Computerized Adaptive Suicidal Risk Stratification and Prediction

ABSTRACT In the next 60 minutes, at least 57 people in the United States (US) will attempt to kill themselves, and five will die. Over 80% of these individuals encountered a healthcare provider in the 12 months before his or her death, but their risk will have gone undetected or, if detected, poorly quantified and not sufficiently monitored longitudinally. The proposed research will address this deficit by building and validating the Computerized Adaptive Test for Suicide Scale -- Expanded (CAT-SS EXPANDED), software that can screen for, quantify, and monitor suicide risk in a way far superior to existing instruments. It will enable at least three key Zero Suicide performance elements, including systematic, standardized frontline suicide risk screening and measurement (Identify), tracking and communicating risk across locations of care (Transitions), and monitoring changes in patient outcomes resulting from continuous quality improvement activities (Improve). In a paradigm shifting innovation, the CAT-SS EXPANDED will not only comprise the first multi-dimensional CAT of its kind, it will import suicide risk indicators that have already been validated by the NIMH-funded Mental Health Research Network (MHRN) directly from a health system’s electronic health record (EHR). This SBIR Fast Track has two phases. Phase 1 will build the CAT-SS EXPANDED, starting with well-established preliminary work and using iterative user testing with 20 clinician-patient dyads to refine the features. Phase 2 will have two Aims. The first Aim will validate the CAT-SS EXPANDED against an independent research clinician’s suicide risk stratification and suicidal behavior 24 weeks after the index visit (n=700). The second Aim will complete two-way integration of the validated CAT-SS EXPANDED into UMass’ EHR production environment and will evaluate clinical usability and acceptability with a new sample of ~30 suicide-risk enriched patients. Innovation and Impact: While traditional suicide screening tools exist, they are very limited, a conclusion supported by systematic reviews, the National Action Alliance for Suicide Prevention, and the NIMH. Because of these limitations, strong, well-validated clinical decision support to guide appropriate levels of care do not exist, reinforcing non-standardized and inefficient workflow, such as a tendency to conservatively order unnecessary emergency psychiatric evaluations. The CAT-SS EXPANDED will address these well-known measurement limitations using cutting edge strategies embodied in multi-dimensional CAT and EHR-derived risk indicators, an unprecedented capability impossible to achieve with existing measures. While this study focuses on the ED as a starting point because of its well-known risk burden and emerging literature supporting the value of universal suicide risk screening combined with brief interventions, the CAT-SS EXPANDED will be specifically designed and later tested as the first approach to suicide risk monitoring suitable for enterprise deployment across settings and locations of care, which is transformational for suicide prevention and fundamental to the Zero Suicide model’s emphasis on holistic system change.NARRATIVE Having a psychometrically valid suicide risk instrument that efficiently screens for risk, provides risk stratification, and can be used across settings and easily integrated into the electronic health record (EHR) has tremendous potential to enable the enterprise implementation of the Zero Suicide model across an entire healthcare system, regardless of location of care. Because the software we will create and validate, the Computerized Adaptive Test for Suicide Scale – Expanded, blends computerized adaptive testing (CAT) and EHR suicide risk indicators previously validated using advanced analytic approaches by the NIMH-funded Mental Health Research Network, it is markedly different from existing instruments, which focus on a specific setting or patient population, rely on obsolete administration and psychometric methods, contain only screening capability with no monitoring components, and are not EHR-integrated in such a way as to tap into the data stored in the EHRs. This SBIR Fast Track proposal is necessary to keep pace with the burgeoning adoption of electronic patient reported outcomes and EHRs across healthcare and is foundational to the next step of designing better suicide interventions.