Inter-rater reliability of the Bereavement Risk Assessment Tool

Abstract

Objective: The Bereavement Risk Assessment Tool (BRAT) was designed to consistently communicate information affecting bereavement outcomes; to predict the risk for difficult or complicated bereavement based on information obtained before the death; to consider resiliency as well as risk; and to assist in the efficacy and consistency of bereavement service allocation. Following initial development of the BRAT's 40 items and its clinical use, this study set out to test the BRAT for inter-rater reliability along with some basic validity measures. Method: Case studies were designed based on actual patients and families from a hospice palliative care program. Bereavement professionals were recruited via the internet. Thirty-six participants assessed BRAT items in 10 cases and then estimated one of 5 levels of risk for each case. These were compared with an expert group's assignment of risk. Results: Inter-rater reliability for the 5-level risk scores yielded a Fleiss’ kappa of 0.37 and an intra-class correlation (ICC) of 0.68 (95% CI 0.5-0.9). By collapsing scores into low and high risk groups, a kappa of 0.63 and an ICC of 0.66 (95% CI 0.5-0.9) was obtained. Participant-estimated risk scores yielded a kappa of 0.24. Although opinion varied on the tool's length, participants indicated it was well organized and easy to use with potential in assessment and allocation of bereavement services. Limitations of the study include a small sample size and the use of case studies. Limitations of the tool include the subjectivity of some items and ambiguousness of unchecked items. Significance of results: The collapsed BRAT risk levels show moderately good inter-rater reliability over clinical judgement alone. This study provides introductory evidence of a tool that can be used both prior to and following a death and, in conjunction with professional judgment, can assess the likelihood of bereavement complications.