APPROPRIATE STATISTICAL-METHODS TO ACCOUNT FOR SIMILARITIES IN BINARYOUTCOMES BETWEEN FELLOW EYES

Purpose. Many ocular measurements are more alike between fellow eyes t han between eyes from different individuals. To make appropriate infer ences using data from both eyes rather than the best or worst eye, sta tistical methods that account for the association between fellow eyes must be used. Methods. Marginal and conditional regression models acco unt for the association between fellow eyes in different ways. The aut hors compare and contrast these methods using data from a series of pa tients with retinitis pigmentosa in whom the primary object is to iden tify risk factors, some of which are subject specific and some of whic h are eye specific, for visual acuity loss (as a binary outcome) among affected subjects. Results. Odds ratios for age, gender, presence of posterior subcapsular cataract, and genetic type of retinitis pigmento sa obtained from the marginal model were all larger than those from th e conditional model. Familial aggregation of visual acuity loss was st atistically significant in the marginal, but not in the conditional, m odel. Conclusions. The estimates and interpretation of the association between an ocular outcome and risk factors can differ significantly b etween these two approaches. The choice of model depends on the scient ific questions of interest rather than on statistical considerations. Computer programs are available for implementing both models.