When Feedback Backfires: Influences of Negative Discrepancies Between Physicians' Self and Assessors' Scores on Their Subsequent Multisource Feedback Ratings.

INTRODUCTION: With multisource feedback (MSF) physicians might overrate their own performance compared with scores received from assessors. However, there is limited insight into how perceived divergent feedback affects physicians' subsequent performance scores. METHODS: During 2012 to 2018, 103 physicians were evaluated twice by 684 peers, 242 residents, 999 coworkers, and themselves in three MSF performance domains. Mixed-effect models quantified associations between the outcome variable "score changes" between first and second MSF evaluations, and the explanatory variable "negative discrepancy score" (number of items that physicians rated themselves higher compared with their assessors' scores) at the first MSF evaluation. Whether associations differed across assessor groups and across a physician's years of experience as a doctor was analyzed too. RESULTS: Forty-nine percent of physicians improved their total MSF score at the second evaluation, as assessed by others. Number of negative discrepancies was negatively associated with score changes in domains "organization and (self)management" (b = -0.02; 95% confidence interval [CI], -0.03 to -0.02; SE = 0.004) and "patient-centeredness" (b = -0.03; 95% CI, -0.03 to -0.02; SE = 0.004). For "professional attitude," only negative associations between score changes and negative discrepancies existed for physicians with more than 6-year experience (b6-10yearsofexperience = -0.03; 95% CI, -0.05 to -0.003; SE = 0.01; b16-20yearsofexperience = -0.03; 95% CI, -0.06 to -0.004; SE = 0.01). DISCUSSION: The extent of performance improvement was less for physicians confronted with negative discrepancies. Performance scores actually declined when physicians overrated themselves on more than half of the feedback items. PA score changes of more experienced physicians confronted with negative discrepancies and were affected more adversely. These physicians might have discounted feedback due to having more confidence in own performance. Future work should investigate how MSF could improve physicians' performance taking into account physicians' confidence.

Exploring Validity Evidence Associated With Questionnaire-Based Tools for Assessing the Professional Performance of Physicians: A Systematic Review.

PURPOSE: To collect and examine-using an argument-based validity approach-validity evidence of questionnaire-based tools used to assess physicians' clinical, teaching, and research performance. METHOD: In October 2016, the authors conducted a systematic search of the literature seeking articles about questionnaire-based tools for assessing physicians' professional performance published from inception to October 2016. They included studies reporting on the validity evidence of tools used to assess physicians' clinical, teaching, and research performance. Using Kane's validity framework, they conducted data extraction based on four inferences in the validity argument: scoring, generalization, extrapolation, and implications. RESULTS: They included 46 articles on 15 tools assessing clinical performance and 72 articles on 38 tools assessing teaching performance. They found no studies on research performance tools. Only 12 of the tools (23%) gathered evidence on all four components of Kane's validity argument. Validity evidence focused mostly on generalization and extrapolation inferences. Scoring evidence showed mixed results. Evidence on implications was generally missing. CONCLUSIONS: Based on the argument-based approach to validity, not all questionnaire-based tools seem to support their intended use. Evidence concerning implications of questionnaire-based tools is mostly lacking, thus weakening the argument to use these tools for formative and, especially, for summative assessments of physicians' clinical and teaching performance. More research on implications is needed to strengthen the argument and to provide support for decisions based on these tools, particularly for high-stakes, summative decisions. To meaningfully assess academic physicians in their tripartite role as doctor, teacher, and researcher, additional assessment tools are needed.

Validation of the INCEPT: A Multisource Feedback Tool for Capturing Different Perspectives on Physicians' Professional Performance.

INTRODUCTION: Multisource feedback (MSF) instruments are used to and must feasibly provide reliable and valid data on physicians' performance from multiple perspectives. The "INviting Co-workers to Evaluate Physicians Tool" (INCEPT) is a multisource feedback instrument used to evaluate physicians' professional performance as perceived by peers, residents, and coworkers. In this study, we report on the validity, reliability, and feasibility of the INCEPT. METHODS: The performance of 218 physicians was assessed by 597 peers, 344 residents, and 822 coworkers. Using explorative and confirmatory factor analyses, multilevel regression analyses between narrative and numerical feedback, item-total correlations, interscale correlations, Cronbach's α and generalizability analyses, the psychometric qualities, and feasibility of the INCEPT were investigated. RESULTS: For all respondent groups, three factors were identified, although constructed slightly different: "professional attitude," "patient-centeredness," and "organization and (self)-management." Internal consistency was high for all constructs (Cronbach's α ≥ 0.84 and item-total correlations ≥ 0.52). Confirmatory factor analyses indicated acceptable to good fit. Further validity evidence was given by the associations between narrative and numerical feedback. For reliable total INCEPT scores, three peer, two resident and three coworker evaluations were needed; for subscale scores, evaluations of three peers, three residents and three to four coworkers were sufficient. DISCUSSION: The INCEPT instrument provides physicians performance feedback in a valid and reliable way. The number of evaluations to establish reliable scores is achievable in a regular clinical department. When interpreting feedback, physicians should consider that respondent groups' perceptions differ as indicated by the different item clustering per performance factor.