No Value of Standardized Rating Scales for Suicide Risk
reviewing Quinlivan L et al. Br J Psychiatry 2017 Mar 16. Carter G et al. Br J Psychiatry 2017 Mar 16.
New reports show that rating scales do not effectively predict risk for repeat self-harm or suicide attempts.
Recent research suggests that standardized rating scales are superior to less quantifiable clinician judgment for assessing outcomes over time in patients with depression and anxiety. Now, two reports show that rating scales do not improve assessment of suicide.
Quinlivan and colleagues used six different scales as well as clinician assessment to evaluate the risk for repeat self-harm within 6 months of such an episode in 464 adults. Clinician assessments measured risk in a qualitative manner by cataloguing known risk factors. Repeat self-harm at 6 months occurred in 30% of patients. Positive predictive values (PPVs) for the various scales were poor to modest at best (range, 13%–47%). No scale outperformed global clinician assessment, and many performed worse. Carter and colleagues performed a systematic review and meta-analysis of 70 studies evaluating standardized risk assessment in a range of patients with self-harm or suicide attempt histories. The pooled PPV for all predictive instruments was highest for a combined outcome of self-harm plus suicide at 35.9%. The authors did not evaluate global clinician assessments.
COMMENT
These sobering results are important because accurate risk assessment could guide deployment of therapeutic resources to those at highest risk. Instead, the results suggest that, largely because of the low prevalence of self-harm and suicide, any measure is of limited value (the large majority of patients with positive scores were not at highest risk). Nonetheless, clinicians should continue to enumerate specific risk factors, to target their interventions toward ameliorating specific symptoms and disorders or removing weapons and other means of suicide, and to try to enhance protective factors (e.g., improving social support).
reviewing Quinlivan L et al. Br J Psychiatry 2017 Mar 16. Carter G et al. Br J Psychiatry 2017 Mar 16.
New reports show that rating scales do not effectively predict risk for repeat self-harm or suicide attempts.
Recent research suggests that standardized rating scales are superior to less quantifiable clinician judgment for assessing outcomes over time in patients with depression and anxiety. Now, two reports show that rating scales do not improve assessment of suicide.
Quinlivan and colleagues used six different scales as well as clinician assessment to evaluate the risk for repeat self-harm within 6 months of such an episode in 464 adults. Clinician assessments measured risk in a qualitative manner by cataloguing known risk factors. Repeat self-harm at 6 months occurred in 30% of patients. Positive predictive values (PPVs) for the various scales were poor to modest at best (range, 13%–47%). No scale outperformed global clinician assessment, and many performed worse. Carter and colleagues performed a systematic review and meta-analysis of 70 studies evaluating standardized risk assessment in a range of patients with self-harm or suicide attempt histories. The pooled PPV for all predictive instruments was highest for a combined outcome of self-harm plus suicide at 35.9%. The authors did not evaluate global clinician assessments.
COMMENT
These sobering results are important because accurate risk assessment could guide deployment of therapeutic resources to those at highest risk. Instead, the results suggest that, largely because of the low prevalence of self-harm and suicide, any measure is of limited value (the large majority of patients with positive scores were not at highest risk). Nonetheless, clinicians should continue to enumerate specific risk factors, to target their interventions toward ameliorating specific symptoms and disorders or removing weapons and other means of suicide, and to try to enhance protective factors (e.g., improving social support).
I don't have an anger problem, I have an idiot problem.
