Applying Computer Adaptive Testing Methods to Suicide Risk Screening in the Emergency Department

UMMS Affiliation

Department of Emergency Medicine; Department of Psychiatry; Department of Quantitative Health Sciences

Publication Date


Document Type



Behavior and Behavior Mechanisms | Emergency Medicine | Health Information Technology | Health Services Administration | Mental and Social Health | Mental Disorders | Psychiatry and Psychology


OBJECTIVE: Combine test theory with technology to develop brief, reliable suicide risk measures in the emergency department.

METHODS: A computer adaptive test for suicide risk was built using the Beck Scale for Suicide Ideation and tested among the emergency department population. Data were analyzed from a sample of 1,350 patients in several Massachusetts emergency departments. The test was built as outlined by the National Institutes of Health Patient-Reported Outcomes Measurement Information System.

RESULTS: Of 1,350 patients, 74 (5%) scored above the cutoff of BSS > 2. Item 2, "Wish to die", was the most informative item. When using only Item 2, 20% (n = 15/74) of at-risk patients and 3% (n = 40/1,276) of not-at-risk patients were misclassified. Patients were classified after four items with computer adaptive testing trait estimates highly comparable to those of the full scale. The precision rule model did not reduce the scale.

CONCLUSIONS: This study models the creation of a computer adaptive test for suicide ideation and marks the start of the development of computer adaptive tests as a novel suicide risk screening tool in the emergency department. Computer adaptive tests hold promise for revolutionizing behavioral health screening by addressing barriers including time and knowledge deficits.

DOI of Published Version



Suicide Life Threat Behav. 2018 Aug 6. doi: 10.1111/sltb.12493. [Epub ahead of print] Link to article on publisher's site

Journal/Book/Conference Title

Suicide and life-threatening behavior

Related Resources

Link to Article in PubMed

PubMed ID