1993 Volume 36 Issue 4 Pages 261-268
This study proposes a method of categorizing clinical findings which can be used for selecting variables in a multivariate analysis of risk factors for diabetic retinopathy. Eye ground findings and 19 other clinical parameters at the initial visit were analyzed in 1, 200 non-insulin-dependent diabetics seen at the Third Department of Internal Medicine, Tokyo University.
Based on correlations between these findings and associations with retinopathy, the 19 clinical findings were classified into the following 7 categories: “gender”, “age” and “duration”, epidemiological components: “obesity” represented by body mass index; “hyperglycemia”, represented by plasma glucose level; “hypertension” represented by blood pressure;and “atherosclerosis”, represented by cholesterol. Duration of diabetes, systolic blood pressure, initial body mass index (inverse) and plasma glucose level at 120 min. on GTT were identified as independent risk factors for diabetic retinopathy by multivariate analysis using the 7 major variables representing each category.
Our method, which selects variables from categorized clinical findings, is valid since it takes into account the complexity of interrelated factors among the miscellaneous clinical findings noted at the initial visit.
