- aDepartment of Clinical Pharmacology and Toxicology, Canberra Hospital, PO Box 11, Woden, ACT 2606, Australia
- bPharmacy Department, Canberra Hospital
- Correspondence to: N A Buckley
- Accepted 4 October 2002
Several studies over the past 15 years have compared the number of fatal poisonings due to antidepressant drugs in the United Kingdom with drug use statistics to derive a fatal toxicity index: deaths per million prescriptions. 1 2 Greater than 10-fold differences in the index have been shown between tricyclic antidepressants and even larger differences between some tricyclics and newer antidepressants. Explanations have focused on preference for noradrenaline or serotonin reuptake blockade, although only weak correlations have been observed2 and the explanation is toxicologically implausible.1 In the late 1990s the use of newer serotoninergic antidepressants increased dramatically. Some data show that venlafaxine in particular may not be as safe in overdose as other serotoninergic drugs, with reports of deaths, arrhythmias, and seizures.3 We aimed to establish the relative frequency with which venlafaxine and other new antidepressants result in fatal poisoning.
Methods and results
We obtained the number of deaths in Scotland, England, and Wales due to acute poisoning by a single drug, with or without co-ingestion of alcohol, from the General Register Office for Scotland and the Office for National Statistics for the years 1993-9. We used the number of prescription items for England, Wales, and Scotland supplied by the respective departments of health for these years as a measure of relative drug use. Use in hospital is not included, but prescribing of antidepressants overwhelmingly occurs in general practice. For each drug we calculated a fatal toxicity index expressed as deaths per million prescriptions. We calculated the lower and upper 95% confidence limits for the index by using exact confidence intervals for the deaths.1
The table lists the drugs in descending order of fatal toxicity index withinBritish National Formulary drug classes. The serotoninergic drug class overall had a much lower index than the tricyclic antidepressants and monoamine oxidase inhibitors, but venlafaxine had a higher index than the individual and combined results of other serotoninergic drugs.
The most striking new observation is that the fatal toxicity index for venlafaxine is higher than those for other serotoninergic antidepressants and similar to those for some less toxic tricyclic antidepressants. This raises the question of whether venlafaxine should continue to be a first line drug in patients with suicidal ideation. Our results also confirm previously reported large differences in fatal toxicity index between other antidepressant drugs. 1 2
This sort of analysis is open to several criticisms.1 Using the fatal toxicity index as a measure of lethality in overdose makes some assumptions, including that mortality data are not influenced by previous literature and that drugs are taken in overdose with similar frequency and in similar amounts. The perceived risk of overdose has the potential to confound by altering several variables. For example, “less toxic” drugs may be preferentially prescribed to patients at higher risk of poisoning and suicide,4 but they are also less likely to be listed as the sole cause of death from overdose.
Toxicity in overdose should be an important consideration in the choice of first line treatment but should be based on data for each individual drug and not on the therapeutic class or on measures such as serotonin or noradrenaline selectivity that do not directly lead to toxicity in overdose. Poisoning with antidepressants accounts for only about 4-7% of all suicides, but the proportion of suicides from antidepressant poisoning in people prescribed antidepressants is much higher.5 Assuming that an average prescription is for one month's treatment, the fatal toxicity index of venlafaxine suggests that it will cause a death from poisoning about every 6000 patient years of use. Clinicians need to consider whether factors in their patients reduce or compensate for this risk before prescribing venlafaxine.
We thank Zoe Uren of the Office for National Statistics; Graham Jackson of the General Register Office for Scotland; Bill Gold of ISD, Primary Care Information Unit, Scotland; Andy Savva of the Statistics Division of the Department of Health, England; and Sandra Hennefer, information officer at Health Solutions, Wales, for supplying the data on which this analysis is based.
Contributors: NB drafted the paper and performed the statistical analyses. Both authors performed data extraction, wrote the paper, and agreed on the final version. NB is the guarantor.
Competing interests None declared.