Abstract
OBJECTIVE: To explore what percentage of suicide decedents (SDs) vs controls were assessed for suicidality at medical appointments in the year before death. PATIENTS AND METHODS: Using the Rochester Epidemiology Project, 66 SDs dying in Olmsted County, Minnesota, between January 1, 2000, and December 31, 2009, were identified and matched with 141 age- and sex-matched controls. Blinded chart review determined how often providers screened and subjects endorsed suicidal ideation (SI). Positive indicators included chart notes recording SI and/or Patient Health Questionnaire-9 scored more than 0 on question 9. RESULTS: We found that only 29 of 66 (43.9%) SDs and 14 of 141 (9.9%) controls had been screened at any point by any means (P < .001). Only 25.8% (17 of 66) of SDs expressed SI, whereas 58.6% of screened SDs (17 of 29) did so, though none at final appointments before death. No control ever expressed SI. While the majority of both cases and controls went unscreened, providers were more likely to screen SDs (P < .001; odds ratio [OR], 9.0; 95% CI, 3.6-22.0), even with controlling for mental health diagnoses (P = .02; OR, 3.6; 95% CI, 1.2-10.6). CONCLUSIONS: With providers screening less than half of SDs at any point in the year before death, and less than 60% of SDs ever endorsing SI, including none at final appointments, the findings of this naturalistic study bring into question both current screening practices and screening effectiveness. Nonetheless, when SDs were screened, they were significantly more likely to endorse SI than were controls, not 1 of whom ever expressed SI. Taken together, these data suggest that patients expressing SI at any point are at elevated risk for eventual suicide.