Journal of the New Zealand Medical Association, 24-October-2003, Vol 116 No 1184
Antidepressant poisoning deaths in New Zealand for 2001
David Reith, John Fountain, Murray Tilyard and Rebecca McDowell
Antidepressant medications have been associated with poisoning deaths due to the increased risk of suicide in depression and the toxicological profile of the older antidepressant medications, particularly the tricyclic antidepressants (TCAs). Within the group of antidepressant medications there is variability in the relative toxicity of individual agents, most marked in comparing the selective serotonin reuptake inhibitors (SSRIs) with the TCAs.1–3 However, even within these classes of drugs there appears to be variability in relative toxicity.
Despite the introduction of newer antidepressant medications, prescription rates for TCAs have not fallen to the degree that would be expected from the increased use of SSRIs.4 TCAs are relatively cheap and newer agents do not appear to offer any marked advances in efficacy. TCAs may also be effective at lower doses than those commonly used by prescribers.5 TCAs are also being increasingly used for indications other than depression. In this context, the continuing use of those TCAs that have greater toxicity in overdose would result in greater mortality, and therefore action is required by regulatory agencies, purchasers of medicines and individual prescribers to limit the use of those agents with greater relative toxicity.
Information about toxicity in overdose is not generally available when medications are first marketed, and such information is often not collected during post-marketing surveillance, where the emphasis is upon adverse events at normal dosing. However, toxicity in overdose is particularly relevant with psychotropic medications. Identification of medicines with greater toxicity in overdose would enable prescribers to make more informed judgements of the risk–benefit profiles of their potential treatments.
This study aimed to compare the death rate per prescription, tablet and defined daily dose (DDD) for antidepressant medications used in New Zealand in the year 2001, separately for the overall exposure or for the primary agent involved.
Deaths from antidepressant poisonings were identified from the reports of coronial inquiries for New Zealand in 2001. The cases were identified from coronial chemical injury cases collected from the Coronial Services Office (CSO) in Wellington as of 13 January 2003. From previous experience there may be delay of over a year in the reporting of deaths from coroners and it is estimated that 90–95% of the poisoning deaths for 2001 were recorded by this date.6 Toxicology data were obtained from Institute of Environmental Science and Research (ESR) toxicology reports that were present in approximately 95% of the coroners’ files and, where the toxicology report was absent, the substances involved were extracted from the coroner’s report, the pathology report, police statement or the statement of family or friends. All the substances detected were recorded in the chemical injury database with the exceptions of ethanol (where the blood level was less than 20 mg/dl) and lignocaine (a drug commonly given in resuscitation). Intentional deaths (suicides and homicides) were separated from unintentional deaths according to the judgement of the coroner. The analysis included both intentional and accidental deaths. The primary substance involved in the fatality was determined by the coroners’ reports and the ESR reports.
Prescriptions for antidepressant medications were identified from the PharmHouse database from 1 January 2001 to 31 December 2001. The PharmHouse database is a subset of the New Zealand Health Information System database and contains records of all the claims for medicines dispensed within New Zealand. The records include the drug name, formulation and strength, the type of prescriber and the prescriber’s New Zealand Medical Council number. The data were imported into Stata® for data management to enable tabulation of the prescription numbers by drug type.7 Analyses were also performed using tablet or capsule numbers, and defined daily doses (DDD) as the denominators.8 These separate analyses were performed because the number of tablets required to treat a patient for a day varies considerably between medications, and the number of tablets per prescription also varies considerably. Rates and their 95% confidence intervals were calculated using the command ‘cii’ and the Poisson distribution in Stata®.
As of 16 January 2003, there were 200 deaths due to chemical injury recorded for the year 2001 in New Zealand. There were 146 intentional and 54 unintentional deaths, and these occurred in 144 males and 56 females. There were 41 deaths involving antidepressant medications, and for 23 of these the antidepressant medication was the primary agent involved (Table 1). For the remaining deaths the primary agent was considered to be an alternative agent, such as carbon monoxide in three cases, dextropropoxyphene in three cases, ethylene glycol in two cases, and morphine in two cases. Individual antidepressants were involved in death on 57 occasions. On 29 occasions a single antidepressant was involved, on ten occasions two were involved and on two occasions four were involved.
For the corresponding time period there were 1 560 990 prescriptions for antidepressant medication dispensed in New Zealand. This represented a total of 38 472 206 DDDs, or sufficient antidepressants to treat 105 403 patients for one year (around 3% of the population of New Zealand). The most commonly prescribed antidepressant medications in New Zealand were paroxetine, fluoxetine, amitriptyline, citalopram and dothiepin (Table 1).
Table 1. Prescription volumes and poisoning deaths for antidepressants in New Zealand, 2001
P = prescriptions; C = combined; PA = primary agent; SSRI = selective serotonin reuptake inhibitor; TCA = tricyclic antidepressant; MAOI = monoamine oxidase inhibitor
The rate ratio (95% CI) for deaths per prescription for SSRIs versus TCAs was 0.45 (0.25–0.79) indicating a significantly decreased rate with SSRIs. The medications with the highest rates of death per volume of drug dispensed (as measured by deaths per 100 000 prescriptions) were venlafaxine, desipramine, amoxapine, dothiepin, nortriptyline and trimipramine (Table 1). However, the confidence intervals were extremely wide for the venlafaxine estimates, reflecting the low usage of this medication in New Zealand. When measured by numbers of tablets dispensed or by DDD the ranking was similar (Table 2), except that, when measured by numbers of tablets or capsules dispensed, citalopram and trimipramine had a greater rate of death than dothiepin and nortriptyline and, when measured by DDDs dispensed, dothiepin had a greater rate of death per DDD than amoxapine.
Table 2. Deaths per number of tablets/capsules prescribed and deaths per defined daily dose (DDD)
SSRI = selective serotonin reuptake inhibitor; TCA = tricyclic antidepressant; MAOI = monoamine oxidase inhibitor
The present study is an example of an ‘ecological’ study whereby characteristics of an overall population, in this case numbers of deaths in relation to numbers of prescriptions, are compared but the data cannot be related back to an individual. The main problem with this study design is that it is not possible to correct for confounders, such as degree of depression or co-medication, in the statistical analysis. On many occasions more than one antidepressant was ingested and for the multiple-agent deaths it is difficult to attribute causality to an individual agent. The indications for prescribing were unknown and may represent use for indications other than depression. Factors such as the length of time a patient was medicated and their compliance with medication were not available in the database. In addition, the late reporting of deaths could introduce some bias, although there is no indication that any systematic bias in reporting would occur. Such bias would therefore decrease the precision of our estimates of the rate of death for each medicine. The major advantage of the design is that the data are reliable, compared with other data sources, and cover an entire country. Also, when prescription volumes are used as the denominator, an assessment of relative fatal toxicity can be performed.
The main finding of the present study is that SSRIs have a lower rate of death per volume of drug dispensed than TCAs for both single-agent ingestions and multiple-agent ingestions. This is consistent with previous reports in the literature for the United Kingdom,1–3,9,10 but there are limited supporting data from other populations.11 Much of the information from the United Kingdom has been obtained from re-analysis of the same population, often including the same data. The rates of death per volume of drug dispensed calculated in the present study are similar to those calculated in the previous studies1–3,9–11 and, taken together, strengthen the conclusion that SSRIs are less toxic in overdose than TCAs.
The present study did not examine suicidal deaths by poisons other than antidepressants, or by means other than poisoning. It is possible that SSRIs may still be associated with an increased risk of suicide (by means other than SSRI poisoning)12 or of self-harm.13 However, it is also possible that prescribers may be treating patients with higher suicidality with SSRIs in preference to TCAs because of a perception of lesser inherent toxicity in overdose. This would bias cohort studies of suicide and self-harm in antidepressant users against SSRIs.
Differences between the TCAs in rates of death per volume of drug dispensed are well described in the literature with amoxapine, viloxazine, desipramine and dothiepin having greater toxicity than other TCAs.9 The toxicity of individual tricyclic antidepressants appears to relate to their individual cardiotoxicity and potency as GABAA antagonists rather than their relative potencies as noradrenergic or serotoninergic reuptake inhibitors.9 In the context of overdose, TCA toxicity is also related to the potential to induce seizures and arrhythmias rather than noradrenergic or serotoninergic reuptake inhibition.14 Hence, the use of TCAs with lesser toxicity would not be expected to result in reduced efficacy.
The toxic potential of SSRIs relates to their ability to induce the serotonin syndrome and seizures in overdose.15 There is limited evidence that some SSRIs, particularly venlafaxine, have greater toxicity in overdose than others.1,16,17 Unlike the TCAs, this greater potential may relate to greater potency and clinical efficacy.18,19 Clearly, a more detailed analysis is required to determine whether increased toxicity is offset by greater efficacy with the SSRIs.
The ingestion of SSRIs in combination with other drugs may lead to a more severe clinical presentation, and greater risk of death, than ingestion of SSRIs alone.15 The implication being that the examination of rates of death per volume of drug dispensed for multiple ingestion is relevant, in addition to examination of single-agent rates of death per volume of drug dispensed. The difficulty lies in identifying which medication has the greater risk for death. In the absence of large-scale cohort studies, it is not possible to correct for confounders such as suicidality or co-ingestion, and analysis of rates of death per volume of drug dispensed represents the best method for comparing the clinical toxicity of different drugs.
It is possible that choice of antidepressant medication may be influenced by their perceived characteristics.20 Patients more at risk of suicide may be prescribed particular medications in order to achieve a more rapid response. This would result in over-expression of these agents in suicides. The present study, and also previous studies reporting rates of death per volume of drug dispensed, do not correct for severity of depression, or suicidality, hence it is not possible to correct for this bias. However, the results of the studies are consistent, both within and between different countries, and this would suggest greater risk of fatal toxicity with particular medications.
The differences between the antidepressants in rate of death are accentuated when comparisons are made using DDDs and reduced when comparisons are made using numbers of tablets or capsules dispensed. This can be explained by the fact that SSRIs require fewer tablets or capsules to make up a DDD. It is possible that the perception of SSRIs as having lesser toxicity has resulted in a greater number of DDDs being provided per prescription. In addition, their daily dosing has been developed as one rather than several tablets or capsules per day. The implication is that when antidepressants are compared on the basis of equivalent therapeutic potency, the differences in toxicity become even more apparent.
Overall, the death rate from antidepressant poisoning in New Zealand is low but could be further reduced by restricting the availability of the more toxic drugs within classes. It has been previously demonstrated that limiting the availability of medications with greater toxicity may result in a reduction in overall mortality.21,22 Limitation of pack sizes has also contributed to a reduction in ingested dose in self-poisoning.21 The factors influencing the availability of antidepressants should include safety in addition to efficacy, tolerability and economic considerations.23
Author information: David Reith, Senior Lecturer, Department of Paediatrics and Child Health, Dunedin School of Medicine; John Fountain, Medical Toxicologist, New Zealand National Poisons Centre; Murray Tilyard, Elaine Gurr Professor of General Practice, Department of General Practice, Dunedin School of Medicine, University of Otago, Dunedin; Rebecca McDowell, Health Information Analyst, Population and Environmental Health, Institute of Environmental Science & Research, Porirua
Acknowledgements: We thank Justine Broadley of the Best Practice Advisory Centre.
Correspondence: Dr David Reith, Senior Lecturer, Dunedin School of Medicine, 3rd Floor Children’s Pavilion, Dunedin Hospital, Great King Street, Dunedin. Fax: (03) 474 7817; email: email@example.com
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals