 SDLP changes SDLP changes

8

7

6 T1 r = 0.51 p = 0.005 5

4

C

[Morphine] T1 [Morphine] 3

2

1

0 -3 -2 -1 0 1 2 Mean speed changes

Figures 3 A, B, C. Scatter plots showing relationships between (A) morphine concentration at T1 and SDLP changes, (B) codeine concentration at T1 and SDLP changes, (C) morphine concentration at T1 and mean speed changes. SDLP: standard deviation of lateral position.

No significant correlations were found between the PVT parameters and the blood sample concentrations collected at T2.

Discussion

This study was designed to examine the effects of three growing therapeutic doses of codeine/paracetamol on the parameters used to quantify simulated driving performance. Our study did not reveal any effects of codeine/paracetamol, at the 3 doses tested, on the SDLP, the most sensitive parameter used to investigate drug effects on driving performance (O'Hanlon et al. 1982). Moreover, the other driving parameters, i.e. speed variability and the number of road exits were also unaffected. The only modification of behavior was observed on the mean lateral position and with the lowest dose. The lateral position to the left of the road lane in comparison to placebo could be hypothetized as a sort of safety margin used by the subjects when

140 DRUID 6th framework programme Deliverable 1.2.2 driving. Such behavior has already been observed after hypnotic use in urban driving (Meskali et al. 2009). Overall results suggested that up to 60 mg codeine and 1 200 mg paracetamol in a single intake, codeine/paracetamol drug did not impair monotonous driving performance in healthy young subjects. Among the different parameters studied, only one can be compared with the previous study (Linnoila and Hakkinen 1974). The authors did not find any effect of the 50 mg of codeine on the mean speed during a 40 min monotonous driving test. This can be compared to our monotonous driving test although they included accidents during this task contrarly to our test. Our results agreed with this study regarding the monotonous driving test. However, (Linnoila and Hakkinen 1974) found that the 50 mg dose of codeine increased the risk of accidents in emergency situations. As our objective was to determine a dose-effect relationship with the standard and sensitive test used in the field of assessment of drugs on driving performance (Vermeeren 2004), i.e. the monotonous driving test, previously used in our laboratory (Bocca et al. Minor revision; Bocca et al. 1999), we did not perform such an evaluation of behavior, and it is thus difficult to really compare these studies. Nevertheless, as we have developed a methodology to evaluate accident scenarios in urban driving with a simulator, and have shown its usefulness (Meskali et al. 2009), it would be interesting to evaluate the effects of different doses of codeine on such situations in a further study. Consequently, we can conclude that, in our study, the 3 doses of codeine/paracetamol did not impair monotonous driving performance in healthy young subjects. The absence of a codeine/paracetamol effect on driving performance is supported by the similar absence of impairment of performance in the PVT and on the subjective measure to characterize variations in driver fatigue and sedative effect of analgesics. All of these results indicated, despite the high sensitivity of the PVT test in various conditions (sleep loss, age, caffeine, etc) (Dinges and Powell 1985; Lim and Dinges 2008), that codeine had no effect on arousal and attentional state at the doses used, which could explain the absence of driving performance impairment in the monotonous driving test. Moreover, the absence of effects on the subjective feelings of alertness, KSS and driving assessment, revealed that the subjects felt well, and judged their own driving quality as nearly normal. Only a significant difference between the two doses of codeine 40 and 60 mg was observed with the KSS, which is difficult to interpret. All of these results are in agreement with previous findings (Walker and Zacny 1998). Indeed, this study evaluated the effects of 60 and 120 mg of codeine and did not find any effect of codeine on VAS, reflecting subjective feelings and various psychomotor performance tests. Our results suggest that codeine at therapeutic doses, i.e. up to 60 mg in a single intake in healthy young subjects, did not impair both driving performance and vigilance. The serum concentrations collected at the theoretical Cmax and at the end of the morning sessions served to evaluate the elimination course of codeine/paracetamol and codeine metabolites, i.e. morphine, and helped to interpret our behavioral results. Firstly, as morphine was found in all samples, the absence of impairment with codeine doses could not be linked to the fact that our subjects might be poor metabolizers. Secondly, as codeine concentrations were found to be higher at the end of the experiment (4 hours after the oral administration) than 1 hour after oral administration (theoretical Cmax) although no significant difference between T1 and T2 concentrations was found, this suggests that the effect of codeine may have not been maximal during the driving test performed during the 1 hour just after the first sample. This delay could be due to the fact that codeine/paracetamol was ingested in capsules for blind ingestion, which could delay the maximal concentration observed in the classical galenic form, i.e. pill. This hypothesis is also supported by the time effect found in the alertness factor, with alertness decreasing across the morning. Nevertheless, as

141 DRUID 6th framework programme Deliverable 1.2.2 subjects performed two tests consecutively during the morning session (driving and PVT), we cannot exclude the possibility that the decreased alertness could also be due to an increase in fatigue. Thirdly, despite non significant effects of the 3 doses of codeine/paracetamol tested in this study, some significant correlations were found between behavioral measures and blood concentrations of codeine or morphine. Such relationships have been described with alcohol, with increasing impairment of driving performance with increasing dose (Louwerens et al. 1987; Verster et al. 2009). It should be noted that positive trend were found between SDLP changes and both morphine and codeine concentrations. This suggests that higher concentrations of morphine and codeine corresponded to the subjects driving outside the road. Moreover, these relationships indicate that the more the subjects drove outside the road the more they reduced their speed. This could be a safety mechanism developed by the subjects. Our results also indicate that evaluating the effects of treatments on performance did not reflect the individual variations in drug pharmacokinetics. Indeed, on average, the different doses of codeine/paracetamol did not impair the driving performance even though subjects with higher concentrations of morphine or codeine in the blood had slightly more impaired driving performance. This result is interesting since codeine is metabolized into morphine via the CYP2D6, which is subject to genetic polymorphism (Sindrup and Brosen 1995). We can hypothetize that, for a small fraction of the population who are fast metabolizers, the proportion of morphine in the blood will be greater and thus should increase the risk of accidents. It would be interesting to evaluate if the genetic differences in codeine metabolism could determine the extent of the impairment, by comparing the genotype and the phenotype (ratio morphine/codeine) of the volunteers. However, due to a small number of subjects (n=16), this hypothesis cannot be verified in the present study. Interestingly, we found correlations with morphine at T1 although the morphine concentration should not be maximal at this time, probably due to the fact that morphine is a metabolite of codeine. The correlations revealed that although the level of morphine is low, in agreement with (Shah and Mason 1990) results, a link between morphine level and performance can be observed. Stronger relationships need to be hypothesized between driving parameters and morphine if blood samples are collected after T1. Interestingly, we also performed correlations for each dose and found that only with the 60 mg dose positive and significant correlations were found between SDLP difference and both morphine and codeine (p<0.05, for both). These results suggest that with higher doses (greater than 60 mg) behavioral impairments are likely to be observed. Finally, these correlations suggest that both morphine and codeine seem to have a concentration-dependent effect that may lead to impairment of driving performance. These results may be compared to those of Bachr et al (Bachs et al. 2003), who suggested that codeine effects may be independent from those of morphine but their cut-off value of morphine concentration was higher than ours (15 ng/mL in the study of Bachs et al. (2009) and 1 ng/mL in the present study). Our results indicate that morphine may lead to impairments even at low levels. Contrary to the driving parameters, no correlations were found between the 3 PVT parameters (RT, SDRT and % of lapses) and codeine/morphine/paracetamol T2 concentrations. This absence of relationships is due to the fact that PVT was performed at the end of the morning, 2.5 hours after capsule intake. Consequently, the concentrations in serum collected at T2 could not reflect the drug effects measured 1.5 hours before the samples were collected. Another study was performed to assess the effects of another opioid analgesic using a methodology similar to that used in our study (Verster et al. 2006). The authors revealed that oxycodone/paracetamol did not

142 DRUID 6th framework programme Deliverable 1.2.2 impair monotonous driving performance, the psychomotor task or the mood. Although oxycodone is a different compound than that tested in our study, it is similar to codeine in that it reduces pain by acting centrally, i.e. on the central nervous system. Our results, in addition to a previous study (Verster et al. 2006), could provide information for general practioners, pharmacists and patients. This information suggests that, to date, no noticeable driving impairment has been demonstrated after the use of single, usual doses of these opioid analgesics. Nevertheless, both studies were performed on healthy young subjects. Despite the fact that the choice of assessing a dose-effect relationship in young subjects was justified by the fact that this is the classic methodology to evaluate drug effects and that 26 percent opioid consumers are under 25 (Cadet-Taïrou et al. 2008), it would be useful to assess the effects of codeine in aged subjects before generalizing these results. Indeed, it has been shown that with age the amount of mu receptors decreases and the affinity increases, leading to an increased sensitivity to opioids (Wilder-Smith 2005).

In conclusion, our study, which was performed to establish a dose-effect relationship with the usual therapeutic doses of codeine/paracetamol in a single intake in healthy young subjects, did not reveal any impairment of driving and vigilance. From a road safety point of view, it must be noted that a concordance between feelings of driving quality and objective measures of performance was found. Additional research is needed to assess the effects on other critical driving situations such as urban accident scenarios, on aged drivers, and at higher doses since linear relationships have been observed between behavioral measures and blood concentrations of codeine, morphine and paracetamol.

Acknowledgements This work was conducted as part of the Driving under the influence of drugs, alcohol, and medicines (DRUID) research consortium funded by European Union grant TREN-05- FP6TR-S07.61320-518404- DRUID. This report reflects only the authors’view. The European Community is not liable for any use of the information contained herein. None of the authors have any financial or other relationships that might lead to a conflict of interest.

References

Akerstedt T, Gillberg M (1990) Subjective and objective sleepiness in the active individual. Int J Neurosci 52: 29-37 Alvan G, Bechtel P, Iselius L, Gundert-Remy U (1990) Hydroxylation polymorphisms of debrisoquine and mephenytoin in European populations. Eur J Clin Pharmacol 39: 533-7 Bachs L, Skurtveit S, Morland J (2003) Codeine and clinical impairment in samples in which morphine is not detected. Eur J Clin Pharmacol 58: 785-9 Bachs LC, Engeland A, Morland JG, Skurtveit S (2009) The risk of motor vehicle accidents involving drivers with prescriptions for codeine or tramadol. Clin Pharmacol Ther 85: 596-9 Bocca M, Marie S, Lelong-Boulouard V, Bertran F, Couque C, Desfemmes T, Berthelon C, Amato J, Moessinger M, Paillet-Loilier M, Coquerel A, Denise P (Minor revision) Zolpidem and zopiclone impair

143 DRUID 6th framework programme Deliverable 1.2.2 similarly monotonous driving performance after a signle nighttime intake in aged subjects. Psychopharmacology Bocca ML, Le Doze F, Etard O, Pottier M, L'Hoste J, Denise P (1999) Residual effect of zolpidem 10 mg and zopiclone 7.5 mg versus flunitrazepam 1 mg and placebo on driving performance and ocular saccades. Psychopharmacology 143: 373-9 Bond A, Lader M (1974) the use of analogue scales in rating subjective feelings. Br J Med Psychol 47: 211- 218 Brookhuis KA, De Vries G, Van Wijngaarden PP, Veenstra G, Hommes M, Louwerens JW, O'Hanlon JF (1985) The effects of increasing doses of (100, 200, 400 mg) and Glafenine (200mg) on driving performance. . Verkeerskundig Studiecentrum: Traffic Research Center, Haren, pp 40 Cadet-Taïrou A, Gandilhon M, Toufik A, Evrard I (2008) Phénomènes émergents liés au drogues en 2006/2007. Huitième rapport national du dispositif TREND. In: OFDT (ed). Observatoire Français des drogues et des toxicomanies, Saint Denis, pp 189 Caraco Y, Sheller J, Wood AJ (1999) Impact of ethnic origin and quinidine coadministration on codeine's disposition and pharmacodynamic effects. J Pharmacol Exp Ther 290: 413-22 Chesher GB (1985) The influence of analgesic drugs in road crashes. Accid Anal Prev 17: 303-9 Coles R, Kushnir MM, Nelson GJ, McMillin GA, Urry FM (2007) Simultaneous determination of codeine, morphine, hydrocodone, hydromorphone, oxycodone, and 6-acetylmorphine in urine, serum, plasma, whole blood, and meconium by LC-MS-MS. J Anal Toxicol 31: 1-14 De Gier JJ (1995) Drugs other than alcohol and driving in the European Union. In: Communities DGfTotCotE (ed). The institute for human psychopharmacology, The Netherlands Dinges DF, Powell JW (1985) Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. BRM 17: 652-655 Dubois S, Bedard M, Weaver B (2010) The association between opioid analgesics and unsafe driving actions preceding fatal crashes. Accid Anal Prev 42: 30-7 Engeland A, Skurtveit S, Morland J (2007) Risk of road traffic accidents associated with the prescription of drugs: a registry-based cohort study. Ann Epidemiol 17: 597-602 Fishbain DA, Cutler RB, Rosomoff HL, Rosomoff RS (2002) Can patients taking opioids drive safely? A structured evidence-based review. J Pain Palliat Care Pharmacother 16: 9-28 Gaulier JM, Marquet P, Lachatre G (2003) Traitements de substitution et conduite automobile. Opiate addiction treatment and driving. Ann Toxicol Anal 15: 123-127 Horne JA, Ostberg O (1976) A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol 4: 97-110 Huestis MA, Smith ML (2009) Pharmacokinetics and pharmacodynamics of drugs abused in driving. In: Verster JC, Pandi-Perumal SR, Ramaekers JG, De Gier JJ (eds) Drugs, Driving and Traffic Safety. Birkhäuser Verlag, switzerland, pp 151-185 Kelly E, Darke S, Ross J (2004) A review of drug use and driving: epidemiology, impairment, risk factors and risk perceptions. Drug Alcohol Rev 23: 319-44 Kim I, Barnes AJ, Oyler JM, Schepers R, Joseph RE, Jr., Cone EJ, Lafko D, Moolchan ET, Huestis MA (2002) Plasma and oral fluid pharmacokinetics and pharmacodynamics after oral codeine administration. Clin Chem 48: 1486-96

144 DRUID 6th framework programme Deliverable 1.2.2 Leveille SG, Buchner DM, Koepsell TD, McCloskey LW, Wolf ME, Wagner EH (1994) Psychoactive medications and injurious motor vehicle collisions involving older drivers. Epidemiology 5: 591-8 Lim J, Dinges DF (2008) Sleep deprivation and vigilant attention. Ann N Y Acad Sci 1129: 305-22 Linnoila M, Hakkinen S (1974) Effects of diazepam and codeine, alone and in combination with alcohol, on simulated driving. Clin Pharmacol Ther 15: 368-73 Lotsch J, Skarke C, Schmidt H, Rohrbacher M, Hofmann U, Schwab M, Geisslinger G (2006) Evidence for morphine-independent central nervous opioid effects after administration of codeine: contribution of other codeine metabolites. Clin Pharmacol Ther 79: 35-48 Louwerens JW, Gloerich ABM, De Vries G, Brookhuis KA, O'Hanlon JF (1987) The relationship between drivers' blood alcohol concentration (BAC) and actual driving performance during high speed travel. In: Noordzij PC, Roszbach, R. (Eds.) (ed) Alcohol, Drugs and Traffic Safety-T86. Excerpta Medica, Amsterdam, pp 183-186 Meskali M, Berthelon C, Marie S, Denise P, Bocca ML (2009) Residual effects of hypnotic drugs in aging drivers submitted to simulated accident scenarios: an exploratory study. Psychopharmacology 207: 461-467 Michon JA (1985) A critical view of driver behavior models: What do we know, what should we do? In: Evans L, Schwing RC (eds) Human behavior and traffic safety. Plenum Press, New-York, pp 485-520 O'Hanlon JF, Haak TW, Blaauw GJ, Riemersma JB (1982) Diazepam impairs lateral position control in highway driving. Science 217: 79-81 O'Neal CL, Crouch DJ, Rollins DE, Fatah A, Cheever ML (1999) Correlation of saliva codeine concentrations with plasma concentrations after oral codeine administration. J Anal Toxicol 23: 452-9 Quiding H, Persson G, Ahlstrom U, Bangens S, Hellem S, Johansson G, Jonsson E, Nordh PG (1982) Paracetamol plus supplementary doses of codeine. An analgesic study of repeated doses. Eur J Clin Pharmacol 23: 315-9 Rainville P (2002) Brain mechanisms of pain affect and pain modulation. Curr Opin Neurobiol 12: 195-204 Salvucci DD, Boer ER, Liu A (2007) Toward an Integrated Model of Driver Behavior in Cognitive Architecture. T R R: 9-16 Schmidt EA, Schrauf M, Simon M, Fritzsche M, Buchner A, Kincses WE (2009) Drivers' misjudgement of vigilance state during prolonged monotonous daytime driving. Accid Anal Prev 41: 1087-93 Shah JC, Mason WD (1990) Plasma codeine and morphine concentrations after a single oral dose of codeine phosphate. J Clin Pharmacol 30: 764-6 Sindrup SH, Brosen K (1995) The pharmacogenetics of codeine hypoalgesia. Pharmacogenetics 5: 335-46 Somogyi AA, Barratt DT, Coller JK (2007) Pharmacogenetics of opioids. Clin Pharmacol Ther 81: 429-44 Vermeeren A (2004) Residual effects of hypnotics: epidemiology and clinical implications. CNS Drugs 18: 297-328 Verster JC, Pandi-Perumal SR, Ramaekers JG, de Gier JJ (2009) Drugs, Driving and Traffic Safety Verster JC, Veldhuijzen DS, Volkerts ER (2006) Effects of an opioid (oxycodone/paracetamol) and an NSAID (bromfenac) on driving ability, memory functioning, psychomotor performance, pupil size, and mood. Clin J Pain 22: 499-504 Walker DJ, Zacny JP (1998) Subjective, psychomotor, and analgesic effects of oral codeine and morphine in healthy volunteers. Psychopharmacology 140: 191-201 Wilder-Smith OH (2005) Opioid use in the elderly. Eur J Pain 9: 137-40

145 DRUID 6th framework programme Deliverable 1.2.2 Zacny JP (1995) A review of the effects of opioids on psychomotor and cognitive functioning in humans. Experimental and Clinical Psychopharmacology 3: 432-466

146 DRUID 6th framework programme Deliverable 1.2.2 Chapter 7 : Dose related effects of dronabinol on actual driving performance of occasional and heavy cannabis users

W.M. Bosker*, A. Surinx*, R. Blankespoor*, J.G. Ramaekers*

Maastricht University, Faculty Psychology and Neuroscience, Dept Neuropsychology and Psychopharmacology, Experimental Psychopharmacology Unit, Maastricht, The Netherlands

Email: [email protected] Tel: +31 43 3881518 Fax: +31 43 3884560

147 DRUID 6th framework programme Deliverable 1.2.2 Abstract

Dronabinol (Marinol®) is a cannabinoid that is used for the treatment of chronic pain, anorexia in AIDS and other wasting diseases, and as an antiemetic medication in cancer patients undergoing chemotherapy. The active ingredient dronabinol is synthetic ¨9-tetrahydrocannabinol (THC), which is know to cause driving impairment when smoked. The effects of dronabinol on driving have never been assessed before. This study investigated the effects of dronabinol (Marinol®) on actual and simulated driving performance and the Field Sobriety Tests (FST). Participants, 12 occasional and 12 daily cannabis users, were administered 10 and 20 mg dronabinol and placebo according to a double-blind, randomized, 3-way cross-over design. Performance assessments consisted of actual driving tests and the FST. The aim was to assess the acute effects of THC on actual driving performance in occasional cannabis users and the degree of behavioural tolerance to such effects in heavy cannabis users. Results demonstrated that single doses of dronabinol impaired road tracking performance of occasional cannabis users during on-the-road driving tests in a dose related manner. Relative to placebo, SDLP significantly increased by approximately 2 and 3 cm respectively, after dronabinol 10 and 20 mg. The upper limits of the 95%CI associated with change SDLP exceed the alcohol criterion limit indicating that dronabinol effects after both doses on SDLP were comparable to the effects of a BAC=0.5 mg/ml. The effects of dronabinol on driving performance of heavy cannabis users however were less pronounced or even absent. Overall, dronabinol did not affect any driving measure as compared to placebo in superiority tests. Equivalence testing however indicated that the 95%CI associated with change scores in SDLP after dronabinol 10 and 20mg contained both the alcohol criterion. The latter basically indicates large individual variation in change SDLP after both doses dronanbinol. This suggest that tolerance to the impairing effects of dronabinol is not complete and may cause impairment in some regular users of cannabis but not in others. In general it is concluded that dronabinol impaired driving performance in occasional and heavy users in a dose-dependent way. Equivalence tests demonstrated that dronabinol induced increments in SDLP were bigger than impairment associated with BAC of 0.5 mg/mL in occasional and heavy users, although the magnitude of driving impairment was generally less in heavy users.

148 DRUID 6th framework programme Deliverable 1.2.2 Introduction

Cannabis is one of the most widely used drugs. Past month users were 15.2 million persons in 2008 in the US, which is about 6% of the general population aged 12 and older (Substance Abuse and Mental Health Services Administration, 2009). In Europe this number was estimated to be 12 million (3.6%) in recent years (European Monitoring Centre for Drugs and Drug Addiction, 2009). Because of its widespread use, the prevalence of cannabis in the general driving population is also one of the highest after alcohol. A recent European survey (European Monitoring Centre for Drugs and Drug Addiction, 2008) showed that 2.4% of the general driving population had ever driven under the influence of cannabis. However, amongst young drivers this was raised to 30%. This is a cause for concern, especially since about 10% of drivers involved in an accident were found to have cannabis in their blood. Several studies have shown that cannabis causes impairment in cognition, psychomotor performance and actual driving (e.g. Fernandez-Serrano, Perez-Garcia, & Verdejo-Garcia, 2010; Grotenhermen, et al., 2007; Lamers & Ramaekers, 2001; Ramaekers, Berghaus, van Laar, & Drummer, 2004; Ramaekers, Kauert, Theunissen, Toennes, & Moeller, 2009; Ramaekers, et al., 2006; Ramaekers, Robbe, & O'Hanlon, 2000). The actual driving impairment caused by cannabis is generally equal to or greater than that caused by alcohol at a BAC of 0.5 mg/mL (Ramaekers, et al., 2004; Ramaekers, et al., 2000). Since this BAC is the legal limit in most European countries it makes the impairment caused by cannabis clinically relevant. Cannabis is becoming more and more accepted as a medicine. Not only smoked cannabis, but also orally administered cannabis is being developed for this purpose. Dronabinol (Marinol®) for example is a cannabinoid that is used to treat anorexia in AIDS patients and other wasting diseases, anti-emesis in cancer patients undergoing chemotherapy and for chronic pain and is administered orally in capsules. The active ingredient dronabinol is synthetic ǻ9-tetrahydrocannabinol (THC) that also occurs as a natural component of Cannabis sativa L (Marijuana), which is recreationally smoked. The therapeutic range of dronabinol is between 2.5 and 20 mg/day. Only a few studies have investigated the pharmacodynamic effects of orally administered dronabinol in an experimental design. Almost all studied the subjective effects of dronabinol and found significant positive effects with doses ranging from 5 to 15 mg (Curran, Brignell, Fletcher, Middleton, & Henry, 2002; Gray, Hart, Christie, & Upadhyaya, 2008; Kirk & de Wit, 1999). In these studies subjects generally scored higher on items like “Good drug effect”, “Feel drug” and “Feeling stoned” when under the influence of dronabinol. Kirk and de Wit (1999) studied frequent and infrequent users and found that frequent users were more tolerant to these subjective effects than infrequent users. However, other studies did not find these subjective effects with doses of 7.5 mg in daily and weekly users (Goodwin, et al., 2006) and even 20 mg in occasional users (Menetrey, et al., 2005). In both of these studies only 6 respectively 8 subjects participated, which could have decreased statistical power. However, Menetrey et al. (2005) did find an effect on willingness to drive. Subjects indicated on a VAS scale that they were significantly hampered in their willingness to drive while under the influence of 20 mg dronabinol. Moreover, they also showed significant impairment on a tracking task in a driving simulator. Psychomotor performance as measured with the digit symbol substitution task (DSST) was not impaired in another study (Gray, et al., 2008). Neither was perceptual priming and working memory, but episodic memory and learning were impaired in a dose

149 DRUID 6th framework programme Deliverable 1.2.2 dependent way (Curran, et al., 2002). Cardiovascular measures are generally not affected by dronabinol (Goodwin, et al., 2006; Gray, et al., 2008). Thus, it seems that subjective effects of dronabinol are generally positive and there are no indications for cardiovascular effects. Psychomotor and memory effects are more inconsistent and need to be investigated more thoroughly. Oral administration of cannabis has a different pharmacokinetic profile than smoked cannabis. Dronabinol has an onset of action of approximately 0.5 to 1 hour. Its peak effect is at 2 to 4 hours and lasts for 4 to 6 hours. The bioavailability is only 4-12%. Smoked cannabis on the other hand has on onset of action within 5-10 minutes, when it reaches peak plasma levels (McGilveray, 2005; Ramaekers, et al., 2006). Subjective effects also peak within 5-10 minutes and significantly decrease after 1 hour (Ramaekers, et al., 2009). The bioavailability of smoked THC is on average 30% (McGilveray, 2005). Hence, oral administration of cannabis has a slower onset of action and is less effective, but the effects last longer compared to smoked cannabis. Due to the differences in pharmacokinetics between orally administered and smoked cannabis, the pharmacodynamic effects might differ as well. Since smoked cannabis is known to cause driving impairment, it is important to study the effects of orally administered dronabinol on driving performance. We therefore designed a study that investigated the effects of dronabinol on actual driving, simulated driving and the Field Sobriety Test (FST) in occasional and daily cannabis users. The daily users represent a model for chronic use of dronabinol. The current paper will only focus on the actual driving tests. Therefore, the objective was to assess the differential effect of dronabinol on actual driving in both user groups. We expected the daily users to be tolerant to the effects of dronabinol. Furthermore, it was hypothesized that the occasional users would show driving impairment. Results on simulated driving and the FST will be reported elsewhere.

150 DRUID 6th framework programme Deliverable 1.2.2 Method

Subjects Twelve recreational occasional cannabis users and twelve daily users participated in this study. Fourteen males and ten females equally divided over both user groups received placebo, 10 mg and 20 mg dronabinol on three separate occasions. Their mean (SE) age was 23.6 (0.6) years. The mean (SE) lifetime cannabis use was 274.1 (89.6) times for the occasional users and 2444.2 (708.8) times for the daily users. Subjects were recruited by advertisements at Maastricht University and were paid upon completion of the study. Before enrollment, all subjects were screened by means of a telephone interview to determine whether they qualified for the study. The inclusion criteria were experience with cannabis, i.e., 5-36 times in the last year for occasional users and >160 times in the last year for daily users; free from psychotropic medication; good physical health as determined by a medical examination; absence of any major medical, endocrine, and neurological condition; body mass index between 18 and 28; possession of valid driving license without revocation; and written informed consent. The exclusion criteria were history of drug abuse or addiction as assessed by means of a medical questionnaire by the physician at the medical checkup; pregnancy or lactation; cardiovascular abnormalities on electrocardiogram; excessive drinking, i.e., more than 20 alcoholic consumptions a week; hypertension, i.e., systolic blood pressure over 170 mmHg or diastolic blood pressure over 100 mmHg; history of or current psychiatric disorder; susceptibility to simulator sickness; and allergy to sesame oil. If subjects met the inclusion criteria, they received a medical history and a drug questionnaire to get a more precise view on their health and drug use. Finally, subjects underwent a medical examination and took part in a training session to get familiar with the tests. This study was conducted according to the code of ethics on human experimentation established by the declaration of Helsinki (1964) and amended in Seoul (2008). Approval for the study was obtained from the Medical Ethics committee of the Academic Hospital of Maastricht and Maastricht University. A permit for obtaining, storing, and administering cannabis was obtained from the Dutch drug enforcement administration.

Study design The study was conducted according to a double-blind, placebo-controlled, randomized, three-way, cross-over design. Treatments consisted of single doses of placebo, 10 and 20 mg dronabinol. Treatment orders were balanced over subjects and treatment periods. Placebo and dronabinol were administered orally in four identically appearing capsules. The wash-out period between treatments was at least 1 week for occasional users and at least four days for daily users.

Procedure Subjects were asked to refrain from any drugs 1 week before the medical examination until study completion. Daily users were allowed to continue their cannabis consumption as they were used to before enrollment in the study. Subjects were not allowed to drink alcohol and caffeine or smoke tobacco during a 24-h period prior to testing. Subjects were always tested for alcohol and drugs, i.e., tetrahydrocannabinol, opiates, amphetamine/ecstasy, benzodiazepines, cocaine, and methamphetamine/ecstasy, in breath and urine respectively upon arrival (between 11:00 and 11:30 a.m.) at the laboratory on test days. Baseline

151 DRUID 6th framework programme Deliverable 1.2.2 measures, i.e. bloodpressure, heartrate, temperature, subjective measures, blood and saliva samples, were taken between 11:30 a.m. and 12:00 p.m. At 12:00 p.m. subjects received placebo or one of the two dronabinol doses after which they had a break of half an hour and lunch between 12:30 and 1:00 p.m. After lunch performance testing started and the timeline of this is displayed in Fig. 1. A testing day ended at 6:15 p.m. at which time subjects were driven home.

Figure 1: Study time table

Actual driving tests The road tracking test (O'Hanlon, 1984) consists of driving in a specially instrumented car with a constant speed of 95 km/h and as straight as possible on the right lane of primary highway during a 1 hour test ride. A video-camera mounted on the rear end of the car registers its lateral position relative to the road delineation. The images are recorded onto a hard drive in the car with a frequency of 4 Hz and are transformed into a file containing the measures of the lateral position. An off line editing routine is applied for removal of all data segments that reveal signal loss, disturbance or occurrence of passing maneuvers. The edited dataset is then used to calculate means and variances for lateral position. The primary dependent measures of this test is the standard deviation of lateral position (SDLP; i.e. a measure of weaving). Speed and standard deviation of speed (SDSP) are recorded as secondary control measures. The highway driving test has been calibrated in a manner allowing expression of any sedative drug effect in terms of the BAC required to achieve the equivalent level of driving impairment (Louwerens, Gloerich, de Vries, Brookhuis, & O’Hanlon, 1987). The alcohol calibration curve demonstrates that drinkers’mean SDLP rises exponentially with BAC. Results from the alcohol calibration study can be used for describing drugs’effects on SDLP in terms of respective BAC equivalencies. The change in SDLP at a BAC of 0.5 mg/ml (i.e. 2.4 cm) has been used as a criterion level to quantify drug effects. Any drug induced changes in SDLP that exceed this criterion value are defined as clinically relevant impairing drug effect in the present study. The car-following test (Brookhuis, de Waard, & Mulder, 1994; Ramaekers, Muntjewerff, & O'Hanlon, 1995) consists of two cars driving in tandem on a secondary road. The leading vehicle is operated by a study staff member, the following vehicle is operated by the subject who is accompanied by a driving instructor. The test begins with the two vehicles traveling in tandem at speeds of 100 km/h on a primary highway. Subjects attempt to drive 50-60 m behind the preceding vehicle and to maintain that headway as it executes a series of deceleration maneuvers. During the test, the speed of the leading car is automatically controlled by a modified cruise-control system. At the beginning it is set to maintain a constant speed of 100 km/h, and by activating a microprocessor, the investigator can start sinusoidal speed changes reaching an amplitude of -10% and returning to the starting level within 50 sec. The maneuver is repeated 6-10 times. Speed signals collected during speed maneuvers enter a power spectral analysis for yielding phase-delay between the vehicle’s velocities at the maneuver cycle frequency (0.02) Hz. Phase delay converted to a measure of Time to Speed Adaptation (TSA, in s) is the primary measure. Gain and coherence are secondary control

152 DRUID 6th framework programme Deliverable 1.2.2 measures. Gain is the amplification factor between both speed signals collected from the leading and following vehicle and indicates the magnitude of overshoot in reaction. Coherence is a measure to control for correspondence between both speed signals. Test duration is 25 minutes.

Subjective measures Subjective measures were taken at baseline and throughout a testing day. They consisted of two questionnaires: the revised Addiction Research Centre Inventory Marijuana Scale (ARCI-M) (Chait, Fischman, & Schuster, 1985) and a Visual Analogue Scale (VAS). The revised ARCI-M is a twelve item true- false questionnaire related to marijuana intoxication. The dependent measure was a composite score that was calculated by adding the answers on the twelve items, with false scored as 0 and true as 1. The seven item VAS consisted of a 100 mm line anchored with “not at all”and “most ever”. The subjects were asked to indicate to what extent they experienced “good drug effect”, “high”, “stoned”, “stimulated”, “sedated”, “anxious”and “depressed”and their score was registered in mm. At the end of a testing day subjects were asked to indicate on the treatment questionnaire which drug they thought they had received that day, i.e. placebo, 10 or 20 mg dronabinol.

Pharmacokinetic assessment Blood samples ( 10 mL) were collected 3 times during a testing day, i.e 1.5, 4.25 and 6 hours post- drug. From daily users an extra sample was taken as a baseline measure, because they continued smoking their own cannabis in between testing days as they were used to. THC, 11-OH-THC and THC-COOH concentrations were determined in serum afterwards. The blood sample was centrifuged and the resulting serum was frozen at -20ºC until analysis.

Statistical analysis

All statistical analyses were conducted by means of SPSS 18.0 for Mac. Statistical analyses consisted of 2 steps: 1) Assessment for overall treatment effects by means of superiority testing; 2) Equivalence or non-inferiority testing of drug effects based on difference scores from placebo (within group) relative to the pre-established alcohol criterion, and 3) determination of concentration effect relations. Steps 2 and 3 were only conducted in case of treatment effects, and only for the primary measures of driving performance. During step one all data entered the general linear model (GLM) repeated measures ANOVA procedures with Drug (3 levels) and Time (4 levels in case of subjective measures) as main within subject factors. User group (2 levels) were entered as between subject factors. If the sphericity assumption was violated or not applicable, the Greenhouse-Geisser correction was used. In case of an overall effect of Drug, separate drug-placebo contrast analyses were conducted for each dronabinol dose. Since the objective of this study was to investigate the differential effects of dronabinol on driving performance in occasional and daily users and because we had an apriori hypothesis we repeated the same analysis on the two user groups separately whenever a main effect or interaction was found. Step two assessed whether a pre-established alcohol criterion falls within the 95% confidence interval (CI) of the drug effect. If yes, than the drug effect was considered to be equivalent or bigger than a

153 DRUID 6th framework programme Deliverable 1.2.2 BAC of 0.5 mg/mL, and thus relevant for traffic safety. If the 95% CI was lower than the alcohol criterion value, than a drug effect was considered of no clinical relevance. In step 3, concentration-effect relations were determined according to the following procedure. Data collected during different doses of a drug were converted to change scores from placebo for analyses of the association between drug concentration and performance. A linear regression analysis was conducted to establish linear relationships between changes (from placebo) in task performance during drug treatment and log-transformed drug concentrations in serum. The total number of data points included in these equations was defined by the number of subjects x maximal number test repetitions x the number of drug doses. Individual drug concentrations in serum prior to performance assessments in each of the drug dose conditions were divided over three mutually exclusive categories covering the full range of drug concentrations. The concentration ranges in serum were 0-50, 50-100 and > 100 ng/mL during evening sessions and 0-25, 25-50 and >50 ng/mL during morning sessions. The concentration ranges in oral fluid were 0-250, 250-1000 and >1000 ng/mL during evening sessions and 0-100, 100-500 and >500 ng/mL during morning sessions Corresponding change scores of task performance were then classified either as showing “impairment”or “no impairment”for all individual cases within each of these categories. Impairment was defined as a positive change score from placebo. Binomial tests were applied to measure whether the proportion of observations showing impairment or no impairment significantly differed from the hypothesized proportion. It was hypothesized that in case of no effect of a drug on task performance the proportion of observations showing impairment or no impairment would be equal, i.e. 50%.

154 DRUID 6th framework programme Deliverable 1.2.2 Results

Dropouts and missing data In total 3 occasional users dropped out and they were replaced by other subjects. In all cases subjects did not want to continue because of the strong drug effects, which they experienced as unpleasant. Three occasional users had missing data in the highway driving. This was due to heavy rain, which made it impossible to measure SDLP (placebo condition), computer failure (10 mg condition) and in one case the subject felt unable to drive because of the medication (20 mg condition). In case of missing data in the placebo condition the training measurement was taken instead. The other two subjects with missing data were not used in the analysis. During car following computer failure caused missing data in 6 cases of occasional users and in 4 cases of daily users. In 3 of these cases of daily users data was missing in 2 conditions. Subjects with missing data were excluded from the analysis. Two occasional users and one daily user had missing data on the VAS and were not used in the analysis.

Prematurely terminated driving tests In three cases the highway driving test was prematurely terminated by the driving instructor. In one case because of heavy snowfall. This condition was replaced on a later day by the same subject. The other two cases were terminated because subjects were falling asleep while driving. In one case the available data entered the analysis. In the other case the training measurement was used instead, because it happened in the placebo condition.

Table 1: Mean (SE) performance on road tracking test and car following test and percentage of subjects impaired on SFST in every treatment condition. Significance indicated by p-value. Drug ANOVA (overall) Test Cannabis Placebo 10 mg 20 mg Dronabinol x Can- Dronabinol use history dronabinol dronabinol nabis use history Road tracking

SDLP (cm) Occasional 17.9 (0.8) 20.4 (1.2) 22.1 (1.4) 0,008 NS Heavy 19.7 (1.3) 21.0 (1.2) 21.4 (1.2) NS Mean speed Occasional 96.1 (0.4) 95.9 (0.3) 96.1 (0.6) NS NS (km/h) Heavy 95.8 (0.2) 95.8 (0.4) 95.6 (0.3) NS SD Speed Occasional 2.1 (0.1) 2.2 (0.2) 2.1 (0.1) NS NS (km/h) Heavy 2.2 (0.2) 2.3 (0.2) 2.3 (0.2) NS Car Following

TSA (s) Occasional 2.5 (0.4) 4.1 (0.3) 2.9 (0.3) 0,011 NS Heavy 3.1 (0.1) 3.6 (0.4) 3.8 (0.8) NS Coherence Occasional 0.9 (0.01) 0.9 (0.02) 0.9 (0.01) NS NS Heavy 0.9 (0.01) 0.9 (0.02) 0.9 (0.01) NS Gain Occasional 1.2 (0.07) 1.2 (0.05) 1.2 (0.07) NS NS Heavy 1.2 (0.06) 1.2 (0.04) 1.4 (0.10) NS Ȥ2

SFST (% Occasional 8.3 8.3 8.3 NS impaired) Heavy 0.0 16.7 16.7 NS

155 DRUID 6th framework programme Deliverable 1.2.2 Table 1: Mean (SE) performance on road tracking test and car following test and percentage of subjects impaired on SFST in every treatment condition. Significance indicated by p-value. HGN (% Occasional 0.0 8.3 0.0 NS impaired) Heavy 0.0 0.0 0.0 NS WAT (% Occasional 8.3 33.3 33.3 NS impaired) Heavy 16.7 41.7 50.0 NS OLS (% Occasional 16.7 16.7 25.0 NS impaired) Heavy 8.3 25.0 16.7 NS

Actual driving performance The main dependent variable for driving performance is the SDLP of the highway driving test. The results showed that SDLP increased linearly from placebo to 20 mg dronabinol in a dose dependent fashion for both user groups (Table 1). Repeated measures analysis indicated a main effect on SDLP for drug (F2,40=7.812 , p=0.001). Simple contrasts indicated that this effect was due to a significant difference between placebo and 10 mg dronabinol (p=0.014) as well as a significant difference between placebo and 20 mg dronabinol (p=0.001). Even though the interaction between drug and user group was not significant, the analysis for the two user groups separately showed that the drug effect was significant for occasional users (F2,18=6.493 , p=0.008), but not for the daily users (p=0.218). Simple contrasts for the occasional user group indicated that the effect was significant for 10 mg (0.039) as well as 20 mg versus placebo (p=0.016). The two control measures of the highway driving test (mean speed and SDSP) did not show a significant effect on drug (p=0.947 and p=0.460 respectively).

Figure 2. Mean change in SDLP (95%CI) after doses of 10 and 20mg dronabinol in occasional and daily cannabis users.

The equivalence testing for SDLP (Figure 2) shows that occasional users are significantly impaired compared to placebo, because zero is not included in the 95% confidence intervals (CIs). It also shows that

156 DRUID 6th framework programme Deliverable 1.2.2 their impairment is relevant for traffic safety, since the ǻSDLP values of at least 0.5 mg/mL BAC are included in the CIs of both doses. In case of 20 mg even 0.8 mg/mL BAC falls within the CI of the occasional users. Equivalence tests also demonstrated that in dialy users, the 95%CIs for 10 and 20 mg did exceed the alcohol criterion limit. In case of 10 mg dronabinol, the CI actually included 0 as well as the 0.5 mg/mL criterion. (Figure 3).

Figure 3. Individual SDLP change scores from placebo after dronabinol 10 and 20 mg in occasional and daily users

The primary dependent measure of the car following test is TSA. Results showed a trend effect for drug on TSA (F2,24=3.083 , p=0.064). Simple contrasts indicated that this effect was significant for placebo versus 10 mg dronabinol (p=0.005). The interaction between drug and user group was non-significant. However, the same analysis for the user groups separately showed a significant effect of drug on TSA for occasional users (F2,10=7.269 , p=0.011) and not for daily users (p=0.609). Simple contrasts for the occasional user group indicated that this was due to the difference between placebo and 10 mg (p=0.029). The two control measures of the car following tests (coherence and gain) did not show any significant effect.

Table 2: Mean (SE) scores on the subjective questionnaires. Significance indicated by p-value. Measure User Occasional Daily Occasional Daily Occasional Daily

157 DRUID 6th framework programme Deliverable 1.2.2 Table 2: Mean (SE) scores on the subjective questionnaires. Significance indicated by p-value. Measure Scale ARCI-M Total VAS Good drug effect VAS High 1 0.00 (0.00) 0.42 (0.15) 0.4 (0.8) 1.2 (0.8) 0.5 (0.8) 1.2 (0.8) 2 0.25 (0.25) 0.08 (0.08) 5.9 (6.0) 7.1 (6.0) 3.6 (4.5) 5.8 (4.5) Placebo 3 0.08 (0.08) 0.00 (0.00) 0.5 (0.3) 1.0 (0.3) 0.8 (0.7) 1.9 (0.7) 4 0.08 (0.08) 0.00 (0.00) 1.1 (0.2) 0.6 (0.2) 0.7 (0.1) 0.4 (0.1) 1 0.00 (0.00) 0.25 (0.18) 0.5 (1.1) 1.5 (1.1) 0.5 (1.0) 1.4 (1.0) 10 mg 2 3.75 (1.05) 1.75 (0.54) 16.5 (6.4) 17.0 (6.4) 12.6 (4.2) 10.8 (4.2) Drug dronabinol 3 3.92 (1.10) 0.25 (0.13) 15.4 (1.3) 3.4 (1.3) 12.0 (1.2) 3.4 (1.2) 4 2.42 (0.67) 0.00 (0.00) 6.6 (0.4) 1.1 (0.4) 3.7 (0.4) 1.0 (0.4) 1 0.00 (0.00) 0.25 (0.18) 0.4 (0.8) 1.5 (0.8) 0.6 (0.3) 0.8 (0.3) 20 mg 2 4.25 (0.94) 2.17 (0.59) 27.4 (9.1) 28.2 (9.1) 23.4 (8.8) 28.9 (8.8) dronabinol 3 4.42 (0.91) 1.33 (0.53) 29.6 (3.8) 12.6 (3.8) 19.2 (3.0) 11.0 (3.0) 4 4.17 (0.93) 0.33 (0.26) 25.3 (1.2) 2.1 (1.2) 12.5 (0.9) 2.2 (0.9)

Drug 0,000 0,009 0,000 0,021 0,004 0,006 WS ANOV Time 0,000 0,010 0,002 0,012 0,001 0,010 A Drug x Time 0,031 0,006 NS 0,099 NS 0,020

BS Drug x User 0,001 0,098 NS ANOV Time x User 0,000 NS NS A User 0,000 NS NS

Subjective measures Results on the VAS showed a main effect for drug on all scales except depression, i.e. good drug effect, high, stoned, stimulated, sedated, anxious and a main effect of time on all scales except anxious and depression (see Table 2). Simple contrasts indicated that for all significant drug effects placebo differed significantly from 10 mg as well as 20 mg dronabinol and for all significant time effects all measures differed significantly from baseline. Interaction between time and user group were significant for sedated and anxious. Contrasts indicated that this was due to a difference from baseline for measure 3 and 4 for sedated and a difference from baseline for measure 2 and 3 for anxious. The interaction between drug and time was also significant for all measures except anxious and depression, meaning that the drug effect differed for the various time points. Contrasts generally showed that for the 10 mg dronabinol versus placebo less time points differed significantly from baseline than for the 20 mg dronabinol versus placebo. The between subjects factor user group was significant on the anxious scale. When these analyses were repeated for the two user groups separately, results were generally confirmed (Figure 4). The main effect of drug was significant in both user groups for good drug effect, high, stoned and stimulated. The difference between daily and occasional users was that for occasional users this effect was true for the 10 mg as well as 20 mg dronabinol versus placebo and for the daily users only the difference between placebo and 20 mg dronabinol reached significance. Daily users also had a significant drug effect on the sedated scale for both drug conditions. The same pattern is visible for the main effect of time. Good drug effect, high, stoned and sedated differed significantly from baseline at all time points for

158 DRUID 6th framework programme Deliverable 1.2.2 occasional users and at all time points except the last one for daily users. On the scale stimulated occasional and daily users were comparable with a significant difference from baseline for all time points except the last one. The drug by time interaction was non-significant for the occasional users, but the daily users had a significant interaction on the high, stoned and sedated scales. Contrasts showed that only 20 mg dronabinol versus placebo differed significantly from baseline at time points 2 and 3. Additionally, on the sedated scale 10 mg versus placebo differed significantly from baseline at time point 2.

Figure 4: subjective ratings after 10 and 20mg dronabinol in occasional users (upper pane;) and daily users (lowe panel) as a function of time after dosing

The repeated measure analysis of the ARCI-M showed a main effect of drug (F2,44=25.118, p=0.000), a main effect of time (F3,66=16.921, p=0.000), an interaction of drug by user (F2,44=8.728, p=0.001), an interaction of time by user (F3,66=8.799, p=0.000) as well as a drug by time interaction (F2.752,60.555=5.929, p=0.002). Simple contrasts indicated that both drug conditions were significantly different from placebo for the main drug effect and the interaction with user group and that all measures differed significantly from baseline in the main effect of time as well as the interaction with user group. The drug by time interaction was also significant for all drug levels and at all time points compared to placebo and baseline respectively. The between subjects factor user group was also significant (F1,22=17.770, p=0.000). The analysis for the two user groups separately generally confirmed the above findings (see Table 2 and Figure 5 for details). The results for the occasional users did not change, but in the daily user group the main effect of time was only significant for the change from baseline at time 2. The drug by time interaction in the daily user group was significantly different from baseline at time 2 for 10 mg dronabinol versus placebo and at time 2 and 3 for 20 mg dronabinol versus placebo. The main drug effect of the daily users stayed the same as above.

159 DRUID 6th framework programme Deliverable 1.2.2 Figure 5. Mean ARCI-M after all treatments in occasional and daily users.

Pharmacokinetic assessment Pharmacokinetic data was missing on 18 occasions. The mean (SE) values for THC, 11-OH-THC and THC- COOH are displayed in Table 3. The baseline measures for the daily users did not differ significantly between the three drug conditions (F1.086,8.686=1.971, p=0.196).

Table 3: THC, 11-OH-THC and THC-COOH concentrations in blood. User Occasional Daily Measure THC (ȝg/L) 11-OH-THC THC- THC (ȝg/L) 11-OH-THC THC-COOH ȝg/L) COOH(ȝg/L) ȝg/L) ȝg/L) Placebo 1 - - - 15.3 (8.5) 5.1 (2.3) 58.2 (17.5) 2 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 6.4 (2.7) 2.5 (0.8) 37.4 (9.9) 3 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 4.9 (1.7) 1.8 (0.5) 37.2 (11.6) 4 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 4.4 (1.6) 1.5 (0.4) 31.2 (9.4) 10 mg dronabinol 1 - - - 7.6 (3.6) 2.9 (1.2) 41.8 (11.2) 2 2.4 (0.7) 3.5 (0.7) 20.8 (5.5) 9.2 (2.1) 6.2 (0.9) 54.4 (8.8) 3 1.2 (0.5) 1.6 (0.3) 18.5 (2.8) 4.6 (1.1) 3.1 (0.4) 43.4 (7.4) 4 0.9 (0.2) 1.7 (0.3) 16.8 (1.3) 4.2 (1.0) 2.3 (0.4) 37.2 (7.6) 20 mg dronabinol 1 - - - 11.2 (5.3) 3.9 (1.7) 48.1 (13.9) 2 4.5 (1.3) 5.0 (1.4) 30.5 (10.3) 10.8 (2.0) 7.7 (1.0) 57.5 (9.7) 3 1.4 (0.3) 2.6 (0.5) 24.0 (3.7) 6.8 (1.2) 4.7 (0.5) 54.8 (9.3) 4 2.6 (0.6) 3.1 (0.5) 30.2 (4.3) 5.7 (1.2) 3.6 (0.4) 57.5 (9.9)

160 DRUID 6th framework programme Deliverable 1.2.2 Discussion

This study was designed to assess the effects of two doses dronabinol, i.e. 10 and 20 mg, on driving performance in occasional and daily cannabis users. These doses are within the normal range prescribed for patients suffering from anorexia in AIDS and other wasting deseases, emesis due to chemotherapy in cancer patients and for chronic pain. Results of this study showed that occasional cannabis users were impaired on the highway driving test as well as on the car following test when driving under the influence of dronabinol. Their SDLP increased significantly with on average 2.5 cm in the 10 mg dronabinol condition and 4.2 cm in the 20 mg dronabinol condition compared to placebo. This means that occasional users had increased weaving and thus diminished automatic control over the car to an extent that is relevant for traffic safety. In the car following test occasional users had increased TSA. This indicates that they reacted slower to the changes in speed of the car in front of them when under the influence of 10 mg dronabinol compared to placebo. Interestingly, performance under the influence of 20 mg dronabinol did not reach significance and was better than that with 10 mg dronabinol. This is probably due to the large amount of missing data. Only 6 subjects had complete data and entered the analysis. If the other 6 would have been included, results could have been different. The overall effect, when occasional and daily users are used as a between subjects factor, also just missed significance, probably due to the same reason. Therefore, this result should be interpreted with caution. However, the fact that occasional users in the 10 mg condition did reach significance with this amount of subjects, points toward a strong effect. In daily users, mean SDLP did not differ between treatments according to superiority tests. However, equivalence test demonstrated that the 95% CI associated with change SDLP after both doses of dronabinol included the criterion value equivalent to BAC of 0.5 mg/mL. A relatively wide range of these 95% CIs was caused by large inter-individual variations in change SDLP observed after both doses. Inspection of individual data indicated that approximately 25% of the heavy users demonstrated impairment in road tracking performance that was equivalent or worse than that observed after BAC of 0.5 mg/mL. Together, these data show that THC induced impairments of road tracking performance in heavy cannabis are less as compared to those observed in occasional users. This reduction in sensitivity to the impairing effects of THC in heavy users has been reported before and has been interpreted as a demonstration of behavioral tolerance that develops after repeated cannabis use Ramaekers et al. (2009, 2010). This study however also demonstrates that behavioral tolerance was not complete in every heavy cannabis user as indicated by large inter-individual differences in driving impairments observed in these drivers. Tolerance to the effects of dronabinol in chronic cannabis users has been shown in a recent study as well (Bedi, et al., 2010). HIV positive patients who were also daily cannabis users were evaluated on caloric intake as well as subjective and cognitive measures. Within 16 days tolerance developed to the therapeutic effects, i.e. caloric intake and sleep quality, in these patients who received 4 times per day 10 mg dronabinol. However, the mood-enhancing effects did not disappear over the 16 days test period. Thus, tolerance developed for some specific effects and not for others. We used the daily cannabis user group as a model for chronic dronabinol use. Chronic use of dronabinol might be necessary in patient populations for which the drug has been proven useful, i.e. AIDS patients, cancer patients undergoing chemotherapy and chronic pain patients. Considering the fast development of tolerance in the study of Bedi et al. (2010) and the results of the current study, this model

161 DRUID 6th framework programme Deliverable 1.2.2 seems valid. The consequence of these results is that chronic users might need higher doses to reach therapeutic effects. Although we did not show a significant effect of dronabinol on driving performance in daily cannabis users, some subjects in this group should be classified as impaired. Compared to Bedi et al. (2010) our dose was low: 4 times per day 10 mg versus once 20 mg as our highest dose. Therefore, if chronic users need even higher doses than used in that study, traffic safety might be comprised for these individuals. It seems that even with the current dose in our study some subjects are not tolerant to the effects on driving performance. Higher doses would probably increase the number of subjects showing impairment. Subjective effects were also assessed in this study to determine tolerance to these effects as well in daily users. We used two measures for subjective effects: ARCI-M and VAS. The ARCI-M showed significant effects on all factors. In general the subjects experienced marijuana related effects when under the influence of both doses dronabinol. Effects lasted up to 6 hours after drug intake. There were differences between occasional and daily users. In general occasional users scored higher on the ARCI-M indicating that they experienced more marijuana related effects, i.e. more items were answered positively, than the daily users. The effects generally lasted longer for occasional users as well. Daily users experienced longer lasting effects in the 20 mg than in the 10 mg dronabinol condition. The VAS showed a similar effect: in general the positive drug effects, i.e. good drug effect, high, stoned and stimulated, were increased under the influence of dronabinol. Occasional users reported these effects for both doses, whereas daily users only for the 20 mg dronabinol condition. Again these effects were experienced throughout a testing day especially in occasional users. The daily users generally reported the effects up to 4.25 hours after drug intake. User group differed significantly for the negative effects, i.e. sedation and anxiety. Daily users reported more sedation under the influence of dronabinol than occasional users, which lasted longer in the 20 than in the 10 mg condition. Occasional users experienced more anxiety than daily users, especially toward the end of a testing day, i.e. 4.25 and 6 hours after drug intake. From these results it is clear that daily cannabis users were not tolerant to the subjective effects of dronabinol. They experienced less effects though, that were less intense and shorter lasting than occasional cannabis users. Other studies found the same in various population. The study with HIV positive patients mentioned earlier found that these patients, who were daily cannabis users, experienced positive mood effects and elevated ratings of drug high, good drug effect, drug liking and wanting to take the drug again up to 16 days with 4 times daily 10 mg dronabinol compared to placebo (Bedi, et al., 2010). As in our study these patients also experienced sedation, especially in the second testing week (day 9-16). The authors concluded that the subjects were not tolerant to these subjective effects. Another study compared frequent and infrequent users on subjective measures and found that frequent users reported higher ratings on feel drug, high and ARCI-M in a low dose condition, i.e. 7.5 mg dronabinol, than infrequent users (Kirk & de Wit, 1999). In the high dose condition, i.e. 15 mg dronabinol, both user groups reported elevated scores on these measures. The infrequent users also reported more negative effects, i.e. sedation, compared to frequent users and lower ratings on stimulant-like effects and drug liking compared to placebo in the high dose condition, which was not found in the frequent users. Based on this the authors conclude that the frequent users might be tolerant to some of the subjective effects, especially the negative effects that were only found in infrequent users. However, their results clearly point out that frequent users are not tolerant to the positive effects as was found in the current study. The discrepancies between this study and ours, especially on sedation which was more elevated in our daily users than our occasional users, is probably due to

162 DRUID 6th framework programme Deliverable 1.2.2 differences in dose. We used 10 and 20 mg dronabinol, whereas they used 7.5 and 15 mg. Moreover, the frequent users of this study used less cannabis than our daily users: 3.5 versus 13.4 joints per week. The conclusion from our study is that it is unsafe to drive under the influence of a therapeutic dose of dronabinol. Chronic users might become less sensitive to the impairing effects to some extent. However, not all daily users had developed tolerance and some drove comparable or worse than subjects under the influence of 0.5 mg/mL BAC, which is the legal limit in most European countries.

163 DRUID 6th framework programme Deliverable 1.2.2 Acknowledgement This work was conducted as part of the Driving under the influence of drugs, alcohol and medicines (DRUID) research consortium funded by European Union grant TREN-05-FP6TR-S07.61320-518404- DRUID. This report reflects only the author's view. The European Community is not liable for any use that may be made of the information contained therein.

References Bedi, G., Foltin, R. W., Gunderson, E. W., Rabkin, J., Hart, C. L., Comer, S. D., et al. (2010). Efficacy and tolerability of high-dose dronabinol maintenance in HIV-positive marijuana smokers: a controlled laboratory study. Psychopharmacology (Berl), 212(4), 675-686. Brookhuis, K. A., de Waard, D., & Mulder, B. (1994). Measuring driving performance by car-following in traffic. Ergonomics, 37(3), 427-434. Chait, L. D., Fischman, M. W., & Schuster, C. R. (1985). 'Hangover' effects the morning after marijuana smoking. Drug Alcohol Depend, 15(3), 229-238. Curran, H. V., Brignell, C., Fletcher, S., Middleton, P., & Henry, J. (2002). Cognitive and subjective dose-response effects of acute oral Delta 9-tetrahydrocannabinol (THC) in infrequent cannabis users. Psychopharmacology (Berl), 164(1), 61-70. European Monitoring Centre for Drugs and Drug Addiction (2008). EMCDDA Insights Series No 8: Drug use, impaired driving and traffic accidents.: Luxembourg: Office for official publications of the European Communities. European Monitoring Centre for Drugs and Drug Addiction (2009). Annual report 2009: the state of the drugs problem in Europe: Luxembourg: Publications Office of the European Union. Fernandez-Serrano, M. J., Perez-Garcia, M., & Verdejo-Garcia, A. (2010). What are the specific vs. generalized effects of drugs of abuse on neuropsychological performance? Neurosci Biobehav Rev. Goodwin, R. S., Gustafson, R. A., Barnes, A., Nebro, W., Moolchan, E. T., & Huestis, M. A. (2006). Delta(9)-tetrahydrocannabinol, 11-hydroxy-delta(9)-tetrahydrocannabinol and 11-nor-9-carboxy- delta(9)-tetrahydrocannabinol in human plasma after controlled oral administration of cannabinoids. Ther Drug Monit, 28(4), 545-551. Gray, K. M., Hart, C. L., Christie, D. K., & Upadhyaya, H. P. (2008). Tolerability and effects of oral Delta9-tetrahydrocannabinol in older adolescents with marijuana use disorders. Pharmacol Biochem Behav, 91(1), 67-70. Grotenhermen, F., Leson, G., Berghaus, G., Drummer, O. H., Kruger, H. P., Longo, M., et al. (2007). Developing limits for driving under cannabis. Addiction, 102(12), 1910-1917. Kirk, J. M., & de Wit, H. (1999). Responses to oral delta9-tetrahydrocannabinol in frequent and infrequent marijuana users. Pharmacol Biochem Behav, 63(1), 137-142. Lamers, C. T., & Ramaekers, J. G. (2001). Visual search and urban driving under the influence of marijuana and alcohol. Hum Psychopharmacol, 16(5), 393-401. Louwerens, J., Gloerich, A., de Vries, G., Brookhuis, K., & O’Hanlon, J. (1987). The relationship between drivers’blood alcohol concentration (BAC) and actual driving performance during high speed travel. Alcohol Drugs Traffic Saf, 86, 183-186. McGilveray, I. J. (2005). Pharmacokinetics of cannabinoids. Pain Res Manag, 10(A), 15A-22A.

164 Menetrey, A., Augsburger, M., Favrat, B., Pin, M. A., Rothuizen, L. E., Appenzeller, M., et al. (2005). Assessment of driving capability through the use of clinical and psychomotor tests in relation to blood cannabinoids levels following oral administration of 20 mg dronabinol or of a cannabis decoction made with 20 or 60 mg Delta9-THC. J Anal Toxicol, 29(5), 327-338. O'Hanlon, J. F. (1984). Driving performance under the influence of drugs: rationale for, and application of, a new test. Br J Clin Pharmacol, 18 Suppl 1, 121S-129S. Ramaekers, J. G., Berghaus, G., van Laar, M., & Drummer, O. H. (2004). Dose related risk of motor vehicle crashes after cannabis use. Drug Alcohol Depend, 73(2), 109-119. Ramaekers, J. G., Kauert, G., Theunissen, E., Toennes, S., & Moeller, M. (2009). Neurocognitive performance during acute THC intoxication in heavy and occasional cannabis users. J Psychopharmacol. Ramaekers, J. G., Moeller, M. R., van Ruitenbeek, P., Theunissen, E. L., Schneider, E., & Kauert, G. (2006). Cognition and motor control as a function of Delta9-THC concentration in serum and oral fluid: limits of impairment. Drug and Alcohol Dependence, 85(2), 114-122. Ramaekers, J. G., Muntjewerff, N. D., & O'Hanlon, J. F. (1995). A comparative study of acute and subchronic effects of dothiepin, fluoxetine and placebo on psychomotor and actual driving performance. Br J Clin Pharmacol, 39(4), 397-404. Ramaekers, J. G., Robbe, H. W., & O'Hanlon, J. F. (2000). Marijuana, alcohol and actual driving performance. Hum Psychopharmacol, 15(7), 551-558. Ramaekers, J. G., Theunissen, E. L., de Brouwer, M., Toennes, S. W., Moeller, M. R., & Kauert, G. (2010). Tolerance and cross-tolerance to neurocognitive effects of THC and alcohol in heavy cannabis users. Psychopharmacology (Berl). Substance Abuse and Mental Health Services Administration (2009). Results from the 2008 National Survey on Drug Use and Health: National Findings: Office of Applied Studies, NSDUH Series H-36, HHS Publication No. SMA 09-4434). Rockville, MD.

165 Chapter 8 : Effects of analgetic medication on actual driving

Markus Schumacher*, Anja Knoche*, Mark Vollrath‡, Frank Petzke°, Ricarda Jantos†, Eric Vuurman˜, Jan Ramaekers˜

* Federal Highway Research Institute (BASt) Section U3 / Traffic Psychology, Traffic Medicine Bruederstraße 53 D-51427 Bergisch Gladbach, Germany

‡Technische Universität Braunschweig Institut für Psychologie Ingenieur- und Verkehrspsychologie Gaussstr. 23 D-38106 Braunschweig, Germany

° Georg-August-Universität Schmerz-Tagesklinik und -Ambulanz Zentrum Anaesthesiologie, Rettungs- und Intensivmedizin Robert-Koch-Str. 40 37075 Göttingen, Germany

†Universität Heidelberg Institut für Rechts- und Verkehrsmedizin Voß-Straße 2 D- 69115 Heidelberg, Germany

˜Maastricht University Dept of Neuropsychology & Psychopharmacology, Faculty of Psychology and Neuroscience Universiteitssingel 40 6229 ER Maastricht, The Netherlands

166 Abstract

Nowadays morphine-like drugs, called opioids (e.g. fentanyl, oxycodone, buprenorphine, and hydromorphone) are widely used to treat chronic non-cancer pain. Some studies have already demonstrated that long term treatment with opioids for non-cancer pain does not impair driving related skills. But there is still a lack of studies investigating actual driving performance. Twenty patients suffering from chronic non-cancer pain, on long term treatment with opioid analgesics (fentanyl, buprenorphine, oxycodone, hydromorphone or morphine), aged between 35 and 68 years, participated in this study. Their performance in an on-the-road driving test was compared to the performance of an age independent sample of healthy controls aged between 23 years to 58 years. Performance of healthy controls with blood alcohol concentration (BAC) of 0.5‰ was used as reference. The on-the-road driving tests consisted of a road-tracking test and a car-following test. The primary outcome measure was standard deviation of lateral position (SDLP). Time to speed adaption (TSA) and brake reaction time (BRT) was measured in the Car-following test. Whereas alcohol increased SDLP in healthy controls by about 2.4 cm, no difference in SDLP between patients and controls was found. Neither TSA nor BRT differed between patients and controls. Also no effect of alcohol on these measures was found. The data indicates that patients with chronic non-cancer pain are not necessarily unfit to drive but due to the individual variability of test results an individual assessment is recommended.

Keywords Chronic non-cancer pain, opioids, driving ability, on-the-road driving

Introduction

According to a recent review of publications the prevalence of chronic pain in the general population ranges from 10% to over 40% (Nickel & Raspe, 2001). Pain is classified being chronic when it persists for three month or longer (Mersky & Bogduk, 1994). The pharmacological management of chronic pain has changed over the last decade. Based on an optimistic assessment of the beneficial effects of morphine-like drugs, called opioids (e.g. fentanyl, oxycodone, buprenorphine, and hydromorphone) on pain intensity and functional capabilities, various opioids were widely used to treat chronic non-cancer pain conditions (Portenoy, 2000). More recently it became evident that the uncritical use of opioids may also cause considerable problems (e.g. dose escalation due to lack of relevant effect, development of tolerance with loss of efficacy). Therefore Guidelines like LONTS (Long-Term Treatment with Opioids for Non-Cancer Pain by the German Association for the Study of Pain (Schmidt, Lowenstein, Hesselbarth, Klinge, & Michels, 2010) have formulated a more cautious approach to opioid therapy in patients with chronic non-cancer pain. The availability and clinical evidence for efficacy of better tolerated neuromodulatory drugs like gabapentin and pregabalin, as well as both traditional and newer antidepressant agents like and duloxetine have broadened the pharmacological treatment spectrum. These agents are particularly effective in conditions where neuropathic or mixed pain aetiology was found as well as in musculoskeletal pain syndromes where a central hypersensitivity may be relevant. These developments have led to combination therapy in many patients with chronic pain. Although there is

167 little scientific evidence to support this approach, it is widely established and accepted in clinical practice. Hence today monotherapy with opioids is the exception rather than the rule. It is well documented that cognitive functions like attention are impaired in chronic pain patients (e.g. Dick & Rashiq, 2007; Eccleston, Crombez, Aldrich, & Stannard, 1997; Grisart & Plaghki, 1999) whereas relief of pain might diminish these impairments (Dick et al., 2007; Eccleston et al., 1997; Grisart, 2009). Attention deficits may lead to reduced fitness to drive. Lagarde, Chastang, Lafont, Coeuret-Pellicer, & Chiron (2005) showed that pain alone or in combination with pain treatment was associated with an increased risk of being involved in traffic accidents. However it was not possible to distinguish in this study between the negative effects of pain itself and those of pain treatment. Opioids are often prescribed in the treatment of chronic pain (McQuay, 1999) but are associated with side effects in the gastrointestinal system as well as in the central nervous system. Drowsiness and sleepiness are symptoms frequently reported (Kalso, Edwards, Moore, & McQuay, 2004; Moore & McQuay, 2005). Several studies showed that driving under the influence of sedating medications or drugs, especially of alcohol, increases accident risk (Drummer et al., 2004; Movig et al., 2004). The prevalence of drivers tested positive for opioids in a recent study ranged from 1% to 5%. Overall, the correlation between opioid use and culpability in a recent study was very low (Drummer et al., 2004). Studies published earlier support the opinion that the prevalence of opioids use in traffic is very low and is not associated with motor vehicle accidents or motor vehicle fatalities (Fishbain, Cutler, Rosomoff, & Rosomoff, 2002; Fishbain, Cutler, Rosomoff, & Rosomoff, 2003; Lenné, Dietze, Rumbold, Redman, & Triggs, 2000). Beside these epidemiological studies, a limited number of experimental studies have been published aiming at measuring driving ability of chronic pain patients (see Kress & Kraft, 2005 for a review and Veldhuijzen, Karscha, & van Wijck, 2009 for an overview). No performance decrement, measured in a set of tests, was found in cancer pain patients treated with stable doses of oral morphine (Vainio, Ollila, Matikainen, Rosenberg, & Kalso, 1995). Sabatowski et al. (2003) used the computerized test set that is also used for testing traffic offenders in Germany. They assessed driving ability of chronic non-cancer pain patients treated with transdermal fentanyl and found that their performance was non-inferior to that of a control group. The results of two other studies with the same experimental design and the same test set but with different opioid analgesics were similar: No overall impairment was found for the long-term use of transdermal buprenorphine (Dagtekin et al., 2007) and also not for the long-term use of controlled release oxycodone (Gaertner et al., 2006). However, particular patients had poor results. The assumption, that no general impairment occurs due to long-term treatment with opioids was also supported by Strumpf, Willweber-Strumpf, Herberg, & Zenz (2005). Gaertner et al. (2008) examined the effects of opioid drug dose changes on cognitive and psychomotor performances relevant for driving. Seven days after dose change, no deterioration was observed. Tassain et al. (2003) analyzed the performance of pain patients before the initiation of opioid therapy with slow release morphine. This initial performance was compared to the performance after 3, 6 and 12 month under stable medication. Results indicate that stable opioid medication did not lead to a deterioration of cognitive functioning during long-term application. Patients with chronic pain often report cognitive complaints (McCracken & Iverson, 2001) and it is well known that an association exists between fatigue and pain (Fishbain et al., 2003). The performance impairments might be caused by side effects of opioids but these side effects decline over time

168 (Dellemijn, van Duijn, & Vanneste, 1998; Tassain et al., 2003). In addition, those who use opioids occasionally or habitually (treatment of chronic pain or opioid abusers) are much less likely to be impaired (Zacny, 1995). Test indicators of cognitive ability and psychomotor function increase over time (Jamison et al., 2003). Moreover, it was assumed that the impairing effect of pain is more profound than the impairment caused by side effects of the opioids (Sjogren, Olsen, Thomsen, & Dalberg, 2000). The assumption that pain itself is impairing is supported by the study of Veldhuijzen et al. (2006). In this study an on-the-road driving test was used which was developed in the nineteen eighties and was applied in numerous studies on the impairments caused by psychoactive drugs and alcohol (Verster & Ramaekers, 2008). The standard deviation of lateral position (SDLP), measuring the range of weaving, was used as dependant measure in this test. The difference in SDLP of pain patients and healthy controls who were matched for age, education and driving experience, corresponded to the difference in SDLP observed in healthy volunteers driving sober vs. those driving with blood alcohol concentration (BAC) of 0.8‰ . Psychotropic drug use was not allowed in this study and the difference between patients and controls remained significant even when patients who had used paracetamol and/or NSAIDs during the study were excluded from the data analysis. In contrast to opioid analgesics both substances are not psychoactive. So the study demonstrated that chronic pain itself impaired driving performance. The results underline the need for effective pain treatment in order to enhance traffic safety. No assumption could have been drawn from this study on the impairments caused by psychoactive substances prescribed for pain treatment. Sabatowski, Kaiser, & Gossrau (2010) stated that to this date no decision could be drawn explicitly from the available studies whether slow release opioid formulations lead to impaired fitness to drive in pain patients or not. This assumption was mainly based on studies in which driving related skills were assessed. Up to now no study exists in which driving performance of pain patients under long term combination therapy was assessed in an on-the-road driving test. In order to better reflect the target group of pain patients treated with combination therapy, chronic pain patients treated with a variable mixture of other compounds in addition to a level 3 opioid were recruited for this study. The sample reflects the current standard of pharmacological treatment of chronic pain. It is more helpful to study this combination therapy, with the opioid being the most potent drug affecting the ability to drive, to gain a realistic and clinically relevant picture of driving ability of patients suffering from chronic pain, than focusing on the specific effects of a given opioid.

Methods

This study compared the driving ability of patients receiving opioid analgesic of WHO level 3 for chronic non-cancer pain treatment with a group of healthy controls. Performance of the controls driving with BAC of 0.5‰ was used as a reference for the impairment. Driving ability was measured by two standardized driving scenarios widely used in the field of drug research: Road tracking test (O'Hanlon, 1984) and Car-following test (Brookhuis, de Waard, & Mulder, 1994).

Subjects All subjects gave written informed consent, possessed a valid driver’s license for passenger cars, travelled at least 2000 km in the preceding 12 month and were driving on a regular basis (at least once a week). This study aimed at comparing the performance of pain patients to a representative

169 cross-section of the driving population. Therefore controls were stratified for age (20-30, 30-40, 40-50, 50-65). Since prevalence of chronic pain increases with age and men are more often affected by chronic pain than women (Nickel et al., 2001) the average age differs between both groups as can be seen in table 2. All subjects were given financial compensation for participating in this study.

Patients No study related change in prescribed medication was made. 13 male and 7 female patients (see table 2) of the pain outpatient department of the University Hospital of Cologne aged between 35 and 68 (m= 54, sd = 8.91) suffering from non-cancer pain responsive to opioids were enrolled in the study. They have been treated for at least four weeks with either transdermal fentanyl (• 12 ȝg/h), transdermal buprenorphine (10ȝg/h), slow release oxycodone (• 10 mg/day), slow release oxycodone combined with naloxone (• 10 mg/day), slow release hydromorphone (• 4 mg /day) or slow release morphine (• 20 mg/day). The dosage of the analgesic remained unchanged for at least 14 days prior to the assessment. All patients received co-medication on a stable dose at least during the past 14 days. Regular intake of benzodiazepines (• 4 times per week) and (> 3 times per week) was not allowed. In addition, the daily intake of antidepressants and anticonvulsant had to be below a given dose limit (see table 1). Patients had to abstain from intake of immediate release opioids, benzodiazepines, barbiturates and alcohol for at least two days prior to the assessment.

Table 1: Co-medication dose limits. antidepressants anticonvulsants amitryptilin ” 75mg carbamazepin ” 1200mg, doxepin ” 75mg oxcarbazepin ” 1800mg, imipramin ” 75mg gabapentin ” 2400mg, trazodon ” 100mg pregabalin ” 600mg sertralin ” 50mg fluoxetin ” 20mg fluvoxamin ”75mg duloxetin ” 120mg venlafaxin ” 225mg citalopram ”10mg

Controls 12 male and 7 female healthy controls participated in this study (see table 2). Their mean (sd) age was 43 (11.07) ranging from 23 years to 58 years. Participants were selected from a pool of volunteers that had given consent to participation in studies conducted by BASt in advance. Exclusion criteria were history of alcohol addiction or abuse, intake of medicaments prohibiting the consumption of alcohol or psychoactive substances with known impairing effect on driving ability. Women who were

170 pregnant or breast feeding or women of child-bearing potential who were not using a highly effective contraception method with a pearl-index ” 1 were also excluded due to the alcohol administration in the course of the study. All male but only five of seven female drivers completed the driving test sober as well as under the influence of alcohol. Two female controls terminated their participation due to personal reasons not related to the study. The total number of controls of which data of both conditions (sober and under alcohol influence) is available is 17 (89 %).

Table 2: Sample characteristics. characteristic patients controls (N = 20) (N = 19) age (years) mean (sd) 54 (8.91) 43 (11.07) gender male N (%) 13 (65%) 12 (63%) female N (%) 7 (35%) 7 (37%) driving experience median km (range) 8500 12000 (km/last 12 month): (1200 - 25000) (2000 - 50000) driving license median (range) 34 (17 - 50) 25 (5 - 40) (years) duration of pain median (range) 12.5 (5 - 27) - (years) diagnosis (N) musculoskeletal 16 - visceral pain 2 - other 2 - opioid analgesic N fentanyl 5 (25 µg/h; 25 µg/h – 75 µg/h) (median; dose buprenorphine 1 (87.5 µg/h; --) range) oxycodone 5 (140 mg/d; 30 mg/d – 150 mg/d) hydromorphone 6 (26 mg/d; 16 mg/d – 56 mg/d) morphinesulfate 3 (120 mg/d; 60 mg/d – 140 mg/)

Study design This study was conducted according to a mixed within-between design. Treatment consisted of alcohol administration before the driving test vs. no alcohol administration and was only done in the control group. The medication of involved pain patients was not changed during the study. Hence, patients did the driving test once, whereas controls did the driving test twice. To fulfil all safety requirements according to the protocol, the first driving test of controls always had to be done sober. Accordingly, the sequence of driving could not be balanced but it was adhered to a two weeks delay between both driving sessions.

171 Procedure General remarks This study was conducted in collaboration with Maastricht University that was responsible for the driving test. Therefore the use of two instrumented cars of the Federal Highway Research Institute (BASt) was allowed to Maastricht University. All participants were Germans, medically screened at the pain outpatients department of the University Hospital of Cologne. It was verified by breath test and by urine drug screening that subjects did not use alcohol or other psychoactive substances besides those prescribed for pain treatment before each driving test. All participants completed a computer-based test of driving related skills there before taking part in the driving test. At the day of the driving test participants were picked up at Cologne (see table 3 for the timeline), brought to Maastricht and afterwards back home. All driving tests, as well as familiarization of drivers with the car, were accompanied by a Dutch licensed driving instructor. This driving instructor met participants at a rest area near the Dutch boarder. From there participants drove in the instrumented car to Maastricht University which took about 30 minutes (30 km) and involved driving on a primary highway. On their way to Maastricht participants received a training driving test aiming to learn test procedures. At Maastricht University patients were served lunch. Alcohol was administered to participants after arrival if the under-influence driving (alcohol condition) was on agenda. Driving tests started and ended at the University building. Depending on traffic conditions it took about 15 minutes to get from the University building to the starting point of the Road tracking test on the highway.

Alcohol administration The amount of alcohol needed to obtain the intended BAC of 0.5‰ was calculated by the Watson formula. This formula takes gender and weight of the individual into account as well as the time span until the intended BAC should be reached (Watson, Watson, & Batt, 1981). The total amount of alcohol was split into three parts, mixed with orange juice and administered -60, -40 and -20 minutes before the driving test. BAC was checked -40 and -20 minutes before the driving test as well as during the test. In case BAC did not rise as expected additional alcohol was administered. The last dose was given 20 minutes before the start of the driving test because a 20 minutes drive was necessary to get to the highway section on which the driving test was done.

Ethical issues The study was conducted according to the code of ethics on human experimentation established by the declaration of Helsinki (2008). The German Federal Institute for Drugs and Medical Devices and the Ethics Committee of the University Hospital of Cologne gave approval to this investigation. It was also approved by the Ethics Committee of Maastricht University.

172 Table 3: Timeline of the driving test for patients and controls.

Start End Patients Controls [hh:mm] [hh:mm] 08:30 09:00 Taking of blood, saliva and urine --- sample, check of inclusion criteria 09:00 10:30 Transfer to Dutch boarder 11:00 11:45 Meeting with driving instructor, training driving test 11:45 12:15 Lunch break at Maastricht Alcohol administration at University Maastricht University 12:15 12:30 Participant drives to starting point of driving test 12:30 13:30 Road tracking test 13:30 13:45 Return to Maastricht University 13:45 14:00 Break Alcohol administration 14:00 14:05 Participant drives to starting point of Car-following test 14:05 14:45 Car-following test 14:45 15:00 Return to Maastricht University 15:00 15:10 Break 15:10 16:45 Transfer to Cologne

Actual driving tests The driving test was conducted in an instrumented car. It comprised two different driving tests: (1) Road tracking test and (2) Car-following test. The Road tracking test was conducted on a primary highway (A79) near Maastricht (see figure 1) between the exits Bunde and Heerlen-Centrum. The distance between both exits was 17 km. In this test participants drove along that section six times (= three rounds from/to Bunde) which equates a distance of 100 km. The Car-following test was done on A2 between the exits Gronsveld and Moelingen. The distance between both exits was 7.5 km. For the Car-following test participants drove along that section four times.

Figure 1: Map of the route Maastricht (Bunde) - Heerlen (A79) used for the Road tracking test.

(1) Road tracking test For the Road tracking test (O'Hanlon, 1984) subjects drove 100 km in a specially instrumented car on a primary highway. Their task was to maintain a constant speed (95 km/h) and a steady lateral position on the right (slower) traffic lane. An optical measurement device mounted at the grill (line- scan camera) continuously measured lateral distance between the vehicle and the left road marking. Drivers were allowed to overtake slower vehicles and were advised to change back to the right traffic lane after overtaking as soon as possible. Data on lateral position was stored together with data on

173 vehicle movements (e.g. speed, lateral and longitudinal acceleration) and inputs by the driver (e.g. steering wheel angle, throttle position) on a computer in the car. The technical equipment also provided video recordings of the drive. In the off-line editing routine the signal which was digitized at a rate of 30 Hz was down sampled to 10 Hz. To eliminate noise the signal was then filtered by a digital low-pass filter (cut-off frequency: 2 Hz). Afterwards all data segments with signal loss as well as overtaking manoeuvres were eliminated. Sections where the highway was left and entered again in the opposite direction were also removed. The remaining data was then used to calculate means and standard deviation of lateral position for every 5-km-segment of the track (equates three per section). The values of all of these segments were then averaged to get an overall measure of standard deviation of lateral position (SDLP) for the whole driving test. SDLP is a measure of weaving (see figure 2). It was used as the primary outcome measure. In addition average speed and standard deviation of speed were computed in the same way for the whole track. They were used as control measures. The test duration was 1 hour.

Figure 2: Description of SDLP (Verster, Volkerts, & Verbaten, 2002, S. 263).

(2) Car-following test The Car-following test (Ramaekers, Muntjewerff, & O'Hanlon, 1995) involved the use of two vehicles. The preceding vehicle was under an investigator's control. The subject was driving in the following vehicle. The test began with the two vehicles driving behind each other at 95 km/h. Subjects were told to follow the leading car at a distance of 50 m (time distance: 2 seconds) and to maintain that headway constant. During the test, the speed of the leading car was controlled by a modified cruise- control system. At the beginning this system was set to maintain a constant speed. The investigator started sinusoidal speed changes. The duration of one such cycle was 50 seconds: within 25 seconds the leading car decelerates 10% of the actual driving speed and then accelerates to reach the initial speed within 25 second again. Such speed change cycles were repeated 12 times at random sections during the Car-following test. Between theses speed change manoeuvres, the investigator in the leading car randomly activated the brake lights of his vehicle. The brake lights then light up for 3 seconds whereas the speed of the car remained constant. The subject was told to react to the onset of brake lights by removing his/her foot from the gas pedal as fast as possible. This procedure was repeated 12 times throughout the Car-following test.

174 Headway was continuously recorded by a radar distance sensor. This device was mounted in the grill of the instrumented car (= the following vehicle). The velocity of the leading vehicle and the initiation times of speed changes and brake lights were transmitted by Wi-Fi to the following instrumented vehicle and were stored there on a computer along with the velocity of the following vehicle, distance headway and response time to the onset of brake lights. To analyze data of the speed changes a power spectral analysis of the speed signals of both cars was done. Therefore the cycles were put in a direct sequence. The power spectral analysis revealed phase-delay between the speed cycles of both vehicles (cycle frequency = 0.02 Hz). Phase delay was converted to an indicator of time-to-speed adaptation (TSA, in seconds). Gain and coherence are control measures and were also provided by power spectral analysis. Gain is the amplification factor between both speed cycles. It indicates the amount of overshoot when participants adapt driving speed to the speed changes of the leading car. Coherence is a measure to control for correspondence between both speed signals. Brake reaction time (BRT, in seconds) is as secondary outcome measure. It was calculated of the delay between the onset of the brake lights of the leading car and release of the throttle by the subject who was driving the following car. Single values for all events were averaged across the whole driving test to get a measure of brake reaction time for the whole drive. Test duration was 25 minutes.

Karolinska sleepiness scale Subjective ratings of sleepiness were assessed by Karolinska sleepiness scale (Akerstedt & Gillberg, 1990) before and after the Road tracking test and the Car-following test as well as after each section of the Road tracking test immediately before the highway was left at the exit to change direction. Scores on the scales were ranging from 1 “extremely alert”to 9 “very sleepy, great effort to keep alert, fighting against sleep”.

Rating Scale Mental Effort (RSME; Zijlstra, 1993) With this scale participants have had to indicate the amount of effort they invested in the task by flagging a 15cm long axis ranging from 0 to 150. Segments of the axis were tagged with verbal categories indicating different levels of effort. Effort was assessed after the Road tracking test and after the Car-following test.

Additional self rating scales Visual analogue scales (10cm long) were used to assess self-ratings of pain intensity, quality of driving and intensity of alcohol effects. The latter was only done for controls whereas pain intensity was only assessed in patients. During the Road tracking test quality of driving was also assessed at the end of each section while approaching the exit.

Collection of body fluids For quantification of opioid concentrations in body fluids a blood sample (10 mL) was taken of every patient at the University of Cologne before he/she was brought to Maastricht. Dried blood spots were prepared as well as serum by centrifugation of half of the whole blood sample. Whole blood samples

175 and serum samples were frozen at -20°C and forwarded to the Heidelberg University where samples were analysed using solid phase extraction and gas chromatography with mass spectrometric detection. Dried blood spots were stored in a cool dark place. StatSure saliva sampler was used to collect oral fluids for the quantitative analysis of analgesics concentrations by GC-MS. Until analysis was done at Gent University these samples were also stored at -20°C.

Statistical analysis

SPSS 19 for Windows was used for all statistical analysis. The testing procedure consisted of three steps. 1) Superiority testing by ANOVA for between group comparisons of patients and controls (sober). In order to find out if differences between these groups are caused by opioid treatment, by the underlying pain disorder or by age differences, age and pain intensity were used as covariates in the ANOVA. 2) Superiority testing by general linear model (GLM) repeated measures ANOVA with alcohol (two levels: 0.5‰ vs. sober) as main within subject factor to compare performance of controls driving sober and under influence of alcohol. 3) In case superiority was proven, equivalence testing based on difference scores was conducted in addition (patients’SDLP – average SDLP of the sober controls). An alcohol criterion was calculated (= averaged difference scores of SDLP of controls alcoholized – controls sober). Equivalence testing assessed whether this alcohol criterion falls within the 95% CI for the difference scores of patients. If that was the case the difference between groups was considered equivalent to a BAC of 0.5‰ and therefore relevant for traffic safety. If the 95% CI was below the alcohol criterion value, the difference between patients and healthy controls was not considered to be relevant.

Results

Between March 2010 and September 2010 20 pain patients and 19 healthy controls completed the driving test. Because of the great expenses of the driving test only participants fulfilling the inclusion criteria were included into the driving test.

Alcohol levels Dräger Alcotest 6510 was used to measure BAC in the course of the driving test. On average the participants received 47 mL alcohol (96 Vol.) before the Road tracking test. Depending on gender and weight of the participants this amount differs between individuals (SD = 10.61; min = 32.5 mL; max = 62.9 mL). The intention was to raise drivers’BAC to 0.5‰ before the Road tracking test and to keep it on that level for the duration of both driving tests. Therefore BAC was checked four times in the course of the Road tracking test and also at the beginning and at the end of the Car-following test. As can be seen from table 4, the intended BAC was reached before both driving tests. In case it was below 0.45‰ after the first round of the Road tracking test, additional alcohol was administered during a break at the roadside. This was necessary for seven (41%) of the drivers. Additional alcohol had to be administered to all drivers in the break between Road tracking test and Car-following test which was done at the University building.

176 Table 4: BAC (‰ ) measured during testing (data of controls).

Time of measurement ǻt Average BAC SD BAC N [min.] [‰ ] Alcohol 1st dose -60 -- -- 17 administration 2nd dose -40 -- -- 17 3rd dosed -20 -- -- 17 Road tracking Start 0 0.49 0.11 17 test after 1st round 23 0.49 0.11 17 after 2nd round 46 0.48 0.07 17 after 3rd round 69 0.39 0.07 17 (end) Break (alcohol administration) -- -- 17 Car-following Start 105 0.45 0.07 17 test End 140 0.35 0.05 17

Toxicology Toxicological analysis of the concentrations of the active agents in whole blood (B) and plasma (p) revealed the following concentrations in the different specimen (see table 5).

Table 5: Average substance concentrations (range) in whole blood (B) and in plasma (p); N = number of samples available. active agent B [ng/mL] p [ng/mL] N buprenorphine 0.15 ( – ) 0.30 1 fentanyl 0.48 (0.18 – 4.0) 0.61 (0.20 – 4.24) 5 hydromorphone 10.80 (2.31 – 20.17) 9.85 (2.41 – 19.71) 6 morphine 210.56 (97.77 – 215.65) 205.75 (91.78 – 206.70) 3 oxycodone 70.02 (47.17 – 122.55) 54.33 (30.62 – 89.84) 5

Traffic conditions On-the-road driving tests are exposed to influences of traffic conditions. High traffic density during the Road tracking test might reduce fatigue of the driver because driving in dense traffic is less monotonous. The number of overtaking manoeuvres was used as an indicator of traffic density. As could be seen from figure 3 the drivers had to overtake equally often across all conditions indicating comparable traffic density. Speed changes events in the Car-following test could only be initiated by the driver of the leading car in case no slower vehicles were ahead and no faster cars were approaching. For both events both cars have to drive at the same speed. The investigator in the leading car had the instruction to initiate at least 8 speed cycles and at least 8 brake events. Figure 4 indicates that these requirements had been fulfilled across all three conditions.

177 Figure 3: Average number of overtaking Figure 4: Mean (± SD) of number of maneuvers in the Road tracking test as an successful speed cycles and brake events in indicator of traffic density reveals the Car-following test over conditions comparability between conditions (P = indicates comparability (P = patients, C- = patients, C- = controls sober, C+ = controls controls sober, C+ = controls 0.5‰ ). 0.5‰ ).

Premature termination of driving tests Only one Road tracking test had to be terminated prematurely. This was done on request of the driver, a healthy volunteer driving sober. He felt too sleepy to go on driving after 2/3 of the designated driving time. None of the Road tracking tests had to be terminated prematurely by the driving instructor due to security concerns, neither of patients nor in the alcohol condition. All Car-following tests have been completed successfully.

Performance measures Mean (SD) performances on the primary measure SDLP are shown in figure 5. Table 6 shows the numbers and the results of superiority testing. Alcohol significantly impaired SDLP (F1,16 = 18.39, p ” .001). SDLP increased on average by 2.43 cm when controls drove under influence of alcohol (0.5‰ ). Although SDLP has proven to be sensitive to impairments caused by alcohol, no significant difference between patients and controls was found (F1,35 = 2.01, n.s). Neither age (F1,35 = 2.28, n.s) nor pain intensity (F1,35 = 1.39, n.s) significantly affected SDLP. Drivers have been instructed to keep the velocity of 95 km/h. Average speed was used as control measure. Data analysis showed that drivers followed the instruction across all conditions (see table 6). Standard deviation of speed did not differ between conditions, too (see table 6). Neither age of the driver nor ratings of pain intensity affected both speed measures.

178 Figure 5: Mean (± SD) of lateral position (SDLP in cm; P = patients, C- = controls sober, C+ = controls 0.5‰ ).

The results of the Car following test can be found in table 6. No impairing effects of alcohol were found in the measures of the Car-following test. Therefore the measures have not proven to be sensitive to the alcohol effect in the study at hand. Similarly no performance differences between patients and controls were found.

Table 6: Summary of means (SD) and effects for the three conditions (P = patients, C- = controls sober, C+ = controls 0.5‰ ) in the Road tracking test and the Car-following test.

groups (m (sd)) ANOVA (p ” ) P C- C+ P vs. C- C- vs. C+ SDLP (cm) 20.53 (4.29) 17.96 (4.04) 20.74 (3.32) .166 .001 speed (km/h) 95.24 (1.34) 94.56 (0.94) 94.42 (0.75) .090 .781 SD speed 2.36 (0.66) 1.91 (0.69) 2.23 (0.95) .153 .173 (km/h) TSA 3.17 (1.00) 3.26 (0.79) 3.57 (0.91) .095 .090 (seconds) gain 1.19 (0.06) 1.20 (0.15) 1.26 (0.10) .888 .427 coherence 0.96 (0.03) 0.96 (0.02) 0.95 (0.01) .241 .465 BRT 0.93 (0.29) 0.86 (0.15) 0.95 (0.16) .320 .097 (seconds)

Subjective measures All drivers made self-ratings on several aspects related to their performance in the driving test. Fatigue and quality of driving were assessed after each section of the track. Unless otherwise stated, the following results refer to the averaged ratings of all six sections in the Road tracking test and of all four sections in the Car-following test. Other aspects like the extent of effort needed to perform the tests successfully were assessed only once after the completion of the Road tracking test as well as of the Car-following test. In the alcohol condition the degree of impairment caused by alcohol was assessed. All patients assessed the intensity of pain by rating and informed on their opinion concerning the sources of impairment.

179 Alcohol impairment Before the Road tracking test and before the Car-following test all drivers were asked if they feel fit to drive. 14 of 17 drivers (82%) felt unfit to drive, three were not sure (12%) and only one driver (6%) rated himself being still fit to drive. Before the Car-following test 15 drivers (88%) felt unfit to drive. Only two held the opinion that they were still able to drive a car safely (12%). Drivers’ratings on a 10 cm visual analogue scale ranging from zero (no effect of the alcohol at all) to 10 (strongest effect of the alcohol) showed that the effect of alcohol was clearly noticeable during both parts of the driving test (see figure 6). Accordingly they felt impaired in both parts of the driving test (see figure 7).

Figure 6: Mean (± SD) self-rated intensity of Figure 7: Mean (± SD) self-rated impairment the alcohol effect before and after both parts by alcohol on performance (0 = no of the test (0 = no effect at all / 10 = strongest impairment at all / 10 = strongest effect). impairment).

Performance The rating scales used for assessment of performance showed a slight alcohol induced decline in performance in the Road tracking test (F1,16 = 6.491, p ” .022; see also table 7) and in the Car- following test (F1,16 = 6.683, p ” .020; see also table 8). There was no significant difference between the ratings of patients and sober controls. Throughout all conditions drivers judged their performance being good: all averaged ratings were above six on a 10 cm visual analogue scale ranging from zero (very bad) to 10 (very well). Age and ratings of pain intensity did not affect these subjective ratings of performance. Pain intensity slightly increased during the Road tracking test as well as during the Car-following test (see figure 8). Patients traced back the impairment equally on pain and on pain treatment (figure 9). Overall, the intensity of pain as well as the extent of impairment caused by pain and by the side effects of the analgesics seemed to be low.

180 Figure 8: Mean (± SD) self-rated intensity of Figure 9: Mean (± SD) self-rated impairment pain before and after Road tracking test and by pain and by analgesic on performance in Car-following test (0 = no pain at all / 10 = the Road tracking test and in the Car- strongest pain). following test (0 = no impairment at all / 10 = strongest impairment).

Sleepiness Ratings on the Karolinska Sleepiness Scale (ranging from 1 “extremely alert”to 9 “very sleepy, great effort to keep alert, fighting against sleep”) differed significantly between conditions in the Road tracking test (see table 7). Controls felt slightly sleepier when driving under influence of alcohol (F1,16 =

4.486, p ” .050). Patients felt sleepier than controls in this part of the driving test (F1,35 = 5.905, p ” .020). However all averaged ratings ranged between three and five across all conditions, indicating ratings of “neither awake nor sleepy” up to “awake”. The covariates age and pain intensity did not affect ratings of sleepiness. Statistical analysis revealed no significant differences in ratings of sleepiness in the Car-following test.

Effort By means of a 15cm long scale participants indicated the effort they needed to complete the driving task. As depicted in table 7 more effort was needed to fulfil the requirements of the Road tracking test under influence of alcohol (F1,16 = 12.690, p ” .003). The same applies to the Car-following test (F1,16 = 24.586, p ” .000; see table 8). Ratings of patients and of controls did not differ. All in all low effort was necessary across all conditions.

Table 7: Means (SD) and results of superiority test of subjective measures taken in the Road tracking test (P = patients, C- = controls sober, C+ = controls 0.5‰ ).

group ANOVA (p ” ) P C- C+ P vs. C- C- vs. C+ (N = 20) (N = 19) (N=17) Performance 7.45 (1.28) 7.16 (1.22) 6.48 (1.56) .345 .022 Sleepiness 3.19 (1.39) 4.14 (1.08) 4.70 (1.69) .020 .050 (KSS) Effort 3.99 (2.94) 4.09 (2.19) 5.52 (2.15) .214 .003

181 Table 8: Means (SD) and results of superiority test of subjective measures taken in the Car- following test (P = patients, C- = controls sober, C+ = controls 0.5‰ ).

group ANOVA (p ” ) P C- C+ P vs. C- C- vs. C+ (N = 20) (N = 19) (N=17) Performance 7.66 (1.27) 7.25 (1.22) 6.57 (1.48) .299 .020 Sleepiness 3.58 (1.79) 3.51 (1.21) 4.04 (1.33) .064 .192 (KSS) Effort 3.26 (2.49) 2.64 (1.33) 4.31 (1.56) .089 .000

Correlations to primary outcome measure SDLP Correlation analysis revealed a positive relation between age and SDLP (see table 9). According to the data the older driver is the more he is weaving. This indicates a performance decrement with age. Neither gender nor driving experience or frequency of driving are significantly correlated to this primary outcome measure of the Road tracking test. Patients included into the present study were treated with different opioid analgesics at different dose levels. These analgesics differ in strength of the analgesic effect as well as in strength of the expected side effects. One way to compare different analgesics is to calculate the amount of morphine necessary to get the same analgesia. These morphine equivalence dosages are not significantly correlated to SDLP (see table 10). Correlation analysis also showed that performance of patients was not related to the duration of pain and also not related to ratings of pain intensity.

Table 9: Correlation (Pearson) between SDLP and sample characteristics (data of patients and controls driving sober).

r ” p (two-sided) N Characteristic (Pearson) Age 0.401 .011 39 Gender 0.234 .152 39 Driving experience -0.215 .188 39 (km last 12 month) Frequency of 0.232 .155 39 driving

182 Table 10: Correlation (Pearson) between SDLP, duration of pain, morphine equivalence dosage and pain intensity (only data of patients).

r ” p (two-sided) N Characteristic (Pearson) Duration of pain -0.042 .861 20 (years) morphine 0.119 .618 20 equivalence dosage pain intensity before -0.360 .119 20 driving (cm)

Discussion

In line with published studies (e.g. Verster et al., 2008) the Road tracking test has proven to be sensitive to the impairments caused by BAC of 0.5‰ . Alcohol deteriorated road-tracking performance which was indicated by a significant increase in SDLP relative to sober driving. In contrast road- tracking performance of pain patients under long term treatment with opioid analgesics of WHO level 3 was not impaired. There was no statistical significant difference in SDLP between patients and an age-independent group of healthy controls. Moreover none of the Road tracking tests of patients had to be terminated prematurely although drowsiness and fatigue are well known side effects of opioid treatment. In a recently published study the Road tracking test was used to study the effects of MDMA on actual driving performance before and after sleep deprivation (Ramaekers, 2011). Approximately 20% of driving tests had to be terminated prematurely when participants drove after a night of sleep deprivation. These subjects felt too sleepy to fulfil the monotonous driving task.

To date only one study is available in which driving performance of pain patients was assessed by the Road tracking test (Veldhuijzen et al., 2006). Other than in the present study only some of the patients had been treated with analgesics. None of them was treated with opioids but with paracetamol and/or NSAIDs. Both agents are known to have only minor side effects on cognitive tasks (Bradley & Nicholson, 1987; Wysenbeek, Klein, Nakar, & Mane, 1988) whereas opioid analgesics are associated with cognitive side effects. Overall, the mean SDLP of patients involved in the study of Veldhuijzen et al. (2006) was higher compared to that of healthy controls. Hence the results indicated worse highway driving performance of pain patients. The comparison of the absolute values from both studies showed, that SDLP was higher in both groups in the study of Veldhuijzen et al. (2006) than in the present study (25.2 cm (SD = 4.6) vs. 20.7 cm (SD = 3.4), study at hand: 20.53 cm (SD = 4.29) vs. 17.96 cm (SD = 4.04)). The deviations of absolute values in both studies might be traced back to the different characteristics of the instrumented vehicles as well as to the different technical equipment used for the assessment of lateral position. The study of Veldhuijzen et al. (2006) shows that the difference in SDLP of pain patients and matched healthy controls equates the increase in SDLP caused by 0.8‰ alcohol in healthy drivers. The study at hand has found no such difference which suggests that effective pain treatment my improve fitness to drive. Pain itself is associated with fatigue (Fishbain et al., 2003)

183 which might cause an impairment as long as pain remains untreated. However, while drawing conclusions from the actual study about road tracking performance of pain patients under stable medication, one has to consider two aspects. First of all, it was very hard to find patients who were willing to take part in the driving test. One can assume that only those who had no reasonable doubts about their fitness to drive agreed to undergo the driving test. In addition, the standard deviation of SDLP within the group of patients is rather high. This indicates performance differences between individuals.

In the Car-following test the time needed to adapt driving speed to speed changes of a leading car was measured. In addition, drivers had to react to the brake lights flash of the leading car by releasing the accelerator. Although the BAC of 0.5‰ was reached before the Car-following test and drivers felt impaired by the alcohol, both measures of reaction time showed no difference between sober driving and driving under influence of alcohol. Therefore in the actual study those measures have not proven to be sensitive to alcohol induced impairments. There was also no difference evident between patients and healthy controls. Accordingly, the ability of pain patients to react was not impaired.

Several studies investigated driving ability of chronic pain patients receiving opioid analgesics by measuring driving related skills by the aid of computer-based test systems (Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003). Compared to healthy controls with BAC of 0.5‰ no impairment was found in those studies. Therefore the results of computer-based tests of driving related skills are in line with the results of this on-the-road test.

Future studies will need to examine a larger sample of chronic pain patients in order to find out if the results could be generalized. In the present study patients treated with different analgesics at different dosages were involved. Due to the low sample size no conclusions on impairments caused by a certain active agent at a certain dosage could have been drawn. Therefore future studies must examine a sample that is more comparable related to the psychoactive agents used for pain treatment. Since age and increase in SDLP are positively correlated, more information is needed on the relation between age and performance in the Road tracking test.

One major problem while conducting this study was to find a sufficient number of patients who were willing to participate in the driving test. The obvious question is why so many patients refused to participate. More information on their performance is needed. Moreover both the Road tracking test and the Car-following test comprise tasks drivers have to cope with while driving on highways. Therefore the influence of pain and pain treatment on skills drivers need to deal with tasks typical for city traffic has to be subject of future studies. All in all the study at hand suggests that patients with chronic pain on stable doses of opioid analgesics are not necessarily unfit to drive. Nevertheless it is very important to take into consideration that individual differences in cognitive and motor performance might exist.

184 References

Akerstedt, T. & Gillberg, M. (1990). Subjective and objective sleepiness in the active individual. 27. Int.J.Neurosci., 52, 29-37. Bradley, C. M. & Nicholson, A. N. (1987). Studies on performance with aspirin and paracetamol and with the centrally acting analgesics meptazinol and pentazocine. Eur.J.Clin.Pharmacol., 32, 135-139. Brookhuis, K., de Waard, D., & Mulder, B. (1994). Measuring driving performance by car-following in traffic. Ergonomics, 37, 427-434. Dagtekin, O., Gerbershagen, H. J., Wagner, W., Petzke, F., Radbruch, L., & Sabatowski, R. (2007). Assessing cognitive and psychomotor performance under long-term treatment with transdermal buprenorphine in chronic noncancer pain patients. Anesth.Analg., 105, 1442-1448. Dellemijn, P. L. I., van Duijn, H., & Vanneste, J. A. L. (1998). Prolonged treatment with transdermal fentanyl in neuropathic pain. Journal of Pain and Symptom Management, 16, 220-229. Dick, B. D. & Rashiq, S. (2007). Disruption of attention and working memory traces in individuals with chronic pain. Anesth.Analg., 104, 1223-9, tables. Drummer, O. H., Gerostamoulos, J., Batziris, H., Chu, M., Caplehorn, J., Robertson, M. D. et al. (2004). The involvement of drugs in drivers of motor vehicles killed in Australian road traffic crashes. Accid.Anal.Prev., 36, 239-248. Eccleston, C., Crombez, G., Aldrich, S., & Stannard, C. (1997). Attention and somatic awareness in chronic pain. Pain, 72, 209-215. Fishbain, D. A., Cole, B., Cutler, R. B., Lewis, J., Rosomoff, H. L., & Fosomoff, R. S. (2003). Is pain fatiguing? A structured evidence-based review. Pain Med., 4, 51-62. Fishbain, D. A., Cutler, R. B., Rosomoff, H. L., & Rosomoff, R. S. (2002). Can patients taking opioids drive safely? A structured evidence-based review. J.Pain Palliat.Care Pharmacother., 16, 9-28. Fishbain, D. A., Cutler, R. B., Rosomoff, H. L., & Rosomoff, R. S. (2003). Are opioid- dependent/tolerant patients impaired in driving-related skills? A structured evidence-based review. J.Pain Symptom Manage., 25, 559-577. Gaertner, J., Elsner, F., Radbruch, L., Kolibay, F., Theisohn, M., Berghaus, G. et al. (2008). Influence of changes to daily dose of opioids on aspects of cognitive and psychomotor performance involved in driving. Schmerz., 22, 433-441. Gaertner, J., Radbruch, L., Giesecke, T., Gerbershagen, H., Petzke, F., Ostgathe, C. et al. (2006). Assessing cognition and psychomotor function under long-term treatment with controlled release oxycodone in non-cancer pain patients. Acta Anaesthesiol.Scand., 50, 664-672. Grisart, J. (2009). [Cognitive effects of opioids]. Rev.Med.Suisse, 5, 1356-3. Grisart, J. M. & Plaghki, L. H. (1999). Impaired selective attention in chronic pain patients. Eur.J.Pain, 3, 325-333. Jamison, R. N., Schein, J. R., Vallow, S., Ascher, S., Vorsanger, G. J., & Katz, N. P. (2003). Neuropsychological effects of long-term opioid use in chronic pain patients. J.Pain Symptom.Manage., 26, 913-921. Kalso, E., Edwards, J. E., Moore, R. A., & McQuay, H. J. (2004). Opioids in chronic non-cancer pain: systematic review of efficacy and safety. Pain, 112, 372-380. Kress, H. G. & Kraft, B. (2005). Opioid medication and driving ability. Eur.J.Pain, 9, 141-144.

185 Lagarde, E., Chastang, J. F., Lafont, S., Coeuret-Pellicer, M., & Chiron, M. (2005). Pain and pain treatment were associated with traffic accident involvement in a cohort of middle-aged workers. Journal of Clinical Epidemiology, 58, 524-531. Lenné, M., Dietze, P., Rumbold, G., Redman, J. R., & Triggs, T. J. (2000). Opioid dependence and driving ability: a review in the context of proposed legislative change in Victoria 8. Drug and Alcohol Review, 19, 427-439. McCracken, L. M. & Iverson, G. L. (2001). Predicting Complaints of Impaired Cognitive Functioning in Patients with Chronic Pain. Journal of Pain and Symptom Management, 21, 392-396. McQuay, H. (1999). Opioids in pain management. Lancet, 353, 2229-2232. Mersky, H. & Bogduk, N. (1994). Classification of chronic pain: Descriptions of chronic pain syndromes and definitions of pain terms. (2 ed.) Seattle: IASP Press. Moore, R. A. & McQuay, H. J. (2005). Prevalence of opioid adverse events in chronic non-malignant pain: systematic review of randomised trials of oral opioids. Arthritis Res.Ther., 7, 1046-1051. Movig, K. L., Mathijssen, M. P., Nagel, P. H., van Egmond, T., DE GIER, J. J., Leufkens, H. G. et al. (2004). Psychoactive substance use and the risk of motor vehicle accidents. Accid.Anal.Prev., 36, 631-636. Nickel, R. & Raspe, H. H. (2001). Chronischer Schmerz: Epidemiologie und Inanspruchnahme. Nervenarzt, 72, 897-906. O'Hanlon, J. F. (1984). Driving performance under the influence of drugs: Rationale for, and application of, a new test. Br J Clin Pharmacol, 18 (Suppl. 1), 121-129. Portenoy, R. K. (2000). Current pharmacotherapy of chronic pain. J.Pain Symptom.Manage., 19, 16- 20. Ramaekers, J. G. (2011). Effects of stimulant drugs on actual and simulated driving (Rep. No. 1.2.1 (under review)). Maastricht: Maastricht University. Ramaekers, J. G., Muntjewerff, N. D., & O'Hanlon, J. F. (1995). A comparative study of acute and subchronic effects of dothiepin, fluoxetine and placebo on psychomotor and actual driving performance. Br J Clin Pharmacol., 39, 397-404. Sabatowski, R., Kaiser, U., & Gossrau, G. (2010). Chronische Schmerzen: Fahrsicherheitsrelevante Aspekte beim Einsatz von Opioiden [Opioids in the management of chronic pain and driving ability]. Anasthesiol.Intensivmed.Notfallmed.Schmerzther., 45, 384-385. Sabatowski, R., Schwalen, S., Rettig, K., Herberg, K. W., Kasper, S. M., & Radbruch, L. (2003). Driving ability under long-term treatment with transdermal fentanyl. J Pain Symptom Manage., 25, 38- 47. Schmidt, H., Lowenstein, O., Hesselbarth, S., Klinge, V., & Michels, H. (2010). [S3 guidelines LONTS (Long-term administration of opioids for non-tumor pain)]. Schmerz., 24, 71-72. Sjogren, P., Olsen, A. K., Thomsen, A. B., & Dalberg, J. (2000). Neuropsychological performance in cancer patients: the role of oral opioids, pain and performance status. Pain, 86, 237-245. Strumpf, M., Willweber-Strumpf, A., Herberg, K. W., & Zenz, M. (2005). Sicherheitsrelevante Leistungsfähigkeit von Schmerzpatienten unter Langzeitopioidtherapie. Der Schmerz, 19, 426-433. Tassain, V., Attal, N., Fletcher, D., Brasseur, L., Degieux, P., Chauvin, M. et al. (2003). Long term effects of oral sustained release morphine on neuropsychological performance in patients with chronic non-cancer pain. Pain, 104, 389-400.

186 Vainio, A., Ollila, J., Matikainen, E., Rosenberg, P., & Kalso, E. (1995). Driving ability in cancer patients receiving long-term morphine analgesia. The Lancet, 346, 667-670. Veldhuijzen, D. S., Karscha, A. M., & van Wijck, A. J. M. (2009). Drugs, driving and traffic safety in acute and chronic pain. In J.C.Verster, S. R. Pandi-Perumal, J. G. Ramaekers, & J. J. DE GIER (Eds.), Drugs, driving and traffic safety (pp. 355-370). Basel: Birkhäuser. Veldhuijzen, D. S., van Wijck, A. J., Wille, F., Verster, J. C., Kenemans, J. L., Kalkman, C. J. et al. (2006). Effect of chronic nonmalignant pain on highway driving performance. Pain, 122, 28-35. Verster, J. & Ramaekers, J. G. (2008). The on-the-road driving test. In J.C.Verster, S. R. Pandi- Perumal, J. G. Ramaekers, & J. J. DE GIER (Eds.), Drugs, driving and traffic safety (pp. 83-90). Basel: Birkhäuser. Verster, J. C., Volkerts, E. R., & Verbaten, M. N. (2002). Effects of Alprazolam on driving ability, memory functioning and psychomotor performance: A randomized, placebo-controlled study. Neuropsychopharmacology, 27, 260-269. Watson, P. E., Watson, I. D., & Batt, R. D. (1981). Prediction of blood alcohol concentrations in human subjects. Updating the Widmark Equation. J.Stud.Alcohol, 42, 547-556. Wysenbeek, A. J., Klein, Z., Nakar, S., & Mane, R. (1988). Assessment of cognitive function in elderly patients treated with naproxen. A prospective study. Clin.Exp.Rheumatol., 6, 399-400. Zacny, J. P. (1995). A review of the effects of opioids on psychomotor and cognitive functioning in humans. Experimental and Clinical Psychopharmacology, 3, 432-466. Zijlstra, F. R. H. (1993). Efficiency in work behavior. A design approach for modern tools. PhD thesis, Delft University of Technology, Delft, The Netherlands.

187 Chapter 9: Effects of opioid analgesics on driving related skills

Markus Schumacher*, Anja Knoche*, Frank Petzke‡, Ricarda Jantos†,

* Federal Highway Research Institute (BASt) Section U3 / Traffic Psychology, Traffic Medicine Bruederstraße 53 D-51427 Bergisch Gladbach, Germany

‡ Georg-August-Universität Schmerz-Tagesklinik und -Ambulanz Zentrum Anästhesiologie, Rettungs- und Intensivmedizin Robert-Koch-Str. 40 37075 Göttingen, Germany

† Universität Heidelberg Institut für Rechts- und Verkehrsmedizin Voß-Straße 2 D- 69115 Heidelberg, Germany

188 Abstract

Nowadays morphine-like drugs, called opioids (e.g. fentanyl, oxycodone, buprenorphine, and hydromorphone) were widely used to treat chronic non-cancer pain conditions. Some studies already have demonstrated that long term treatment with opioids for non-cancer pain does not impair driving related skills. Whereas today many pain patients are treated with drug combinations, these studies have focused on specific opioids. Twenty-six patients suffering from chronic non-cancer pain on long term treatment with opioid analgesics (fentanyl, buprenorphine, oxycodon, hydromorphone or morphine) aged between 35 and 68 years participated in this study. Their performance in a computer- based test of driving related skills (Vienna Test System) was compared to the performance of an age independent sample of healthy controls. Their mean (sd) age was 43 (10.68) years. Performance of 21 controls with blood alcohol concentration (BAC) of 0.5‰ was used as reference for the impairment. The test battery assessed stress tolerance, visual orientation ability, concentration, attention and reaction speed. These areas of performance are considered to be relevant for fitness to drive according to the German Driving Licensing act (FeV). A combination score was calculated of the set of tests as primary outcome measure. Findings Patients performed worse than controls with respect to this primary outcome measure. No alcohol induced impairment could have been demonstrated. Chronic pain patients under long term treatment with opioids show impairments in driving related skills. Since there was no impact of alcohol on performance, no conclusion on the size of the impairment can be drawn.

Keywords Chronic non-cancer pain, opioids, driving ability, psychomotor performance, Vienna test system

189 Introduction

According to a recent review of publications the prevalence of chronic pain in the general population ranges from 10% to over 40% (Nickel & Raspe, 2001). Pain is classified being chronic when it persists for three month or longer (Mersky & Bogduk, 1994). The pharmacological management of chronic pain has changed over the last decade. Based on an optimistic assessment of the beneficial effects of morphine-like drugs, called opioids (e.g. fentanyl, oxycodone, buprenorphine, and hydromorphone ) on pain intensity and functional capabilities, various opioids were widely used to treat chronic non- cancer pain conditions (Portenoy, 2000). More recently it became evident that the uncritical use of opioids may also cause considerable problems (e.g. dose escalation due to lack of relevant effect, development of tolerance with loss of efficacy). Therefore Guidelines like LONTS (Long-Term Treatment with Opioids for Non-Cancer Pain by the German Association for the Study of Pain (Schmidt, Lowenstein, Hesselbarth, Klinge, & Michels, 2010) have formulated a more cautious approach to opioid therapy in patients with chronic non-cancer pain. The availability and clinical evidence for efficacy of better tolerated neuromodulatory drugs like gabapentin and pregabalin, as well as both traditional and newer antidepressant agents like amitriptyline and duloxetine have broadened the pharmacological treatment spectrum. These agents are specifically effective in conditions where neuropathic or mixed pain aetiology was found as well as in musculoskeletal pain syndromes where a central hypersensitivity may be relevant. These developments have led to combination therapy in many patients with chronic pain. Although there is little scientific evidence to support this approach, it is widely established and accepted in clinical practice. Hence today monotherapy with opioids is the exception rather than the rule. It is well documented that cognitive functions like attention are impaired in chronic pain patients (e.g. Dick & Rashiq, 2007; Eccleston, Crombez, Aldrich, & Stannard, 1997; Grisart & Plaghki, 1999) whereas relief of pain might diminish these impairments (Dick et al., 2007; Eccleston et al., 1997; Grisart, 2009). Deficits in attention may lead to reduced fitness to drive. Lagarde, Chastang, Lafont, Coeuret-Pellicer, & Chiron (2005) showed that pain alone or in combination with pain treatment was associated with an increased risk of being involved in traffic accidents. However it was not possible in this study to distinguish between the negative effects of pain itself and those of pain treatment. Opioids are often prescribed in the treatment of chronic pain (McQuay, 1999) but are associated with side effects in the gastrointestinal system as well as in the central nervous system. Drowsiness and sleepiness are frequently reported side effects (Kalso, Edwards, Moore, & McQuay, 2004; Moore & McQuay, 2005). Several studies showed that driving under the influence of sedating medications or drugs, especially of alcohol, leads to an increase in accident risk (Drummer et al., 2004; Movig et al., 2004). The prevalence of drivers tested positive for opioids in a recent study ranged from 1% to 5%. Overall the association between opioids and culpability was very low (Drummer et al., 2004). Reviews on studies published earlier supported this notion that the prevalence of opioid use in traffic is very low and is not associated with motor vehicle accidents or motor vehicle fatalities (Fishbain, Cutler, Rosomoff, & Rosomoff, 2002; Fishbain, Cutler, Rosomoff, & Rosomoff, 2003; Lenné, Dietze, Rumbold, Redman, & Triggs, 2000). Beside these epidemiological studies a limited number of experimental studies have been published aiming at measuring driving ability of chronic pain patients

190 under medical treatment (see Kress & Kraft, 2005 for a review and Veldhuijzen, Karscha, & van Wijck, 2009 for an overview). No performance decrement in skills related to driving was found in cancer pain patients treated with stable doses of oral morphine (Vainio, Ollila, Matikainen, Rosenberg, & Kalso, 1995). Sabatowski et al. (2003) used the Vienna Test System (VTS), a computerized test battery that is also used for the assessment of fitness to drive of traffic offenders in Germany. They assessed driving ability of non- cancer pain patients under long-term treatment with transdermal fentanyl. Performance of patients was compared to the performance of matched healthy controls tested with blood alcohol concentration (BAC) of 0.5‰ . The results showed that performance of patients was non-inferior to that of controls. The results of two other studies which were done with the same experimental design and the same test battery but with different opioids were similar. No overall impairment was found for the long-term use of transdermal buprenorphine (Dagtekin et al., 2007) and also no impairment was found for the long-term treatment with controlled release oxycodone (Gaertner et al., 2006). The authors conclude from their studies that there is no general impairment due to long-term treatment with opioids. This conclusion is in line with the conclusion of Strumpf, Willweber-Strumpf, Herberg, & Zenz (2005). Gaertner et al. (2008) examined the effects of opioid drug dose changes on cognitive and psychomotor performances relevant for driving. Seven days after dose change, no deterioration was observed. Tassain et al. (2003) analyzed the performance of pain patients before the initiation of opioid therapy with slow release morphine. This initial performance was compared to the performance after 3, 6 and 12 month under stable medication. Stable opioid medication did not lead to a deterioration of cognitive functioning during long-term application. Patients with chronic pain often report cognitive complaints (McCracken & Iverson, 2001) and it is well known that there exists an association between fatigue and pain (Fishbain et al., 2003). The performance impairments might be caused by side effects of opioids. These side effects decline over time (Dellemijn, van Duijn, & Vanneste, 1998; Tassain et al., 2003) or test scores on cognitive ability and psychomotor function are even increasing over time (Jamison et al., 2003). In addition, those who use opioids occasionally or habitually (treatment of chronic pain or opioid abusers) are much less likely show signs of impairment (Zacny, 1995). Moreover it was proposed that the impairing effect of pain is more profound than the impairment caused by side effects of the opioids (Sjogren, Olsen, Thomsen, & Dalberg, 2000). This is further supported by a study by Veldhuijzen et al. (2006).They found pain itself to significantly affect driving performance. In their study an on-the-road driving test was used to measure fitness to drive. In summary Sabatowski, Kaiser, & Gossrau (2010) state that to this date no decision can be drawn explicitly from the available data whether slow release opioid formulations lead to impaired fitness to drive or not.

Methods

In order to better reflect the collective of pain patients treated with combination therapy, in this study chronic pain patients were enclosed treated with a variable mixture of other compounds in addition to a level 3 opioid. For this reason the sample reflects the current standard of pharmacological management of chronic pain. It is more helpful to study this combination therapy, with the opioid being the most potent drug affecting the ability to drive, to gain a realistic and clinically relevant insight into

191 driving ability of patients suffering from chronic pain, than focusing on the specific effects of a specific opioid. Performance on driving related skills of patients receiving opioid analgesic of WHO level 3 for chronic non-cancer pain treatment is compared to performance of healthy controls. Performance of the control group with BAC of 0.5‰ was used as a reference. Driving ability was measured by a computerized test of driving related skills that fulfils the legal regulations in Germany for the assessment of fitness to drive. Whereas patients had to do this computer-based test once, controls had to do it twice: once sober and once under the influence of alcohol. Due to organisational restrictions it was not possible to test half of the controls first sober and the others first under influence of alcohol. In order to compensate for potential learning effects it was ensured that there was a time span of at least two weeks between both test sessions.

Subjects All subjects gave written informed consent, possessed a valid driver’s license for passenger cars, travelled at least 2000 km by car in the last preceding 12 month and were driving on a regular basis (at least once a week). This study aimed at comparing the performance of pain patients to a representative cross-section of the driving population. Therefore controls were matched to four age groups (20-30, 30-40, 40-50, 50-65). The prevalence of chronic pain increases with age (Nickel et al., 2001). Overall women are more often affected by most of the chronic pain syndroms but men are more likely to receive high-dose opioids (Mailis-Gagnon et al., 2011). Accordingly the average age differs between both groups as can be seen in table 2. All subjects were given financial compensation for participating in this study to account for loss of earnings and travel costs.

Patients No study related changes in prescribed medication were made. 15 male and 11 female patients (see table 2) of the pain outpatient department of the University Hospital of Cologne aged between 35 and 68 (m= 54, sd = 8.28) suffering from non-cancer pain responsive to opioids were enrolled into the study. They had been treated for at least four weeks with either transdermal fentanyl (• 12 ȝg/h), transdermal buprenorphine (10ȝg/h), slow release oxycodone (• 10 mg/day), slow-release oxycodone combined with naloxone (• 10 mg/day), slow release hydromorphone (• 4 mg /day) or slow release morphine (• 20 mg/day). Table 2 shows the actual frequencies of the analgesics and the duration of opioid treatment. The dosage of the opioids remained unchanged for at least 14 days prior to the assessment. To meet the inclusion criteria co-medication with NSAIDs, anticonvulsants and antidepressants had to be on a stable dose during the past 14 days. In addition the daily intake of antidepressants and anticonvulsant had to be below a given dose limit (see table 1) and patients had to restrain from intake of immediate release opioids, benzodiazepines, barbiturates and alcohol at least two days prior to the assessment.

192 Table 1: Dose limits for co-medication. antidepressants anticonvulsants amitryptilin ” 75mg carbamazepin ” 1200mg, doxepin ” 75mg oxcarbazepin ” 1800mg, imipramin ” 75mg gabapentin ” 2400mg, trazodon ” 100mg pregabalin ” 600mg sertralin ” 50mg fluoxetin ” 20mg fluvoxamin ”75mg duloxetin ” 120mg venlafaxin ” 225mg citalopram ”10mg

Controls 13 male and 8 female healthy controls participated in this study. Their mean (sd) age was 43 (10.68). Participants were selected from a pool of volunteers that previously had given consent to participate in studies conducted by BASt. Exclusion criteria were a history of alcohol addiction or abuse, intake of medications prohibiting the consumption of alcohol or medicinal or recreational use of psychoactive substances with known impairing effect on driving ability. Women who were pregnant or breast feeding or women of child-bearing potential who were not using a highly effective contraception method with a pearl-index ” 1 were also excluded due to the alcohol administration during the course of the study. 12 male (92%) and 6 female (75%) participants completed the test in both conditions: sober as well as under the influence of alcohol. The total number of controls of which data of both conditions is available was 18 (86%).

Study design This study was conducted according to a mixed within-between design. Treatment consisted of alcohol administration before the test vs. no alcohol administration and was only done in the control group. The amount of alcohol necessary to obtain the intended BAC of 0.5‰ was calculated by the Watson formula. This formula takes into account gender and weight of the individual as well as the time span until the intended BAC should be reached (Watson, Watson, & Batt, 1981). Alcohol was mixed with orange juice before administration. Pain patients were enclosed with their existing medication and no study related change in medication was done. Hence patients did the computer-based test once, whereas controls did the test twice. The first test was always done sober, the second test at least two weeks later under the influence of alcohol.

193 Table 2: Sample characteristics. characterisitic Patients Controls (N = 26) (N = 21) age (years) mean (sd) 54.00 (8.28) 43.10 (10.68) gender male N (%) 15 (58) 13 (62) female N (%) 11 (42) 8 (38) driving experience median km (range) 7500 12000 (km/last 12 month): (1200 - 25000) (2000 - 50000) driving license median (range) 34 (17 - 50) 25 (5 - 40) (years) duration of pain median (range) 12.50 (5 - 40) - (years) duration of opioid median (range) 27.00 (2-192) treatment (month) diagnosis (N) musculoskeletal 20 - visceral pain 3 - other 3 - opioid analgesic N fentanyl 5 (25µg/h; 25µg/h - 75µg/h) (median; dose range) buprenorphine 4 (43.75µg/h; 35µg/h - 88µg/h) oxycodone 5 (140mg/d; 30mg/d – 150mg/d) hydromorphone 8 (20mg/d; 8mg/d – 56mg/d) morphinesulfate 4 (130mg/d; 60mg/d – 200mg/)

Procedure The study was conducted according to the code of ethics on human experimentation established by the declaration of Helsinki (2008). The German Federal Institute for Drugs and Medical Devices (BfARM) and the ethics committee of the University Hospital of Cologne gave approval for this investigation. After the participants had been informed about the study and had given their written informed consent, their demographic data, data on their driving experience and medical history was recorded and fulfilment of the inclusion and exclusion criteria was checked. Prior to the computer based testing blood, saliva, and urine samples were taken from all patients. Urine of all participants was screened for psychoactive substances to control for intake of medications not prescribed or recreational drug use. Also an alcohol breath test was done in all subjects. Pain intensity was rated by patients before and after the testing. For the controls in the alcohol condition the total amount of alcohol was split into two parts, mixed with orange juice and administered 30 min and 15 min before the test. In case the actual BAC level was too low at the testing time, additional alcohol was administered and BAC level was checked again after an adequate time span. Completion of all eight designated tests of the test battery took 90 minutes. In order to prevent excessive exhaustion patients were allowed a 10 minutes break after they had completed the first five

194 tests (after 60 minutes approximately). Testing of healthy volunteers in the alcohol condition was also interrupted at the same time. During this break BAC was checked and additional alcohol was administered in case BAC was below 0.45‰ .

Vienna Test System (VTS) Brief description After withdrawal of a driver’s license (e.g. due to drunk driving) in Germany an expert assessment of the fitness to drive is needed before a driver’s license can be reinstated. According to the German Driving Licensing Act (FeV) five skills considered to be relevant for safe driving must be assessed by using accredited computer-based testing devices. These skills are: stress tolerance, visual orientation ability, concentration, attention and reaction speed (BASt, 2009; Schuhfried GmbH, 2009). The Vienna Test System used in this study is such a device accredited to be eligible within the scope of the assessment of fitness to drive. Five different performance tests are recommended by the manufacturer to assess the skills mentioned above. Table 3 gives an overview of the respective skills and the tests appropriate to assess them (Schubert, Schneider, Eisenmenger, & Stephan, 2005; Schuhfried GmbH, 2009). In this study three additional tests were used to obtain a deeper insight into potential performance decrements caused by sedating psychoactive substances. 2HAND test was used to measure potential impairments in eye-hand and hand-hand coordination. VIGIL test was used to measure potential attention decrements under monotonous tasks conditions. According to the latest revision of FeV VIGIL can be done additionally if participants are prone to daytime sleepiness. As sleepiness is one of the common side effects of opioids this test was included. WRBTV was done to quantify the level of risk accepted in traffic situations.

Table 3: Overview of skills required by German Driving Licensing Act (FeV) and tests of the test battery covering these skills (Schubert et al., 2005; Schuhfried GmbH, 2009). If a test assesses several skills, the skills the tests focuses on are highlighted (Schuhfried GmbH, 2009). Not mandatory tests according to FeV and the guidelines of the manufacturer are identified (last column); the tests are sorted in the testing sequence; the 10 minute break was placed before the LVT. stress visual concentration attention reaction mandat tolerance orientati speed ory? on

DT X X X yes COG X X yes TAVTMB X X X yes

2HAND - - - - - no

VIGIL - - - - - no LVT X X yes

RT X yes

WRBTV - - - - - no

195 The Vienna Test System consists of a central processing unit, a display, a response panel and response pedals (see figure 1). All tests were presented in a standardized way and always in the same sequence by the computer system. Each test started with an explanation on the monitor. Data collection began after a training phase had been completed successfully. In case the training goal was not reached, the investigator was informed by the system and gave additional explanations. Testing was only performed, if the training phase was completed successfully. It took about 90 minutes to complete all tests.

Figure 1: Vienna Test System.

In the following section the different tests of the test battery are briefly described. All pictures are taken from the website of the manufacturer (www.schuhfried.com). A more detailed description of all tests can also be found there. The duration of the tests varies depending on the test form (some of them adapt to the working speed of the participant) and the time needed for instruction and training phase.

Determination Test (DT) Visual and acoustic stimuli are presented to the participant. The participants’task is to press the corresponding buttons on the response panel or the pedals. In the present study test form S2 was used. Since the test is adaptive to the working speed of the participant overall duration varied between 10 to 15 minutes.

Cognitrone (COG) It is the respondents’task to compare abstract figures and to indicate whether there are corresponding figures or not by pressing a button. Test form S11 was used. The working speed depends on performance of the participant. Therefore duration was 5 to 8 minutes.

196 Adaptive Tachistoscopic Traffic Perception Test (TAVTMB) In this test pictures of different traffic situations are presented to the participants for very short time periods. The participants then indicated what they had seen on those pictures by choosing the correct answers out of five options presented to them. Duration was 8 to 10 minutes.

Two-Hand Coordination Test (2HAND) The participant has to move the red dot along the paths by turning buttons with his left and right hand. Turning one button moves the dot in vertical direction whereas turning the other button moves the button in horizontal direction. Test form S3 was used. Duration was 5 to 10 minutes.

Vigilance Test (VIGIL) In this monotonous monitoring task a white dot moves along a circle of open dots. Sometimes the white dot skips one dot of the circle. The participant has to indicate these jumps by pressing a button. Test form S1 was used. Duration was 25 minutes.

Visual Pursuit Test (LVT) In this test an array of lines is presented. The participant has to indicate where the tagged line ends by pressing the corresponding number on the response panel as fast as possible. Test form S3 was used. Test duration was 3 to 5 minutes.

197 Reaction Test (RT) This test involves the presentation of visual and acoustic stimuli. The respondent has to press a button only in case the visual target stimuli and the acoustic target stimuli are presented together. Test form S3 was used and duration was 4 to 7 minutes.

Vienna Risk Taking Test Traffic (WRBTV) Videos of 24 traffic situations are presented on the screen. Each traffic situation is presented twice. At the second presentation the participant has to indicate when the situation becomes too dangerous by stopping the video. Duration was 15 to 20 minutes.

Testscores and calculation of sum score (primary outcome measure) For the different tests, specific main variables, secondary variables and additional results were used to describe the participants’performance. These variables have been defined by the manufacturer and are specific for the test and the test version. The performance rating is based on the main and secondary variables hence only these variables are described in table 4 and were used in the further analysis. A more detailed description of all variables and the available additional results can be found in the manuals referred to in that table. Besides raw values, percentile ranks for the individual’s performance with respect to an age- independent reference group are provided for all main and secondary variables by the test system. According to FeV absence of fitness to drive was assumed if the results of an individual were below the 16th-percentile in any of the five tests (DT, COG, TAVTMB, LVT, RT) in any main or secondary variable. In line with that rationale all tests were classified as passed if the main and, if available, the secondary variables were above the 16th-percentile.

Comparable to published studies on driving ability of pain patients in which the same testing device was used (e.g. Sabatowski et al., 2003), a sum score was defined as a primary endpoint measure in advance. In the present study this score was calculated of all main and secondary variables of the tests after z-transformation. Since only DT, COG, TAVTMB, LVT and RT are obligatory according to FeV only those tests were included into this sum score. Hence the score differs from the one used in the study referred to above but fulfils the recommendations of the German Driving Licensing Act (FeV) as close as possible. In case there were several main and/or secondary variables available for a specific test, an average score was calculated of these variables after z-transformation and then this averaged scores was included into the sum score instead of the main and secondary variables. The

198 overall sum score of all tests as well as the scores for specific tests were calculated in the way that higher values represent a better performance. 2HAND, VIGIL and WRBTV are not mandatory according to FeV. Therefore these tests were not included into the overall sum score.

Rating scales Karolinska sleepiness scale Subjective ratings of sleepiness were assessed by the Karolinska sleepiness scale (Akerstedt & Gillberg, 1990) before and after completion of the computerized test. Scores on the scales are ranging from 1 “extremely alert”to 9 “very sleepy, great effort to keep alert, fighting against sleep”. Brief pain inventory (BPI) The German version of the Brief Pain Inventory (Radbruch et al., 1999) is a comprehensive instrument for the assessment of pain intensity and impairments caused by pain. Rating Scale Mental Effort (RSME; Zijlstra, 1993) This scale asks participants to indicate the amount of effort they had invested into a task by flagging a 15cm long axis ranging from 0 to 150. Segments of the axis are tagged with verbal categories indicating different levels of effort.

Additional self rating scales Visual analogue scales (10cm long) were used to assess self-ratings of pain intensity, quality of performance and intensity of alcohol effects. The latter was only done for controls whereas pain intensity was only assessed in patients.

Collection of body fluids For quantification of analgesics concentrations in body fluids a blood sample (10 mL) was taken of every patient at the day of testing. Dried blood spots were prepared as well as serum by centrifugation of half of the whole blood sample. Whole blood and serum samples were frozen at -20°C until analysis using solid phase extraction and gas chromatography with mass spectrometric (GC-MS) detection. Dried blood spots were stored in a cool dark place. The StatSure™ saliva sampler was used to collect oral fluids for the quantitative analysis of opioid analgesic concentrations by GC-MS. Until analysis was done these samples were also stored at -20°C.

199 Table 4: Brief description of main and secondary variables of all tests and abbreviations referred to in the following sections (*= inverted for integration into sum score). Test Main Secondary Score (s) Abbreviation Reference variable variable DT X number of correct reactions DTMV (Neuwirth & X number of false reactions* DTSV1 Benesch, 2007) X number of missings* DTSV2 COG X average reaction time* COMV (Wagner & Karner, 2008) TAVTMB X number of traffic situations TAMV (Sommer & without errors Neurwirth, X number of correct answers TASV1 2007) across all traffic situations X number of errors across all TASV2 traffic situations* 2HAND X average time needed to 2HMV1 (Puhr, 2008a) pass the track X average % of time outside 2HMV2 track X ratio 2HMV1 to 2HMV2 2HMV3 VIGIL X total number of correct VIMV1 (Puhr, 2008b) reactions X total number of false VIMV2 reactions X average reaction time VIMV3 (correct reactions) LVT X number of correct answers LVMV (Biehl, 2008) within limited time frame RT X average reaction time* RTMV1 (Prieler, 2008) (time span between signal and release of rest button) X Average motor time* (time RTMV2 span between release from rest button and button press) X variation reaction time* RTSV1 X Variation motor time* RTSV2 WRBTV average time distance WRMV (Hergovich, (sec) to critical situation Arendasy, when videos were stopped Sommer, & by participant Bognar, 2007)

200 Statistical analysis

SPSS 19 for Windows was used for all statistical analysis. The testing procedure consisted of three steps. 1) Superiority testing by ANOVA for between group comparisons of patients and controls (sober). 2) Superiority testing by general linear model (GLM) repeated measures ANOVA with alcohol (two levels: 0.5‰ vs. sober) as main within subject factor to compare performance of controls driving sober as well as under influence of alcohol. 3) Only for main measures, in case superiority of controls over patients was proven, equivalence testing based on difference scores was conducted in addition (patients’scores – average score of the sober controls). In addition an alcohol criterion was calculated (= averaged difference scores controls alcoholized – controls sober). Equivalence testing assessed whether this alcohol criterion fell within the 95% CI for the difference scores of patients. If that was the case the difference between groups was considered equivalent to a BAC of 0.5‰ and therefore relevant for traffic safety. If the 95% CI was below the alcohol criterion value, the difference between patients and healthy controls was not considered to be relevant.

Results

Between December 2009 and July 2010 a total number of 26 pain patients completed the test. The alcohol calibration proceeded from January 2010 to September 2010. Only data of patients and controls fulfilling the inclusion criteria was included into data analysis.

Alcohol levels For the alcohol calibration the intended BAC of controls was 0.5‰ . The mean levels measured by breath analysis (Dräger Alcotest 6510) before and after the two parts of the test battery are shown in table 5. At the beginning of both parts of the computer based test the intended BAC-level was reached. As expected, BAC decreased during the time of testing. Therefore additional alcohol had to be administered to 11 (62%) controls between the first and second part of the test.

Table 5: BAC (‰ ) measured during testing (data of controls). Time of measurement Average BAC SD BAC N Start part 1 0.52 0.04 18 End part 1 0.36 0.09 18 Start part 2 0.51 0.12 18 End part 2 0.41 0.06 18

201 Toxicology Toxicological analysis of the concentrations of different active agents in whole blood (B) and plasma (p) revealed the following concentrations (see table 6). Table 6: Average substance concentrations (range) in whole blood (B) and in plasma (p); N = number of samples available. active agent B p N Buprenorphine 0.36 (0.21 – 0.48) 0.34 (0.18 -0.47) 4 Fentanyl 39.04 (20.39 – 110.55) 27.38 (13.66 – 82.31) 5 Hydromorphone 5.56 (2.31 – 12.11) 5.64 (2.41 – 9.89) 8 Morphine 298.10 (82.36 – 479.10) 285.00 (85.74 – 468.85) 4 Oxycodone 39.04 (20.39 – 110.55) 27.38 (13.66 – 82.31) 5

Subjective measures Self-ratings on several aspects related to performance in the computer based test were taken. Since the tests were not done at once but in two parts with a break in between, self-ratings were assessed for both parts separately. Unless otherwise stated, the following results refer to the averaged ratings of these two parts.

Alcohol impairment Before starting the test in the alcohol condition, controls were asked if they feel fit to drive. All of them felt unfit to drive. In addition they had to indicate the intensity of the alcohol effect before and after both parts of the test on a 10 cm visual analogue scale ranging from zero (no effect at all) to 10 (strongest effect). Furthermore self ratings of the amount of impairment caused by alcohol had been assessed for both parts of the test. As can be seen in figure 2 during the whole time of testing participants’ratings are near the middle of the scale with a slight decrease in reported intensity at the end of part two. As shown in figure 3 a medium intensity of impairment was reported for both parts of the test.

Figure 2: Mean (± SD) self-rated intensity of Figure 3: Mean (± SD) self-rated impairment by alcohol effect before and after both parts of the alcohol on performance (0 = no impairment at all test (0 = no effect at all / 10 = strongest effect). / 10 = strongest impairment).

202 Pain and impairments related to pain and pain treatment Patients reported a moderate intensity of pain during the course of the computer-based test (see figure 4). Pain intensity did not change over time (F3, 75 = 0.625, p ” .601). They rated the impact of pain and of their analgesic medication as being low (see figure 5).

Figure 4: Mean (± SD) self-rated intensity of pain Figure 5: Mean (± SD) self-rated impairment before and after both parts of the test (0 = no caused by pain and by analgesic (0 = no pain at all / 10 = strongest pain). impairment at all / 10 = strongest impairment).

Effort By means of a 15cm long scale (RSM) participants indicated the effort they needed to complete the task. More effort was needed by controls to perform the test battery under influence of alcohol (F1, 17 = 5.760, p ” .028; see also table 7). No statistical significant differences in ratings were found between patients and controls. Sleepiness Scores on the Karolinska Sleepiness Scale are ranging from 1 “extremely alert”to 9 “very sleepy, great effort to keep alert, fighting against sleep”. As can be seen in table 7 on average participants rated themselves being neither alert nor sleepy in all conditions.

Performance A 10 cm visual analogue scale ranging from zero (very bad) to 10 (very well) was used to assess self- ratings of quality of performance (see table 7). Ratings of patients did not differ from those of healthy controls. But controls rated their performance worse under influence of alcohol (F1, 17 = 22.564, p ” .000).

Table 7: Means (SD) and results of superiority test of subjective measures (P = patients, C- = controls sober, C+ = controls 0.5‰ ). groups (m (sd) ANOVA (p ” ) P C- C+ P vs. C- C- vs. C+ (N = 26) (N = 21) (N=18) Effort 5.47 (2.49) 4.34 (1.35) 5.62 (2.57) .067 .028 Sleepiness 4.73 (1.74) 4.76 (1.32) 5.56 (1.98) .936 .150 (KSS) Performance 6.08 (1.31) 6.07 (1.61) 4.42 (1.29) .990 .000

203 Passed tests Table 8 shows the percentage of participants having passed all five tests. The percentage differs between groups. Only 2 out of 26 patients (8%) passed all five tests (DT, COG, TAVTMB, LVT, RT) and therefore fulfilled the prerequisites of FeV for being fit to drive. Even though the percentage of controls having passed all tests in the sober condition is higher (33%), even 14 controls (67%) failed to pass all tests. Under influence of 0.5‰ alcohol the percentage of controls having passed all tests is lower. Only 4 out of 18 controls passed all five tests (22%).

Table 8: Percentage of participants having passed / failed all five tests according to FeV between groups. Passed? Patients Controls Controls (sober) (0.05‰ ) N 26 21 18 Yes 8% 33% 22% No 92% 67% 78%

The percentage of patients and controls scoring above the 16th percentile for all main and secondary variables is shown in table 9 for all tests. It becomes immediately obvious that the percentage of participants who meet the criteria to pass the Determination test is very low in all conditions. Overall, the percentage of patients having passed the test was below that of the reference group. Alcohol seems to improve performance of healthy subjects since more of them passed LVT under influence of alcohol. In the WRBTV more patients met the 16th-percentile criteria indicating that they accept lower levels of risk in critical traffic situations.

Table 9: Percentage of participants having passed a test. For description of main and secondary variables see table 4; passed means all main and secondary variables of a test above the 16th percentile. test name patients controls controls (sober) (0.05‰ ) N 26 21 18 DT 46% 48% 39% COG 96% 100% 100% TAVTMB 69% 95% 100% LVT 88% 100% 100% RT 42% 71% 61% 2HAND 75% 76% 83% VIGIL 58% 71% 61% WRBTV 92% 71% 56%

When taking into account only those five mandatory tests, patients passed three (m = 3.42, sd = 1.03) tests on average, controls passed four tests sober (m = 4.14, sd =.793) as well as under influence of alcohol (m = 4.00, sd =.686).

204 Participants have to score above the 16th percentile on 12 main and secondary variables in total (six main variables and six secondary variables; see table 4) to meet the criteria of FeV for being fit to drive. To pass all eight tests of the test battery, participants have to score above the 16th percentile on 19 main and secondary variables in total. On average, patients scored above the 16th percentile on nine of 12 variables of the mandatory tests (see table 10). They performed worse than controls tested sober (F1, 45 = 7.637, p ” .008). Performance of patients was also worse than performance of controls under influence of alcohol (F1, 42 = 4.591, p ” .038). Alcohol did not affect the number of passed variables in the control group (F1, 17 = 0.680, n.s.). Patients’performance is also below the performance level of controls when taking into account all variables of all tests of the test battery (F1, 45 = 6.327, p ” .016). Sum score (primary endpoint) A sum score was defined as primary endpoint in advance. This score comprises the z-transformed raw values of all main and secondary variables of all five tests mandatory according to FeV. Higher values of this score indicate higher performance levels. As can be seen from figure 6 and table 10, performance of controls is nearly equal whether they are under influence of 0.5‰ BAC or not. Patients perform worse than controls (F1, 45 = 14.983, p ” .000). An additional equivalence test was done based on difference scores from control group (sober). Basically, this procedure assesses whether the alcohol criterion value (average difference sum score between controls sober and under influence of alcohol) falls within the 95%CI of difference scores between patients and controls. Figure 7 indicates that the performance difference between patients and controls is not equal to a performance change caused by alcohol.

Figure 6: Mean (± SD) sum score of all Figure 7: Average difference scores for patients mandatory tests in all groups (P = patients, C- = (95% CI) to sober controls. The alcohol criterion controls sober, C+ = controls 0.5‰ ). Higher (horizontal line = average difference scores values indicating better performance. between controls sober and under 0.5‰ BAC) is not included in the 95% CI.

205 Table 10: Summary of means (SD) and effects in the computer based test (P = patients, C- = controls sober, C+ = controls 0.5‰ ). groups (m (sd)) ANOVA (p ” ) P C- C+ P vs. C- P vs. C+ C- vs. C+ Number of variables > 16th 9.42 (2.02) 10.81 (1.21) 10.56 (1.15) .008 .038 .421 percentile (FeV only) Number of variables > 16th 15.23 (2.89) 17.00 (1.58) 16.22 (1.86) .016 .207 .076 percentile (all tests) Sum Score (z -1.82 (2.75) 1.11 (2.35) 1.33 (2.28) .000 .000 .0197 transformed)

There was a negative correlation between age and the sum score of all tests as can be seen in table 11 only for controls. This indicates that performance decreases with age. Neither gender nor driving experience or frequency of driving was significantly correlated to this primary measure in both groups. Patients included in the present study were treated with different opioid analgesics and with different dosage. Moreover, they differed in duration of pain. The strength of the analgesic effect of different opioids can be compared with reference to morphine. Therefore morphine equivalent dosages were calculated. Neither morphine equivalent dosage nor duration of pain was correlated to the performance in the test battery (see table 12).

Table 11: Correlation (Pearson) between sum score and sample characteristics of patients and of controls. Patients Controls Characteristic r (Pearson) p (two-sided) N r (Pearson) p (two-sided) N Age -.158 .441 26 -.672 .001 21 Gender -.204 .318 26 -.095 .681 21 Driving experience .085 .681 26 -.211 .358 21 (km last 12 month) Frequency of .202 .323 26 -.331 .142 21 driving

206 Table 12: Correlation (Pearson) between sum score duration of pain and morphine equivalent dose (only patient group). r p (two-sided) N Characteristic (Pearson) duration of pain .105 .609 26 morphine equivalent .244 .229 26 dose

Performance profiles In the following section the performance of patients and sober controls is compared. This section starts with the mandatory tests. Since superiority of the control group was already proven for the sum scores, testing can be regarded as closed procedure. Therefore no adjustment was made for multiple comparisons. T-test for independent samples was used to compare performance in both groups.

Determination Test (DT) DT measures stress tolerance and reaction speed. For the test, the respondent has to distinguish different stimuli (colours, sounds) and press the corresponding response buttons. As can be seen from table 13 the number of correct reactions was lower in the group of patients. This indicates that their stress tolerance and reaction speed was diminished compared to healthy controls. The number of false or missed reactions did not differ between both groups.

Table 13: Main and secondary variables of DT. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Number of correct 388.81 494.43 3.673 45 .000 reactions (103.92) (90.07) Number of false 30.15 29.81 -.053 45 .479 reactions (22.18) (21.96) Number of missings 32.42 28.95 -.726 45 .236 (16.81) (15.63)

Cognitrone (COG) Cognitrone measures attention and concentration. The respondent’s task is to compare figures and indicate if they correspond or not. No significant difference was found between patients and controls for the average time needed for correct reactions (see table 14).

207 Table 14: Main variable of COG. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Average reaction time 2.94 2.66 -1.515 45 .068 (0.69) (0.54)

Adaptive Tachistoscopic Traffic Perception Test (TAVTMB) By briefly presenting pictures of traffic situations TAVTMB measures the observational ability of the respondent. After the short presentation, the respondent has to indicate which information had been given on these pictures (multiple choice). Patients performed worse in all three variables used to describe performance in this test (see table 15). On average the number of traffic situations for which the patients indicated all relevant information correctly was lower for patients than for controls. Moreover they ticked less correct answers and made more errors in the test. Hence the patients’ ability to gain an overview of traffic situations is diminished compared to healthy controls.

Table 15: Main and secondary variables of TAVTMB. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Number of traffic 10.23 12.76 3.317 45 .001 situations without (2.82) (2.30) errors Number of correct 45.81 48.52 2.372 45 .011 answers across all (4.53) (2.94) traffic situations Number of errors 3.73 2.10 -2.002 45 .026 across all traffic (3.47) (1.55) situations

Visual Pursuit Test (LVT) LVT assesses visual orientation ability. The respondent has to track simple visual elements in a complex environment. On average, the number of correct responses of patients is lower than the number of correct responses of controls. Hence visual orientation ability of patients is lower (see table 16) because they are worse than healthy controls in obtaining an overview in a complex environment quickly and accurately.

208 Table 16: Main variable of LVT. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Number of correct 10.65 13.67 2.417 45 .010 answers within limited (4.76) (3.51) time frame

Reaction Test (RT) In order to measure reaction time, visual and acoustic stimuli are presented to the respondent in this test. A reaction is only required in case both stimuli are presented together. This test distinguishes between reaction time and motor time. Table 17 shows the results. The time span between stimulus and release of the rest button, referred to as reaction time, is longer for patients than for controls. Also the time span between release from rest button and button press, referred to as motor time, is longer for patients than for controls. Moreover, the variation of both measures is higher within the group of patients. To sum up, patients react slower to the onset of stimuli than healthy controls. Table 17: Main and secondary variables of RT. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Average reaction time 481.15 444.24 -1.784 45 .041 (70.51) (70.55) Average motor time 193.15 149.29 -2.666 45 .005 (59.29) (51.79) Variation reaction time 78.96 63.76 -2.642 45 .006 (20.41) (18.55) Variation motor time 29.65 22.48 -2.017 45 .025 (12.15) (12.09)

Two-Hand Coordination Test (2HAND) Due to motor impairments in the hands, two patients were not able to do the Two-Hand Coordination Test. For further analysis, their missing scores were replaced by mean values of their group. The mean time for passing the track was longer in the patient-group than in the control group. Patients moved the dot longer outside the track than controls (table 18). Compared to healthy controls patients are impaired in motor-coordination. They are not as good as controls in transferring visual information about deviations into motor reactions for compensation.

209 Table 18: Main and secondary variables of 2HAND. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Average time needed 49.66 37.66 -1.879 43 .034 to pass the track (1) (23.99) (17.91) Average % of time 2.80 1.65 -1.412 43 .083 outside track (2) (3.57) (1.19) Ratio (1) to (2) 5.46 5.47 .003 43 .499 (5.08) (4.99)

Vigilance Test (VIGIL) Vigilance Test revealed no differences between patients and controls (see table 19) with respect to the difference between their ability to sustain attention under monotonous task conditions.

Table 19: Main and secondary variables of VIGIL. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Total number of correct 97.65 97.43 -.232 45 .409 reactions (3.01) (3.67) Total number of false 3.92 2.33 -1.478 45 .073 reactions (4.64) (1.83) Average reaction time 0.47 0.47 -.251 45 .401 (correct reactions) (0.09) (0.07)

Vienna Risk Taking Test Traffic (WRBTV) Willingness to take risk is measured by the time distance to the critical situation in a video at which the video file is stopped by the respondent. The average time distance to the critical traffic situations at which the video files were stopped by the patients is lower than by the controls (see table 20). This indicates that patients accepted a higher level of risk in traffic situations. Table 20: Main variable of WRBTV. Variable Patients Controls T df ” p (one m (sd) m (sd) sided) Average time distance 6.68 8.02 2.805 45 .004 (sec) (1.52) (1.75)

210 Discussion

The present study demonstrated that patients with chronic pain treated with stable doses of opioid analgesics show some impairment in driving related skills compared to healthy controls. Driving ability was measured by a computerized test that fulfils the German law regulations for the assessment of fitness to drive (Vienna Test System). To fulfil the criteria of being fit to drive according to the German Driving Licensing Act (FeV), respondents have to score above the 16th percentile of an age- independent reference group on five standardized tests assessing five areas of performance: stress tolerance, visual orientation ability, concentration, attention and reaction speed (BASt, 2009; Schuhfried GmbH, 2009). The impairment becomes obvious when looking at the tests with respect to this criteria, as well as when looking at the whole set of mandatory test. In addition a sum score was defined in advance as primary outcome measure. This score included all raw values of the performance measures specified by the manufacturer for the tests fulfilling German regulations for the assessment of fitness to drive. With respect to this sum score patients perform worse than a group of age-independent healthy controls. When looking at the individual tests of the test battery it becomes obvious that stress tolerance and reaction speed was diminished compared to healthy controls as well as observational ability, visual orientation ability and motor-coordination. Therefore the results of this study are not in line with findings of studies on driving ability of pain patients treated with opioid analgesics published before (Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003). The authors of those studies conclude from their results that driving ability of patients treated with transdermal Fentanyl (Sabatowski et al., 2003), transdermal Buprenorphine (Dagtekin et al., 2007) or controlled release Oxycodone (Gaertner et al., 2006) is not impaired compared to healthy controls. Even though the same computer-based tests were used, there are some aspects in which these studies are different from the present study. These are: primary outcome measure, composition of the patient group and composition of the control group.

Primary outcome measures Overall it becomes immediately obvious that in the study at hand the percentage of patients as well as the percentage of controls that fulfil the strict requirements of being fit to drive is low: Among healthy controls only 70% passed the test measuring reaction speed which is one of the areas of performance considered being relevant for being fit to drive according to the German law. Only half of them passed the test measuring stress tolerance. Table 20 compares the percentage of patients who have passed the test in this study to the results of the published studies referred to before. Overall the percentage of patients having passed a test in the intent-to-treat group in these studies is higher.

211 Table 20: Percentage of patients having passed a test in this study and in the other published studies (xx = not assessed / no results available; results of intent-to-treat-group). test name this study Dagtekin et al., Gaertner et al., Sabatowski et 2007 5 2006 al., 2003 6 N 26 30 30 30 DT 46% 80% 66.7% 95% COG 96% 85% 72.4% 80% TAVTMB 69% 85% 82.1% 82% LVT 88% xx xx xx RT 42% xx xx xx 2HAND 75% 85% 82.8% xx VIGIL 58% 75% 79.3% xx WRBTV 92% xx xx xx

However in this study the percentage of patients who passed the tests is lower than the percentage of controls. Due to the high failure rate in some of the tests of the test battery, the percentage of patients and controls who meet the performance criteria for all tests was low, too. Therefore it seemed reasonable to assume, that this narrow interpretation of the German law regulations is too strict. Law regulations also allow that a bad performance in one test can be counterbalanced by a good performance in another test. This takes into account that most of the mandatory tests assess more than one area of performance. The main methodological shortcoming of this approach is that then a case by-case decision by experts would be necessary. As in the studies mentioned before a sum score was defined in advance as primary outcome measure. This score integrates all main and secondary variables recommended by the test system manufacturer. Herewith this score is very close to law regulations. Performance of patients is also below that of controls with reference to this sum score. The way this sum score is calculated in the present study differs from the way this is done in the studies mentioned before. Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003 integrated only three tests (DT, COG, TAVT) into this score whereas in this study the results of DT, COG, TAVT, LVT and RT are integrated in order to fulfil the recommendations of the manufacturer of the computer based testing device and the recommendations of FeV. Moreover Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003 did not use all main and secondary variables as recommended by the manufacturer but used two variables of each test representing performance speed and quality of performance (Stachwitz, 2006).

Composition of the patient group Whereas Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003 included only patients treated with one specific opioid analgesic in one study, the sample in this study is very heterogeneous. Overall five opioid analgesics at different dose levels have been included. Due to the low sample size no results concerning the different opioids have been possible. So in the present study several

5 numbers estimated from figure. 6 numbers estimated from figure. 212 different opioid analgesics at different dose levels were included. Moreover all patients were treated with additional medications. But correlation analysis in the present study revealed that morphine equivalence dosage and test performance were not related.

Composition of the control group and type of study The performance of patients here was compared to the performance of a reference group stratified for age. Whereas the performance of patients was compared to the performance of matched controls in the published studies(Dagtekin et al., 2007; Gaertner et al., 2006; Sabatowski et al., 2003), both groups in this study differ in mean age (patients m = 54.00; sd = 8.28 / controls m = 43.10; sd = 10.68). Correlation analysis of the data revealed that performance decreased with age so age and performance were related which might account for the different results. In addition the studies which have been published on the same topic were designed as non-inferiority studies. In those studies performance of patients was compared to historic performance data of healthy volunteers performing under influence of 0.5‰ alcohol. Instead of testing the controls under influence of alcohol virtual values that are equivalent to test performance under influence of 0.5‰ alcohol had been calculated. This calculation was based on the effect size from a vigilance test controls did under influence of 0.5‰ alcohol.

This study was designed to compare the amount of impairment to that caused by 0.5‰ alcohol in healthy volunteers. But it was not possible to show an alcohol induced performance decrement, although an impairment could have been expected according to a review on the effects of low doses of alcohol on skills related to driving (Moskowitz & Dary, 2000). But due to organizational restrictions it was not possible to counterbalance the sequence of sober and alcoholized testing so all controls first performed the test sober. Although the intended BAC of 0.5‰ was successfully reached in the alcohol condition and maintained stable during testing, no performance decrement could be found. According to their self-ratings controls felt impaired by the alcohol and rated their performance being worse than in the sober condition. So it seems that the alcohol effect was not strong enough to outweigh the training effect, although there was a delay of at least two weeks between both test sessions and there was a training phase before data was acquired right at the beginning of each test.

The results of this study demonstrated that patients suffering from chronic non-cancer pain, treated with stable doses of opioid analgesics, are impaired in driving related skills. But it could not be concluded if the impairment is comparable to the impairment caused by alcohol in healthy people. Therefore further research is needed and no general statement can be given about patients’fitness to drive under opioid treatment. It is necessary to make a by-case decision taking into account several aspects.

213 References

Akerstedt, T. & Gillberg, M. (1990). Subjective and objective sleepiness in the active individual 27. Int.J.Neurosci., 52, 29-37. BASt (2009). Begutachtungsleitlinien zur Kraftfahrereignung (Berichte der Bundesanstalt für Straßenwesen (BASt), Mensch und Sicherheit, Heft M 115). Bremerhaven: Wirtschaftsverlag NW. Biehl, B. (2008). LVT Linienverfolgungstest. Mödling: Schuhfried. Dagtekin, O., Gerbershagen, H. J., Wagner, W., Petzke, F., Radbruch, L., & Sabatowski, R. (2007). Assessing cognitive and psychomotor performance under long-term treatment with transdermal buprenorphine in chronic noncancer pain patients. Anesth.Analg., 105, 1442-1448. Dellemijn, P. L. I., van Duijn, H., & Vanneste, J. A. L. (1998). Prolonged treatment with transdermal fentanyl in neuropathic pain. Journal of Pain and Symptom Management, 16, 220-229. Dick, B. D. & Rashiq, S. (2007). Disruption of attention and working memory traces in individuals with chronic pain. Anesth.Analg., 104, 1223-9, tables. Drummer, O. H., Gerostamoulos, J., Batziris, H., Chu, M., Caplehorn, J., Robertson, M. D. et al. (2004). The involvement of drugs in drivers of motor vehicles killed in Australian road traffic crashes. Accid.Anal.Prev., 36, 239-248. Eccleston, C., Crombez, G., Aldrich, S., & Stannard, C. (1997). Attention and somatic awareness in chronic pain. Pain, 72, 209-215. Fishbain, D. A., Cole, B., Cutler, R. B., Lewis, J., Rosomoff, H. L., & Fosomoff, R. S. (2003). Is pain fatiguing? A structured evidence-based review. Pain Med., 4, 51-62. Fishbain, D. A., Cutler, R. B., Rosomoff, H. L., & Rosomoff, R. S. (2002). Can patients taking opioids drive safely? A structured evidence-based review. J.Pain Palliat.Care Pharmacother., 16, 9-28. Fishbain, D. A., Cutler, R. B., Rosomoff, H. L., & Rosomoff, R. S. (2003). Are opioid- dependent/tolerant patients impaired in driving-related skills? A structured evidence-based review. J.Pain Symptom Manage., 25, 559-577. Gaertner, J., Elsner, F., Radbruch, L., Kolibay, F., Theisohn, M., Berghaus, G. et al. (2008). Influence of changes to daily dose of opioids on aspects of cognitive and psychomotor performance involved in driving. Schmerz., 22, 433-441. Gaertner, J., Radbruch, L., Giesecke, T., Gerbershagen, H., Petzke, F., Ostgathe, C. et al. (2006). Assessing cognition and psychomotor function under long-term treatment with controlled release oxycodone in non-cancer pain patients. Acta Anaesthesiol.Scand., 50, 664-672. Grisart, J. (2009). [Cognitive effects of opioids]. Rev.Med.Suisse, 5, 1356-3. Grisart, J. M. & Plaghki, L. H. (1999). Impaired selective attention in chronic pain patients. Eur.J.Pain, 3, 325-333. Hergovich, A., Arendasy, M., Sommer, M., & Bognar, B. (2007). WRBTV Wiener Risikobereitschaftstest Verkehr. Mödling: Schuhfried. Jamison, R. N., Schein, J. R., Vallow, S., Ascher, S., Vorsanger, G. J., & Katz, N. P. (2003). Neuropsychological effects of long-term opioid use in chronic pain patients. J.Pain Symptom.Manage., 26, 913-921. Kalso, E., Edwards, J. E., Moore, R. A., & McQuay, H. J. (2004). Opioids in chronic non-cancer pain: systematic review of efficacy and safety. Pain, 112, 372-380. Kress, H. G. & Kraft, B. (2005). Opioid medication and driving ability. Eur.J.Pain, 9, 141-144. 214 Lagarde, E., Chastang, J. F., Lafont, S., Coeuret-Pellicer, M., & Chiron, M. (2005). Pain and pain treatment were associated with traffic accident involvement in a cohort of middle-aged workers. Journal of Clinical Epidemiology, 58, 524-531. Lenné, M., Dietze, P., Rumbold, G., Redman, J. R., & Triggs, T. J. (2000). Opioid dependence and driving ability: a review in the context of proposed legislative change in Victoria 8. Drug and Alcohol Review, 19, 427-439. Mailis-Gagnon, A., Lakha, S. F., Ou, T., Louffat, A., Yegneswaran, B., Umana, M. et al. (2011). Chronic noncancer pain: Characteristics of patients prescribed opioids by community physicians and referred to a tertiary pain clinic. Can.Fam.Physician, 57, e97-e105. McCracken, L. M. & Iverson, G. L. (2001). Predicting Complaints of Impaired Cognitive Functioning in Patients with Chronic Pain. Journal of Pain and Symptom Management, 21, 392-396. McQuay, H. (1999). Opioids in pain management. Lancet, 353, 2229-2232. Mersky, H. & Bogduk, N. (1994). Classification of chronic pain: Descriptions of chronic pain syndromes and definitions of pain terms. (2 ed.) Seattle: IASP Press. Moore, R. A. & McQuay, H. J. (2005). Prevalence of opioid adverse events in chronic non-malignant pain: systematic review of randomised trials of oral opioids. Arthritis Res.Ther., 7, 1046-1051. Moskowitz, H. & Dary, F. (2000). A Review of the literature on the effects of low doses of alcohol on driving-related skills. Washington: U.S. Department of Transportation, National Highway Traffic Safety Administration. Movig, K. L., Mathijssen, M. P., Nagel, P. H., van Egmond, T., DE GIER, J. J., Leufkens, H. G. et al. (2004). Psychoactive substance use and the risk of motor vehicle accidents. Accid.Anal.Prev., 36, 631-636. Neuwirth, W. & Benesch, M. (2007). DT Determinationstest. Mödling: Schuhfried. Nickel, R. & Raspe, H. H. (2001). Chronischer Schmerz: Epidemiologie und Inanspruchnahme. Nervenarzt, 72, 897-906. Portenoy, R. K. (2000). Current pharmacotherapy of chronic pain. J.Pain Symptom.Manage., 19, 16- 20. Prieler, J. (2008). RT Reaktionstest. Mödling: Schuhfried. Puhr, U. (2008a). 2HAND Zweihand Koordination. Mödling: Schuhfried. Puhr, U. (2008b). VIGIL Vigilanz. Mödling: Schuhfried. Radbruch, L., Loick, G., Kiencke, P., Lindena, G., Sabatowski, R., Grond, S. et al. (1999). Validation of the German version of the Brief Pain Inventory. Journal of Pain and Symptom Management, 18, 180- 187. Sabatowski, R., Kaiser, U., & Gossrau, G. (2010). Chronische Schmerzen: Fahrsicherheitsrelevante Aspekte beim Einsatz von Opioiden [Opioids in the management of chronic pain and driving ability]. Anasthesiol.Intensivmed.Notfallmed.Schmerzther., 45, 384-385. Sabatowski, R., Schwalen, S., Rettig, K., Herberg, K. W., Kasper, S. M., & Radbruch, L. (2003). Driving ability under long-term treatment with transdermal fentanyl. J Pain Symptom Manage., 25, 38- 47. Schmidt, H., Lowenstein, O., Hesselbarth, S., Klinge, V., & Michels, H. (2010). [S3 guidelines LONTS (Long-term administration of opioids for non-tumor pain)]. Schmerz., 24, 71-72.

215 Schubert, W., Schneider, W., Eisenmenger, W., & Stephan, E. (2005). Begutachtungsleitlinien zur Kraftfahrereignung. Kommentar. (2 nd. ed.) Bonn: Kirschbaum. Schuhfried GmbH (2009). Verkehrspsychologie: Testbatterie nach FeV Anlage 5 Nr. 2 (Bedienungsanleitung). Mödling: Schuhfried. Sjogren, P., Olsen, A. K., Thomsen, A. B., & Dalberg, J. (2000). Neuropsychological performance in cancer patients: the role of oral opioids, pain and performance status. Pain, 86, 237-245. Sommer, M. & Neurwirth, W. (2007). TAVTMB Tachistoskopischer Verkehrsauffassungstest. Mödling: Schuhfried. Stachwitz, P. (2006). Fahrsicherheit unter analgetischer Medikation mit retardiertem Oxycodon bei Patienten mit chronischen nicht-tumorbedingten Schmerzen. Medizinsche Fakultät der Universität Köln. Strumpf, M., Willweber-Strumpf, A., Herberg, K. W., & Zenz, M. (2005). Sicherheitsrelevante Leistungsfähigkeit von Schmerzpatienten unter Langzeitopioidtherapie. Der Schmerz, 19, 426-433. Tassain, V., Attal, N., Fletcher, D., Brasseur, L., Degieux, P., Chauvin, M. et al. (2003). Long term effects of oral sustained release morphine on neuropsychological performance in patients with chronic non-cancer pain. Pain, 104, 389-400. Vainio, A., Ollila, J., Matikainen, E., Rosenberg, P., & Kalso, E. (1995). Driving ability in cancer patients receiving long-term morphine analgesia. The Lancet, 346, 667-670. Veldhuijzen, D. S., Karscha, A. M., & van Wijck, A. J. M. (2009). Drugs, driving and traffic safety in acute and chronic pain. In J.C.Verster, S. R. Pandi-Perumal, J. G. Ramaekers, & J. J. DE GIER (Eds.), Drugs, driving and traffic safety (pp. 355-370). Basel: Birkhäuser. Veldhuijzen, D. S., van Wijck, A. J., Wille, F., Verster, J. C., Kenemans, J. L., Kalkman, C. J. et al. (2006). Effect of chronic nonmalignant pain on highway driving performance. Pain, 122, 28-35. Wagner, M. & Karner, T. (2008). COG Cognitrone. Mödling: Schuhfried. Watson, P. E., Watson, I. D., & Batt, R. D. (1981). Prediction of blood alcohol concentrations in human subjects. Updating the Widmark Equation. J.Stud.Alcohol, 42, 547-556. Zacny, J. P. (1995). A review of the effects of opioids on psychomotor and cognitive functioning in humans. Experimental and Clinical Psychopharmacology, 3, 432-466. Zijlstra, F. R. H. (1993). Efficiency in work behavior. A design approach for modern tools. PhD thesis, Delft University of Technology, Delft, The Netherlands.

216 Chapter 10: Risperidone effects on real driving performance compared with the effects of alcohol

P.A. Sardi*, G.M. Sardi*, C. Signoretti*, G. Skopp†, R.P.J. Freeman‡

* SIPSiVi Via Matteotti, 2 12100 Cuneo Italy

† Institute of Legal Medicine University Hospital Heidelberg, Germany

‡ Institute of Education University of London 20 Bedford Way, London, United Kingdom

217 Abstract

People with psychotic disorders can be treated using antispsychotic medication. Unfortunately, such medication has side effects that can include impairment in driving performance. The present study was designed to assess the effects of one particular antipsychotic medication – Risperidone – on performance in road tracking and reaction to a sudden event on a closed-circuit test track. Fifteen patients taking Risperidone (3-4 mg/die) and fifteen matched volunteers participated, with the volunteers driving under the double-blind administration of the legal Blood Alcohol Concentration (BAC) of 0.5 g/l and a placebo. The mean lateral position relative to the white-line at the edge of the road was significantly worse for those drivers under the influence of alcohol, but not for Risperidone. However, drivers under the influence of Risperidone showed greater Standard Deviation of Lateral Position when monitoring their speed and on straight sections of the track compared to drivers receiving the placebo or the legal level of alcohol. In the sudden event task, both drivers under the influence of a legal level of alcohol and Risperidone were significantly slower in reacting compared to the placebo, but there was no difference between these two conditions. It is concluded that the effects of alcohol and risperidone in general did not effect most driving measures. However, those measures that were sensitive to alcohol and risperidone indicated that the effects of risperidone were comparable or worse than effects assiociated with a BAC of 0.5 mg/ml.

Key words Driving under the influence of drugs, DUID, antipsychotics, risperidone

218 Introduction There is a wide consensus on considering all antipsychotic medicines to cause a serious impairment of driving capacity. A notable exception is Lithium, which is considered to have the least impact on driving performance, but is unfortunately unable to reduce all kinds of psychiatric symptoms. As confirmed by the DRUID deliverable 4.1.1 “Review of existing classification efforts”, issued by the DRUID WP4 on the 7th of February 2008, there is not yet an internationally accepted classification of these driving impairments and the existing classifications are largely based on expert opinions rather than on experimental evidence. In addition to the prescription of antipsychotic medicines, modern antipsychotic therapies include a focus on social integration as a basic approach. A particular example is employment, which often requires driving a car, especially in countries where the public transport network is not well developed. For this reason, psychiatric teams trying to socially integrate their patients through working, and hence driving, are particularly interested in prescribing antipsychotic medicines that cause the least impairment in driving performance. Attempting to study the driving performance of psychotic patients with experimental methods produces unique challenges. For example, the more emotional aspects of this disease, that are very important in determining the degree of impairment and consequential risks, are not sufficiently activated by the aseptic environment of the laboratory. As a result, the use of a driving simulator has poor validity. However, studying driving in real traffic situations raises serious ethical issues and legal responsibilities on the psychiatric teams. An intermediate approach can be found by using a closed circuit where similar tasks to real driving can be studied while protecting all the participants. Such an approach facilitates the study of the main difficulties of a real road environment, but under controlled conditions in order to allow mistakes that will not result in road accidents. This is also an environment where insurance companies will accept, explicitly and without reservation, to provide cover for the risks incurred by these patients when driving. Such a study proved possible with the participation of the Administration of the Health Unit Rome 7 and its Psychiatric Department (who are particularly engaged in patients’socialization through the choice of suitable jobs). Second, the Automobile Club of Italy (a public body) was interested in studying road risks, and owns a closed circuit in Rome Vallelunga, which is located in the catchment area of the above mentioned psychiatric department. Third, the Engineering Faculty of Parma, and its spin-off Vislab were also available to participate. The Vislab system enables the accurate measurement of various factors in driving: manoeuvres (including braking, steering, speed etc), the physical relationship between the car and its environment (including distances to obstacles, road layout, white lanes etc.). All these data can be recorded every tenth of a second, stored in an in-car computer and made available for later analysis. Fourth, suitable psychotic patients were available and willing to participate in such a study. Such patients face two risks - to lose their driving license and to lose their insurance cover – and these concerns usually prevent them from participating. For this reason, previously the effects of antipsychotic medicines on driving have been seldom tested, and predominantly on healthy volunteers. An example is the IMMORTAL project, where antidepressant medicines were tested on persons mourning as a reaction to a real loss, not under a psychotic depression. But the European Commission insisted on the need to have a scientific assessment on the degree of impairment caused by essential medicines, in order to overcome a situation where the pharmaceutical industries are able to declare “possible”impairing effects on driving capacity so as not 219 to be prosecuted for indirectly causing road accidents, legal bodies legislate generically against driving under the influence of these medicines, but leave responsibility for any exceptions to psychiatrists and patients finding a doctor authorizing such exceptions, without a sound scientific basis to such decisions. The DRUID project was initially titled “ENABLE”to highlight in the same word the aims of formal authorization and the effective ability to drive. The pharmaceutical industries have already tried to address these problems by proposing a second generation of antipsychotics. Soyka, et al. (2005) notes that there “is some but limited evidence that patients under novel atypical neuroleptics show less impairment compared to conventional neuroleptics. More clinical and experimental studies are warranted”. Earlier, a review by Meltzer and Gurk (1999) noted “clear improvements against the cognitive dysfunction in schizophrenia, especially on verbal and work memory, attention, executive function, motor coordination”in Risperidone, one of these new antipsychotics, which performs better than the other new generation antipsychotics. Nevertheless, there is not yet consensus about the impairment produced by Risperidone on driving ability. In the above mentioned “Review of existing classification efforts”, it appears that Risperidone (N05AX08) is not classified under Wolschrijn, et al. (1991), classified II (Likely to produce minor or moderate adverse affects) by ICADTS (International Council on Alcohol, Drugs and Traffic Safety) and classified also II (nearly the same meaning as ICADTS) in Belgium, Spain I, Spain II, Portugal and France I. However, France II classifies it as III (Attention, danger: do not drive). This final classification, introduced by a decree published in the official bulletin on the 2nd of August 2005, is legally binding and expressed by a label on its boxes. Greece II classifies it “+”, that means “affecting driving ability”, and Norway “w”, which means “drug with warning in Norwegian Catalogue”. Other classifications ignore it. In practice, only France II prohibits to drive under Risperidone. The producers of Risperidone state that it “May affect your driving ability; therefore, do not drive or operate machinery before talking to your healthcare professional”. Therefore, the psychiatric team of Rome 7 asked their patients to participate in the experiment while being treated with various types of antipsychotic medicines, not just those treated with Risperidone. The fact that the availability of patients treated with Risperidone was sufficiently high that 15 took part in our experiment, might be evidence for an increase in their cognitive capacity, allowing them to understand that they could trust that their results would be confidential and that there was no risk of losing their licence and hence their employment as a result. The psychiatric team enabled the participation of two psychiatrists and four nurses during all the experimental sessions.

Method

Subjects Fourteen males and one female composed the experimental group (N = 15). Their average age was 39 years. The patients were recruited through the psychiatric team of the Health Department of Rome-F whose catchment area includes the closed circuit of ACI-Vallelunga. The inclusion criteria were: patients treated with the same antipsychotic medicine (Risperidone/Paliperidone), alone or with mood stabilizing medicines; regular, fixed doses for at least three months (3-4 mg/die); their diagnoses include schizophrenia, paranoia, bipolar psychosis. Other inclusion criteria were the possession of a normal driving license, driving a car at least once a week, 220 absence of other severe health problems, and unimpaired liver function enabling normal metabolization of their medicine. The regular consumption and metabolization was also tested by the assessment of the metabolite in their samples, sent to the Heidelberg laboratory who analyzed them. A group of these patients also agreed to undergo the Vienna driving test. The psychiatric team provided detailed diagnosis and symptoms for each subject. The exclusion criteria were: acute cases; any other physical and/or psychological impairment; other medicines affecting driving; other substance consumption (illicit drugs, alcohol abuse); clinical, legal or social contraindications. The experimental group received a form containing a description of the experiment trials and of the overall aims of the DRUID project. Then they were asked to sign a consent form to agree to participate in the experiment. For reasons of privacy, the experimental subjects were allocated codes and in no cases did the researchers know the names of the patients, that were coded and stored by the health personnel using their usual confidential procedure. This experimental study received the approval of the Ethical Committee of the Health Department “USL RM/F”. The ethical proposal was drafted in accordance with the DRUID ethical manual approved by the DRUID Consortium. At the end of each experimental session the subjects received reimbursement in petrol vouchers. A matched sample of 15 healthy volunteers were recruited as reference group and to enable a comparison with the effects of alcohol. The average age of the reference group was 38 years. The reference group was matched with the experimental group based on: gender, age, weight and overall distance driven per year. The healthy volunteer subjects were tested for illicit drug consumption (six substances) before each driving session. The subjects were recruited by leaflets advertising the study circulated in the ACI Vallelunga Circuit, the Health Department Rome-F and by emails to SIPSiVi networks. The inclusion criteria were that they matched the characteristics of the experimental subjects.

Study design The study was conducted using a double-blind, matched design. The experimental group was taking a regular dose of Risperidone (inclusion criteria: 3-4 mg/die). The subjects in the reference group were given 98% pure alcohol mixed with 200ml of bitter orange juice according with the weight/concentration formula in order to reach a Blood Alcohol Concentration (BAC) of 0.5 g/l. Their placebo consisted of bitter orange juice with the order of presentation of the placebo and alcohol balanced over the subjects in the reference group.

Experimental group (Risperidone) procedure The experimental group (Risperidone) subjects were taken to the Vallelunga circuit by the health personnel in groups with a maximum of five subjects per session (08:00 meeting time). Subjects received a briefing and a description of the project and then signed a consent form. The health personnel team was two psychiatrists and four nurses. The nurses collected the blood samples (serum, whole blood and blood dry spot) before the driving sessions. The blood sample were coded by the health department for privacy reasons, centrifuged and transported in a cooled container box (- 20C) to University Hospital, Heidelberg in Germany for analyses (for details see pharmacokinetic assessment in Appendix 1). Each subject was instructed on the driving task and practised to enable

221 learning saturation before each trial. All sessions were carried out in the morning and ended before 14:00 with each trial lasting about 45 minutes.

Reference group (alcohol calibration) procedure The healthy volunteers for alcohol calibration were asked to arrive at Vallelunga Circuit by 08:00 with a maximum of five subjects per session. Subjects were tested for drug consumption (urine) and only subjects who tested negative were enrolled in the study. Subjects were asked to be present at the circuit for two sessions on two different dates at least one week apart. In each session subjects were given 200 ml of bitter orange juice 20-30 minutes before the driving session. In one session it was a placebo (only orange juice) and in the other session the orange juice was combined with 98% pure alcohol according with weight of the subject so they would reach 0.5 BAC. Half of the subjects received the alcohol in the first session and the other half received the alcohol in the second session - with the other half receiving the placebo. All subjects were monitored with a breathalyzer before and after the driving session. All the subjects that consumed alcohol beverage were required to have 0.5 BAC at the beginning and at the end of the session for their data to be included in the analysis. Only in one case was a subject below 0.5 BAC at the end of the driving session; their data for that session were excluded and the session repeated on another date. Each subject was instructed on the driving task and practised to enable learning saturation before each trial. All sessions were carried out in the morning and ended before 14:00 with each trial lasting about 45 minutes.

Driving tests Road Tracking Louwerens, Gloerich, de Vries, Brookhuis, & O’Hanlon (1987) constructed the road tracking test with the following characteristics: having the subject driving at a constant speed of 95 km/h as straight as possible on the right lane of a highway (left-hand drive) for a one-hour test drive, with the car equipped in order to take measures of several parameters: the primary dependent measure of this test is the Standard Deviation from Lateral Position (SDLP), while speed and standard deviation of speed are recorded as secondary control measures. This test enabled calibration of the effects of any sedative drug in relation to the BAC required to achieve the equivalent level of driving impairment. The alcohol calibration curve demonstrates that the intoxicated drivers’mean SDLP rises exponentially with BAC. Results from the alcohol calibration study can be used for describing drug effects on SDLP in terms of respective BAC equivalencies. The change in SDLP at BAC of 0.5 mg/ml has been used as a criterion level to quantify drug effects. The SDLP changes of the alcohol calibration study have been used as reference intervals (0.0 and 0.5 BAC values) to assess the equivalent effect of Risperidone in the driving tasks of experimental subjects. In the present study a variation of the Road Tracking test was used, taking into account the different scenario used. We considered driving performance separately on sharp curves (that are not usually so sharp on a highway) and on straight sections. Due to the sharp curves on the circuit, the fixed speed we used was only 30 km/h and there were two conditions: with and without a speedometer. In the absence of real traffic, it was possible to require the driver to keep a fixed distance of 30 cm from the white line on the right hand side. This distance was chosen to emulate the

222 distance to be kept from the line by the driver in order not to hit pedestrians or bicycles sheltered by that white lane, when restricted into a narrowed space. The car used was equipped with a Driver Performance Monitor (DPM) a specific device designed for data collection, developed by the VisLab research group (www.vislab.it). This system includes a colour camera installed behind the vehicle's front windscreen, an embedded computer placed in the car trunk and a touch screen monitor on the passenger's side which allows the selection of different tests and the evaluation of the current test. The DPM permits data acquisition with no scenario restrictions: a feature increasing the validity of the performance analysis. Moreover, powered by advanced algorithms, it is able to manage five different tests in order to extract a complete profile of the driver's performance. In this study the main data that have been collected and used are the vehicle's position relative to the white line, the vehicle’s speed and the driver's reaction time.

Sudden event The driver reaches the speed of 40 km/h, and passes besides a frame from which a “bobby car”suddenly appears in a randomised set of trials. The bobby car is released when the front of the driver’s car crosses the path of a photo-cell, but the Vislab tool starts measuring the reaction time only when the Bobby car becomes visible from the driver’s car (as happens in real driving) and stops when the driver starts braking. The tool also records the length of the track after braking and the movements of the steering wheel (angle and speed of rotation).

Statistical analyses

All statistical analyses were conducted using SPSS 18.0. Statistical analyses consisted of two steps: 1) Assessment for overall treatment effects by means of superiority testing; 2) Equivalence or non-inferiority testing of Risperidone effects based on the difference of scores compared to the matched sample when under the influence of alcohol. In step 1, the data were entered into the General Linear Model (GLM) repeated measures ANOVA for the 15 matched participants (i.e. 15 controls and 15 Risperidone users) with the control group measured both without and with alcohol. If the sphericity assumption was violated or not applicable, the Greenhouse-Geisser correction was used.

Results Dropouts and missing data One of the Risperidone participants admitted to his psychiatrist after his testing session that he had discontinued his Risperidone for three days before his test session, because he was afraid of his driving licence being withdrawn or suspended due to his driving performance. Subsequently, blood analyses confirmed the absence of Risperidone in his blood for that session. Approximately six months later, the same participant took part in another test session while under treatment with Risperidone (and the presence of Risperidone was confirmed by blood analysis).

Driving tests For the four straight line measures (mean speed, standard deviation of the speed, mean lateral position and standard deviation of lateral position - SDLP), no differences were significant, but

223 SDLP approached significance, (F1.5, 20.4 = 3.60, p = .058), with a trend for greater variation in the distance from the line for Risperidone. For the speed monitoring task, there was a significant effect for lateral position (F2, 28 = 3.79, p = .035), with a significant difference between the alcohol and Risperidone conditions (t14 = 2.17, p = .048). Those drivers under the influence of alcohol drove closer to the line maintaining a mean difference of 22 cm compared with placebo 29 cm and Risperidone 32 cm as can be seen in Figure 1. The 95% CI (-0. 046 – 0.101) associated with the difference in lateral position between the risperidone and placebo did not include the mean difference induced by alcohol (i.e. -0.069). LP in the risperidone group thus was equivalent to placebo controls.

Figure 1. Mean Lateral Position from white line (m) for the three conditions of the speed monitoring task with the target distance of 30 cm marked in red.

The SDLP in the speed monitoring task also showed a significant effect, (F2, 28 = 3.65, p = .039) with a significant difference between the alcohol and Risperidone conditions (t14 = 2.65, p = .019) with greater variation in the distance from the line for Risperidone.

Considering only the straight sections, the mean lateral position approached significance (F1.43, 20.04 = 3.36, p = .069) with again those under the influence of alcohol driving closer (25 cm) to the white line than either in the Placebo (28 cm) or Risperidone (33 cm) conditions. SDLP was significant (F2, 28 = 6.23, p = .006) with a significant difference between the alcohol and Risperidone conditions (t14 = 3.04, p = .009). Once again those using Risperidone showed greater variation (.12) compared to Placebo (.10) and alcohol (.10). For the curved sections only, the standard deviation of speed was significant (F1.34, 18.78 = 8.29, p = .006) with a significant difference between the alcohol and Risperidone conditions (t14 = 2.83, p = .013). Those using Risperidone (1.9) showed greater variation in speed than either Placebo (1.2) or alcohol (1.2). In addition, SDLP approached significance (F1.28, 17.92 = 3.53, p = .068) with again a trend to greater variation for Risperidone (.25) compared to Placebo (.22) and alcohol (.20). All other measures were not significant.

224 Sudden event For the sudden event, the reaction time to begin braking was significant (F2, 28 = 3.52, p = .043), but there was not a significant difference between the alcohol and Risperidone conditions (t14 = .66, p = .518). Both alcohol (340 ms) and Risperidone (300 ms) mean reaction times were slower than the Placebo (210 ms) as can be seen in Figure 2. The 95% CI associated with the difference between risperidone and placebo (i.e. -0.00560 thru 0.18560) included the mean difference between alcohol and placebo (i.e. = 0.131). This implies that the increment in RT in the risperidone group was of clinical relevance and comparable/bigger than that produced by a BAC=0.5 mg/ml.

Figure 2. Reaction time (ms) to the sudden event for the three conditions.

Vienna test Psychologists based in Rome administered the Vienna test to eleven patients (three patients were not available to undergo that check of their driving ability, mainly because of their fear of losing their licence). Nine out of these eleven produced results indicating they were able to drive safely. One of the two who did not had already been prevented from driving by relatives who were worried because of frequent road accidents. The correlation between the Vienna test and the performances on the circuit was rather strong (0.67) if these two patients are included, but much weaker when the patient who performed worst on both tests was removed.

Discussion

The Results present a broadly consistent picture. The basic results for the Road Tracking task that required drivers to maintain a fixed distance of 30 cm from the right-hand white line showed that drivers under the influence of alcohol were closer to the line whereas drivers on risperidone and those not under the influence of alcohol were able to maintain the appropriate distance. In addition, drivers under risperidone drove with more ‘weaving’than drivers under the influence of alcohol or the placebo. 225 Both drivers under the influence of alcohol and Risperidone showed impairment in reaction times in the sudden event task compared to the placebo condition. However, the absence of a difference between alcohol and Risperidone suggests that the impairing effect of Risperidone is broadly equivalent to that of a legal level of alcohol (i.e. BAC 0.5). Taken together these results provide empirical data that enable us to provide a judgement of the effect of risperidone on driving performance. Drivers using risperidone drove with a lateral position that was comparable to placebo. However the standard deviation of lateral position and reaction time to sudden events were signifanctly increased and comparable or bigger than those observed after a blood alcohol concentration of 0.5 mg/ml. The present data thus seems to indicate that drivers under the influence of risperidone do demonstrate impairments that should be considered of clinical relevance.

Acknowledgments This study was conducted within the DRUID research consortium, WP 1.2, granted by the European Commission, DG-TREN (TREN-05-FP6TR-S07.61320-518404 – DRUID)

References

Louwerens J.W., Gloerich, A.B.M., de Vries, G., Brookhuis, K.A. and O’Hanlon, J.F. (1987). The relationship between drivers’blood alcohol concentration (BAC) and actual driving performance during high speed travel. In P.C. Noordzij and R. Roszbach (Eds.), Proceedings of the 10th International Conference on Alcohol, Drugs and Traffic Safety (pp. 183-192). Amsterdam: Excerpta Medica.

Meltzer, H.Y. and McGurk, S.R. (1999). The effect of clozapine, risperidone, and olanzapine on cognitive function in schizophrenia. Schizophrenia Bulletin, 25, 233 - 255.

Ramaekers, J. G. (2003). Antidepressants and driver impairment: empirical evidence from a standard on-the-road-test. Journal of Clinical Psychiatry, 64(1), 20-29.

Soyka, M., Kagerer, S., Brunnauer, A., Laux, G. and Moller, H.J. (2005). Driving ability in schizophrenic patients: effects of neuroleptics. International Journal of Psychiatry Clinical Practice, 9(3), 168-174.

Wolschrijn, H., De Gier, J. J. and De Smet, P. A. G. M. (1991). Drugs and Driving: A New Categorization System for Drugs Affecting Psychomotor Performance. Technical Report. Limburg, The Netherlands: Institute for Drugs, Safety and Behavior, University of Limburg.

226 Chapter 11: Blood to serum ratios of hypnotics, opioid and non-opioid analgesics as well as antipsychotics and amphetamine-like drugs and their analysis in dried blood spots

Ricarda Jantos, Gisela Skopp

Institute of Legal Medicine and Forensic Medicine, University Hospital, Voss-Str. 2, 69115 Heidelberg, Germany

Telephone ++49 6221 568920 Fax ++49 6221 561300 E-mail: [email protected]

227 Abstract

Blood is prevalently used for a toxicology investigation including DUID cases whereas analysis on pharmacokinetic samples is performed on separated serum or plasma, as a rule. Most drugs, however, do not equally distribute between blood and plasma. Dried blood spots (DBS) are increasingly used in drug analysis due to their ease of collection, shipping and storage as well as the reduced risk of infection. Therefore, the following objectives were pursued: a. determination of drug concentration in whole blood and corresponding plasma samples to estimate ex vivo blood to plasma (b/p) ratios, and b. comparison of drug levels in whole blood and corresponding DBS. Analytes were hydromorphone, morphine, fentanyl and norfentanyl, oxycodone and noroxycodone, alprazolam, zopiclone, temazepam, THC, 11-OH-THC and THC-COOH, MDMA and MDA, d- amphetamine as well as risperidone and 9-OH-risperidone. Analysis was performed on whole blood, plasma and DBS by LC/MS/MS except THC and metabolites, where GC-MS was used. All analytical methods were validated according to international guidelines for each respective matrix. B/p ratios were derived from corresponding blood and plasma samples. Bland Altman analysis was used to test agreement of concentrations determined from whole blood and corresponding DBS. The mean obtained b/p ratios were: hydromorphone: 1.04, morphine: 1.03, fentanyl: 0.87, norfentanyl: 1.19, oxycodone: 1.48, noroxycodone: 1.73, alprazolam: 0.81, zopiclone: 0.89, temazepam: 0.71, MDMA: 1.19, MDA: 1.72, d-amphetamine: 0.89, risperidone: 0.65, 9-OH-risperidone: 0.73. Ratios were close to reliably established ratios published in the current literature as far as available, except for MDA. For all analytes except zopiclone, the mean ratio of the DBS and blood concentrations and their relative standard deviations indicated that DBS analysis is as reliable as analysis from whole blood. Bland-Altman difference plots for the several substances supported this thesis. Zopiclone is expected to undergo degradation even in DBS. Dividing the concentrations of the analytes blood by the obtained b/p ratios may give a reasonably good estimate of the coexisting concentration in plasma, except for MDA. For law enforcement purposes, it is recommended to consider inherent biological variations in the b/p relationship and to take the range of individual b/p ratios as a basis. The MDA b/p ratio is recommended to be analyzed in a separate study after administration of MDA itself. There is sound evidence that the DBS assay has potential as a precise and inexpensive option for the determination of the investigated analytes in small blood samples. However, stability of zopiclone in DBS has to be tested and compared to the stability in whole blood specimens using a stability indicating method.

228 Introduction

Distribution of drugs between whole blood and serum or plasma Drug analysis in forensic and postmortem toxicology including DUID (driving under the influence of drugs) cases is usually performed on whole blood whereas serum or plasma is preferably used in clinical facilities and pharmacological studies. Blood is a complex biological fluid consisting of a buffered clear fluid containing proteins, fats, solids and suspended cells. The major constituents – the red cells – can be separated from the clear fluid by centrifugation. If blood is allowed to stand without the addition of an anti coagulating agent, then red cells will clot and the resultant fluid can be decanted. If anticoagulants are added, plasma can subsequently be prepared. Serum is in most respects similar to plasma except that it does not contain soluble factors that lead to blood clotting (1). The proportion of blood volume that is occupied by red cells is referred to as the hematocrit value (%). Normally, it averages 48% for men and 38% for women, and ranges from 35–54% in blood from healthy adults (2). Most drugs are not equally distributed between the sub compartments of blood; hence, the concentration in serum or plasma may differ from that in whole blood. Blood to plasma (b/p) concentration ratios may not only vary between different compounds with the same core structure, but also between the parent drug and corresponding metabolites (3) or depend on the hematocrit value (4, 5). To know the distribution of drugs into the major sub compartments of blood is mandatory in order to reliably compare whole blood to plasma or serum levels that have been derived from controlled pharmacokinetic studies.

Generally, distribution of compounds between whole blood and plasma is determined using in vitro or, to a much lesser extent, ex vivo procedures (5-7). In the conventional in vitro method, the drug is incubated with a whole blood specimen at a known hematocrit value – which does not apply to DUID samples. Following equilibration, an aliquot of the whole blood specimen is put on the side while plasma is prepared from the remaining sample. It is not known how far off from the true values the in vitro determined ones are likely to be. In the ex vivo method, the blood sample taken from a drug user for analysis is portioned, and plasma is prepared from an aliquot of the original specimen. In both methods, drug concentrations in whole blood and plasma will be measured separately using separate standards for whole blood and plasma, respectively.

This investigation compares b/p distribution ratios of opioid and non-opioid analgesics, hypnotics and antipsychotics as well as commonly used illegal drugs using authentic samples from healthy volunteers conducting DUID experiments. Partners were instructed to collect 5-10 mL blood from the median cubital vein or the anterior forearm by venipuncture into a vacuum tube or into a syringe and needle with subsequent transfer to a VacutainerTM or a MonovetteTM containing potassium oxalate/sodium fluoride as an anticoagulant/preservative (e.g. grey top VacutainerTM, DIN ISO 6710). Plasma should be prepared by centrifuging the sample at 2000-3000 g for 10-15 minutes at 20-22°C or at 4°C (labile compounds). Samples should be stored frozen and shipped on dry ice.

229 Dried blood spots

Dried blood spots (DBS) have routinely been used in neonatal metabolic screening for over two decades, and have recently established themselves as a valuable tool in therapeutic drug monitoring (8-13). Despite a limited sample size of 10-100 µL blood, analysis of DBS specimens has become feasible with the advent of increasingly sensitive MS technologies (14). DBS can be stored at room temperature and shipped by regular mail, in contrast to whole blood or plasma specimens. Use of DBS is an appropriate method to reduce virus infection risk to a minimum which is a major concern handling samples of drug users (15, 16). Being readily accessible also in subjects with limited venous access, such as e.g. injecting drug users, it represents a valuable and less invasive alternative to taking of a blood sample. The simple sampling can also be performed by non-medical personnel. In addition, the use of DBS makes labile compounds such as ester type drugs less susceptible to degradation (15). A blood spot card was designed for collection of DBS in the present investigation. The face and the back of the card are shown in Figure 1. The 903 specimen collection paper (GE Healthcare, Dassel, Germany) used for the custom made card is an FDA listed class II medical device. It is manufactured from 100% pure cotton linters with no wet-strength additives. Both, the manufacturing and post-printing quality of the paper were checked.

Fig. 1: Blood spot card, face and back of the card

230 Methods

Figure 2 gives an overview on the study design:

Whole blood

Plasma Blood spot

Fig. 2: Plasma and blood spot specimens are obtained from the corresponding whole blood specimen for measurement of blood/plasma ratios and to evaluate measurement from DBS

Measurement of drug concentration in whole blood and corresponding plasma samples enables an estimation of ex vivo b/p ratios. The knowledge of these ratios and of their range is mandatory to valuably compare drug concentrations measured either in whole blood or plasma. A comparison of drug levels in whole blood and corresponding DBS allows demonstrating whether drug measurement from DBS is as accurate as that from the whole blood specimen. DBS analysis will make handling of toxicological samples much easier.

Extraction and determination of the analytes in plasma, whole blood and DBS

Extraction and determination: Plasma, blood and DBS were extracted to determine the concentrations of the following analytes: hydromorphone, morphine, fentanyl, norfentanyl, oxycodone, noroxycodone, alprazolam, temazepam, zopiclone, tetrahydrocannabinol (THC), 11-hydroxy-tetrahydrocannabinol (11-OH-THC), tetrahydrocannabinol carboxylic acid (TCH-COOH), 3,4-methylenedioxy-methamphetamine (MDMA), 3,4-methylenedioxyamphetamine, d-amphetamine, risperidone and 9-hydroxy-risperidone (9-OH-risperidone). DBS were prepared by spotting a 100 µL aliquot of whole blood the custom made DRUID card (GE Healthcare, Dassel, Germany), which were dried at room temperature for at least 3 h. Then, each sample was packed in a plastic bag together with a desiccant pack and stored at ambient temperature (20-24°C) whereas whole blood and plasma samples were kept frozen (-20°C) until analyzed. Before extraction, DBS were cut out completely and transferred into plastic tubes. Analysis was performed by liquid chromatography/tandem mass spectrometry (LC/MS/MS) following liquid/liquid extraction except cannabinoids which were determined by GC-MS. Only hydromorphone was isolated by solid phase extraction. Each sample was extracted twice. Table1 gives an overview on the extraction procedures and mass spectrometry conditions developed for the different analytes.

231 For quantitation, spiked plasma, whole blood or DBS samples were prepared and analyzed in the same way, except for the dronabinol samples. For the determination of THC, 11-OH-THC and THC-COOH 1000 µL plasma were extracted. Calibration lines were constructed with linear least squares regression using the ratio of the target analyte peak area to the corresponding internal standard peak area.

Validation: Imprecision, extraction efficiency and bench top stability (24 h) were investigated according to the FDA Guidance for Industry (17). Carryover was checked as described by Bansal and DeStefano (18). Ion suppression or enhancement was determined according to Matuszewski et al. (19). The lower limit of detection (LLOD) and quantitation (LLOQ) was estimated from the calibration curves according to DIN 32465 at a probability of 95% (20). Data analysis was done using Microsoft Excel®. Agreement of the whole blood and DBS methods was further assessed by Bland-Altman plots (21). Therefore, mean difference between the blood and DBS concentrations was calculated. 95% of the values are required to lie between the limits of agreement, which were calculated as the mean difference ±1.96xSD (standard deviation).

Materials and instrumentation: Zopiclone was purchased from Rhône Poulenc Rorer (Cologne, Germany). Risperidone, 9-OH-risperidone and didehydromethylrisperidone were supplied by Janssen- Cilag (Neuss, Germany). Hydromorphone, morphine, fentanyl, norfentanyl, oxycodone, noroxycodone, alprazolam, temazepam, THC, 11-OH-THC, THC-COOH, MDMA, MDA, and amphetamine as well as their deuterated standards and lorazepam-d4 were obtained from LGC. Wesel, Germany. High-pressure liquid chromatography (HPLC)-grade acetonitrile and methanol as well as ethyl acetate (•99.5 %), toluene (•99.5 %), isopropanol (•99.5 %), solid NaOH (•99 %), ammonium acetate (•98 %), acetic acid (100 %) were from Roth (Karlsruhe, Germany). Isoamylalcohol (•99 %), dichloromethane (•99.8 %), ammonium hydroxide (25 %), hydrochloric acid (25 %), sodium carbonate (•99.5 %), sodium hydrogen carbonate (•99.5 %), potassium chloride •99.5 %) und boric acid (•99.8 %) were supplied by Merck (Darmstadt, Germany). Double distilled water was obtained from Braun (Melsungen, Germany). Drug-free whole blood and plasma for preparation of calibration lines and validation experiments were purchased from the local blood bank of the University Hospital of Heidelberg. For solid phase extraction of hydromorphone, Bond-Elut C8 1 mL columns were used (Varian, Darmstadt, Germany). LC-MS/MS analysis was performed on an API 4000 tandem MS with a TurboIon ionization source operated in the positive-ion mode (AB Sciex, Darmstadt, Germany). It was interfaced to an HPLC pump equipped with an autosampler 1100 series, Agilent, Waldbronn, Germany). GC-MS determination was carried out using a 5973N mass selective detector interfaced with a 6890N GC system equipped with a 7683 series injector (Agilent, Waldbronn, Germany).

232 Table 1: Analysis of hydromorphone, morphine, fentanyl, norfentanyl, oxycodone, noroxycodone, alprazolam, zopiclone, temazepam, MDMA, MDA, d-amphetamine, risperidone and 9-OH-risperidone: extraction, chromatography and mass spectrometry conditions analyte adjustment of internal standard extracting agent [vol%] mobile flow retention column transition used transition the pH-value for (IS) phase [µL/min] time for quantitation IS extrac-tion A:B:CI [min] [v:v:v] hydromorphone1 carbonate hydromorphone-d3 dichloromethane/ 50:10:40 300 1.18 # 286à185 289à185 buffer pH 9.0 isopropanol/ conc. NH3 80:20:2 morphine2 borate buffer morphine-d3 ethyl acetate 50:10:40 220 2.00 Ƈ 286à152 289à152 pH 8.5 fentanyl 5 %NH3 fentanyl-d5 ethyl acetate 40:12:48 250 1.69 # 337à188 342à188 norfentanyl norfentanyl-d5 1.36 233à84 238à84 oxycodone 5 %NH3 oxycodone-d6 ethyl acetate 50:10:40 300 1.20 # 316à298 322à304 noroxycodone norxycodone-d6 1.11 302à284 305à287 THC3 0.25 M acetic THC-d3 n-hexane/ethyl acetate helium 50 5.5-6.5 ǻ 386, 371,303 389 11-OH-THC acid 11-OH-THC-d3 9:1 mL/min 7.2-7.6 371, 474, 459 374 THC-COOH THC-COOH-d3 7.65-8.2 371, 473, 488 374 alprazolam borate buffer alprazolam-d5 toluene/isoamyl alcohol 45:11:44 250 3.70 # 309à205 314à210 pH 8.5 95:5 zopiclone borate buffer lorazepam-d4 toluene/isoamyl alcohol 40:12:48 300 1.15 # 389à245 325à307 pH 8.5 95:5 temazepam borate buffer temazepam-d5 toluene/isoamyl alcohol 40:12:48 300 3.02 # 301à255 306à260

233 pH 8.5 95:5 MDMA4 0.01 M NaOH MDMA-d5 ethyl acetate 60:8:32 220 2,15 Ƈ 194à163 199à165 MDA4^ MDA-d5 2,08 180à163 180à135 d-amphetamine amphetamine-d5 1.71 136à91 141à124 risperidone borate buffer didehydromethyl- ethyl acetate 50:10:40 250 2.00 # 411à191 421à201 9-OH- pH 8.5 risperidone 1.90 427à207 risperidone

I unless stated otherwise: A: 4 mM ammonium acetate buffer pH 3,2; B: at three characteristic fragments within their respective retention time windows. methanol; C: acetonitrile THC and its metabolites were separated by a temperature gradient. # Phenomenex Luna C18 2,0 mm x 150 mm, particle size 5 µm, Phenomenex, 4: the organic phase was acidified with 50 µL of methanol HCl (49:1, v:v) prior Aschaffenburg, Germany to evaporation to dryness Ƈ Agilent Zorbax Eclipse XDB-C8 2,1 mm x 150 mm. particle size 5 µm, Agilent, Waldbronn, Germany ǻ CP-Sil5 CB 0.2 mm x 12.5 m, film thickness 0.4 µm, Varian. Darmstadt, Germany

1 solid phase extraction 2 ultrasonication (5 min) was applied following addition of borate buffer and IS 3: analysis was performed using GC-MS in the selected ion monitoring mode. Identification of the analytes was performed in the single ion monitoring mode

234 Results and discussion

Evaluation

Table 2 gives an overview on the validation results determined in plasma. Additionally, matrix effect, extraction efficiency and 24 h bench top stability were checked; all values were in an acceptable range (data not shown). Carryover could not be observed for any analyte. There were no significant differences between the validation results in plasma, blood and DBS; all parameters were in the same range as presented in table 2 or better. It could be observed that matrix effects in DBS are of a lesser extent than in the other media.

Table 2: validation results for the most important substances determined in plasma analyte LLOD LLOQ between-run within-run precision [%] linearity [ng/mL] [ng/mL] precision [%] hydromorphone 0.9 3.1 4 ng/mL: 4 ng/mL: 2-20 ng/mL 5.4 3.5 r=0.9998 12 ng/mL: 4.1 12 ng/mL: 4.1 morphine 6.2 22.0 50 ng/mL: 8.3 50 ng/mL: 9.0 50- 250 ng/mL: 5.6 250 ng/mL: 5.6 500 ng/mL r=0.9999 fentanyl 0.04 0.16 0.25 ng/mL: 4.7 0.25 ng/mL: 4.7 0.1-10 ng/mL 6.5 ng/mL: 4.7 6.5 ng/mL: 2.9 r=1.0000 norfentanyl 0.01 0.03 0.25 ng/mL: 3.5 0.25 ng/mL: 2.9 0.1- 2.0 ng/mL: 3.7 2.0 ng/mL: 3.7 4.0 ng/mL r=1.0000 oxycodone 1.9 6.4 10 ng/mL: 6.8 10 ng/mL: 6.4 5-100 ng/mL 45 ng/mL: 4.2 45 ng/mL: 3.0 r=0.9997 noroxycodone 1.5 5.7 10 ng/mL: 3.7 10 ng/mL: 3.7 5-100 ng/mL 45 ng/mL: 2.2 45 ng/mL: 1.9 r=0.9989 zopiclone 1.9 6.7 20 ng/mL: 5.5 20 ng/mL: 0.4 10-50 ng/mL r=0.9999 temazepam 12.7 44.1 100 ng/mL: 4.9 100 ng/mL: 4.1 50- 500 ng/mL: 2.5 500 ng/mL: 2.0 500 ng/mL r=0.9997 alprazolam 0.8 3.0 10.0 ng/mL: 10.0 ng/mL: 5.1 2.5-50 ng/mL 5.7 % r=0.9999 MDMA 0.9 3.1 50 ng/mL: 3.3 50 ng/mL: 2.5 A: 50- 250 ng/mL: 4.1 250 ng/mL: 2.8 400 ng/mL r=0.9997 B: 5-

235 40 ng/mL r=0.9999 MDA 0.25 0.93 50 ng/mL: 2.4 50 ng/mL: 1.9 A: 5- 150 ng/mL: 6.4 150 ng/mL: 5.4 40 ng/mL r=0.9992 B: 0.5- 4 ng/mL r=0.9992 d-amphetamine 0.6 2.3 50 ng/mL: 5.1 50 ng/mL: 2.2 5-40 ng/mL r=0.9998 risperidone 0.7 2.5 6.7 ng/mL: 6.3 6.7 ng/mL: 2.9 5-40 ng/mL 19.7 ng/mL: 6.0 19.7 ng/mL: 6.0 r=0.9995 9-OH- 1.2 4.4 20.4 ng/mL: 6.2 20.4 ng/mL: 4.8 5-60 ng/mL risperidone 64.0 ng/mL: 10.4 64.0 ng/mL: 4.4 r=0.9999

Concentrations of the analytes in blood, plasma and DBS

Table 8 in the appendix provides complete information on the results of the analysis of all drugs. Drugs where the sample size with concentrations >LLOQ did not exceed n=7 were not considered in this part of the document, but results are also shown in table 8 in the appendix. For the dronabinol samples, analysis of the blood specimens is not yet completed. Due to a sample volume of 100 µL for DBS instead of 1000 µL for plasma and blood, the extraction method presented in table 1 turned out to be not applicable to DBS samples without modifications. Method development for DBS extraction is planned to be completed in May 2011, the results of the validation procedures and the determination of the assays of THC, 11-OH-THC and THC-COOH in the dronabinol samples might be expected in August 2011.

Blood/plasma ratios

Table 3 gives an overview on the b/p ratios and their relative standard deviations (RSD) determined in the investigations for the DRUID project and reference values (if available).

Table 3: b/p ratios obtained in the presented studies and published values; n.a.: not available, -: no reference analyte b/p ratio; range b/p ratio; mean reported values reference ± RSD [%] hydromorphone 0.91-1.22 1.04 ± 8.11% 1.35 (22) morphine 0.96-1.07 1.03 ± 3.59% 1.02 (23) fentanyl 0.62-1.02 0.87 ± 13.9% 1.0 (24) norfentanyl 1.03-1.29 1.19± 6.81% n.a. - oxycodone 1.29-1.76 1.48 ± 8.19% n.a. - noroxycodone 1.28-2.09 1.73 ± 13.5% n.a. -

236 alprazolam 0.73-0.90 0.81 ± 5.84% 0.80 (25) zopiclone 0.66-1.29 0.89 ± 16.1% 1.0 (26) temazepam 0.58-0.87 0.71± 12.0% 0.53 (27) MDMA 0.98-1.46 1.19 ± 8.03% 1.16 (28) MDA 1.00-3.02 1.72 ± 31.1% 1.27 (28) amphetamine 0.65-1.14 0.89 ± 10.94 % 0.91 (29) risperidone 0.56-0.73 0.65 ± 7.52% 0.67 (30) 9-OH-risperidone 0.61-0.91 0.73 ± 12.3% n.a. -

Distribution ratios are generally derived from in vitro partition experiments where plasma water, plasma proteins and red blood cells are pooled. Some caution is advisable using these data. When spiked blood is diluted with autologous plasma water, erythrocytes may discharge the compound over proportionally compared to plasma proteins. Concentration ratios between blood and plasma may vary from 0.5 to 2.0 such as e.g. for phenytoin and maprotiline, respectively. In table 3, b/p ratios reported in the bold typed references are not comparable to those obtained from the studies performed in the DRUID project. For hydromorphone, the b/p ratio obtained in our study differs significantly from the reference value. Parab et al. give neither an information on the determination of the b/p ratio nor provide a reference for the published factor (22). Also, the b/p ratio of fentanyl significantly differs from the published value. Bower and Hull (24) investigated the distribution of fentanyl between plasma and packed erythrocytes. The obtained ratio was later cited as a b/p ratio. Unlike packed erythrocytes, whole blood also contains plasma proteins, which has not been considered referring the erythrocyte/plasma ratio as a b/p ratio. Thus, the b/p ratio obtained in our study cannot be compared to the value reported by Bower and Hull. Bramness et al. published a b/p ratio of 1.0 for zopiclone in the context of a fatal zopiclone intoxication (26). It is not apparent if the ratio has been determined in post-mortem blood; also, no particular reference is provided. Whenever the b/p ratio has been determined from post-mortem blood it is impossible to compare this value to those obtained in the present study. It is well known that significant changes occur in post-mortem blood, due to autolysis and hemolysis. Especially hemolysis may cause differences of the b/p ratio compared to blood samples collected from living subjects. Hence, a comparison of the b/p ratios obtained in the DRUID study with the reference values is not considered to be reasonable. Osselton et al. investigated the b/p ratio of temazepam on blood samples of volunteers receiving drug therapy (27). The authors give no information on the number of samples investigated. In addition, all participants suffered from drug addiction which leads to hepatic or renal insufficiency affecting the plasma or erythrocyte fraction of the blood or the amount and quality of plasma proteins. The b/p ratios of morphine, alprazolam, MDMA and amphetamine and risperidone were in accordance with the reported values and showed acceptable RSD. For 9-OH-risperidone, the slightly higher b/p ratio in contrast to risperidone appears to be in line with its increased hydrophilic properties due to the hydroxyl group and its decreased plasma protein binding compared to risperidone. Despite of the smaller concentration range, the amphetamine b/p results are in line with those determined in the VTI study from Sweden. Table 4 shows the analytical data of both studies:

237 Table 4: comparison of TNO and VTI amphetamine studies carried out during the DRUID project Study of TNO, Study of VTI, The Netherlands Sweden positive findings in blood n=29 (100%) n=37 (62%) positive findings in plasma n=29 (100%) n=37 (62%) Concentration range, blood 10.75-40.65 0-123.5 concentration range, plasma 12.20-41.13 0-112.0 mean +/- SD, blood 20.61 +/- 7.08 21.78 +/- 29.60 mean +/- SD, plasma 23.21 +/- 7.60 22.92 +/- 29.40 b/p, range 0.64-1.14 0.65-1.10 b/p mean +/- RSD 0.89 +/- 11.24 % 0.91 +/- 13.49 %

MDA b/p ratio from samples of the study from RugPsy differed significantly from the value determined in a previous study performed in Maastricht. Results from both studies are summarized in table 5:

Table 5: comparison of the b/p ratios of MDA determined in two MDMA studies carried out during DRUID project Study of Maastricht Study of RugPsy, Groningen positive findings in blood n=15 (24%) n=29 (78%) positive findings in plasma n=27 (43%) n=30 (79%) concentration range, blood 5.0-10.4 1.1-24.4 concentration range, plasma 5.7-16.3 1.0-13.5 mean +/- SD, blood 7.5 +/- 1.4 9.4 +/- 6.4 mean +/- SD, plasma 8.1 +/- 2.4 5.1 +/- 3.3 b/p, range 1.01-1.77 1.00-3.02 b/p mean +/- RSD 1.27 +/- 15.8 % 1.72 +/- 31.1 %

More positive findings were obtained from samples collected during the RugPsy study, and the particular concentration range was considerably wider compared to samples obtained from the Maastricht study. The low concentrations of MDA probably affect b/p ratios leading to relatively high RSD. The low concentrations are likely to explain the differences which could not be observed for MDMA. In both studies, MDA was determined as a metabolite of MDMA, which can be detected in a concentration of 5-10 % of the MDMA concentration (31). MDA itself is used as an amphetamine-like drug and is expected to lead to higher blood levels than after administration of MDMA. The b/p ratio should be investigated after administration of MDA in dosages near to those of recreational use. In general it has to be noted that the obtained b/p ratios cannot be applied without restriction to over- therapeutic or toxic blood concentrations since all participants of the particular studies received therapeutic dosages or, in case of drug consumption, only moderate amounts.

238 Agreement of the determination in blood and DBS

With respect to the advantages mentioned in the introduction, DBS may be a suitable method for roadside blood sampling in case of suspicion of DUID. Before use of DBS as a reliable method can be realized, it must be shown that DBS analysis is able to provide results that are as reliable as those using whole blood samples. Therefore, blood and DBS results of the determination of the drugs investigated in the presented studies were compared using the DBS/blood ratio (DBS/b) and Bland- Altman analysis. Accordingly, the respective mean of the corresponding results determined with the two different methods were plotted on the x-axis and their difference on the y-axis. In contrast to a scatter plot of blood and corresponding DBS concentrations, the Bland-Altman difference plot shows the distribution of the differences over the complete concentration range. The mean of the differences indicates over- or underestimation by one of the two methods. Besides the mean of the differences, the 95 % limits of agreement were calculated as the mean difference ±1.96xSD. These values define the range within which most differences between measurements by the two methods will lie. In case of normal distribution, 95 % of the differences are expected to lie within these limits (21). Ideally, the mean DBS/b ratio should be equal to 1.00, which means that results from whole blood and DBS analysis do not differ. All analytes and their corresponding DBS/b ratios including respective RSD are summarized in table 6:

Table 6: summary of the DBS/b ratios analyte n DBS/b DBS/b RSD [%] range mean hydromorphone 15 0.86-1.08 0.99 6.36 morphine 7 0.95-1.02 0.99 2.22 fentanyl 13 0.90-1.09 1.00 8.56 norfentanyl 13 0.84-1.05 0.97 5.94 oxycodone 12 0.98-1.10 1.02 3.52 noroxycodone 12 0.89-1.06 1.00 4.32 alprazolam 22 0.92-1.25 1.02 6.71 zopiclone 45 0.63-1.22 0.86 15.86 temazepam 9 0.85-1.11 0.97 7.68 MDMA 35 0.97-1.09 1.01 2.46 MDA 30 0.85-1.09 1.02 5.11 d-amphetamine 29 0.94-1.15 1.05 5.23 risperidone 10 0.86-0.97 0.93 3.51 9-OH-risperidone 14 0.91-1.03 0.97 4.56

For all samples containing opioid type drugs the estimated DBS/b ratios were in an acceptable range and showed a small range of variation. Also, ratios and RSD for the benzodiazepine-type drugs alprazolam and temazepam as well as for the amphetamine-like drugs MDMA, MDA and d-amphetamine account for equality of analysis between the two media.

239 For zopiclone, a DBS/b ratio of 0.86 indicates an overestimation of the results from blood compared to DBS. This overestimation is supported by the mean difference calculated for the Bland-Altman difference plot of zopiclone (figure 3).

zopiclone: Bland-Altman difference plot

15,0

13,0

11,0

9,0

7,0

5,0

3,0

1,0

-1,0 difference (blood-DBS) [ng/mL] (blood-DBS) difference -3,0

-5,0 0 5 10 15 20 25 30 35 40 45 50 mean [(blood+DBS)/2] [ng/mL]

Figure 3: Bland-Altman difference plot for zopiclone. The solid line illustrates the mean difference of 3.99 ng/mL, the dotted lines indicate the limits of agreement set to 1.96xSD (-3.62 and 11.59 ng/mL).

Whole blood samples were stored at -20°C until analysis, whereas DBS were kept at ambient temperature. With respect to the different temperatures of storage, degradation of zopiclone to 2-amino-5-chloropyridine might have occurred which has recently been published for whole blood samples by Nilsson et al (32). Currently, stability investigations are carried out to compare the degradation in whole blood and DBS and to draw a conclusion concerning the better sample matrix and the best conditions for storage. Final results can be expected in September 2011. First investigations from residual samples of the P35 study indicate zopiclone degradation in whole blood specimen of 21.8 % after 30 days of storage at -20 %. In DBS stored under the same conditions the initial zopiclone concentration was decreased by only 10.0 %. The mean DBS/b ratio of risperidone was not as close to 1.00 as the respective ratios of the other analytes (except zopiclone). The risperidone Bland-Altman difference plot (figure 4) may, however, be useful to evaluate whether determination in DBS is equivalent to that in whole blood:

240 Figure 4: Bland-Altman difference plot for risperidone. The solid line illustrates the mean difference of 0.83 ng/mL, the dotted lines indicate the limits of agreement set to 1.96xSD (-0.67 and 2.32 ng/mL).

risperidone: Bland-Altman difference plot

2,50

2,00

1,50

1,00

0,50

0,00

difference (blood-DBS) [ng/mL] (blood-DBS) difference -0,50

-1,00 0 5 10 15 20 25 mean [(blood+DBS)/2] [ng/mL]

The Bland-Altman difference plot of risperidone clearly indicates that the methods of determination of risperidone in either blood of DBS are comparable. None of the measured quantities was outside the limits of agreement. A mean difference of 0.83 ng/mL is quite low, and only accounts for 7.4 % in relation to a mean blood concentration of risperidone of 11.1 ng/mL. Overall, both methods can be regarded as equivalent. MDMA could be quantified in DBS as reliable as in whole blood specimens which was already evident from the deliverable of July 2010 (29). The results obtained in the RugPsy study confirmed again equality of both methods. Moreover, equivalence of the methods could be proven for MDA: The DBS/b ratio of 1.02 is very close to 1.00; a very low coefficient of variation of 5.11 % and a very small mean difference between both methods could be estimated from the results. Figure 5 shows the Bland- Altman difference plot of MDA:

241 Figure 5: Bland-Altman difference plot for MDA. The solid line illustrates the mean difference of 0.02 ng/mL, the dotted lines indicate the limits of agreement set to 1.96xSD (-1.36 and 1.40 ng/mL).

MDA: Bland-Altman difference plot

3,5

3,0

2,5

2,0

1,5

1,0

0,5

0,0

difference (blood-DBS) [ng/mL] (blood-DBS) difference -0,5

-1,0

-1,5

-2,0 0,0 5,0 10,0 15,0 20,0 25,0

mean [(blood+DBS)/2] [ng/mL]

All values except for a single one are within the limits of agreement. No trend of the differences between the results obtained from either blood or DBS values over the whole concentration range could be observed. Especially the very small mean difference leaves no doubt that analysis of MDA from DBS is as reliable as from whole blood. In addition to the DBS/b ratios presented in table 6, table 7 gives the results of the Bland-Altman analyses for all analytes:

242 Table 7: Summary of results of the Bland-Altman difference plots analyte mean difference mean-1.96xSD mean +1.96xSD mean blood-DBS [ng/mL] [ng/mL] difference/mean [ng/mL] blood concentration [%] hydromorphone 0.14 -0.90 1.17 1.53 morphine 2.12 -8.30 12.53 0.88 fentanyl -0.03 -0.20 0.14 -2.16 norfentanyl -0.006 -0.068 -0.055 -1.34 oxycodone -1.24 -4.46 1.98 -2.07 noroxycodone 0.27 -3.26 3.80 0.63 alprazolam 0.07 -1.76 1.90 1.08 zopiclone 3.99 -3.62 11.60 15.30 temazepam 2.23 -16.19 20.66 2.71 MDMA -3.55 -14.34 7.25 -1.94 MDA 0.02 -1.36 1.40 0.17 d-amphetamine -1.03 -3.32 1.25 -5.01 risperidone 0.83 -0.67 2.32 7.44 9-OH-risperidone 0.64 -1.13 2.40 4.15

Obviously, for no analyte except zopiclone the mean difference exceeded ±10 % of the mean blood concentration. Therefore, the DBS and whole blood methods for the analytes investigated during the DRUID project can be regarded to be equivalent.

Conclusion

The b/p ratios of morphine, alprazolam, MDMA, amphetamine and risperidone obtained in the present studies closely fit to reported values. For hydromorphone, fentanyl, zopiclone and temazepam the calculated b/p ratios cannot be compared to published data, since some publications either provide no detailed information on the investigation or on the samples size. Some of the values were cited as b/p ratios but have been determined from other media such as packed erythrocytes. In the cases of oxycodone, noroxycodone and norfentanyl no reference data are currently available. The b/p ratio of MDA has assessed once again. The small concentrations resulting after administration of MDMA were close to the limit of quantification resulting in relatively high coefficients of variation. The DBS assay has potential as a precise and inexpensive option for the determination of several analytes in small blood samples. The DBS/b ratios were very close to 1.00, and the relative standard deviations ” 8.56 %. Measures of under-/overestimation were provided by Bland-Altman difference plots, and 95 % of all differences between the concentrations determined from either whole blood or DBS were within the limits of agreement. Also, differences were uniformly distributed across the concentration range. Except zopiclone, which is very sensitive to degradation, all substances

243 investigated in the presented studies could be determined in DBS as reliable as in whole blood specimens. A zopiclone stability study in blood and DBS samples is running to compare the extent of degradation in both media: then, conclusions can be drawn concerning the optimum sample material and optimum temperature for storage.

244 Appendix: Table 8: Summary of plasma, blood and DBS concentrations of all analytes including major metabolites and their b/p and DBS/b ratios analyte n range mean media range mean media range mean media range mean range mean remark plasma plasm n blood blood n DBS DBS n DBS b/p b/p DBS/b DBS/ s [ng/mL] a plasm [ng/mL] [ng/mL blood [ng/mL] ng/mL] [ng/m b [ng/mL a ] [ng/m L] ] [ng/m L] L] hydromorphone 16 2.4-19.7 8.4 6.3 2.3-20.2 8.8 7.0 2.4-19.3 8.7 7.6 0.91- 1.04 0.86- 0.99 1 1.22 1.08 morphine 7 85.7-468.9 232.7 206.7 81.4- 240.2 215.7 82.2- 238.1 209.2 0.96- 1.03 0.95- 0.99 479.1 472.6 1.07 1.02 fentanyl 13 0.20-8.67 1.53 0.71 0.18-6.10 1.24 0.52 0.18-6.27 1.26 0.49 0.62- 0.87 0.90- 1.00 1.02 1.19 norfentanyl 13 0.07-1.82 0.40 0.17 0.08-2.21 0.46 0.21 0.07-2.28 0.47 0.21 1.03- 1.19 0.84- 0.97 2 1.29 1.05 oxycodone 12 11.6-89.8 41.6 33.4 17.8- 59.9 53.7 19.3- 61.2 57.1 1.29- 1.48 0.98- 1.02 122.6 123.9 1.76 1.10 noroxycodone 12 11.0-71.1 26.4 19.5 20.6-91.2 42.9 33.6 20.4-86.5 42.1 33.8 1.28- 1.73 0.89- 1.00 3 2.09 1.06 buprenorphine 6 0.30-0.84 0.53 0.45 0.09-0.47 0.26 0.21 0.17-0.48 0.31 0.24 0.50- 0.55 0.99- 1.07 1 0.63 1.13 6 0.58-2.35 1.17 0.91 0.24-2.41 1.01 0.82 0.26-2.39 1.02 0.82 0.95- 0.98 0.98- 1.02 1, 4 1.03 1.09 tramadol 2 75.4-322.8 204.1 - 93.4- 211.0 - 90.9- 240.4 - 1.15- 1.19 0.97- 1.00 382.6 389.8 1.24 1.02

245 O- 2 11.9-69.9 40.9 - 16.7-65.5 41.1 - 16.0-67.3 41.7 - 0.94- 1.17 0.96- 0.99 5 desmethyltramadol 1.40 1.03 N- 2 148.1- 161.6 - 190.7- 201.4 - 195.2- 201.1 - 1.21- 1.25 0.92- 1.00 5 desmethyltramadol 175.1 212.1 206.9 1.29 1.09 1 - 2.16 - - 2.38 - - 2.21 - - 1.10 - 0.93 nortilidine 1 - 64.5 - - 70.3 - - 70.6 - - 1.09 - 1.00 6 bisnortilidine 1 - 133.8 - - 204.1 - - 205.4 - - 1.53 - 1.01 6 THC 31 0.8-92.0 6.4 3.2 ------7 2 11-OH-THC 31 0.5-25.0 3.6 2.5 ------7 2 THC-COOH 31 3.0-175.0 38.5 29.0 ------7 2 alprazolam 52 3.11-51.3 6.7 8.6 3.7-20.7 6.6 5.8 4.2-20.1 6.9 6.0 0.73- 0.81 0.92- 1.02 8 0.90 1.25 zopiclone P35 45 11.9-50.4 29.8 29.6 10.1-49.1 26.1 24.9 4.7-39.8 22.1 21.5 0.66- 0.89 0.63- 0.86 9 1.29 1.22 zopiclone P31 8 ------10 diazepam 2 215.9- 270.7 - 200.6- 232.2 - 216.0- 240.6 - 0.81- 0.87 1.00- 1.04 325.5 263.7 265.2 0.93 1.08 nordiazepam 2 184.5- 218.3 - 185.4- 200.5 - 177.5- 233.2 - 0.86- 0.93 0.96- 1.15 11 252.2 215.6 288.8 1.00 1.34 temazepam 10 114.6- 198.6 209.8 82.5- 141.3 154.2 70.2- 139.0 155.5 0.58- 0.71 0.85- 0.98 11 271.2 191.7 184.6 0.87 1.11 oxazepam 16 4.6-430.0 103.0 44.6 12.1- 79.0 28.3 7.4-356.8 79.7 20.8 0.40- 0.93 0.61- 0.92 11, 12

246 355.7 2.65 1.16 midazolam 6 13.8-46.4 25.1 24.3 10.3-27.6 16.9 15.7 7.3-35.1 18.8 17.7 0.60- 0.70 0.71- 1.06 0.89 1.35 clonazepam 1 - 18.8 - - 14.4 - - 15.4 - - 0.77 - 1.07 desalkylflurazepam 3 48.8-60.3 53.0 - 25.2-30.0 28.1 - 29.8-42.1 36.8 - 0.42- 0.54 1.18- 1.30 13, 14 0.60 1.45 hydroxyethylflurazep 2 ------13, 15 am lormetazepam 4 2.2-7.7 4.6 4.2 1.0-4.5 2.6 2.5 1.6-4.2 2.5 2.1 0.46- 0.55 0.71- 1.10 0.62 1.56 nitrazepam 2 57.5-58.0 57.8 - 45.3-50.4 47.9 - 41.9-54.4 48.2 - 0.79- 0.83 0.92- 1.00 0.87 10.8 zolpidem 4 15.6-54.1 33.4 31.8 19.5-54.7 33.2 29.4 23.9-75.6 43.8 37.8 0.78- 1.09 1.14- 1.28 1.41 1.38 MDMA 39 5.5-388.6 157.8 145.7 7.0-444.3 180.5 167.2 7.0-444.5 186.4 173.7 0.98- 1.19 0.97- 1.01 16 1.46 1.09 MDA 39 0.97-13.5 5.1 5.2 1.12-24.4 9.4 9.6 0.54-24.1 8.4 7.9 1.00- 1.72 0.85- 1.02 17 3.02 1.09 d-amphetamine 29 12.20- 23.21 21.60 10.75- 20.61 19.60 10.90- 21.65 20.35 0.65- 0.89 0.94- 1.05 41.13 40.65 43.90 1.14 1.15 risperidone 16 1.7-36.8 16.0 9.8 4.3-22.7 11.1 6.6 1.2-20.8 9.5 6.2 0.56- 0.65 0.86- 0.93 18 0.73 0.97 9-OH-risperidone 16 4.9-46.5 21.3 22.2 4.1-29.8 15.3 15.5 4.2-30.8 14.7 14.7 0.61- 0.73 0.91- 0.97 19 0.91 1.03 2: as a metabolite of fentanyl 4: as a metabolite of buprenorphine 1: one sample < LLOQ 3: as a metabolite of oxycodone 5: as a metabolite of tramadol

247 6: as a metabolite of tilidine 17: as an active metabolite of MDMA; 7: DBS analysis still in progress; plasma plasma: 37 samples, 29>LLOQ; blood: 38 samples>LLOQ: THC: 178, 11-OH-THC: 193, samples, 30>LLOQ; DBS: 38 samples, 34>LLOQ THC-COOH: 200 18: plasma and DBS: 11 samples >LLOQ, 8: plasma: 42>LLOQ; blood: 33 samples, blood: 10 samples >LLOQ 24>LLOQ; DBS: 33 samples, 22>LLOQ 19: active metabolite of risperidone; in each media 14 samples >LLOQ -: not applicable 9: suspected degradation of zopiclone in DBS; stability study to compare the degradation in whole blood and DBS is still running 10: concentrations in the P31 study were too low to provide reliable results 11: in 2 samples as a metabolite of diazepam 12: in 8 samples as a metabolite of temazepam; wide range of concentrations, but not normally distributed (compare great discrepancies between the mean and the median concentrations in the three different media) 13: active metabolite of flurazepam 14: in one samples accidental detection although not the planned medication in the study

15: only detectable in traces 16: plasma: 37 samples, 34>LLOQ; blood+DBS: 38 samples, 35>LLOQ

248 References

1. J. Chamberlain. The analysis of drugs in biological fluids. CRC Press, Boca Raton, Florida, USA, (1995). 2. C. Geigy. Geigy Scientific Tables. Ciba Geigy Limited, Basle, Switzerland, (1997). 3. K. P. Maguire, G. D. Burrows, T. R. Norman and B. A. Scoggins. Blood/plasma distribution ratios of psychotropic drugs. Clin Chem 26: 1624-1625 (1980). 4. G. Skopp, L. Pötsch, B. Ganßmann, R. Aderjan and R. Mattern. A preliminary study on the distribution of morphine and its glucuronides in the subcompartments of blood. J. Anal. Toxicol. 22: 261-264 (1998). 5. G. Skopp, K. Klinder, L. Poetsch, G. Zimmer, R. Lutz, R. Aderjan and R. Mattern. Postmortem distribution of and metabolites in a fatal case of dihydrocodeine intoxication. Forensic Sci. Int. 95: 99-107 (1998). 6. A. W. Jones and H. Larsson. Distribution of diazepam and nordiazepam between plasma and whole blood and the influence of hematocrit. Ther. Drug Monit. 26: 380-385 (2004). 7. C. Giroud, A. Ménétrey, M. Augsburger, T. Buclin, P. Sanchez-Mazas and P. Mangin. ǻ9- THC, 11-OH-ǻ9-THC and ǻ9-THCCOOH plasma or serum to whole blood concentrations distribution ratios in blood samples taken from living and dead people. Forensic Sci. Int. 123: 159-164 (2001). 8. D. H. Chace, T. A. Kalas and E. W. Naylor. Use of tandem mass spectrometry for multianalyte screening of dried blood spot specimens from newborns. Clin. Chem. 49: 1797-1817 (2003). 9. P. M. Edelbroek, J. v. d. Heijden and L. M. L. Stolk. Dried blood spot methods in therapeutic drug monitoring: methods, assays, and pitfalls. Ther. Drug Monit. 31: 327-336 (2009). 10. R. Guthrie and A. Susi. A simple phenylalanine method for detecting phenylketonuria in large populations of human infants. Pediatrics 32: 338-343 (1963). 11. J. V. Mei, J. R. Alexander, B. W. Adam and W. H. Hannon. Use of filter paper for the collection and analysis of human whole blood specimens. J. Nutr. 131: 1631S-1636S (2001). 12. S. P. Parker and W. D. Cubitt. The use of dried blood spot sample in epidemiological studies. J. Clin. Pathol. 52: 633-639 (1999). 13. N. Spooner, R. Las and M. Barfield. Dried blood spots as a sample collection technique for the determination of pharmacokinetics in clinical studies: considerations for the validation of a quantitative bioanalytical method. Anal. Chem. 81: 1557-1563 (2009). 14. W. Li and F. L. S. Tse. Dried blood spot sampling in combination with LC-MS/MS for quantitative analysis of small molecules. Biomed. Chromatogr. 24: 49-65 (2010). 15. G. Skopp and L. Poetsch. Nachweis von Cocain in Blutspots. Ach. Kriminol. 207: 81-88 (2001). 16. G. Skopp. Preanalytic aspects in postmortem toxicology. Forensic Sci. Int. 142: 75-100 (2004). 17. FDA Guidance for Industry: Bioanalytical Method Validation. U.S. Department of Health and Human Services. Center for Drug Evaluation and Research. Food and Drug Administration, 2001. 18. S. Bansal and A. DeStefano. Key elements of bioanalytical method validation for small molecules. The AAPS Journal 9: E109-E114 (2007).

249 19. B. K. Matuszewski, M. L. Constanzer and C. M. Chavez-Eng. Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Anal. Chem. 75: 3019- 3030 (2003). 20. Deutsches Institut für Normung. DIN 32645. Chemische Analytik - Nachweis- Erfassungs- und Bestimmungsgrenze unter Wiederholbedingungen - Begriffe, Verfahren, Auswertung. Beuth, Berlin, 2008. 21. J. M. Bland and D. G. Altman. Measuring agreement in method comparison studies. Stat. Methods Med. Res. 8: 135-160 (1999). 22. P. V. Parab, W. A. Ritschel, D. E. Coyle, R. V. Gregg and D. D. Denson. Pharmacokinetics of hydromorphone after intravenous, peroral and rectal administration to human subjects. Biopharm. Drug Dispos. 9: 187-199 (1988). 23. C. W. Hand and R. A. Moore. Comparison of whole blood and plasma morphine. J. Anal. Toxicol. 12: 234-235 (1988). 24. S. Bower and C. J. Hull. Comparative pharmacokinetics of fentanyl and alfentanyl. Br. J. Anaesth. 54: 871-877 (1982). 25. D. J. Crouch, D. E. Rollins, D. V. Canfield, D. M. Andrenyak and J. E. Schultis. Quantitation of alprazolam and Į-hydroxyalprazolam in human plasma using liquid chromatography electrospray ionization MS-MS. J. Anal. Toxicol. 23: 479-485 (1999). 26. J. G. Bramness, M. Arnestad, R. Karinen and T. Hilberg. Fatal overdose of zopiclone in an elderly woman with bronchogenic carcinoma. J. Forensic Sci. 46: 1247-1249 (2001). 27. M. D. Osselton, M. D. Hammond and A. C. Moffat. Distribution of drugs and toxic chemicals in blood. J. Forens. Sci. Soc. 20: 187-193 (1980). 28. R. Garcia Boy, J. Henseler, J. G. Ramaekers, R. Mattern and G. Skopp. A comparison between experimental and authentic blood/serum ratios of 3,4-methylenedioxymethamphetamine and 3,4-methylenedioxyamphetamine. J. Anal. Toxicol. 33: 283-286 (2009). 29. G. Skopp and R. Jantos. Blood to plasma ratios of amphetamine and MDMA and analysis of dried blood spots. DRUID project deliverable. (2010). 30. G. Mannens, W. Meuldermans, E. Snoeck and J. Heykants. Plasma protein binding of risperidone and its distribution in blood. Psychopharmacology 114: 566-572 (1994). 31. H. J. Hemlin, K. Bracher, D. Bourquin, D. Vonlanthen and R. Brenneisen. Analysis of 3,4- methylendioxymethamphetamine (MDMA) and its metabolites in plasma and urine by HPLC-DAD and GC-MS. J. Anal. Toxicol. 20: 432-440 (1996). 32. G. H. Nilsson, F. C. Kugelberg, R. Kronstrand and J. Ahlner. Stability tests of zopiclone in whole blood. Forensic Sci. Int. 200: 130-135 (2010).

250