Scoring standardized patient examinations: lessons learned from the development and administration of the ECFMG Clinical Skills Assessment (CSA®)

Throughout the 40 year history of standardized patient assessments and OSCEs, there have been numerous advancements, including many that involve scoring the simulated clinical encounters. While there is no clear agreement on how examinees' performance should be documented or scored in an encounter, there is a consensus that several well-chosen SP encounters are required to produce reliable examinee scores. There also continues to be some debate as to who should do the scoring on an SP-based assessment. While logistics and cost will certainly play a role, it is probably best to use the person who is most familiar with the domain being assessed. In some instances this will be the SP; in others, an outside observer or content expert. Finally, with the growing use of OSCEs for summative purposes (e.g. certification, licensure), special attention must be paid to fairness issues. Since the same test form cannot be used day after day, examinee scores must be 'equated', taking into account the psychometric properties of scores from individual cases and individual SPs. To date, the CSA has been one of the highest-volume, high-stakes, standardized patient assessments to be developed and successfully administered. In 2003 alone, over 11 500 IMGs were tested. The early conceptual framework for this assessment was synthesized from the research endeavours of several notable individuals, including, amongst many others, Harden et al. 1975, Swanson & Stillman, 1990, Newble & Swanson, 1988, Vu et al. 1992 and Colliver, 1995. The early prototype administrations of the CSA, including many operational research studies, were supported and guided by Dr Friedman Ben-David, Friedman et al. 1991, 1993, Stillman et al. 1992, and Sutnick et al. 1993, 1995.