Measuring progression of lens opacities for longitudinal studies

A number of classifications schemes for the grading of lens opacities have been developed but none of them have been examined for reliability and validity in prospective or longitudinal studies. The purpose of this study was to develop and test a method for determining progression which was sufficiently sensitive to true change and insensitive to measurement error. The method consists of assessing various sources of measurement error, determining the likely magnitude of the error, and setting boundaries for changes in opacification which incorporate this error. Measurement error was assessed in 203 nuclear lens photographs graded using a decimal system and 136 cortical lens photographs graded in 1/16th area. 95% confidence intervals were calculated of the distribution of differences between photographs and graders and the width of this intervals defined as measurement error. The magnitude of the measurement error was +/- 0.7 units for nuclear opacities and +/- 2/16 for cortical opacities. The utility of this approach was tested in a series of 84 clinic patients followed at irregular intervals for up to 3 years. Only one eye showed "improvement" and progression rates ranged from 1.8 to 5.5% for nuclear opacities and 0 to 7.4% for cortical opacities in this clinic-based series. This approach appears to be useful for discarding noise and enabling the assessment of progression for use in longitudinal studies.