Omission of Suicide Data

To the Editor: Drs. Bostwick and Pankratz reanalyzed data from earlier reports by Guze and Robins (1) and Goodwin and Jamison (2) to compare two sets of statistics regarding “proportionate mortality prevalence” and “case fatality prevalence” (Table 1, p. 1927). They compared the two prevalence estimates among different subgroups: 1) affective disorder outpatients (24.6% and 2.0%, respectively), 2) affective disorder inpatients (20.0% and 4.1%), and 3) suicidal inpatients (31.7% and 6.0%). In conclusion, the authors asserted that case fatality prevalence is a better estimate for the risk of suicide than proportionate mortality prevalence, which was used in the studies by Guze and Robins (1) and Goodwin and Jamison (2). They also concluded that there is a hierarchical layering of suicide risk among affective disorder patients with different treatment histories and that suicidal inpatients are at the highest risk of suicide. We think this study provides great insight and affirmation of the strength of the association between affective disorders and suicide. However, the studies that Drs. Bostwick and Pankratz reexamined were conducted completely on clinical groups. We would like to highlight our findings from a community sample.

The Baltimore Epidemiologic Catchment Area (ECA) study was a community-based longitudinal study based on 13 years of follow-up. The first interviews were conducted in 1981, and the most recent interview wave was conducted between 1993 and 1995. The incidence of mental disorders and various causes of death (including suicide) were estimated. (For more background information about the Baltimore ECA study, refer to the report by Eaton et al.[3]).

We obtained the following results. For all Baltimore ECA study participants, the proportionate mortality prevalence for suicide was 0.9% (95% confidence interval [CI]=0.19–1.61), and the case fatality prevalence was 0.2% (95% CI=0.04–0.36). For the ECA participants with a history of affective disorders at the baseline wave, the proportionate mortality prevalence for suicide was 5.3% (95% CI=0.00–15.61), and the case fatality prevalence was 0.5% (95% CI=0.00–1.53). For the ECA participants with a history of suicide attempts, the proportionate mortality prevalence for suicide was 16.7% (95% CI=0.00–38.73), and the case fatality prevalence was 1.8% (95% CI=0.00–4.31). Although prevalence estimates among the three groups were not significantly different, there was a hierarchical layering similar to that found by Drs. Bostwick and Pankratz; history of suicide attempts had the strongest association with eventual suicide deaths. Estimates made by use of the proportionate mortality prevalence tended to be four to 10 times as high as the estimates made by using case fatality prevalence. Overall, our findings showed a pattern of suicide risk comparable to that found by Drs. Bostwick and Pankratz, even though our community-based prevalence estimates were much lower and our study included patients with less severe illness than those in the clinical groups.