Development and validation of a risk prediction algorithm for the recurrence of suicidal ideation among general population with low mood

- The first study to predict recurrence of suicidal ideation at national level.
- The first study to predict suicidal ideation among population with low mood.
- The final model only included 6 predictors which made it easy for future application.
- Emptiness was the most relevant factor in predicting the recurrence of suicidal ideation.

BackgroundSuicidal ideation is one of the strongest predictors of recent and future suicide attempt. This study aimed to develop and validate a risk prediction algorithm for the recurrence of suicidal ideation among population with low moodMethods3035 participants from U.S National Epidemiologic Survey on Alcohol and Related Conditions with suicidal ideation at their lowest mood at baseline were included. The Alcohol Use Disorder and Associated Disabilities Interview Schedule, based on the DSM-IV criteria was used. Logistic regression modeling was conducted to derive the algorithm. Discrimination and calibration were assessed in the development and validation cohorts.ResultsIn the development data, the proportion of recurrent suicidal ideation over 3 years was 19.5 (95% CI: 17.7, 21.5). The developed algorithm consisted of 6 predictors: age, feelings of emptiness, sudden mood changes, self-harm history, depressed mood in the past 4 weeks, interference with social activities in the past 4 weeks because of physical health or emotional problems and emptiness was the most important risk factor. The model had good discriminative power (C statistic=0.8273, 95% CI: 0.8027, 0.8520). The C statistic was 0.8091 (95% CI: 0.7786, 0.8395) in the external validation dataset and was 0.8193 (95% CI: 0.8001, 0.8385) in the combined dataset.LimitationsThis study does not apply to people with suicidal ideation who are not depressed.ConclusionsThe developed risk algorithm for predicting the recurrence of suicidal ideation has good discrimination and excellent calibration. Clinicians can use this algorithm to stratify the risk of recurrence in patients and thus improve personalized treatment approaches, make advice and further intensive monitoring.