Detecting significant change in wavefront error: how long does it take?

Measurement noise in ocular wavefront sensing limits detection of statistically significant change in high-order wavefront error (HO WFE). Consequently, measurement noise is problematic when trying to detect progressive change in HO WFE. Our aims were to determine the necessary amount of time to detect age-related change in HO WFE given measurement variability and HO WFE composition and magnitude; and to minimise the length of time necessary to detect change.

Five subjects with 0.26 to 1.57 micrometres root mean square HO WFE (HORMS) over a 6 mm pupil were measured 12 times in 10 to 15 minutes using a custom Shack-Hartmann wavefront sensor. Each individual's standard deviation of measures was used to calculate the 95% confidence interval around their mean HO RMS. Data previously reported on the rate of change in the HO RMS due to normal ageing and pupil diameter were used to calculate time to detect change exceeding this interval given measurement variability.

Single measurements limit statistical detection to a range of eight to 30 years.Increasing the number of WFE measurements per visit decreases time to detection (for example, seven measurements reduce the range to three to 14 years). The number of years to detect a change requires consideration of the subject's measurement variability,level and distribution of aberrations and age. Uncertainty in locating pupil centre accounts for 39 +/- 8 per cent of the total variability.

The ability to detect change in HO WFE over a short period due to normal ageing is difficult but possible with current WFE measurement technology. Single measurements of HO WFE become less predictive of true HO WFE with increasing measurement variability. Multiple measurements reduce the variability. Even with proper fixation and instrument alignment, pupil centre location uncertainty in HO WFE measurements is a non-trivial contributor to measurement variability.