To the Editor: Kirsi Suominen, M.D., Ph.D., et al. (1) followed a cohort of 100 self-poisoned Helsinki patients and found out suicides continued to accumulate 37 years after the index attempts. Although it had a long follow-up period, the analysis they carried out was completely based on a clinical study group. We would like to highlight our findings based on a community sample from the United States.

The Baltimore Epidemiologic Catchment Area Study was a community-based longitudinal study starting in 1981. History of suicide attempts was assessed at baseline. Mortality and causes of death were identified by death certificates after deaths were confirmed by searching the National Death Index. Our study analyzed deaths confirmed up to 1998. By the end of 1998, 861 of the 3,481 baseline participants were deceased. Death certificates of 762 deceased participants (88.5%) were collected, and causes of death were identified, with 729 having died of natural causes. Thirty-three had died of nonnatural causes (4.3% among all known causes of death), including 19 from accidents, six from homicides, and two undetermined. The remaining seven died from suicide (0.9% among all known causes of death). The method was firearms in 57% of the suicide deaths, hanging in 29%, and drug overdoses in 14%. Only two suicides were women (29%). The year of suicide did not cluster in any specific years. History of suicide attempts was not associated with overall mortality after sociodemographic characteristics were adjusted.

In our sample, only 1.7% of the participants with a history of attempts eventually completed suicide in the 17-year period. Compared to people without a history of attempts, former attempters had a much higher odds of completing suicide (odds ratio=11.3, p<0.001). In a multiple logistic regression model with adjustment for sociodemographic variables, only history of attempts and gender were associated with later suicide: men were six times more likely to commit suicide than women (odds ratio=5.9, p<0.001). History of suicide attempts was not associated with the likelihood of other nonnatural deaths.

Although our analysis had a lower percent of eventual suicide than the study by Suominen et al. (13%), our findings overall showed a comparable, although less significant, pattern between gender, history of attempts, and eventual suicide. Our sample also included a comparison group (nonattempters) that the study by Suominen et al. did not have. Our less significant result might have come from our shorter follow-up period and other intrinsic population differences. The newest wave of the Baltimore Epidemiologic Catchment Area Study began in 2004 and will offer more information on longitudinal association between suicide attempts and suicide.

The Epidemiological Catchment Area was supported by grant R01-MH-47447. Points of view and opinions in this letter do not necessarily represent the official positions of the United States Government, the Medical and Health Research Association of New York City, Inc., or the National Development Research Institutes, Inc.