Colditz and colleagues [7] used the Nurse's Health Study cohort of 114 824 women to quantify the relations between change in adult weight and the risk for non-insulin-dependent diabetes during a 14-year follow-up. They observed that the risk for developing diabetes increased monotonically with levels of body mass index greater than 22 kg/m². They also noted that weight gain after age 18 years was strongly related to risk, as was weight gain during middle age; the relative risks for the latter were weaker. In contrast, women who lost more than 5 kg from early adult life had a significantly reduced risk for diabetes. Indeed, the risk for diabetes was substantially increased even at a body mass index of 25 kg/m², a level well below the usual criterion for a diagnosis of obesity [4-6]. Colditz and colleagues concluded that the risk for diabetes with even modest typical adult weight gain is substantial, and they stressed the importance of maintaining a constant body weight throughout adult life. This recommendation differs from the 1990 U.S. Department of Agriculture guidelines, which imply that substantial weight gain at age 35 years and a body mass index as high as 27 kg/m² are healthy [6].

The results of Colditz and colleagues' large prospective study are in agreement with those of smaller studies [8-10] that showed a relation between weight gain and risk for diabetes in adolescents [8] and adults [8-10]. This increased risk is therefore not unique to women. In another relatively large study [10], men and women who gained 10 pounds or more between ages 40 and 60 years had a similar associated risk for developing diabetes. Indeed, in this study, weight gain after age 18 years was still associated with an increased risk for diabetes after the investigators controlled for weight at age 18 years. The relative risk for diabetes was 1.4 for every 17% to 31% increase in body weight after age 18 years [10]. These studies emphasize the continuum that exists between weight gain in adulthood and the risk for developing non-insulin-dependent diabetes mellitus in both men and women.

Colditz and colleagues [7] used body mass index as a measure of adiposity. As the authors acknowledge, body mass index is an imperfect measure of obesity, and the relatively strong relation between the index and the development of non-insulin-dependent diabetes mellitus may be an underestimate. Indeed, persons with a preponderance of abdominal fat tend to be insulin-resistant and are especially prone to develop non-insulin-dependent diabetes mellitus [4, 11-13]. This may relate to the metabolic characteristics of abdominal adipocytes, which are more resistant to the actions of insulin and more sensitive to lipolytic hormones than adipocytes from the gluteo-femoral region [14, 15]. Increased release of free fatty acids into the portal system could impair first-pass metabolism of insulin. Persons who are prone to develop non-insulin-dependent diabetes mellitus and are not overweight but have "excessive body fat content" should also be considered. Reaven [15] has suggested that as many as 25% of nonobese adult persons have insulin resistance that may be related to excessive relative body fat content [16]. Although use of the waist-to-hip ratio and determinations of skinfold thickness are better gauges of relative adiposity, computed tomographic scanning of the abdomen is the current gold standard. These scanning techniques have clearly shown that visceral fat is most closely associated with insulin resistance, dyslipidemia, and non-insulin-dependent diabetes mellitus [4, 11-13]. Thus, future prospective studies of the relation between adiposity and the development of non-insulin-dependent diabetes mellitus should determine fat distribution and relative adiposity as well as body mass index.

Colditz and colleagues [7] did not examine changes in physical activity levels, which are an important determinant of insulin sensitivity [17]. Indeed, improved tissue insulin sensitivity has been shown in obese persons whose maximum oxygen consumption significantly increased during physical training, even with insignificant changes in body weight and fat composition [17]. Thus, age-related decreases in physical activity may also have played a role in the association of weight gain and the development of non-insulin-dependent diabetes mellitus. Persons who assume a sedentary lifestyle during adulthood are probably more likely to gain weight, particularly in the form of adipose tissue [16]. This interaction is particularly relevant because obesity and physical inactivity are both preventable and reversible.

Another limitation of Colditz and colleagues' study is the racial constitution of the study population—98% of the participants were white. Indeed, considerable evidence suggests that black and Hispanic women are more likely than white women to develop non-insulin-dependent diabetes mellitus [2, 3, 18, 19]. Race- and sex-specific data from the Coronary Artery Risk Factor study (CARDIA), an ongoing longitudinal study of a biracial cohort of more than 5000 young adults [19], have shown that black women have the highest insulin levels, which suggests a predisposition to non-insulin-dependent diabetes mellitus. This predisposition may be related to the higher prevalence of obesity in this cohort, particularly visceral obesity [18, 19]. In the San Antonio Heart Study [19], young Hispanic persons were found to have decreased insulin sensitivity, hyperinsulinemia, and a greater prevalence of non-insulin-dependent diabetes mellitus in adulthood. These observations suggest that the relation between obesity and subsequent development of non-insulin-dependent diabetes mellitus shown in Colditz and colleagues' study may have been even more striking had minority populations been represented in proportion to their representation in the population of the United States.