A paper which questions the predictive validity of the C-SSRS (Columbia-Suicide Severity Rating Scale) suicide risk screening tool in emergency departments—asserting that it is “insensitive”—has won the Academy’s coveted Dlin/Fischer Clinical Research Award for 2020.

The Joint Commission (JC) endorses the use of C-SSRS for suicide screening in emergency departments, among other tools. But the JC has also urged all US medical institutions to use a standardized ‘one tool for all’, the Ask Suicide-Screening Questions tool.

The Dlin/Fischer award is presented for the best-scoring oral paper submitted for presentation at the annual meeting.

The research paper, entitled The Columbia Suicide Rating Scale is Insensitive to Suicide Risk After Emergency Department Visit, is the first to describe the C-SSRS screener’s predictive validity for self-harm and suicide.

It says the screener proved insensitive to suicide risk among emergency department patients: only about one in 300 patients who screened positive died by suicide within 90 days, and most suicides occurred among patients who screened negative.

The paper was compiled by Christian Goans, PhD, Veteran’s Affairs Eastern Colorado Heath Care System; and Denver Health colleagues, Molly Middleton, MPH; Ryan Loh, PhD; Alicia Dalton, MS; Karen Ryall, PhD; and Scott Simpson MD, MPH, co-chair of ACLP’s Emergency Psychiatry SIG.

Background

More than 40% of individuals who die by suicide have emergency department (ED) encounters in the year before death. To guide suicide prevention efforts for these patients, the JC requires EDs to assess suicide risk among behavioral health patients using a standardized screening tool.

One recommended tool was the C-SSRS—Clinical Practice Screener. The C-SSRS screener is a three- to seven-item instrument that assesses suicidal ideation and behavior over recent months. “Despite the JC’s endorsement, the screener’s predictive validity has not been described,” says the paper.

So, the researchers aimed to:

• Describe the C-SSRS screener’s ability to predict suicide among ED patients.
• Describe the screener’s ability to predict self-harm among ED patients.
• Examine the screener’s utility when other clinical variables are considered.

The researchers identified patients’ first encounter in the Level I Trauma Center at Denver Health from April 2016 through June 2018, when the C-SSRS screener was administered to all adult patients.

Main findings

Of 117,017 adult ED encounters during the study period, 92,643 (79%) met inclusion criteria. Among patients with screener data, there were 63 (0.068%) suicides within 365 days of index encounter and 11 (0.011%) suicides within 30 days after discharge; the former represents an incidence of 30.6 suicides per 100,000 person-years.

“The screener had low sensitivity and positive predictive ability to predict suicide at all time points after ED discharge,” says the paper. “Most patients (73%) who died by suicide screened negative, and most deaths (67/84, 80%) among patients screening positive were due to causes other than suicide.”

Among 24,347 patients missing C-SSRS screener data, 19 patients died by suicide within one year. Out of 65 patients who died by suicide in one year and had either a negative or missing C-SSRS screener score, only four (6%) received a psychiatric assessment.

A return visit for self-harm occurred within a year for 421 (0.6%) of 68,755 analyzable encounters. The C-SSRS screener was more sensitive to detecting ED return visits for self-harm outcomes compared to suicide.

For predicting death by suicide, the area under the receiver operating characteristic curve (AUC) was greatest at 90 days after the ED visit—suggesting the screener most strongly correlates with outcome in this timeframe—so this outcome was used for multivariate analyses. Bivariate analyses showed no impact of psychiatric assessment or hospitalization on altering outcomes after screening.

The screener remained a significant predictor of suicide in 90 days after controlling for age, sex, homelessness, mental health diagnosis, substance use diagnosis, self-harm at index visit, presence of psychiatric evaluation, and hospital admission. Similarly, for predicting self-harm, the highest AUC was obtained at 30 days after discharge, at which point there was no effect of psychiatric assessment or hospitalization on outcomes and the screener remained correlated with self-harm in multivariate analyses.

Conclusions

“These findings raise questions about the current role of suicide screening and psychiatric services in EDs,” says the paper. “The C-SSRS has been criticized for its focus on suicidal ideation and behavior which constitute but one part of a patient’s risk profile.

“Our finding that most patients who die by suicide deny suicide ideation in their clinical encounter is consistent with smaller prior studies. We are concerned that use of this screener may have misdirected psychiatric intervention: most patients who died by suicide screened negative (or were not screened at all) and then did not receive psychiatric care in the ED.”

It adds: “Psychiatric consultation to the ED must be reimagined, particularly as disparities in access to care drive heavy ED utilization among underserved patients. The current focus on suicidal ideation in suicide risk screening is limited. Inclusion of patients’ prior history, current treatments, and diagnoses may improve screening sensitivity sufficiently to guide intervention. But even the best suicide prevention interventions will be inadequate given that patients with suicide ideation are more likely to die from causes other than suicide.

“Consulting psychiatrists should consider opportunities to reduce mortality among ED patients including through enhancing treatment adherence for somatic illnesses, substance use treatment, and behavioral therapies for risky behaviors.”