A psychophysical technique for estimating the accuracy and precision of retinoscopy

The accuracy and precision of retinoscopy need to be evaluated in an objective and unambiguous manner to judge an individual's performance in this task during training or in research for consistent measurements of refractive error. This study describes and evaluates a psychophysical technique for obtaining simultaneous and unbiased estimates of accuracy and precision in retinoscopy.

Subjects with zero to 12 years of experience with retinoscopy performed the psychophysical task on a model eye (75 subjects) and on a cyclopleged human eye with spherical refractive error (30 subjects). Subjects made forced choice judgments of 'with' or 'against' for lens powers within ± 0.5 D of expected neutrality (in 0.12 D steps), each placed 20 times before the eye in random order. Accuracy and precision were determined from the mean and standard deviation of the resultant psychometric function.

Subjects could be qualitatively divided into those with good and poor accuracy and precision based on the task outcomes. The median and inter-quartile range of accuracy (no experience: 0.16 ± 0.34 D; four or more years of experience: 0.06 ± 0.11 D) and precision (no experience: 0.30 ± 0.39 D; four or more years of experience group: 0.13 ± 0.08 D) improved with task experience (p < 0.001). Median accuracy and precision in the human eye were similar to the model eye (p > 0.8). Accuracy and precision were poorly correlated with each other for both the human eye and model eye (|ρ| ≤ 0.20; p ≥ 0.09 for all).

The psychophysical retinoscopic task could differentiate subjects based on their accuracy and precision and also capture key elements of improvement in retinoscopic performance with experience. Retinoscopic performance using this task was similar to previous reports using routine clinical retinoscopy. Therefore, the psychophysical technique may be used to evaluate and monitor objectively retinoscopic performance.