Radiation Dose Responses, Thresholds, and False Negative Rates in a Series of Cataract Surgery Prevalence Studies...

In order to find imperfect sensitivity or the false-negative rate of cataract surgery due to latent clinically significant or severe cataract and a dose-response threshold, cataract surgery prevalence data analysis was made for each of the two-year periods from 1986 through 2005 among atomic bomb survivors. Using the latent variable regression model published earlier, cataract prevalence studies allowing for false-negative and/or false-positive rates were conducted in each of the 10 two-year periods during 1986 to 2005. As the best statistical model for prevalence data, a logistic model with a non-negligible false negative rate was selected for analysis. The commonly used naïve logistic analysis resulted in an average odds ratio (OR) at 1 Gy of 1.33 (95%CI: 1.28, 1.38) for cataract surgery with no linear time trend (p = 0.334), and the OR at 1 Gy with the model allowing for sensitivity was 1.48 (95%CI: 1.40, 1.56) for clinically significant or severe cataract with no linear time trend (p = 0.263). Cataract surgery is an imperfect surrogate for clinically significant cataract, and the sensitivity increased from 0.15 to 0.50 during the 20 y with increasing rate of sensitivity per 2-y period of approximately 22%. The dose-response threshold based on a naïve logistic model for cataract surgery ranged from 0.04-1.03 Gy (simple average of 0.41 Gy) with no linear time trend (p = 0.620) in the 10 2-y periods compatible with the no dose response threshold model in all periods.