Reliability of performance-based clinical skill assessment of emergency medicine residents

Objective: To test the overall reliability of a performance-based clinical skill assessment for entering emergency medicine (EM) residents. Also, to investigate the reliability of separate reporting of diagnostic and management scores for a standardized patient case, subjective scoring of patient notes, and interstation exercise scores. Methods: Thirty-four first- year EM residents were tested using a 10-station standardized patient (SP) examination. Following each 10-minute encounter, the residents completed a patient note that included differential diagnosis and management. The residents also were asked to read an ECG or chest x-ray (CXR) associated with each case. History, physical examination, and interpersonal skills were scored by the SPs. The patient note, CXR, and ECG readings were scored by faculty emergency physicians. Intercase reliability was determined for the residents. Results: Global score reliability, Cronbach's a = 0.85. Reliabilities for the other components were: history, 0.77; physical examination, 0.83; and interpersonal skills, 0.80. Differential diagnosis and management reliabilities were 0.61 and 0.66, respectively. Subjective scoring of the patient note resulted in acceptable reliability for legibility (0.80), history completeness (0.80), and history organization (0.81). Physical examination completeness and organization reliabilities were 0.74 and 0.73. For ECG and CXR readings, a = 0.74 and 0.34, respectively. Conclusions: SPs can be used to reliably assess bedside clinical skills of EM residents. While component reliability levels are slightly lower than the global clinical skill reliability coefficient, they are still high enough to use for identification of individual strengths and weaknesses.