Background. The aim of this study was to estimate the extent to which anxiety disorders contribute to an increase in suicidal behaviour after controlling for both observed and non-observed sources of confounding.

Method. Data were collected from the Christchurch Health and Development Study (CHDS), a 25-year longitudinal study of over 1000 participants. Measures of anxiety disorders [phobia, generalized anxiety disorder (GAD), panic disorder], major depression (MD), substance use disorders, conduct/antisocial personality disorder, stressful life events, unemployment, and suicidal ideation/attempts for subjects aged 16–18, 18–21 and 21–25 years were used to fit random and fixed effects regression models of the associations between anxiety disorders and suicidal behaviours.

Results. Anxiety disorders were strongly associated with suicidal ideation/attempts. Any single anxiety disorder increased the odds of suicidal ideation by 7·96 times [95% confidence interval (CI) 5·69–11·13] and increased the rate of suicide attempts by 5·85 times (95% CI 3·66–9·32). Control for co-occurring mental disorders, non-observed fixed confounding factors and life stress reduced these associations [suicidal ideation odds ratio (OR) 2·80, 95% CI 1·71–4·58; suicide attempts incidence rate ratio (IRR) 1·90, 95% CI 1·07–3·39]. Rates of suicidal behaviour also increased with the number of anxiety disorders. Estimates of the population attributable risk suggested that anxiety disorders accounted for 7–10% of the suicidality in the cohort.

Conclusions. Anxiety disorders may be a risk factor for suicidality, even after controlling for confounding, with risks increasing with multiple anxiety disorders. Management of anxiety disorders may be an important component in strategies to reduce population rates of suicide.