A statistical analysis of Modified Clinical Technique vision screening of preschoolers by optometry students.

Fourth year optometry students screened 745 preschoolers using a slightly altered Modified Clinical Technique (MCT) under the supervision of a faculty doctor. Children who failed the MCT were randomly selected and then matched by age, sex, and ethnic origin to children who had passed the screening battery. The 61 screening failures and 45 matched controls were later given full eye examinations with cycloplegia by University of Alabama at Birmingham faculty doctors who were unaware of the screening results. The positive predictive value (PPV) (0.52) and negative predictive value (NPV) (0.78) of the MCT were calculated directly from the 2 x 2 contingency table crossing screening results and a standard diagnosis. Sensitivity [0.50, k(1,0) = 0.29], specificity [0.79, k(0,0) = 0.30], efficiency [0.70, k(0.5,0) = 0.29] of the MCT, and the prevalence (0.30) of children failing the standard diagnosis were estimated using statistics appropriate to the prospective sampling design. The reproducibility of the diagnosis, estimated by analyzing multiple, independent diagnosis of each study child by seven doctors was moderate (kappa D 0.58). Statistics summarizing the agreement between the MCT and the diagnosis by the individual study doctor are similar to those obtained with comparison to the standard diagnosis. The characteristics of the MCT may be generalized only to similar populations that are screened by clinicians with similar experience, using the same tests.