Purpose: To examine factors necessitating pupil dilation to achieve gradable diabetic screening photographs using a digital non-mydriatic camera and to establish techniques to predict the need for dilation and to validate them.
Methods: Prospective clinic-based cross-sectional study with follow-up validation study. The participants' involved consecutive patients attending the diabetic retinopathy screening clinic at a University Hospital. Best corrected visual acuity, age, sex, pupil size, mean spherical equivalent, cataract grade and the requirement for dilation to achieve gradable photographs in 90 patients were recorded. Data analysis using principal component analysis and multivariate analysis of variance derived a set of equations to predict the requirement for dilation. The predictive powers of these equations were validated in an independent group of 51 patients.
Results: Smaller pupil size, denser nuclear colour, older age, poorer best-corrected visual acuity, cortical lens opacity and posterior subcapsular lens opacity were associated with the need for dilation (P<0.001 in all). Single variables used in isolation had a poorer predictive value than combining variables. Dilating patients with either a pupil size>3.75 mm or age>59 years correctly allocates 83 and 78% of patients, respectively to dilation or not. Combining pupil size with age produces a decision table that improves the predictive value to 84%. In the validation study this table had a predictive value of 80%.
Conclusion: We have produced and validated criteria based on a range of clinical variables for application in a clinical setting that allows for the development of targeted mydriasis.