The finding:
This study examined the utility of the Columbia-Suicide Severity Rating Scale (C-SSRS) screener’s sensitivity and specificity for identifying patients who enter the emergency department for risk of suicide death (the primary outcome) and suicide attempt/intentional self-harm (the secondary outcome). These outcomes were assessed at 30 days, 90 days, 180 days, and 365 days after the index encounter using multiple methods. The C-SSRS was found to be an insensitive marker for suicide death (i.e. there were many suicide deaths after a negative screen) at all time points. Nonetheless, a positive screen on the C-SSRS was found to confer a statistically significant increased risk of suicide death at 90 days and intentional self-harm/suicide attempt at 30 days. Of particular relevance, neither a psychiatric assessment nor hospitalization at the time of the index encounter was correlated with suicide death outcomes, though hospitalization at the index encounter was correlated with a reduction in self-harm.

Strength and weaknesses:
This research starts to fill a glaring need: “real-world” data on the performance of screeners such as the C-SSRS in settings where suicide risk is increased by the nature of the encounter (an emergency department visit). Deaths by suicide were identified through the Colorado Department of Public Health and Environment; while this would of course miss patients who might have moved across state lines, this would presumably miss only a small percentage of the total index encounters and is acknowledged within the paper. Suicide attempts or intentional self-harm was identified through the diagnostic code and obtained through the Colorado Hospital Association allowing for identification of these outcomes beyond a single hospital system. This would potentially miss patients whose emergency department presentations were not properly coded or identified, though this is a normal peril for this type of research. What is particularly astounding is the comprehensiveness of this research in the context of a medical outcome considered to have an incredibly low event rate (suicide death). The statistical analysis is rigorous, with an acknowledgment of the method of binary categorization (positive/negative screen) as opposed to a more intensity tiered categorization (low/moderate/high risk). The binary categorization was utilized after discovering the binary categorization performed as well as an intensity-tiered approach (and the data on the intensity-tiered approach is available in the Data Supplement). The paper additionally acknowledges the method of categorizing a negative screen as a negative response to any of the first three questions as potentially influencing the outcomes (as opposed to a negative screen being a negative response to all of the first three questions), though a re-analysis of data with a more strict interpretation of a negative screen did not alter the findings.

Relevance:
This research is an urgent call to attention on the Joint Commission’s current mandate on screening for suicidal ideation. As discussed in this paper, the C-SSRS screener has not been validated in its ability to appropriately detect patients at increased risk of suicide nor for self-harm. With the outcomes of this study in mind, universal screening then becomes a significant burden on limited resources without clear effectiveness in catching a multitude of individuals who would benefit from help. This paper demands the attention of psychiatrists as well as hospital systems to creatively consider new methods of identifying individuals at risk of suicide or self harm; as discussed in the paper, utilization of machine learning and artificial intelligence will likely be a crucial aspect of improving our detection of patients at risk as our current limited risk assessments leave much to be desired in helping to detect and to diminish suicide risk.

Type of study (EBM guide):
Retrospective cohort