Reliability of the longitudinal experts all data (LEAD) methodology for determining the presence of elder mistreatment.

Developing reliable screening tools to identify elder mistreatment requires an accurate and reproducible reference standard. This study sought to investigate the reliability of the Longitudinal, Experts, All Data (LEAD) methodology as a reference standard in confirming presence of elder mistreatment. We analyzed data from a large, emergency department-based study that used a LEAD panel to determine the reference standard. For this study, a second, blinded LEAD panel reviewed clinical material for 40 patients. For each panel, five content experts voted on whether elder mistreatment was present. We found moderate agreement between the two LEAD panels in determining presence of elder mistreatment: 85% agreement; k = 0.58; 95% Confidence Interval 0.28-0.87. Individual raters for both LEAD panels reported being mostly certain or certain >90% of votes. Efforts to further characterize and improve the reliability of the LEAD methodology in this context are warranted.