Predicting suicide attempts with the SAD PERSONS scale: a longitudinal analysis

J Clin Psychiatry. 2012 Jun;73(6):e735-41. doi: 10.4088/JCP.11m07362.


Objective: The SAD PERSONS scale is a widely used risk assessment tool for suicidal behavior despite a paucity of supporting data. The objective of this study was to examine the ability of the scale in predicting suicide attempts.

Method: Participants consisted of consecutive referrals (N=4,019) over 2 years (January 1, 2009 to December 31, 2010) to psychiatric services in the emergency departments of the 2 largest tertiary care hospitals in the province of Manitoba, Canada. SAD PERSONS and Modified SAD PERSONS (MSPS) scale scores were recorded for individuals at their index and all subsequent presentations. The 2 main outcome measures in the study included current suicide attempts (at index presentation) and future suicide attempts (within the next 6 months). The ability of the scales to predict suicide attempts was evaluated with logistic regression, sensitivity and specificity analyses, and receiver operating characteristic curves.

Results: 566 people presented with suicide attempts (14.1% of the sample). Both SAD PERSONS and MSPS showed poor predictive ability for future suicide attempts. Compared to low risk scores, high risk baseline scores had low sensitivity (19.6% and 40.0%, respectively) and low positive predictive value (5.3% and 7.4%, respectively). SAD PERSONS did not predict suicide attempts better than chance (area under the curve =0.572; 95% confidence interval [CI], 0.51-0.64; P value nonsignificant). Stepwise regression identified 5 original scale items that accounted for the greatest proportion of future suicide attempt variance. High risk scores using this model had high sensitivity (93.5%) and were associated with a 5-fold higher likelihood of future suicide attempt presentation (odds ratio =5.58; 95% CI, 2.24-13.86; P<.001).

Conclusion: In their current form, SAD PERSONS and MSPS do not accurately predict future suicide attempts.

Publication types

  • Clinical Trial
  • Multicenter Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adolescent
  • Adult
  • Canada
  • Female
  • Humans
  • Longitudinal Studies
  • Male
  • Middle Aged
  • Predictive Value of Tests*
  • Psychiatric Status Rating Scales / statistics & numerical data*
  • ROC Curve
  • Sensitivity and Specificity
  • Suicide, Attempted / psychology*