The two study groups may not be comparableBMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.6996.56a (Published 01 July 1995) Cite this as: BMJ 1995;311:56
- David Gunnell
- Lecturer Department of Epidemiology and Public Health Medicine, Health Care Evaluation Unit, University of Bristol, Bristol BS8 2PR
EDITOR,--Susan S Jick and colleagues conclude their detailed study of suicide and prescribing of antidepressants by stating that their data are consistent with the hypothesis that those who are determined to commit suicide will do so even if one particular means (drug overdose) is made less available.1 Such a view is also supported by J Guy Edwards in his accompanying editorial.2 The data presented do not support these assertions.
In Jick and colleagues' study overall crude suicide rates were twice as high in the group who took other antidepressants (14.9/10000 person years) as in those taking tricyclic antidepressants (7.4/10000). This suggests that there may be differences between the two groups that were not adjusted for in the analysis (as adjustment for age, sex, and calendar year made little difference to the relative risks quoted). Such differences may include diagnosis and severity of depression, which may in turn affect risk of suicide; the effects of these factors were examined in only a subsample of patients. Unadjusted rates of suicide due to an overdose were no higher in those prescribed tricyclic depressants (2.9/10000 person years) than in those prescribed the four other antidepressants included in the analysis (3.1/10000). Eight of the 104 suicides in those taking tricyclic antidepressants were a result of overdose of antidepressant while none of the 39 suicides in those taking non-tricyclic antidepressants were; the difference does not reach conventional levels of significance (two tailed Fisher's exact P=0.1). Patients taking non-tricyclic antidepressants used other drugs more often for overdose. The incidence of failed suicide attempts is not reported. The authors' inference is that patients taking antidepressants other than tricyclic antidepressants knew that their drugs were less toxic in overdose. These observational data do not support or refute the “method availability hypothesis”; only by more complete adjustment for confounding can this issue in this group of patients be examined.
Edwards also suggests that people who are determined to kill themselves will find a means of doing so. He cites data on time trends in rates of suicide by particular methods; such trends may, however, be subject to different interpretations. Rates of suicide by particular methods differ by age and sex.3 Increases in the use of violent methods and decreases in suicide due to overdose may be associated with the increase in suicide among young males (who more often use violent methods or carbon monoxide poisoning) and the decrease in suicide among females (who more often use overdose).
Evidence suggests that limitations on the ready availability of methods deter some people from committing suicide.4 Changes in domestic gas supply as well as reduced use of barbiturates are thought to have prevented over 6700 suicides in the 1960s in Britain.5 Restricting the availability of methods exploits the ambivalence felt by many would be suicides and buys time during which the suicidal impulse may pass. There is clearly more to preventing suicide than limiting access to means. Social policy measures and recognition and treatment of depression may be equally important. Possible interventions to reduce suicide by decreasing the availability of particular methods have been described; some of these measures include modifications to prescribing practices, and the evidence presented by Jicks and colleagues and Edwards does not refute these suggestions.4 6