Effect of intensive therapy on the development and progression of diabetic nephropathy in the Diabetes Control and Complications Trial

A cause-and-effect relation between blood glucose concentrations and microvascular complications in patients with insulin-dependent diabetes mellitus has not been established. We randomly assigned 102 patients with insulin-dependent diabetes mellitus, nonproliferative retinopathy, normal serum creatinine concentrations, and unsatisfactory blood glucose control to intensified insulin treatment (48 patients) or standard insulin treatment (54 patients). We then evaluated them for microvascular complications after 18 months and 3, 5, and 7.5 years. Mean (+/- SD) glycosylated hemoglobin values were reduced from 9.5 +/- 1.3 percent to 7.1 +/- 0.7 percent in the group receiving intensified treatment and from 9.4 +/- 1.4 percent to 8.5 +/- 0.7 percent in the group receiving standard treatment (P = 0.001). In 12 of the patients receiving intensified treatment (27 percent of those included in the analysis) and 27 of those receiving standard treatment (52 percent), serious retinopathy requiring photocoagulation developed (P = 0.01). Visual acuity decreased in 6 patients receiving intensified treatment (14 percent) and in 18 receiving standard treatment (35 percent) (P = 0.02). Nephropathy (urinary albumin excretion, > 200 micrograms per minute) developed in one patient in the group receiving intensified treatment, as compared with nine patients in the group receiving standard treatment (P = 0.01). No patient in the intensified-treatment group had nephropathy with subnormal glomerular filtration rates, as compared with six patients in the standard-treatment group (P = 0.02). The conduction velocities of the ulnar, tibial, peroneal, and sural nerves decreased significantly more in the standard-treatment group than in the intensified-treatment group. The odds ratio for serious retinopathy was 0.4 (95 percent confidence interval, 0.2 to 1.0; P = 0.04) in the intensified-treatment group as compared with the standard-treatment group. The corresponding odds ratio for nephropathy was 0.1 (95 percent confidence interval, 0 to 0.8; P = 0.04). Long-term intensified insulin treatment, as compared with standard treatment, retards the development of microvascular complications in patients with insulin-dependent diabetes mellitus.

We studied whether microalbuminuria (urinary albumin excretion rates of 15 to 150 micrograms per minute) would predict the development of increased proteinuria in Type I diabetes. We also studied the influence of glomerular filtration rate, renal blood flow, and blood pressure on the later development of proteinuria. Forty-four patients who had had Type I diabetes for at least seven years and who had albumin excretion rates below 150 micrograms per minute were studied from 1969 to 1976, and 43 were restudied in 1983. Of the 14 who initially had albumin excretion rates at or above 15 micrograms per minute, 12 had clinically detectable proteinuria (over 500 mg of protein per 24 hours) or an albumin excretion rate above 150 micrograms per minute at the later examination. Of the 29 who initially had albumin excretion rates below 15 micrograms per minute, none had clinically detectable proteinuria at the later examination, although four had microalbuminuria. Those whose condition progressed to clinically overt proteinuria had elevated glomerular filtration rates and higher blood pressures at the initial examination than did those in whom proteinuria did not develop. Renal blood flow was not elevated in these patients. We conclude that microalbuminuria predicts the development of diabetic nephropathy and that elevated glomerular filtration rates and increased blood pressure may also contribute to this progression.

The question of how to analyze unbalanced or incomplete repeated-measures data is a common problem facing analysts. We address this problem through maximum likelihood analysis using a general linear model for expected responses and arbitrary structural models for the within-subject covariances. Models that can be fit include standard univariate and multivariate models with incomplete data, random-effects models, and models with time-series and factor-analytic error structures. We describe Newton-Raphson and Fisher scoring algorithms for computing maximum likelihood estimates, and generalized EM algorithms for computing restricted and unrestricted maximum likelihood estimates. An example fitting several models to a set of growth data is included.