Assessing doctors' competencies using multisource feedback: validating a Japanese version of the Sheffield Peer Review Assessment Tool (SPRAT)

Abstract

OBJECTIVE: To assess the validity and reliability of the Sheffield Peer Review Assessment Tool (SPRAT) Japanese version for evaluating doctors' competencies using multisource feedback. METHODS: SPRAT, originally developed in the UK, was translated and validated in three phases: (1) an existing Japanese version of SPRAT was back-translated into English; (2) two expert panel meetings were held to develop and assure content validity in a Japanese setting; (3) the newly devised Japanese SPRAT instrument was tested by a multisource feedback survey, validity was tested using principal component factor analysis, and reliability was assessed using generalisability and decision studies based on generalisability theory. RESULTS: 86 doctors who had been practising for between 2 and 33 years participated as assessees and were evaluated with the SPRAT tool. First, the doctors identified 1019 potential assessors who were each sent SPRAT forms (response rate, 81%). The mean number of assessors per doctor was 9.7 (SD=2.5). The decision study showed that 95% CIs of ± 0.5 were achieved with only 5 assessors. 85 of the 86 doctors achieved scores that could be placed with 95% CI above the 4 expected standard. Doctors received lower scores from more senior assessors (p<0.001) and higher scores from those they had known longer (p<0.001). Scores also varied with the job role (p<0.05). CONCLUSIONS: Following translation and content validation, the Japanese instrument behaved similarly to the UK tool. Assessor selection remains a primary concern, as the assessment scores are affected by the seniority of the assessor, the length of the assessor-assessee working relationship, and the assessor's job role. Users of the SPRAT tool need to be aware of these limitations when administering the instrument.