We are pleased to respond to the letter by Orchard et al. (1), especially since they first raised the following question: Do composite measures of chronic glycemia correlate or predict complications better than individual components? Orchard et al. reported evidence against the hypothesis, while we (2) reported evidence for the hypothesis. Having considered their suggestions, we offer an explanation for why their conclusions differed from ours.

Orchard et al. (3) compared the fit from two models, one consisting of only the composite and the other consisting of a regression model that included both components. The regression model is a linear combination of the two components in which the weights are chosen to obtain an optimal fit; thus, the regression model itself is a composite, though one in which the fit to the data should be better than A1 months (which is exactly what they found).

Since comparing two composites was not the goal of our study (2), we approached the analyses differently. We developed one regression model including all variables that were significant in the multivariate modeling, including the composite as well as individual components, as candidates for the model. Each partial R2 measures the explanatory value of the corresponding variable beyond the prediction already available from all the other variables in the model. Except for severity of retinopathy at baseline, we found that the composite was consistently the best predictor and that the individual components added little, if anything.

We agree that age at onset and duration added together equal the age of the patient at the time of study, although the appropriate weights for these two time periods in predicting the outcome may differ, and determining whether the weights significantly differ would be of interest. However, this was not a focus of our study.

We also agree that the patient population under study and the choice of outcomes to be analyzed can influence the results and that a continuous neuropathy measure is desirable. Although use of a common outcome measure would assist in comparing our results with those of Orchard et al. (3), such a comparison was not the focus of our study (2). Finally, determining the threshold of chronic glycemia, which induces complications, is a worthy goal, but before we do this we want to include studies of normal subjects and glucose-impaired individuals currently being studied.

Orchard TJ, Costacou T, Miller RG, Prince CT, Pambianco G: Modeling chronic glycemic exposure variables as correlates and predictors of microvascular complications of diabetes (Letter).
Diabetes Care
Dyck PJ, Davies JL, Clark VM, Litchy WJ, Dyck PJB, Klein CJ, Rizza RA, Pach JM, Klein R, Larson TS, Melton LJ III, O’Brien PC: Modeling chronic glycemic exposure variables as correlates and predictors of microvascular complications of diabetes.
Diabetes Care
Orchard TJ, Forrest KY, Ellis D, Becker DJ: Cumulative glycemic exposure and microvascular complications in insulin-dependent diabetes mellitus: the glycemic threshold revisited.
Arch Intern Med