Research A prospective cohort study of trends in self- poisoning, Newcastle, Australia, 1987–2012: plus ça change, plus c’est la même chose

Despite ntentional poisoning is a major Abstract massive public health problem and gener- Objective: To examine inhospital mortality and morbidity associated with I ally occurs in the context of delib- self-poisoning with different drug classes over an extended period. increases in erate self-harm and drug misuse. Design, setting and participants: A prospective cohort study over 26 years the number of There are 60 International Statistical (1987–2012) with limited follow-up of patients presenting consecutively to a primary and tertiary referral toxicology centre covering Newcastle, Lake antidepressant Classification of Diseases and Related Health Problems, 10th revision (ICD- Macquarie and Port Stephens, Australia. prescriptions, 10) codes for drug-related deaths. In Main outcome measures: Hospital length of stay, types of drugs ingested, neither rates 2009, these codes together accounted intensive care unit (ICU) admission, requirement for ventilation, inhospital deaths and rates of antidepressant drug use in Australia. of self-harm for 6.4% of male and 5.5% of female total years of potential life lost in Results: Over the study period, there were 17 266 admissions of nor severity of Australia.1 Most deaths are in young patients poisoned by 34 342 substances (16 723 drugs available only on people, and drug-related deaths ac- prescription). The median length of stay was 16 hours, 12.2% of patients antidepressant (2101/17 266) were admitted to an ICU, 7.4% (1281/17 266) were ventilated count for a large proportion of lost poisoning and 78 (0.45%) died in hospital. Patient demographics, social and years of life — causing about 25% of psychiatric factors remained stable over the 26-year period, but case 1 increased completed suicides. The estimated fatality decreased (from 0.77% [15/1955] to 0.17% [7/4060]) as did ICU markedly Australian rate of people hospitalised admissions (19.2% [376/1955] to 6.9% [280/4060]), ventilation (13.7% for self-poisoning of 119 per 100 000 [268/1955] to 4.8% [193/4060]) and LOS. The most frequently ingested population per year2 substantially substances were alcohol, benzodiazepines, paracetamol, antidepressants underestimates total numbers as and antipsychotics. There was a substantial fall in some highly toxic drugs many patients are not admitted or (tricyclic antidepressants, barbiturates, conventional antipsychotics and do not present.3 Most poisonings are theophylline), but increases in less toxic selective serotonin reuptake inhibitors, serotonin–noradrenaline reuptake inhibitors and paracetamol. in young adults and are impulsive A greater than sixfold increase in community antidepressant use was or unplanned. Morbidity and mor- accompanied by only minor changes in overall and antidepressant self- tality from poisoning has proved poisoning rates. surprisingly responsive to targeted Conclusion: Over two decades, there were decreases in poisonings by public health interventions to reduce many highly toxic drugs which were associated with substantial reductions the availability of means to poison in morbidity and inhospital deaths. Despite massive increases in the oneself accidently or deliberately.4,5 number of antidepressant prescriptions, neither rates of self-harm nor the Identification of drugs causing dis- proportion of antidepressant poisonings increased markedly. proportionate numbers of poison- Nicholas A Buckley ings, morbidity or deaths is thus a BMed, FRACP, MD1 key aspect of an effective toxicovigi- and antidepressant self-poisoning, 24 hours/day toxicology treatment Ian M Whyte lance system. and examined changes in prescrib- MB BS(Hons), FRACP, service for a population of about 2,3 FACMT We aimed to examine inhospital mor- ing of and self-poisoning with sev- 500 000. From 1992, ambulances di- eral drugs that have previously been Andrew H Dawson bidity and mortality associated with verted all poisoning presentations 4 MB BS, FRCP, FRACP poisoning in the greater Newcastle identified as having higher relative in the lower Hunter to this service. toxicity in overdose: short-acting bar- Geoffrey K Isbister region over 26 years and to broadly HATS currently has direct clinical 2,3 BSc, FACEM, MD determine what factors have (and biturates, dextropropoxyphene, chlo- responsibility for all poisoned adult ral hydrate, dothiepin, thioridazine, have not) changed over this time, patients in all hospitals in the greater 1 University of Sydney, pheniramine, temazepam, amisul- Sydney, NSW. during which there have been sub- Newcastle region and provides a ter- 2 Calvary Mater Newcastle, pride, alprazolam, venlafaxine and tiary referral service to Maitland and stantial changes in medication use. 10-15 Newcastle, NSW. citalopram. the Hunter Valley. HATS routinely 3 University of Newcastle, In particular, the use of psychotropic Newcastle, NSW. drugs has changed and there have records data on patients who present 4 Royal Prince Alfred Hospital, been large increases in antidepres- to hospital (even if the poisoning is Sydney, NSW. Methods sant prescribing in Australia over uncomplicated).16 Previous studies geoff.isbister@ 6,7 on poisoning in Newcastle17,18 have gmail.com the past three decades. Favourable This is a cohort study of patients pre- and unfavourable effects on popula- senting consecutively after self-poi- shown that no patients were treated tion suicide rates have been postu- soning to the Hunter Area Toxicology exclusively in private hospitals or by doi: 10.5694/mja14.01116 lated for antidepressants.8,9 Thus we Service (HATS) between January their family doctor, indicating that assessed the effect of the increase in 1987 and December 2012. Since 1987, most presentations to medical care Online first 27/04/15 antidepressant prescribing on total HATS has provided a comprehensive facilities are recorded. This cohort

438 MJA 202 (8) · 4 May 2015 Research

does not comprehensively cover increased over the first 8 years, but 1 Morbidity and mortality due to self- unintentional childhood (age < 14 since 1995 has been quite stable poisoning, 1987–2012, measured by need years) poisonings. The local human (HATS became well established in for intensive care and ventilation, length research ethics committee has previ- 1994) (Appendix 1). of stay and fatality* ously granted an exemption regard- Of the total admissions, 15 327 (88.8%) Admitted to intensive care unit ing use of the database and patient were attempts at self-harm and the Ventilated information for research. remainder were a mixture of unin- A structured data collection form tentional, iatrogenic and recreational 20% is used by HATS to prospectively self-poisonings. (Data are generally 15% capture information on patient presented for admissions, and may demographics (age, sex, postcode), thus include the same patient with 10% drugs ingested (including doses), different poisonings.) The median co-ingested substances, regular age of admitted patients was 32 5% medications and management and years (range, 14–97 years) and the complications of poisoning.19 At dis- female : male ratio was 1.6 : 1 (10 514

Proportion of admissions 0 charge, further information is col- female, 6711 male and 39 transgender lected (eg, hospital length of stay patient admissions). The median LOS 20 [LOS], psychiatric and substance was 16 hours (interquartile range, misuse diagnoses). Data are rou- 9.3–25.7 hours; total time spent in 15 tinely entered into a fully relational hospital, 15 688 hours). Of the total Microsoft Access database separate admissions, 2101 involved admission 10 (hours) to the hospital’s main medical record of the patient to an intensive care unit Length of stay у 5 system. Data on all patients aged 14 (ICU) (12.2%) and 1281 involved ven- Median length of stay years who presented following self- tilation of the patient (7.4%). There 0 poisoning were analysed. were 78 inpatient deaths (0.45% of admissions). 1.0% Analyses of population-referenced Fatality data (ie, rates) were restricted to We investigated the changes in mor- 0.8% postcodes that predominantly cover bidity and mortality over the study Newcastle, Lake Macquarie and Port period by dividing admissions into 0.6% Stephens. Changes in total self-poi- those in the first 6 years, reported 0.4% soning rates in the four statistical sub- previously,20 and four subsequent divisions in this area were examined 5-year periods (Box 1). Over this died in hospital 0.2% between 1991 and 2011. Changes in period, the rate of admission to ICU 0 rates of self-poisoning using the main dropped from 19.2% (376/1955) to Proportion of patients who antidepressant drug classes (tricy- 6.9% (280/4060) and rate of mechani- clic antidepressants [TCAs], selective cal ventilation from 13.7% (268/1955) 1987–19921993–1997 serotonin reuptake inhibitors [SSRIs], to 4.8% (193/4060). The fatality rate 1998–20022003–20072008–2012 serotonin–noradrenaline reuptake dropped from 0.77% (15/1955) to inhibitors [SNRIs], monoamine oxi- 0.17% (7/4060). The median LOS * Data were divided into five periods with roughly equal patient numbers. dase inhibitors [MAOIs] and other) decreased from 20.5 hours in the  were also examined. Data on rates of first 6 years to about 16 hours in all antidepressant drug use in these drug subsequent 5-year periods. 7.7%), d ivorce d (1302, 7.5%), w idowe d classes (standardised by the defined In the 17 266 admissions, a previous (409, 2.4%), in de facto relationships daily dose [DDD]) in Australia history of psychiatric illness (9692, (79, 0.5%), other (13, 0.1%); and data from 1991 to 2011 were taken from 56.1%), previous admission for a psy- were missing for the remainder (467, Australian government publications. chiatric episode (6426, 37.2%), previ- 2.7%). These demographic, social and We have previously shown that these ous suicide attempt (9665, 56.0%) psychiatric factors remained stable, data agreed within two significant and history of alcohol or drug mis- except for a slight rise in the propor- digits with Newcastle-specific data use (8466, 49.0%) were commonly tion of patients reporting previous for a range of medications.11 recorded. Few admitted patients self-harm (Appendix 2). were in full-time paid work (2421, Including co-ingested alcohol, 34 342 Results 14.0%); some were unemployed (3622, substances were involved (mean, 1.99 21.0%), pensioners or retired (4163, per patient; range, 1–18 per patient). Over the study period, there were 24.1%), students (1058, 6.1%), doing The major groups of agents involved 17 266 admissions of patients who had home duties (920, 5.3%) or other in self-poisonings and ICU admis- self-poisoned and 11 049 individual (703, 4.1%); data were missing for the sions are shown in Appendix 3 (drugs patients; the median number of ad- remainder (4379, 25.4%). Only 23.8% usually available on prescription) and missions per patient was one (range, (4104) were married; others were Appendix 4 (non-prescription drugs 1–115). The number of admissions single (9567, 55.4%), separated (1325, and other substances). The most

MJA 202 (8) · 4 May 2015 439 Research

2 Use of sedatives and psychotropic drug classes in self-poisoning, 5). This is only partly explained by 1987–2012 falling prescriptions for these agents (Appendix 6). 50% Tricyclic antidepressants Monoamine oxidase There was a more than sixfold inhibitors increase in antidepressant DDDs per 40% Selective serotonin reuptake inhibitors 1000 people per day between 1991 Serotonin–noradrenaline and 2010 (from 12 to 77). However, reuptake inhibitors the increase in the proportion of poi- 30% Other antidepressants sonings due to antidepressants was Other hypnotics very modest (about 1.34-fold). There Benzodiazepines 20% were no corresponding changes in Barbiturates the population rates of self-poisoning Atypical antipsychotics

Proportion of ingested agents Conventional antipsychotics (Box 4, Appendix 6), which fluctu- 10% Anticholinergics ated around the long-term mean in Anticonvulsants each district. There was thus a large Amphetamines decrease in the rate of self-poisoning 0 per DDD prescribed per 1000 popu- 07 0 2002 lation per day for antidepressants –1997 –

1987–1992 1993 (Appendix 7). In Box 4, the increase 1998 2003–2 2008–2012 in total antidepressant prescriptions is illustrated by the total shaded area and changes in individual classes are illustrated by the coloured shading. 3 Use of alcohol, analgesics and selected other drug classes in self- poisoning, 1987–2012 50% Discussion Antibiotics Cardiac drugs The rates of self-poisonings in 40% Newcastle were stable over the past Antihistamines two decades, and the features of the Opioids population presenting with self-poi- 30% Non-steroidal anti- inflammatory drugs soning were constant. This suggests a long-term ongoing and reasonably Salicylates 20% predictable need for clinical toxicol- Paracetamol ogy treatment and ancillary psychi- Alcohol atric and drug and alcohol support Proportion of ingested agents 10% services. Despite large increases in prescriptions for drugs used to treat psychiatric illness (and a range of 0 other major mental health interven- 07 0 1997 2002 tions), there appears to have been no –1992 – – –2 87 19 1993 positive result in terms of reducing 1998 2003 2008–2012 episodes of self-harm. Interestingly, there was a more than commonly ingested substances were drugs and sedatives. Psychotropic sixfold increase in the use of antide- benzodiazepines (5470, 15.9%), alco- drugs consistently accounted for pressants, and while the agents taken hol (5461, 15.9%), paracetamol (4619, about 50% of all drugs ingested but in overdose changed substantially, there were only small changes in 13.5%), antidepressants (4477, 13.0%), newer antidepressants and atypical rates of antidepressant overdoses. antipsychotics (3180, 9.3%), anticon- antipsychotics have largely replaced Interpreting this surprising finding is vulsants (1514, 4.4%), opioids (1232, the older drugs (TCAs and conven- not straightforward. It probably indi- 3.6%), non-steroidal anti-inflamma- tional antipsychotics). Several of the cates that antidepressants are increas- tory drugs (1104, 3.2%) and antihista- drug classes for which frequency of mines (743, 2.2%). Prescription items ingly being prescribed for patients ingestion declined (eg, barbiturates, who have minimal risk of self-harm. accounted for 18 950 agents (55.2%), of theophylline, TCAs) have dispropor- Reassuringly, there is no evidence in which 14 445 (76.2%) were known to tionately high toxicity (Appendix 3). our population to support concerns have been prescribed for the patient. In some drug classes, there were about pro-suicidal effects of new anti- There were major changes over time in larger declines in individual drugs depressant prescriptions. The lack the patterns of drugs ingested (Box 2, identified as having greater toxic- of any change in overall self-harm Box 3), especially for psychotropic ity in the mid 1990s10-13 (Appendix rates also suggests that increased

440 MJA 202 (8) · 4 May 2015 Research

antidepressant use for depression is 4 Rates of antidepressant prescribing (1990–2011, shaded areas) and total rates of self- not an effective public health strategy poisoning (1992–2011, solid lines and symbols)* to reduce rates of self-harm. The only Other antidepressants Newcastle strategy to prevent fatal poisoning Monoamine oxidase inhibitors West Lake Macquarie with consistent supporting evidence 300 Serotonin–noradrenaline reuptake inhibitors 300 is restricting the availability of high- Selective serotonin reuptake inhibitors East Lake Macquarie Antidepressant defined daily doses prescribed Tricyclic antidepressants Port Stephens lethality methods.4,5

Identification of high toxicity in over- per 1000 population day dose is a problem that can only be studied after approval for marketing 200 200 is granted. Postmarketing surveil- 000 people per year lance by pharmaceutical compan- ies of toxicity in overdose is not a requirement for drug registration in Australia or any other country. There 100 100 is little incentive for voluntary sur- veillance. Although most companies record case reports of overdoses of their drugs, this does not facilitate Self-poisonings per 100 comparisons. Reporting biases mean 0 0 that such cases may be atypical of the 1990 1995 2000 2005 2010 usual clinical picture. Year

The many new psychiatric medica- * Total numbers of admissions for self-poisoning before 1993 are slightly lower as the Hunter Area Toxicology Service was just tions coming onto the market should being established. Dotted lines indicate 20-year mean for each district.  mandate coordinated collection of timely information on self-poisoning centres in Australia and overseas will the highlighted drugs (barbiturates and suicide. Ideally, this should be be needed to accelerate collection of and chloral hydrate were withdrawn done at three levels. First, a national data on self-poisoning. in 1994–1995, thioridazine was with- coronial register of drug-related drawn in 2009). Further, there has deaths is essential to enable an analy- Most deaths due to poisoning occur been no drop yet in prescriptions of sis of relative mortality (as done in the outside hospital. Any significant or poisoning with drugs identified United Kingdom21). Second, data on decrease in mortality from self- as having greater toxicity in the mid poisonings reported to poison centres poisoning will result from primary 2000s14,15 (data not shown). are essential, particularly for child- or secondary prevention. Efforts at hood poisonings that rarely require decreasing morbidity and mortality Our data show large drops in rates admission and for assessing the effect from self-poisoning should continue of poisoning with some of the more of primary prevention measures. to target drugs that are frequently lethal drugs, such as TCAs and bar- Poison centre data have limitations taken or are lethal in overdose. biturates (Box 2, Box 4). The intro- due to referral bias, lack of uniform- The falls in prescriptions and poison- duction of less toxic antidepressants ity of assessment and lack of clinical ings with several drugs with greater and sedatives dramatically changed 22 information. Third, for these rea- relative toxicity occurred several prescribing, which in turn changed sons, systematic use of clinical data- years after the problems relating to the types of drugs taken in self-poi- bases to record hospital admissions overdoses with these drugs were iden- soning. This trend has presumably for cases of poisoning is needed to tified in the HATS database around also been reflected in changes in measure relative clinical toxicity. 1994–1995 (Appendix 6, Appendix 8). drug-related deaths. However, the The HATS clinical database has For example, large reductions were nature of coding in official death identified disproportionate effects due to withdrawal of drug subsi- statistics means that there are no in overdoses with many drugs. dies (those for dextropropoxyphene published Australian data to sup- Translation of HATS data into clinical and thioridazine were withdrawn port this contention. For example, risk assessment and guidelines has in 2000) and removal of a formula- poisoning by “antiepileptic, seda- occurred, but with lengthy delays. tion that was misused (temazepam tive-hypnotic, antiparkinsonism and For example, the risk of QT prolon- gel-filled capsules were withdrawn psychotropic drugs” are all lumped gation and torsades de pointes with in 2001). Publication of toxicity data together under one code in Australia.1 thioridazine, citalopram and escit- alone had limited (if any) effect in Most fatal poisonings are classified 10,14,23 terms of reducing prescriptions of alopram were detected in the as due to unspecified or multiple the more toxic drugs. No drugs were database 5–10 years after these drugs agents. Some improvement in cod- banned in Australia on the basis of became available. However, to reduce ing of drug-related deaths should HATS data, although manufacturers the time to identify toxicological not be difficult. Much finer detail is did voluntarily withdraw some of problems to 1–2 years, collaborating provided in other ICD-10 codes. For

MJA 202 (8) · 4 May 2015 441 Research

example, the Australian Bureau of measured in most patients, although little impact on rates of self-harm or Statistics records deaths from croc- previous research on specific drugs antidepressant poisoning. Changes odile and rat bites (three and zero, has found the patient history of drugs in antidepressant classes (generally respectively, between 1999 and 2008) ingested to generally be confirmed by from more to less toxic) have had sig- separately from those caused by other an appropriate assay.24 nificant effects on morbidity and mor- animal bites.1 The development of The key strengths of our study are tality from antidepressant poisoning, the National Coronial Information its long duration and the consist- and therefore from all poisonings. We System, launched in 2000, may make ent core data fields. There are few were also able to generate informa- fatal poisoning comparisons pos- similar attempts to gather data lon- tion regarding relative toxicity and sible in future, but only if data are gitudinally on self-poisoning over patient management. However, for accurately and consistently coded. prolonged periods. Many have ret- many rare poisons, gaining sufficient Assessing the impact that clinical tox- rospective identification of cases numbers of patients to generate reli- icology has on direct patient manage- from hospital coding and thus rely able information about management ment and on public health is hindered entirely on the completeness of medi- and prognosis will take decades by the lack of reliable epidemiological cal records.25 Others have not been from one centre. We believe that, and clinical data. collected continuously26 or have in Australia and overseas, there is Our data have inherent limitations. focused on psychiatric factors and a need for a coordinated approach There is likely to be selection bias treatment (rather than drugs ingested to address the toxicity of drugs in against less severe poisonings (the and toxicity)27 or were conducted in overdose. The public health benefits types of cases where patients might developing countries where agents would greatly outweigh the modest not present for medical attention) and ingested differ substantially from costs of enhancing postmarketing rapidly lethal poisonings (cases in those used in Australia.28 surveillance through more wide- which patients die outside hospital). However, the uniqueness of the HATS spread systematic collection of poi- In surveys of self-harm and anecdo- database also highlights a weakness soning and overdose data. tal reports from patients, it has been of our study. As there are no compa- estimated that a significant propor- Acknowledgements: Geoff rey Isbister is supported rable current datasets, it is difficult by an NHMRC Fellowship (1061041). Andrew Dawson tion of people who have self-poisoned to determine the extent to which the is supported by an NHRMC Practitioner Fellowship (5%–15%) don’t present for medical (1059542). Ongoing toxicovigilance studies are Hunter experience represents that care.3 Also, although there is a pro- supported by an NHMRC Program Grant (1055176). of the developed world. Expansion A great many people have contributed to the HATS spective data collection form, retro- database — including past registrars, advanced trainees, of the database could be facilitated spective review of medical records fellows, nursing staff , research assistants and computer by database systems integrated with programmers — and this work would not have been is often required to complement pro- electronic medical records. possible without their input. spectively collected data. Data on the Competing interests: No relevant disclosures. ingested drugs were based on patient We identified interesting and impor- history, including corroborating his- tant patterns relating to drug pre- Provenance: Not commissioned; externally peer tory obtained from ambulance offic- scriptions, epidemiology of overdose reviewed. ers and accompanying people and patients and importance of relative Received 08 Aug 2014, accepted 9 Jan 2015.  from information on drug contain- toxicity. A massive increase in anti- ers. Drug concentrations were not depressant prescriptions has had References are available online at www.mja.com.au.

442 MJA 202 (8) · 4 May 2015 Research

1 Australian Bureau of Statistics. Causes 11 Buckley NA, Whyte IM, Dawson AH, et 21 Hawton K, Bergen H, Simkin S, et al. of death, Australia, 2012. Canberra: al. Correlations between prescriptions Toxicity of antidepressants: rates of ABS, 2014. (ABS Cat. No. 3303.0.) and drugs taken in self-poisoning. suicide relative to prescribing and non- 2 Australian Institute of Health and Implications for prescribers and drug fatal overdose. Br J Psychiatry 2010; Welfare. Suicide and hospitalised regulation. Med J Aust 1995; 162: 196: 354-358. self-harm in Australia: trends and 194-197. 22 Wolkin AF, Martin CA, Law RK, et al. analysis. Injury research and statistics 12 Buckley NA, Dawson AH, Whyte IM, Using poison center data for national series No. 93. Canberra: AIHW, 2014. O’Connell DL. Relative toxicity of public health surveillance for chemical (AIHW Cat. No. INJCAT 169.) http:// benzodiazepines in overdose. BMJ 1995; and poison exposure and associated www.aihw.gov.au/publication- 310: 219-221. illness. Ann Emerg Med 2012; 59: 56-61. detail/?id=60129549729 (accessed 13 Buckley NA, Dawson AH, Whyte IM, 23 van Gorp F, Whyte IM, Isbister Apr 2015). Henry DA. Greater toxicity in overdose GK. Clinical and ECG effects of 3 Martin G, Swannell SV, Hazell PL, et al. of dothiepin than of other tricyclic escitalopram overdose. Ann Emerg Med Self-injury in Australia: a community antidepressants. Lancet 1994; 343: 2009; 54: 404-408. survey. Med J Aust 2010; 193: 506-510. 159-162. 24 von Mach MA, Weber C, Meyer MR, 4 Roberts DM, Karunarathna A, Buckley 14 Isbister GK, Bowe SJ, Dawson A, et al. Comparison of urinary on-site NA, et al. Influence of pesticide Whyte IM. Relative toxicity of selective immunoassay screening and gas regulation on acute poisoning deaths serotonin reuptake inhibitors (SSRIs) chromatography-mass spectrometry in Sri Lanka. Bull World Health Organ in overdose. J Toxicol Clin Toxicol 2004; results of 111 patients with suspected 2003; 81: 789-798. 42: 277-285. poisoning presenting at an emergency 5 Carlsten A, Allebeck P, Brandt L. Are 15 Isbister GK, O’Regan L, Sibbritt D, department. Ther Drug Monit 2007; 29: suicide rates in Sweden associated Whyte IM. Alprazolam is relatively 27-39. with changes in the prescribing of more toxic than other benzodiazepines 25 Rafnsson SB, Oliver JJ, Elton RA, medicines? Acta Psychiatr Scand 1996; in overdose. Br J Clin Pharmacol 2004; Bateman DN. Poisons admissions in 94: 94-100. 58: 88-95. Edinburgh 1981–2001: agent trends and 6 Mant A, Rendle VA, Hall WD, et 16 Whyte IM, Dawson AH, Buckley NA, et predictors of hospital readmissions. al. Making new choices about al. A model for the management of Hum Exp Toxicol 2007; 26: 49-57. antidepressants in Australia: the long self-poisoning. Med J Aust 1997; 167: 26 Lund C, Vallersnes OM, Jacobsen D, view 1975–2002. Med J Aust 2004; 181 142-146. et al. Outpatient treatment of acute (7 Suppl): S21-S24. 17 Hardwicke C, Holt L, James R, Smith AJ. poisonings in Oslo: poisoning pattern, 7 Hollingworth SA, Burgess PM, Trends in self-poisoning with drugs in factors associated with hospitalization, Whiteford HA. Affective and anxiety Newcastle, New South Wales, 1980– and mortality. Scand J Trauma Resusc disorders: prevalence, treatment and 1982. Med J Aust 1986; 144: 453-454. Emerg Med 2012; 20: 1. antidepressant medication use. Aust N 18 Myers JB, Smith AJ, Elliott RL, MacAskill 27 Kapur N, Steeg S, Webb R, et al. Z J Psychiatry 2010; 44: 513-519. P. Self-poisoning with drugs: a 3 1/2- Does clinical management improve 8 Baune B, Hay P. Suicide rates and year study in Newcastle, NSW. Med J outcomes following self-harm? Results antidepressant prescribing: a casual or Aust 1981; 2: 402-405. from the multicentre study of self- harm in England. PLOS One 2013; 8: causal relationship? PLOS Med 2006; 19 Buckley NA, Whyte IM, Dawson AH, 3: e220. Reith DA. Preformatted admission e70434. 9 Healy D. Did regulators fail over charts for poisoning admissions 28 Dawson AH, Eddleston M, Senarathna selective serotonin reuptake inhibitors? facilitate clinical assessment and L, et al. Acute human lethal toxicity of BMJ 2006; 333: 92-95. research. Ann Emerg Med 1999; 34: agricultural pesticides: a prospective cohort study. PLOS Med 2010; 7: 10 Buckley NA, Whyte IM, Dawson AH. 476-482. e100035 7  Cardiotoxicity more common in 20 Buckley NA, Whyte IM, Dawson AH, thioridazine overdose than with other et al. Self-poisoning in Newcastle, neuroleptics. J Toxicol Clin Toxicol 1995; 1987-1992. Med J Aust 1995; 162: 33: 199-204. 190-193.

MJA 202 (8) · 4 May 2015 443