Potential Drug Interactions and Potentially Inappropriate Medications in Daily Radiooncology Practice : a Risk Assessment

From the Department of Radiotherapy and Radiation Oncology of Marienhospital Herne – University Hospital – of the Ruhr-Universität Bochum Director: Prof. Dr. med. I. A. Adamietz

Potential drug interactions and potentially inappropriate medications in daily radiooncology practice – a risk assessment

Inaugural-Dissertation for the Attainment of a Doctor´s Degree in Medicine at the high Medical Faculty of the Ruhr-Universität Bochum

Presented by Nina Sibylle Lamberty from Düsseldorf 2013

Dean of Faculty: Prof. Dr. med Klaus Überla Referee: Prof. Dr. med. Irenäus A. Adamietz Co-Referee: PD Dr. med Oliver Lindner Date of oral exam: 03.02.2015

Abstract

Lamberty Nina Potential drug interactions and potentially inappropriate medications in daily radiooncology practice – a risk assessment

Background: Radiation treatment of malignant disease includes patients with various extents of tumor and different comorbidities. Due to bad general condition approximately 20 % of irradiated subjects are in-patients. Beyond elimination of tumor and treatment of its side effects drug interactions and potentially inappropriate medications constitute a major challenge in these radiooncologic patients. The layering of polypharmacy with age and radiation related physiological and functional changes and comorbidities might increase the risk for potential drug drug interactions (pDDs). This study attempted to quantify the frequency of pDDIs and potential inappropriate medications (PIMs) not related to i.v. chemotherapy in patients undergoing radiation treatment aiming the assessment of risk by medication.

Methods: Medication profiles were analyzed by reviewing discharge letters of 120 cancer patients who had been admitted to the Marienhospital, Herne between November 2010 and November 2011. An improvement of the Karnofsky index during hospital treatment had been registered. The medication of all patients (n = 120) was screened for pDDIs at hospital admission and at discharge using the ABDA drug interaction software (Pharmatechnik®). Potential interactions, graded by their levels of severity were identified. The patient’s medication profile was evaluated additionally in terms of potentially inappropriate medications due to an unacceptable risk-to-benefit ratio. Potential inappropriate medications in cancer patients were identified using the Priscus list (PIMs) and the Cave module (Indi- vidual inappropriate medications (IIMs). Logistic regression was applied to determine odds ratios for specific risk factors of pDDIs and potentially inappropriate medications i.e. age, gender and number of medications.

Results: The evaluation of the admission and discharge medication revealed that there was a significant risk for being harmed by potential drug interactions although the Karnofsky index improved after radiooncologic treatment (Mean admission: 67.9; Mean discharge: 75.4). At hospital discharge significantly more pDDIs per patient (2.25) were detected than at hospital admission (1.59) (p = 0.001). Most potential drug interactions (46.7 %) involved non- anticancer agents such as antihypertensive drugs, corticosteroids, anticoagulants and NSAIDs. According to the Frechen-Score the most frequently involved drugs in therapeutically relevant pDDIs were cardiovascular drugs, insulin, corticosteroids and NSAIDs, whereas antipsychotics, antidepressants and antiemetics were rarely involved in potential drug interactions. In multivariate analysis, increased risk of receiving drug combinations in which there were potential drug interactions was associated with receipt of increasing numbers of drugs (p = 0.001). According to the Cave module 362 prescribed IIMs were inappropriate due to increased age (39 %) and underlying metabolic (25 %) or cardiovascular diseases (13.2 %). With increasing age (p = 0.003), number of comorbid diseases (p = 0.005) and the number of medications (p < 0.001) the proportion of patients receiving IIMs increased. The three most common IIMs were antihypertensives (18.3 %), NSAIDs (11.3 %), and corticosteroids (10.3 %). Of 79 patients aged > 65 46.8 % were taking at least one PIM, as defined by the german Priscus list at hospital discharge.

Discussion: Although, there was an improvement of general performance status after hospital treatment the present study recorded a high prevalence of pDDIs and PIMs in the radiooncologic setting. Additional medication, including supportive therapies and concomitant medications should be weighed carefully for benefit versus risk of ADEs in the context of existing regimens prior to start of radiation. In the context of risk management recommendation guidelines for day-to-day routine were developed for radiooncologic patients. The risk by medications should be assessed periodically to reduce the overall risk potential.

To my parents

Table of contents

1 Introduction ...... 10 1.1 Background 10 1.2 Risk management and outline of the thesis 11 1.3 Literature review and introduction into the topic 12 1.3.1 Drug related problems 14 1.3.2 Drug drug interactions 17 1.3.3 Potential inappropriate medications (PIMs) 28 1.3.4 Medicine safety management 30 1.3.5 Documentation of drug interactions and inappropriate medications 32 1.3.6 Management of drug interactions and inappropriate medications 33

2 Aim of the thesis ...... 36

3 Material und methods ...... 38 3.1 Material, records 38 3.2 Patients informed consent 38 3.3 Study population 38 3.4 Study design 39 3.5 Data ascertainment 39 3.5.1 Cave module 40 3.5.2 Steps of data extraction 40 3.5.3 Patient characteristics and medical conditions 42 3.5.4 Drug combinations examined 44 3.5.5 Chronic medications and supportive medications 45 3.5.6 Definition of the classification pattern used to estimate the risk potential for potential drug interactions 45 3.5.7 Evaluating potential drug interactions 47 3.5.8 Potential drug interactions classification and flagging 47 3.6 Potentially inappropriate medications 47 3.6.1 Cave module 48 3.6.2 Priscus list 48 3.7 Statistical analysis 48

4 Results ...... 49 4.1 Patient characteristics 49 4.2 Comorbidities 51 4.2.1 Comorbidities in the study cohort 51

4.2.2 Comorbidity scores: Karnofsky index and Charlson comorbidity index 52 4.3 Prescribed medications 55 4.4 Analysis of potential drug interactions 60 4.4.1 Number of potential drug interactions 60 4.4.2 Classification of potential drug interactions 62 4.4.3 Correlation analysis 63 4.4.4 Medication classes involved frequently in potential drug interactions 66 4.4.5 Score predicting the interaction potential 73 4.4.6 Effects induced by potential drug interactions 75 4.5 Potentially inappropriate medications 79 4.5.1 Cave module: Frequency of IIMs and correlation analysis 79 4.5.2 Pricus list medications 85

5 Discussion ...... 87 5.1 Patient population and prevalence of potential drug interactions 87 5.2 Clinical relevance of drug combinations commonly involved in pDDIs in the radiooncologic setting 89 5.2.1 Major potential drug interactions (Category II) 90 5.2.2 Moderate potential drug interactions (Category III and Category IV) 91 5.2.3 Minor potential drug interactions (Category V) 96 5.2.4 Potential drug interactions with oral anticancer drugs 97 5.2.5 Score predicting the interaction potential (Gaertner et al., 2012) 97 5.3 Factors associated with the presence of pDDIs 100 5.4 Potentially inappropriate medications 102 5.4.1 Cave module and other classification systems (Beers criteria and STOPP criteria) 102 5.4.2 Priscus list 107 5.5 Critical appraisal of own methods 108

6 Key results, management strategies and conclusions ...... 111 6.1 Key results 111 6.2 Management strategies 112 6.2.1 Potential drug interactions 112 6.2.2 Potentially inappropriate medications 118 6.3 Conclusions and outlook 118

Bibliography ...... 120

Appendix ...... 137 A ABDA database 137 A 1.1 ABDA database 137 A 1.2 Evaluating drug interactions by the ABDA database 137 A 1.3 The ABDA database mode of action 138

A 2 Cave module 138 B Study data 140 B 1 Drug interactions 140 B 2 Potential inappropriate medications 145 B 3 Case report 149 B 3.1 Potential drug interactions with a CYP P450 inducer 149 B 3.2 Drug disease interactions – Example 1 151 B 3.3 Drug disease interaction – Example 2 152 C Potentially inappropriate medications 153

Abbreviations

ABDA Bundesvereinigung Deutscher Apothekerverbände (Federal organization of the German pharmacist associations)

ACE Angiotensin-converting enzyme

ADE Adverse drug event

ADR Adverse drug reaction

ATC Anatomical therapeutical chemical

BMI Body mass index

CAD Coronary artery disease

CI Confidence interval

CNS Central nervous system

CSEs HMG-CoA reductase inhibitors

Corticost. Corticosteroids

CPOE Computerized physician order entry

CDSS Clinical decision support system

CYP Cytochrome P450 isoenzyme

DDI Drug-drug interaction

DIF Drug Interaction Facts

DR Drug-Reax

DRP Drug-related problem

FPH Foederatio Pharmaceutica Helvetiae

GFR Glomerular filtration rate

IAC-C Interactions between chronic and chronic medications

IAS-C Interactions between supportive and chronic medications

ICD-10 International classification of diseases, 10th revision

IIM Individual inappropriate medication

INR International normalized ratio

NPV Negative predictive value

NSAIDs Non-steroidal antiinflammatory drugs

OR Odds ratio

ORCA OpeRational ClassificAtion

OTC drugs Over the counter drugs (selfmediction)

PAD Peripheral artery disease

PCNE Pharmaceutical Care Network Europe pDDIs potential drug-drug interactions

P-gp P-glycoprotein

PIMs Potentially inappropriate medications

PPIs Proton pump inhibitors

PV Pharmavista

RINV Radiation induced nausea and vomiting

SERM Selective estrogen receptor modulator

SSRI Selective serotonin reuptake inhibitor

TCAs Tricyclic antidepressants

WHO World Health Organization yr Year

List of figures

Figure 1-1. Procedure of the applied risk management. 11 Figure 1-2. Prescribing Cascade according to (Rochon et al.1997). 16 Figure 1-3. Age related impairment of organ function and polymorbidity frequently lead to an increased risk of adverse drug reactions in the elderly. 21 Figure 1-4. The Swiss cheese model adopted from (Perneger et al. 2005). 31 Figure 3-1. Study design. 39 Figure 4-1. Clinical symptoms at hospital admission. 51 Figure 4-2. Number of comorbidities in the study cohort stratified according to age (left), different types of comorbidities (right). 52 Figure 4-3. Karnofsky index score in the study collective stratified according to age groups. 53 Figure 4-4. Changes in score values in the Karnofsky index during hospital admission and discharge. 54 Figure 4-5. The Charlson comorbidity index score (age adjusted and unadjusted). 54 Figure 4-6. Number of medications per patient in the admission and discharge medication portfolio. 56 Figure 4-7. Number of prescribed medications in the study cohort. 56 Figure 4-8. Number of patients with prescriptions belonging to the drug classes according to the anatomical therapeutical chemical index (ATC) classification at hospital admission and hospital discharge. 57 Figure 4-9. Prevalence of the most frequently prescribed medications in 120 patients stratified by age group at hospital admission. 58 Figure 4-10. Prevalence of the most frequently prescribed medications in 120 patients stratified by age group at hospital discharge. 59 Figure 4-11. Prescription frequency of supportive medications. 60 Figure 4-12. Number of potential drug interactions between chronic medications (IAC-C); and supportive and chronic medications (IAS-C). Hospital admission (left), hospital discharge (right). 62 Figure 4-13. Distribution and classification of potential interactions according to the classification system of the ABDA drug interaction module. 62 Figure 4-14. Distribution of the detected pDDIs among similar ABDA monographies. 63 Figure 4-15. Number of pharmacologically active substances prescribed and prevalence of potential drug interactions (pDDIs) stratified by age group at hospital admission. 65 Figure 4-16. Number of pharmacologically active substances prescribed and prevalence of potential drug interactions stratified by age group at hospital discharge. 66 Figure 4-17. Most frequently prescribed medications which have been involved in potential drug interactions. 67/68 Figure 4-18. Potential drug interactions between supportive and chronic medications as well as supportive and supportive medications 69 Figure 4-19. Effects which might have been induced by potential drug interactions in the study cohort. 76 Figure 4-20. Distribution of effects of potential drug interactions over the patient collective. 77

Figure 4-21. Medication groups most frequently involved in IIM prescribing (left). Antihypertensive drugs most frequently involved in IIM prescribing (right). 82 Figure 4-22. Individual inappropriate medications due to underlying diseases. 83 Figure 4-23. Priscus medications detected in the study collective 86

List of tables

Table 1-1. Operational Classification of Drug interactions (ORCA) innovated by Hansten and Horn adopted by the ABDA database (Indermitte, 2006). 18 Table 3-1. Definition of the Charlson comorbidity index (CCI) (Charlson et al., 1987). 43 Table 3-2. Classification of comorbidities in four disease stages (Charlson et al., 1987). 44 Table 4-1. Patient characteristics. 49 Table 4-2. Age distribution in the patient collective. 50 Table 4-3. Distribution in CCI score. 55 Table 4-4. Correlation between Charlson comorbidity index and number of potential drug interactions. 55 Table 4-5. Number of patients with n = (1, 2, 3, 4, 5-18) potential drug interaction combinations at hospital admission and discharge. 61 Table 4-6. Factors significantly associated with an increased risk of clinically relevant potential drug interactions at hospital discharge using univariate and multivariate regression analysis. 63 Table 4-7. Prevalence of pDDIs in radiooncologic patients treated with oral anticancer drugs. 70 Table 4-8. Eight most frequently prescribed drug classes in the patient cohort and the most prevalent clinically relevant potential drug- drug interactions (pDDIs) identified in radiooncologic patients. 71 Table 4-9. Drugs with a high potential for potential drug interactions. 74 Table 4-10. Drugs with a low potential for potential drug interactions. 74 Table 4-11. Drugs with a changing potential for potential drug interactions. 75 Table 4-12. Most frequently involved drug pairs and possible induced effects. 77 Table 4-13. Frequency of IIMs in total, differentiated in IIMs contraindicated and restriction on use. 79 Table 4-14. Frequency of variables (Gender, age and number of comorbid diseases) in relation to the number of IIMs in total. 80 Table 4-15. Frequency of variables (Gender, age and number of comorbid diseases) in relation to the number of contraindicated IIMs. 80 Table 4-16. Frequency of variables (Gender, age and number of comorbid diseases) in relation to the number of IIMs restricted on use. 81 Table 4-17. Frequency of number of IIMs at hospital admission and discharge. 81 Table 4-18. Inappropriate individual medications due to age inappropriateness detected by the Cave module. 83 Table 4-19. Prevalence of Priscus related drugs at admission and discharge in the elderly sub-population. 86 Table 5-1. Most commonly prescribed potentially inappropriate medications (PIMs) as defined per STOPP criteria. 103 Table 6-1. Overview of the most frequently detected potential drug interaction pairs found in the study cohort. 114

Table 6-2. List B Management guidelines for most frequently identified drug drug interactions modified and adopted from [Hansten, 2012, Hansten, 2013]. 115 Table A-1. Classification of Cave module. 139 Table B-1. Overview of the most frequently detected potential drug interactions of Category II and III. 140 Table B-2. Overview of the most frequently detected potential drug interactions of Category IV. 141 Table B-3. Overview of the most frequently detected potential drug interactions of Category V and VI. 143 Table B-4. Drugs (IIMs) to be avoided with certain diseases / conditions (Metabolic system) evaluated by the Cave module. 145 Table B-5. Drugs (IIMs) to be avoided with certain diseases / conditions (Cardiovascula system) evaluated by the Cave module. 146 Table B-6. Drugs (IIMs) to be avoided with certain diseases / conditions (Gastroenterologic system, Pulmonolgy, Neurological system) evaluated by the Cave module. 147

1 Introduction

1.1 Background The human and economical burden associated with drug related problems is staggering. “Many patients die due to medical treatment and not due to underlying chronic diseases.” This quota- tion from Moliere (1673 Le malade imaginaire) still reflects the difficulties of drug therapy in the present century (Fischalek, 2009, Horn and Hansten, 2009). One third of all hospital admissions are linked to drug related problems, many of which are drug drug interactions that could have been avoided (Juurlink et al., 2004). The prescription of medications is becoming more complex (Anthierens et al., 2010). The permanent risk of adverse reactions and interactions rises because of the ageing population, the increasing polypharmacy in cancer and multimorbid patients and the pharmacological complexity of modern drugs (Anthierens et al., 2010). Mil- lions of drug interactions occur annually despite major efforts to minimize exposure to known hazards (Malone et al., 2003).

Prescription of many drugs in ambulatory care results out of independent not coordinated- therapy regimens prescribed by different physicians (Shimp et al., 1985). An increase in the number of drugs (Rochon and Gurwitz, 1997) is caused by too long lasting therapies and treatments of masked side effects with additional medications. Furthermore effective drugs are combined with ineffective drugs without discontinuing drugs leading not to an appropriate effect. Self-medication with over the counter drugs and complementary medicine enhances the risk for drug interactions.

Drug–drug interactions (DDIs) and potential inappropriate medications (PIMs) comprise an important problem in medical oncology practice (Riechelmann et al., 2005). Cancer patients need close supervision of drug therapy. Apart from the complex systemic antineoplastic therapy, supportive care is a firm component of therapy. Cancer patients frequently take many additional medications to treat treatment-induced toxicity and cancer-related syndromes and to treat other comorbid illnesses. The risk for drug interactions rises steadily with the growing complexity of radiooncologic treatment. This can have pharmacological, clinical pharmacokinetic or physicochemical reasons. A recent study (Riechelmann et al., 2007) revealed that one third of cancer patients is endangered by potential drug interactions primarily from treatments for comorbid conditions or supportive cancer care.

Tools for the identification of potential drug- drug interactions (pDDIs) and potentially inappropriate medications (PIMs) exist since some years. Computerised alerts systems for drug interactions enable important assistance. Through a timeous identification of pDDIs and PIMs a timely

intervention in drug therapy within the scope of risk management is possible and hence contrib- utes to an evidence based risk reduction (Bertsche et al., 2010).

1.2 Risk management and outline of the thesis Risk management is the identification, assessment, and prioritization of risks (defined in ISO 31000 as the effect of uncertainty on objectives, whether positive or negative) followed by coordinated and economical application of resources to minimize, monitor, and control the probability and/or impact of unfortunate events (Hubbard, 2009).

Risk management related to potential drug interactions and potentially inappropriate medications contains the following elements (Figure 1-1):

Assess

Identify Determine

Risk managment

Reduce Review risk

Figure 1-1. Procedure of the applied risk management.

1. Identify: Recognizing the risk of being harmed by pDDIs and PIMs in cancer patients receiving radiation treatment 2. Assess: Characterization of pDDIs and PIMs with drug screening programs and identification of risk factors for pDDIs and PIMs 3. Determine: Evaluating potential adverse outcomes of pDDIs and PIMs 4. Identify ways to reduce risks: Evaluating recommendations for clinical practice including laboratory tests in the clinical risk management of pDDIs and developing strategies for PIM reduction based on lists like Priscus list, STOPP criteria 5. Prioritize risk reduction measures based on a strategy: Review management guidelines and update if necessary in future studies

In this study the risk management procedure described above was applied.

The research thesis is structured as follows. An outline of current research that has been performed in the field of drug interactions and potentially inappropriate medications in different

clinical settings is given in a subchapter following this overview. Risk factors for the occurrence of pDDIs have been identified. The newest classification system of potential drug interactions (ORCA) was described. (STEP 1: Identify risk of beeing harmed by pDDIs).

In chapter 2 the aim of the preliminary retrospective study is presented.

In chapter 3 materials and methods used in the thesis are presented. Different approaches to evaluate the overall risk of mortality and morbidity are presented in cancer patients (Karnofsky index, Charlson comorbidity index). The evaluation of the overall risk is important in order to identify patients which are at an increased risk of pDDIs and PIMs. Further the exact flow of data ascertainment is presented in detail. The interaction potential of drugs was identified using the flag score firstly implemented by (Frechen et al., 2012). The flag score calculation is presented in detail.

In chapter 4 the results of the retrospective survey are presented (STEP 2: Assess the risk for pDDIs and PIMs).

In chapter 5 the discussion section is included. The clinical relevance of the 12 most frequently detected pDDIs was assessed by reviewing current literature. A comparison with the recommendations found in the literature will reveal possibilities for improvement of drug therapy in elderly cancer patients (STEP 3 Evaluate the risk).

In chapter 6 key results and management strategies are presented. Preliminary recommendations for daily clinical practice are given (Step 4 Identify ways to reduce risks). Recommendations for future studies including review procedures are given in the conclusion and outlook section (Step 5 Review management guidelines).

In the annex basic study data are listed. In addition three examples for the occurrence of pDDIs and drug disease interactions are given reflecting the clinical relevance of this issue.

1.3 Literature review and introduction into the topic Actual data in Germany document that 93054 drugs with twelve thousand active ingredients are on the market in 2013. Besides that 46662 non-prescription drugs are available for treating illnesses (Bundesinstitut für Arzneimittel und Medizinprodukte, 2013). Each patient takes in the mean 1250 tablets per year.

A recently performed study pointed out that one third of all insured patients takes between 5-8 drugs daily, whereas 20 % in the age group between 75 to 80 take between 9-12 drugs and 13 % of these even up to 20 drugs (Corcoran, 1997). Interactions of these drugs on a pharmacologic basis are probable but often unknown (Hoffmann and Beyer, 2006). In case they are not recognized as they may lead to adverse clinical events, DDIs comprise a significant cause of morbidity and mortality worldwide. Potential drug drug interactions may result in a decrease or an inac-

tivation of the therapeutic effect of a drug, and may enhance drug toxicity and indirectly compromise treatment outcomes and adherence (Riechelmann and Del Giglio, 2009). Not all drug interactions can be predicted. Those that are predictable are not always avoidable (Blower et al., 2005). The uncritical exposure to drugs implicates 28.000-58000 cases of death annually (Focus Verlag GmbH, 2012). A recent study found out that 2 % of hospitalised patients had a drug interaction as cause of hospital admission (Miranda et al., 2011). A large Canadian study reported that the risk of hospitalisation for people over 65 years substantially increased when exposed to a drug interaction (Peterson, 2011). In a Norwegian study 4 % of cancer related deaths in hospitalized patients were associated with severe drug interactions (Riechelmann et al., 2007). A scientific study of the AOK in 2012 revealed that in the first quarter of 2012, in 206000 cases, drug combinations have been prescribed which should not have been administered together. In recent studies the estimated number of patients receiving drugs with a potential to interact and leading to a change in therapeutic efficacy vary between 0.63 % and 53 % (Bucsa et al., 2013, Vonbach et al., 2008, Zhan et al., 2005). According to (Becker et al., 2007) 0.57 % of hospital admissions, 0.12 % of re-hospitalisations and 0.054 % of emergency department visits are related to the occurrence of drug drug interactions. In relation to the study setting the drugs involved in DDIs and the reactions caused by the DDIs differ.

Recent studies revealed that there exists an exponential correlation between the number of drugs administered and the number of potential drug drug interactions (Johnell and Klarin, 2007, Mateti et al., 2011). In this context patient’s undergoing polypharmacy have a higher risk to be endangered by ADRs. Radiooncologic treatment often requires multivalent cancer regimens, which include radiation, supportive therapies and medications to counter the side effects and squeals of treatment and disease (Lapi et al., 2010). (Goldberg et al., 1996) showed that patients taking two drugs face a 13 % risk of adverse drug- drug interactions, rising to 38 % for four drugs and even up to 82 % in case seven or more drugs are given concurrently (Goldberg et al., 1996, Page et al., 2010). Karas reported that the actual risk of interaction increases from 16 % with 3 medications to 72 % with 6 medications and 100 % with seven or more (Karas, 1981). Therefore cancer patient’s particular those over 65 years of age are at a high risk for potential drug interactions.

Pharmacotherapy in cancer patients is often expanded and complicated by a patient´s baseline medication profile, OTC drugs and increasingly complementary and alternative medicines. Es- pecially the application of evidence based medicine tends to increase the number of drugs described to treat special diseases. The prescribing physician and the number of prescribing physicians are a main risk factor for polypharmacy in cancer patients (Green et al., 2007, Tamblyn et al., 1996). The layering of polypharmacy with age and radiation related physiological and functional changes and comorbidities increase the risk for potential drug drug interactions. Potential drug interactions and potentially inappropriate medications compromise the health outcome in

radiooncologic patients who are already weekend by paraneoplastcic surgery, toxic treatment regimens and a deprived immune status (Rochon, 2013). Studies which evaluated the prevalence of polypharmacy in geriatric oncology patients are rare. A study by (Riechelmann et al., 2005) revealed that 63 % in this study had at least one potential drug interaction. More than half of the drug interactions detected were classified as moderate and the plurality of patients received eight medications on average.

In addition cancer patients are exposed to mental and physical stresses during radiation therapy and chemotherapy. These therapies often have therapy related side effect’s which have a high influence on the quality of life of these patients. If these patients are additionally at risk for drug drug interactions the outcome of therapy can be influenced negatively. The economic impact of DDIs in medical oncology practice is unknown. It is important to bear in mind in this context that not only serious DDIs leading to hospitalization be evaluated for costs but also minor DDIs, which may make the oncologist repeat blood tests, prescribe medications to treat new symptoms and require more visits to the hospital (Riechelmann and Del Giglio, 2009).

1.3.1 Drug related problems 1.3.1.1 Polypharmacy Polypharmacy derives from two Greek root words: poly, meaning many, and pharmakeia meaning medicines or drugs (Rambhade et al., 2012, Torrey, 2010). Multiple definitions are used for polypharmacy and initially referred to the use of different drugs used at the same time to treat coexisting diseases (Chan et al., 2009, Lees and Chan, 2011, Maggiore et al., 2010).

There are different definitions for polypharmacy found in the literature such as the concurrent use of two or more drugs for more than 240 days (Veehof et al., 2000), use of four or more medications, and use of five and more medications (Maggiore et al., 2010). The term polypharmacy was used in studies performed in the United States to imply the oral intake of unnecessary or redundant medications as well as taking medications which represent a risk for the occurrence of adverse drug events (Ewachiw et al., 2008). In the last years more clinical relevant issues such as specific medications like potentially inappropriate medications (PIMs) defined by the Beers criteria were included in the definition of polypharmacy. PIM defines medications which are not appropriate for a given patient due to age or concurrent illness. Regardless of which definition is applied, polypharmacy seems to have a high prevalence among oncologic patients. (Fulton and Allen, 2005) claimed that the use of a strict definition of polypharmacy (five or more drugs) as a sign for drug related problems in daily clinical practice is not helpful to approach the problem due to the fact that treatment for example in cancer patients with underlying comorbid diseases has a clear benefit and is necessary to ensure survival.

Changes in medication at the transition point from outpatient to inpatient care and back may increase the frequency of drug-related problems such as pDDIs and inappropriate medications (Himmel et al., 1996, Smith et al., 1997, Vonbach, 2007). In this context drug modifications shortly before hospital discharge may be of great importance because the monitoring of patients significantly declines after hospital discharge and adverse drug reactions in the ambulatory care setting are not unmasked as drug drug interactions (Ellitt et al., 2010, Wanrooy, 2011). Accord- ing to Forster et al. 11 % of discharged patients developed an ADE within 24 days (Forster et al., 2005).

Independent prescribing patterns (by general practitioners and oncologic specialists) may increase the risk for drug related problems (Corcoran, 1997, Lees and Chan, 2011). As a consequence different medications to achieve the same therapeutic endpoint might be prescribed. The risk of receiving an inappropriate drug combination is directly related to the number of physicians prescribing for elderly people (Seymour and Routledge, 1998, Tamblyn et al., 1996). Finding the right balance between reducing polypharmacy and still maintaining the quality of life (relief of symptoms and pain) of these patients is the major task.

The possibility for polypharmacy is decreased in case the primary medical record is accessible to each treating physician and the patient gets one physician as main coordinator for medical issues. Patients not enclosing all medicines in their medical histories run at risk that drug interactions or drug side effects are attributed to other causes. A recent study revealed that of 218 cancer patients who were interviewed at the H. Lee Moffitt Cancer Centre & Research Institute in Tampa, Florida, 47 % were taking non-prescription items but did not report these products as medications on a 3-day medication history (Corcoran, 1997). Of the daily non-prescription items, 65.9 % were classified as vitamins and supplements and 25.4 % as analgesics.

1.3.1.2 Prescribing cascade Adverse drug reactions are often treated with additional drugs, leading to a prescribing cascade (Hajjar et al., 2007). The prescribing cascade (Figure 1-2) starts, when an adverse drug reaction is misinterpreted as a new medical condition (Mallet et al., 2007). At first a drug is prescribed and results in the occurrence of an adverse drug effect that is mistakenly diagnosed as a new medical condition. The patient is placed at risk of developing additional adverse effects relating to this potentially unnecessary treatment as soon as a new drug is prescribed (Rochon, 2013, Rochon and Gurwitz, 1997).

Drug 1

Adverse drug effect – misinterpreted as new medical condition

Drug 2

Adverse drug effect

Drug 3

Figure 1-2. Prescribing Cascade according to Rochon et al.1997.

1.3.1.3 Drug related problems All circumstances which potentially or actually impair the optimal result of pharmacotherapies are called problems associated with pharmacotherapy or drug related problems (DRPs) (Krähenbühl-Melcher, 2005). They can be illustrated by the intersections of three circles repre- senting medication errors, ADE and ADRs. Medication errors include every mistake on behalf of the medication process (prescribing, administration of drugs). Only a minority of medication errors result in ADR and ADEs.

Problems that occur even when no mistakes are made within the drug distribution process are called adverse drug reactions (ADRs). (Aronson, 2000) defines an adverse drug reaction as “An appreciably harmful or unpleasant reaction, resulting from an intervention related to the use of a medicinal product, which predicts hazard from future administration and warrants prevention or specific treatment, or alteration of the dosage regimen, or withdrawal of the product” (Aronson, 2000). DRPs due to medication errors such as drug-drug interactions (DDIs) are included in the definition of an ADR.

An injury resulting from the use of a drug is called an adverse drug event (ADE). Under this definition the ADE includes harm caused by the drug (adverse drug reactions and overdoses) and harm from the use of the drug including dose reduction and discontinuation of drug therapy (Chelan-Douglas Health District, 2012). Adverse drug events are any unfavourable medical event that occurs in association with the use of a certain medication, but which is not necessarily causally related to this medication (Krahenbuhl-Melcher and Krahenbuhl, 2005, Leape, 1995). Drug interactions are recently the reason for adverse drug reactions. Studies and reports often conflict in description or evaluation of the clinical significance or relevance of the interaction.

The combination of these two elements makes the identification and management of drug interactions a difficult challenge for the clinician.

1.3.2 Drug drug interactions 1.3.2.1 Classification of drug interactions A differentiation between clinically important and unimportant drug drug interactions is difficult due to the multitude of drugs on the market. Books and software which are used in order to evaluate drug interactions used classification systems for simplification. The most classification systems in the past like Pharmavista® classified interactions based on severity such as ‘severe’(life-threat, permanent harm, discontinuation of the drug), ‘moderate’ (frequent therapeutic problem, close monitoring required when drug combination is administered) or ‘minor’ (decreased or increased drug effect, only specific subgroups affected) (Indermitte et al., 2007, Indermitte, 2006). Other classification schemes classified drugs according to frequency (common, uncommon, rare, and very rare) or based on availability of the ADRs.

In the last 20 years the implementation of drug interaction screening programs revealed that many of the old classification systems based on frequency of interactions, severity and degree of documentation seemed to be problematic in daily clinical practice. Uncertainties in the interpretation of results and the assignment to individual patients gave cause for the development of a new classification system. (Bergk et al., 2004) introduced a management-oriented algorithm according to four decision layers for systemic evaluation of drug interactions. The four decision layers took into account severity, manageability, risk benefit assessment and patient related risk factors (Bergk et al., 2004).

Hansten and Horn initiated the classification of drug interactions according to necessary measures. The so-called ‘Operational Classification for drug interactions’ (ORCA) was developed by the Drug Interaction Foundation with input from an international group of physicians (Hansten and Horn, 2012, Indermitte, 2006). They aimed to improve the clinical utility of classification systems and perceived the deficiencies of the older European systems. This classification helps physicians to decide ultimately on a course of action (or interaction) for each potential drug interaction, giving them clinical assistance on management options that can reduce patient risks. According to the ORCA classification system implemented by Hansten and Horn in 2009 the ABDATA- Pharma-Service (ABDATA-Pharma-Service, 2012) developed a private classification system based on European data. After each single monograph has been revised the new management oriented classification system displaced the old classification system of the ABDATA in January 2009 (Table 1-1).

Table 1-1. Operational Classification of Drug interactions (ORCA) innovated by Hansten and Horn adopted by the ABDA database (Indermitte, 2006).

Category Definition Characterization I Contraindicated Risk of combination outweighs benefit II Contraindicated as a precaution Use under special circumstances: . Interactions for which there are preferable alternatives for one or both drugs . Interactions to avoid unless the benefit is juged outweighs the risk

III Monitoring or adjustment necessary Adopting measures are necessary in case of certain in certain cases conditions: risk factors (renal impairment), high dosage, longer lasting therapy IV Monitoring or adjustment necessary Assess risk and take one of the following actions: . Consider alternatives: alternatives which are less likely to interact . Circumvent (take action to minimize risk of interaction) . Monitor: early detection can minimize risk for an adverse outcome

V Monitor as a precaution The interaction is theoretically possible but until now not documented, or occurred only in isolated cases, whereby nor risk factors have been apparent VI No measures required Risk of adverse event appears small

1.3.2.2 Mechanism of drug interactions Drug drug interactions are the modification of the effect of one drug (the object drug) by the prior administration of another drug (precipitant drug).

Pharmacodynamic interactions

“Pharmacodynamics is defined as what the drug does to the body or the response of the body to the drug” (Ruskin, 2012). Pharmacodynamic interactions generally involve additive, synergistic or anatgonistic effects of drugs acting on the same receptors or physiological systems (Iacobellis, 2006). Pharmacodynamic responses can be affected by disorders including genetic mutations, thyrotoxicosis or malnutrition. These disorders can lead to changes in receptor binding, alter the level of binding proteins, or decrease receptor sensitivity (Moroney, 2012). Through alterations in receptor binding or in post receptor response aging tends to affect pharmacodynamic responses (Ruskin, 2012). Pharmacodynamic drug drug interactions result in competition for receptor binding sites or alter postreceptor response (Moroney, 2012).

Drugs with opposing pharmacological effect

Interactions resulting from the use of 2 drugs with opposing pharmacological effect e.g. in case a diuretic (thiazide) is prescribed for a diabetic patient the hypoglycaemic action of the antidiabetic drug may be counteracted, necessitating a dose adjustment (Leary and Reyes, 1984).

Drugs with similar pharmacological effect

The concurrent use of antipsychotic drugs (e.g promethazine) and a tricyclic antidepressant (amitriptyline) resulting in an additive anticholinergic effect may result in an atropine-like delirium that could be mistakenly interpreted as a worsening of psychiatric symptoms or the presence of delirium

Pharmakokinetic interactions

Pharmacokinetic interactions are more complicated and difficult to predict because the interacting drugs often have unrelated actions. Pharmacokinetic interactions occur when one drug af- fects the other in terms of its absorption, distribution within the body, metabolism or elimination (National Medicines Information Centre, 2008). This result in altered plasma concentrations of the affected drug (decreased or increased) or altered duration of effect (enhanced clearance, metabolism) with the potential for enhanced toxicity or reduced efficacy (Iacobellis, 2006).

Absorption

Drugs need to reach their target to induce a pharmacological effect. Oral delivery of drugs ne- cessitates consideration of various factors which can influence drug absorption and leads to altered bioavailability. The mechanisms of action are the following (Iacobellis, 2006).

. Changes in GI ph value: e.g. dissolution of ketoconazole is reduced due to the increasing stomach ph caused by PPIs, resulting in reduced absorption of ketokonazole. . Chelation / complexing mechanisms: e.g. Tetracyclines can combine with metal ions such as calcium, magnesium, aluminium and iron in the gastrointestinal tract to form complexes that are poorly absorbed. . Altered gastrointestinal function (acceleration or slowing of gastric emptying, mucosal damage due to gastritis, change in vascularity of the mucosa): e.g. metoclopramide enhances motility and accelerates gastric emptying leading to faster absorption of NSAIDs.

Distribution

Distribution interactions involve competition between two drugs for binding sites on tissue or plasma proteins. The blood flow to the target tissue and the binding properties of the drug to plasma proteins (such as albumin, lipoproteins and immunoglobulin) determine drug distribution to the target site. The biologically active fraction of a drug is the unbound drug which can

exert its effect on the pharmacological target. Binding to plasma proteins leads to a limitation of a drugs activity.

Metabolism

Drug interactions could be related to the metabolic interaction at the P-glycoprotein level or to an interference with metabolism at the cytochrome system level (CYP P 450).

Cytochrome P 450

The cytochrome P 450 system plays an important role in the oxidative breakdown of drugs. The cytochrome P 450 enzyme systems comprise a family of isoenzymes with clinical relevance with at least 15 human liver enzymes involved in the metabolism of drugs such as CYP3A4, CYP2D6, CYP2C9, CYP1A2, CYP2C19 and CYP2E1 (Ogu and Maxa, 2000). The activity and capacity of CYP P 450 can be increased (enzyme induction) or decreased (enzyme inhibition) by different drugs.

. Enzyme induction takes place by means of the activation of nuclear receptors. In the cell nucleus an increased transcription rate leads to an increased enzyme production (Remmer, 1972). A maximal inductive effect is reached within 1-2 weeks. During this period drug therapy should be adapted stepwise to the modified CYP-activity and controlled by therapeutic drug monitoring (McAlpine et al., 2011). Per time unit more drugs are degraded as a consequence of hepatic induction, as a result of which a shortened half life and a dimin- ished plasma concentration is reached. . In the case of enzyme inhibition the capacity of enzymes and the degradation of drugs are reduced per time unit. Main courses of enzyme inhibition are competitive inhibition and feedback inhibition. In case of competitive inhibition the enzyme is inhibited by the inactive substrate because it competes with the real substrate for the active site (Clackamas Community College, 2001). In feedback inhibition there is a second binding site on the enzyme. The presence or absence of the inhibitor at this second binding site changes the con- formation of the enzyme so that the active site is made available or unavailable to the substrate. Substances that inhibit CYP metabolism can increase serum concentrations of sub- strates for the inhibited enzyme (Clackamas Community College, 2001).

P- glycoprotein

Recent publications indicate that P-glycoprotein mediates the transcellular transport of many drugs in addition to anticancer compounds (Yu, 1999). Its potential role in clinically significant drug-drug interactions has just begun to be realized (Yu, 1999). P-glycoprotein is an intracellu- lar tissue-specific transport system that belongs to the adenosine 5'-triphosphate (ATP)-binding cassette (ABC) superfamily. By functioning as an efflux pump, P-gp prevents the cellular up- take of toxic or foreign substances, such as xenobiotics (Holtzman et al., 2006). Because P-gp is

ATP dependent, it is able to transport a variety of chemical compounds against a concentration gradient. One mechanism of drug-drug interaction is that the interacting drug inhibits or induces P-gp, thus affecting its ability to serve as a transporter of a substrate drug or affecting the exposure of the drug to metabolizing enzymes (Marchetti et al., 2007).

1.3.2.3 Physiology of aging An elderly patient is at risk for drug related problems, because the physiologic changes that occur due to aging make the body more sensitive to the effects of medications (Ginsberg et al., 2005, Turnheim, 2003). Appreciating the physiologic changes influencing both pharmacokinetic and pharmacodynamic parameters is important. If drug therapy is not adapted these changes may be responsible for the more frequent occurrence of ADRs in the elderly (Figure 1-3). The prevalence of disease is increasing with age in addition to age related physiological changes (McLean and Le Couteur, 2004). As the figure depicts comorbidity is correlated with polypharmacy presenting a risk factor for pDDIs that may result in ADRs (Figure 1-3). Poor adherence to drug treatment increases the risk for pDDIs. Especially in radioncologic therapy different drug regimens increase the risk for poor adherence.

Pharmacokinetic changes

ORGAN FUNCTION Pharmacodynamic IMPAIRED changes

Homeostasis im- ADR paired or

ADE Potentially inappropriate drugs

POLYMORBIDITY Polypharmacy pDDI

Poor adherence

Figure 1-3. Age related impairment of organ function and polymorbidity frequently lead to an increased risk of adverse drug reactions in the elderly. Due to a negative benefit–risk-ratio or due to an exacerbation of underlying diseases some drugs may be inappropriate in elderly patients. ADE =Adverse drug event; ADR = adverse drug reaction (adopted from (Egger, 2007, Egger et al., 2007)

In case of pharmacokinetic drug interactions absorption, distribution, metabolism, or elimination of a drug is altered by another drug (Egger, 2007). Age related changes in body composition, hematologic reserves, and nutritional and disease related changes in plasma protein bind-

ing affect drug distribution and might lead to altering drug concentrations (Browener and Delman, 2008). Gastrointestinal motility, acidity, and mucosal stem cells decline with advancing age affecting the pharmacokinetic and pharmacodynamic properties of a drug (Browener and Delman, 2008). With increasing age the percentage of body fat increases and the total body water decreases due to a decrease in muscle mass. Drugs being lipid-soluble (Thiopental, Tra- zadone) have a longer half life in the body because they can be distributed in a number of fat stores. Serum levels of water soluble drugs go up due to a decreased volume of distribution. Cachexia, oedema and malignant aszites as cancer related symptoms have an additional impact on drug distribution (Hoyumpa and Schenker, 1982).

The activity of the CYP P 450 microsomal enzyme pathways, hepatic clearance and blood flow sustain a loss in function with advancing age (Kinirons and Crome, 1997, Wauthier et al., 2007). As the CYP 3A4 enzyme is critical to the metabolism of drugs, polypharmacy increases the competition for enzyme activity and leads to a decline in metabolisation of certain medications (Fulton and Allen, 2005). Drugs such as tricyclic antidepressants and beta-blockers showing a significant hepatic first pass metabolism may have a faster onset and higher bioavailability. Dose administrations need to be adopted (Kinirons and Crome, 1997, Page et al., 2010).

An age dependant descent in the glomerular filtration rate leads to alterations in drug excretion and a delay in the clearance of drugs. In patients aged ≥ 75 a 50 % decline in the renal clearance is presented. Renal excretion can be impaired by concomitant diseases such as diabetes and hypertension having an additional influence on the cardiac output as well as on tubular absorption and secretion (Ewachiw et al., 2008).

The second area of age related clinical changes are pharmacodynamic changes, primarily affecting the blood brain barrier (Beers, 1999). As a result of changes in the blood brain barrier, drugs having an influence on the CNS may have far greater impact in elderly than on younger patients. A change to the drug receptor site increases or decreases the patient´s sensitivity to certain medications (Mangoni and Jackson, 2004).

1.3.2.4 Radiooncologic treatment Two categories of radiation effect’s can be distinguished. Effects occurring during treatment and within 2-3 weeks after its completion are defined as acute effects. After a period of months or even year’s late effects appear in slowly growing tissues such as lungs, kidney’s and the heart.

The supportive medication needed in radiooncologic patients to reduce side effects increases the risk for potential drug interactions thus patients additionally rely on several other drugs for an- algesia, anaemia, neutropenia and depression. This patient population is predominantly at risk of polypharmacy.

Anaemia is one of the most common blood abnormalities of cancer. According to (Ludwig and Strasser, 2001) cancer patients with low haemoglobin levels do not respond as well to radiotherapy as non anaemic-patients. Potential drug interactions leading to bleeding complications decrease the haemoglobin level additionally and put the patient at risk for negative outcomes during radiation therapy. A study of head and neck cancer patients who received radiation therapy found that stroke rates were five times greater than expected (Dorresteijn et al., 2002, Faloon, 2011). “The average time between radiation treatment and stroke was 10.9 years, but the increased risk of strokes persisted for 15 years after radiation therapy” (Dorresteijn et al., 2002, Faloon, 2011).

Women receiving radiation of the breast with exposure of the heart due to diagnosed breast cancer have been shown to have an increased risk for cardiovascular diseases (Gyenes et al., 1998). Drug interactions with drugs constituted to treat high blood pressure as well as being suitable for stroke prophylaxis might lead to decreased plasma levels of antihypertensive medications, or even inactivation of active compounds. Interactions not being recognized at the time of radiation and especially after discharge from hospital put the patient at risk of experiencing cardiac complications earlier because the baseline therapy is insufficient and the patient is sus- ceptible to negative outcomes.

Nausea and vomiting are two of the most distressing side effects of radiotherapy and cytotoxic drugs, which are currently often combined to treat tumours. Inadequate control of these symptoms may result in significant patient suffering and a decrease in the patient's quality of life, which has been shown to decrease patient’s compliance to treatment, with a potential impact on disease outcome (Horiot, 2004). 70 % of patients treated for cancer of the prostate and other malignancies in the pelvic region suffer from acute inflammatory small intestine changes (Resbeut et al., 1997). Gastrointestinal mucositis characterised by diarrhoea, abdominal pain, results in impaired absorption. Later on increased plasma clearance accompanied by an increased volume of distribution, thus leading to further prolongation of elimination half life. The volume of distribution may be decreased due to dehydration (Haas, 2004). Apparent changes in volume of distribution represent changes in bioavailability of a drug in case of oral administration.

1.3.2.5 Risk factors for drug interactions Well-known risk factors for developing drug interactions are known in daily clinical practice. Co-prescription with drugs which have a narrow therapeutic index (were a small margin exists between therapeutic and toxic drug levels) may lead to clinically significant drug interactions e.g. digoxin, warfarin, ciclosporin, aminoglycosides. Drugs with a steep dose-response curve (i.e. a small rise in drug levels leads to a large increase in effect) are very likely to cause DDIs (National Medicines Information Centre, 2008).

In the presence of a drug interaction drugs with a saturable metabolism are more likely to accu- mulate. Drugs with these drug characteristics are phenytoin and theophylline (National Medicines Information Centre, 2008). Apart from drug characteristics associated with a high risk for DDIs patient characteristics have been identified which lead to an increased risk for DDIs (Bjerrum et al., 2008). A major risk factor for drug interactions is increasing age (Corcoran, 1997). In addition to age-related physiological changes the prevalence of disease is also increasing with age (Egger et al., 2007). The presence of multiple diseases is correlated with polypharmacy, which is a risk factor for DDIs that may result in ADRs. Polypharmacy (concomitant use of > 5 drugs) is often necessary to properly manage chronic diseases (National Medicines Information Centre, 2008). The risk of DDIs in patients taken 2-5 drugs has been estimated to be 19 %, but the risk arises to 80 % for those taking > 6 drugs (National Medicines Information Centre, 2008).

The complex nature of parenteral and enteral nutrition increases the risk for serious drug- nutrient interactions. Unpredictable side effects and incompatibilities are caused by the exist- ence of various drug dosage forms, metabolites, and inactive ingredients. Nutrients and medications occasionally share metabolic pathways, thus changes in dietary composition can influence hepatic metabolism or renal clearance (Sacks, 2004).

The clearance of certain hepatically-cleared agents like propranolol is accelerated by high protein diets (Sacks, 2004). Patients with poor nutritional intake (i.e., low protein consumption) may be more likely to experience adverse effects of renally eliminated drugs (Mayhew and Christensen, 1993). In the presence of protein-calory malnutrition, decreased renal clearance of penicillins and methotrexate occurred which increased the risk for drug toxicity. Low fat mass and edema may interfere with transdermal medication delivery (e.g. fentanyl patch) and require an alternative route of medication administration (Sacks, 2004). Clinicians must consider basing their dose calculations on an expanded weight adjusted volume of distribution in order to obtain therapeutic serum drug concentrations.

Although there do not exist definitive studies about pharmacokinetic changes in oncologic patients it might be assumed that the pharmacokinetic properties of a drug might be distorted due to impaired drug absorption, mucosites, altered excretion in patients due to renal or hepatic impairment and changes in plasma distribution resulting from oedema, dehydration and malnutrition (Riechelmann and Del Giglio, 2009). All these changes can increase the risk for potential drug interactions because of the altered pharmacokinetic properties.

1.3.2.6 Factors minimizing potential for drug interactions According to (Hansten and Horn, 2012) many drug interactions have identifiable factors that mitigate potential drug interactions and render it unlikely to produce adverse consequences. Identifying these factors early in the medication process can save considerable time and helps to

disregard the potential drug interaction from further considerations. Published literature in clinical pharmacology underlines important issues in the medication process.

Dosage of drugs

The dose of one drug or both drugs can be so low to make the risk of an adverse interaction minimal (Horn and Hansten, 2009).

Duration of therapy

Sometimes a drug is not given long enough to cause adverse interactions.

Dosing times

In case of gastrointestinal absorption interactions might not occur between drugs because time intervals between administration have been long enough- antacids and concurrent medication 2- 4 hour-time interval.

Order of administration

Interactions are decreased because the sequence of administration of the drugs mitigates the interaction (Horn and Hansten, 2009)

Pharmacogenetics

Patient characteristics concerning pharmacogenetic properties are a useful parameter to forecast DDIs (Iacobellis, 2006).

Monitoring

Adverse outcomes can be avoided by carefully monitoring the patient. If a patient is started on a therapy of digoxin monitoring of elevating digitoxin levels is of utmost importance in case an additional drug with P-glycoprotein inhibition is administered additionally (Horn and Hansten, 2009).

1.3.2.7 Clinical Relevance of drug interactions (potential drug interactions) (Haddad et al., 2007) pointed out that the exposure to DDIs was associated with a significantly increased risk of hospitalization. As reported by the U.S. food and drug administration 2 million hospitalized patients in the U.S. suffer from serious adverse drug reactions each year, attributa- ble to polypharmacy. (Lepori et al., 1999) showed that 21 % of all drug-related hospital admissions in a Swiss hospital were caused by DDIs. A study by (Rivkin, 2007) reported that more than half of all ADRs in the ICU resulted from DDIs and concluded that more than 100 % of DDIs could be prevented by appropriate drug management.

The first drug interaction was documented 1946 in the USA (Zagermann-Muncke and P., 2009). The Tuberkustatikum p-Aminobenzoicacid reduced the renal elimination of salicylates and in-

creased its plasma concentration. After having noticed the implication of drug interactions for the healthcare system first scientific studies have been performed. Broad studies have been performed on potential drug interactions in settings of internal medicine (Egger et al., 2007, Köhler, 2001, Vonbach, 2007). The prevalence of pDDIs ranged between 40-65 %. Drug interactions are a neglected issue in oncology. Due to a large number of medications that cancer patients receive (radiotherapy supportive care agents and drugs to treat comorbid illnesses) it is of prime importance to address drug interactions in this population. More than a quarter of cancer patients are at risk of potentially serious drug interactions according to a recent study in the Journal of the National Cancer Institute (Riechelmann et al., 2007). First studies in this field have been performed by (Riechelmann et al., 2005). The proportion of potential adverse events turning into clinical consequences is unknown. (Miranda et al., 2011) sought to evaluate how many hospital admissions in oncology were due to drug-drug interactions (DDI) or adverse drug reactions (ADR). Each hospitalization was evaluated by two blinded investigators who classified reasons for hospital admission by their probability to be associated with either a DDI or an adverse drug reaction. Among unplanned admissions (n = 298), 39 (13.0 %, 95 % CI 9.4 - 17.4 %) were considered to be associated with an adverse drug event: 33 (11.0 %, 95 % CI 7.7 - 15.2 %) were associated with an ADR and 6 (2 %, 95 % CI 0.7 - 4.3 %) with a DDI (Miranda et al., 2011).

(Riechelmann et al., 2005) evaluated the frequency of potential drug interactions among 100 consecutive patients with solid tumours and haematological malignancies. The study revealed that 63 % of patients were exposed to at least one potential drug interaction. The average number of drugs prescribed to a cancer patient was 5, with a range of 0-23 (Riechelmann et al., 2005). In a second study the frequency of drug combinations leading to potential drug interactions has been investigated in ambulatory patients with solid tumours receiving standard cancer directed therapy (Riechelmann et al., 2007). According to (Riechelmann et al., 2007) 276 potential drug interactions were identified in 109 (27 %) out of 405 patients. This study revealed that the majority of drug interactions took part between the drugs used to treat comorbid illnesses such as blood pressure and diabetes or those medications interacting with the medications for supportive cancer care. Medications like warfarin (used to prevent blood clots), anti- hypertension drugs, aspirin and anticonvulsants were the drugs that most commonly interacted with cancer medications. The majority of potential interactions involved non-anticancer agents (87 %), and were classified as either major or moderate (86 %) and had good evidence in the literature (Riechelmann et al., 2007).

A retrospective study examined the frequency of potential DDIs among patients with advanced cancer who were receiving supportive care exclusively (Riechelmann et al., 2008). They found that one-third (29 %) of patients were exposed to drug combinations with the potential to inter-

act and that most potential drug interactions involved common medications used to treat comorbid illnesses (Riechelmann et al., 2008).

(Haddad et al., 2007) proposed that understanding the mechanism of drug interactions, avoid- ance of dangerous drug combinations and the substitution for safer alternatives could potentially decrease the risk for drug interactions. In oncologic patients cytochrome CYP3A4 accounts for a great risk for potential drug interactions (Haddad et al., 2007). Common medications used in palliative medicine (dexamethasone, prednisone, midazolam, triazolam, fentanyl, haloperidol) are substrate drugs for CYP 3A4. Certain anti-seizures drugs, selective serotonin receptor inhibitors, macrolides and azoles up regulate or inhibit CYP3A4 activity. Understanding drug pharmacokinetics and interactions through CYP3A4 is important to avoid adverse outcomes in oncologic patients (Haddad et al., 2007).

(Buajordet et al., 1995) tried to investigate fatal adverse events by means of aggregated medical records, autopsies and pre and post-mortem drug analyses. The incidences of fatal adverse drug events were 18.2 %. Mainly those drugs used for treating comorbid pulmonary diseases, an- tithrombotic drugs and drugs for treating coronary heart disease and heart failure were suspected to cause or contribute to fatal outcomes.

In 2011 (Bosch et al., 2006, Espinosa-Bosch et al., 2012) performed a literature review (January 1990- September 2008) in order to study the prevalence of drug interactions in hospital healthcare. The search generated 436 articles out of which 47 were included into the study. 42 articles focused on potential drug interactions. 3 articles investigated real drug interactions with clinical consequences and 2 described both. (Espinosa-Bosch et al., 2012) found out that a large number of studies focussing on the prevalence of potential drug interactions in hospitals report widely varying results. In patients with heart diseases and elderly people the prevalence of pDDIs is higher (Espinosa-Bosch et al., 2012). The highest prevalence was found in discharged patients higher than in outpatients or patients treated in emergency departments (Forster et al., 2005, Triller et al., 2005).

A recent study from 2012 performed by (Bucsa et al., 2013) assessed the prevalence of pDDIs. Furthermore (Bucsa et al., 2013) studied the prevalence of real DDIs which resulted in clinical visible adverse drug reactions. A number of 478 pDDIs were found at hospital admission in 161 patients while the number rose to 801 in 217 patients at hospital discharge. Only fourteen drug interactions led to 13 clinical identifiable ADRs. (Bucsa et al., 2013) identified a statistically significant association between ADRs related to DDIs and a correlation between the lengths of hospital stay in the presence of medications belonging to the group of cardiovascular medications.

(Frechen et al., 2012) performed a study with the primary aim to assess the potential for drug- drug interactions in dying patients by identifying drug combinations and risk factors associated

with a high risk of pDDIs. In two hospices charts of 364 dying patients were reviewed retro- spectively. Drugs which have been prescribed during the last 2 weeks of life were screened for pDDIs by the electronic database of the Federal Union of German Associations of pharmacists, which classified pDDIs by therapeutic measures required to reduce possible adverse events according to the ORCA system (operational classification of Drug interactions). This study used the same database like our study. 223 patients in the study had potential pDDIs (61 %). The drugs most frequently prescribed were antipsychotics, antiemetics (e.g. metoclopramide, antihistamines) antidepressants, and cardiovascular drugs, NSAIDs, glucocorticoides and insulin. The study revealed that the most prevalent potential adverse effects constituted of additive anticholinergic, antidopaminergic and cardiac effects (QT prolongation).

A recent study by (Lea et al., 2013) investigated the prevalence of potential drug interactions in an acute geriatric setting. Pharmacists identified 245 pDDIs of major (n = 13) or moderate (n = 232) severity in 80 (63.5 %) of the 126 patients included on admission and/or during hospitalization. “In 94 of 162 cases where the pharmacists alerted the geriatricians (58.0 %), prescribing changes or monitoring actions were implemented” (Lea et al., 2013). For 22 % of patients a causal relationship between hospitalisation and DDIs was assessed as possible. A study by (Hasan et al., 2012) showed that 64 (15.9 %) pDDIs were clinically significant.

(Lao et al., 2013) performed a study in 2013 to evaluate the prevalence of PIM use and pDDIs among elderly (> 65y) nursing home residents in Macao. PIM use was determined by using the screening tool of older person´s prescription (STOPP) criteria and pDDIs were detected using the preset criteria of two compendia, Drug Reax and Lexi-Interact. Only pDDIs occurring in both systems were included into the study: 46.5 % of patients used one or more PIMs. The prevalence of pDDIs was 37.8 % among the 111 patients included into the study. An increased number of drugs was identified as an independent risk factor associated with PIM use and pDDIs (p < 0.05) (Lao et al., 2013).

1.3.3 Potential inappropriate medications (PIMs) The definition of potentially inappropriate medications (PIM) includes several elements. The term potentially inappropriate prescription encompasses when a medication´s use introduces a significant risk of an adverse drug event when a potentially equal or more effective medication with a potentially lower-risk profile exists (Page et al., 2010). The use of a drug is potentially inappropriate

. that has the wrong indication . that has no indication . that has a high risk of Adverse Drug Reaction (ADR) i.e. adverse drug-drug or drug- disease interactions or Adverse Drug Event (ADE) . for too short or too long a time period

In 1991 gerontologist mark H. Beers has been advocating the use of explicit criteria for identifying inappropriate use of medications (Beers et al., 1991). He summarized these medications on the Beer´s list. Because of age-related physiological changes and impaired homeostatic mechanisms drugs on the Beer´s list are associated with a higher risk for ADRs due to alterations of their pharmacokinetic and/or pharmacodynamic properties. The Beer´s list has been used to evaluate clinical drug use and to study the effect of intervention on reducing adverse events in the past (Doubova et al., 2007, Zhan et al., 2005). An updated Beer´s list was developed for all older patients (Fick et al., 2003). The disadvantages of the Beers list were that more than 50 % of the drugs listed were not available in Europe and that some issues like duplicate prescribing, drug drug interactions were neglected. (Zhan et al., 2005) categorized the drugs as drugs that should be avoided, were rarely appropriate or needed special supervision. The association between the use of drugs on the Beer’s list and healthcare outcomes has been examined in retrospective cohort studies (Fick et al., 2008).

A systematic review of these retrospective studies revealed an association between the use of drugs from the Beer’s list and hospitalization in elderly but couldn’t be statistically proven so far. Further prospective studies with focus on health outcome such as mortality should be evaluated (Hanlon et al., 2006).

Due to the perceived deficiencies of Beers criteria Gallagher et al. developed a new set of PIM criteria in older people, called STOPP criteria (Screening Tool of Older Person’s potentially inappropriate prescriptions) (O'Mahony and Gallagher, 2008). STOPP criteria are based on commonly encountered PIMs. They are listed according to physiological systems and include 65 instances of important PIMs that predispose to ADRs in older people (Hamilton et al., 2011, O'Mahony and Gallagher, 2008).

Recent studies showed that clinically significant ADRs were caused by PIMs listed in STOPP criteria 2.54 times more often than by PIMs listed on the Beer’s list. The risk of a severe, avoidable ADE was increased significantly with STOPP medications (OR=1.85, 95 % CI 1.51 - 2.26, p < 0.001). The risk of a severe, avoidable ADE was not increased significantly with Beer`s medications (OR=1.28, 95 % CI 0.94 - 1.72, p = 0.11) (O'Mahony and Gallagher, 2008). STOPP criteria are already implemented in guidelines managing prescription behavior for elderly people (Deutsche Gesellschaft für Allgemeinmedizin und Familienmedizin (DEGAM) and Leitliniengruppe Hessen, 2013) in general medicine.

A number of PIM lists already existed for countries other than Germany based on the Beer’s criteria and the STOPP criteria (STOPP criteria, 2011). Due to differences between countries with respect to drug approval, prescribing practices and treatment guidelines these lists cannot be directly transferred to Germany. In 2009/2010 experts created the Priscus list for Germany (Holt et al., 2010, Schubert et al., 2013), proceeding on the basis of a literature search and a

qualitative analysis of existing PIM lists from other countries, judged 83 medications to be potentially inappropriate for elderly. The Priscus list can be seen as an important aid in medical decision-making and can be used as a quality measure (Projektverband Priscus, 2011). There are times when the drugs listed on the Priscus list can be administered but they are never first line therapy and administration should be closely monitored for potential adverse events with documentation reflecting the patient’s response to therapy.

The application of the Priscus criteria and other tools for identifying PIM use will continue to enable physicians to plan interventions for decreasing both drug related costs and thus minimize drug related problems (Fick et al., 2003). Since March 2012 health insurances (e.g. Techniker health insurance) publish data about the prescription figures of “Priscus list medications” in patient profiles in their monthly drug health report. A statistical evaluation of the Techniker health care insurance revealed that every fourth patient received in 2011 medications belonging to the Priscus list (Techniker Krankenkasse - Pressestelle, 2012).

According to recent guidelines (Deutsche Gesellschaft für Allgemeinmedizin und Familienmedizin (DEGAM) and Leitliniengruppe Hessen, 2013) the Priscus list and the STOPP criteria are valuable tools in routine clinical practice to evaluate potentially inappropriate drugs and should be used in combination.

In this thesis the Cave module was used in addition to the Priscus list to detect inappropriate medications in the elderly. The Cave module is a new supplementary module implemented in the ABDA database which offers key data for automatic patient specific drug risk checks in the distribution of drugs focusing on drug disease interactions and age and gender inappropriate medications (ABDATA Pharma-Daten-Service, 2007). In this program personal properties of patients are deposited in the database (Age, gender, allergies and comorbid illnesses) and matched with the medications administered.

According to the ABDA, the data taken as a basis for the CAVE module were developed by studying the summary of product characteristics of each medication, evaluating new studies on the market and enquiring literature sources. The Beers list and the STOPP criteria have not been taken as a basis. More detailed information is presented in Annex A1.

1.3.4 Medicine safety management The dominant paradigm for analyzing medical errors and patient safety incidents has become Reason’s Swiss cheese model (Horn and Hansten, 2004, Perneger, 2005).

Risks Prescriber’s Knowledge Computer DEFENCES Screening

Pharmacist‘s Knowledge e.g. Different prescri- bers Patient Risk Factors

e.g. Inadequate evaluation Drug Admi- nistration e.g. Inadequate training Monitoring

e.g. Lack of supervision

GAPS e.g Inapropriate medication

e.g. Poor monitorization

Adverse Event

Figure 1-4. The Swiss cheese model adopted from (Perneger et al. 2005).

The image of “Swiss cheese” was suggested by James Reason to illustrate the occurrence of system failures, such as medical mishaps. According to this metaphor, a series of barriers help to prevent damage and harm of patients (Figure 1-4). Each barrier consists of inadvertent deficiencies and leaks comparable to the appearance of Swiss cheese (Perneger, 2005). These leaks are inconstant e.g., the holes open and close at random. Perfect systems do not exist and the holes in the Swiss cheese represent gaps in the defences (Horn and Hansten, 2004).

The Swiss cheese model has been adapted to the problem of drug interactions. The initiating effect is a drug interaction between drug A and drug B. In an instance the potential drug interaction is detected by the pharmacist / physician and the medication is not dispensed. In the other instance the error trajectory passes through all the safety barriers and the patient suffers from an adverse event. The adverse event resulting e.g. from a drug drug interaction could have been prevented by improvements in various aspects of the medicine management process (Avery et al., 2012). This improvement process can take part on different levels:

. Physician knowledge of DDIs . Computerized order entry . Pharmacist´s knowledge of DDIs

. Pharmacy Computer Screening Programs (Malone et al., 2003) . Proactive intervention approaches to reduce medication-related adverse events have yielded promising results. Educational outreach (Avorn and Soumerai, 1983), use of computerised prompts and active intervention by pharmacists (Zermansky et al., 2001, Zermansky et al., 2002) have all been shown to have beneficial effects upon the prescribing behaviour. Drug interactions randomly result in clinical manifestations. Coordinated sequence of drug administration, dose adjustment and monitoring of drug therapy reduce the risk for drug interactions.

Reviews have supported the potential of CPOE (computerized physician order entry) and CDSS (clinical decision support system) in improving physician performance and in reducing medication errors especially when targeting high-risk populations and high-risk drugs (Cresswell et al., 2007). A study found CPOE and CDSS to be effective in reducing prescribing rates of contraindicated medication in a sample of elderly outpatients (Cresswell et al., 2007, Smith et al., 1997). Advantages include computerized warnings of possible contraindications (e.g. the use of corti- sone in diabetes), drug–drug interactions (e.g.), a systematic way of data entry as well as inte- gration of the prescription with the patient's history (e.g. allergies, comorbidities) (Cresswell et al., 2007).

1.3.5 Documentation of drug interactions and inappropriate medications Important tools for the documentation of drug-related problems and following interventions are coding systems. They should be suitable for the broader implementation of Pharmaceutical Care in hospital and pharmacies. The working conference of the Pharmaceutical Care Network Eu- rope constructed a classification scheme for drug related problems (DRP) (PCNE V6.2) (Pharmaceutical Care Network Europe Foundation, 2010). A drug related problem is defined as an event or circumstance involving drug therapy that actually or potentially interferes with de- sired health outcomes (Pharmaceutical Care Network Europe Foundation, 2010).

According to this definition a drug interaction can be considered to be potential in the constella- tion of a patient’s drug therapy or manifest when leading to an adverse event (Indermitte, 2006). DRPs were classified according to a PCNE system which defined the drug-related issues as lack of treatment effect (P1 of PCNE version 6.2), adverse drug event (P2) and provision of information (Indermitte, 2006). Underlying causes of a lack of treatment effect or an adverse drug event were classified as resulting from an interaction (inappropriate combination; C1.3), or inappropriate drug, inclusively contraindication (C1.1). Drug interactions have been listed separately in the problem section as P5 until PCNE version 6.1. It has been distinguished in P 5.1 Potential interaction and P 2. Manifest interaction.

The classification should help health care professionals to document DRP information in the pharmaceutical care process and can be used as a process indicator in experimental studies of

Pharmaceutical Care outcomes (Zaman Huri and Fun Wee, 2013). Four items are at least attributed to each observation: (a) coding for the problem itself, (b) the actual or suspected cause(s) of the problem, (c) the intervention(s) required to resolve the DRP and (d) its outcome (Eichenberger et al., 2010). As the PCNE has been created for documentation of DRPs in public pharmacies certain items are lacking for inpatients. DRPs such as incompatibilities, application errors or faulty transcriptions cannot be coded satisfactory. (Eichenberger et al., 2010, Lampert et al., 2008) used this classification in an hospital setting and implemented an additional Cate- gory P 7 technical DRPs. Technical DRPs are related to prescription quality and impedes to unambiguously dispense a drug in the correct dose, dosage form and package size (e.g. unread- able prescription, missing specifications). (Lampert et al., 2008) considered the PCNE system to be a practical tool in the hospital setting, which demonstrates the values of a clinical pharmacy service in terms of identifying and reducing DRPs and also has the potential to reduce prescribing costs.

The Pharmacovigilance Division continuously informs about adverse drug reactions and interactions related to the use of medicinal products which have become known and ensure that patients and medical doctors are made aware of existing risks as well as of possibilities to decrease them (BfArM, 2006). If an unexpected reaction is observed in a patient it might be difficult to establish it’s causality and if it has resulted from a drug-drug interaction. It is important to consider possible causes for the reaction such as concomitant medications and patient’s underlying disease and the nature of the reaction, the timing of the reaction in relation to drug administration and the relationship to the dose administered in order to assess causality. In the radiooncological setting a causality assessment of adverse drug reactions is very difficult. In case an adverse reaction induced by a DDI takes place in a patient undergoing radiotherapy the causality assessment is difficult. The onset of medication is unclear. Therefore the timing of the reaction in relation to drug administration cannot be defined. Radiooncological treatment has an influence on lab results of patients. Changes in lab values can either be attributed to ongoing therapy or to a possible drug interactions or inappropriate medications. This dilemma shows how difficult it is to define manifest drug interactions under radiation treatment.

1.3.6 Management of drug interactions and inappropriate medications Drug interaction detection involves awareness of the problem, adequate medication records, and access to reference materials. The manifestations of drug interactions are often quite varied and can represent anything from a minor to a major clinical problem. Memorising all potential drug interactions that have been identified up to date and new interacting drug pairs which are identified every month is nearly impossible. Drug interaction compendia in the form of books, computer or personal digital assistant (PDA) software or online databases are offered to pharmacists and doctors to cope with this task. Studies in the past reference the US-database by Thompson

Micromedex or the British Stockley’s drug interactions compendia which can be considered as standard referenced information sources. Apart from this the SPC (summary of product characteristics) claims to be the basis for information for health professionals on how to use drugs safely and effectively. Since 1980 sophisticated computerized systems for maintaining medication records and for detecting drug interaction problems are already in use in many hospitals and community-based pharmacies in Germany. In Germany a drug interaction database is implemented in the drug information system Pharmatechnik. It is adapted from the German AB- DATA database which is used in all German pharmacies. The usefulness of computerized screening methods depend on the quality, including proper validation, of data held in the system (Mateti et al., 2011). (Blix et al., 2008) compared two methods for identification of drug interactions-computerized screening by means of a drug interaction software and prospective bedside recording with regard to identify drug drug interactions. Among 827 patients, computer screening identified DDIs in 544 patients, whereas only in 73 patients DDIs were found by bedside recording (Blix et al., 2008). Computer screening seems to overestimate concerning clinical relevant DDIs. Computerized drug interaction screening detects a large number of drug-drug interactions of questionable significance. Studies revealed that many doctors and pharmacists ignore most drug interaction alerts provided by the computer screening (Mille et al., 2008). (Abarca et al., 2006) concluded in his study that quality improvement efforts should focus on improving the performance of these systems in flagging DDIs with a high probability of true positive adverse clinical effects and ignoring DDI that have a high probability of having no adverse clinical effect. The deficiency of computer systems is the lack of data concerning the date and time of administration (Mille et al., 2008). Computer screenings are based on a list of active ingredients prescribed rather than strict periods of overlap in the administration of different drugs (Mille et al., 2008). Therefore it became obvious that it would be of utmost importance to develop monographs within the database which are classified according to the appropriate clinical management necessary to avoid the interaction or reduce the risk (operational classification) (Hansten et al., 2001). Monographs should include short information on the interacting substance group, significance of interaction, type of interaction and detailed information on effect, mechanism of action and measures needed to minimize the risk. Recent studies showed that the inclusion and severity grading of pDDIs was largely inconsistent with use of different drug compendia (Vonbach et al., 2008).

A recent study in Germany showed a new technique helping to prevent drug interactions and adverse drug reactions and to improve patient’s compliance. In the clinical setting as well as in public pharmacies the Brown Bag method has been established. A public pharmacy (Schlenk, 2012) in Bavaria offered every patient a bag with a questionnaire on the back focusing on chronic medications, over the counter products and dietary supplements. The patients were en- couraged to put all their medications in the bag, completing the medication registration sheet

and handing it to the pharmacy store. The pharmacy performed a drug interaction screening by means of the ABDA drug interaction software. The brown bag approach has already been used in a study performed by (Colt and Shapiro, 1989). Self reporting of medications helps to monitor medication compliance, preventing drug interactions and reducing polypharmacy. The study revealed that medications taken on a long term basis were mentioned more frequently than medications taken on an as needed basis and as short term medication. A recent study using the Brown Bag method in public pharmacies (Schlenk, 2012) reported that every sixth patient in Bavaria was affected by pDDIs. Especially in radiooncologic patients supportive medication is suspected to present a risk factor for continuous drug interactions (Caskie and Willis, 2004) because the medications are taken on an as needed basis and are frequently not recognized by the patients.

2 Aim of the thesis

The study aims at analysing the prevalence of potential drug drug interactions (pDDIs) and potentially inappropriate medications (PIMs) in patients in the radiooncology setting. Cancer patients are a growing population with a high risk for adverse effects resulting from medical treatment. In the scope of risk assessment this research is performed as a preliminary retrospective study including 120 patients to both evaluate the clinical relevance of pDDIs and PIMs in the radiooncolgic setting and give preliminary recommendations for daily clinical practice as well as for future studies. The frequency of pDDIs and PIMs at hospital admission and discharge is assessed by reviewing and screening the medication profiles listed in the discharge letters in patients receiving radiooncologic treatment in the Marienhospital Herne, Germany. The patterns of pDDIs and PIMs of this study assessed by the ABDA drug interaction software Pharmatechik®, the Cave module and the Priscus list should be compared to the results of extensive studies performed in other settings e.g internal medicine and emergency departments.

The objectives of the retrospective study in terms of pDDIs can be summarized as follows:

. Identifying the prevalence and the risk for pDDIs in the admission and discharge medication profile using the ABDA drug interaction software Pharmatechnik® . Identifying the most frequent drug classes involved in pDDIs . Evaluating risk factors for pDDIs in patients by assessing a possible correlation between the Charlson comorbidity index (CCI) and the occurrence of pDDIs as well as by assessing the correlation between variables like number of medications, age and comorbidities and the occurrence of pDDIs . Evaluating whether the subjective well being of patients (symptoms of discomfort and deterioration of health) assessed by using the Karnofsky index score at hospital admission and discharge might generally allow to seek hints for being harmed by DDIs in future studies (although no manifest DDIs are identified) . Evaluating whether an additional onset of supportive medications during radiotherapy increases the risk for pDDIs . Determining the prevalence of pDDIs caused by chronic-chronic and chronic-supportive medications . Determining the risk of frequently used medications causing pDDIs by using the Flag score method (Frechen et al. 2012) with particular regard to supportive medications and cardiovascular medications

The objectives of the retrospective study in terms of PIMs can be summarized as follows:

. Identifying the prevalence and the risk for PIMs in the admission and discharge medication profile using the Cave module (individual inappropriate medications) and the Priscus list . Evaluating risk factors of PIMs in patients by assessing the correlation between variables like gender, age, number of medications and the occurrence of PIMs detected via the Cave module . Identifying the most frequent drug classes involved in drug disease interactions using the Cave module . Identifying the most frequently prescribed drugs from the Priscus list

3 Material und methods

3.1 Material, records The study was a retrospective study conducted at the Department of Radiooncology at the Ma- rienhospital, Herne, Germany in a twelve-month period between November 1st, 2010 and No- vember 1st, 2011. The hospital is a 475-bed teaching institution providing care to a population of approximately 150000 inhabitants.

3.2 Patients informed consent The data were stored expressively in an anonymized manner and evaluated statistically. Due to the fact that the data collection took place within the scope of quality assurance, the informed consent of each single patient was not considered to be necessary.

3.3 Study population The selection of patient’s records has been made by chance. Recorded were all patients who had been on the wards fulfilling the following eligibility criteria at the time of data assessment.

Inclusion criteria

. Patients being treated in the Department of Radiooncology in the time of November 1st, 2010 and November 1st, 2011 . Patients aged >18 who had been prescribed two or more medications in the admission or discharge medication had been enrolled into the study

120 out of 172 patients (69.8 %) had been included into the evaluation because all inclusion criteria were fulfilled.

Exclusion criteria

. Patients falling under the age limit of 18 . Incomplete patient documentation in the discharge letter for medications or incomplete history of comorbidities . Patients who died before hospital discharge during the hospital stay

52 patients could not be enrolled into the study because they did not meet the inclusion criteria. 3 patients were aged under 18 and were excluded from the study. 19 patients were excluded because of insufficient information about medications in the discharge medication and 10 pa-

tients because of an incomplete medical history. Regarding the analysis of prescriptions at hospital discharge, 20 patients were excluded because these patients died during hospitalization.

3.4 Study design Figure 3-1 illustrates the data collection flow and analysis methods throughout the study.

Review of discharge letters

PRISCUS list ABDA screening

CAVE module Drug interaction software

Extraction of data and determination of variables

Statistical analysis

Critical evaluation and interpretation

Recommendations for clinical practice

Figure 3-1. Study design.

3.5 Data ascertainment Data collection parameters included (in alphabetical order):

. Admission diagnosis . Age . Comorbidities were maintained from the patient’s discharge letter. Patient’s drug regimen and disease states were analysed to profile the study cohort and record their comorbidities and associated pharmacotherapy . Gender . General condition (Karnofsky Index, Charlson Comorbidity index) . Main cancer diagnosis (according to the international classification of diseases, 10th revision (ICD-10)),

. Medication profile: Admission and discharge medication were evaluated separately; each medication was assigned an eight- digit drug code using the international classification of the WHO, the ATC Code.

Patient drug profiles at hospital admission and discharge were screened by the newest version of the ABDA interactive drug interaction software, a drug interaction program that is used predominantly in public pharmacies to predict pDDIs (potential drug interactions) (detailed information Annex A1). The Pharmatechnic © interaction analysis is based on the interaction databases produced by the Federal Union of German Associations of Pharmacists (ABDATA- Pharma-Service, 2012, Lendrath, 2009). The program was chosen as a result of the evaluation of frequently used drug interaction screening programs. (Vonbach et al., 2008) recommended Pharmavista © as the program with the highest sensitivity for detecting pDDIs, for its high negative and positive predictive values. Pharmavista© uses the same ABDA database as information source as Pharmatechnik, the program we used in our study (Lendrath, 2009). In addition other sources of information such as Hansten and Horn drug interactions Analysis and Management 2012, as well as monographies published by the Bundesverband der Phar- mazeutischen Industrie e.v. and the “Rote Liste” were used.

3.5.1 Cave module As a second task the medications on the discharge letters were screened by the Cave module. “Cave is a new supplementary module for the ABDA database which offers key data for automatic patient specific drug risk checks in the distribution of drugs” (ABDATA Pharma-Daten- Service, 2007). By means of the Cave module patient specific risk factors can be implemented in drug therapy. The patient represents the central figure, and drug therapy is adjusted to his individual risk constellations. The Cave data are provided by the ABDATA Pharma Service and implemented by the different software providers (Pharmatechnik®).

In case the Cave module detects a drug-disease contraindication, it generates one of two mes- sages indicating severity levels: absolute contraindication (CI), restriction on use (ROU). The figure in the annex A1 shows how the Cave module is generated and gives detailed information on the system.

3.5.2 Steps of data extraction The data extraction consisted of six main steps:

As a first step in data acquisitation a customer card for every patient has been created in an anonymous form in the database consisting of the patient’s initials of first and last name. Age and gender were additional relevant data which were saved on the customer card. The card con- taining individual patient data was continuously deposited in the database and it could be referred to these data with access at any time.

In a second step individual patient characteristics, in particular comorbidities and allergies were stored additionally on the customer card by means of the CAVE module. Comorbidities, which might have an influence on drug therapy such as asthma, diabetes, hypertension, renal dysfunction were considered in total up to 214 eligible diseases. As a basis the indexing of patient’s risk factors by means of MIV codes (medication relevant parameters) was used. MIV codes provided the coding of diseases and symptoms and considered special circumstances of life such as pregnancy, smoking. Additionally the duration of each single disease were specified by lodging the status in the database. The status was either static or temporarily. After having implemented all relevant basic data into the database, the installation of a new customer card was completed.

In a third step the entrance medication was implemented into the database. A drug risk analysis was started. Three main groups of potential risk factors were checked by the Cave module: Gender related indication; inappropriate medications with regard to underlying diseases, as well as invalid age ranges. Types and degrees of seriousness were listed. This classification helps to decide ultimately on a course of action for each potential inappropriate medication giving clinical assistance on management options that can reduce patient’s risks.

In a fourth step a drug interaction screening with the ABDA Drug interaction database was performed separately for the entrance and discharge medication. This software runs in parallel to Cave as soon as the patient’s individual medication therapy is implemented into the database.

In a fifth step the drug interactions found with Pharmatechnik © ABDA interaction analysis were entered into SPSS and a separate EXCEL work sheet with the following information was evaluated : total number of drug interactions, distinction of drug interactions according to severity rating:

1. total number of drug interactions Category I “Contraindicated” 2. total number of drug interactions Category II “Contraindicated as a precaution” 3. total number of drug interactions Category III “Monitoring or adjustment necessary in certain cases” 4. total number of drug interactions Category IV “Monitoring or adjustment necessary” 5. total number of drug interactions Category V “Monitor as a precaution” 6. total number of drug interactions Category VI “ No measures required”

Furthermore the total number of drug interactions which occurred only between chronic medications and between supportive and chronic medications was calculated.

For each single patient drug interactions identified by the software were listed and evaluated in the following manner:

. drug groups leading to drug drug interactions e.g. beta-blockers- diuretics . time of occurrence (hospital admission/discharge or both)

. substance groups involved in drug drug interactions e.g. dexamethasone- hydrochlorothia- zide . classification of drug interactions (according to Category I-VI) . theoretical effect of drug interactions . type of interactions (pharmacokinetic or dynamic) . theoretical mechanism of drug interactions . ABDA No. . evidence of drug interactions (does the drug interaction result in a revisable mode of action Yes/No)

In a six step inappropriate medications (IIMs) found with Cave analysis (STEP 3) were entered into an Excel database worksheet with the information listed below:

. total number of IIMs . total number of IIMs differentiated in (a) contraindicated and (b) restriction on use . differentiation in IIMs restricted on use due to age inappropriateness or disease at different time points (hospital admission and discharge)

In case were individual inappropriate medications occurred twice that means they occurred in the admission as well as in the discharge medication, the IIM was counted as single medication.

For each patient the following data were evaluated for each identified IIM:

. Inappropriate indication due to an underlying disease (6 disease categories were distinguished: 1. Cardiovascular disease, 2. Gastrointestinal Disease, 3. Age inappropriateness, 4. Endocrinological and metabolic diseases 5. Neurological disease 6. Pulmonology) . the time point of occurrence (admission or discharge) . the name of the active drug substance . Category: Restriction on use or contraindication . undesired side effects induced by the use of the inappropriate drug substance (e.g. fluctuations in blood sugar levels, electrolyte imbalances, changes in lab values)

The total amount of IIMs might have been higher than those calculated when adding IIMs from the six categories per patient together, because IIMs might belong to other than the six examined underlying disease classes. The total amount was received by manually summing up all IIMs listed on the Cave print out for each patient, independently of the affiliation to one of the six disease classes.

3.5.3 Patient characteristics and medical conditions Clinical conditions were classified with a modern version of the Charlson comorbidity index (CCI) on the basis of the hospital discharge diagnosis coded by the International Classification of Diseases (ICD 10 Codes) (Table 3-1).

Table 3-1. Definition of the Charlson comorbidity index (CCI) (Charlson et al., 1987). Condition (comorbidity) Weighed score (Age unadjusted)

Cerebrovascular disease 1

Chronic pulmonary disease 1

Congestive Heart Failure 1

Connective tissue disease 1

Dementia 1

Mild liver disease 1

Myocardial infarction 1

Peptic ulcer disease 1

Peripheral vascular disease 1

Any tumor (within the last 5 years) 2

Diabetes with end-organ damage 2

Diabetes with no end-organ damage 2

Hemiplegia from any cause 2

Leukemia 2

Lymphoma 2

Moderate/severe renal disease 2

Moderate or severe liver disease 3

AIDS 6

Metastatic solid tumor 6

The Charlson comorbidity index score (CCI score) uses 19 weighted categories, primarily defined using ICD 10 diagnosis codes to predict the likelihood of 1 year mortality. Each Category has an associated weighted score that is based on an adjusted risk of 1 year mortality. For simplification we used a table which contained ICD 10 and ICD 9 Coding algorithms for Charlson comorbidities e.g. Dementia: F00.x, F03.x, F05.1x and G30.x.

Two types of CCI scores need to be distinguished, the age adjusted and age unadjusted score. A strong influence has been seen to exist with comorbidity. Thus it is recommended for the age adjusted score that 1 point needs to be added to the total score for each decade above the age of 40. Between the age of 50-59 1 point needs to be added, to the score 60-69 2 points, 70 to 79 3 points, 80 to 89 4 points.

Charlson's original study showed that increasing CCI scores were significantly correlated with increased 10-year mortality within a breast cancer cohort (χ2 = 163, p < 0.0001). CCI scores of 0, 1, 2, and 3 predicted 10-year survival rates of 93 %, 73 %, 52 %, and 45 % (Charlson et al., 1987). Three levels of comorbidity were defined based on the Charlson index score 0 (low), 1-2 moderate, and 3 or more (high) (Hall et al., 2004, Schmidt et al., 2010, Wang et al., 2007).

Table 3-2. Classification of comorbidities in four disease stages (Charlson et al., 1987). Charlson index 0 1-2 3-4 ≥ 5

Grade of comorbidi- 1 2 3 4 ty

Illness degree No Mildly ill Moderately ill Severely ill

Finding that CCI has been widely used and validated throughout the oncologic literature (Hall et al., 2004) implemented an electronic application for rapidly calculating the Charlson comorbidity index. The provided Microsoft Excel (MS Excel) macro for the rapid and accurate calculation of CCI scores was used in this study.

The Karnofsky scoring system is used to decide which patients are physically suitable for radiooncological treatment and is an important tool in predicting patient’s life expectancy (Podichetty et al., 2003, Simon et al., 2011). The Karnofsky Performance Status Scale is de- signed to assess independent function and appears to have substantial validity as an indicator of overall physical status. Comprehensive geriatric assessment (CGA) should be performed before radiotherapy to establish baseline functioning. Performance status scores are widely used in oncologic practice because they correlate with patient survival duration (Albain et al., 1991) and response to treatment (Sengelov et al., 2000), as well as their quality of life (Blagden et al., 2003, Finkelstein et al., 1988). Functional status is an important predictor of response to therapy (Haas and Kuehn, 2001).

3.5.4 Drug combinations examined As a matter of course not all medications licensed for the German market could be examined for their potential to induce pDDIs. The retrospective data analysis considered drugs which were often prescribed in ambulatory and hospital care settings (antihypertensive, CSE antagonists, antibiotics) and drugs which were randomly prescribed but were known to lead to severe clinical outcomes (e.g. Fluconazole).

Potential drug interactions with narcotics, chemotherapies, vaccinations were not considered. Supportive medications taken by the patients on behalf of medical advice in the hospital in order to reduce side effects of oncologic radiation were included. Co-medication unknown to the treating physician (OTC products bought by patient himself) as well as other OTC drugs

(NSAID use) were negotiated and could be the subject of further studies. Furthermore not subject of the study were homeopathic and anthroposophical drugs, as well as cytostatic drug preparations except oral anticancer drugs.

Different routes of administration were considered in the medication profile. All systemically routes of application such as oral, rectal, and nasal applications were included into the study. Topically applied drugs with presumed topical effects were not evaluated, other than inhaled drugs for the treatment of airway diseases. Anaesthetics patches which continuously release drugs for pain relief were part of this study.

The screening for potential drug interactions was performed using the active drug substance name. The description of potential interactions referred to the substance group to which the drug substance belonged. In case two drugs with identical active agent had been prescribed, the interaction counted once. In case of non identical active agents, belonging to the same substance group, the count of the corresponding interactions was done separately. Details concerning day dosage and individual dosages were unreliable and were therefore not taken into account for further analysis.

3.5.5 Chronic medications and supportive medications Two kinds of medication had to be distinguished. Chronic medication was defined by the oral intake period. The exact oral intake period in the past could not be inferred from the discharge letters. As continuously taken medications we therefore defined medications listed in the entrance medication used to treat comorbid illnesses unless otherwise noted.

Supportive medication was not fixed in a patient’s pharmacotherapy profile. The patient could have decided depending on the degree of subjective complaints, about the necessity and frequency of taking medications to treat side effects and receiving relieve of pain (Pain medication, laxatives). NSAIDs, supportive gastroprotecting agents (antacids) and corticosteroids were regarded as supportive medication. As supportive GI medications were considered: Dopamine- antagonists: Metoclopramide and 5-HT3 antagonists e.g palonosetrone. As ulcer prophylaxis the following drugs were considered: H2 blockers (e.g. ranitidine) and PPIs (Omeprazole, pantoprazole). The following drug classes were enclosed in the supportive radiooncologic medication: antitussive agents (e.g. acetylcysteine, codein), laxatives (e.g. bisacodyl, macrogol) sleeping aids (e.g. zoplicon, benzodiazepines) and antiemetics and anti-diarrheal agents (antacids, dimenhydrinate, loperamide). The administration of opioids was regarded as chronic medication because for effective treatment they should be administered on a regular basis.

3.5.6 Definition of the classification pattern used to estimate the risk for potential drug interactions The ABDA Drug interaction software classified severities of DDIs into VI categories (Hansten and Horn, 2012):

Category I: Contraindicated (serious events likely)

Category I includes drug interactions which may be life-threatening, or intoxication or permanent damage may be induced. The administration of these drug pairs must be avoided because serious consequences on health are documented. The risk of adverse patient outcomes precludes the concomitant administration of these drugs (Hansten and Horn, 2012).

Category II: Contraindicated as a precaution

Category II includes drugs which should not be administered concurrently because serious consequences need to be expected on a theoretical basis. Under special circumstances the benefit of the drug combination outweighs the risk. Monitoring and adjustment of therapy is necessary.

Category III: Monitoring or adjustment necessary in certain cases

Category III drug interactions may cause reduced or increased effects of drug therapy. Patients with organ dysfunction are exposed to an increased risk for pDDIs. Patients with renal failure and hepatic insufficiencies require special supervision. Changes in the route of administration and in drug dosages that can circumvent or minimize the risk for potential drug interactions are suggested (Hansten and Horn, 2012).

Category IV: Monitoring or adjustment necessary

Category IV includes drug interactions between drug combinations which lead to difficulties in drug therapy. The combination of these drug combinations can be administered in case careful monitoring of lab values and clinical symptoms is guaranteed (Monitoring laboratory parameters: INR, blood cell counts). Adjustment measures are necessary. Considering alternative drug products, timely separation of oral intake periods, dose adjustment and dose constraints as well as supervision with regard to undesired effects needs to be undertaken under all times. The benefit of the combinations outweighs the risk.

Category V: Monitor as a precaution

Category V includes drug interactions that are theoretically possible, but have not been documented in the literature. Only isolated cases are cited or even postulated, and their clinical relevance is unclear.

Category VI: No measures required

Category VI includes drug interactions for which a risk of adverse outcome is very small. In case medication is used as recommended, there is no increased risk for drug interactions. No specific action is needed other than to be aware of the possibility of the drug interaction.

The ability to customize drug interaction alert severity is a desirable feature and was included in our data base. For our study no sections of the drug interaction alert system were switched off. In this study pDDIs of all severities were included for analysis.

3.5.7 Evaluating potential drug interactions For interacting drug combinations the ABDA drug interaction software provides extensive information on each potential drug interaction. Moreover, for each identified pDDI it provides a monograph with specific information on expected adverse outcomes, underlying mechanisms of interaction (e.g. pharmacokinetic, pharmacodynamic, etc.), management recommendations and recent literature. Details and examples are listed in Annex A2.

3.5.8 Potential drug interactions classification and flagging In accordance with the study of (Frechen et al. 2012) a ratio was calculated to distinguish the individual potential of the substance for pDDIs. In case a drug was involved in a potential DDI

(Category I-VI) the drug substance received one point in the score (ni). The total number of points was then divided by the number of patients who had prescribed the drug in their medication portfolio (np). By means of the ratio the likelihood of generating relevant pDDIs in clinical practice was assessed. The score values had been adopted from (Frechen et al., 2012, Gaertner et al., 2012). Score calculation: number of class 1-6 pDDIs per patient exposed:

n Flag score = i ⋅100 (1) n p “Red flag”: Flag score ≥50: Beware of therapeutically relevant pDDIs, avoid administration if possible.

“Yellow flag”: Flag score 25-50: Alternatives should be considered, withdrawal should be considered.

“Green flag”: Flag score 0-25: No serious clinical events due to pDDIs are expected in general.

3.6 Potentially inappropriate medications In the following section potentially inappropriate medications in elderly patients were evaluated. As mentioned in the literature different criteria can be used to define PIMs. For the evalua-

tion in this preliminary study two different criteria were used: the Cave module and the Priscus list.

In the Priscus list the term PIM is used as an established expression. In order to avoid confusion the potentially inappropriate medications evaluated by the Cave module were called “Individual inappropriate medications”. In the Priscus list most of the medications were rated as PIM independent of diagnosis or clinical condition.

3.6.1 Cave module The application oft the Cave module was described in detail in Annex A2. Drug disease interactions can occur in all age categories. Therefore all 120 patients were included into the study.

3.6.2 Priscus list In our study cohort all patient´s medication profiles aged over 65 were screened for the 83 medications, listed on the Priscus list at hospital admission and at hospital discharge (Projektverband Priscus, 2011). The drug substances for which a dose limitation is given in the Priscus list such as benzodiazepines the discharge letters were manually retrieved for the exact doses. Patients below the age of 65 were excluded from this statistical evaluation. 79 patients in the study cohort were aged over 65 and were included into the study.

3.7 Statistical analysis Microsoft Excel 2007 (Microsoft Corporation, Redmond, WA, USA) and SPSS (computer software packages IBM SPSS Statistics 21.0 - August 2012) were used for data analysis.

Descriptive statistics were utilized to report the prevalence of PIM use and pDDIs. To describe the distribution of other variables means with corresponding 95 % confidence intervals (CIs), medians and ranges, or proportions were calculated. The frequency of potential drug interactions was determined by means of worksheet calculations and expressed in absolute frequencies.

To compare the distribution of categorical data of independent samples inferential statistics including Chi Square test and/or Fisher´s Exact test were used. The distributions of continuous data were compared using the Student’s t-test (independent two-sample comparisons).

Multivariate logistic regression analysis was performed to identify the independent risk factors which were associated with PIM use and pDDIs. Independent variables tested in the study included: age (continuous variable), presence of comorbid diseases, and number of medications (continuous variable). In the multivariate regression analysis predictors were considered to be statistically significant if the corresponding p-value was ≤ 0.05.

4 Results

4.1 Patient characteristics A total of 120 elderly patients were enrolled in the study. The patient characteristics are dis- played in Table 4-1. The 120 patients included in the study had a total of 48 diagnoses and were prescribed a number of 700 drugs in total at hospital admission. The number of drugs in the discharge medication profile increased up to 1024. In the study evaluation 667 medications in the admission medication and 918 in the discharge portfolio were analysed in detail because the study only considered the main drug classes.

Table 4-1. Patient characteristics (n = 120). Sex, number(%) Male 66 (55 %) Female 54 (45 %) Age Mean 68.75 95 % Confidence Interval 66.8 – 70.7 Median 70 Range 38 – 94 CCI score age unadjusted Mean 5.69 Median 6.0 Standard deviation 2.365 Range (Min-Max) 2 – 12 CCI score age adjusted Mean 8.20 Median 9.0 Standard deviation 2.800 Range (Min-Max) 2 – 14 Karnofsky index (admission) Mean (Admission) 67.92 Median 70 Standard deviation 14.55 Range (Min-Max) 30-100 Karnofsky index (discharge) Mean (Discharge) 75.42 Median 80 Standard deviation 15.06 Range (Min-Max) 30-100 Main cancer diagnosis Laryngeal cancer 18 Bronchial carcinoma 17 Breast Cancer 14 Others 12 Gynaecological cancer 8 Prostate cancer 7 Renal cell carcinoma 6 Braintumor 5 CUP 6 Hypopharynx carcinoma 5 Metastatic breast cancer 4 Haemotological cancer 4 Rectal cancer 4 Oesophageal cancer 4 Ovarial cancer 2 Parotis cancer 1 Hepatocellular carcinoma 1

Medication admission Mean 6.12 (No. of patients: 120 valid) Median 5.0 Standard deviation 3.700 Range Min-Max 0 – 21 Medication discharge Mean 8.53 (No.of patients: 120 valid) Median 8.50 Standard deviation 3.689 Range Min-Max 0 – 21 Number of prescribed medications Hospital admission 700 / 667 (in total) Hospital discharge 1024 / 918 Potential drug interactions at admission Mean 1.59 95 % CI 1.02 – 2.16 Median 0 Range 0-21 Potential drug interactions at discharge Mean 2.25 95 % CI 1.63 – 2.87 Median 1.0 Range 0 – 21 Number of drug interaction alerts 288 (according to the ABDA classification system)

The study group comprised sixty six men (55.0 %) and fifty four (45 %) women.

Table 4-2. Age distribution in the patient collective. Gender Age Total < 55 years 55 – 65 years 66 – 75 years > 75 years Female 6 (5 %) 15 (12.5 %) 15 (12.5 %) 18 (15 %) 54 (45 %) Male 8 (6.7 %) 14 (11.7 %) 24 (20 %) 20 (16.6 %) 66 (54 %) Total 14 (11.7 %) 29 (24.2 %) 39 (32.5 %) 38 (31.6 %) 120 (100 %)

Table 4-2 shows the age distribution of the population in the study. The age of the patients was based on the time of data analysis (deadline 01.11.2010). The mean age of the male patients was 68.5 years. The female patients were averaged 69 years. 77 patients (64.2 %) of the study were aged over 65, while 38 patients (31.6 %) were aged over 75. Only 14 patients (11.7 %) were under 55 years old. On average the patients over 75 years had significantly more diagnoses (p = 0.001) and more pharmacologically active substances prescribed than patients below 75 years of age.

The most common symptoms leading to an admission to the radiooncology ward were deterioration of health status (82.5 %), pain (24.2 %), dysphagia (21.7 %), nausea (14.2 %) , inability to retain food (18.3 %), dyspnoea (11.7 %), and a deterioration of lab values (15 %) (e.g. eleva- tion of liver enzymes, drop in leucocytes).

There was a malignant diagnosis in all 120 patients. The most common cancer diagnosis was laryngeal cancer (n = 18), followed by small cell lung cancer (n = 17) and breast cancer (n = 14). Patients aged over 65 were primarily affected by small cell lung cancer whereas laryngeal cancer mainly affected patients between 55-65 years of age. 46 patients (38.3 %) did not show any signs of metastasis.

66 patients (55 %) did not show any signs of severe clinical symptoms at hospital admission. 12 patients (10 %) showed signs of dehydration. Loss of appetite and vomiting were the major reasons for dehydration during radiation therapy. 2 patients (1.6 %) were admitted due to acute severe diarrhoea (Figure 4-1).

10 patients (8.3 %) had symptoms of tachycardia like a rapid pulse rate (Beets / min > 100), heart palpitations, chest pain and dizziness. 7 patients (5.8 %) showed arrhythmic heart beats while 3 patients (2.5 %) showed signs of bradycardia. None of the patients had been admitted to hospital with signs of bleeding, ulcers or thrombosis according to the discharge letters.

Haemoptysis Hypocalcemia Anemia Low level of medica tion Fluctuations in blood sugar level Diarrhoea Syncopes Dehydration Hypokalemia Tachycardia Bradycardia Hypertensive episode Cardiac arrthymias QT prolongation 0 2 4 6 8 101214 Number of patients

Figure 4-1. Clinical symptoms at hospital admission.

4.2 Comorbidities

4.2.1 Comorbidities in the study cohort Patients showed a number of 48 coded diagnoses. In total 309 comorbid diseases occurred in the study cohort.

8 * Age group MS < 55 y 7 NEPH ** Age group 6 55 - 65 y PSY NEU 5 Age group 66 - 75 y MD 4 Age group > 75 y GI 3 Type of comorbidity P 2

Number ofNumber comorbid diseases CA 1 0 10203040506070 0 2 4 6 8 10121416 Proportion of patients with Number of patients comorbidities [%]

Figure 4-2. Number of comorbidities in the study cohort stratified according to age (left) (*: highest number of comorbid diseases in the age group 66-75 years, **: highest number of comorbid diseases in the age group > 75 years), different types of comorbidities (right) (COM = Comorbidity MS: Musculoskeletal system; NEPH: Nephrology, PSY: Psychiatric conditions, NEU: Neurological disease, MD: Metabolic disease, GI: Gastrointestinal system, P: Pulmonology, CA: Cardiovascular disease.

Only 9 patients (7.5 %) had no comorbid conditions at the time of diagnosis. 57 patients (47.5 %) had 1-2 comorbid conditions at hospital admission. 54 (45 %) patients had more than 2 comorbidities. Figure 4-2 (left) illustrates the frequency of different comorbidities in the study cohort. The most prevalent co-morbidities besides cancer were cardiovascular diseases with a prevalence of 62.5 % (n= 75) (Figure 4-2 (right)).

Hypertension, coronary artery disease and diabetes mellitus were the three most prevalent comorbidities in the whole study population with overall prevalences of 32.5 %, 22.5 % and 16.7 %. There was a significant increase in the prevalence of hypertension, coronary artery disease, arrhythmia and diabetes with advancing age (p = 0.001). The prevalence of psychiatric disorders (e.g. depression) was quite low in radiooncologic patients (5.8 %). 18.3 % of the patients had underlying comorbidities influencing the pharmacokinetic metabolism of medications. 19 patients (15.8 %) were affected from renal impairment, while 3 patients (2.5 %) showed signs of liver cirrhosis.

4.2.2 Comorbidity scores: Karnofsky index and Charlson comorbidity index Karnofsky index score

The study included radiooncologic patients with various diagnosis of cancer. Figure 4-3 shows a score distribution between 30 and 100 %. 97 patients (80.83 %) had a score exceeding 50 % implying occasional assistance in daily living. Figure 4-3 reveals that patients with very low scores in the Karnofsky index (30 %) were younger than 65. Figure 4-3 depicts that patients aged 66-75 years most frequently presented with Karnofsky scores between 60-80%.

Patients having a score between 80 and 100 % (46; 38.3 %) were able to carry on normal activity and no special care was needed. Patients with a score between 50 and 70 % (65; 54.2 %) were often unable to work, being able to live at home and cared for most personal needs with varying amount of assistance.

30% severely disabled Age group < 55 y 40% disabled Age group ** 55 - 65 y 50% assistance; medical care * Age group 66 -75 y 60% occasional assistance Age group >75 y 70% cares for self 80% normal acticity with effort Karnofsky index index score Karnofsky 90% minor signs of disease 100% normal no complaints

02468101214 Number of patients

Figure 4-3. Karnofsky index score in the study collective stratified according to age groups (*: worst Karnofsky index score in age group 66-75 years, **: worst Karnofsky index score in age group > 75 years).

Figure 4-4 depicts that during radiooncologic treatment the scores in the Karnofsky index improved between hospital admission and discharge. The mean value of the Karnofsky index at hospital admission was 67.9 with a confidence interval of (CI 95 % 65.28-70.54); whereas the mean increased at hospital discharge up to 75.4 with a confidence interval of (CI 95 % 72.69- 78.13). A KPS improvement up to 60 % indicating to regain the ability to function mostly independent of help in all key areas of live, was observed in 12 out of 120 patients. At hospital discharge the number of patients with a score of 90 % indicating an improvement towards being able to perform normal activity with minor symptoms increased from n = 9 (7.5 %) up to n = 26 (21.6 %) at hospital discharge. Another pronounced increase was seen for a score of 100 %, where the prevalence increased from 0.8 % (n = 1) to 7.5 % (n = 9). In the paired T- test it has been shown that significant improvement in the Karnofsky performance score occurred under radiooncologic treatment (p = 0.0001).

100

50 Admission Karnofsky index score index Karnofsky 40 Discharge 30

0 5 10 15 20 25 30 35 40 Number of patients

Figure 4-4. Changes in score values in the Karnofsky index during hospital admission and discharge.

Charlson comorbidity index

The Charlson comorbidity index (CCI) score, unadjusted and adjusted by age, was calculated in each patient. In the cohort of 120 patients comorbid conditions were common with CCI scores ranging from 2-14. Figure 4-5 shows the different values for the Charlson comorbidity index classified in an age adjusted and age unadjusted score.

14 13 Age adjusted 12 Age unadjusted 11 10 9 8 7 CCI score 6 5 4 3 2 0 5 10 15 20 25 30 Number of patients

Figure 4-5. The Charlson comorbidity index score (age adjusted and unadjusted).

As soon as the CCI score is age adjusted the points in the CCI score shift to higher numbers. The mean value in the CCI score age unadjusted was 5.69, the median 6.0, whereas the mean value in the CCI age adjusted score was 8.20 and the median 9.0. The CCI score was highly correlated with increasing age (Pearson = 0.434, P = 0.001).

Figure 4-5 shows that 15 patients (12.5 %) in the study cohort had an age unadjusted score between 0 and 2 that implies a moderate level of comorbidity. The majority of patients (n = 105; 87.5 %) at hospital admission had a score higher than 3 and therefore a higher risk for overall 54

mortality. In order to make a statistically valuable statement, another CCI sore classification was applied. The patient cohort was divided into patients with a CCI score < 4, CCI score between 5 and 8, and > 8.

Table 4-3. Distribution in CCI score. Age adjusted Age unadjusted < 55 y 55 – 65 y 66 – 75 y > 75 y Total < 55 y 55 – 65 y 66 – 75 y > 75 y Total ≤ 4 5 7 5 1 18 7 11 13 10 41 5 – 8 6 15 10 9 40 6 15 20 24 65 > 8 3 7 24 28 62 1 3 6 4 14 Total 14 29 39 38 120 14 29 39 38 120

41 (34.2 %) patients had a CCI score age unadjusted less or equal than 4, 65 patients (54.2 %) an age unadjusted score between five to eight and 14 patients (11.7 %) an age unadjusted score > 8 (Table 4-3). For the age adjusted score the numbers can be described as follows: 18 patients (15 %) had a score less or equal than 4, 40 patients (33.3 %) a score between five to eight, and 62 patients (51.7 %) a score > 8.

The logistic regression analysis showed that patients with a CCI age adjusted score > 8 were significantly affected by potential drug interactions according to univariate analysis (Table 4-4). The multivariate analysis considered age (< 55, 55-65, 66-75, > 75), number of medications, gender and comorbidities. No statistical significance could be detected in multivariate analysis.

Table 4-4. Correlation between Charlson comorbidity index and number of potential drug interactions.

CCI No. of patients Univariate Multivariate with potential DDIs

Yes No OR 95 % CI P-value OR 95 % CI P-value

< 4 8 10 0.48 0.17 – 1.31 0.150

4 - 8 20 20 0.54 0.25 – 1.17 0.116

> 8 44 18 2.62 1.23 – 5.56 0.012 1.03 0.40 – 2.64 0.954

4.3 Prescribed medications Figure 4-6 shows the number of medications which patients received per patient during hospital admission and discharge. The number of medications increased when patients were discharged from hospital after radiooncologic treatment. In total the number of prescribed drugs increased from 700 (667) up to 1024 (918) medications. The number of concomitant prescription medications ranged from 0 to 21. The mean total number of medications per patient was 6.12 at hospital admission and 8.53 at hospital discharge. The number of drugs prescribed at hospital discharge was significantly higher than at hospital admission (p < 0.0001).

20 18 Admission 16 Discharge 14 12 10 8 Frequency Frequency 6 4 2 0 0 1 2 3 4 5 6 7 8 9 101112131415161718192021 Number of medications per patient

Figure 4-6. Number of medications per patient in the admission and discharge medication portfolio.

35.8 % of patients (n = 43) received between 0-4 medications at hospital admission, while 64.2 % (n = 77) received more than 5 medications. These patients were affected by polypharmacy ( > 5 medications). At hospital discharge the proportion of patients affected by polypharmacy increased up to 84.2 % (n = 101).

Figure 4-7 shows that in the medication group “9-20” the number of patients affected by polypharmacy increased steadily with increasing age (< 55 y – > 75 y) at hospital admission and discharge. The number of patients aged over 75 receiving 9-20 medications, increased from 17.5 % (n = 21) at hospital admission up to 23.4 % (n = 28) at hospital discharge. A subgroup analyses revealed that the number of medications over all age groups increased significantly with advancing age (p = 0.003).

0 - 4 (AD) * Age group Admission < 55 y 5 - 8 (AD) Age group 55 - 65 y 9 - 20 (AD) Age group ** 66 -75 y Age group 0 - 4 (D) > 75 y Discha rge 5 - 8 (D)

Number of prescribed ofNumber prescribed medications 9 - 20 (D) * ** 0 5 10 15 20 25 30 Number of patients

Figure 4-7. Number of prescribed medications in the study cohort (AD: Admission, D: Discharge / *: highest number of prescribed medications in the age group 66-75 years, **: highest number of prescribed medications in the age group > 75 years).

100 90 Admission 80 Discharge 70 60 50 40 30 20

Patients with prescriptions [%] with prescriptions Patients 10 0 ABCGH J LMNR Anatomical main group

Figure 4-8. Number of patients with prescriptions belonging to the drug classes according to the anatomical therapeutical chemical index (ATC) classification at hospital admission and hospital discharge; n = number of patients; anatomical main groups: A = alimentary tract and metabolism, B = blood and blood forming organs, C = cardiovascular system, D = genito urinary system and sex hormones, H = systemical hormonal preparations, exclusive sex hormones and insulines, J = antiinfectives for systemic use, L = antineoplastic and immunomodulating agents, M = musculo-skeletal system, N = nervous system, R = respiratory system.

Figure 4-8 shows the number of patients with at least one prescription belonging to the specified drug class (according to the anatomical therapeutical chemical (ATC) classification) adopted from (Vonbach, 2007). At hospital discharge the prevalence of patients receiving drugs affecting the alimentary tract and metabolism (ATC A) as well as drugs regulating the nervous system (ATC N) slightly increased from 80.9 % to 94.2 % for drugs of group ATC A and from 58.3 % to 70.9 % for drugs belonging to ATC class N. Both drug classes contained supportive medications such as drugs to treat disorders of the gastrointestinal tract or metabolic diseases (mineral supplements, proton pump inhibitors, stomach protectors) for ATC A and primarily analgesics benzodiazepines and sleeping medications for ATC N. A subgroup analysis revealed that the total number of medications from ATC A increased from 176 (26.4 %) up to 291 (31.7 %) at hospital discharge while for ATC N the total values were 135 (20.2 %) at hospital admission and 209 (22.7 %) at discharge. The other two most prevalent anatomical groups were drugs affecting the blood and blood forming organs (admission 35.8 %, discharge 41.7 %) and drugs affecting the cardiovascular system (admission 66.6 %, discharge 69.2 %).

Since arterial hypertension and coronary heart disease were the most prevalent comorbidities, ACE inhibitors (admission: 32.5 %; discharge: 36.7 %) beta-blockers (admission: 29.2 %; discharge 35.0 %), angiotensin receptor blockers (admission: 5.8 %; discharge 7.5 %), and thiazide or loop diuretics (admission: 24.2 %; discharge: 30.8 %) were the drugs most often prescribed concomitantly in the patient collective at hospital admission and discharge (n = 120). The pre-

scription frequency of beta-blockers, ACE inhibitors and calcium channel blockers increased significantly with advancing age (p = 0.001).

Opioids Age group Non-opioid analgesics < 55 y Non-steroidal anti-inflammatory drugs Age group 55 - 65 y Tetracyclic antidepressants Age group Selective serotonin reuptake inhibitors 66 - 75 y Tricyclic antidepressants Age group > 75 y Neuroleptics Vitamines Supportive GI medication Supportive radiooncologic medication

Ulcer prophylaxis * Corticosteroids Other antihypertensives Diuretics ** Angiotensin II receptor antagonists Calcium channel blockers Beta-blockers ACE inhibitors 0 5 10 15 20 25 30 35 40 45 Frequency

Figure 4-9. Prevalence of the most frequently prescribed medications in 120 patients stratified by age group at hospital admission (*: highest frequency of medications in age group 66 -75 years, **: highest frequency of medications in age group > 75 years).

Figure 4-9 reveals that patients in the age group of 66-75years had a higher prescription frequency of ulcer prophylaxis medications and supportive medications at hospital admission in comparison to the other age groups (steroids, gastric protecting agents, vitamins). In the age group > 75 diuretics had the highest prescription frequency. At hospital discharge the prescription frequency of supportive medications increased in the other age groups due to the high prescription frequency after radiooncologic treatment (Figure 4-10). The prescription frequency of antidepressants (14.2 %) and neuroleptics (2.5 %) at hospital admission was very low in this study and increased at hospital discharge (antidepressants: 24.2 %; neuroleptics: 6.6 %). The most frequently prescribed opioids had been morphin, oxycodon and fentanyl. At hospital discharge the prescription frequency of morphin increased from 8 to 22 % whereas fentanyl increased only from 2.5 to 6.7 %.

Opioids Age group Non-opioid analgesics < 55 y Non-steroidal anti-inflammatory drugs Age group Tetracyclic antidepressants 55 - 65 y Selective serotonin reuptake inhibitors Age group 66 - 75 y Tricyclic antidepressants Age group Neuroleptics > 75 y Vitamines Supportive GI medication Supportive radiooncologic medication ** Ulcer prophylaxis * Corticosteroids Other antihypertensives Diuretics Angiotensin II receptor antagonists Calcium channel blockers Beta-blockers ACE inhibitors

0 5 10 15 20 25 30 35 40 45 Frequency

Figure 4-10. Prevalence of the most frequently prescribed medications in 120 patients stratified by age group at hospital discharge (*: highest frequency of medications in age group 66 -75 years, **: highest frequency of medications in age group > 75 years).

Figure 4-11 shows the detailed prescription frequency of supportive medications in the patient collective. The figure reveals that most of the drugs prescribed belonged to the following classes: (I) laxatives (Bisacodyl, macrogol, and lactulose), (II) antitussive agents (ACC, codeine) and (III) antacids (aluminium hydroxide). Medications to treat diarrhoea (loperamide, Saccha- romyces cerevisiae) were prescribed rarely. 15 % of the patients received acid blockers at hospital admission. At hospital discharge the prescription rate increased up to 24.1 %. The application of pantoprazol increased from 26.7 % (n = 32) to 56.7 % (n = 68), whereas the administration of omeprazole was reduced nearly by half at hospital discharge (n = 13). Omeprazole is a drug substance with a high interaction potential. NSAIDs most frequently prescribed to treat pain were diclofenac, paracetamol, ibuprofen and etoricoxib (Figure 4-11). The prescription frequency of diclofenac and ibuprofen did not change essentially during admission and discharge. The prevalence of patients receiving etoricoxib, a drug with a high interaction potential increased up to 4.2 % (n = 5) at hospital discharge.

Pantoprazole Metoclopramid Dexamethasone Macrogol Admission Zopiclon Aluminium hydroxide Discharge Omeprazole Esomeprazole Potassium/Magnesium Acetylcysteine Ibuprofen Diclofenac Magnesium oxide Prednisolone Palonosetron Bisacodyl Codein Etoricoxib Dexpanthenol Lactulose Cortisol Ondansetron Ranitidin Loperamide Paracetamol Acetylsalicylic acid (ASS) Dimenthidenmaleat Saccharomyces cerevisiae Alpha lipoic acid (ALA) Silymarin 0 10203040506070

Figure 4-11. Prescription frequency of supportive medications.

4.4 Analysis of potential drug interactions

4.4.1 Number of potential drug interactions A total of 466 potential drug interactions were found in the study cohort. In the admission medication portfolio 191 potential drug interactions were detected. This corresponds to an average of 1.59 interactions per patient (CI 95 % 1.02-2.16). In the discharge medication portfolio the number of potential drug interactions increased up to 275 (2.25 interactions per patient; CI 95 % 1.63-2.87). At hospital discharge significantly more pDDIs per patient (2.25) were detected than at hospital admission (1.59) (p = 0.0001).

A subgroup analysis revealed that 13 potential drug interactions were detected exclusively in the admission medication. 178 potential drug interactions occurred in the admission as well as in the discharge medication, while changes in the prescription behaviour during hospitalisation lead to 97 (33.7 %) new potential drug interactions in the discharge medication. 288 relevant

potential drug interactions were detected in the study cohort in case the potential drug interactions detected in the admission and discharge medication were counted only once as medication.

The prevalence of pDDIs at hospital admission, as well as at hospital discharge is shown in detail in Table 4-5.

Not every patient had a drug interaction risk in his portfolio. The share of patients without any pDDIs was 58.3 % at hospital admission and 40.0 % at hospital discharge. Among the 120 patients, 50 patients had at least one pDDI at hospital entry resulting in a prevalence value of 41.7 %. This figure increased at hospital discharge up to 60 %. At hospital admission 14 patients (11.7 %) were affected twice by a possible potential drug interaction, 7 patients (5.8 %) three times, 5 patients (4.2 %) four times and 5 patients (4.2 %) five times. 8 (6.7 %) patients received more than 5 potential drug interaction combinations within their medication portfolio.

Table 4-5. Number of patients with n = (1, 2, 3, 4, 5-18) potential drug interaction combinations at hospital admission and discharge. No. of Interac- Admission Discharge tions Frequency Frequency n [%] Cum.-[%] n [%] Cum.-[%] 0 70 58.3 58.3 48 40.0 40.8 1 11 9.2 67.5 18 15.0 55.8 2 14 11.7 79.2 14 11.7 67.5 3 7 5.8 85.0 12 10.0 77.5 4 5 4.2 89.2 9 7.5 85.0 5 5 4.2 93.3 7 5.8 90.0 6 2 1.7 95.0 3 2.5 92.5 7 - - 95.0 2 1.7 94.2 8 2 1.7 96.7 2 1.7 95.8 9 1 0.8 97.5 1 0.8 96.7 10 1 0.8 98.3 1 0.8 97.5 16 - - 98.3 1 0.8 98.3 18 1 0.8 99.2 1 0.8 99.2 21 1 0.8 100.0 1 0.8 100.0 Total 120 100 120 100

Table 4-5 indicates an increase in the prevalence of patients with 1 possible interaction from 9.2 % up to 15 % and with three interaction combinations from 5.8% to 10 % at hospital discharge. In the discharge medication 12 patients at hospital discharge had more than 5 possible jeopardizing drug interaction combinations in their portfolio.

Figure 4-12 indicates that at hospital discharge the number of potential drug interactions between chronic medications increased (IAC-C: interactions between chronic-chronic medications). At hospital admission 43 (35.8 %) patients showed potential drug interactions between chronic medications in their medication portfolio while in the discharge medication profile the number of patients affected by interactions between their chronic medications increased up to 56 (46.7 %) implicating an increase of 11 %. Figure 4-12 reveals that more than 6 potential drug interactions were frequently caused by interactions between chronic medications. This trend is seen at hospital admission and at hospital discharge. A subgroup analysis revealed that patients

with more than 6 potential interactions at hospital discharge were affected by the same drug interaction also in their admission medication.

19 19 IA S-C IA S-C 17 17 IA C-C IA C-C 15 15 13 13 11 11 9 9 7 7 5 5 3 3 Number of potential interactions of potential Number interactions of potential Number 1 1 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20 Number of patients Number of patients

Figure 4-12. Number of potential drug interactions between chronic medications (IAC-C); and supportive and chronic medications (IAS-C). Hospital admission (left), hospital discharge (right).

4.4.2 Classification of potential drug interactions The distribution and classification of the potential drug interactions according to the classification system of the ABDA drug interaction module is presented in Figure 4-13. In terms of severity, the majority of potential drug interactions belonged to Category IV and V. 169 out of 191 pDDIs (88.5 %) detected in the admission medication belonged to Category IV and V, while it had been 239 out of 275 (86.9 %) at hospital discharge. At hospital discharge 70 new pDDIs belonging to Category IV and V occurred.

150 Admission 125 Discharge 100

Number of potential drug interactions drug interactions of potential Number 0 Category I Category II Category III Category IV Category V Category VI Classification according to the ABDA interaction module

Figure 4-13. Distribution and classification of potential interactions according to the classification system of the ABDA drug interaction module: Category I: (Contraindicated), Category II: (Contraindicated as a precaution), Category III (Monitoring or adjustment necessary in certain cases), Category IV (Monitoring or adjustment necessary), Category IV: (Monitor as a precaution), Category VI (No measures required).

The documented 288 different drug interactions in this study referred to 106 ABDA drug interaction monographies available from the ABDATA Service constituted on November 2010. The distribution of the interaction monographies within the 6 categories can be derived from Figure 4-14. For drug interaction alerts of Category II 3 different drug combinations had been responsible belonging to two different ABDA monographies. Drug interaction alerts of Category III had been caused by drug combinations of 12 interactions monographies, while in Category IV 133 drug combinations were involved which belonged to 53 interaction monographies. 38 different ABDA interaction monographies were responsible for 118 pDDIs of Category V.

140

120 ABDA monogra phies 100

80 Total monographies

20 Number of potential drug drug interactions of potential Number 0 Category I Category II Category III Category IV Category V Category VI Classification according to the ABDA interaction module

Figure 4-14. Distribution of the detected pDDIs among similar ABDA monographies.

4.4.3 Correlation analysis Predictors of potential DDIs

All 120 patients were included into the logistic regression model in order to determine risk factors associated with potential drug interactions. The logistic regression model was built using a stepwise selection of variables. Dependent variable of the model was the variable "Interaction(s) at discharge yes/no". As independent variables gender, age (in groups: "-55", "56-65", "66-75" and "> 75"), number of medications (0-4, 5-8, 9-20) at discharge, CCI score and special comorbidities were considered.

Table 4-6. Factors significantly associated with an increased risk of clinically relevant potential drug interactions at hospital discharge using univariate and multivariate regression analysis. Variables No. of poten- Univariate Multivariate tial DDIs Yes No OR 95 % CI P-value OR 95 % C P-value Age 75 years (Ref- 31 7 4.43 1.75-11.20 0.002 2.45 0.87-6.95 0.091 erence < 75 y) ≥65 51 26 2.06 0.96-4.40 0.064 Gender Male 66 54 0.80 0.38-1.67 0.549 0.67 0.27-1.67 0.388

Variables No. of poten- Univariate Multivariate tial DDIs Yes No OR 95 % CI P-value OR 95 % C P-value Number of medications 0-4 1 18 0.02 0.01-0.18 0.001 5-8 22 19 0.31 0.31-1.44 0.308 9-20 49 11 7.17 3.11-16.53 0.0001 4.27 1.60-11.35 0.004 (Reference <9)

Total number 1.441 1.24-1.67 0.001 of medications Comorbidity COMHK 56 19 5.34 2.40-11.91 0.0001 2.167 0.83-5.69 0.116 COMR 13 10 0.837 0.33-2.10 0.705 0.66 0.22-1.94 0.446

Sex in our study was not significantly related to the occurrence of pDDIs. There is general agreement on the fact that female gender is a risk factor for pDDIs (Cruciol-Souza and Thomson, 2006). Univariate analysis revealed that the age of the patients in the age group over 75 significantly correlated with the number of potential drug interactions (p = 0.002). The multivariate analysis showed no significance in the age group ≥ 75 (Table 4-6).

A subgroup analysis compared the number of interactions in the age group > 75 to the remaining study population at hospital admission and discharge. The Fisher exact test with a p-value of 0.048 revealed that in the age group over 75 significantly more interactions occurred than in the age groups below 75. The odds ratio was 2.26 at hospital admission implying that patients over 75 had a 2.26 fold higher risk of being affected by a pDDI (95 % CI 1.03-4.95) compared to patients < 75 years. At hospital discharge a similar trend was seen. The correlation coefficient was R = 1.48 indicating a positive correlation implying that the older the patient the more drug interactions were detectable. The odds ratio was 4.43 implying that patients over 75 had a 4.5 fold higher risk of experiencing a pDDI at hospital discharge than the remaining patients (95 % CI 1.75-11.20; p = 0.002).

The total number of medications at discharge was significantly related to the occurrence of pDDIs according to univariate analysis (p = 0.001). Table 4-6 reveals that in the univariate regression analysis a significant correlation could be detected for the number of pDDIs and patients receiving 0-4 or more than 9 medications. In multivariate regression analysis significance could be detected for 9-20 medications (p = 0.004). In univariate analysis a significant correlation could be proven for underlying cardiovascular disease and the number of pDDIs (p = 0.0001).

Bearing in mind that there exists a significant correlation between the number of medications and the number of pDDIs Figure 4-15 provides detailed information about the dependency of drug interactions stratified according to different number of medications < 5, 6-9, > 9 at hospital admission and discharge. Due to statistical reasons the age distribution for the calculation had to be modified (age group ≤ 65, 66-75, > 75).

100

40 < = 65 y 20 66-75y > 75 y 0 6 - 9 > 9

Patients with potential interactions[%] <=5 Medications per patient Figure 4-15. Number of pharmacologically active substances prescribed and prevalence of potential drug interactions (pDDIs) stratified by age group. The figure depicts the proportion of patients with a pDDI stratified by age and number of active substances prescribed at hospital admission.

Figure 4-15 shows in detail the proportion of patients (95 % CI confidence interval) with pDDIs stratified by age and number of substances prescribed at hospital admission. The prevalence of patients with relevant potential DDIs increased with increasing numbers of pharmacologically active substances. In patients aged ≤ 65 years the prevalence of pDDIs increased from 4.5 % (95 % CI 0.12-22.84 %) when < 5 medications were prescribed to 100 % (95 % CI 39.7-100 %) in patients receiving more than 9 medications concomitantly. The increase was similarly pronounced in the other age groups.

Figure 4-15 shows that patients aged ≤ 65 receiving less than 5 medications had fewer potential drug interactions whereas the patients aged over 75 had a very high incidence of potential drug interactions (95 % CI 11,8 -88.2 ). In patients receiving more than 6 medications there is no significant increase with age. The proportion of potential drug interactions is similar in the age group of ≤ 65 and 66-75 when 6-9 medications were prescribed.

Like at hospital admission the prevalence of patients with relevant potential DDIs increased with increasing numbers of pharmacologically active substances at hospital discharge. In patients aged ≤ 65 years the proportion of potential DDIs increased from 0 % when < 5 medications were prescribed to 75 % (95% CI 42.8-94.5 %) in patients receiving more than 9 medications concomitantly. The increase was even more pro-nounced in patients aged > 75 years, reaching 88 % (95 % CI 68.7-97.45 %) when 9 medications or more were prescribed.

100 < =65 y 80 66- 75 y > 75 y 60

Patients with potential interactions[%] 0 <=5 6 - 9 > 9 Medications per patient Figure 4-16. Number of pharmacologically active substances prescribed and prevalence of potential drug interactions stratified by age group. The figure depicts the proportion of patients with pDDIs stratified by age and number of active substances prescribed at hospital discharge.

In case more than 6 substances were prescribed there was no significant increase of patients with potential drug interactions detectable comparing the age groups < 65 and 66-75 years. A pronounced increase of potential drug interactions with increased age can be recorded for the age group over 75 years. In the discharge portfolio the proportion of patients affected by pDDIs when receiving 6-9 medications accounted for 55 % in the age group ≤ 65 and 52.9 % in the age group 66-75. 78.5 % of the patients aged over 75 were affected by pDDIs when 6-9 medications were prescribed. In the admission portfolio the proportion of patients in all age groups affected by pDDIs when 6-9 medications were prescribed ranged between 50-58 %.

4.4.4 Medication classes involved frequently in potential drug interactions In total 96 different medication class combinations, 173 different drug combinations and 104 single substances were responsible for the potential drug interactions in this study. 19 medication classes were respectively mainly involved with a share between 79.1 % in the admission medication and 75.3 % in the discharge medication portfolio. For the evaluation of the most frequently detected pDDIs classified according to the ABDA classification system the 288 different relevant potential drug interactions were applied as a basis. The drugs most frequently involved in moderate (Category III and IV) pDDIs were ACE inhibitors with the drug substances ramipril and lisinopril. Drugs acting on the cardiovascular system accounted for 41.9 % of potential drug interactions at hospital admission and 36.5 % for hospital discharge. Those drugs acting on the central nervous system accounted for 8.11 % at hospital admission and 8.9 % at hospital discharge.

The majority of the 288 different pDDIs were postulated to occur through a pharmacodynamic mechanism, (n = 194, prevalence 67.4 %), followed by a pharmacokinetic (n = 79, prevalence

27.4 %) mechanism. The number of potential drug interactions for which the mechanism was unknown, accounted for n = 15 (5.2 %) for the 288 potential drug interactions. A comparison of the frequency with which these drugs were responsible for drug interactions with the overall prescribing frequency in the study cohort reveals that only few substance groups were disproportionately often involved in potential drug interactions in relation to their prescribing frequency.

Figure 4-17 reveals that the frequent involvement of ACE inhibitors, diuretics and anticoagulants in potential drug interactions can be related to their high prescription frequency. It be- comes obvious that supportive medications are not frequently involved in potential drug interactions, although they are prescribed in high numbers especially at hospital discharge (supportive radiooncologic medication, ulcer prophylaxis). The numbers for hospital admission and discharge (Figures 4-17) slightly increased for ACE inhibitors and diuretics. The frequency of interactions increased in comparison to hospital admission.

ACE inhibitors Anti-platelet agents ADPF Beta-blockers pDDI Calcium channel blockers Diuretics Potassium-sparing diuretics Oral anticoagulants Other antihypertensives Sympathomimetics Ulcer prophylaxis Supportive radiooncologic medication Supportive GI medication Oral antidiabetics Insulin Corticosteroids Tricyclic a ntidepressa nts Selective serotonin reuptake inhibitors Neuroleptics Opioids

0 20406080100120 Frequency

ACE inhibitors * Anti-platelet agents DPF Beta-blockers pDDI Calcium channel blockers Diuretics * Potassium-sparing diuretics Oral anticoagulants Other antihypertensives Sympathomimetics Ulcer prophylaxis Supportive radiooncologic medication Supportive GI medication Oral antidiabetics Insulin Corticosteroids * Tricyclic antidepressants Selective serotonin reuptake inhibitors Neuroleptics Opioids 0 20 40 60 80 100 120 Frequency

Figure 4-17. Most frequently prescribed medications which have been involved in potential drug interactions. The black bars indicate an involvement in potential drug interactions and the white bars present the frequency of prescriptions in the patient cohort at the time point of hospital admission (top) and (below) hospital discharge. DPF = discharge prescription frequency, ADPF = admission prescription frequency, pDDI = potential drug drug interaction (*: increased pDDI rate at discharge).

For Category I (recommendation: contraindicated) no pDDIs were identified. For Category II (recommendation: contraindicated as a precaution) 3 different pDDIs were reported for 3 patients regarding the combination of anticoagulants (e.g. clopidogrel) and PPIs (e.g. omeprazole) and opioid agonists (e.g. morphine) versus opioid partial agonists/antagonists (e.g. buprenorphin). In all three potential drug interactions the underlying mechanism was a pharmacokinetic mechanism.

The drugs most commonly involved in potential drug interactions of Category III were NSAIDs and antihypertensives (n = 19) including NSAIDs and ACE inhibitors (n = 8), NSAIDs and diuretics (n = 6) and NSAIDs and other antihypertensive drugs (like beta-blockers, AT1 antagonists) (n=5) predominantly prescribed at hospital discharge. The most frequent NSAIDs at risk of drug interactions involved ibuprofen (35.5 %), etoricoxib with a prevalence of 12.9 % and 9.7 % for diclofenac. The relevant ACE inhibitors were ramipril, captopril and enalapril. The most common pharmacological mechanism underlying the potential drug interactions was pharmacodynamic (n = 19).

In Category IV (recommendation: monitoring or adjustment necessary) the four most frequently identified pDDIs were related to non-steroidal anti-inflammatory drugs (NSAIDs) and cardio-

vascular medications: a) the combination of corticosteroids and NSAIDs [12 patients (10 %)], b) ACE inhibitors and xanthine oxidase inhibitors [10 patients (8.3 %)], c) beta-blockers and beta symphatomimetic agents [5 patients (4.2 %)] and d) cardiac glycosides and diuretics [7 patients (5.8 %)]. Detailed information is given in Annex B1 (Table B-2. Overview of the most frequently detected potential drug interactions of Category IV).

The most common Category V pDDIs (recommendation: monitor as a precaution) were related to: the combinations of a) ACE inhibitors and diuretics [16 patients (13.3 %)], b) diuretics and steroids [12 patients (10 %)], c) beta-blockers and calcium-antagonists [9 patients (7.5 %)] and d) antidiabetics and ACE inhibitors [8 patients (6.7 %)]. Detailed information is given in Annex B1 (Table B-3. Overview of the most frequently detected potential drug interactions of Catego- ry V and VI.). The number of interactions including supportive medications increased from 44 (44/191 (23.1 %)) up to 94 (94/275 (34.2 %)). The study revealed that corticosteroids in total (288) were responsible for 14.2 % of pDDIs due to premedication (n = 41). A subgroup analysis indicated that they belonged primarily to Category IV (n = 20) and Category V (n = 21) (corticosteroids and NSAIDs, corticosteroids and diuretics). Figure 4-18 gives an overview over the most frequently involved pDDIs due to supportive medications. Figure 4-18 implicates that antiinfectives like antimycotics and antibiotics induce potential drug interactions at hospital discharge.

Corticost. + diuretics Corticost. + NSAIDs Corticost. + antidiabetics SSRIs + MCP Admission PPIs + antibiotics Discharge SSRIs + NSAIDs PPIs + antimycotics Corticost. + immunosup. Corticost. + hormones Corticost. + antiepileptics Antihypertensives + NSAIDs CSH + azole antimycotics Opioids+ azole antimycotics PPI + protein kinase inhibitors Corticost. + anticoagulants Corticost. + cardiac glycosides Antidiabetics + NSAIDs Antidepressants + H1 blockers

0481216 Frequency

Figure 4-18. Potential drug interactions between supportive and chronic medications as well as supportive and supportive medications. Corticost. = Corticosteroids, NSAIDs = non-steroidal anti-inflammatory drugs, SSRIs = selective serotonin reuptake inhibitors, PPIs = proton pump inhibitors, immunosup = immunosuppressants.

In the study the prevalence of pDDIs in radiooncologic patients treated with oral anticancer drugs was very low, with a prevalence of 3.3 %. The prescription frequency of the following drugs at hospital admission and discharge were cytostatics (admission (2.5 %); discharge (4.2 %); hormones (e.g. Tamoxifen) (admission: 8.3 %, discharge: 9.16 %); tyrosine kinase inhibitors (erlotinib) admission = discharge: 1.7 %) and selective estrogen receptor modulators SERMs (admission = discharge: 3.3 %). Table 4-7 shows an overview of the interactions most frequently detected.

Table 4-7. Prevalence of pDDIs in radiooncologic patients treated with oral anticancer drugs. Potential drug interactions involving anticancer Frequency Category ABDA No. Description of adverse reaction drugs

Tamoxifen (SERM) + Fluoxetin (SSRI) 1 IV 1174 Drug combinations can prolong QT interval

Tamoxifen (SERM) + Phenprocoumon (antico- 1 IV 408 Bleeding may occur due to an agulants) increase of hypoprothrombinemic effects

(Es)omeprazole (PPI) + Erlotinib (tyrosine 2 IV 1144 Proton pump inhibitors may kinase inhibitor) decrease the plasma concentration of tyrosine kinase inhibitors

Phenytoin (anticonvulsant) + Temozolamid 1 IV 431 Anticonvulsants may decrease (cytostatic drugs) the plasma concentration of cytostatic drugs

Abbreviations; SERM: Selective estrogen receptor modulator; SSRI: Selective serotonin receptor reuptake inhibitor, PPI: Proton pump inhibitor.

Table 4-8. Eight most frequently prescribed drug classes in the patient cohort and the most prevalent clinically relevant potential drug-drug interactions (pDDIs) identified in radiooncologic patients.

Drug or drug class Number Number of Interacting drugs Category Mechanism Expected ADR due to DDI of pDDIs patients reported affected n (%) n (%) ACE inhibitors 19 (6.6) 16 (13.3) Loop/thiazide diuretics V Additive effects Hypotension (first dose) 10 (3.5) 10 (8.3) Xanthine oxidase inhibitors IV Unknown Increased risk of immunological reactions, skin reactions 9 (3.13) 9 (7.5) Antidiabetics V Increased insulin sensitivity Hypoglycemia 8 (2.8) 8 (6.7) NSAIDs III Reduction of prostaglandins synthesis Hypertension 6 (2.1) 6 (5.0) Potassium sparing diuretics IV Additive effects Hyperkalemia 2 (0.7) 2 (1.7) NSAIDs IV Additive impairment of kidney function Hyperkalemia/ Renal failure Antidepressants, anti- 5 (1.7) 5 (4.2) Antiparkinson agents V Additive effects Neuroleptics block dopamine psychotics receptors-increase of extrapyramidal symptoms 5 (1.7) 3 (2.5) Dopamin-antagonists e.g. metoclopra- IV Additive effects Increased risk for extrapy- mide ramidal side effects 1 (0.3) 1 (0.8) Other anticholinergic drugs IV Additive anticholinergic effects Xerostomia, blurred vision, urinary retention, confusion, tachycardia, delirium, cognitive impairment Antihypertensives 11 (3.8) 9 (7.5) Beta-blockers + calcium antagonists V Additive effects Increased antihypertensive effect, hypotension

8 (2.8) 5 (4.2) Beta-2-symphatomimetics + beta- IV Antagonistic effects Decreased effect of beta- blockers sympathomimetic agents

7 (2.4) 7 (5.8) Beta-2-symphatomimetics + diuretics V Additive effects Hypokalemia 5 (1.7) 5 (4.2) NSAIDs III Reduction of prostaglandins synthesis Hypertension 6 (2.1) 6 (5.0) Diuretics, NSAIDs III Reduction of prostaglandins synthesis Hypertension

Drug or drug class Number Number of Interacting drugs Category Mechanism Expected ADR due to DDI of pDDIs patients reported affected n (%) n (%) Cardiac glycosides 7 (2.4) 7 (5.8) Loop/thiazide diuretics IV Enhanced inhibition of NA-K-ATP áse Digoxin toxicity Corticosteroids 15 (5.2) 12 (10) Diuretics V Water and sodium retention (additive effects) Risk of hypokalemia 14 (4.9) 12 (10) NSAIDs IV Damage of the gastric mucosa, steroids mask Increased risk of GI bleeding symptoms of mucosa damage 2 (0.7) 2 (1.7) Antiepileptics(hydantoins) IV CYP 450 induction Decreased efficacy of corticosteroids Opioid agonists 1 (0.3) 1 (0.8) Opioid partial agonists/antagonists II Competitive inhibition Reduced analgesic effect with withdrawal symptomes Oral anticoagulants 5 (1.7) 5 (4.2) Low dose ASS, clopidogrel, NSAIDs IV Additive effects (platelet aggregation inhibition, Bleeding gastric erosion for NSAIDs) 4 (1.4) 4 (3.3) PPI e.g omeprazole, esomeprazole V Increase of the hypoprothrombinemic response Bleeding to phenprocoumon 4 (1.4) 4 (3.3) Antidepressants e.g citalopram V Inhibition of the oxidative metabolism, Bleeding decreased metabolism 2 (0.7) 2 (1.7) PPIs e.g. esomeprazole, omeprazole II Unknown effect Bleeding Oral antidiabetics 4 (1.4) 4 (3.3) Diuretics V Unknown effect Increased blood glucose lowering effect with the risk of hypoglycemia 4 (1.4) 4 (3.3) Corticosteroids IV Increased gluconeogensis, Decreased blood sugar low- Decreased insulin-sensitivity ering effect, risk of hyperglycemia 3 (1) 3 (2.5) Beta-blockers V Masking of a hypoglycemic effect Severe hypoglycemia 3 (1) 3 (2.5) Thyroid hormones V Reduced glucose induced insulin secretion Increased risk of hyperglycemic episodes ACE = Angiotensin converting enzyme, PPIs = proton pump inhibitors, ASS = acetylsalicylic acid

4.4.5 Score predicting the interaction potential The drugs prescribed to the entire study cohort were classified according to a score firstly introduced by (Gaertner et al., 2012). (Gaertner et al., 2012) implemented a scoring system related to a traffic light system. This scoring system helps to distinguish substances with a high probability of causing DDI related adverse events (Red flag), from those with a low or moderate interaction potential (Green flag or Yellow flag) (Frechen et al., 2012, Gaertner et al., 2012).

Table 4-9 lists the drugs with a high potential for drug interactions, limited to n = 5 prescription drugs per patient at hospital admission and at hospital discharge. Etoricoxib and amitriptyline were listed additionally because the changes in the prescription frequency at admission and discharge seemed to be of importance. The study of (Gaertner et al., 2012) primarily focused on Category I-IV as clinically relevant drug interactions. A recent article by (Ziegelmeier and Goller, 2013) emphasized that for improving drug safety the announcement of Category IV to V are of utmost importance. Therefore, all categories (I-VI) had been included in the study of this thesis.

Table 4-9 reveals that ACE inhibitors and diuretics were associated with an increased risk of pDDIs at hospital admission and discharge. According to the flag score enalapril exhibited a more favourable DDI profile compared to ramipril, captopril and lisinopril. Steroids like prednisolone and dexamethasone frequently used as premedication prior to radiation therapy were associated with a high risk inducing relevant potential drug interactions (Table 4-9). A comparison of the listed tables reveals that opioids (Oxycodone, morphine), the prescription frequency of which increased significantly at hospital discharge, did not account for high numbers of pDDIs. In contrast to that classical NSAIDs like Ass, diclofenac and ibuprofen had a very high risk for pDDIs. The non opioid analgesic metamizol which was prescribed as pain treatment frequently in radiooncologic patients (40.8 % at discharge) was not involved in one single potential drug interaction. Anticoagulants like phenprocoumon were associated with a high risk for pDDIs due to an increased risk of bleeding complications. In our study antiemetics acting via the dopamine pathway (e.g metoclopramide) and of the antihistamine group (diphenhydramine) exhibited a low risk for pDDIs (Table 4-10). Laxatives like bisacodyl and PPIs, especially useful as coadministered drugs in opioid therapy and PPIs as gastric protection in steroid therapy had a low interaction potential (Frechen et al., 2012). Furthermore Table 4-11 reveals that neuroleptics (haloperidole) and antidepressants especially SSRIs (venlafaxine, sertraline) accounted for very small numbers of potential drug interactions (green flag score) at hospital admission in comparison to other studies (Frechen et al., 2012, Gaertner et al., 2012).

Table 4-9. Drugs with a high potential for potential drug interactions (red flag score (ratio ni/np × 100 ≥ 50) adopted from (Frechen et al., 2012, Gaertner et al., 2012). Drug Admission Discharge Patients DDI Ratio Flag Patients DDI Ratio Flag a a affected classes ni/np score affected classes ni/np score I-VI I-VI np ni np ni Dexamethasone 20 22 1.100 110.0 34 30 0.882 88.2 Metoprolol 20 27 1.350 135.0 24 36 1.500 150.0 Ass 19 20 1.052 105.2 24 23 0.958 95.8 Ramipril 18 27 1.5 150.0 22 30 1.363 136.4 Allopurinol 15 8 0.533 53.4 15 12 0.8 80 Furosemide 14 11 0.785 78.6 18 18 1 100 Formoterol 11 10 0.909 90.90 11 12 1.090 109.1 HCT 9 11 1.222 122.2 10 17 1.7 77.8 Lisinopril 8 7 0.875 87.5 9 7 0.777 77.8 Metformin 8 8 1 100 6 10 1.66 166.7 Ibuprofen 7 14 2 200 8 16 2 200 Torasemide 7 16 2.28 228.6 9 15 1.66 166.7 Phenprocoumon 6 17 2.83 283.33 7 21 3 300 Prednisolone 6 8 1.33 133.3 5 8 1.6 160 Captopril 5 10 2 200 5 13 2.6 260 Diclofenac 5 4 0.8 80 6 6 1 100 Glimepiride 5 3 0.6 60 3 6 2 200 Insulin 5 7 1.4 140 8 9 1.125 112.5 Valproic acid 5 4 0.8 80 3 4 1.33 133.3 Amitriptyline 2 3 1.5 150 10 6 0.6 60 Etoricoxib 1 2 2 200 5 14 2.8 280 a Flag score: number of class I-VI pDDIs per patients exposed (ratio ni/np × 100). HCT = hydrochlorothiazide

Table 4-10. Drugs with a low potential for potential drug interactions (green flag score (ratio ni/np × 100 = 0-25) Drug Admission Discharge Patients DDI Ratio Flag Patients DDI Ratio Flag a a affected classes ni/np score affected classes ni/np score I-VI I-V np ni np ni Pantoprazole 32 4 0.125 12.5 68 11 0.161 16.17 Omeprazole 20 3 0.15 15 13 3 0.230 23.07 Metoclopramide 15 1 0.066 6.66 39 5 0.128 12.82 Morphine 9 0 0 0 26 1 0.038 3.846 Budesonide 7 1 0.142 14.28 7 1 0.142 14.28 Natriumpicosulfat 5 1 0.2 20 0 1 0 0 a Flag score: number of class I-VI pDDIs per patients exposed (ratio ni/np × 100).

Table 4-11 presents different scores for specific medications at hospital admission and discharge. A comparison of the score reveals that the scores for antidepressants (e.g. sertraline, venlafaxine (SSRIs) were subject to fluctuations. Although the number of patients affected were similar at hospital admission and discharge, the number of pDDIs increased from 1 to 4 at hospital discharge. This indicates that in case SSRIs are combined with an inappropriate medication the risk for potential drug interactions can ultimately change from low to high. A subgroup analysis revealed that NSAIDs (Ibuprofen, etoricoxib) and supportive medications (metoclopramide) were involved in the potential drug interactions at hospital discharge.

Table 4-11. Drugs with a changing potential for potential drug interactions (flag score). Framing of the numbers of the flag score with three lines means red flag, framing with 2 lines yellow flag and framing with one line green flag score. Drug Admission Discharge Patients DDI Ratio Flag Patients DDI Ratio Flag score a affected classes ni/np score affected classes ni/np I-VI I-V np ni np ni Alendronic acid 4 2 0.5 50 5 2 0.4 40 Benzodiazepines 5 4 0.8 80 11 4 0.363 36.36 Ciprofloxacin 4 3 0.75 75 8 3 0.375 37.5 Enoxaparin 10 5 0.5 50 15 4 0.266 26.67 Repaglinide 1 1 1 100 1 0 0 0 Esomeprazole 4 2 0.5 50 9 2 0.22 22.22 Amlodipine 9 4 0.44 44.44 9 6 0.666 66.66 Calcium 4 1 0.25 25 2 7 3.5 350 Candesartan 3 1 0.33 33.33 4 2 0.5 50 Enalapril 7 2 0.258 28.57 8 4 0.5 50 Levotyroxine 14 6 0.428 42.85 19 11 0.578 57.89 Magaldrate 8 2 0.25 25 14 10 0.714 71.42 Nifedipine 4 1 0.25 25 5 3 0.6 60 Bisoprolol 4 0 0 0 4 1 0.25 25 Carvedilol 5 1 0.2 20 7 2 0.285 28.57 Erlotinib 1 0 0 0 1 2 2 200 Haloperidole 1 0 0 0 1 1 1 100 Oxycodone 3 0 0 0 8 2 0.25 25 Sertraline 1 0 0 0 1 4 4 400 Venlafaxine 1 0 0 0 1 1 1 100 a Flag score: number of class I-VI pDDIs per patients exposed (ratio ni/np × 100).

4.4.6 Effects induced by potential drug interactions The ABDA system reported for 120 patients 191 theoretically possible DDIs at hospital admission and 275 at hospital discharge. An overall analysis revealed that 108 of the detected potential clinical drug interactions at hospital admission and 164 at hospital discharge required clinical attention (Category I-IV) while 1-2 % of the potentially drug interactions needed to be assessed as life threatening (e.g. parallel administration of ACE inhibitors and potassium increased risk of hyperkalemia). A subgroup analysis revealed that 159 of the potential drug interactions could be detectable by measuring laboratory parameters or identifying clinical symptoms (bradycardia, hypotension, fluctuations in blood sugar levels). 129 pDDIs could not be verified due to an underlying pharmacodynamic mechanism with antagonism and synergism.

Figure 4-19 depicts the potential effects [%] which might have been caused by potential drug interactions in the study. Besides influencing the efficacy of drug treatment (increased or decreased effect) the most frequent side effects caused by potential drug interactions in our study were: hypotension/bradycardia (10.76 %), influence on blood sugar levels (hypoglycemia, hyperglycemia, 8.7 %). gastrointestinal bleeding (8.68 %) complications and hypokalemia (8.3 %). With a prevalence of 4.51 % patients were at risk of arrhythmias (tachycardia/bradycardia). The analysis of the admission diagnosis in the discharge letters showed that none of the patients had

been admitted due to gastric bleeding complications. 16.6 % of the patients showed signs of arrythmic potentials in their 12-channel-ECGs.

Reduced effect Increased effect Hypotension/Bradycardia Gastrointestinal bleeding Hypokalemia Hypoglycemia/Hyperglycemia Hypercalcemia Risk of immunological reactions Increased absorption of aluminium Increased incidences of myopathies Bleeding risk increased Ventricular tachycardia Serotonin syndrome-eytrapyramidale side effects Hypertension Additive cardiodepressive effect Bronchoconstriction Seizures

0 5 10 15 20 25 Potentia l drug intera ctions [%]

Figure 4-19. Effects which might have been induced by potential drug interactions in the study cohort.

Fig. 4-20 depicts that some patients were affected several times by a potential drug interactions with certain effects (GI bleeding, hypokalemia). In case of GI bleeding 4 patients (3.3 %) were affected by two concurrent potential drug interactions both leading to an increased risk of GI bleeding complications. In these patients the risk of manifest consequences increases significantly. A similar distribution was found for the effect “hypokalemia”. 2 patients received between three to six inappropriate drug combinations concurrently leading to hypokalemia.

7 6 5 4 3 2 1 6 5 4 3 2

Hypokalemia bleeding GI 1 7 6 5 4 3 2 Bradycardia Hypotension/ Hypotension/ 1 7 6 5 4 3 2

Hypoglycemia/ Hypoglycemia/ Hyperglycemia 1 7 Effect of potential drug interactions drug of potential Effect 6 5 4

effect 3

Increased 2 1 7 6 5 4 3 2

Reduced effect 1 0 2 4 6 8 10 12 14 16 18 Number of patients affected

Figure 4-20. Distribution of effects of potential drug interactions among the patient collective; numbers = number of potential drug interactions with the same effect per patient.

Table 4-12 gives an overview of the most frequent drug combinations which occurred in the study collective stratified according to induced effects. Two drug combinations were primarily identified leading to hypotension and bradycardia. ACE inhibitors and diuretics (10 %) and beta-blockers and calcium channel blockers (7.5 %).

Table 4-12. Most frequently involved drug pairs and possible induced effects.

Drug pairs No. of Patients Drugs frequently involved Effect of drug combina- pDDIs affected tion reported.

Cardiac glycosides + diure- 7 7 (5.8 %) Digitoxin + furosemide tics Increased effect Anticoagulants + PPIs 4 4 (3.3 %) Clopidogrel + omeprazole

NSAIDs + antihypertensives 9 7 (5.8 %) Ramipril, metoprolol + Decreased effect

Drug pairs No. of Patients Drugs frequently involved Effect of drug combina- pDDIs affected tion reported.

etoricoxib, ibuprofen

Beta2 sympathomimetic 8 5 (4.2 %) Salbutamol, fenoterol + agents + beta-blockers metoprolol

ACE inhibitors + NSAIDs 8 8 (6.7 %) Ramipril + ibuprofen Decreased effect

Thyroid preparations + cati- 6 5 (4.2 %) Levothyroxine + magaldra- ones te

Neuroleptics + dopamin- 5 3 (2.5 %) Melperon + levodopa agonists

Diuretics + NSAIDs 4 4 (3.3 %) Furosemide + ibuprofen

Antidiabetics + ACE inhi- 9 9 (7.5 %) Metformin + captopril bitors Hypoglycemia ACE inhibitors + diuretics 3 3 (2.5 %) Ramipril + triamterene

ACE inhibitors + diuretics 16 13 Ramipril; furosemide (10.8 %)

Beta-blockers + calcium 11 9 (7.5 %) Metoprolol + Hypotension, Brady- channel blockers amlodipine, nifedipine cardia

Cardiac glycosides + beta- 4 4 (3.3 %) Digitoxin + metoprolol blockers

ACE inhibitors + potassium- 6 6 (5 %) Ramipril + Hyperkalemia sparing diuretics triamterene, spironolactone

Aluminium-salts + citric acid 6 6 (5 %) Aluminiumoxid + Citric Renal impairment acid

Corticosteroids + diuretics 15 12 (10 %) Dexamethasone + furosemide Hypokalemia Diuretics + sympathomime- 7 7 (5.8 %) Torasemide + salbutamol tic agents

Corticosteroids + NSAIDs 14 12 (10 %) Dexamethasone + ibuprofen, Gastrointestinal blee- NSAIDs + SSRIs 4 3 (2.5 %) Diclofenac + citalopram ding complications Anticoagulants + NSAIDs 5 5 (4.2 %) Phenprocoumon + diclofenac, etoricoxib ACE: Angiotensin converting enzyme inhibitor; PPIs = proton pump inhibitor, SSRI = selective serotonin reuptake inhibitor.

4.5 Potentially inappropriate medications

4.5.1 Cave module: Frequency of IIMs and correlation analysis The following section gives an overview of all individual inappropriate medications prescribed to the patient cohort analyzed via the Cave module. To calculate IIMs in total, IIMs which occurred in the admission and discharge medication were only counted once. The distribution of age used for calculation analysis was differing from the age categories used for pDDIs due to statistical reasons.

Table 4-13. Frequency of IIMs in total, differentiated in IIMs contraindicated and restriction on use. Variables Value Frequency %* Number of IIMs (in total) 0 40 33.3 1 5 4.2 2 16 13.3 3 13 10.8 4 7 5.8 5 11 9.2 6 12 10.0 7 7 5.8 8 5 4.2 9 2 1.7 12 2 1.7 Number of IIMs (Contraindicated) 0 89 74.2 1 19 15.8 2 7 5.8 3 3 2.5 4 2 1.7 Number of IIMs (Restriction on use) 0 40 33.3 1 6 5.0 2 20 16.7 3 15 12.5 4 11 9.2 5 8 6.7 6 11 9.2 7 5 4.2 8 1 0.8 9 2 1.7 10 1 0.8 * percentages relate to the number of valid values of each variable

80 patients (66.7 %) had been prescribed an IIM. In total 362 individual inappropriate medications had been prescribed. 327 IIMs were analyzed in detail. 31 (25.8 %) patients received in total 50 medications which were contraindicated due to an underlying disease or due to age. 312 medications were prescribed for 80 (66.6 %) patients belonging to the Category: Restrictions on use. Table 4-14 shows that there was no significant gender difference in the number of patients receiving no IIMs [35.3 % females, 31.8 % males]. More man (40.9 % versus 25.9 %) received between one to four IIMs while slightly more females received more than 4 IIMs. According to the CAVE module 48.4 % of all patients aged 60-69 received between 1-4 IIMs. 39 % of the patients aged 70-79 received more than 4 IIMs. This percentage further increased up to 62 % in

the patients aged > 79. These values show the clear tendency that over the age of 70 there exists a significant increase in IIM prescription (p = 0.001).

Table 4-14. Frequeny of variables (Gender, age, number of comorbid diseases) in relation to the number of IIMs. Variable Value no IIM 1 - 4 IIMs > 4 IIMs No. %* No. %* No. %* Gender Female 19 35.2 14 25.9 21 38.9 Male 21 31.8 27 40.9 18 27.3 Age (years) < 59 15 55.6 8 29.6 4 14.8 60 - 69 10 32.3 15 48.4 6 19.4 70 - 79 13 31.7 12 29.3 16 39.0 > 79 2 9.5 6 28.6 13 61.9 Number of comorbid diseases 0 7 77.8 1 11.1 1 11.1 1 13 46.4 11 39.3 4 14.3 2 7 24.1 12 41.4 10 34.5 3 5 26.3 9 47.4 5 26.3 > 3 8 22.9 8 22.9 19 54.3 * The percentages refer to IIM shares of each Category within the values of the variables (line sum = 100 %).

For all variables except gender (p = 0.195) statistical differences have been found. With increasing age (p = 0.003), number of comorbid diseases (p = 0.005) and the number of medications (p < 0.001) the proportion of patients receiving IIMs increased. In a subgroup analysis the statistical correlations were calculated separately for the number of contraindicated IIMs and IIMs with restriction on use.

Table 4-15. Frequency of variables (Gender, age and number of comorbid diseases) in relation to the number of contraindicated IIMs. Variables Value No contraindicated At least one contra- IIM indicated IIM No. %* No. %* Gender Female 37 68.5 17 31.5 Male 52 78.8 14 21.2 Age (years) < 59 23 85.2 4 14.8 60 - 69 28 90.3 3 9.7 70 - 79 26 63.4 15 36.6 > 79 12 57.1 9 42.9 Number of comorbid diseases 0 8 88.9 1 11.1 1 25 89.3 3 10.7 2 24 82.8 5 17.2 3 13 68.4 6 31.6 > 3 19 54.3 16 45.7 * The percentages refer to IIM shares yes/no within the values of the variables (line sum = 100 %).

Table 4-15 shows that with increasing age (p = 0.015), number of comorbid diseases (p = 0.013) and the number of medications (p = 0.007) the proportion of patients receiving contraindicated PIMs tendentially increased. This development was not consistent over all categories of variables. For all variables except gender (p = 0.216) differences were found.

Table 4-16. Frequency of variables (Gender, age and number of comorbid diseases) in relation to the number of IIMs restricted on use Variables Value No restrictions on 1 - 3 restrictions > 3 Restriction on use IIM on use IIMs use IIMs No %* No %* No %* Gender Female 19 35.2 14 25.9 21 38.9 Male 21 31.8 27 40.9 18 27.3 Age (years) < 59 15 55.6 8 29.6 4 14.8 60 - 69 10 32.3 15 48.4 6 19.4 70 - 79 13 31.7 12 29.3 16 39.0 > 79 2 9.5 6 28.6 13 61.9 Number of 0 7 77.8 1 11.1 1 11.1 comorbid- 1 13 46.4 11 39.3 4 14.3 diseases 2 7 24.1 12 41.4 10 34.5 3 5 26.3 10 52.6 4 21.1 > 3 8 22.9 7 20.0 20 57.1 * The percentages refer to IIM shares of each Category within the values of the variables (line sum = 100%.

A correlation analysis of IIMs with restriction on use revealed similar results. For all variables except gender (p = 0.199) differences were found. With increasing age (p = 0.002), number of comorbid diseases (p = 0.001) and the number of medications (p < 0.001) the share in patients receiving IIMs (restriction on use) increased (Table 4-16).

Table 4-17 shows the difference in the IIM prescribing pattern between hospital admission and discharge. The number of patients receiving contraindicated IIMs due to their underlying disease increased marginally at discharge (26 (21.7 %) versus 30 (25 %) patients). The prevalence of patients receiving IIMs restricted on use increased from n= 60 (50 %) up to n= 68 (56.7 %). None of the patients received contraindicated medications due to age according to the Cave module. Most of the patients received only one single contraindicated medication at hospital admission and discharge and with 1.7 % the patients receiving more than 3 contraindicated medications due to their underlying disease were quite rare. Medications mostly contraindicated in the collective were antihypertensive agents e.g. beta-blockers, loop diuretics, blood thinners, and antidepressants / anti-psychotics (see Annex B 2).

Table 4-17. Frequency of number of IIMs at hospital admission and discharge. Variables Value Hospital admission Hospital discharge n %* n %* Number of contraindicated IIMs due to 0 94 78.3 90 75.0 disease 1 15 12.5 21 17.5 2 9 7.5 7 5.8 4 2 1.7 2 1.7 Number of contraindicated IIMs due to 0 120 100.0 120 100.0 age Number of IIMs restriction on use due to 0 60 50.0 52 43.3 disease 1 23 19.2 17 14.2 2 12 10.0 21 18.3 3 13 10.8 13 10.0 4 5 4.2 8 6.7 5 2 1.7 3 2.5 6 2 1.7 3 2.5

Variables Value Hospital admission Hospital discharge n %* n %* 7 2 1.7 1 0.8 9 1 0.8 2 1.7 Number of IIMs restriction on use due to 0 72 60.0 62 51.7 age 1 19 15.8 23 19.2 2 16 13.3 16 13.3 3 11 9.2 13 10.8 4 2 1.7 6 5.0 * The percentages refer to IIM shares admission/discharge within the values of the variables (line sum = 100 %).

A subgroup analysis indicated that antihypertensive drugs, NSAIDs, steroids and opioids were most frequently responsible for drug disease interactions. 18.3 % of the prescribed IIMs were antihypertensives, 14.1 % dietary supplements, 11.3 % NSAIDs and 10.4 % corticosteroids. Figure 4-21 indicates that beta-blockers (e.g. metoprolol, carvedilol) (48.3 %), ACE inhibitors (35 %) mainly ramipril, calcium-antagonists (e.g. amlodipin) (11.67 %) were most frequently involved in inappropriate drug prescribing.

Antihypertensives Moxonidine Dietary supplements Candesartan Losartan HCT NSAIDs Lerca nidipine Steroids Nifedipine Opioids Amlodipine Sedatives Enalapril Anti-platelet agents Ramipril Diuretics Propranolol GI medication Bisoprolol Atenolol Antidepressants Timolol Inhalers Celiprolol Antipsychotics Metoprolol Antidiabetics Carvedilol

0 102030 0102030 Prescriptions [%] Prescriptions [%]

Figure 4-21. Medication groups most frequently involved in IIM prescribing [%] (left). Antihypertensive drugs most frequently involved in IIM prescribing [%] (right).

Most drugs were used inappropriately due to advanced age. 129 IIMs (39.4 %) should not have been administered to the patient cohort because patients were aged over 65-75 years (Figure 4- 22). 82 IIMs (25 %) were used inappropriately in patients with under-lying metabolic dysfunction (diabetes mellitus, renal impairment). Drug disease interactions considering cardiovascular diagnosis and conditions occurred with a prevalence of 13.1 % (Figure 4-22).

Age

Endocrinological disease

Cardiovascular disease

Gastroenterologic disease

Underlying disease Underlying Pulmonary disease

Neurological disease

0 10203040 Individual inappropriate medication [%]

Figure 4-22. Individual inappropriate medications due to underlying diseases.

Table 4-18 shows the most frequent side effects induced by therapy with the listed inappropriate drugs in patients aged ≥ 65. Steroids lead to a deterioration or induction of osteoporosis because corticosteroids have several adverse effects on bone metabolism (Direct inhibition of osteoblast function, direct enhancement of bone resorption inhibition of gastrointestinal calcium absorption, increases in urine calcium loss). This increases the risk for falls and hip fractures. Patients aged over 65 are prone to paradoxical reactions of antidepressants and antipsychotics.

Table 4-18. Inappropriate individual medications due to age inappropriateness detected by the Cave module.

Potential induced effects of inappropriate drugs Drugs Frequency* Deterioration or induction of osteoporosis Dexamethasone 16 Prednisolone 1 Increased risk of rhabdomyolysis Simvastatin 11 Increased risk of extrapyramidale side effects, alterations of Haloperidol 2 the conduction sytem of the heart Promethazine 1 Heparin 3 Delayed excretion of metabolites Novaminsulfon 18 Dose adjustment Amlodipin 6 Bisoprolol Morphin Doxepin 9 Trimipramin Citalopram Capecitabin 1 Theophyllin 1 Allopurinol 5 Increased risk of side effects (paradoxic side effects, agita- Amitriptylin 22

Potential induced effects of inappropriate drugs Drugs Frequency* tion, aggressiveness increased incidences of falls) Levomepromazin Zoplicon Lorazepam Lorazepam Induction of respiratory depression Morphin Oxycodon 6 Fentanyl Increased sensibility to medication efficacy Ramipril, Fentanyl 14 Glibenclamid Cardiotoxicity, electrolyte imbalance Digitoxin 2 Xipamide Deterioration of kidney function, fluid retention, increased NSAIDs (etoricoxib, 4 bleeding risk diclofenac) *Frequency of IIMs implies the potential frequency; no statement can be made regarding the real manifestation of side effects

Table B-4 (Annex B 2) gives an overview over the drug disease interactions considering endocrinologic and metabolic diagnosis and conditions. Diabetes and Renal failure were the main primary diseases leading to drug disease interactions. 40 medications were inappropriate due to an underlying diabetes mellitus. In the patient collective 25 IIMs caused alterations in blood glucose levels due to an underlying diabetes mellitus. The drugs which can mainly cause blood sugar alterations (hyperglycemia as well as hypoglycemia) were antihypertensive drugs (40 %), corticosteroids (28 %), NSAIDs (12 %) and medications for psychiatric disorders (20 %). Blood sugar alterations with hypoglycemic episodes put the patients at risk of falls due to dizziness. In diabetic patients receiving antihypertensive drugs like furosemide, ramipril or heparin (enoxaparin) the potassium levels need to be closely monitored according to the CAVE module. Patients with known hyperuricemia additional treated with diuretics are at risk of experiencing acute gout attacks. 3 patients (2.5 %) received beta-blockers which might mask symptoms of hypoglycemia.

32 medications should not have been prescribed to patients with renal insufficiency (15.8 %). These medications should only be administered when close monitoring of the creatinin clearance can be guaranteed. Under a clearance of 30-60 ml/min the administration of these drugs should have been avoided or adopted in dosing (Table B-4 Annex B.2).

Table B-5 (Annex B 2) gives an overview over the drug disease interactions considering cardiovascular diagnosis and conditions. In patients with hypertension corticosteroids can lead to or- thostatic dysregulation with episodes of hypotension. The use of levotyhroxine and testogel need as a prerequisite a well programmed blood pressure, otherwise the use should be avoided.

The use of beta-2-symphatomimetics in patients with hypertension requires the monitoring of the heart rate (risk of tachyarrhythmia). 6 patients (5 %) with underling coronary artery disease were at an increased risk of thromboembolic events in case of NSAID intake (e.g. ibuprofen, diclofenac). Thyroid preparations and novaminsulfon, a frequently used pain medication, might lead to an enhancement of CAD symptoms manifesting in shortness of breath and typical angina pectoris symptoms. As Table B-5 (Annex B.2) shows beta-2-sympathomimetic agents were inappropriate in patients with arrhythmia because these drug substances might induce tachyarrhythmia and put the patient at additional risk for cardiovascular side effects like strokes and myocardial infarction.

Table B-6 (Annex B 2) gives an overview over the drug disease interactions considering gastro- enterological, pulmonological, neurological and urological diagnosis and conditions. 6 patients (5 %) with underlying ulcers received blood thinning preparations or corticosteroids. Deteriora- tion of existing ulcers with the potential to produce gastric bleeding complications could be a serious consequence. In patients with constipation the administration of opioids (e.g. morphine, oxycodone) and diuretic agents (furosemide, xipamide) is inappropriate due to the fact that the electrolyte imbalance perpetuates the vicious circle of constipation due to hypokalemia and hyponatremia.

4.5.2 Pricus list medications In the patient collective 700 medications were prescribed in total as admission medication and 1024 at discharge. 79 patients were aged ≥ 65 and were included into the evaluation. 3 % of the total medications prescribed at hospital admission and 5.0 % of the medications at discharge turned out to be potentially inappropriate and were listed on the Priscus list. A subgroup analysis revealed that 18 patients (22.8 %) received one drug from the Priscus list at hospital entrance, while 3 patients (3.79 %) received even 2 drugs from the Priscus list. At hospital discharge the number of patients receiving PIMs increased up to 37 (46.8 %) while 8 patients (10.1 %) received > 2 potentially inappropriate medications. 13 patients (16.5 %) received PIMs from the Priscus list at hospital admission and discharge. Figure 4-23 indicates that five medications were of primary significance. The most frequently prescribed PIMs were drugs belonging to the class of antidepressants and hypnotic agents. In the Priscus list most of the medications were rated as PIMs independent of diagnosis or clinical condition. Figure 4-23 reveals that hypnotic and sedative drugs were most frequently used in the study collective. Zoplicon a drug substance with a sedative pharmacological mechanism was prescribed most frequently. 12 patients (15.2 %) were affected by zoplicon prescriptions > 7.5 mg at hospital discharge. Figure 4-23 shows that the prescription rate of amitriptyline changed during admission and discharge. The prescription rate of amitriptyline was five times higher at hospital discharge when at the time of admission.

Amitriptyline Dimenhydrinate Dimetindene Admission Doxepin Etoricoxib Discharge Haloperidol Levomepromazine Lorazepam > 2mg Nifedipine Sotalol Tolterodine Trimipramine Zoplicon 024681012 Frequency

Figure 4-23. Priscus medications detected in the study collective.

A similar trend was seen for etoricoxib prescriptions. Benzodiazepines are another drug class which should not be prescribed according to the Priscus list. For benzodiazepine prescriptions there exists a dose limitation which should not be exceeded. A subgroup analysis showed that 12 patients (15.2 %) received a benzodiazepine (lorazepam), but only 4 patients received it in a dosage over > 2 mg, which is considered to be precarious in elderly patients. 1 patient received it in the admission medication and in 3 cases it was newly prescribed as treatment recommendation after discharge from the hospital. No gender difference could be detected in the study collective regarding prescriptions from the Priscus list. For zoplicon a significant higher prescription frequency could be seen in female patients at hospital discharge. But no statement can be made regarding gender difference because at hospital admission the prescription frequency was exactly the opposite (higher frequency in male patients).

Table 4-19. Prevalence of Priscus related drugs at admission and discharge in the elderly sub-population of n=79 elderly aged ≥ 65. Drug Prevalence in study cohort [%] Admission Discharge All Men Women All Men Women Zoplicon 8.86 5.06 3.08 15.19 2.53 12.66 Nifedipine 5.06 3.80 1.27 6.33 3.80 2.53 Amitriptyline 2.53 0 2.53 12.66 5.06 7.59 Doxepin 2.53 1.27 1.27 2.53 1.27 1.27 Etoricoxib 1.27 1.27 0 0 0 0 Sotalol 1.27 1.27 0 2.53 1.27 1.27 Toltderodine 1.27 0 1.27 2.53 0 2.53 Trimipramine 1.27 1.27 0 1.27 0 1.27 Haloeridol 1.27 1.27 0 1.27 0 1.27 Lorazepam 1.27 0 1.27 5.06 2.53 2.53 Dimetinden 0 0 0 1.27 0 1.27 Dimehydrinat 0 0 0 5.06 3.80 1.27 Levomepromazin 0 0 0 2.53 1.27 1.27

5 Discussion

5.1 Patient population and prevalence of potential drug interactions To the best of my knowledge this is the first study assessing the potential for pDDIs in hospitalised patients in a radiooncology setting. Similar studies were performed in 2012 by (Frechen et al., 2012) in hospice inpatients in Germany and by (Lapi et al., 2010) in a cohort of patients undergoing radiodiagnostic procedures in 2010.

The mean age of the male patients was 68.5 years. The female patients were averaged 69 years. Because the wards used for data collection had their main focus on radiooncologic treatment (cancer related) the proportions of elderly patients were as high as expected. The median age of the sample (70 years) was slightly lower to that reported in other studies conducted in the hospital setting, 75 years in the study by (Glintborg et al., 2005). In the study by (Kannan et al., 2011) the median age was low with a median of 56 years. The median age is dependent on the study sample and the underlying comorbidities. Cancer is primarily a disease of older people, with incidence rates increasing with age for most cancers. More than three out of five (63 %) cancers are diagnosed in people aged 65 and over, and more than a third (36 %) are diagnosed in the elderly (aged 75 and over) (Cancer Research UK, 2010).

The prevalent conditions on the radioonologic ward for example multimorbid patients with cardiovascular diseases lend themselves to use very complex drug regimens with a high interaction potential. A study by (Evans et al., 2005) showed that having three or more comorbidities can increase the risk for severe ADRs by 2, 9-12, 6 fold. The risk of hospitalization and potentially preventable hospitalization strongly increased with the number of chronic conditions (both p < 0.0001) (Braunstein et al., 2003).

In our study the most common comorbidities were cardiovascular diseases (62.5 % of patients), metabolic diseases primarily diabetes (16.7 % of patients) and neurological diseases (22.5 %). A study by (Merrill and Elixhauser, 2005) showed that the leading comorbidities were hypertension (29.4 %), chronic obstructive lung disease (COPD) 12,1 %, diabetes mellitus (11.8 %), fluid/electrolyte disorders (11.7 %) and heart failure (5.7 %). A study by (Riechelmann et al., 2008) focussing on potential drug interactions in cancer patients receiving supportive medications exclusively yielded similar results. The number of psychiatric disorders (e.g. depression) in the study by (Riechelmann et al., 2008) was as low as in our study (5 % versus 5.8 %). The estimated prevalence of depression in 15 subgroup studies in a literature review by (Walker et al., 2013) was as follows: 5 % to 16 % in outpatients, 4 % to 14 % in inpatients, 4 % to 11 % in

mixed outpatient and inpatient samples and 7 % to 49 % in palliative care (Walker et al., 2013). Our study results fall within the range of 4 to 14 % for inpatients.

At hospital discharge the number of patients receiving medications affecting the alimentary tract and metabolism (ATC A) as well as drugs regulating the nervous system (ATC N) were increased (ATC A: Admission: 80.9 %; Discharge: 94.2 %; ATC N Admission: 58.3 %; Dis- charge: 70.9 %). Both drug classes contain supportive medications such as PPIs, stomach protectors and antidepressants. This can possibly be explained by the increase in prescriptions after radioncologic treatment. In the discharge medication patients with prescriptions belonging to the drug class ATC B contains substances which are used to treat anaemia. Anaemia can be a result of radiation therapy (Harrison et al., 2001). (Shasha et al., 2004) showed that 48 % of patients presented to their department with anaemia (defined as haemoglobin <12 g/dl) and a total of 57 % ultimately became anaemic by the end of radiation therapy. In addition (Warde et al., 1998) found evidence in a retrospective study that pre-treatment haemoglobin levels were one of several independent prognostic factors for local failure after radiation therapy. The prem- ise that anaemia decreases tissue radiosensitivity has long been accepted by radiotherapists, and correction of the anaemia has been recommended as a means of increasing the effectiveness of irradiation (O'Brien et al., 1968). At discharge from hospital patients receive anaemia treatment to improve their haemoglobin level for the next cycle of radiation therapy. A study by (Vonbach, 2007) found that the three most prevalent anatomical groups were drugs affecting the alimentary tract and metabolism (admission 46 %, discharge 72 %), drugs affecting the blood and blood forming organs (admission 45 %, discharge 60 %) and drugs affecting the cardiovascular system (admission 61 % discharge 71 %). These values correlate with the incidence rates found in our study except for ATC B (blood forming organs). The prescription frequency of diarrhoea medications is quite low. This fact is astonishing because usually radiation induced diarrhoea is a frequent complication of radiation. Even treatment with probiotics as prevention of radiation induced diarrhoea is scarce.

The mean number of prescription drugs was 6.12 at hospital admission, which was consistent with studies conducted in similar settings (Lao et al., 2013, Vonbach, 2007). The preliminary study showed that the number of medications prescribed per patient increased significantly from admission to discharge while the number of potential drug interactions as well increased. At hospital entry 50 patients (41.7 %) had at least one pDDI. This figure increased at hospital discharge up to 60 %. Studies that have looked at prevalences of interactions among hospitalised patients have yielded similar results (Cruciol-Souza and Thomson, 2006, Egger et al., 2007, Ismail et al., 2013, Vonbach, 2007). A present study by (Sepehri et al., 2012) found that the total frequency of potential drug-drug interactions (both severe and moderate pDDIs) in prescriptions in a general hospital in Zarand City was almost 59.1 %, which was higher than the frequency in Europe (Wysowski and Bacsanyi, 1996). Studies looking at the prescription fre-

quency at hospital discharge estimate this proportion as 41.1- 69.7 % (Fokter et al., 2010). Alt- hough our methodology might have differed from those in others studies, the results fall within the range cited previously. The range of prevalence values may be explained by factors such as methodology, characteristics of the population, number of medications prescribed, prescriber behaviour, and database used to screen drug interactions.

It is problematic that interactions between two drugs can present in clinical practice with various degrees of severity: The interaction between ACE inhibitors and diuretics may remain without clinical symptoms, may lead to an asymptomatic hyperkalemia only detectable via lab tests or can lead to severe cardiac arrhythmia resulting in haemodialysis (Köhler, 2001). This helps to explain the different attitudes and assessments of different publishers.

In our study cardiovascular medications (ACE inhibitors, diuretics, beta-blockers) and anti- platelet agents (Phenprocoumon, ASS) were most frequently involved in potential drug interactions. This data have been confirmed by other studies in hospital and ambulatory settings (Becker et al., 2007, Cruciol-Souza and Thomson, 2006, Ismail et al., 2013, Lao et al., 2013, Magro et al., 2008, Teweleit et al., 2001).

A comparison of the frequency with which drugs were responsible for drug interactions with the overall prescribing frequency in the study cohort reveals that only few substance groups were disproportionately often involved in potential drug interactions in relation to their prescription frequency. It can be assumed that most of the drugs were often involved in potential drug interactions due to their high prescription frequency. In a study by (Köhler, 2001) ACE inhibitors, cardiac glycosides and potassium excreting diuretics occurred disproportionally often in potential drug interactions. In the study by (Köhler, 2001) the high involvement of ACE inhibitors and diuretics in potential drug interactions could not be explained by their high prescription frequency. In our study the prescription frequency of cardiac glycosides was very low with a prevalence of 3.3 %. The clinical significance of cardiac glycosides can have more impact than many other classes of medications due to the narrow therapeutic index. Diuretics can indirectly interact with digoxin because of their potential for decreasing plasma potassium levels (i.e. pro- ducing hypokalemia). Because potassium competes with digoxin for binding sites on the Na+/K+-ATPase, hypokalemia results in increased digoxin binding and thereby enhances its therapeutic and toxic effects. Cardiac glycosides with their high risk potential for drug interactions might have lead to an increase in the interaction frequency in the study by (Köhler, 2001).

5.2 Clinical relevance of drug combinations commonly involved in pDDIs in the radiooncologic setting The study data revealed that 288 potential interactions occurred in the study collective in 120 patients. An evaluation showed that the interactions were distributed over a broad range of dif-

ferent interactions including 173 different drug pair combinations. The question which needs to be raised is how dangerous drug interactions are really and which consequences do they have for the patient.

(Riechelmann et al., 2007) showed in two studies in an oncologic setting that almost one third of patients were receiving drug combinations with the potential to interact. The large majority of potential drug interactions (87 %) was due to the presence of non-anticancer drugs and involved mainly medications to treat comorbid illnesses e.g. cardiovascular medications, anticonvulsants and anticoagulants (Riechelmann et al., 2007, Riechelmann et al., 2008). In our study the number of patients receiving anticonvulsants was low with a prevalence of 10.8-13.3 %. The reason might lie in the cancer types occurring in the study collective. In our study brain tumours occurred in only 5 % of cases often treated concurrently with anticonvulsants.

The values of pDDIs were small in our study due to the small sample size, but the trend with regard to the occurrence of pDDIs yielded similar results as other studies (Girre et al., 2011, Juurlink et al., 2003, Riechelmann, 2007).

In the next section the clinical relevance of the 12 most frequently detected drug interaction pairs should be reviewed by data from the literature. The potential risk for the patient should be estimated.

5.2.1 Major potential drug interactions (Category II) Clopidogrel – PPIs (omeprazole)

In our study 2 two major drug interactions belonging to Category II (contraindicated) were detected, which placed the patients at risk and should not be administered concurrently. The first precautious contraindicated drug combination relates to the concurrent administration of clopidogrel and omeprazole (ABDA No. 1162). Clopidogrel requires metabolisation via CYP 3A 4/5 and CYP 2 C19 to an active metabolite (Kazui et al., 2010). The formation of clopidogrels active metabolite may be reduced because omeprazole is an inhibitor of CYP 2 C19 leading to an increased risk for thrombosis (Ford and Taubert, 2013). In the study it is apparent that the number of patients receiving omeprazole at hospital admission was 16.7 % while it decreased at hospital discharge up to 10.8 %. The prevalence of patients receiving pantoprazole increased from 26.7 % up to 56.7 % at hospital discharge. The official recommendation is to use an alternative acid lowering drug with less CYP 2 C19 inhibitory effect such as pantoprazole, a histaminergic H2 blocker or an antacid. The study data revealed that in the Marienhospi- tal Herne the relevance of this issue is known and the recommendations for the clinical setting are implemented already in daily clinical practice. As recent studies reveal that plasma levels of both drugs are short lived, separation by 12-20 h should in theory prevent competitive inhibition

of CYP metabolism and minimize any potential for clinical interactions. PPIs may be administered before breakfast and clopidogrel at bedtime (Laine and Hennekens, 2010).

Opioid agonists- opioid partial agonists

The other major drug interaction referred to the combination of opiod agonist and opioid partial agonists (ABDA No. 433). The underlying mechanism is a pharmacodynamic one leading to an antagonism at the receptor side. In radiooncologic treatment the frequency of patients receiving opioids at hospital discharge increased up to 57.5 %. In our study the interaction occurred only once, but due to the high prescription frequency the issue seems to be of concern. The effect of pure opioid agonists like morphine and hydromorphon can be reduced by concomitant treatment with buprenorphin or nalbuphin because opioids with antagonistic properties block the receptors by competitive inhibition (Harrington and Zaydfudim, 2010). The dose effect curve is shifted to higher dosages. The concomitant treatment should be avoided and p.o. administration of fast response morphine preparations should be administered.

5.2.2 Moderate potential drug interactions (Category III and Category IV) NSAIDs - antihypertensives

The most common potential drug interaction classified as Category III was the interaction between NSAIDs and antihypertensives. This kind of interaction appeared in 17 patients. Category III contains pDDIs were risk factors (renal impairment, heart failure) contribute to an increased risk of pDDI occurrence. Almost 20.71 % of the patients suffered from pain in the course of radioconcologic treatment which justified therapy with NSAIDs. Four different ABDA monographies need to be distinguished (ABDA No. 1, 211, 45, 915). NSAIDs (e.g. ibuprofen, diclofenac, ASS) can interfere with the lowering effect of many classes of antihypertensives because NSAIDs can increase blood pressure (Grossman and Messerli, 2012) by decreasing the renal response to loop diuretics mainly by interfering with the formation of prostaglandins (Opie 2012). NSAIDs cause sodium retention and vasoconstriction which can increase blood pressure (Pavlicevic et al., 2011). Under NSAID therapy the peripheral resistance is increasing based on an inhibition of vasodilative prostaglandin synthesis. A study by (Gyamlani and Geraci, 2007) showed that even small increases in blood pressure (5-10 mmHg) sustaining over a longer period of time (as a result of treatment with NSAIDs) may significantly increase cardiovascular risk associated with an increased risk for myocardial infarction and strokes. Recent published data revealed that interactions between cardiovascular drugs and NSAIDs are the most common interactions leading to adverse event outcomes in elderly patients (Bacic-Vrca et al., 2010, Becker et al., 2007, Dormann et al., 2013, Ismail et al., 2013). A recent study by (Venturini et al., 2011) showed that two third of drug interactions involved angiotensin-converting enzyme inhibitors, non steroidal anti-inflammatory, loop and thiazide diuretics.

The ABDA mongraphy No. 1 focussing on this issue, mainly involved ACE inhibitors and NSAIDs like ibuprofen and diclofenac. In patients with reduced kidney perfusion (liver cirrhosis, heart failure), renal hemodynamic is maintained by the vasodilative effect of prostaglandins. In case prostaglandin synthesis is inhibited, kidney function might deteriorate. Patients with these risk factors were more likely to be affected by an increase in blood pressure. Increased age and increased salt sensitivity are considered as further risk factors for this pDDI. The blood pressure lowering effect of calcium antagonists is less affected and therefore calcium antagonists constitute a therapeutic alternative.

The ABDA monography No. 45 refers to the drug combination between beta-blockers and NSAIDs (e.g. ibuprofen, diclofenac) and monography No. 915 refers to the drug combination between AT-1-antagonists (candesartan) and Cox 2 inhibitors (etoricoxib). The effect and pharmacological mechanism mentioned above can be applied to these drug combinations. The blood pressure increase with beta-blockers was slightly higher than with ACE inhibitors (6 mm Hg versus 3 mmHg) (Aljadhey et al., 2012). Under treatment with etoricoxib the blood pressure increase was 5 mmHg.

The potential interaction between loop diuretics and NSAIDs emerged as important in our study (ABDA No. 211). NSAIDs like ibuprofen decrease the natriuretic effect of loop diuretics resulting in a decreased antihypertensive effect or decreased diuresis (Weir, 2012). A recent study by (Lapi et al., 2013) showed that triple therapy combination consisting of diuretics with angiotensin converting enzyme inhibitors or angiotensin receptor blockers and NSAIDs was associated with an increased risk of acute kidney injury. Therefore radiooncologic patients should be closely monitored and where necessary diuretic dosages need to be increased and anti-inflammatory agents replaced (ibuprofen versus paracetamol). As recommendation the dose of the NSAIDs should be adjusted as low as possible and the salt intake should be reduced.

The study data revealed that the interaction between NSAIDs and antihypertensives emerged as an important issue in our study. A study in 2006 by the Bavarian association of pharmacists (Mayer and Schneider, 2006) showed that 50 % of potential drug interactions in community pharmacies were caused by 10 potential drug interactions. The drug interactions listed above (No. 1. 211, 45) belonged to the top 10 drug interactions found in pharmacies emphasizing the importance of this issue. NSAIDs are administered on prescriptions and over the counter in high numbers in self medication. The interaction combination between NSAIDs and antihypertensives might have been underestimated in our study because the patient’s use of over the counter medications was neglected. The data by (Hanigan et al., 2008) revealed that 96 % of the subjects used prescription medications within three days prior to radiation therapy, 71 % took OTC drugs and 69 % used vitamins, herbs or supplements. Radiooncologic patients should be asked at each appointment, prior to radiation, for the consumption of OTC drugs because the risk for

pDDIs increases significantly. Further it should be figured out whether the intake took place simultaneously and the medication was prescribed for the first time.

ACE inhibitors - diuretics

The combination of ACE inhibitors and diuretics was the most frequent potential interaction (25/288) detected in our study which appeared in 22 patients (18.3 %). Depending on the drug substance it appeared in Category IV and V. 2 different drug combination pairs need to be distinguished mainly the combination between ACE inhibitors and loop or thiazid diuretics (19/288) (ABDA No.232) and between ACE inhibitors and potassium–sparing diuretics (6/288) (ABDA No.186). The drug combination most frequently involved was the combination of ACE inhibitors and diuretics. This result is consistent with findings in recent studies performed in different clinical settings (Fokter et al., 2010, Lapi et al., 2010, Lubinga and Uwiduhaye, 2011, Riechelmann et al., 2005, Sica et al., 2011). In a literature review (Espinosa-Bosch et al., 2012) found out that the interaction pair in the highest number of studies (7; 16.27 %) was the combination of an ACE inhibitor and a diuretic agent, responsible for 366 interactions in a population of 2415 patients which equals a prevalence of 15.5 %. In this study the prevalence was 22.5 %. The inhibition of conversion of angiotensin I to angiotensin II by ACE inhibitors leads to low aldosterone levels and subsequently to sodium and water depletion (Lubinga and Uwiduhaye, 2011, Mignat and Unger, 1995). In the clinical setting patients who had been sodium depleted by diuretic therapy (thiazide or loop diuretics) or other reasons, the initiation of ACE inhibitors can produce a severe transient postural hypotension (Mignat and Unger, 1995) called first dose effect. Risk factors increasing the risk of acute hypotensive episodes are preexisting high blood pressure, high levels of renin and angiotensin II, and congestive heart failure (Hansten and Horn, 2012). Another risk factor for this drug interaction seemed to be the electrolyte imbalance and sodium depletion due to diarrhoea and vomiting during radiooncologic treatment (Yeoh, 2008). Our study showed that only few patients received treatment for diarrhoea but many patients received infusion therapy for dehydration. Therefore this drug interaction seemed to be of importance. Renal function may be affected by ACE inhibitors in the presence of sodium depletion. As a recommendation underlying volume depletion should be corrected prior to initiation of ACE inhibitors in radiooncologic patients. In patients were the combination is necessary ACE inhibitors should be titrated upwards towards the targeted dose (Bicket, 2002). It might be necessary to withdraw the diuretic temporarily, before staring the ACE inhibitor. Recent studies revealed that the addition of potassium-sparing diuretics, e.g. spironolactone to an existing treatment of ACE inhibitors may lead to hyperkalemia (Mateti et al., 2012). In our study, 7.5 % of the patients were exposed to potential DDIs that could have resulted in hyperkalemia. In patients with cardiovascular comorbidities such as congestive heart failure an addition of 25 mg spironolactone has proven to reduce mortality (Pitt et al., 1999). Acute renal failure (ARF) can be one of the many complications associated with malignancy in radiooncologic patients

(Givens and Crandall, 2010). Etiologies, such as tumor lysis syndrome (TLS) and thrombotic microangiopathy (TMA) are unique threats faced by the radioncologic patient (Lyman et al., 2013) after radiation therapy leading to a deterioration of kidney function. Acute renal failure itself leads to hyperkalemia. The risk of hyperkalemia induced by pharmacotherapy puts the patient at additional risk to cardiac adverse events like fatal arrhythmias and neurological side effects like parasthesia (Wilson et al., 2009).

According to (Pitt et al., 1999) a plasma creatinine level exceeding 220 mmol/l, an estimated GFR (glomerular filtration rate) less than 30 ml/minute/1.3 m2 area and a serum potassium exceeding 5 mmol/l are red flags with respect to treatment with potassium sparing diuretics and ACE inhibitors. Blood tests for renal function (creatinine, blood urea nitrogen), glucose and occasionally creatinine kinase and cortisol should be performed. (Espinosa-Bosch et al., 2012) studied the prevalence of drug interactions in hospital healthcare by reviewing literature. The combination of ACE inhibitors and diuretics was the pair appearing in the highest number of studies (7/16.27 %) responsible for 366 interactions in 2415 patients. (Espinosa-Bosch et al., 2012).

Corticosteroids - NSAIDs

The concurrent use of corticosteroids and NSAIDs constituted another frequent interaction belonging to Category IV (Monitoring or adjustment necessary). The prevalence of this interaction rose from 3.5 % (10/288) at hospital admission to 4.5 % (13/288) at hospital discharge. The main drugs prescribed in the corticosteroid group were dexamethasone and prednisolone for premedication in radiooncologic treatment. Drug interactions (ABDA No. 189) were associated with an increased risk of gastrointestinal toxicity, including ulceration, perforation inflammation and bleeding (Hansten and Horn, 2012). During drug treatment with steroids and NSAIDs particular physicians should be alert for evidence of GI ulceration and bleeding e.g. (FOBT=Faecal occult blood test) and in cases where it seems to be necessary misoprostol should be administered. Depending on the indication, NSAIDs should be replaced by paracetamol or COX 2 inhibitors and a PPI should be coadministered during radiooncologic treatment. Risk factors for gastrointestinal bleeding complications are advanced age, high dosages and pre existing anamnesis of gastric ulcers.

ACE inhibitors – xanthine oxidase inhibitors (allopurinol)

The combination of ACE inhibitors and allopurinol was the second most frequent (10/288) interaction of Category IV (ABDA No. 288). Allopurinol in combination with an ACE inhibitor might cause a hypersensitivity reaction (Elasy et al., 1995). The reactions appear to be rare and unpredictable and there exists no established mechanism for the interactions (Hansten and Horn, 2012). Milder cases of hypersensitivity (arthralgia, myalgia, fever) and exfoliative facial derma- titis have been reported. A case of anaphylaxis and myocardial infarction was reported in a pa-

tient taking enalapril who was given allopurinol (Ahmad, 1995). Usually the drug interaction appears in the first three months of treatment (Gerdemann et al., 2008). The patients receiving combination therapy of these two drugs must be closely monitored for signs of hypersensitivity or low white cell counts, especially if they have renal impairment. As recommendation radiooncologic patients should be asked about the onset of therapy with allopurinol. Patients with hematologic, rather than solid organ, malignancies who are undergoing radiation therapy are at risk of experiencing the tumor lyse syndrome, a syndrome with metabolite and electrolyte abnormalities (e.g. hyperuricemia). Hyperuricemia is often treated with allopurinol. In our study population the prescription frequency of allopurinol had not changed during admission and discharge. The prevalence of haematological disorders was very low in our study with 3.3 %. In populations with haematological cancer types the drug interaction between allopurinol and ACE inhibitors can be of importance due to the commencement of allopurinol therapy and therefore this interaction should be beared in mind in radiooncologic treatment.

Beta-blockers – beta-2-sympathomimetics

The concomitant use of non-cardio selective beta-blockers with beta-2-symphatomimetics occurred frequently in our study (ABDA No. 111). It appeared 8 times in 5 different patients.

Non-cardioselective beta-blockers not only decrease the efficacy of beta-2-symphatomimetics, but are also mostly contraindicated in the patients where beta-2-symphatomimetics were indicated (Vonbach, 2007). If beta-blockers are to be used in such patients, cardioselective beta- blockers should be preferred but only under close monitoring of the pulmonary function. Im- portant to notice is that in our study the beta-2-symphatomimetics were mostly administered via pulmonary inhalation. The pulmonary inhalation of beta-2-symphatomimetics was not distinguished from systemic use of beta-2-symphatomimetics and thus the combination of metoprolol and salbutamol was designated as a moderate pDDI (Category IV) leading to an overestimation of the number of pDDIs.

Digitoxin - diuretics

In our study the frequency of digitoxin prescription and the prevalence of pDDIs involving digitoxin were similar at hospital admission and discharge (ABDA No. 97). Although it was prescribed only 4 times in the patient cohort, it was involved in 7 pDDIs. Digitoxin and Digoxin are drugs frequently involved in serious pharmacodynamic drug interactions due to a narrow therapeutic range and its pharmacodynamic properties (Budnitz et al., 2007). Due to an age related decline in kidney function leading to an decrease in digoxin clearance and an enhanced susceptibility to digitoxin at therapeutic concentrations digitoxin toxicicty may be enhanced in oncologic patients (Miura et al., 2000). In our study pDDIs found in patients treated with digitoxin were combinations with loop diuretics (5.83 %), antiarrythmics (1.6 %), betablockers (2.5 %) and steroids (0.83 %). A study by (Wang et al., 2010) yielded similar results. Loop diu-

retics can produce potassium and magnesium deficiencies, leading to an enhancement of the inhibitory effect of cardiac glycosides on the NA-K-ATP`ase and a reduction of digitoxin clearance leading to toxicity (Hansten and Horn, 2012). The concurrent treatment with hypopotas- semic agents like steroids in radiooncologic treatment as supportive medication presents an additional risk factor for patients treated with digitoxin (ABDA No. 097). Beta-blockers may enhance the bradycardic effect of cardiac glycosides (Opie 2012).

Thyroid drugs – cationic agents

The drug interaction with the fifth highest prevalence in Category IV was the interaction between thyroid drugs and cationic agents (ABDA No. 564) with a prevalence of 2.1 %. Cationes included in supportive medications like antacids bind thyroxine and trijodtyhronine in the intestine, thus impairing absorption of thyroids (Singh et al., 2001). According to (Hansten and Horn, 2012) available evidence form the literature suggests that 4-5 hours should elapse between administration of thyroids and antacids. The dosing interval should remain constant.

5.2.3 Minor potential drug interactions (Category V) Steroids - diuretics

In our study another common pDDI was the combination of diuretics and steroids (15/288; 5.2 %) belonging to Category V (ABDA No. 231). 10 % of the patients at hospital discharge received steroids and dexamethasone was the drug most commonly added. At hospital discharge steroid treatment increased up to 34.2 %. Steroids as supportive medications are often prescribed as adjuvant analgesic, to treat nausea and as an appetite stimulant (34 % reported appetite problems) (Navari, 2003). Steroids can cause sodium retention leading to an antagonisation of the general effect of diuretics that of natriuresis (Opie 2012). In our study steroids were responsible for 14.2 % of pDDIs due to premedication. A study by (Lapi et al., 2010) investigating patients undergoing a radiodiagnostic procedure revealed that the most frequent interaction related to the drug combination steroids and acetyl salicylic acid (NSAIDs) followed by steroids and hydrochlorothiazide (diuretic). 12.9 % of the 93 pre-treated patients incurred in a pDDI directly due to pre-treatment with steroids.

Beta-blockerscalcium channel blockers

Many patients need more than two antihypertensive agents. In our study 9.2 % of patients received the drug combination beta-blockers and calcium-antagonists (ABDATA No. 453). With the aim to decrease arterial pressure and improve control of hypertension the prescription of more than two antihypertensives is a planned exploitation of known and expected pharmacodynamic interactions between antihypertensive drugs (Bacic-Vrca et al., 2010). Close clinical and laboratory monitoring and dose adjustment is necessary. Radioncologic patients are often weekend by the treatment process. Additional hypotensive episodes due to drug treatment put the

patients at risk for additional falling episodes. Blood pressure levels should continuously be monitored twice a day during radioncologic treatment.

Antidiabetics – ACE inhibitors

The last important potential drug interaction of Category V had been the pDDI between antidiabetics and ACE inhibitors with a prevalence of 3.1 % (ABDA 28). 15.8 % of the patients in the study were diabetic patients and the data revealed that more than half of the diabetic patients were effected by this drug interaction. A retrospective study showed that 13.6 % of hospital admissions in diabetic patients were attributed to the administration of ACE inhibitors (Triplitt, 2006). Especially at the beginning of drug treatment the coadministration can lead to hypoglycaemic episodes resulting in symptoms like sweating, tachycardia and dizziness. Radiation therapy itself may cause blood glucose levels to change (rise or decrease). In addition, lack of physical activity, stress, and some treatment side effects, such as vomiting and the inability to eat because of nausea might affect blood glucose levels.

5.2.4 Potential drug interactions with oral anticancer drugs A study by (van Leeuwen et al., 2013) investigated the prevalence of potential drug interactions in cancer patients treated with oral anticancer drugs. In 143 patients (16 %) major pDDIs were identified between concurrent medication and oral anticancer drugs. In our study collective 3.3 % of patients were at risk of potential drug interactions due to oral anticancer drugs. The pDDIs found by (van Leeuwen et al., 2013) were similar to the findings in our study. The study by (van Leeuwen et al., 2013) assessed the prevalence of pDDIs in ambulatory cancer patients on oral anticancer treatment. This explains the higher numbers of pDDIs found in his study. In radiooncologic patients the relevance of this issue depends on the underlying cancer type and the medication treatment. In case patients receive oral anticancer drugs and radiation, the administration of SERMs, cytostatics and anticoagulants should be closely monitored.

5.2.5 Score predicting the interaction potential (Gaertner et al., 2012) The score calculated according to (Frechen et al., 2012, Gaertner et al., 2012) provided a tool to estimate the interaction potential of drug combinations. Two studies were performed using the score one in a palliative setting (Gaertner et al., 2012) and the other one in a hospice setting (Frechen et al., 2012).

The study by (Gaertner et al., 2012) aimed to identify the combinations of substances with a high potential for drug interactions in a palliative care setting. High potential for drug interactions in this study included combinations of scopolamine, neuroleptics (e.g. levomepromazine, melperone, promethazine) antihistamines (dimenhydrinate), metoclopramide, NSAIDs, antidepressants (amitriptyline, benzodiazepine) and diuretics (Furosemide, spironolactone). Principal- ly antidepressants are prescribed as anti-depressants and co-analgesics, but they may also be

given as sedative/hypnotics, anxiolytics or anti-panic agents in the palliative setting. In comparison to that the number of patients receiving neuroleptics and antidepressants was low in our study cohort (neuroleptics: admission (2.5 %), discharge (6.6 %), antidepressants: (admission (14.16 %), discharge (24.16 %). This fact was astonishing at a first glance because usually the prescription frequency in oncologic patients is higher. A recent study by (Maurer et al., 2012) showed that more than one third of the radiooncologic patients suffered from positive or marginally positive symptoms of anxiety and depression prior to radiation treatment (Maurer et al., 2012). This fact would allow the assumption that the prescription frequency of antidepressants would increase prior to radiation treatment to improve patient’s quality of life under irradiation. According to our data the prescription frequency marginally increased at hospital discharge. Depression is a frequent problem among cancer patients and had been estimated to occur with a frequency of approximately 30 %-40 % (Mitchell et al., 2011). Furthermore depression is an independent predictive risk factor for cancer related mortality according to a study by (Satin et al., 2009). Mortality rates were up to 25 % higher in patients experiencing depressive symptoms, and up to 39 % higher in patients diagnosed with major or minor depression (Satin et al., 2009). Therefore more frequent medical treatment would have been assumed. Future studies need to address this issue.

Another study by (Frechen et al., 2012, Gaertner et al., 2012) using the similar score was performed in a hospice setting. Antipsychotics (haloperidol, promethazine) and dopamine antagonists were the most frequently found drug interaction combinations in the study cohort (17.9 %), followed by tricyclic antidepressants and antipsychotics. In our study the number of patients receiving antipsychotics ranged between 2.5-6.6 % and the number of pDDIs was low. This higher prescription frequency of antidepressants and neuroleptics in the hospice setting explains the high interaction potential of antidepressants in the study by (Frechen et al., 2012) and explains the low prevalence of drug interaction combinations in our study.

Table 4-11 reveals that especially SSRIs (e.g.sertraline) and SNRIs had a changing interaction behaviour in our study. Although the number of patients affected were similar at hospital admission and discharge the number of pDDIs increased from 1 to 4 at hospital discharge. Dependent on the drug administered concurrently the risk for potential drug interactions ultimately changed from low to high. A subgroup analysis revealed that NSAIDs (e.g. ibuprofen, etoricoxib) and supportive medications (meto-clopramide) were involved in the potential drug interactions at hospital discharge. Dopamine antagonists (metoclopramide) and SSRIs (e.g. citalopram) were frequent drug interactions in hospice patients (Frechen et al., 2012) leading to increased extrapyramidal adverse events and an increased risk for the serotonin syndrome. 32.5 % of patients received dopamine antagonists and histamine H 1 blockers frequently as supportive medication after radiooncologic treatment but no pDDIs were detected. In case the prescription frequency of antidepressants and antipsychotics changes to higher values the interaction potential of do-

pamine antagonists, tricyclic antidepressants should not be under-estimated. The findings concerning drugs with a low interaction potential are consistent with the results of (Frechen et al., 2012, Gaertner et al., 2012). Despite their frequent use (admission 39.1 %; discharge: 64.1 % related to 120 patients) opioids did not account for a large number of pDDIs. Our study revealed that opioids subject to CYP 3A4.1 metabolism (e.g. fentanyl) accounted for an even lower rate of pDDIs to that of opioids metabolised via UGT (e.g. morphine). In the study by (Frechen et al., 2012) morphine and fentanyl accounted for similar rates of pDDIs. Opioids detected in the study by (Frechen et al., 2012) like dihydrocodein and levomethadone with a high risk of potential interactions were of limited relevance in our study.

These findings were in contrast to a recent study by (Hasan et al., 2012) who assessed the prevalence of pDDIs in an intensive care setting. Opioid analgesics were the most common drug class associated with the occurrence of pDDIs, while morphine was the most frequently involved drug. The five most frequent drug combinations were ranitidine and morphine (n = 21, 5.2 %), morphine and magnesium sulphate (n = 13, 3.2 %), midazolam and magnesium sulphate (n = 11, 2.7 %), erythromycin and ranitidine (n = 9, 2.2 %), dobutamine and noradrenaline (n = 8, 2.0 %). The medications involved in the morphine interactions are highly specific for intensive care units and explain the high prevalence. This example illustrates the high reliance of the occurrence of pDDIs on the setting and the patient population.

Laxatives (e.g. natriumpicosulfat, flag score discharge:0) and PPIs (e.g. pantoprazole, flag score discharge: 16.17) with a widespread use in radiooncologic patients were identified as being safe in terms of inducing drug interactions (Table 4-10).

The other two drugs showing a change in their risk portfolio at hospital discharge were calcium and Magaldrat (Aluminium hydroxide) (Table 4-11). Although the number of patients affected is reduced by half in case of calcium prescriptions at hospital discharge the number of potential drug interactions increased by seven. This example demonstrates that one inappropriate drug can lead to a high risk interaction portfolio. This result is consistent with the findings of our study that the interaction between antacids and cationes play an important role as interaction pair in Category III.

Diuretics and cardiovascular medications had a red flag score ranging between (100 and 300). This implicates a high interaction potential. These findings match with the single interactions most frequently found in our study distributed over the different categories (Table 4-9).

Furthermore Figure 4-19 reveals that most of the patients suffering from concurrent potential drug interactions with same effects received drug combinations leading to reduced or increased effects of medical treatment. Providing general recommendations for the detection of these potential drug interactions seems to be difficult. Detecting symptoms indicating an adverse event due to an underlying interference between 2 drugs is not easy. A sophisticated computer screen-

ing program is necessary to detect potential drug interactions leading to increased or decreased effects. Based on this analysis each single drug needs to be taken into account, the underlying mechanism needs to be evaluated and then it should be soughed for symptoms which could have been caused by a pDDI. The potential adverse outcomes in the study by (Ismail et al., 2013) included ototoxicity, nephrotoxicity hepatotoxicity, hypoglycemia, hyper-glycemia, risk of thrombosis, hypotension, cardiac arrythmias and reduction in therapeutic effectiveness. These results were consistent with the findings in our study apart from ototoxicity and risk of thrombosis. Our preliminary study showed that 159 of the potential drug interactions could be detectable by measuring laboratory parameters or identifying clinical symptoms (bradycardia, hypotension, fluctuations in blood sugar levels). A recent study by (Geerts et al., 2009) showed that the frequency of DDIs requiring laboratory tests increased with age and number of drugs, but was not sex related. Laboratory tests for renal function (42.2 %), electrolytes (20.1 %) and co- agulation (13.1 %) were the most frequently required tests. The study by (Geerts et al., 2009) and our findings agree that lab tests for the assessment and detection of manifest DDIs could be helpful. Future studies should be performed proving this issue.

5.3 Factors associated with the presence of pDDIs Other studies already indicated that the number of drugs, comorbidity and advanced age are independent predictors of pDDI occurrence (Frechen et al., 2012, Gagne et al., 2008, Lapi et al., 2010, Vonbach et al., 2007). A recent study by (Lao et al., 2013) identified an increased number of medications as an independent risk factor associated with PIM use and pDDIs (p < 0.05). In a study by (Cruciol-Souza and Thomson, 2006) females demonstrated a significant 23 % higher risk in developing pDDIs. There was not found a statistically difference between men and women regarding pDDIs in our study. The lack of significance in our study might probably be explained by the smaller sample size in comparison to previous studies.

In the present study a statistically highly significant association was found regarding age and number of pDDIs in the age group > 75years in univariate analysis (p = 0.002) corresponding to findings of other studies (Gagne et al., 2008). The multivariate analysis showed no significance in the age group ≤ 75 (Table 4-6). The lack of significance might be attributed to various factors considered in multivariate regression analysis.

As expected and as shown in previous studies we could identify polypharmacy as a risk factor for pDDIs. The figure depicting the proportion of patients with a pDDI stratified by age and number of active substances prescribed at hospital admission and discharge yielded similar results as the study by (Egger et al., 2007). (Egger et al., 2007) evaluated age-specific differences in the prevalence of clinically relevant pDDIs in 2742 ambulatory dyslipidemic patients treated with a statin. Dyslipidemia is one of the most important modifiable risk factors for coronary

100

heart disease. Coronary artery disease was one of the main comorbidities in our patient cohort resulting in medical treatment with various cardiovascular medications. Due to the fact that we focused primarily on non anticancer drugs in radiooncologic patients the comparison with these data seemed to be reasonable. In patients aged ≤ 65 years the prevalence of pDDIs at admission increased from 4.5 % (95 % CI 0.12-22.84 %) when < 5 medications were prescribed to 100 % (95 % CI 39.7-100 %) in patients receiving more than 9 medications concomitantly. In the study by (Egger et al., 2007) the tendency was similar. In patients aged < 54 years, the prevalence of pDDIs increased from 3.4 % (95 % CI 1.3-7.2 %) when 2-3 active substances were prescribed to 22.7 % (95 % CI 11.5-37.8 %) in patient receiving > 7 concomitantly at hospital admission. The increase was pronounced in all other age groups. This exemplifies that the prevalence of patients with clinically relevant pDDIs increases with an increasing number of pharmacologically substances allocated by both studies.

It is apparent that in our study the range of the confidence interval is much broader than in the study by (Egger et al., 2007) and approaches 100 % in our study. The length of the confidence interval reflects the dispersion of values in the patient collective. The pronounced leap in the age group > 75 with less than 5 medications can be related statistically to the low number of patients in the study. The confidence interval for the patient group > 75 in case less than 5 medications was quite expanded ranging from 11.8 to 88.2. In the study by (Egger et al., 2007) the proportion of patients with drug interactions in the age group < 54 receiving more than 9 medications was significantly lower than in the other age groups ≥ 65 years (22.5 % versus 35.4 % (55-64 y) and 33.8 %) implicating that the risk of drug interactions is significantly lower in this age group. In our study we evaluated only one age group ≥ 65 years. In our study the proportion of patients with pDDIs receiving more than > 9 medications ranged between 80-100 %. There was not a significant difference between the age groups under and over 65. It needs to be investigated in further studies whether explicitly the age group under 54 has a lower prevalence for potential drug interactions.

The higher prevalence of pDDIs observed in patients aged > 75 years compared to younger patients may partly be explained by the higher numbers of comorbidities and drugs prescribed in total. The prescription of drugs with a higher potential for pDDIs may be another reason for higher numbers of pDDIs with advancing age. Patients over 75 receive more cardiovascular drugs and diuretics due to underlying comorbidities like heart failure which have a high potential to interact. Comparing the total number of pDDIs with the number of pDDIs caused by cardiovascular drugs reveals that five of the most frequent drug interactions of Category V involve cardiovascular drugs, mainly ACE inhibitors, beta-blockers, calcium-antagonists.

101

5.4 Potentially inappropriate medications On the basis of expert consensus, lists of potentially inappropriate medications (PIMs) for elderly patients were developed to identify and prevent risks more easily. Well-known examples are the Beers list, the STOPP and START criteria, and since 2010 in Germany, the Priscus list.

The data of this study suggested that most of the patients used multiple medications, a factor which was associated with a higher probability of potentially inappropriate medication use. In our study 66.7 % of the patients received potentially inappropriate medications according to the Cave module. Correlation analysis showed that with increasing age (p = 0.003), number of comorbid diseases (p = 0.005) and the number of medications (p < 0.001) the number of IIM prescribing according to the CAVE module increased significantly.

5.4.1 Cave module and other classification systems (Beers criteria and STOPP criteria) Although recent other studies report an association between polypharmacy and PIMs, their definitions do not allow a direct comparison of the data of our preliminary study with other studies performed in different settings. In most of the studies the Beers Criteria were applied to assess the use of PIMs among the investigated sample (Gorzoni et al., 2008, Lao et al., 2013, Lucchetti et al., 2011, Oliveira et al., 2012). In others studies the prevalence of inappropriate medications according to Beers were: 9.8‐38.5 % in primary care, 34 % in secondary care and 40.3 % in Nursing Homes.

Another tool to estimate potential inappropriate medications in the elderly were the START and STOPP criteria developed by (Gallagher et al., 2008). Recent studies reported prevalences of 21.4 % in primary care (Ryan et al., 2009), 34.5 % in secondary care (Gallagher et al., 2007) and 49.8.-55 % in nursing homes (Lam and Cheung, 2012). A recent study by (Lao et al., 2013) using the STOPP criteria to determine PIMs in elderly chinese nursing home residents in Macao showed that 46.5 % of them regularly used one or more PIMs.

STOPP and Beers criteria have several areas of overlaps with the Cave module. All three sets of criteria emphasize the high risk of adverse drug reactions in older people with the use of long- acting benzodiazepines, tricyclic antidepressants, anticholinergic drugs and non-cyclooxygenase

-2-selective non-steroidal anti-inflammatory drugs.

A comparison revealed that the Cave module included some potentially inappropriate medications listed by the STOPP criteria. For some inappropriate medications the STOPP criteria spec- ifies dose ranges (e.g. Digoxin at a long term dose of > 125 ug/day with impaired renal function (< 50 ml) and duration of treatment periods ( > 1 month for neuroleptics and benzodiazepines, PPI > 8 weeks). In our study these data were not available and therefore these STOPP criteria for inappropriate medication use could not be assessed. Furthermore some STOPP criteria re-

102

quire lab values to make an assignment to certain criteria e.g. SSRIs with a history of clinically significant hyponatremia (non iatrogenic hyponatremia < 130 mmol/l within the previous 2 months). In our study no lab values were collected and therefore an assignment was impossible. Some medications listed in the STOPP criteria were not available on the German market e.g. warfarin, dipyramidole, diphenoxylate.

Table 5-1. Most commonly prescribed potentially inappropriate medications (PIMs) as defined per STOPP criteria a (STOPP criteria, 2011).

Disease condition STOPP Criteriaa Frequencyc Cardiovascular system Thiazide diuretic with a history of gout (may exacerbate 4 gout). Beta-blocker with chronic obstructive pulmonary disease 3 (COPD) (risk of increased bronchospasm. Use of diltiazem or verapamil with NYHA Class III or IV 1 heart failure (may worsen heart failure). Calcium channel blockers with chronic constipation (may 1 exacerbate constipation). Central nervous system and TCAs with cardiac conductive abnormalities (pro- 1 psychotropic drugs arrhythmic effects). TCAs with constipation (likely to worsen constipation). 1 TCAs with prostatism or prior history of urinary retention 1 (risk of urinary retention). Long-term (i.e. > 1 month), long-acting benzodiazepines e.g. Xb chlordiazepoxide, fluazepam, nitrazepam, chlorazepate and benzodiazepines with long-acting metabolites e.g. diazepam (risk of prolonged sedation, confusion, impaired balance, falls). Anticholinergics to treat extra-pyramidal side-effects of 1 neuroleptic medications (risk of anticholinergic toxicity). Gastrointestinal system PPI for peptic ulcer disease at full therapeutic dosage for X b > 8 weeks (dose reduction or earlier discontinuation indicated). Musculoskeletal system Non-steroidal anti-inflammatory drug (NSAIDs) with histo- 1 ry of peptic ulcer disease or gastrointestinal bleeding, unless with concurrent histamine H2 receptor antagonist, PPI or misoprostol (risk of peptic ulcer relapse). NSAID with moderate-severe hypertension (moderate: 1 160/100mmHg – 179/109mmHg; severe: ≥180/110mmHg) (risk of exacerbation of hypertension NSAID with chronic renal failure* (risk of deterioration in 0 renal function). Endocrine system Glibenclamide or chlorpropamide with type 2 diabetes melli- Xb tus (risk of prolonged hypoglycemia). Beta-blockers in those with diabetes mellitus and frequent hypoglycemic episodes i.e. ≥ 1 episode per month (risk of masking hypoglycemic symptoms Drugs that adversely affect those prone to falls (≥ 1 fall in past three months) Benzodiazepines (sedative, may cause reduced sensorium, Xb impair balance). First generation antihistamines (sedative, may impair senso- Xb rium).

103

Disease condition STOPP Criteriaa Frequencyc Vasodilator drugs known to cause hypotension in those with Xb persistent postural hypotension i.e. recurrent > 20mmHg drop in systolic blood pressure (risk of syncope, falls) a STOPP: Screening Tool of Older People’s potentially inappropriate Prescriptions Xb = no data concerning length of intake or number of falls were available, therefore no statement could be made, c number of PIM instances, TCAs = tricyclic antidepressants.

Table 5-1 reveals STOPP criteria which have been detected via the CAVE module. The most frequent drug disease interactions were thiazide diuretics with a history of gout, beta-blockers with chronic obstructive pulmonary disease (COPD) and tricyclic antidepressants with dementia or cardiac abnormalities. In total only 15 potential inappropriate medications were detected using the STOPP criteria considering the limitations listed above. A study by (Hamilton et al., 2011) using the STOPP criteria showed that the most common prescribed potentially inappropriate medications were medications considering treatment durations. 128 IIMs were proton pump inhibitors for uncomplicated peptic ulcer disease at full therapeutic dosages for > 8 weeks, while 56 medications were benzodiazepines in patients who had > 1 fall in the past 3 months. The three most commonly identified PIMs by two other recent studies by (Ryan et al., 2013, Wahab et al., 2012) were benzodiazepines, proton pump inhibitors and duplicate drug classes. The study by (Lao et al., 2013) showed similar results. Instead of PPI prescription the use of dipyramidole as monotherapy for cardiovascular secondary prevention was inappropriate. In our study pantoprazole and benzodiazepines were prescribed frequently, but due to a lack of information (Length of intake, number of falls) no assignment could be made to the STOPP criteria leading consequently to an underestimation of PIMs as per STOPP criteria in our study. It is apparent that the use of aspirin in our study was very low. The reason might lie in the mye- losuppression under radiation therapy and premature settling of aspirin prior to radiation therapy.

The disadvantage of the CAVE module is that anamnestic patient information except age and comorbidities is not considered but this is due to the field of application. The Cave module is a system used in public pharmacies where the access to amnestic patient information is limited. In daily clinical practice difficulties might occur. Data about the duration of treatment are often missing in the hospital setting. Primarily the family physician knows since how many months special medications like PPIs or benzodiazepines are prescribed. The importance of the STOPP criteria in general medicine has already been taken into account. The STOPP criteria are already implemented in new guidelines for general practitioners in Hessen (Deutsche Gesellschaft für Allgemeinmedizin und Familienmedizin (DEGAM) and Leitliniengruppe Hessen, 2013).

The most common potential drug disease interactions in a study by (Lindblad et al., 2005) were calcium-channel blockers and heart failure, beta-blockers and diabetes, aspirin (NSAIDs) and peptic ulcer disease, beta-blockers and peripheral vascular disease, and beta-blockers and chronic obstructive pulmonary disease. In our study the most frequent drug disease interactions oc-

104

curred with underlying metabolic and endocrinologic diseases. The inappropriate use of steroids and beta-blockers in diabetic patients was the most frequent drug disease interaction in our study (28/327 (8.6 %)). A study by (Wami et al., 2013) showed that the chronic use of NSAIDs and corticosteroids in diabetic patients either separately or in combination, was significantly associated with the worsening of HbA1c levels. (Andersson et al., 2012) proved that in overweight, cardiovascular high-risk patients with type 2 diabetes, increasing HbA1c concentrations were associated with increasing risks of cardiovascular adverse outcomes and all-cause mortality. These data show that the issue in the radiooncologic setting is of relevance and good metabolic control is very important due to the fact that radiation therapy itself leads to rising blood sugar levels. Giving steroids in multiple doses throughout the day instead of a single bolus dose or administering the total daily steroid dose intravenously over 24 hours can assist in controlling hyperglycemia (Psarakis, 2006).

In diabetic patients the use of beta blockers is a matter of debate. Due to the possibility that the symptoms associated with hypoglycemia are mitigated and glycogenolysis is impaired in diabetic patients the risk for hypoglycemia may be increased. Cardioselective beta-blockers can be considered to be safe in diabetics. Therefore it is advisable to avoid non-cardioselective (like propranolol) beta-blockers in diabetics and to prefer cardioselective beta-blockers (e.g. bisoprolol). Patients should be advised that tachycardia may not develop despite hypoglycemia, but that sweating may be an indicator for a hypoglycemic reaction (Vonbach, 2007).

Another frequent drug disease interaction identified in radiooncologic patients was dose adoption due to an underlying chronic renal impairment (32/327 (9.8 %)) or due to age (22/327 (6.7 %)). 15.8 % of the patients were affected by renal impairment. One of the most clinically important age-related changes in elderly is a decline in renal function. Chronic renal failure (CRF) and end-stage renal disease (ESRD) alter drug disposition by affecting tissue and protein binding and therefore reducing systemic clearance of renally cleared drugs of plasma proteins (Dreisbach and Lertora, 2003).

The most frequent age inappropriate medications according to the Cave module were medications increasing the risk of side effects like paradoxical side effects, agitation, aggressiveness and increasing the incidences of falls (22/327 (6.7 %)) (McKenzie and Rosenberg, 2010). The main drugs were Z drugs (zoplicon), benzodazepines (lorazepam, diazepam) and antidepressants (citalopram, amitripytline) or antipsychotics (haloperidol). These drug classes also play a major role in the classifications according to Beers and (Gallagher et al., 2007). Especially elderly > 65 are prone to paradoxical reactions. Elderly patients with chronic agitation treated with benzodiazepines are at an increased risk of falls, sedation and cognitive impairment (Ray et al., 1989).

105

The second most frequent age inappropriate medication, detected in our study was metamizol natrium, the most frequent pain medication in radiooncologic patients. In elderly patients the excretion of metabolites of metamizol is delayed due to a decline in renal function. Metamizol sodium is rapidly and almost completely absorbed in the gastrointestinal tract and rapidly cleaved to 4-methylaminoantipyrine (MAA) via non-enzymatic hydrolysis. The elimination half-life of MAA increases from 2.6 h (12 volunteers, 21–30 years of age) to 4.5 h (9 volunteers, 73–90 years of age) in elderly patients (ABZ Pharma, 2013). Therefore the dose needs to be adopted in radiooncologic patients.

A recent paper by the EMEA showed that diclofenac increases the risk of arterial thromboembolic events in patients who have serious underlying heart or circulatory conditions, such as heart failure, heart disease, circulatory problems or a previous heart attack or stroke. The absolute cardiovascular risk with any NSAID depends on a person’s underlying risk factors, such as high blood pressure and cholesterol levels (European Medicines Agency, 2013). Patients receiving diclofenac in doses > 150 mg and for a longer period of time should not be treated with diclofenac. The Cave module had not implemented this information but it seems to be necessary to add this information in this thesis.

The most already known and expected drug disease interaction was the administration of corticosteroids in patients with osteoporosis. Steroids can further induce osteoporosis (CIO) in long term treatment, especially in patients treated with steroids > 6 months. In all patients who received a daily dose of at least 7.5mg equivalent prednisone and a treatment that is expected to last at least 3months, prevention or treatment of osteoporosis should be considered (Briot and Roux, 2013, Soen, 2011). Minimal efficacious doses should be administered to prevent CIO and treatment of calcium, vitamin D and gonadal hormones insufficiencies should be considered. Patients at risk of fracture, as post-menopausal women with prevalent fractures, should receive a bisphosphonate (Levine, 2006). Steroids in elderly should be avoided but the radiooncologic setting shows that in cancer therapy steroids had become indispensable in cancer therapy. Therefore a risk benefit assessment in each individual case needs to be performed.

Lipid lowering drugs like statins can lead to rhabdomolysis in elderly patients (Tamraz et al., 2013). Polypharmacy and altered drug metabolism both put the elderly patient at increased risk of myotoxicity, when drugs in the HMG-CoA reductase inhibitor class are administered (Sica and Gehr, 2002). Rhabdomyolysis describes the breakdown of muscle fibres and the release of their myoglobin into the bloodstream. Acute tubular necrosis or kidney failures are induced by the breakdown of myoglobin in harmful compounds. Fatigue and muscle weakness are attributed to aging, but in clinical practice it is a sign for an underlying drug drug or drug disease interaction (Marusic et al., 2012).

106

5.4.2 Priscus list The retrospective evaluation provides new information on the prescribing behaviour of potential inappropriate medications listed on the Priscus list (Projektverband Priscus, 2011). The aim of the study was to assess the baseline prevalence of the prescribing of PIMs as defined by the Priscus list in the hospital radiooncologic setting.

79 patients were aged ≥ 65 and included into the evaluation. At hospital admission 18 patients (22.8 %) were affected by PIMs from the Priscus list while at hospital discharge the number of patients affected increased up to 37 (46.8 %). 13 different drugs were mainly involved in PIM prescribing. Recent studies showed that the PIM prevalence was higher in women than in men (Amann et al., 2012, Holt et al., 2010). Our study data revealed different results, only for zoplicon a higher prescription rate could be detected in female patients. The most commonly prescribed PIMs were zoplicon (admission: 8.9 %, discharge: 15.2 %), amitriptyline (admission 2.5 %, discharge: 12.7 %), and lorazepam (admission: 1.3 %, discharge 5.1 %). The proportion of PIMs in relation to the total number of medications (700/1024) prescribed at hospital admission and discharge seemed to be low with 3-5 %.

PIM prevalences of 12 % or more in the outpatient setting and up to 40 % in care homes have been reported in review articles in studies in Europe and USA (Hamilton et al., 2011, Siebert et al., 2013). According to the Barmer Drug Report 2012 a quarter of patients received one medication from the Priscus list in 2011, women more often than men (Barmer GEK, 2012). Psycho- leptics active agents (NSAIDs and antidepressants) and drugs to treat cardiovascular diseases were most frequently prescribed. The drug class that includes psycholeptics such as hynpotics and sedatives presents the largest class of the Priscus list with 31 drugs included.

(Amann et al., 2012) examined in 2007 the medications of 804000 elderly patients. 201472 (25 %) patients received at least one PIM. (Amann et al., 2012) reported that long term benzodiazepines like bromazepam and diazepam, the short acting benzodiazepines lorazepam and oxazepam and the Z-drugs such as zoplicon and zolpidem were most frequently prescribed. These findings agree with results of other German studies based on the Priscus list and support the findings in our study (Dormann et al., 2013, Schubert et al., 2013). According to the Priscus list the risk of falling and hip fractures is increased in patients taking Z-agents. The reaction time is delayed and psychatric recations such as paradoxical e.g. agitation, irritability, hallucinations and psychosis might be induced. Due to the risk of withdrawal symptoms and physical as well as physiological dependency, the maximum duration of treatment with Z-drugs (benzodiazepine like drugs) should be limited to 4 weeks. Dose tapering might be necessary, when treatment is continued for a longer period. As our data indicated 5 patients received a new prescription at hospital discharge. Their exists a high risk that in the transition point from hospital to ambulatory care the prescription is not discontinued after a 4 weeks period. A lot of radiooncologic pa-

107

tients have insomnia due to radiation therapy and it might be assumed that they take it for a longer time. Monitoring the discontinuation of supportive medications in the ambulatory setting would be reasonable.

For benzodiazepine prescriptions there exists an upper dose limitation which should not be exceeded. The Priscus list gives upper dose limits beyond they are rated as PIMs. A subgroup analysis showed that 12 patients (15.2 %) received a benzodiazepine (e.g. lorazepam), but only 4 patients (5.1 %) received it in a dosage over > 2 mg which is considered to be precarious in elderly patients. The fact that only 4 patients received an overly high dose in the hospital setting suggests that the clinic was already familiar with benzodiazepine prescriptions in elderly people.

In other national studies by (Fick et al., 2003) the high prevalence of usage of antidepressants has been reported. 12.7 % of the patients received amitriptyline at hospital discharge. 10.1 % of the patients received it as new prescription at hospital discharge. Amitriptyline is most often used in the pharmacological management of depressive illness. The assessment of antidepressants in the Priscus list focused on depression as indication. Amitripty- line is regarded as appropriate in older people when given at a low dose for various neuropa- thies and pain control in palliative care. In radiation therapy probably amitriptyline is used for both indications. After radiation pain might increase due to radiation induced inflammations and depressive symptoms might be enhanced by stresses of therapy. In future studies the indication for amitripytline needs to be figured out. In our study PIM prescription of amitriptyline might be slightly overestimated due to the unknown indication.

Other studies (Holt et al., 2010) showed that cardiovascular medications like acetyldigoxin, doxazosin and nifedipine (immediate release formulation) usually were prescribed between 10- 20 % in a study cohort. Acetyldigoxin and doxazosin were not prescribed in our study cohort. Recent studies have been performed in cardiac patients receiving these medications more frequently.

A recent study by (Siebert et al., 2013) evaluated the Priscus list in clinical practice. While hospitalized, the mean number of administered PIMs per patient was 1.2 based on STOPP criteria and 0.5 based on the Priscus list. The lowest number of PIMs per patient was detected by apply- ing the Beers list (0.4 PIMs). The amendment of diagnosis-related STOPP criteria to the Priscus list would be useful to significantly advance therapeutic success and drug safety in the elderly (Siebert et al., 2013).

5.5 Critical appraisal of own methods The retrospective preliminary study is affected by several limitations. First, the population is limited to a single institution and, therefore, the generalizability to other cancer care settings is unknown. The study focused on radioncologic patients and the findings may not be generalisa-

108

ble outside the settings of the Marienhospital Herne without further studies. Other international studies were performed with more patients (1000-80000 patients). The aim of the preliminary study was to assess trends regarding the prescription frequency of pDDIs and PIMs in the radiooncologic setting and should give suggestions for future studies. Studies with more patients need to be performed in the future to increase the statistically explanatory power.

The principal limitations of this study were the retrospective design and the use of discharge letters as data source. The reliability and completeness of the information obtained could not be ascertained with any accuracy and therefore retrospective data collection can cause bias. Medi- cations and concurrent diseases were not controlled by interviewing the patients again. There- fore it might be possible that the number of potential drug interactions were over or underre- ported. Furthermore the temporal connection of drug intake was neglected in the study. Espe- cially potential drug interactions influencing absorption and metabolism could have been overestimated in our study. The study only focused on medications which were listed in the discharge letters. Medications which patients had been taken in self medication such as herbal remedies, sleeping aids (diphenhydramine, valerianae radix) and over-the-counter drugs (NSAIDs) were neglected. Therefore the overall prevalence of potential drug interactions could be underestimated in our study. These limitations are inherent to other retrospective studies which focused on potential drug interactions and potentially inappropriate medications. Another neglected issue were food drug interactions. Depending on the radiation side e.g. laryngeal cancer, oesophagus cancer food intake might be reduced. In the study 28.3 % of the patients were feed by parenteral nutrition in order to reduce the risk for nutritional deficiencies during radiation therapy. Interactions between nutrients and medications may be significant, resulting in treatment failure or adverse events and the number of potential drug interactions may be underestimated in our study.

Another limitation is that potential interactions between intravenous chemotherapeutic agents and chronic or supportive medications were not screened in the study. Chemotherapy and chronic medications could be a source of additional interactions (anti-inflammatory drugs) with a higher risk of toxicity in the case of hypoalbuminemia commonly observed in elderly radioncologic patients. Chemotherapy often has an influence on metabolic functions, lab values and blood pressure. Therefore it might be difficult to assess causality of manifest drug interactions in radiooncologic patients with chemotherapy because the consequences of drug interactions like blood sugar fluctuation or leukopenia can be attributed to either both or are masked by chemotherapy. A preliminary study was performed in radiooncologic patients focussing on pDDIs without investigating pDDIs between i.v. chemotherapeutic agents and chronic medications. At first it was evaluated whether interactions between chronic medications put the patient at risk for interactions. In future studies focussing on manifest drug interactions in cancer patients, the difficulties relating to chemotherapy shall be taken into account.

109

In the study the evaluation of clinical manifestations of drug interactions were not included and therefore the potential drug interaction expression was introduced. The study evaluated whether laboratory parameters might give evidence for underlying potential drug interactions. The results revealed that 159 potential drug interactions might lead to changes in lab values e.g. hy- percalemia, blood sugar changes, skin reactions and deterioration in symptoms (dyspnoea). In future studies drug charts should be screened for clinical manifest symptoms in order to evaluate the real clinical relevance of drug interactions. A causality assessment of DDIs using the Naranjo algorithm should be performed.

110

6 Key results, management strategies and conclusions

Medical treatment was screened in this study for potential drug interactions using the ABDA drug interaction software and for potential inappropriate medications using the Cave module and the Priscus list.

6.1 Key results The key results of the study can be summarized as follows:

. The study showed that in the radiooncologic setting the prevalence for pDDIs is high and increased significantly at hospital discharge (p = 0.0001) (Page 60). 33.7 % of all pDDIs at discharge were created by medication changes during hospitalization (Page 60). . The most frequently drug classes involved in pDDIs were antihypertensive drugs (ACE inhibitors, beta-blockers, and diuretics), NSAIDs and anticoagulants (Figure 4-17, Page 68), Table 4-8, Page 71). . Patients with a CCI age adjusted score > 8 are significantly affected by pDDIs according to univariate analysis (p = 0.012) (Table 4-4, Page 55). . Risk factors for pDDIs in this study were age and the number of concomitant medications. The prevalence of clinically relevant pDDIs is significantly higher in patients aged over 75 years (p < 0.001). The patients of this age group show a greater number of diseases and pharmacologically active substances (p < 0.001) (Table 4-6, Page 63). Multivariate regression analysis showed that the number of pDDIs increased significantly with the number of medications (9-20; p = 0.004, Table 4-6). . The subjective well being of patients assessed by the Karnofsky index does not allow to seek for hints of being harmed by pDDIs. The Karnofsky index improved after radiooncologic treatment, while the prevalence of pDDIs increased at hospital discharge (Figure 4-4, Page 54). Finding manifest pDDIs in future studies might be difficult. . The study proved that pDDIs most frequently occurred between chronic medications. This finding is supported by findings of other recent studies (Riechelmann et al., 2005) (Figure 4-12, Page 62). . Supportive medications (e.g. supportive GI medications and radiooncologic medications) have a low interaction potential (except corticosteroids) assessed by a low flag score (green) (Table 4-9, Table 4-10, Page 74). Most of the cardiovascular medications have a high interaction potential (red flag score.).

111

. The pDDIs detected in this study are similar to the potential interactions detected in extensive studies performed in other settings e.g. internal medicine and emergency departments (Page 89-100). . Radiooncologic patients were discharged with a slightly higher prevalence of IIMs according to the Cave module at hospital discharge. IIMs were inappropriate due to increased age (39.4 %) and underlying metabolic (25.0 %) or cardiovascular diseases (13.1 %) (Figure 4- 22, Page 83). . With increasing age (p = 0.003), number of comorbid diseases (p = 0.005) and the number of medications (p < 0.001) the proportion of patients receiving IIMs increased (Table 4-14, Page 80). . The prevalence of medications listed on the Priscus list increased at hospital discharge (46.8 %). The three most prevalent inappropriate drugs were long- acting benzodiazepines, antidepressants and hypnotics (zoplicon) (Figure 4-23, Page 86).

6.2 Management strategies Preliminary recommendations for daily clinical practice with special considerations for reducing pDDIs and PIMs are developed based on the results and findings of the preliminary retrospective study in the radioncologic setting confirmed by more extensive studies in other clinical settings. A novel management strategy for easy and fast detection of pDDIs were developed and can easily been implemented in daily clinical practice. In the risk management process this procedure belongs to step four mentioned in Chapter 1.2.

6.2.1 Potential drug interactions In terms of reducing pDDIs a complete medication history of each patient is important. The data of the study showed that NSAIDs were most frequently involved in pDDIs. The brown bag approach (Colt and Shapiro, 1989) – bringing all medications to the first visit at hospital admission-seems to be a useful tool in identifying polypharmacy and reducing the risk for pDDIs.

Aiming at decreasing the risk of pDDIs the following procedure is proposed:

In a first step the treating radiooncologist screens a patient`s medication portfolio at hospital admission. In case medications appear on the recommended list (A) (Table 6-1) the physician needs to evaluate whether the drug is administered with a concurrent drug on the list which might result in a pDDI. All pDDIs with a frequency of n ≥ 4 have been included in list A (X: frequency ≥ 4, XX: frequency ≥ 8, no symbol: no occurrence). The value n = 4 has been arbi- trarily decided. The results proved that this systematology could clearly been established implying a high probability that pDDIs can be detected. In the right upper part of Table 6-1 frequen-

112

cies of potential drug interactions can be derived, while in the left lower part the classification of the drug interaction pairs into a Category (I-VI) can be derived (Table 6-1, Chapter 3.5.6.)

Supportive information about potential adverse outcomes and management strategies of pDDIs are listed on list B. List B is arranged in alphabetical order per Category (I-VI).

In a second step the treating radiooncologist further checks whether potential adverse outcomes of a pDDI can be excluded in the individual patient. In case symptoms can be attributed to a DDI the physician announces in a third step the ADR to the risk and error management of the clinic (manifest DDI). Further the physician refers to list B to get advice in how to deal with the manifest DDI. Often one drug of the interacting combination needs to be cancelled immediately and replaced by an appropriate drug. Some potential adverse outcomes are difficult to prove by means of symptoms e.g. decreased plasma biovailability. Methods need to be established in the future to easily detect DDIs.

113

Table 6-1. Overview over the most frequently detected potential drug interaction pairs found in the study cohort; (X: Frequency ≥ 4, XX: Frequency ≥ 8), e.g.II: Category II in List B (Table 6-2) (Category I: Contraindicated, Cat II: Contraindicated as a precaution, Cat. III: Monitoring or adjustment in certain cases, Cat. IV: Monitoring or adjustment necessary, Cat. V: Monitor as a precaution, Cat. VI: No measures required). inhibitors ACE Aluminium salt Anticoagulants Antidepressants Antidiabetics Antiparcinson Antihypertensives Beta-blockers Beta-symphatomimetics channel blockersCalcium glycosides Cardiac Cationes Citric acid Clopidogrel Corticosteroids antagonists Dopamin- Diuretics Neuroleptics NSAID Opioid agonists agonists Opioid-partial diuretics sparing Potassium PPIs SSRIs TAIs Thyreoid preparations Xanthine oxidase inhibitor ACE inhibitors XX XX XX X XX Aluminium salts X Anticoagulants X X XX Antidepressants V Antidiabetics V XX X Antiparkinsonian X Antihypertensives X Beta-blockers XX XX Betasymphatomimetics IV X Calcium channel blockers V Cardiac glycosides X Cationes X Citric acid III Clopidogrel X Corticosteroids IV XX XX Dopamin-antagonists X Diuretics V V V IV V X Neuroleptics V NSAIDs III III IVIII Opioid agonists X Opioid partial agonists II Potassium sparing diuretics IV PPIs V II SSRIs IV TAIs IV Thyreoid preparations III V IV Xanthine oxidase inhibitor IV

114

Table 6-2. List B Management guidelines for most frequently identified drug drug interactions modified and adopted from((Hansten and Horn, 2012), (Bacic-Vrca et al., 2010, Frechen et al., 2012, Gagne et al., 2008, Griese, 2012, Ismail et al., 2013)) Category I: Contraindicated, Cat II: Contraindicated as a precaution, Cat. III: Monitoring or adjustment in certain cases, Cat. IV: Monitoring or adjustment necessary, Cat. V: Monitor as a precaution, Cat. VI: No measures required), I: first drug; II; second interacting drug Cat ABDA Interactions Potential adverse outcomes Management guidelines II 433 Opioid agonists- + partial ago- Reduced analgesic effect Pure opioid agonists should be combined with paracetamol or rapid-acting agonists like mor- nists with withdrawal symptoms phine or hydromorphon. I: Morphin II: Buprenorphin II 1162 PPIs + clopidogrel Increased risk of infarctions It is recommended to use an alternative acid-lowering drug with less CYP 2C19 inhibitory effect I: Omeprazole ,esomeprazole due to reduced anticoagulant such as pantoprazole, an antacid or a H2-blocker (ranitidine). II: Clopidogrel effectivness III 1 ACE inhibitors+ NSAIDs Decreased antihypertensive Patient´s blood pressure and hemodynamic parameters need to be monitored daily. In case of the I: Captopril, enalapril, lisinopril, ramipril effect occurrence of an ADR, NSAIDs need to be replaced by non opioid analgesics (metamizole) II:Ass, coxibs, ibuprofen, piroxicam paracetamol, tramadol or ACE inhibitors by an AT1-antagonist. III 446 Aluminium salts + citric acid Increased absorption of Decreased bioavailability of citric acid. When both drugs are coadministered, a time interval of 3 I: Aluminiumhydroxid, magaldrat aluminum salts -4 hours should be maintained. II: Citric acid in dissolving tablets III 36 Anticoagulants+ thyroid hor- Decreased bioavailability of Under therapy with anticoagulants, modifications in the thyroid hormone status due to dose mones thyroid hormones changes might lead to changes in clotting parameters (INR needs to be closely monitored). I:Phenprocoumon II: L-thyroxine III 915 Antihypertensives (AT1 antag- Decreased blood pressure Blood pressure needs to be monitored and dose adopted. NSAIDs should be replaced by non- onists) + NSAIDs lowering effect opioid analgesics. I: Losartan, candesartan II: Ibuprofen, diclofenac III 45 Antihypertensives (beta- Decreased blood pressure Blood pressure needs to be monitored and dose adopted. NSAIDs should be replaced by non blockers) + NSAIDs lowering effect opioid analgesics. ACE inhibitors might be replaced by calcium channel blockers. I: Metprolol, bisoprolol, atenolol II Ibuprofen, diclofenac, ass III 211 Diuretics + NSAIDs Decreased diuretic and anti- Blood pressure needs to be monitored and dose adopted. NSAIDs should be replaced by non- I: HCT, furosemide, torasemide, hypertensive effect opioid analgesics. xipamide II:Ass, ibuprofen, coxibs, meloxicam, dicofenac IV 186 ACE inhibitors + diuretics Risk of hyperkalemia The hyperkalemia associated with this combination is of special concern in patients with renal (potassium sparing diuretics) impairment or diabetes. Renal function and serum potassium levels need to be monitored. I: Ramipril, lisinopril, captopril Use the minimum effective dose of potassium sparing diuretics like spironolactone (25 mg/day). II: Amilorid, traimteren, spironolactone

115

Cat ABDA Interactions Potential adverse outcomes Management guidelines IV 288 ACE inhibitors + xanthine Increased risk of immuno- Monitoring patients receiving the combination carefully for hypersensitivity reactions and blood oxidase inhibitors logical reactions, skin reac- count changes, e.g neutropenia. I: Ramipril, lisinopril tions Prompt discontinuation of the offending drugs is important. II: Allopurinol IV 183 Antidiabetics+ corticosteroids Decreased blood sugar low- Increased risk of hyperglycemic episodes, at the beginning and at the end of therapy. Close mon- I: Glimepiride, glibenclamid, rosiglitazon, ering effect, risk of hyper- itoring of blood glucose levels, dose adoption of the antidiabetic agent might be necessary. Pro- metformin glycemia nounced diabetogenic effect with prednisolone – replacement by deflazacort. II Betamethason, prednisolone IV 111 Beta-blockers + beta- Decreased effects of beta- Beta-blockers should be avoided in patients receiving beta-agonists. If beta-blockers are re- sympathomimetic agents sympathomimetic agents quired, cardioselective agents are preferable (e.g. atenolol, metoprolol, bisoprolol). I: Metoprolol, bisoprolol II: Salbutamol, formoterol, fenoterol IV 97 Cardiac glycosides + diuretics Diuretica induced hypoka- Monitor potassium and magnesium status of patients on diuretic /digitalis therapy. If needed I: Dig(it)oxin, etacrynic acid lemia may increase the risk replacement of magnesium and potassium and continuous digoxin plasma level measurement. II: HCT, furosemide, xipamide of digitalis toxicity Potassium sparing diuretics can be considered as alternative to potassium – wasting agents. IV 189 Corticosteroids + NSAIDs Increased risk of gastrointes- Be particular alert for evidence of GI ulceration and bleeding in patients. Consider the concur- I: Betamethason, dexamethasone, hydro- tinal bleeding complications rent use of misoprostol and PPIs. cortison,prednisolone II: Ibuprofen, Ass, diclofenac IV 511 Oral anticoagulant s+ TAIs Increased ulcerogenic ef- If any NSAID is used in combination with an anticoagulant, the prothrombin time should be I: Phenprocoumon fects, risk of bleeding monitored carefully and watched for signs of gastrointestinal bleeding (stool samples). II: Ass IV 1005 SSRI + MCP Increased risk of extrapy- This combination might produce extrapyramidal symptoms. Symptoms may consist of an in- I: Citalopram, fluoxetin,sertraline ramidal side effects creased heart rate, shivering, sweating, dilated pupils, myoclonus (intermittent tremor or twitch- II: MCP ing), as well as over responsive reflexes. IV 564 Thyroid hormones + polyvalent Decreased bioavailability of 4 to 5 hours should elapse between administration of cationes and thyroids. A relatively constant cationes thyroid hormones interval should be maintained between the administration of the 2 drugs. Monitoring for altered I: L thyroxine thyroid response (e.g. serum thyroid stimulating hormone concentrations) is necessary, when II: Aluminium, calcium and iron therapy with cationes (e.g. antacids) is started or the dosing interval is changed. Thyroid preparations should be administered with empty stomach in the morning. V 232 ACE inhibitors + diuretics Hypotension, Bradycardia Risk in patients with sodium depletion and hypovolemia. While starting an ACE inhibitor, the (loop/thiazid diuretics) guidelines recommend temporarily discontinuing loop or thiazid diuretics and using a very low I: Ramipril, lisinopril dose in the evening. After the initial dose the blood pressure should be monitored closely for 4 II: HCT, furosemide hours. V 516 Anticoagulants + PPIs Increase of the hypothrom- PPI increases the hypoprothrombinemic response to phenprocoumon, be aware for signs of I: Phenprocoumon binemic response to phen- bleeding and monitor the INR. II: Omeprazole, esomeprazole, pantopra- procoumon zole

116

Cat ABDA Interactions Potential adverse outcomes Management guidelines V 528 Antidepressants + anticoagu- Increased anticoagulant PPI increases the hypoprothrombinemic response to phenprocoumon, be aware for signs of lants effectiveness bleeding and monitoring of INR. I: Citalopram, sertraline, fluoxetin II: Heparin, enoxaparin, clopidogrel, Prasugrel, Ass V 28 Antidiabetics + ACE inhibitors Increased insulin sensitivity, Monitoring patients for altered hypoglycemic effects, if ACE inhibitor therapy is initiated, dis- I: Glimepiride, metformin risk of hypoglycemia continued or changed in dose. Insulin or oral hypoglycemic dosages need to be adjusted. II: Ramipril, lisinopril V 164 Antidiabetics+ diuretics (Thia- Increased blood glucose Increased risk for hyperglycemic episodes especially if low plasma potassium levels are present- zide) lowering effect hypoglyce- ed. In diabetic patients serum glucose and potassium levels need to be closely monitored. As I:Glimepirid, glibenclamide, metformin mia alternative agents loop diuretics like furosemide can be used because the extent of the interaction II:Hydrochlorothiazide is lower than in case of thiazide (hydrochlorothiazide use). V 453 Beta-blockers + calcium antag- Increased antihypertensive Monitoring blood pressure because combination can result in hypotension. Further the calcium onists effect with hypotension antagonist can increase beta-blocker concentrations. I:Metoprolol, propranolol, atenolol II: Nifedipine, amlodipine V 1278 Diuretics + beta- Risk of hypokalemia Monitoring of patients with potassium wasting diuretics and beta 2 agonists for signs of hypoka- sympathomimetic agents lemia (Serum concentration< 3.6 mval/l) including ECG changes, fatigue and muscle pain. Pa- I: HCT, furosemide, xipamide tients predisposed to hypokalemia should receive triamteren or potassium supplementation to II: Formoterol, salbutamol prevent reduction of potassium serum concentrations. V 231 Diuretics+ corticosteroids Risk of hypokalemia Monitoring patients for signs of hypokalemia (Serum concentration < 3.6 mval/l) including ECG I: HCT, furosemide, xipamide changes, fatigue and muscle pain. Patients predisposed to hypokalemia should receive potassium II: Dexamethasone, prednisolone sparing diuretics, like spironolactone or triamteren or potassium supplementation to prevent reduction of potassium serum concentrations. Steroids with a low mineralocorticoid effect like dexamethasone, triamcinolone should be used instead of hydrocortisone or prednisolone. V 213 Neuroleptic agents-anti Parkin- Neuroleptics block dopa- Antagonism at the dopamine receptor, mutual reduction of the effect. Patients with Parkinson`s son medication (dopaminantagonists mine receptors –increase of disease should receive atypical neuroleptics like clozapin, olanzapin or quetiapin. versus dopaminagonist) extrapyramidal symptoms I: Levodopa, pramipexol II: Promethazin, haloperidol, clozapin

117

6.2.2 Potentially inappropriate medications The study results showed that potential inappropriate medications were of concern in the radiooncologic setting. Especially patients with underlying metabolic diseases and renal impairment were affected by PIMs. Future studies are required to gain further data for developing management strategies for reducing PIMs. The comparison of the Beers list, the Cave Module and the STOPP criteria and the Priscus list showed that similar elements of PIMs were found in all four classifications. Overlaps between the four classification methods could be identified. It is difficult to refer only to one list to assess the prevalence of PIMs as already proven by (Siebert et al., 2013). The amendment of diagnosis-related STOPP criteria to the PRISCUS list and the results of the Cave module would be useful to significantly advance therapeutic success and drug safety in the elderly. Based on the data of the preliminary study recommendations for ensuring safe drug prescribing in renally impaired patients can be developed. For ensuring safe prescribing in this patient population, it would be advisable to calculate the creatinin clearance by an equation developed by Prof. Dr. Haefeli in patients with renal impairement (I-IV) in the radiooncologic setting. The following values are necessary for the calculation of the estimated creatinin clearance: age, body weight, serum creatinin and gender. The macro (Annex C.) provided online under (www.dosing.de) calculates an estimated creatinin clearance, an estimated excretion capacity and an estimated half life of the drug substance and gives dosing recommendations on an individual basis for each patient.

6.3 Conclusions and outlook This preliminary study should raise the awareness for the topic drug interactions in the radiooncologic setting. Even if the study is limited in terms of number of patients, the outcomes can be considered as a step forward in risk management in daily clinical practice. The results in the small cohort with highly selected patients directly reflect the results of more extensive studies (Lao et al., 2013).

Despite all concrete drug treatment instructions the fact remains that the fewer medications are used the better. Physicians should keep in mind the correlation between increasing number of drugs and pDDIs. More respect for resident physicians with regard to drug treatment on behalf of the patients is advisable due to the fact that the number of pDDIs increased significantly at hospital discharge. Questionable is whether the general practitioner cancels and modifies the treatment changes or continues therapy unchanged. Prescribing drugs with a low risk for pDDIs as well as careful monitoring for ADRs are important measures to prevent harm associated with pDDIs.

118

Medical treatment with drugs associated with a high potential for pDDIs and or ADRs may not always be avoided depending on the underlying disease. Recent research showed that knowledge about the potential risk can help to take appropriate measures to decrease the probability for an adverse outcome e.g. dose adjustment, monitoring lab values and replacement by an alternative drug.

Periodically drug interaction screenings should be performed to update the preliminary recommendation list developed in this study in order to reduce the overall risk of pDDIs in the scope of risk assessment (STEP 5 Chapter 1.2). Documentation patterns in Germany are sufficient. Therefore future studies aiming at risk assessment of pDDIs and PIMs can be performed retro- spectively and should include a higher number of patients. These studies should contain the most prevalent criteria of the STOPP criteria and the basic elements of the Cave module for identification of PIM prevalences. The application of these criteria should help to optimise drug treatment in the radiooncologic setting.

The preliminary study showed that potential drug interactions can mostly be detected by clinical parameters. The study by (Geerts et al., 2009) yielded similar results. With this background information drug charts should in future studies been screened for clinical manifest symptoms if pDDIs were previously detected by the computer screening in order to evaluate the real clinical relevance of potential drug interactions. A causality assessment of manifest DDIs using the Na- ranjo algorithm should be performed in future studies (Zaki, 2011).

119

Bibliography

Abarca, J., Colon, L. R., Wang, V. S., Malone, D. C., Murphy, J. E. and Armstrong, E. P. (2006). Evaluation of the performance of drug-drug interaction screening software in community and hospital pharmacies. J Manag Care Pharm 12 (5), 383-389

ABDATA Pharma-Daten-Service (2012). CAVE Beschreibung der Inhalte für Apotheken. Werbe- und Vertriebsgesellschaft Deutscher Apotheker mbH, Eschborn

ABDATA Pharma-Daten-Service (2007). CAVE Leitfaden für Apotheken-SWH. Werbe- und Vertriebsgesellschaft Deutscher Apotheker mbH, Eschborn

ABDATA-Pharma-Service (2012). ABDA Datenbank und Aktuelle Info für Apotheken. Werbe- und Vertriebsgesellschaft Deutscher Apotheker mbH, Eschborn

ABDATA-Pharma-Service (2012). ABDATA-Kurzinfo ABDA und CAVE (Englisch). Eschborn

ABZ Pharma (2013). (Zugriff vom 12.06.2013). Novaminsulfon 500 mg/ml oral Drops; DE/H/1189/001/DC. www.hma.eu/.../spc_de118901novaminsulfon.pdf

Ahmad, S. (1995). Allopurinol and enalapril. Drug induced anaphylactic coronary spasm and acute myocardial infarction. Chest 108 (2), 586

Albain, K. S., Crowley, J. J., LeBlanc, M. and Livingston, R. B. (1991). Survival determinants in extensive-stage non-small-cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 9 (9), 1618-1626

Aljadhey, H., Tu, W., Hansen, R. A., Blalock, S. J., Brater, D. C. and Murray, M. D. (2012). Comparative effects of non-steroidal anti-inflammatory drugs (NSAIDs) on blood pressure in patients with hypertension. BMC Cardiovasc Disord 12 (93), 93 91-10

Amann, U., Schmedt, N. and Garbe, E. (2012). Prescribing of potentially inappropriate medications for the elderly: an analysis based on the PRISCUS list. Dtsch Arztebl Int 109 (5), 69-75

Andersson, C., van Gaal, L., Caterson, I. D., Weeke, P., James, W. P., Coutinho, W., Finer, N., Sharma, A. M., Maggioni, A. P. and Torp-Pedersen, C. (2012). Relationship between HbA1c levels and risk of cardiovascular adverse outcomes and all-cause mortality in overweight and obese cardiovascular high-risk women and men with type 2 diabetes. Diabetologia 55 (9), 2348- 2355

Anthierens, S., Tansens, A., Petrovic, M. and Christiaens, T. (2010). Qualitative insights into general practitioners views on polypharmacy. BMC Fam Pract 11, 65

Aronson, J. (2000). Adverse drug reactions: definitions, diagnosis, and management. The Lancet 356, 1255-1259

Avery, T., Barber, N., Ghaleb, M., Franklin, B., Armstrong, S., Crowe, S., Dhillon, S., Freyer, A., Howard, R., Pezzolesi, H. and Serumaga, B. (2012). Investigating the prevalence and causes of prescribing Errors in General practice: The PRACTICE Study A Report for the GMC. University of Nottingham

120

Avorn, J. and Soumerai, S. B. (1983). Improving drug-therapy decisions through educational outreach. A randomized controlled trial of academically based "detailing". N Engl J Med 308 (24), 1457-1463

Bacic-Vrca, V., Marusic, S., Erdeljic, V., Falamic, S., Gojo-Tomic, N. and Rahelic, D. (2010). The incidence of potential drug-drug interactions in elderly patients with arterial hypertension. Pharm World Sci 32 (6), 815-821

Barmer GEK (2012). Barmer GEK Arzneimittelreport 2012. Pressekonferenz, Berlin

Becker, M. L., Kallewaard, M., Caspers, P. W., Visser, L. E., Leufkens, H. G. and Stricker, B. H. (2007). Hospitalisations and emergency department visits due to drug-drug interactions: a literature review. Pharmacoepidemiol Drug Saf 16 (6), 641-651

Beers, M. H. (1999). Aging as a risk factor for medication - related problems : presented at the Gerontological Society of America Annual Meeting Post-Conference Symposium, Philadelphia, Pennsylvania, November 24, 1998. The Consultant Pharmacist 14 (12), 1337-1340

Beers, M. H., Ouslander, J. G., Rollingher, I., Reuben, D. B., Brooks, J. and Beck, J. C. (1991). Explicit criteria for determining inappropriate medication use in nursing home residents. UCLA Division of Geriatric Medicine. Arch Intern Med 151 (9), 1825-1832

Bergk, V., Gasse, C., Rothenbacher, D., Loew, M., Brenner, H. and Haefeli, W. E. (2004). Drug interactions in primary care: impact of a new algorithm on risk determination. Clin Pharmacol Ther 76 (1), 85-96

Bertsche, T., Pfaff, J., Schiller, P., Kaltschmidt, J., Pruszydlo, M. G., Stremmel, W., Walter- Sack, I., Haefeli, W. E. and Encke, J. (2010). Prevention of adverse drug reactions in intensive care patients by personal intervention based on an electronic clinical decision support system. Intensive Care Med 36 (4), 665-672

BfArM (2006). (Zugriff vom 12.11.2012). Pharmacovigilance. http://www.bfarm.de/EN/vigilance/vigilance-node-en.html

Bicket, D. P. (2002). Using ACE inhibitors appropriately. Am Fam Physician 66 (3), 461-468

Bjerrum, L., Gonzalez Lopez-Valcarcel, B. and Petersen, G. (2008). Risk factors for potential drug interactions in general practice. Eur J Gen Pract 14 (1), 23-29

Blagden, S. P., Charman, S. C., Sharples, L. D., Magee, L. R. and Gilligan, D. (2003). Performance status score: do patients and their oncologists agree? Br J Cancer 89 (6), 1022- 1027

Blix, H. S., Viktil, K. K., Moger, T. A. and Reikvam, A. (2008). Identification of drug interactions in hospitals-computerized screening vs. bedside recording. J Clin Pharm Ther 33 (2), 131-139

Blower, P., de Wit, R., Goodin, S. and Aapro, M. (2005). Drug-drug interactions in oncology: why are they important and can they be minimized? Crit Rev Oncol Hematol 55 (2), 117-142

Bosch, J., Yusuf, S., Gerstein, H. C., Pogue, J., Sheridan, P., Dagenais, G., Diaz, R., Avezum, A., Lanas, F., Probstfield, J., Fodor, G. and Holman, R. R. (2006). Effect of ramipril on the incidence of diabetes. N Engl J Med 355 (15), 1551-1562

121

Braunstein, J. B., Anderson, G. F., Gerstenblith, G., Weller, W., Niefeld, M., Herbert, R. and Wu, A. W. (2003). Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 42 (7), 1226-1233

Briot, K. and Roux, C. (2013). [Corticosteroid-induced osteoporosis]. Rev Med Interne 34 (5), 315-323

Browener, S. and Delman, B. (2008). (Zugriff vom 28.05.2013). Polypharmacy in the Cancer Patient With Multiple Comorbid Conditions: Clinical Pharmacology and the Physiology of Aging. http://www.medscape.org/viewarticle/580628_3

Buajordet, I., Ebbesen, J., Erikssen, J., Svaar, H., Brors, O. and Hilberg, T. (1995). Frequency, reporting and classification of drug-related deaths. Tidsskr Nor Laegeforen 115 (19), 2373-2375

Bucsa, C., Farcas, A., Cazacu, I., Leucuta, D., Achimas-Cadariu, A., Mogosan, C. and Bojita, M. (2013). How many potential drug-drug interactions cause adverse drug reactions in hospitalized patients? Eur J Intern Med 24 (1), 27-33

Budnitz, D. S., Shehab, N., Kegler, S. R. and Richards, C. L. (2007). Medication use leading to emergency department visits for adverse drug events in older adults. Ann Intern Med 147 (11), 755-765

Bundesinstitut für Arzneimittel und Medizinprodukte (2013). (Zugriff vom 01.08.2013). Verkehrsfähige Arzneimittel im Zuständigkeitsbereich des BfArM. http://www.bfarm.de/DE/Arzneimittel/4_statistik/statistik-node.html

Cancer Research UK (2010). (Zugriff vom 12.08.2012). Cancer incidence by age. http://www.cancerresearchuk.org/cancer-info/cancerstats/incidence/age/

Caskie, G. I. and Willis, S. L. (2004). Congruence of self-reported medications with pharmacy prescription records in low-income older adults. Gerontologist 44 (2), 176-185

Chan, A., Tan, S. H., Wong, C. M., Yap, K. Y. and Ko, Y. (2009). Clinically significant drug- drug interactions between oral anticancer agents and nonanticancer agents: a Delphi survey of oncology pharmacists. Clin Ther 31 (Theme Issue), 2379-2386

Charlson, M. E., Pompei, P., Ales, K. L. and MacKenzie, C. R. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40 (5), 373-383

Chelan-Douglas Health District (2012). (Zugriff vom 08.08.2012). Adverse drug events, adverse drug reactions, medication errors and side effects. http://www.cdhd.wa.gov/documents/PP3400190AdverseDrugEventsDrugReactionsMedErrorsS ideEffects.doc

Clackamas Community College (2001). (Zugriff vom 14.07.2013). Enzyme inhibition. http://dl.clackamas.cc.or.us/ch106-08/enzyme1.htm

Colt, H. G. and Shapiro, A. P. (1989). Drug-induced illness as a cause for admission to a community hospital. J Am Geriatr Soc 37 (4), 323-326

Corcoran, M. E. (1997). Polypharmacy in the Older Patient With Cancer. Cancer Control 4 (5), 419-428

Cresswell, K. M., Fernando, B., McKinstry, B. and Sheikh, A. (2007). Adverse drug events in the elderly. Br Med Bull 83, 259-274

122

Cruciol-Souza, J. M. and Thomson, J. C. (2006). Prevalence of potential drug-drug interactions and its associated factors in a Brazilian teaching hospital. J Pharm Pharm Sci 9 (3), 427-433

Deutsche Gesellschaft für Allgemeinmedizin und Familienmedizin (DEGAM) and Leitliniengruppe Hessen (2013). Hausärztliche Leitlinie Multimedikation : Empfehlungen zum Umgang mit Multimedikation bei Erwachsenen und geriatrischen Patienten.

Dormann, H., Sonst, A., Muller, F., Vogler, R., Patapovas, A., Pfistermeister, B., Plank-Kiegele, B., Kirchner, M., Hartmann, N., Burkle, T. and Maas, R. (2013). Adverse drug events in older patients admitted as an emergency: the role of potentially inappropriate medication in elderly people (PRISCUS). Dtsch Arztebl Int 110 (13), 213-219

Dorresteijn, L. D., Kappelle, A. C., Boogerd, W., Klokman, W. J., Balm, A. J., Keus, R. B., van Leeuwen, F. E. and Bartelink, H. (2002). Increased risk of ischemic stroke after radiotherapy on the neck in patients younger than 60 years. J Clin Oncol 20 (1), 282-288

Doubova, S. V., Reyes-Morales, H., Torres-Arreola Ldel, P. and Suarez-Ortega, M. (2007). Potential drug-drug and drug-disease interactions in prescriptions for ambulatory patients over 50 years of age in family medicine clinics in Mexico City. BMC Health Serv Res 7, 147

Dreisbach, A. W. and Lertora, J. J. (2003). The effect of chronic renal failure on hepatic drug metabolism and drug disposition. Semin Dial 16 (1), 45-50

Egger, S. (2007). Potential Risk factors for Adverse drug reactions in elderly people; PhD Thesis; http://edoc.unibas.ch/583/. University of Basel / Faculty of Science Basel

Egger, S. S., Ratz Bravo, A. E., Hess, L., Schlienger, R. G. and Krahenbuhl, S. (2007). Age- related differences in the prevalence of potential drug-drug interactions in ambulatory dyslipidaemic patients treated with statins. Drugs Aging 24 (5), 429-440 eHealthMe (2013). (Zugriff vom 18.07.2013). From FDA reports: Effexor and Low blood sugar. www.ehealthme.com/ds/effexor/low+blood+sugar

Eichenberger, P. M., Lampert, M. L., Kahmann, I. V., van Mil, J. W. and Hersberger, K. E. (2010). Classification of drug-related problems with new prescriptions using a modified PCNE classification system. Pharm World Sci 32 (3), 362-372

Elasy, T., Kaminsky, D., Tracy, M. and Mehler, P. S. (1995). Allopurinol hypersensitivity syndrome revisited. West J Med 162 (4), 360-361

Ellitt, G. R., Engblom, E., Aslani, P., Westerlund, T. and Chen, T. F. (2010). Drug related problems after discharge from an Australian teaching hospital. Pharm World Sci 32 (5), 622- 630

Espinosa-Bosch, M., Santos-Ramos, B., Gil-Navarro, M. V., Santos-Rubio, M. D., Marin-Gil, R. and Villacorta-Linaza, P. (2012). Prevalence of drug interactions in hospital healthcare. Int J Clin Pharm 34 (6), 807-817

European Medicines Agency (2013). (Zugriff vom 24.07.2013). New safety advice for diclofenac – CMDh endorses PRAC recommendation. www.ema.europa.eu/.../news_and_events/news/2013/06/news_detail_001830.jsp&mid=WC0b0 1ac058004d5c1

Evans, R. S., Lloyd, J. F., Stoddard, G. J., Nebeker, J. R. and Samore, M. H. (2005). Risk factors for adverse drug events: a 10-year analysis. Ann Pharmacother 39 (7-8), 1161-1168

123

Ewachiw, B. D., Browner, I. S. and Basile, D. A. (2008). (Zugriff vom 02.05.2011). Polypharmacy: Managing Chemotherapy in the Older Cancer Patient with Multiple Comorbid Conditions. http://www.medscape.org/viewarticle/580628

Faloon, W. (2011). (Zugriff vom 5.17.2013). Cancer Establishment Hides Radiation Side Effects. http://www.lef.org/magazine/mag2011/dec2011_Cancer-Establishment-Hides- Radiation-Side-Effects_01.htm

Fick, D. M., Cooper, J. W., Wade, W. E., Waller, J. L., Maclean, J. R. and Beers, M. H. (2003). Updating the Beers criteria for potentially inappropriate medication use in older adults: results of a US consensus panel of experts. Arch Intern Med 163 (22), 2716-2724

Fick, D. M., Mion, L. C., Beers, M. H. and J, L. W. (2008). Health outcomes associated with potentially inappropriate medication use in older adults. Res Nurs Health 31 (1), 42-51

Finkelstein, D. M., Cassileth, B. R., Bonomi, P. D., Ruckdeschel, J. C., Ezdinli, E. Z. and Wolter, J. M. (1988). A pilot study of the Functional Living Index-Cancer (FLIC) Scale for the assessment of quality of life for metastatic lung cancer patients. An Eastern Cooperative Oncology Group study. Am J Clin Oncol 11 (6), 630-633

Fischalek, J. (2009). Interaktionen von Arzneimitteln in der Vertragsärztlichen Versorgung. Profund (4), 6 - 9

Focus Verlag GmbH (2012). (Zugriff vom 15.08.2013). Tödliche Pillencoctails - Zehntausende sterben jährlich an den Wechselwirkungen von Medikamenten: AUDIO. http://www.themenportal.de/audio/toedliche-pillencocktails-zehntausende-sterben-jaehrlich-an- den-wechselwirkungen-von-medikamenten-o-ton-paket

Fokter, N., Mozina, M. and Brvar, M. (2010). Potential drug-drug interactions and admissions due to drug-drug interactions in patients treated in medical departments. Wien Klin Wochenschr 122 (3-4), 81-88

Ford, N. F. and Taubert, D. (2013). Clopidogrel, CYP2C19, and a Black Box. J Clin Pharmacol 53 (3), 241-248

Forster, A. J., Murff, H. J., Peterson, J. F., Gandhi, T. K. and Bates, D. W. (2005). Adverse drug events occurring following hospital discharge. J Gen Intern Med 20 (4), 317-323

Frechen, S., Zoeller, A., Ruberg, K., Voltz, R. and Gaertner, J. (2012). Drug interactions in dying patients: a retrospective analysis of hospice inpatients in Germany. Drug Saf 35 (9), 745- 758

Fulton, M. M. and Allen, E. R. (2005). Polypharmacy in the elderly: a literature review. J Am Acad Nurse Pract 17 (4), 123-132

Gaertner, J., Ruberg, K., Schlesiger, G., Frechen, S. and Voltz, R. (2012). Drug interactions in palliative care--it's more than cytochrome P450. Palliat Med 26 (6), 813-825

Gagne, J. J., Maio, V. and Rabinowitz, C. (2008). Prevalence and predictors of potential drug- drug interactions in Regione Emilia-Romagna, Italy. J Clin Pharm Ther 33 (2), 141-151

Gallagher, P., Barry, P. and O'Mahony, D. (2007). Inappropriate prescribing in the elderly. J Clin Pharm Ther 32 (2), 113-121

124

Gallagher, P., Ryan, C., Byrne, S., Kennedy, J. and O'Mahony, D. (2008). STOPP (Screening Tool of Older Person's Prescriptions) and START (Screening Tool to Alert doctors to Right Treatment). Consensus validation. Int J Clin Pharmacol Ther 46 (2), 72-83

Geerts, A. F., De Koning, F. H., De Smet, P. A., Van Solinge, W. W. and Egberts, T. C. (2009). Laboratory tests in the clinical risk management of potential drug-drug interactions: a cross- sectional study using drug-dispensing data from 100 Dutch community pharmacies. Drug Saf 32 (12), 1189-1197

Gerdemann, A., Griese, N. and Schulz, M. (2008). Allopurinol und ACE-Hemmer. Pharmazeutische Zeitung 153 (18), 32-33

Ginsberg, G., Hattis, D., Russ, A. and Sonawane, B. (2005). Pharmacokinetic and pharmacodynamic factors that can affect sensitivity to neurotoxic sequelae in elderly individuals. Environ Health Perspect 113 (9), 1243-1249

Girre, V., Arkoub, H., Puts, M. T., Vantelon, C., Blanchard, F., Droz, J. P. and Mignot, L. (2011). Potential drug interactions in elderly cancer patients. Crit Rev Oncol Hematol 78 (3), 220-226

Givens, M. L. and Crandall, J. (2010). Renal complications in oncologic patients. Hematol Oncol Clin North Am 24 (3), 567-575

Glintborg, B., Andersen, S. E. and Dalhoff, K. (2005). Drug-drug interactions among recently hospitalised patients--frequent but mostly clinically insignificant. Eur J Clin Pharmacol 61 (9), 675-681

Goldberg, R. M., Mabee, J., Chan, L. and Wong, S. (1996). Drug-drug and drug-disease interactions in the ED: analysis of a high-risk population. Am J Emerg Med 14 (5), 447-450

Gorzoni, M. L., Fabbri, R. M. and Pires, S. L. (2008). Beer-Fick criteria and generic drugs in Brazil. Rev Assoc Med Bras 54 (4), 353-356

Green, J. L., Hawley, J. N. and Rask, K. J. (2007). Is the number of prescribing physicians an independent risk factor for adverse drug events in an elderly outpatient population? Am J Geriatr Pharmacother 5 (1), 31-39

Griese, N. (2012). Arzneimittel sicher kombinieren. Pharmazeutische Zeitung 157 (24), 26-27

Grossman, E. and Messerli, F. H. (2012). Drug-induced hypertension: an unappreciated cause of secondary hypertension. Am J Med 125 (1), 14-22

Gyamlani, G. and Geraci, S. A. (2007). Secondary hypertension due to drugs and toxins. South Med J 100 (7), 692-699; quiz 700, 708

Gyenes, G., Rutqvist, L. E., Liedberg, A. and Fornander, T. (1998). Long-term cardiac morbidity and mortality in a randomized trial of pre- and postoperative radiation therapy versus surgery alone in primary breast cancer. Radiother Oncol 48 (2), 185-190

Haas, L. M. (2004). Utilizing geriatric skills in radiation oncology. Geriatric Nursing (Lond) 25 (6), 355-360

Haas, M. and Kuehn, E. (2001). Head and neck cancers. in Watkins-Bruner, D., Moore-Higgs, G. J. and Haas, M. (Hrsg.). Outcomes in radiation therapy: multidisciplinary management. Jones and Bartlett, Sudbury, MA, 195-213

125

Haddad, A., Davis, M. and Lagman, R. (2007). The pharmacological importance of cytochrome CYP3A4 in the palliation of symptoms: review and recommendations for avoiding adverse drug interactions. Support Care Cancer 15 (3), 251-257

Hajjar, E. R., Cafiero, A. C. and Hanlon, J. T. (2007). Polypharmacy in elderly patients. Am J Geriatr Pharmacother 5 (4), 345-351

Hall, W. H., Ramachandran, R., Narayan, S., Jani, A. B. and Vijayakumar, S. (2004). An electronic application for rapidly calculating Charlson comorbidity score. BMC Cancer 4, 94

Hamilton, H., Gallagher, P., Ryan, C., Byrne, S. and O'Mahony, D. (2011). Potentially Inappropriate Medications Defined by STOPP Criteria and the Risk of Adverse Drug Events in Older Hospitalized Patients. Arch Intern Med 171 (11), 1013-1019

Hanigan, M. H., Dela Cruz, B. L., Thompson, D. M., Farmer, K. C. and Medina, P. J. (2008). Use of prescription and nonprescription medications and supplements by cancer patients during chemotherapy: questionnaire validation. J Oncol Pharm Pract 14 (3), 123-130

Hanlon, J. T., Pieper, C. F., Hajjar, E. R., Sloane, R. J., Lindblad, C. I., Ruby, C. M. and Schmader, K. E. (2006). Incidence and predictors of all and preventable adverse drug reactions in frail elderly persons after hospital stay. J Gerontol A Biol Sci Med Sci 61 (5), 511-515

Hansten, P. D. and Horn, J. R. (2012). Drug Interactions Analysis and Managment. Lippincott Williams & Wilkins London

Hansten, P. D., Horn, J. R. and Hazlet, T. K. (2001). ORCA: OpeRational ClassificAtion of drug interactions. J Am Pharm Assoc (Wash) 41 (2), 161-165

Harrington, C. J. and Zaydfudim, V. (2010). Buprenorphine maintenance therapy hinders acute pain management in trauma. Am Surg 76 (4), 397-399

Harrison, L., Shasha, D., Shiaova, L., White, C., Ramdeen, B. and Portenoy, R. (2001). Prevalence of anemia in cancer patients undergoing radiation therapy. Semin Oncol 28 (2 Suppl 8), 54-59

Hasan, S. S., Lim, K. N., Anwar, M., Sathvik, B. S., Ahmadi, K., Yuan, A. W. and Kamarunnesa, M. A. (2012). Impact of pharmacists' intervention on identification and management of drug-drug interactions in an intensive care setting. Singapore Med J 53 (8), 526- 531

Himmel, W., Tabache, M. and Kochen, M. M. (1996). What happens to long-term medication when general practice patients are referred to hospital? Eur J Clin Pharmacol 50 (4), 253-257

Hoffmann, B. and Beyer, M. (2006). Fehler des Monats: Interaktionen zwischen Medikamenten beachten. Ärztemagazin (21), 21-22

Holt, S., Schmiedl, S. and Thurmann, P. A. (2010). Potentially inappropriate medications in the elderly: the PRISCUS list. Dtsch Arztebl Int 107 (31-32), 543-551

Holtzman, C. W., Wiggins, B. S. and Spinler, S. A. (2006). Role of P-glycoprotein in statin drug interactions. Pharmacotherapy 26 (11), 1601-1607

Horiot, J. C. (2004). Prophylaxis versus treatment: is there a better way to manage radiotherapy- induced nausea and vomiting? Int J Radiat Oncol Biol Phys 60 (4), 1018-1025

126

Horn, J. R. and Hansten, P. (2004). Sources of Error in Drug Interactions: The Swiss Cheese Model. Pharmacy Times (3), 53-54

Horn, J. R. and Hansten, P. D. (2009). Ignoring Drug Interactions for the Right Reasons. Pharmacy Times (11), 33

Hoyumpa, A. and Schenker, S. (1982). Major Drug Interactions: Effect of Liver Disease, Alcohol, and Malnutrition. Annual Review of Medicine 33, 113-149

Hubbard, D. (2009). The failure of risk management: why it's broken and how to fix it. John Wiley & Sons Hoboken, NJ

Iacobellis, G. (2006). Drug-Drug Interactions in the Metabolic Syndrome. Nova Science Pub Inc New York

Indermitte, J., Reber, D., Beutler, M., Bruppacher, R. and Hersberger, K. E. (2007). Prevalence and patient awareness of selected potential drug interactions with self-medication. J Clin Pharm Ther 32 (2), 149-159

Indermitte, J. L. (2006). Potential drug interactions : exposure and management in hospital and ambulatory settings; Thesis. University of Basel / Faculty of Science Basel

Ismail, M., Iqbal, Z., Khattak, M. B., Khan, M. I., Arsalan, H., Javaid, A., Gul, Q. and Khan, F. (2013). Potential drug-drug interactions in internal medicine wards in hospital setting in Pakistan. Int J Clin Pharm 35 (3), 455-462

Johnell, K. and Klarin, I. (2007). The relationship between number of drugs and potential drug- drug interactions in the elderly: a study of over 600,000 elderly patients from the Swedish Prescribed Drug Register. Drug Saf 30 (10), 911-918

Juurlink, D., Mamdani, M., Iazzetta, J. and Etchells, E. (2004). Avoiding drug interactions in hospitalized patients. Healthc Q. 7 (2), 27-28

Juurlink, D. N., Mamdani, M., Kopp, A., Laupacis, A. and Redelmeier, D. A. (2003). Drug-drug interactions among elderly patients hospitalized for drug toxicity. JAMA 289 (13), 1652-1658

Kannan, G., Anitha, R., Rani, V. N., Thennarasu, P., Alosh, J., Vasantha, J., Martin, J. R. and Uma, M. R. (2011). A study of drug-drug interactions in cancer patients of a south Indian tertiary. J Postgrad Med 57 (3), 206-210

Karas, S., Jr. (1981). The potential for drug interactions. Ann Emerg Med 10 (12), 627-630

Kazui, M., Nishiya, Y., Ishizuka, T., Hagihara, K., Farid, N. A., Okazaki, O., Ikeda, T. and Kurihara, A. (2010). Identification of the human cytochrome P450 enzymes involved in the two oxidative steps in the bioactivation of clopidogrel to its pharmacologically active metabolite. Drug Metab Dispos 38 (1), 92-99

Kinirons, M. T. and Crome, P. (1997). Clinical pharmacokinetic considerations in the elderly. An update. Clin Pharmacokinet 33 (4), 302-312

Krahenbuhl-Melcher, A. and Krahenbuhl, S. (2005). Hospital drug safety: medication errors and adverse drug reactions. Praxis (Bern 1994) 94 (24-25), 1031-1038

Krähenbühl-Melcher, A. (2005). Hospital Drug-Safety-Role of the Pharmacist; Thesis. University of Basel / Faculty of Science Basel

127

Köhler, G. I. (2001). Pharmakologische Untersuchungen zu Arzneimittelverodnungen und Arzenimittelinteraktionen am Beispiel der Pharmakotherapie von patienten mit koronarer Herzkrankheit in einer Kohortenstudie; Dissertation. Universität Hamburg / Fakultät für Chemie Hamburg

Laine, L. and Hennekens, C. (2010). Proton pump inhibitor and clopidogrel interaction: fact or fiction? Am J Gastroenterol 105 (1), 34-41

Lam, M. P. and Cheung, B. M. (2012). The use of STOPP/START criteria as a screening tool for assessing the appropriateness of medications in the elderly population. Expert Rev Clin Pharmacol 5 (2), 187-197

Lampert, M. L., Kraehenbuehl, S. and Hug, B. L. (2008). Drug-related problems: evaluation of a classification system in the daily practice of a Swiss University Hospital. Pharm World Sci 30 (6), 768-776

Lao, C. K., Ho, S. C., Chan, K. K., Tou, C. F., Tong, H. H. and Chan, A. (2013). Potentially inappropriate prescribing and drug-drug interactions among elderly Chinese nursing home residents in Macao: [Epub ahead of print]. Int J Clin Pharm (Zugriff vom Jun 28). http://link.springer.com/article/10.1007%2Fs11096-013-9811-y

Lapi, F., Azoulay, L., Yin, H., Nessim, S. J. and Suissa, S. (2013). Concurrent use of diuretics, angiotensin converting enzyme inhibitors, and angiotensin receptor blockers with non-steroidal anti-inflammatory drugs and risk of acute kidney injury: nested case-control study. BMJ 346, e8525-8521 -e8525 -8511

Lapi, F., Vietri, M., Moschini, M., Cecchi, E., Pugi, A., Lucenteforte, E., Banchelli, G., Di Pirro, M., Gallo, E., Mugelli, A. and Vannacci, A. (2010). Potential drug-drug interactions and radiodiagnostic procedures: an in-hospital survey. Pharm World Sci 32 (2), 200-205

Lea, M., Rognan, S. E., Koristovic, R., Wyller, T. B. and Molden, E. (2013). Severity and Management of Drug-Drug Interactions in Acute Geriatric Patients: [Epub ahead of print]. Drugs Aging (Zugriff vom May 17). http://link.springer.com/article/10.1007%2Fs40266-013- 0091-y

Leape, L. L. (1995). Preventing adverse drug events. Am J Health Syst Pharm 52 (4), 379-382

Leary, W. A. and Reyes, A. J. (1984). Drug interactions with diuretics. S Afr Med J 65 (12), 455-461

Lees, J. and Chan, A. (2011). Polypharmacy in elderly patients with cancer: clinical implications and managment. Lancet Oncol 12 (13), 1249-1257

Lendrath, G. (2009). (Zugriff vom 10.06.2013). ABDA Datenbank Handbuch. www.acs- hl.de/webservice/HANDBUCH_K2.pdf

Lepori, V., Perren, A. and Marone, C. (1999). Adverse internal medicine drug effects at hospital admission. Schweiz Med Wochenschr 129 (24), 915-922

Levine, J. P. (2006). Pharmacologic and nonpharmacologic management of osteoporosis. Clin Cornerstone 8 (1), 40-53

Lindblad, C. I., Artz, M. B., Pieper, C. F., Sloane, R. J., Hajjar, E. R., Ruby, C. M., Schmader, K. E. and Hanlon, J. T. (2005). Potential drug-disease interactions in frail, hospitalized elderly veterans. Ann Pharmacother 39 (3), 412-417

128

Lubinga, S. J. and Uwiduhaye, E. (2011). Potential drug-drug interactions on in-patient medication prescriptions at Mbarara Regional Referral Hospital (MRRH) in western Uganda: prevalence, clinical importance and associated factors. Afr Health Sci 11 (3), 499-507

Lucchetti, G., Lucchetti, A. L., Pires, S. L. and Gorzoni, M. L. (2011). Beers-Fick criteria and drugs available through the Farmacia Dose Certa program. Sao Paulo Med J 129 (1), 17-22

Ludwig, H. and Strasser, K. (2001). Symptomatology of anemia. Semin Oncol 28 (2 Suppl 8), 7-14

Lyman, G. H., Khorana, A. A., Kuderer, N. M., Lee, A. Y., Arcelus, J. I., Balaban, E. P., Clarke, J. M., Flowers, C. R., Francis, C. W., Gates, L. E., Kakkar, A. K., Key, N. S., Levine, M. N., Liebman, H. A., Tempero, M. A., Wong, S. L., Prestrud, A. A. and Falanga, A. (2013). Venous Thromboembolism Prophylaxis and Treatment in Patients With Cancer: American Society of Clinical Oncology Clinical Practice Guideline Update. J Clin Oncol 31 (17), 2189- 2204

Maggiore, R. J., Gross, C. P. and Hurria, A. (2010). Polypharmacy in older adults with cancer. Oncologist 15 (5), 507-522

Magro, L., Conforti, A., Del Zotti, F., Leone, R., Iorio, M. L., Meneghelli, I., Massignani, D., Visona, E. and Moretti, U. (2008). Identification of severe potential drug-drug interactions using an Italian general-practitioner database. Eur J Clin Pharmacol 64 (3), 303-309

Mallet, L., Spinewine, A. and Huang, A. (2007). The challenge of managing drug interactions in elderly people. Lancet 370 (9582), 185-191

Malone, D. C., Anthony, M. and Johnson, L. P. (2003). Preventing the Alignment of the Holes in Swiss Cheese :How Multiple Systems Fail To Prevent Drug-Drug Interactions. HealthDay held at the University of Colorado Health Sciences Center on November 14, 2003, Denver. http://www.chainonline.org/CHAINOnline/assets/File/Eisenberg_AZCERT_120927.pdf

Mangoni, A. A. and Jackson, S. H. (2004). Age-related changes in pharmacokinetics and pharmacodynamics: basic principles and practical applications. Br J Clin Pharmacol 57 (1), 6- 14

Marchetti, S., Mazzanti, R., Beijnen, J. H. and Schellens, J. H. (2007). Concise review: Clinical relevance of drug drug and herb drug interactions mediated by the ABC transporter ABCB1 (MDR1, P-glycoprotein). Oncologist 12 (8), 927-941

Marusic, S., Lisicic, A., Horvatic, I., Bacic-Vrca, V. and Bozina, N. (2012). Atorvastatin-related rhabdomyolysis and acute renal failure in a genetically predisposed patient with potential drug- drug interaction. Int J Clin Pharm 34 (6), 825-827

Mateti, U., Rajakannan, T., Nekkanti, H., Rajesh, V., Mallaysamy, S. and Ramachandran, P. (2011). Drug-drug Interactions in Hospitalized Cardiac Patients. J Young Pharm 3 (4), 329-333

Mateti, U. V., Nekkanti, H., Vilakkathala, R., Rajakannan, T., Mallayasamy, S. and Ramachandran, P. (2012). Pattern of Angiotensin-converting enzyme inhibitors induced adverse drug reactions in South Indian teaching hospital. N Am J Med Sci 4 (4), 185-189

Maurer, J., Schäfer, C., Maurer, O. and Kölbl, O. (2012). Angst und Depressivität bei Tumorpatienten im Verlauf der radioonkologischen Behandlung. Strahlentherapie und Onkologie 188 (10), 940-945

129

Mayer, S. and Schneider, J. (2006). Interaktionen erkennen und vermeiden. Pharmazeutische Zeitung 151 (29), 29-36

Mayhew, S. L. and Christensen, M. L. (1993). Pharmacokinetic alterations in malnutrition and obesity. Hosp Pharm 28, 836-837, 840-842, 850

McAlpine, D. E., Biernacka, J. M., Mrazek, D. A., O'Kane, D. J., Stevens, S. R., Langman, L. J., Courson, V. L., Bhagia, J. and Moyer, T. P. (2011). Effect of cytochrome P450 enzyme polymorphisms on pharmacokinetics of venlafaxine. Ther Drug Monit 33 (1), 14-20

McKenzie, W. S. and Rosenberg, M. (2010). Paradoxical reaction following administration of a benzodiazepine. J Oral Maxillofac Surg 68 (12), 3034-3036

McLean, A. J. and Le Couteur, D. G. (2004). Aging biology and geriatric clinical pharmacology. Pharmacol Rev 56 (2), 163-184

Merrill, C. T. and Elixhauser, A. (2005). Hospitalization in the United States : 2002. Agency for Healthcare Research and Quality Rockville, MD

Mignat, C. and Unger, T. (1995). ACE inhibitors. Drug interactions of clinical significance. Drug Saf 12 (5), 334-347

Mille, F., Schwartz, C., Brion, F., Fontan, J. E., Bourdon, O., Degoulet, P. and Jaulent, M. C. (2008). Analysis of overridden alerts in a drug-drug interaction detection system. Int J Qual Health Care 20 (6), 400-405

Miranda, V., Fede, A., Nobuo, M., Ayres, V., Giglio, A., Miranda, M. and Riechelmann, R. P. (2011). Adverse Drug Reactions and Drug Interactions as Causes of Hospital Admission in Oncology. J Pain Symptom Manage 42 (3), 342-353

Mitchell, A. J., Chan, M., Bhatti, H., Halton, M., Grassi, L., Johansen, C. and Meader, N. (2011). Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies. Lancet Oncol 12 (2), 160-174

Miura, T., Kojima, R., Sugiura, Y., Mizutani, M., Takatsu, F. and Suzuki, Y. (2000). Effect of aging on the incidence of digoxin toxicity. Ann Pharmacother 34 (4), 427-432

Moroney, A. (2012). (Zugriff vom 12.07.2013). Overview of pharmacodynamics. http://www.merckmanuals.com/professional/clinical_pharmacology/pharmacodynamics/overvie w_of_pharmacodynamics.html

Murphy, J. E., Malone, D. C., Olson, B. M., Grizzle, A. J., Armstrong, E. P. and Skrepnek, G. H. (2009). Development of computerized alerts with management strategies for 25 serious drug- drug interactions. Am J Health Syst Pharm 66 (1), 38-44

National Medicines Information Centre (2008). Drug interactions I: How they occur. National Medicines Information Centre 14 (4), 1-6

Navari, R. M. (2003). Pathogenesis-based treatment of chemotherapy-induced nausea and vomiting - two new agents. J Support Oncol 1 (2), 89-103

O'Brien, P. H., Moss, W. T., Louwenaar, K., May, J. and Peterson, L. (1968). The effect of acute anemia on the radiosensitivity of V2 rabbit sarcoma. Radiology 91 (6), 1159-1162

130

O'Mahony, D. and Gallagher, P. F. (2008). Inappropriate prescribing in the older population: need for new criteria. Age Ageing 37 (2), 138-141

Ogu, C. C. and Maxa, J. L. (2000). Drug interactions due to cytochrome P450. Proc (Bayl Univ Med Cent) 13 (4), 421-423

Oliveira, M. G., Amorim, W. W., de Jesus, S. R., Rodrigues, V. A. and Passos, L. C. (2012). Factors associated with potentially inappropriate medication use by the elderly in the Brazilian primary care setting. Int J Clin Pharm 34 (4), 626-632

Opie , L. H. (2012). Drug interactions of antihypertensive drugs. S Afr Fam Pract 54 (2, Supp. 1), 23-25

Page, R. L., Linnebur, S. A., Bryant, L. L. and Ruscin, J. M. (2010). Inappropriate prescribing in the hospitalized elderly patient: defining the problem, evaluation tools, and possible solutions. Clin Interv Aging 5, 75-87

Pavlicevic, I., Glavaski, M., Rumboldt, M. and Rumboldt, Z. (2011). Prohypertensive effects of non-steroidal anti-inflammatory drugs are mostly due to vasoconstriction. Coll Antropol 35 (3), 817-822

Perneger, T. V. (2005). The Swiss cheese model of safety incidents: are there holes in the metaphor? BMC Health Serv Res 5, 71

Peterson (2011). Drug interaction dilemmas. Journal of Pharmacy Practice and Research 41 (1), 7-12

Pharmaceutical Care Network Europe Foundation (2010). (Zugriff vom 05.08.2013). PCNE Classification scheme for Drug-Related Problems (14-01-2010vm) V6.2. www.pcne.org/.../PCNE%20classification%20V6-2....

Pitt, B., Zannad, F., Remme, W. J., Cody, R., Castaigne, A., Perez, A., Palensky, J. and Wittes, J. (1999). The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized Aldactone Evaluation Study Investigators. N Engl J Med 341 (10), 709- 717

Podichetty, V. K., Mazanec, D. J. and Biscup, R. S. (2003). Chronic non-malignant musculoskeletal pain in older adults: clinical issues and opioid intervention. Postgrad Med J 79 (937), 627-633

Projektverband Priscus (2011). (Zugriff vom 12.08.2012). Priscus-Liste. www.priscus.net/download/PRISCUS-Liste_PRISCUS-TP3_2011.pdf

Psarakis, H. M. (2006). Clinical Challenges in Caring for Patients With Diabetes and Cancer. Diabetes Spectrum 19 (3), 157-162

Rambhade, S., Chakarborty, A., Shrivastava, A., Patil, U. K. and Rambhade, A. (2012). A Survey on Polypharmacy and Use of Inappropriate Medications. Toxicol Int 19 (1), 68-73

Ray, W. A., Griffin, M. R. and Downey, W. (1989). Benzodiazepines of long and short elimination half-life and the risk of hip fracture. Jama 262 (23), 3303-3307

Recuenco, I., Espinosa, E., Garcia, B. and Carcas, A. (1995). Effect of dexamethasone on the decrease of serum phenytoin concentrations. Ann Pharmacother 29 (9), 935

131

Remmer, H. (1972). Induction of drug metabolizing enzyme system in the liver. European Journal of Clinical Pharmacology 5 (2), 116-136

Resbeut, M., Marteau, P., Cowen, D., Richaud, P., Bourdin, S., Dubois, J. B., Mere, P. and N'Guyen, T. D. (1997). A randomized double blind placebo controlled multicenter study of mesalazine for the prevention of acute radiation enteritis. Radiother Oncol 44 (1), 59-63

Riechelmann, R. P. (2007). Drug combinations with the potential to interact among cancer patients. Support Care Cancer 15 (9), 1113-1114

Riechelmann, R. P. and Del Giglio, A. (2009). Drug interactions in oncology: how common are they? Ann Oncol 20 (12), 1907-1912

Riechelmann, R. P., Moreira, F., Smaletz, O. and Saad, E. D. (2005). Potential for drug interactions in hospitalized cancer patients. Cancer Chemother Pharmacol 56 (3), 286-290

Riechelmann, R. P., Tannock, I. F., Wang, L., Saad, E. D., Taback, N. A. and Krzyzanowska, M. K. (2007). Potential drug interactions and duplicate prescriptions among cancer patients. J Natl Cancer Inst 99 (8), 592-600

Riechelmann, R. P., Zimmermann, C., Chin, S. N., Wang, L., O'Carroll, A., Zarinehbaf, S. and Krzyzanowska, M. K. (2008). Potential drug interactions in cancer patients receiving supportive care exclusively. J Pain Symptom Manage 35 (5), 535-543

Rivkin, A. (2007). Admissions to a medical intensive care unit related to adverse drug reactions. Am J Health Syst Pharm 64 (17), 1840-1843

Rochon, P. A. (2013). (Zugriff vom 18.07.2013). Drug prescribing for older adults : UpToDate. http://www.uptodate.com/contents/drug-prescribing-for-older-adults

Rochon, P. A. and Gurwitz, J. H. (1997). Optimising drug treatment for elderly people: the prescribing cascade. BMJ 315 (7115), 1096-1099

Ruegg, S. (2002). Dexamethasone/phenytoin interactions: neurooncological concerns. Swiss Med Wkly 132, 425-426

Ruskin, M. J. (2012). (Zugriff vom 01.07.2013). Pharmacodynamics in the elderly : the Merck Manual. http://www.merckmanuals.com/professional/geriatrics/drug_therapy_in_the_elderly/pharmacod ynamics_in_the_elderly.html

Ryan, C., O'Mahony, D., Kennedy, J., Weedle, P., Barry, P., Gallagher, P. and Byrne, S. (2009). Appropriate prescribing in the elderly: an investigation of two screening tools, Beers criteria considering diagnosis and independent of diagnosis and improved prescribing in the elderly tool to identify inappropriate use of medicines in the elderly in primary care in Ireland. J Clin Pharm Ther 34 (4), 369-376

Ryan, C., O'Mahony, D., O'Donovan, D. O., O'Grady, E., Weedle, P., Kennedy, J. and Byrne, S. (2013). A comparison of the application of STOPP/START to patients' drug lists with and without clinical information. Int J Clin Pharm 35 (2), 230-235

Sacks, G. S. (2004). Drug-Nutrient Considerationsin Patients Receiving Parenteral and Enteral Nutrition. Practical Gastroenterology 28 (7), 39-48

Satin, J. R., Linden, W. and Phillips, M. J. (2009). Depression as a predictor of disease progression and mortality in cancer patients: a meta-analysis. Cancer 115 (22), 5349-5361

132

Schlenk, M. (2012). (Zugriff vom 03.09.2012). Sicherheitstüte für Arzneimittel : Apotheke ADHOC. http://www.apotheke-adhoc.de/nachrichten/nachricht-detail/sicherheitstüte-fuer arzneimittel

Schmidt, M., Antonsen, S., Hansen, B., Moller, J., Thordal, C. and Sorensen, H. T. (2010). Mortality following acute medical admission in Denmark: a feasibility study. Clin Epidemiol 2, 195-203

Schubert, I., Kupper-Nybelen, J., Ihle, P. and Thurmann, P. (2013). Prescribing potentially inappropriate medication (PIM) in Germany's elderly as indicated by the PRISCUS list. An analysis based on regional claims data. Pharmacoepidemiol Drug Saf 22 (7), 719-727

Sengelov, L., Kamby, C., Geertsen, P., Andersen, L. J. and von der Maase, H. (2000). Predictive factors of response to cisplatin-based chemotherapy and the relation of response to survival in patients with metastatic urothelial cancer. Cancer Chemother Pharmacol 46 (5), 357- 364

Sepehri, G., Khazaelli, P., Dahooie, F. A., Sepehri, E. and Dehghani, M. R. (2012). Prevalence of potential drug interactions in an Iranian general hospital. Indian J Pharm Sci 74 (1), 75-79

Seymour, R. M. and Routledge, P. A. (1998). Important drug-drug interactions in the elderly. Drugs Aging 12 (6), 485-494

Shasha, D., Cremieux, P. and Harrison, L. (2004). Relationship Between Hemoglobin Levels and Quality of Life During Radiation Therapy Plus Concomitant or Sequential Chemotherapy in Patients With Cancer and Anemia Treated With Epoetin Alfa. J Natl Compr Canc Netw 2 (5), 509-517

Shimp, L. A., Ascione, F. J., Glazer, H. M. and Atwood, B. F. (1985). Potential medication- related problems in noninstitutionalized elderly. Drug Intell Clin Pharm 19 (10), 766-772

Shin, J. G., Park, J. Y., Kim, M. J., Shon, J. H., Yoon, Y. R., Cha, I. J., Lee, S. S., Oh, S. W., Kim, S. W. and Flockhart, D. A. (2002). Inhibitory effects of tricyclic antidepressants (TCAs) on human cytochrome P450 enzymes in vitro: mechanism of drug interaction between TCAs and phenytoin. Drug Metab Dispos 30 (10), 1102-1107

Sica, D. A., Carter, B., Cushman, W. and Hamm, L. (2011). Thiazide and loop diuretics. J Clin Hypertens (Greenwich) 13 (9), 639-643

Sica, D. A. and Gehr, T. W. (2002). Rhabdomyolysis and statin therapy: relevance to the elderly. Am J Geriatr Cardiol 11 (1), 48-55

Siebert, S., Elkeles, B., Hempel, G., Kruse, J. and Smollich, M. (2013). [The PRISCUS list in clinical routine. Practicability and comparison to international PIM lists]. Z Gerontol Geriatr 46 (1), 35-47

Simon, C., Bulut, C., Federspil, P. A., Munter, M. W., Lindel, K., Bergmann, Z., Sertel, S., Leitzbach, S. and Plinkert, P. K. (2011). Assessment of peri- and postoperative complications and Karnofsky-performance status in head and neck cancer patients after radiation or chemoradiation that underwent surgery with regional or free-flap reconstruction for salvage, palliation, or to improve function. Radiat Oncol 6, 109

Singh, N., Weisler, S. L. and Hershman, J. M. (2001). The acute effect of calcium carbonate on the intestinal absorption of levothyroxine. Thyroid 11 (10), 967-971

133

Smith, L., McGowan, L., Moss-Barclay, C., Wheater, J., Knass, D. and Chrystyn, H. (1997). An investigation of hospital generated pharmaceutical care when patients are discharged home from hospital. Br J Clin Pharmacol 44 (2), 163-165

Soen, S. (2011). Guidelines on the management and treatment of glucocorticoid-induced osteoporosis. Nihon Rinsho 69 (7), 1310-1314

Spaans, E., van den Heuvel, M. W., Schnabel, P. G., Peeters, P. A., Chin-Kon-Sung, U. G., Colbers, E. P. and Sitsen, J. M. (2002). Concomitant use of mirtazapine and phenytoin: a drug- drug interaction study in healthy male subjects. Eur J Clin Pharmacol 58 (6), 423-429

STOPP criteria (2011). (Zugriff vom 18.07.2013). STOPP: Screening Tool of Older People’s potentially inappropriate Prescriptions. The following drug prescriptions are potentially inappropriate in persons aged ≥ 65 years of age. www.nature.com/clpt/journal/v89/ n6/extref/clpt2011

Tamblyn, R. M., McLeod, P. J., Abrahamowicz, M. and Laprise, R. (1996). Do too many cooks spoil the broth? Multiple physician involvement in medical management of elderly patients and potentially inappropriate drug combinations. CMAJ 154 (8), 1177-1184

Tamraz, B., Fukushima, H., Wolfe, A. R., Kaspera, R., Totah, R. A., Floyd, J. S., Ma, B., Chu, C., Marciante, K. D., Heckbert, S. R., Psaty, B. M., Kroetz, D. L. and Kwok, P. Y. (2013). OATP1B1-related drug-drug and drug-gene interactions as potential risk factors for cerivastatin-induced rhabdomyolysis. Pharmacogenet Genomics 23 (7), 355-364

Techniker Krankenkasse - Pressestelle (2012). (Zugriff vom 26.06.2012). Jeder 4 Senior bekam 2011 potentiell ungeeignete Medikamente-Techniker unterstützt Ärzte mit neuem Service. www.tk.de/tk/pressemitteilungen/archiv-vorjahr/gesundheit.../447278

Teweleit, S., Kuschel, U., Hippius, M., Goettler, M. and Bornschein, B. (2001). Manifestation and prevention of adverse drug reactions (ADR) in the pharmacotherapy of cardiovascular diseases. Med Klin (Munich) 96, 442-450

Torrey, T. (2010). (Zugriff vom 07.07.2013). Polypharmacy. http://patients.about.com/od/glossary/g/polypharmacy.htm

Triller, D. M., Clause, S. L. and Hamilton, R. A. (2005). Risk of adverse drug events by patient destination after hospital discharge. Am J Health Syst Pharm 62 (18), 1883-1889

Triplitt, C. (2006). Drug Interactions of Medications Commonly Used in Diabetes. Diabetes Spectrum 19 (4), 202-211

Turnheim, K. (2003). When drug therapy gets old: pharmacokinetics and pharmacodynamics in the elderly. Exp Gerontol 38 (8), 843-853 van Leeuwen, R. W., Brundel, D. H., Neef, C., van Gelder, T., Mathijssen, R. H., Burger, D. M. and Jansman, F. G. (2013). Prevalence of potential drug-drug interactions in cancer patients treated with oral anticancer drugs. Br J Cancer 108 (5), 1071-1078

Veehof, L., Stewart, R., Haaijer-Ruskamp, F. and Jong, B. M. (2000). The development of polypharmacy. A longitudinal study. Fam Pract 17 (3), 261-267

Venturini, C. D., Engroff, P., Ely, L. S., Zago, L. F., Schroeter, G., Gomes, I., De Carli, G. A. and Morrone, F. B. (2011). Gender differences, polypharmacy, and potential pharmacological interactions in the elderly. Clinics (Sao Paulo) 66, 1867-1872

134

Vonbach, P. (2007). Drug-Drug Interactions in the Hospital; Thesis at Phil.-Nat. Univ. Basel. [s. n.] Basel

Vonbach, P., Dubied, A., Beer, J. H. and Krahenbuhl, S. (2007). Recognition and management of potential drug-drug interactions in patients on internal medicine wards. Eur J Clin Pharmacol 63 (11), 1075-1083

Vonbach, P., Dubied, A., Krahenbuhl, S. and Beer, J. H. (2008). Evaluation of frequently used drug interaction screening programs. Pharm World Sci 30 (4), 367-374

Vonbach, P., Dubied, A., Krahenbuhl, S. and Beer, J. H. (2008). Prevalence of drug-drug interactions at hospital entry and during hospital stay of patients in internal medicine. Eur J Intern Med 19 (6), 413-420

Wahab, M. S., Nyfort-Hansen, K. and Kowalski, S. R. (2012). Inappropriate prescribing in hospitalised Australian elderly as determined by the STOPP criteria. Int J Clin Pharm 34 (6), 855-862

Wakamoto, H., Kume, A., Nakano, N. and Nagao, H. (2006). Benign angiopathy of the central nervous system associated with phenytoin intoxication. Brain Dev 28 (5), 336-338

Walker, J., Holm Hansen, C., Martin, P., Sawhney, A., Thekkumpurath, P., Beale, C., Symeonides, S., Wall, L., Murray, G. and Sharpe, M. (2013). Prevalence of depression in adults with cancer: a systematic review. Ann Oncol 24 (4), 895-900

Wami, W. M., Buntinx, F., Bartholomeeusen, S., Goderis, G., Mathieu, C. and Aerts, M. (2013). Influence of chronic comorbidity and medication on the efficacy of treatment in patients with diabetes in general practice. Br J Gen Pract 63 (609), e267-273

Wang, H. Y., Chew, G., Kung, C. T., Chung, K. J. and Lee, W. H. (2007). The use of Charlson comorbidity index for patients revisiting the emergency department within 72 hours. Chang Gung Med J 30 (5), 437-444

Wang, M. T., Su, C. Y., Chan, A. L., Lian, P. W., Leu, H. B. and Hsu, Y. J. (2010). Risk of digoxin intoxication in heart failure patients exposed to digoxin-diuretic interactions: a population-based study. Br J Clin Pharmacol 70 (2), 258-267

Wanrooy, A. v. (2011). Arzneimittelanamnese und Dokumentation in der elektronischen Pateintenakte. Pharmazeutische Zeitung / Prisma (1), 53-62

Warde, P., O'Sullivan, B., Bristow, R. G., Panzarella, T., Keane, T. J., Gullane, P. J., Witterick, I. P., Payne, D., Liu, F. F., McLean, M., Waldron, J. and Cummings, B. J. (1998). T1/T2 glottic cancer managed by external beam radiotherapy: the influence of pretreatment hemoglobin on local control. Int J Radiat Oncol Biol Phys 41 (2), 347-353

Wauthier, V., Verbeeck, R. K. and Calderon, P. B. (2007). The effect of ageing on cytochrome p450 enzymes: consequences for drug biotransformation in the elderly. Curr Med Chem 14 (7), 745-757

Weir, M. R. (2012). Renal effects of nonselective NSAID´s and coxibs. Cleve Clin J Med 69 (Suppl. I), 53-58

Wilson, N. S., Hudson, J. Q., Cox, Z., King, T. and Finch, C. K. (2009). Hyperkalemia-induced paralysis. Pharmacotherapy 29 (10), 1270-1272

135

Wysowski, D. K. and Bacsanyi, J. (1996). Cisapride and fatal arrhythmia. N Engl J Med 335 (4), 290-291

Yeoh, E. (2008). Radiotherapy: long-term effects on gastrointestinal function. Curr Opin Support Palliat Care 2 (1), 40-44

Yu, D. K. (1999). The contribution of P-glycoprotein to pharmacokinetic drug-drug interactions. J Clin Pharmacol 39 (12), 1203-1211

Zagermann-Muncke and P. (2009). New classification of interactions: ABDA database as guide in the jungle of interactions [in German]. Pharmazeutische Zeitung 154 (1-2), 26-30

Zaki, S. A. (2011). Adverse drug reaction and causality assessment scales. Lung India 28 (2), 152-153

Zaman Huri, H. and Fun Wee, H. (2013). Drug related problems in type 2 diabetes patients with hypertension: a cross-sectional retrospective study. BMC Endocr Disord 13, 2

Zermansky, A. G., Petty, D. R., Raynor, D. K., Freemantle, N., Vail, A. and Lowe, C. J. (2001). Randomised controlled trial of clinical medication review by a pharmacist of elderly patients receiving repeat prescriptions in general practice. BMJ 323 (7325), 1340-1343

Zermansky, A. G., Petty, D. R., Raynor, D. K., Lowe, C. J., Freemantle, N. and Vail, A. (2002). Clinical medication review by a pharmacist of patients on repeat prescriptions in general practice: a randomised controlled trial. Health Technol Assess 6 (20), 1-86

Zhan, C., Correa-de-Araujo, R., Bierman, A. S., Sangl, J., Miller, M. R., Wickizer, S. W. and Stryer, D. (2005). Suboptimal prescribing in elderly outpatients: potentially harmful drug-drug and drug-disease combinations. J Am Geriatr Soc 53 (2), 262-267

Ziegelmeier, M. and Goller, C. (2013). Interaktionsmanagment: Entwicklung einer standardisierten Qualitätssicherung. Deutsche Apotheker-Zeitung 153 (5), 44

136

Appendix

A ABDA database

A 1.1 ABDA database The ABDA database is an interactive electronic drug interaction program with a filter for severity rating (contraindicated, minimise risk, no special precautions) providing referenced information on the clinical picture caused by a given DDI. As the ORCA system recommended by (Hansten et al., 2001) the ABDA database takes into account the potential severity of adverse drug reactions due to the pDDI the factors known to increase or decrease the risk for an adverse drug reaction (risk factors) and the existing management alternatives to avoid the pDDI or to reduce the risk for an adverse drug reaction by other means (ABDATA-Pharma-Service, 2012). An update of the ABDATA Service occurs biweekly.

The ABDA database contains 434 interaction monographies compiled by ABDA in cooperation with the Department of Science of the German Pharmacologists Association (ABDATA- Pharma-Service, 2012). All information concerning individual interactions are taken from scientific literature. The interaction monographies can be subdivided into easily-accessible information, text and information about the substances. The easily-accessible information contains interacting groups, evaluation of the interactions (hazardous, medium, low, negligible, information from other sources), type of interaction (e.g. pharmacodynamic interaction) and a short description of its effects

A 1.2 Evaluating drug interactions by the ABDA database Under section Pharmacological effect symptoms that can be caused in the patient in case a drug interaction occurs are explicitly stated.

Under Clinical significance the type of interaction (Pharmacodynamic or Pharmakokinetic and the clinical procedure is clarified (monitoring of lab values). Sometimes the time course of the interaction is clarified as necessary for clarity and understanding.

In the section Mechanism the underlying pharmacological mechanism of the drug interaction is exemplified which currently thought to be responsible for the drug interaction.

Under Measures it is recommended how to deal with a potential drug interaction. Lab values are specified which should have been monitored and one the basis of which doses should have been adjusted. It includes details such as dose, route of administration. In some cases time delayed

137

oral intake of medications might be sufficient. In case it is reasonable, alternative drugs and other therapeutic options are recommended.

Under section Comment clinical study results and additional information are listed, which facili- tate the assessment of frequency, degree and clinical impact of the drug interaction. Furthermore risk factors are mentioned which predispose special patient groups to suffer from pDDIs (e.g. children, elderly, parenterally fed patients). The latest publications are cited, which were used for the relevant monographs.

A 1.3 The ABDA database mode of action The section below shows an extract from a print out of the ABDA drug interaction database. The short version is listed below. In the extended long version the mechanism and effects are clearly described in detail with reference to data of the literature. Below there is listed one example to clarify the way the ABDA system works.

Possible drug interaction between two supportive medications:

Interaction: Interaction between Ibuprofen (ibuprofen) and Fortecortin (dexamethasone)

Classification: Minimise risk (Category IV)

Effect: Increased incidences of gastric ulcers- increased bleeding risk

Type: Pharmacodynamic interaction (mechanism probable but still uncertain)

Mechanism: Additive effect

ABDATA-No. 00189

Literature reference cited as follows:

Emmanuel, J.H. et al. Postgrad. Med J. 47, 227-232 (1971)

A 2 Cave module The component disease takes into account patient characteristics which could limit or completely prohibit the use of certain medical products (ABDATA Pharma-Daten-Service, 2012, ABDATA-Pharma-Service, 2012).

“The component age warns of any age-related risk in the use of a medicinal product” (ABDATA-Pharma-Service, 2012). The patient’s date of birth is entered into the module and the module checks whether a drug may be used at a patient’s current age, indicating either no use or limited use.

The component allergy has not been used in this thesis.

138

“In the component SEX, a warning is given if sex related fields of application for a medicinal product are not suitable for the patient’s documented sex” (ABDATA-Pharma-Service, 2012). In case the Cave module detects a drug-disease contraindication, it generates one of two mes- sages indicating severity levels: absolute contraindication (CI), and restriction on use (ROU). Table A-1 shows the categories and how the Cave module is generated.

Table A-1. Classification of the Cave module. Classification Clinical relevance Invalid age range Contraindicated medicati- No application of medica- Consultation with physician on (CI) tion Restriction on use (ROU) Use with caution under Precautions of use need to be consideration of specific observed ( monitoring of lab risk factors values necessary) Diseases and spe- Contraindicated medicati- No application of medica- Consultation with physician cial circumstances on (CI) tion of life Restriction on use (ROU) Use with caution under Precautions of use need to be consideration of specific observed ( monitoring of lab risk factors values necessary)

The Cave module mode of action

The section below shows an extract from the print out of the Cave module:

CAVE DISEASE SHORT VERSION: DISEASE RELATED

Drug trade name: Aquex: 20 mg tablets (Xipamid)

Comment: No treatment in case of hyperuricemia

Disease: Gout

Advice: No treatments in case patient suffers from gout, only apply to patients under close monitoring and supervision in patients with hyperuricemia. The tendency to experience a gout attack is increased.

139

B Study data

B 1 Potential drug interactions Category II-III

Table B-1. Overview of the most frequently detected potential drug interactions of Category II and III. Admission Discharge Both Total Effect Type of inter- Type of Mechanism ABDA No. Evidence for action (I) interaction of the inter- drug interac- (II) action tion Clopidogrel + PPIs 0 0 2 2 1 2 3 1 1162 0 Opiod agonist + opioid partial agonists 0 1 0 1 1 2 3 1 433 0 ACE inhibitors + NSAIDs 0 2 6 8 1 1 3 2 1 5 Aluminium/salts + complex acids 0 5 1 6 7 2 3 9 446 0 Diuretics + NSAIDs 0 2 4 6 1 1 3 2 211 5 NSAIDs + Antihypertensives(Others) 1 2 2 5 1 1 3 2 915 5 Anticoagulants +Thyroid Hormones 0 0 2 2 2 1 3 7 36 0 Antiepileptics + TAIs 0 0 2 2 2 1 3 5 235 0 NSAIDs + Antibiotics 0 1 0 1 10 3 2 15 1054 16 Antibiotics + Cationes 0 1 0 1 1 2 1 4 673 0 Diuretics + Steroids 0 0 2 2 1 1 3 2 45 0 Total 1 14 21 36

140

Category IV

Table B-2. Overview of the most frequently detected potential drug interactions of Category IV. Admission Discharge Both Total Effect Type of inter- Type of Mechanism ABDA No. Evidence for action (I) interaction of the inter- drug interac- (II) action tion Corticosteroids + NSAIDs 1 4 9 14 9 1 3 10 189 7 ACE + Xanthine oxidase inhibitors 0 4 6 10 4 3 2 15 288 4 Beta-symphatomimetics + Beta- 0 1 7 8 1 1 1 2 111 0 blockers Cardiac glycosides + Diuretics 0 0 7 7 2 1 1 10 97 18 Thyroid hormones + Cationes 0 4 2 6 1 2 3 4 564 12 ACE inhibitors + Diuretics 1 0 5 6 6 1 1 8 186 6 Anticoagulants + TAIs 0 2 3 5 9 1 3 19 511 7 SSRI + MCP 0 3 1 4 19 1 3 14 1005 11 Corticosteroids + Antidiabetics 0 1 3 4 3 1 1 18 183 3 NSAIDs + SSRI 0 3 0 3 9 1 3 10 1145 1 NSAIDs + TAIs 1 0 2 3 1 1 2 15 967 0 Anticoagulants + NSAIDs 0 1 2 3 9 3 2 15 960 7 Antibiotics + PPIs 0 1 2 3 1 2 3 4 673 0 Antiarrythmics + Beta-Blockers 0 0 3 3 17 1 3 6 610 0 CSEs + Calcium antagonists 0 0 2 2 2 2 3 6 1302 0 Calcium + Thiazides 0 2 0 2 13 1 3 10 1251 19 Proteinkinase inhibitors + PPIs 0 2 0 2 1 2 3 4 1144 9 NSAIDs +Antihypertensives 0 2 0 2 2 1 3 2 1065 5 Opioids + Azol-antimycotics 0 2 0 2 2 2 3 6 855 0 Benzodiazepines + Antiepileptics 0 0 2 2 2 2 3 6 799 0 CSEs + Azol antimycotics 0 2 0 2 18 1 3 6 725 0 Beta-Blockers + Cholinergics 0 1 1 2 5 1 1 10 713 1 Anticoagulants + CSIs 0 0 2 2 2 2 3 5 488 4 Antidiabetics + Beta-blockers 0 1 1 2 2 1 1 3 393 3 Biphosphonates + Cationes 0 0 2 2 1 2 3 4 357 9 ACE inhibitors + Electrolytes 0 1 1 2 6 1 1 10 225 6 Corticosteroids + Antiepileptics 0 1 1 2 1 2 1 7 84 9 Anticoagulants + NSAIDs 0 1 1 2 16 1 3 16 15 7 Antidepressants + Neuroleptics 0 0 1 1 11 1 1 6 1337 14

141

Admission Discharge Both Total Effect Type of inter- Type of Mechanism ABDA No. Evidence for action (I) interaction of the inter- drug interac- (II) action tion ACE inhibitors + Electrolytes 0 1 0 1 14 1 3 10 1251 19 Estrogen modulators + Antidepressants 0 1 0 1 1 2 3 1 1174 9 Hydantoins + Antidepressants 0 0 1 1 2 2 3 7 1068 0 CSEs + Enzyme inductors 0 1 0 1 1 2 3 7 1010 0 Immunosuppressants + At1-antagonists 0 0 1 1 6 1 3 10 949 6 CSEs + Calcium –antagonist 0 0 1 1 18 2 3 6 860 0 Antiarrhythmics + CSIs 0 0 1 1 18 2 3 6 827 0 Antidepressants + Neuroleptics 0 0 1 1 11 1 1 10 751 14 Antidepressants + H1 Blocker 0 1 0 1 11 1 1 10 749 14 Anticoagulants + TAIs 1 0 0 1 2 2 3 10 560 7 Antiepileptics + Antibiotics 0 0 1 1 1 2 3 7 541 8 At1 blocker + Diuretics (potassium 0 0 1 1 6 1 1 10 478 6 sparing diuretics) Antiepileptics + cytostatics 0 0 1 1 1 2 3 4 431 8 Antidepressants + SSRIs 0 0 1 1 2 1 3 6 425 11 NSAIDs + TAIs 0 1 0 1 12 1 3 10 420 0 Immunosuppressants + CSEs 0 0 1 1 18 2 3 6 411 0 Antidepressants + Estrogen modulators 0 1 0 1 9 2 3 6 408 7 Theophyllin + Gyrase inhibitors 0 0 1 1 2 2 1 6 228 0 Neuroleptics + Dopamin-antagonists 0 1 0 1 2 1 3 10 217 21 Antiarrythmics + Cardiac glycosides 0 0 1 1 2 1 3 10 67 18 Alpha 1 antagonists + Beta-Blockers 0 0 1 1 1 1 3 2 61 0 Anticoagulants + Antiarrythmics 0 0 1 1 2 1 3 16 13 7 Alpha 2 antagonists + Antidepressants 0 0 1 1 20 1 3 2 3 5 Total 4 46 81 131

142

Category V-VI

Table B-3. Overview of the most frequently detected potential drug interactions of Category V and VI. Admission Discharge Both Total Effect Type of Type of Mechanism ABDA No. Evidence for interaction interaction of the inter- drug interac- (I) (II) action tion ACE inhibitors + Diuretics 1 5 13 19 5 1 3 10 232 1 Corticosteroids + Diuretics 2 3 10 15 8 1 3 10 231 13 Beta-blockers + Calcium-antagonists 0 6 5 11 5 1 3 6 453 10 ACE inhibitors + Antidiabetics 1 1 7 9 8 1 3 12 28 3 Beta-2-symphatomimetics + Diuretics 0 0 7 7 8 1 3 10 1278 13 Neuroleptics + Antiparkinson med. 0 2 3 5 1 1 3 2 213 0 Anticoagulants + PPIs 0 1 3 4 2 2 3 6 516 7 Antidiabetics + Diuretics 0 2 2 4 3 3 3 3 164 3 Annticoagulants + Antidepressants 1 3 0 4 2 1 2 10 528 7 Antidiabetics + Thyroid hormones 0 1 2 3 20 2 3 2 176 3 Anticoagulants + TAIs 0 0 3 3 16 1 1 10 931 7 PPIs + Antimycotics 0 3 0 3 2 2 3 16 835 0 Antidiabetics + Beta-2- 0 1 2 3 3 1 1 3 256 3 symphatomimetics Cardiac glycosides + Beta-blockers 0 0 3 3 5 1 1 10 192 10 Diuretics + Laxatives 1 1 0 2 8 1 1 10 486 13 Anticoagulants + Xanthine oxidase 0 0 2 2 16 2 3 17 11 7 inhibitors Antiepileptics + Benzodiazepines 0 0 2 2 1 2 3 15 1076 0 Corticosteroids + Hormones 0 1 1 2 2 2 3 6 399 15 Corticosteroids + Immunosuppressants 0 0 2 2 2 1 3 16 370 0 Calcium-antagonists+ H2 Blocker 0 2 0 2 2 2 3 6 233 0 Theophyllin + Beta-2- 0 0 1 1 2 1 3 10 214 0 symphatomimetics Anticoagulants + Antidiabetics 0 0 1 1 2 1 3 3 155 3 Beta-Blockers+ NSAIDs 1 0 0 1 1 1 3 2 45 5 Benzodiazepines + PPIs 0 0 1 1 2 2 3 16 27 0 Antimycotics + Antidepressants 0 1 0 1 11 1 3 16 1125 0 Immunosuppressants + PPIs 0 0 1 1 2 2 3 16 1085 0 Antidiabetics + Beta-blockers 0 0 1 1 3 1 1 3 989 3

143

Admission Discharge Both Total Effect Type of Type of Mechanism ABDA No. Evidence for interaction interaction of the inter- drug interac- (I) (II) action tion Anticoagulants + Antidepressants 0 1 0 1 2 2 3 15 874 7 Beta-blockers + Antidepressants 0 0 1 1 2 2 3 16 813 10 Antidiabetics + Gyrase inhibitors 1 0 0 1 3 2 2 6 728 9 Antimycotics + Calcium antagonists 0 1 0 1 2 2 3 16 606 5 Antidiabetics + NSAIDs 0 1 0 1 2 2 3 17 331 20 Cardiac glycosides + Corticosteroids 0 0 1 1 2 1 1 1 270 0 Antidiabetics + Hormones 0 0 1 1 1 1 3 3 255 3 Anticoagulants + Corticosteroids 0 0 1 1 2 3 2 15 177 0 H2 Blocker + Magaldrate 0 1 0 1 1 2 1 4 179 Total 8 37 76 121 Admission. Drug interactions which only occurred at hospital admission, Discharge: Drug interactions which only occurred at hospital discharge Both. Means the drug interactions occurred as well at hospital admission as at hospital discharge. Due to the fact that it has been the same patient the drug interactions which occurred at both time points has only been counted once. CSEs = HMG-CoA reductase inhibitors, NSAIDs = non-steroidal anti-inflammatory drugs, TAIs = inhibitors of platelet aggregation, SSRIs = selective serotonin reuptake inhibitors, PPIs= proton pump inhibitors. Effect: 1 = Decreased effect, 2= Increased effect, 3 = Hypoglycemia+Hyperglycemia, 4 = increased risk of immunological reactions, 5 = Hypotension+Bradycardia, 6 = Hyperkalemia, 7 = increased risk of aluminium absorption, 8 = Hypokalemia, 9 = gastric bleeding complications, 10 = seizures, 11 = ventricular tachycardia, 12 = modified reaction in single cases, 13 = Hypoglycemia 14 = increased risk of hypercalcaemia, 15 = Bradycardia and bronchoconstriction, 16 = Increased risk of bleeding, 17 = additive cardiodepressive effect, 18 = increased incidence of myopathies, 19 = serotonin syndrome, 20 = hypertension Type of interaction I: 0 = not available, 1 = pharmacodynamic interaction, 2 = pharmacokinetic interaction, 3 = others Type of interaction II: 0 = not available, 1 = mechanism clarified, 2 = not clarified, 3 = not fully elucidated, 4 = probably completely clarified Mechanism of drug interactions: 0 = not available, 1 = decreased metabolism to active metabolite, 2 = antagonism 3 = influence on blood sugar regulation, 4 = deterioration of absorption, 5 = displacement of plasma protein binding, 6 = impaired metabolism, 7 = acceleration of hepatic metabolism. 8 = additive potassium sparing effect, 9 = increased absorption, 10 = synergism, 11 = hypokalemia, 12 = increased insulin sensitivity, 13 = hypercalcaemia, 14 = additive serotonin effect, 15 = unknown , 16 = impaired oxidative metabolism, 17 = impaired excretion. Evidence for drug interactions: 0 = no symptoms, 1 = dyspnoea, 2 = asthmatic attack, 3 = fluctuations in blood sugar levels, 4 = skin reactions, leucopenia, 5 = fluctuations in blood pressure levels, 6 = hyperkalemia, 7 = bleeding, 8 = decreased plasma levels, 9/10 = Bradycardia +Hypotension, 11 = serotonin syndrome, 12 = signs of hypothyroidism, 13 = hypokalemia, 14 = tachycardia, 15 = corticosteroids effect, 16 = seizures, 17 = deterioration of kidney function, 18 = intoxication with cardiac glycosides,19 = hypercalcemia, 20=lactic acidosis

144

B 2 Potential inappropriate medications Table B-4. Drugs (IIMs) to be avoided with certain diseases/conditions (Metabolic system) evaluated by the Cave module (adopted and modified (Fick et al., 2003, Page et al., 2010)).

Disease or condi- Drugs inappropriate Justification for the unfavorable benefit/risk profile Frequency* Classifi- tion cation Anemia Non-opioid analgesics (Novaminsulfon) Blood dyscrasia 2 CI Diabetes Blood thinning agents (Enoxaparine) Potassium level fluctuations, close monitoring 6 ROU ACE inhibitors (Enalapril, furosemide) Beta-blockers (Metoprolol) Disguising of hypoglycemic episodes 3 CI Corticosteroids (Dexamethasone), antiparkinson med- Fluctuations in blood sugar glucose levels 25 ROU ication (Levodopa), ACE inhibitors, formoterol, ven- lafaxin NSAIDs (Etoricoxib, ibuprofen) Increased risk for cardiac side effects 4 ROU Selenase, losartan Renal function needs to be monitored 2 ROU Hyperuricemia Diuretics (Xipamide, furosemide) History of gout as leading to acute gout attacks 2 CI Kidney failure Cytostatics (Capecitabin), NSAIDs (Etoricoxib), Depending on clearance dose needs to be adopted, Clearance under 30 4 CI Antihypertensives (Moxonidine, xipamide ) ml/min listed medications inappropriate due to deterioration of GFR ACE inhibitors (Ramipril), CSE inhibitors (Simvas- Depending on clearance dose needs to be adopted, Clearance under 30 12 ROU tatin) ml/min listed medications inappropriate due to deterioration of GFR Anti-gout drugs (e.g. Allopurinol), Gastric protecting Decreased dosing necessary 8 ROU agents (MCP, omeprazole, ranitidine) Diuretics (Furosemide, xipamide) Increased risk of hyponatremia and hypercalcemia 6 ROU Beta-blockers (Metoprolol), opioids (oxycodone) Increased plasma levels of listed drugs 2 ROU *Frequency of IIMs implies the potential frequency; no statement can be made regarding the real manifestation of side effects

145

Table B-5. Drugs (IIMs) to be avoided with certain diseases/conditions (Cardiovascular system) evaluated by the Cave module (adopted and modified (Fick et al., 2003, Page et al., 2010)).

Disease or conditi- Drugs inappropriate Justification for the unfavorable benefit/risk profile Frequency* Classifi- on cation Arrythmia TCAs (Imipramine, doxepin, amitriptyline) Concerns due to proarrythmic effects and ability to produce QT interval 1 ROU changes.

Beta-2-symphatomimetics (Formoterol, salbutamol) Tachyarrythmia 4 ROU Carvedilol Increased risk of cardiovascular side effects 1 CI Melperon, mirtazapin,torasemide 4 ROU CAD Novaminsulfon, L-thyroxine Enhancement of CAD symptoms 2 ROU Amitriptyline Enhancement of CAD symptoms 1 CI Ibuprofen Increased risk for thromboembolic events 3 ROU Sutinimib, candesartan Risk for cardiovascular events increased 2 ROU Tizanidin In case of bradycardia increased risk of syncopes and falls 1 ROU Heart Failure Beta-blockers (Atenolol) Negative inotropic effect with the potential to promote fluid retention and 1 CI (NYHA II-IV) exacerbation of heart failure Calcium channel blockers (Nifedipine) Negative inotropic effect with the potential to promote fluid retention and 1 ROU exacerbation of heart failure Hypertension NSAIDs (Ibuprofen, diclofenac) Blood pressure is increasing due to sympathomimetic effects 1 ROU L Thyroxine, Testogel Fluctuations in blood pressure (Hypotension-increased risk for falls). As 4 ROU prerequisite for therapy well programmed blood pressure Dexamethasone, Heparin Fluctuations in blood pressure (Hypotension-increased risk for falls) Or- 5 ROU thostatic dysregulation Formoterol Haemodynamic monitoring, HF might increase 1 ROU Myocardial infarc- Calcium channel blockers (Nifedipine) It has been proposed that these deleterious effects are due to repeated 1 CI tion episodes of hypotension and reflex sympathetic activation PAD Beta-blockers (Timolol, celiprolol, metoprolol) Aggravation of PAD 5 CI Stroke Heparin Bleeding risk increased 5 CI

146

Disease or conditi- Drugs inappropriate Justification for the unfavorable benefit/risk profile Frequency* Classifi- on cation Blood clotting Clopidogrel, NSAR, Aspirin, Heparin Increased potential for bleeding complications due to prolonged clotting 2 CI disorder or antico- time and elevated INR values agulant therapy *Frequency of IIMs implies the potential frequency; no statement can be made regarding the real manifestation of side effects; CAD = Coronary artery disease, TCAs = tricyclic antidepressants, PAD = peripheral artery disease

Table B-6. Drugs (IIMs) to be avoided with certain diseases/conditions (Gastroenterologic system, Pulmonolgy, Neurological system) evaluated by the Cave module (adopted and modified (Fick et al., 2003, Page et al., 2010)).

Disease or conditi- Drugs inappropriate Justification for the unfavorable benefit/risk profile Frequency* Classifi- on cation CID Corticosteroids (Dexamethasone) Increased risk of colon perforation 3 ROU ACE inhibitors (Ramipril) Liver cirrhosis Metformin, spironolactone, carvedilol Fluctuations in blood sugar levels, hyponatremia 3 CI MCP, capecitabin Fluctuations in blood sugar levels, hyponatremia 2 ROU Liver dysfunction Anti-gout medication (Allopurinol) Dose reduction due to hepatic metabolism 1 CI Pantoprazol,Tilidin/Naloxon, metoprolol Dose reduction due to hepatic metabolism 4 ROU Obstipation Xipamide, furosemide Electrolyte imbalance, risk of hyponatremia and hypokalemia 2 CI Morphin, oxycodone, aluminiumhydroxide, amitripyt- Deceleration of gastrointestinal passage 11 ROU line Pancreatitis Enoxaparin Bleeding risk increased 1 CI Morphine Bleeding risk increased 2 ROU Gastric ulcers Heparin (Enoxaparin), anticoagulants (Phenprocoum- Deterioration of existing ulcers- potential to produce gastric bleeding, 3 CI on) renal failure, high blood pressure and heart failure Bisacodyl, levodopa, dexamethasone Increased risk of bleeding complications, close monitoring 3 ROU COPD Benzodiazepines (Lorazepam) Inducing respiratory depression 9 ROU Increased risk of bronchospasms Metformin Inducing respiratory depression 3 ROU

147

Disease or conditi- Drugs inappropriate Justification for the unfavorable benefit/risk profile Frequency* Classifi- on cation Beta-blockers, phenhydan, amitriptyline, ibuprofen Inducing respiratory depression 3 ROU Zoplicon Morphin, fentanyl, oxcodon 3 Prostatic hyper- Spironolactone, amitriptyline Leading to urinary outflow obstruction 2 ROU plasia Cognitive impair- Anticholinergics Concern due to CNS altering effects 2 ROU ment Depression Benzodiazepines (Lorazepam) Exacerbation of depression 2 ROU Parkinson Disease Metoclopramide, antipsychotics Cave anticholinergic effects 2 ROU Syncope or falls Short to intermediate- acting benzodiazepines and May induce ataxia, impaired psychomotoric function, syncope and addi- 2 ROU TCAs (Imipramine, doxepin and amitripytline) tional falls *Frequency of IIMs implies the potential frequency; no statement can be made regarding the real manifestation of side effects GFR = Glomerular filtration rate, CI = Medication is contraindicated, ROU = Restriction of use, TCA = Tricyclic antidepressant, COPD = Chronic obstructive pulmonary disease, CID = Chronic infectious diseases, CAD = Coronary artery disease, PAD = Peripheral artery disease, CID = Chronic inflammatory disease.

148

B 3 Case report

B 3.1 Potential drug interactions with a CYP P450 inducer S.L. an 81 year old patient was admitted to hospital after resection of an urinary bladder carcinoma with cerebral metastasis for palliative radiotherapy of the cerebellum. The radiotherapy was continued and well tolerated by the patient. In the medical history COPD, Diabetes type II, depression, and a nephrectomy 20 years ago were known as underlying diseases. After the oper- ation the patient suffered from a grand mal attack. As admission medication the patient received:

. Levetiraectam (Keppra 100 mg 1-0-1) . Pregabalin (Lyrica 25) (1-1-1-1) to treat neuropathic pain . Oxazepam 10 mg 0-0 1 as sleeping aid . Phenytoin (Phenhydan 200 mg 1-1-1) . Atosil drops (20-0-20) The HbA1c was 7.9 at hospital admission. A consult with a diabetologist was initiated to improve diabetic therapy. Insulin therapy was adopted and improved. Prior to radiation therapy the patient received additional medication such as pantoprazole and dexamethasone. For a reactive depression and anxiety the patient had been prescribed amitriyptyline 0-0-0-1 and lorazepam 0, 5 mg directly prior to radiation. A treatment with 2 antibiotics had been necessary to treat an urinary tract infection (Ciprofloaxcin, cefpodoxim). The phenytoin level was determined to monitor drug therapy. Blood investigations during hospitalization revealed a low phenytoin level. The phentyoin level was below the therapeutically relevant border. The drug with the highest interaction potential had been phenytoin. A drug interaction screening of the admission and discharge medication revealed six relevant potential drug interactions. The potential drug interactions are listed below:

. Steroids (Dexamethasone)-Hydantoins (Phenytoin) (Category IV) . Hydantoins (Phenytoin)- Ciprofloxacin (Ciprofloxacin) (Category IV) . Hydantoins (Phenytoin)-Benzodiazepines (Oxazepam) (Category V) . Hydantoins (Phenytoin)-Benzodiazepines (Oxazepam) (Category V) . Hydantoins (Phenytoin)-Antidepressants (Amitripytline) (Category IV) . Antidepressants(Amitripytline)-Neuroleptics (Promethazine) (Category IV) Potential drug interaction 1: The comedication of dexamethasone and phenytoin lead to a reduced efficacy of both drugs. Phenytoin uniformly decreases the level of dexamethasone, accel- erating its metabolism through induction of hepatic microsomal enzymes (Cytochrome P 450 induction). The enzyme induction can last for several weeks after tapering phenytoin. Both increased and lowered (Recuenco et al., 1995) levels of phenytoin were observed under comedication with dexamethasone. The exact mechanisms for these conflicting phenomena remain to be elucidated, but increased levels are attributed to competition on protein binding, whereas de-

149

creased levels may be caused upon induction of hepatic metabolism (Ruegg, 2002). Dexame- thasone reduced the plasma concentration of phenytoin by an unknown pharmacokinetic mechanism and lead to an acceleration of the hepatic metabolism (ABDA No.0084).

Potential drug interaction 2: The patient received an antibiotic ciprofloxacin which can lead to an accelerated clearance of phenytoin by the kidneys, leading to decreased plasma levels (ABDA No. 00541). In patients with reduced renal function the half-life of ciprofloxacin might be slightly prolonged and a dose adaption is necessary.

Potential drug interaction 3/4.: The treatment of patients with benzodiazepines and phenytoin can lead to different scenarios (Murphy et al., 2009). Reports are inconsistent: benzodiazepines can cause serum phenytoin levels to rise (chlordiazepoxide, clobazam, clonazepam, diazepam), occasionally resulting in toxicity; or fall (clonazepam, diazepam); or remain unaltered (alprazo- lam, clonazepam)(Wakamoto et al., 2006). In addition phenytoin may cause a fall in the serum levels of clobazam, clonazepam, diazepam, midazolam (see above) and oxazepam. Indicators of phenytoin toxicity (blurred vision, nystagmus, ataxia or drowsiness) need to be obeyed (ABDA No. 1076) (ABDATA-Pharma-Service, 2012).

Potential drug interaction 5: For treatment of anxiety the patient received amitriptyline. Antide- pressants can increase the efficacy of phenytoin by inhibiting the oxidative metabolism of phenytoin by cytochrome P 450 (Shin et al., 2002, Spaans et al., 2002). Phenytoin itself, as an enzyme inducer, can induce CYP P 450 and the metabolism of the antidepressants is activated resulting in low plasma levels (ABDA No. 1068). In case of comedication of an antidepressant and phenytoin, the phenytoin levels need to be closely monitored.

Potential drug interaction 6: The comedication of amitriptyline and promethazine increases the risk of ventricular tachycardia (Torsade de pointes) (ABDA No. 00751). Torsade de pointes tachycardia might occur with syncopes and cardiac arrest. In addition anticholinergic side effects like constipation, xerostomia and micturition disorders may occur. The degradation of antidepressants and neuroleptics occurs by means of CYP2D6. The competition of both drug substances for CYP2D6 slows down the metabolism of both substances. This phenomena has primary relevance in patients known as slow metabolises of CYP2D6 leading to a plasma level increase of 10-130 %.

This case reveals that the drug interaction phenomena is quite complex. In this case drug interactions lead to low phenytoin levels, whereas other drug interactions (Phenytoin/amitriptyline) might increase plasma levels leading to toxicity. In this patient the phenytoin drug level is too low. The low phenytoin level might have been caused by one or an interplay of the potential drug interactions listed above.

150

B 3.2 Drug disease interactions – Example 1 L.R. a 56 year old patient was admitted to hospital with signs of dyspnoea, fatigue and deterioration of the physical status. He was under current radiation therapy due to a pulmonal and cerebral metastasized malign melanoma firstly diagnosed in 05/2006. His multiple medical conditions included a type 2 diabetes, coronary artery disease with hypertension, hyperlipidemia, and an alcohol toxic cardiomyopathy. His medical history included atria fibrillation with ICD im- plantation in 2006, anaemia, myocardial infarction, and multiple oncologic admissions for chemotherapeutic treatment.

The diabetes is currently being treated with a premixed preparation of (Actraphane 30/70).

The medical examination revealed a heart rate of 110/min. The ICD implanted 2006 was incon- spicuous. As admission medication the patient received the following medications:

. Metoprolol 100 mg 1-0-1-, . Ramipril 20 mg 1-0-0, . Amiodaron 200 mg 1-0-0-0, . Spironolacton 25 mg 1-0-0-0, . Torasemid 10 mg 1-0-0-0 . Insulin (Actraphane 30/70) . Simvastatin 40 mg 0-0-1 As additional medication venlafaxine was prescribed to treat a reactive depression. During radiation therapy the patient received dexamethasone, pantoprazole and metoclopramide. 8 days after hospital admission a syncopal event occurred. Dexamethasone can increase blood sugar levels and oppose the sugar lowering effect of antidiabetic medications. According to the treating physician the dexamethasone therapy was responsible for the derailment of blood sugar levels, although measured blood sugar levels did not prove a derailed glucose metabolism. Re- cent studies indicated that the use of dexamethasone can even lead to low blood sugar levels. A hypoglycemic or hyperglycaemic event could therefore also have caused the syncopal event. According to the Cave module steroids should only be administered to patients with underlying diabetes under daily blood glucose monitoring. Even venlafaxine can have an influence on blood sugar levels requiring monitoring and adoption of dose and therapy (eHealthMe, 2013). The ACE inhibitor ramipril can lower blood glucose levels as well in the initial phase of treatment. Here, ramipril had been taken since years and therefore this effect is unlikely (Bosch et al., 2006). The patient received 100 mg metoprolol twice daily. According to the Cave module the administration of beta-blockers to diabetic patients is inappropriate because the risk of hypoglycemic conditions is concealed under beta-blocker therapy. Presenting features of a hypoglycemic condition are neurological manifestations including coma, convulsions, transient hem- iparesis and stroke, while reduced consciousness and cognitive dysfunction may cause accidents and injuries. Cardiac events may be precipitated, e.g. arrhythmias, myocardial ischemia and cardiac failure. The patient already faced cardiac events (myocardial infarction 2007) in the

151

past. Therefore he belonged to a high risk group with a high potential to receive cardiac events again. Although the patient received 200 mg metoprolol daily a high heart rate of 110 had been documented during hospitalisation. Repeated measures showed similar results, which needed clarification. The Cave module indicated that the use of venlafaxine an antidepressant of the serotonin-norepinephrine reuptake inhibitor (SNRI) class should only be administered under close supervision to patients with underlying cardiac diseases. Venlafaxine administration can lead to a dose dependant increase in blood pressure and heart rate. In this individual case it is questionable, whether the increase of the heart rate can be traced back to the venlafaxine administration or it is a counter regulation of the body to other stimuli (Radiation induced tachycardia, hypoglycemia). This issue should sensitize for the topic of drug disease interactions and inappropriate prescribing under special circumstances.

B 3.3 Drug disease interaction – Example 2 S.I. a 79 year old patient was admitted to hospital due to acute pain in the LWS region after an LWK 2 fracture induced by an underlying plasmozytom. An adoption of pain therapy initially with a combination of Oxycodone/naloxon did not show any efficacy. Due to increased blood pressure levels 190/90 mmHg a shift was performed to morphine. Under clonidin normal blood pressure values were reached. For additional pain therapy the patient received etoricoxib. The drug screening program revealed 4 potential drug interactions with etoricoxib. The drug combinations with etoricoxib lead in all 4 cases to reduced antihypertensive efficacy. The underlying pharmacological mechanism is a functional antagonism and a blocked oxidative metabolism via CYP2D6 (ABDA No. 1065). The drug substance etoricoxib was continuously prescribed at hospital discharge although hypertension is a risk factor for the administration of etoricoxib. During hospitalisation the patient had a hypertensive episode which had been related to the administration of oxycodone/naloxon. A connection between etoricoxib and hypertension was not drawn. The drug screening in this individual case could have helped to detect a potential cause for the hypertensive episode. The continuous administration of etoricoxib leads to an increased risk for strokes and myocardial infarction.

152

C Potentially inappropriate medications

Calculation of kidney function:

()150 - y ⋅ m ⋅ k ECC = SC With ECC: Estimated creatinin clearance [ml/min] y: Age [years] m: Body weight [kg] k: Factor (woman 0.9, man 1.1) SC: Serum creatinin [mg/100ml]

153

Acknowledgments

First and foremost, I would like to gratefully and sincerely thank Prof. Dr. med. Irenäus Anton Adamietz, Director of the Department of Radiotherapy and Radiooncology at the medical faculty of the Ruhr-Universität Bochum for handing over the coordination of the project and his continuous personal encouragement throughout the project. I am very grateful for all his expert help and advice and valuable suggestions during the entire period of the research.

Also my best thanks go to the team of the Department of Radiooncology at the Ma- rienhospital Herne for their support and help in finding eligible medical histories and administrative help.

Furthermore, I wish to express my gratitude to Tasso Weinhold for supporting me with the drug interaction screening program Pharmatechnik® at all times.

Finally, I would like to thank my parents and my sister. Their support, encouragement, quiet patience and unwavering love were undeniably the bedrock upon which the past years of my life have been built. They were always supporting me and encouraging me with their best wishes.

Curriculum Vitae

Nina Lamberty born on January 18, 1977 in Dusseldorf, Germany

Education 1983 – 1987 Brenschenschule, Primary school, Witten, Germany 1987 – 1996 Städtisches Ruhrgymnasium, Secondary school, Witten, Germany 1996 University entrance qualification 1996 – 2001 Studies of Pharmaceutical science, University of Regensburg, Germany 1999 – 2000 Studies of Pharmaceutical science, ETH Zurich, Switzerland, Erasmus- /Socrates scholarship 2001 – 2002 Studies of Pharmaceutical science and research, University of Florida, Gainesville, USA 2002 Degree in Pharmceutical science 2003 – 2005 Studies of Pharmaceutical Medicine, University of Witten-Herdecke and University of Duisburg-Essen, Germany 2005 Master of Science in Pharmaceutical Medicine 2006 – 2012 Studies of Medicine, Ruhr-Universität Bochum, Germany 2012 Physician Diploma

Practical experience 2001 – 2002 Shands-Hospital, University of Florida, Gainsville, USA, Drug information centre and internship in clinical pharmacy Since 2002 Westfalen-Apotheke, Hattingen, Pharmacist in a public pharmacy 2012 Nepean Hosipital, University of Sydney, Australia, Practical training in Medicine