A MASKED, OBJECTIVE METHOD OF MEASURING RETINOSCOPY PERFORMANCE

Title A MASKED, OBJECTIVE METHOD OF MEASURING RETINOSCOPY PERFORMANCE
Author, Co-Author Katherine Clore, David Corliss, Kent Daum, James Beisel, Glenn Hammack, James Marbourg
Topic
Year
1992
Day
Monday
Program Number
Poster 64
Room
Great Hall
Affiliation
Abstract Extensive psychometric measurement techniques have been developed for measuring written examinations, particularly multiple- choice tests; these techniques are not used in practical examinations. The purpose of this study is to describe a psychometrically sound method of examining a basic clinical skill: retinoscopy. The study was completed using identical paradigms in two consecutive years. Year one included 42 students and four instructors; year two included 37 students and four instructors. Autorefractor measurements were also made both years. Refractive errors of patients were chosen to fall within a typical patient population range. Every participant performed retinoscopy on 24 eyes during a round robin consisting of 16 stations with 4 rest stops. Each participant had 4 minutes to complete retinoscopy on two eyes, record the results, and clear the phoropter. None of the participants knew the refractive error of any of the patients and all clinicians, including the instructors, performed the examination under identical conditions. To deal with potential axis measurment errors, we reduced refractive errors to powers in the vertical and horizontal meridians. A measurement error was defined as the absolute difference between the meridional power measured by an individual clinician and the mean of the powers determined by the faculty and the autorefractor. The mean measurement error for the instructors was 0.38D and that for the students was 0.73D (t=5.88, p=0.000). For the entire study, the split-halves Pearson coefficient was 0.45; the Spearman-Brown correlation coefficient of reliability was 0.62. Correlations between each individual measurement error and all other errors suggest that positive spherical refractive errors with cylinders of 1.00D or more provide more reliable measures of performance than negative errors with smaller cylinders. These results show that a psychometrically reliable measurement of a clinician's retinoscopy skill requires usi
Affiliation of Co-Authors
Outline