Henry · BMJ (Clinical research ed.) 1995 · Retrospective epidemiological database linkage study · n=1606 deaths

Relative mortality from overdose of antidepressants.

Cited 292 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective population database linkage study evaluating drug exposure and mortality outcomes.

PubMed 7866123 · doi:10.1136/bmj.310.6974.221 · record verified 2026-08-28

What was done

This retrospective epidemiological review examined antidepressant fatal toxicity in general practice across England, Scotland, and Wales from 1987 to 1992. Prescription data from national health authorities (excluding most private general practices and hospitals) were linked to mortality records from the Office of Population Censuses and Surveys and the General Register Office in Scotland. Fatal toxicity was measured as deaths per million prescriptions and deaths per defined daily dose.

What was found

Among 1,606 total antidepressant overdose deaths, 81.6% (1,310/1,606) were caused by amitriptyline and dothiepin. The overall rate was 30.1 deaths per million prescriptions. Fatal toxicity per million prescriptions varied widely by class: tricyclics had 34.14 deaths (95% CI 32.47 to 38.86; P < 0.001), monoamine oxidase inhibitors had 13.48 (95% CI 6.93 to 22.19; P < 0.001), atypical antidepressants had 6.19 (95% CI 4.04 to 8.80; P < 0.001), and SSRIs had 2.02 (95% CI 0.64 to 4.17; P < 0.001). Amoxapine, dothiepin, and amitriptyline had significantly higher fatal toxicity than expected, while nine other antidepressants had significantly lower fatal toxicity. Analysis per defined daily dose showed consistent results.

Why it matters

This study provides large-scale comparative evidence that older tricyclic antidepressants carry substantially greater fatal toxicity in overdose than SSRIs and newer agents. It supports prioritizing safer agents in suicide prevention strategies for patients requiring antidepressant therapy.

Limits

The dataset excluded private practice and hospital prescriptions. Because this was an ecological, aggregate-level analysis, it could not adjust for patient-level confounding by indication, suicide intent severity, co-ingested substances, or non-fatal overdose incidence.

Cited by