Outcomes for treatment of depression in the Veterans Health Administration: Rates of response and remission calculated from clinical and survey data

Healthcare Effectiveness Data and Information Set (HEDIS) quality measures for depression treatment aggregate Patient Health Questionnaire (PHQ)-9 data from routine clinical assessments recorded in electronic health records (EHR). To determine whether aggregated PHQ-9 data in US Veterans Health Administration (VHA) EHRs should be used to characterize the organization's performance, we compared rates for depression response and remission calculated from EHRs with rates calculated with data representing the underlying Veteran patient population estimated using Veterans Outcome Assessment (VOA) survey data. We analyzed data from initial assessments and 3-month follow-up for Veterans beginning treatment for depression. EHR data were available for only a minority of Veteran patients, and the group of Veterans with EHR data differed from the underlying Veteran patient population with respect to demographic and clinical characteristics. Aggregated rates of response and remission from EHR data were significantly different from estimates based on representative VOA data. The findings suggest that until patient-reported outcome from EHRs are available for a substantial majority of patients receiving care, aggregated measures of patient outcomes derived from these data cannot be assumed to be representative of the outcomes for the overall population, and they should not be used as outcome-based measures of quality or performance.