Suicide Attempt as a Risk Factor for Completed Suicide: Even More Lethal Than We Knew

Abstract

Objective: While suicide attempt history is considered to robustly predict completed suicide, previous studies have limited generalizability because of using convenience samples of specific methods/treatment settings, disregarding previous attempts, or overlooking first-attempt deaths. Eliminating these biases should more accurately estimate suicide prevalence in attempters.

Method: This observational retrospective-prospective cohort study using the Rochester Epidemiology Project identified 1,490 (males, N=555; females, N=935) Olmsted County residents making index suicide attempts (first lifetime attempts reaching medical attention) between January 1, 1986, and December 31, 2007. The National Death Index identified suicides between enrollment and December 31, 2010 (follow-up 3-25 years). Medical records were queried for sex, age, method, and follow-up care for index attempt survivors. Coroner records yielded data on index attempt deaths.

Results: During the study period, 81/1,490 enrollees (5.4%) died by suicide. Of the 81, 48 (59.3%) perished on index attempt; 27 of the surviving 33 index attempt survivors (81.8%) killed themselves within a year. Males were disproportionately represented: 62/81 (11.2% of men, 76.5% of suicides) compared with 19/81 (2.0% of women, 23.5% of suicides). Of dead index attempters, 72.9% used guns, yielding an odds ratio for gunshot death, compared with all other methods, of 140 (95% CI=60-325). When adjusted for covariates, survivors given follow-up psychiatric appointments had significantly lower likelihood of subsequent suicide (odds ratio=0.212, 95% CI=0.089-0.507).

Conclusions: At 5.4%, completed suicide prevalence in this community cohort of suicide attempters was almost 59% higher than previously reported. An innovative aspect of this study explains the discrepancy: by including index attempt deaths-approximately 60% of total suicides-suicide prevalence more than doubled. We contend that counting both index and subsequent attempt deaths more accurately reflects prevalence. Our findings support suicide attempt as an even more lethal risk factor for completed suicide than previously thought. Research should focus on identifying risk factors for populations vulnerable to making first attempts and target risk reduction in those groups.

Figure 1

Steps taken to identify confirmed index suicide events among Olmsted County residents from 1986–2010.

Figure 2

Kaplan-Meier curve corresponding to survival probability from completed suicide attempt either at initial suicide attempt or after an index suicide attempt, separated by sex.