Jump to comment:

• Comment on Favril et al.: Overstatement of the evidence for suicide risk assessment
  C. A. Soper, Matthew M. Large
  Published on: 10 October 2022

• Published on: 10 October 2022

  Comment on Favril et al.: Overstatement of the evidence for suicide risk assessment

  • C. A. Soper, Psychologist Private practice, Lisbon, Portugal
  • Other Contributors:
    • Matthew M. Large, Professor, School of Psychiatry

  In a recent article Favril and associates report a systematic review and meta-analysis of risk factors for suicide derived from psychological autopsy studies that compared community samples of suicide decedents to living or deceased controls. 1 They found a range of risk factors that were, in retrospect, strongly statistically associated with suicide, including the presence of mental disorder (Odds Ratio (OR) = 13.1), depression (OR = 11.0), previous psychiatric treatment (OR = 10.1), previous self-harm (OR = 10.1), and previous suicide attempt (OR = 8.5). While acknowledging methodological weaknesses intrinsic to psychological autopsy studies, the authors maintain a position that “Identifying factors associated with suicide can improve risk stratification and help target interventions for high-risk groups” (p. 1). We consider this conclusion to be premature and fear the article will perpetuate a misplaced confidence in these risk factors as a basis for suicide risk assessment and clinical decision-making.

  Three problems deserve attention. First, more methodologically sound longitudinal studies show much weaker prospective associations between risk factors and suicide. For example, in 2017 Franklin and associates published a survey of 50 years of longitudinal research into factors associated with suicidal thoughts and behaviours, including suicide.2 The top five risk factors for suicide in the Franklin meta-analysis were previous psychiatric hospitalisation (OR = 3...

  Show More

  In a recent article Favril and associates report a systematic review and meta-analysis of risk factors for suicide derived from psychological autopsy studies that compared community samples of suicide decedents to living or deceased controls. 1 They found a range of risk factors that were, in retrospect, strongly statistically associated with suicide, including the presence of mental disorder (Odds Ratio (OR) = 13.1), depression (OR = 11.0), previous psychiatric treatment (OR = 10.1), previous self-harm (OR = 10.1), and previous suicide attempt (OR = 8.5). While acknowledging methodological weaknesses intrinsic to psychological autopsy studies, the authors maintain a position that “Identifying factors associated with suicide can improve risk stratification and help target interventions for high-risk groups” (p. 1). We consider this conclusion to be premature and fear the article will perpetuate a misplaced confidence in these risk factors as a basis for suicide risk assessment and clinical decision-making.

  Three problems deserve attention. First, more methodologically sound longitudinal studies show much weaker prospective associations between risk factors and suicide. For example, in 2017 Franklin and associates published a survey of 50 years of longitudinal research into factors associated with suicidal thoughts and behaviours, including suicide.2 The top five risk factors for suicide in the Franklin meta-analysis were previous psychiatric hospitalisation (OR = 3.57), previous suicide attempt (OR = 2.24), suicidal ideation (OR = 2.22), lower socioeconomic status (OR = 2.20) and life events (OR = 2.18). In contrast to the conclusions of Favril and associates, the authors commented, “No broad category or subcategory [of suicide risk factor] predicted far above chance levels” (p. 187).

  Second, even if the higher odds ratios derived by psychological autopsy were taken at face value, they do not justify the authors’ credence in the utility of suicide risk factors in clinical practice. For example, their assessed OR of 11 for depression, when applied to a base rate of suicide in the order of 10 fatalities per 100,000 per annum in the community samples, points to suicide remaining a highly unlikely outcome for depressed people, even in the long term.

  Third, the authors’ construal of suicide as the “result of a cumulation of multiple risk factors” (p. 7) is contradicted by evidence of the marked weakness suicide risk models based on multiple risk factors. 3.4 Belsher and associates concluded that the accuracy of suicide prediction models for “predicting a future event is near 0” 3. Corke and associates found that suicide risk models using many input variables are no less inaccurate than those using as few as two.4 In this light, suicide emerges as a largely stochastic event, opaque to prediction even in principle (Soper, Malo Ocejo, & Large, 2022).

  A widespread and enduring over-confidence in the utility of suicide risk assessment, whether based on risk factors or by other means, may have the effect of magnifying the perceived risks of suicide, an exaggeration that serves neither the patient nor their clinician. In contrast, we have argued that a brighter future awaits mental healthcare if suicide’s rarity and non-predictability are accepted, acknowledged, and empathically communicated to those in distress. 5

  References

  1. Favril, L., Yu, R., Uyar, A., Sharpe, M., & Fazel, S. (2022). Risk factors for suicide in adults: systematic review and meta-analysis of psychological autopsy studies. Evidence Based Mental Health, ebmental-2022-300549. doi:10.1136/ebmental-2022-300549
  2. Franklin, J. C., Ribeiro, J. D., Fox, K. R., Bentley, K. H., Kleiman, E. M., Huang, X., Nock, M. K. (2017). Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin, 143(2), 187-232.
  3. Belsher, B. E., Smolenski, D. J., Pruitt, L. D., Bush, N. E., Beech, E. H., Workman, D. E., Skopp, N. A. (2019). Prediction models for suicide attempts and deaths: A systematic review and simulation. JAMA Psychiatry, 76(6), 642-651
  4. Corke, M., Mullin, K., Angel-Scott, H., Xia, S., & Large, M. M. (2021). Meta-analysis of the strength of exploratory suicide prediction models; from clinicians to computers. BJPsych open, 7(e26), 1-11.
  5. Soper, C. A., Malo Ocejo, P., & Large, M. M. (2022). On the randomness of suicide: An evolutionary, clinical call to transcend suicide risk assessment. In R. Abed & P. St John-Smith (Eds.), Evolutionary Psychiatry: Evolutionary Perspectives on Mental Health (pp. 134-152). Cambridge, UK: Cambridge University Press and Royal College of Psychiatrists.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top