Risk factors for development of proteinuria in NIDDM analyzed by Poisson regression

To clarify the time dependency of risk factors for the development of diabetic nephropathy, we applied Poisson regression to the analysis of 7167 person-year data in 1447 patients with non-insulin-dependent diabetes mellitus (NIDDM) who were initially free of proteinuria. Significant predictors were found to be annual mean fasting blood glucose (FBG) level, male gender, duration, and age at diagnosis. Hyperglycemia was more influential, while duration was less in the previous year of development of proteinuria than at the initial visit. When more information during longer-year data was used as average, the contribution of FBG level was enhanced. Current age was less associated than was age at diagnosis. Thus, Poisson regression seems to be useful for the analysis of risk variables in chronic diseases.