Photorefraction (PR) methods have been widely used for pediatric vision screening since the 1980's. While PR is easy to implement, the accuracy of refractive error measurements in humans has been unsatisfactory, largely due to the variations of intraocular scattering, the retinal reflectance, and pupil size. The objective of our studies was to clinically evaluate the accuracy of refraction measurements of an improved PR-based device, the Dynamic Ocular Evaluation System (DOES), and to

more »

... e whether the accuracy is affected by the patient's age, race, and pupil size, which are relevant to individual intraocular scattering and retinal reflectance. We performed DOES measurements in 99 volunteers (198 eyes) under two environmental light conditions and using four fixation targets. These results were compared to the standard clinical refraction testing performed the same day. The correlation and standard deviation were determined by Bland-Altman analysis. The influence of intraocular scattering, retinal reflectance, and pupil size was evaluated by comparing results from different age groups, races, and lighting conditions. In the region between -4 diopter (D) and +4D, the device showed a binocular refraction measurement accuracy of 0.45 D, 0.3 D, and 0.18 D root-mean-square (RMS) error (n=1337) for spherical equivalent and cross cylinders J o and J 45 , respectively. Inaccuracy increased at high refraction (>4D). Age, race, and pupil size did not appear to significantly affect DOES PR measurement. This suggests that enhancements in the PR system and analysis may satisfactorily correct intersubject variability that currently limits the clinical use of PR devices and measurements.