Abstract
Reply: We are pleased to receive insightful comments on our work by Brown and coauthors. They raise several important methodology issues with respect to studies of pupil measurement that deserve further discussion and wider consideration. Brown and coauthors suggest that our study was intended to measure the pupil under dark-adapted conditions. This, however, was not our intent. We make a strict distinction between the dark-room conditions described in our study and true dark adaptation, which we reserve for the unique physiological state of rod-driven retinal sensitivity occurring after approximately 35 minutes of complete darkness in the normal eye.1 Brown and coauthors are rightly concerned that the time interval between the first flash exposure and subsequent measurements by the second examiner could have been insufficient for full pupil recovery. If the order of examiner was not randomized, this would lead to a sequence effect whereby the second examiner would measure a consistently smaller pupil size. They provide a plot of pupil size as a function of time in 2 eyes photographed using their recently published technique2 to illustrate this point. In our original publication, we reported that examiner order was arbitrary. In fact, 1 examiner was first in all but 2 cases. This provided an opportunity to test Brown and coauthors' hypothesis. Using a 2-way repeated measures analysis of variance (ANOVA) model, we compared the mean pupil size measured at each light level across examiners and also tested for an interaction between examiner and light level for the digital photographic method of measurement. If the concerns expressed by Brown and coauthors were a factor in our experiments, we would expect to find a significant interaction between these 2 variables. We found no interaction between light level and examiner (Figure 1; P=.09) and no difference between the 2 examiners in pupil size measurements (P=.20).Figure 1.: Mean measured pupil diameter by examiner as a function of light level for the photographic measurement method.Because any observed differences in pupil size could be due to examiner effects or sequence effects, we further explored their concern by testing for differences between examiners in mean pupil diameter with a 2-way ANOVA comparing light level across methods of measurement for each examiner sequence (examiner 1 then examiner 2; examiner 2 then examiner 1). We did not find a significant sequence effect on pupil size when we compared light levels by method interactions or method of measurement for either test sequence and show the comparison of pupil size by light level for each examiner (Figure 1). These findings should satisfy the concerns expressed by Brown and coauthors. Although the concern about an examiner sequence effect is valid and they identify a methodological weakness of our design, it did not influence our reported results. Brown and coauthors suggest that the time between photographs was not controlled or measured. Although we elected not to report these data, they were recorded and provide further evidence to allay their concerns. We show the difference between examiners for pupil diameter measurements as a function of time (Figure 2). The slope of this function was not significantly different than zero by linear regression (−0.018 mm/min; 95% confidence interval −0.050 to +0.014). The time between pupil measurements was 3.5 minutes at the medium light level (5 lux) and 4.3 minutes at the low light level (<0.63 lux).Figure 2.: Difference in measured pupil diameter between examiner 1 and examiner 2 as a function of the time difference between the 2 measurements. Number = 45 (×3 light levels); r 2 = 0.00, P = .98, Y = (0.130)X. The slope of this function was not statistically different than zero.A second hypothesis proposed by Brown and coauthors is that the variations observed in pupil size measurements with the Colvard pupillometer were possibly due to a prolonged tonic proximal accommodative response, and they suggest that we repeat our study with a randomized examiner order to address this issue. This raises another important concern of any pupil measurement study; namely, control of accommodation. The normal time course of pupil recovery from a convergent accommodative stimulus is seconds.3 Given that more than 2 minutes had passed between each examiner's measurements, the pupil should have had sufficient time to completely recover from an accommodative stimulus. We are grateful for the comments of Brown and coauthors and appreciate the opportunity to respond to their concerns. Michael Twa OD Melissa Bailey OD, PhD Mark Bullimore MCOptom, PhD John Hayes PhD
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have