Purpose
To examine the correlations between ophthalmic screening tests in a cohort of American Indian/Alaskan Natives (AI/AN) from the Pacific Northwest.
Methods
A previously described cohort of 429 AI/AN was examined. Subjects underwent a baseline screening examination consisting of refraction/visual acuity, limbal anterior chamber depth assessment, frequency doubling technology perimetry (FDT, c-20-5), confocal scanning laser ophthalmoscopy (CSLO, borderline or abnormal Moorfield’s), nonmydriatic digital photography and tonometry. Subjects with at least one abnormal screening test and a subset of those with normal screening were examined by an ophthalmologist for evidence of ocular disease. We used cross tabulation to examine the correlation between screening tests. To test for association between abnormal screening test results and visually significant eye disease, we implemented univariate analysis and multivariate stepwise logistic regression, with correlations between screening tests interpreted in this context. Finally, the predictive values of screening were determined.
Results
Subjects with best corrected visual acuity <20/40 frequently had abnormal FDT (78.4%) and abnormal nonmydriatic photos (58.8%). Abnormal FDT was also highly correlated with abnormal or difficult CSLO and abnormal photos, with 80.8% and 52.5% of the respective CSLO and 71.7% of abnormal photo groups having an abnormal FDT. The overlap in abnormal testing between abnormal or difficult CSLO and abnormal photos was also relatively extensive (66% and 33% respectively). Stepwise logistic regression identified poor vision and abnormal CSLO, FDT and photos as associated with an outcome of visually significant eye disease. Positive and negative predictive values (PPV, NPV) of screening tests were 55% and 87.8% respectively, with NPV increasing to 93.8% when prior history of eye disease was taken into account.
Conclusions
FDT was the most common abnormal screening test, contributing to our PPV as 1/3 of subjects with a positive screening test, but no eye disease on ophthalmic exam, had abnormal FDT. Despite the frequent overlap in abnormal test results between FDT, CSLO and photographic screening, these tests and poor visual acuity all contributed to the final stepwise regression model to predict visually significant eye disease. We conclude a subset of screening tests may be predictive of ocular disease in a prevalence study.
Keywords: 465 clinical (human) or epidemiologic studies: systems/equipment/techniques •
463 clinical (human) or epidemiologic studies: prevalence/incidence