Empirical variability in the calibration of slope-based eccentric photorefraction

Refraction estimates from eccentric infrared (IR) photorefraction depend critically on the calibration of luminance slopes in the pupil. While the intersubject variability of this calibration has been estimated, there is no systematic evaluation of its intrasubject variability. This study determined the within subject inter- and intra-session repeatability of this calibration factor and the optimum range of lenses needed to derive this value. Relative calibrations for the MCS PowerRefractor and a customized photorefractor were estimated twice within one session or across two sessions by placing trial lenses before one eye covered with an IR transmitting filter. The data were subsequently resampled with various lens combinations to determine the impact of lens power range on the calibration estimates. Mean (±1.96 SD) calibration slopes were 0.99±0.39 for North Americans with the MCS PowerRefractor (relative to its built-in value) and 0.65±0.25 Ls/D and 0.40±0.09 Ls/D for Indians and North Americans with the custom photorefractor, respectively. The ±95% limits of agreement of intrasubject variability ranged from ±0.39 to ±0.56 for the MCS PowerRefractor and ±0.03 Ls/D to ±0.04 Ls/D for the custom photorefractor. The mean differences within and across sessions were not significantly different from zero (p>0.38 for all). The combined intersubject and intrasubject variability of calibration is therefore about ±40% of the mean value, implying that significant errors in individual refraction/accommodation estimates may arise if a group-average calibration is used. Protocols containing both plus and minus lenses had calibration slopes closest to the gold-standard protocol, suggesting that they may provide the best estimate of the calibration factor compared to those containing either plus or minus lenses.