Published in:
01-11-2012 | Refractive Surgery
Can we measure mesopic pupil size with the cobalt blue light slit-lamp biomicroscopy method?
Authors:
Miguel J. Maldonado, Alberto López-Miguel, David P. Piñero, José R. Juberías, Juan C. Nieto, Jorge L. Alió
Published in:
Graefe's Archive for Clinical and Experimental Ophthalmology
|
Issue 11/2012
Login to get access
Abstract
Background
The aim of this work is to assess a previously described slit-lamp biomicroscopy-based method (SLBM) for measuring pupil diameter and compare it to Colvard infrared pupillometry (CIP).
Methods
Two examiners performed three repeated measurements with each instrument in 40 healthy eyes. We determined the agreement of SLBM and CIP, intraobserver and interobserver repeatabilities, and interobserver concordance (kappa) and SLBM ability for detecting pupil sizes over 6.0 mm.
Results
The mean (±standard deviation [SD]) pupil diameter was 5.81 ± 0.70 mm with SLBM and 6.26 ± 0.68 mm with CIP (p = 0.01) averaging both examiner’s results. Mean differences between the SLBM and CIP were –0.60 mm and –0.30 mm for each examiner using the average of the three readings (p = 0.02), and they were very similar using the first reading. Intraobserver reproducibility: the width of the 95% LoA ranged from 1.79 to 2.30 mm. The ICCs were 0.97 and 0.92 for SLBM, and 0.96 and 0.90 for CIP. Interobserver reproducibility: the width of the LoA ranged from 1.82 to 2.09 mm. Kappa statistics were 0.39 and 0.49 for the first and mean SLBM readings, respectively, and 0.45 for both the first and mean CIP readings. Sensitivity and specificity of SLBM for detection of pupils larger than 6 mm ranged from 55.56% to 73.68% and from 76.19% to 95.45%, respectively. The best trade-off between sensitivity and specificity ranged from 5.4 mm to 6.2 mm.
Conclusions
Although the SLBM is quite repeatable, it underestimates mesopic pupil size and shows a too wide range of agreement with CIP. SLBM shows low sensitivity in detecting pupils larger than 6 mm, which may be misleading when planning anterior segment surgery. Previous grading-consensus training strategies may increase interrater reproducibility, and compensation for the systematic underestimation could improve accuracy of the SLBM.