In patients with type 1 diabetes, the debate over the role of careful glycemic control in the evolution of complications has ended, thanks to the Diabetes Control and Complications Trial (DCCT) [1]. The DCCT is the most important clinical study ever conducted in the field of diabetes. This multicenter randomized, controlled clinical trial showed that intensive treatment of type 1 diabetes, with the goal of meticulous glycemic control, dramatically decreases the frequency and severity of retinopathy, nephropathy, and neuropathy. Yet, the extent to which the DCCT results apply to patients with type 2 diabetes has been questioned. The arguments that it does apply are three.

First, extensive epidemiologic studies (as exemplified by the Wisconsin study [2, 3], with its 10 years of follow-up) have shown a strong and consistent relation between glycemia and the incidence and progression of microvascular (diabetic retinopathy, loss of vision, and nephropathy) and macrovascular (amputation and death from cardiovascular disease) complications in patients with type 1 or type 2 diabetes [2, 3]. The similar outcomes in type 1 and type 2 diabetes suggest that glycemia plays a similar role in both conditions.

Second, the DCCT did not find a "glycemic threshold." Rather, there was a continuous relation between glycemic exposure and risk for complications [1, 4-6]. This finding supports the "glucose hypothesis": that overall glycemic exposure drives the biochemical pathways that result in diabetic complications. The pathway should be operative in both types of diabetes.

Third, a small study that was similar in design to the DCCT but involved patients with type 2 diabetes has been performed in Japan [7]. The findings of this study were similar to those of the DCCT in terms of reduction of risk for complications.

For these reasons, the American Diabetes Association Standards of Care recommend striving for meticulous glycemic control in type 1 and type 2 diabetes [7]. However, extensive clinical trial data on type 2 diabetes are lacking. Although it is expected that additional data may emerge from the United Kingdom Prospective Diabetes Study (scheduled for completion in 1998) [8], the experimental design of this study is sufficiently complicated that an unambiguous answer may not be forthcoming.

The absence of clinical trial data has led investigators to other approaches in an attempt to gain further insight into the impact of interventions in type 2 diabetes. One approach is the development of computerized simulation models. Earlier this year, Eastman and colleagues [9] reported the development of a descriptive natural history model examining the multiplicity of complications in type 2 diabetes. This model includes evolution of retinopathy, legal blindness, microalbuminuria, proteinuria, end-stage renal disease, neuropathy, lower-extremity amputation, cardiovascular disease, and death. These investigators also developed an interventional model examining the effect of meticulous glycemic control on development of these complications in patients with type 2 diabetes in terms of health benefits and economics [10]. They found that glycemic control that maintains a hemoglobin A_1c value of 7.2% is predicted to reduce the cumulative incidence of blindness, end-stage renal disease, and lower-extremity amputation by 72%, 87%, and 67%, respectively; the incremental costs are in the range of the costs of interventions that are generally considered cost-effective.

In this issue, Vijan and colleagues [11] describe a simulation model used to estimate the benefits of glycemic control on retinopathy and nephropathy in type 2 diabetes [11]. They focus on the differential benefits (in terms of reduction in years of blindness or end-stage renal disease) in cohorts of different ages at diabetes onset. They note that the greatest benefit is seen in cohorts with a younger age at diabetes onset (40s and 50s); in contrast, much less benefit is seen in cohorts with an older age at onset (60s and 70s). The authors assert that targeting younger patients for better glycemic control may be a better use of health care resources. Yet, it must be appreciated that their model is confined to potential benefits only for the eyes and the kidneys.

It is clearly true that longer glycemic exposure confers greater risk and, therefore, more opportunity for risk reduction. Indeed, secondary prevention of chronic disease complications that depend on duration of exposure to a risk factor (such as blood pressure, glycemia, and cholesterol) is inherently less cost-effective in older persons because duration of exposure is shorter and there is less opportunity to prevent complications. Event rates for blindness and end-stage renal disease are low among persons developing type 2 diabetes in the seventh and eighth decades of life. Eastman and colleagues also noted that treatment may be more cost-effective for patients with longer glycemic exposure (earlier onset of diabetes) and those with higher baseline hemoglobin A_1c values while receiving standard care [10]. However, it should be noted that both models consider age at onset and do not consider effects of treatment of prevalent diabetes in older patients, many of whom already may have retinopathy, nephropathy, or neuropathy amenable to improved control.

Moreover, neither Eastman and colleagues' analyses nor Vijan and associates' study include potential reduction of nonmedical costs as a consequence of improved glycemic control. Testa and Simonson [12] recently used clinical trial data to show short-term economic benefits of improved glycemic control. They found that improved glycemic control in type 2 diabetes enhances employment retention and work productivity and reduces absenteeism, bed days, restricted-activity days, and frequency of physician visits. A preliminary analysis indicated that when these savings are factored into Eastman and colleagues' model, a real cost savings is seen (Eastman RC. Personal communication). In Testa and Simonson's study, the cost of office visits not related to diabetes was reduced in patients with good control [12]. This may be important because a recent study found that in the Kaiser system, 60% of the excess cost of caring for diabetic patients was due to the cost of non-diabetes-related illness [13]. These costs are not considered in the models of Vijan and Eastman and their colleagues. Thus, these models present worst-case analyses, as is appropriate for these types of analyses. They establish the upper limit of cost-effectiveness but do not include all factors that might make treatment more cost-effective.

The use of simulation models provides useful insights into disease evolution and the potential impact of intervention. However, it must be appreciated that models are projections. Although they are based on data, models do not provide the firm evidence gained from controlled clinical trials. These types of theoretical constructs need further validation before they can be used as a basis for allocation of health care resources. Nonetheless, one still might infer from the models of Vijan and Eastman and their colleagues that the care of the elderly patient with type 2 diabetes need not be as aggressive as that of younger patients. For example, if the choice in a 77-year-old patient with a 3-year duration of diabetes is instituting insulin therapy or accepting a hemoglobin A_1c level of 8.3%, one might choose the latter rather than universally taking action with a hemoglobin A_1c level greater than 8% and aiming for a level of 7%, as recommended by the American Diabetes Association [7]. In the end, the practice of medicine is founded on a compilation of information synthesized and distilled from evidence-based clinical trials (which receive the most weight), epidemiologic and observational studies, projections based on pathophysiology or simulation models, and old-fashioned clinical judgment.

There is no doubt that the existing studies show a clear relation between glycemia and diabetic complications in type 2 diabetes. The data suggest that both microvascular and macrovascular complications are influenced by prevailing glycemia. Thus, glycemic control is important in type 2, too. It is now incumbent upon us to design better, safer, and cost-effective treatment programs aimed at meticulous glycemic control.