An Experimental Investigation of Factors that Influence Clinicians’ Decisions to Hospitalize Suicidal Patients

Abstract

Psychiatric inpatient hospitalization and suicide risk share a strong empirical relationship such that frequent users of hospitalization display high rates of post-discharge suicidal behavior. Although research seeking to better understand this relationship is growing, there is a dearth of research evaluating how suicidal individuals end up being hospitalized. One route to exploring this question involves investigating how clinicians come to the decision to hospitalize or not hospitalize their suicidal patients. Existing research evaluating clinical decision-making related to hospitalizing suicidal patients is minimal, and major limitations to the existing research impede generalizability and interpretation. To address some of the limitations of the existing research, this two-part vignette-based study involved designing ecologically valid vignettes depicting suicidal patients, evaluating empirically-supported predictors of hospitalization, and exploring climate-, clinician-, and patient-level variables as potential predictors of hospitalization. An initial pilot study was conducted to finalize study measures and stimuli based on feedback from 30 clinicians. In the main study, clinicians (n= 155) were shown 4 vignettes describing hypothetical suicidal patients and asked to rate each patient’s likelihood of dying by suicide in the next 24 hours, acute risk for suicide using an ordinal scale derived from Rudd’s Acute Suicide Risk Continuum (2006), and whether or not they would hospitalize the patient. Vignettes depicted 4 different patients with combinations of two patient-level variables (diagnosis of borderline personality disorder [yes/no] and history of psychiatric hospitalizations [yes/no]) randomized to vary across vignettes and with vignette-detail combinations counterbalanced across participants. Additional data on clinician-level variables (i.e., professional characteristics, experience and concerns about treating suicidal patients, self-efficacy, burnout) and climate-relevant variables (i.e., organizational norms and perceived behavioral control over the decision to hospitalize) were collected. Results indicated that there was considerable heterogeneity in hospitalization decisions and suicide risk ratings between clinicians, although risk likelihood ratings and hospitalization endorsement were positively correlated. In addition, the experimentally-manipulated patient factors (i.e., BPD diagnosis and hospitalization history) did not significantly predict hospitalization decisions or suicide risk ratings across vignettes. Instead, only clinician- and climate-level factors emerged as significant predictors although the specific factors that were significant varied across outcomes. In particular, deciding to hospitalize suicidal patients was predicted by having an M. D. degree and positive attitudes towards hospitalization, higher ordinal ratings of acute suicide risk were predicted by clinician caseload and work setting, and higher risk likelihood ratings were predicted by greater clinician self-efficacy and intentions to hospitalize. Implications of the findings regarding clinician training and directions for future research are discussed.