Prevalence of Effective Audit-and-Feedback Practices in Primary Care Settings: A Qualitative Examination Within Veterans Health Administration

Objective The purpose of this study is to uncover and catalog the various practices for delivering and disseminating clinical performance in various Veterans Affairs (VA) locations and to evaluate their quality against evidence-based models of effective feedback as reported in the literature. Background Feedback can enhance clinical performance in subsequent performance episodes. However, evidence is clear that the way in which feedback is delivered determines whether performance is harmed or improved. Method We purposively sampled 16 geographically dispersed VA hospitals based on high, low, consistently moderate, and moderately average highly variable performance on a set of 17 outpatient clinical performance measures. We excluded four sites due to insufficient interview data. We interviewed four key personnel from each location (n = 48) to uncover effective and ineffective audit and feedback strategies. Interviews were transcribed and analyzed qualitatively using a framework-based content analysis approach to identify emergent themes. Results We identified 102 unique strategies used to deliver feedback. Of these strategies, 64 (62.74%) have been found to be ineffective according to the audit-and-feedback research literature. Comparing features common to effective (e.g., individually tailored, computerized feedback reports) versus ineffective (e.g., large staff meetings) strategies, most ineffective strategies delivered feedback in meetings, whereas strategies receiving the highest effectiveness scores delivered feedback via visually understood reports that did not occur in a group setting. Conclusions Findings show that current practices are leveraging largely ineffective feedback strategies. Future research should seek to identify the longitudinal impact of current feedback and audit practices on clinical performance. Application Feedback in primary care has little standardization and does not follow available evidence for effective feedback design. Future research in this area is warranted.