Letters

Study is based on unproved assumptions

Since the data on relative mortality due to an overdose of antidepressants do not come from randomised trials they could reflect differences between patients prescribed different antidepressants rather than differences between the drugs themselves. There is evidence to support this interpretation. For example, when Farmer and Pinder derived a similar "toxicity" index using data from Intercontinental Medical Statistics they found that for each antidepressant the index was lower in those aged over 65 than in younger people³; this cannot be reconciled with the suggestion that the index reflects cardiotoxicity in overdose. If the index measures drug toxicity then it should be stable over time; in fact, the index for amitriptyline calculated for the years 1975-84 is 46.5 (95% confidence interval 43.9 to 49.1)⁴ and for the years 1987-92 is 38.9 (35.6 to 42.4).²

If antidepressants do differ in toxicity in overdose it does not follow that routine prescribing of drugs with lower toxicity will reduce the suicide rate. Such a conclusion would be justified only if there was no substitution of method--in other words, if the prescription of a more toxic drug offered an opportunity for suicide that the depressed person would not find elsewhere. The study by Susan S Jick and colleagues substantially undermines this "loaded gun" argument against the tricyclic drugs since its main finding is that prescription of an older tricyclic antidepressant (rather than a new drug) does not constitute a risk factor for suicide.⁵

Contrary to the conclusions of this paper, there is uncertainty about the true difference in toxicity of different antidepressants in overdose and about the effect of substitution of method in a suicide prevention strategy based on prescribing policy. The small number of lives that might potentially be saved by prescribing the selective serotonin reuptake inhibitors must be set against the costs. Our cost effectiveness study showed how expensive such a prescribing policy might be when compared with other public health policies,⁶ even when the calculations were based on assumptions compatible with the data presented by Henry and colleagues. These latest studies do nothing to challenge our conclusion that a widespread move to prescribing specific serotonin reuptake inhibitors as first line treatment for depression cannot be justified on the evidence available.

Consultant Department of Liaison Psychiatry, Leeds General Infirmary, Leeds LS1 3EX

Director NHS Centre for Reviews and Dissemination, University of York YO1 5DD

Research fellow Centre for Health Economics, University of York, York YO1 5DD

Allan House, Trevor Sheldon, Nick Freemantle 

1. Henry J, Alexander C, Sener E. Relative mortality from overdose of antidepressants. BMJ 1995;310:221-4. (28 January.) [Abstract/Free Full Text]
2. Edwards JG. Suicide and antidepressants. BMJ 1995;310:205-6. (28 January.) [Free Full Text]
3. Farmer R, Pinder R. Why do fatal overdose rates vary between antidepressants? Acta Psychiatr Scand 1989;80(suppl 354): 25-35.
4. Cassidy S, Henry J. Fatal toxicity of antidepressant drugs in overdose. BMJ 1987;295:1021-4.
5. Jick SS, Dean A, Jick H. Antidepressants and suicide. BMJ 1995;310:215-8. (28 January.) [Abstract/Free Full Text]
6. Freemantle N, House A, Song F, Mason J, Sheldon T. Selective serotonin reuptake inhibitor prescribing as a strategy for the prevention of suicide. BMJ 1994;309:249-53. [Abstract/Free Full Text]