Developing a Screening Test for Suicide

Our recent article introduces the first computerized adaptive test (CAT) for the assessment of suicidality. The new test, the Computerized Adaptive Test Suicide Scale (CAT-SS), employs a technology that we have developed over the past 15 years of continuous funding from the National Institute of Mental Health. We used multidimensional item response theory to calibrate a large “bank” of symptom items. Next, we used simulated adaptive testing—based on the responses of a large sample of patients and healthy controls—to develop a prototype CAT. The newly developed CAT-SS was then validated against clinician interviews using the Columbia–Suicide Severity Rating Scale. Results revealed that the CAT-SS reproduced the clinician-rated suicidal ideation risk classifications (no risk versus high risk) with near perfect accuracy (sensitivity of 100% and specificity of 95%); active ideation had a sensitivity of 100% and specificity of 92%, and a suicide alert (plan or plan and intent) had a sensitivity of 100% and specificity of 89%. Rather remarkably, the CAT-SS had sensitivity of 58% and specificity of 88% for lifetime suicide attempts, despite the fact that the CAT-SS covers a timeframe for symptoms of the previous two weeks.

What is special about the CAT-SS is that it provides a crosswalk from symptoms of depression and anxiety to suicidality. The CAT-SS was created by first mapping 997 depression, anxiety, and mania symptoms onto 11 suicidality symptoms. We identified approximately 100 depression and anxiety symptoms that were strongly associated with suicidality (none of the mania items were). For example, “feeling that you are at the end of your rope,” loss of interest and pleasure, and hopelessness are all nonsuicidality symptoms that are strongly associated with suicidal ideation and behavior. By including these 100 additional symptoms, the CAT-SS provides the ability to measure suicidality on a continuum, with scores ranging from 0 to 100 and precision of measurement of 5 points on a 100-point scale for individuals taking the test. The net result is that by using an average of 10 adaptively administered symptom-items in approximately 2 minutes, we can measure the severity of suicidality in anyone, even in people who have not yet begun to think about suicide. The paradigm shift of adaptive testing is to fix the precision of measurement but allow the items to vary both in number and content from person to person. This contrasts with traditional mental health measurement, which fixes the items but allows the precision of measurement (which is unknown) to vary across individuals and within individuals who are repeatedly measured over time.

Although the article is now published, work on the CAT-SS continues. Most recently, we have used the CAT-SS to study the severity of suicidality among our nation’s veterans, foster care children in the state of Tennessee, detainees in the Cook County Bond Court, and the entire freshman class at UCLA. Finally, the CAT-SS2 has just been calibrated and adds questions regarding suicidal intent, plan, and behavior to the item bank.

Financial disclosure:Dr Gibbons is the founder of Adaptive Testing Technologies, which distributes the CAT-MH^TM suite of adaptive tests. The terms of this arrangement have been reviewed and approved by the University of Chicago in accordance with its conflict of interest policies. DrKupfer has received speaker/advisory board fees from Servier; is a stock shareholder in Minerva Neuroscience; has equity in Psychiatric Assessments, and HealthRhythms; and has received royalties from the University of Pittsburgh. DrFrank is employed by HealthRhythms, has received grant/research support from NIMH, has received honoraria from Servier, is a stock shareholder in Minerva Neuroscience, and has equity in Psychiatric Assessments and HealthRhythms. DrsBeiser and Boudreaux and MsMoore have no relevant personal financial relationships to report.