Estimated Number of Adults With Prediabetes in the U.S. in 2000

RESEARCH DESIGN AND METHODS

NHANES III was a survey of the civilian, noninstitutionalized U.S. population that followed a complex sampling design, including oversampling of blacks and Mexican-Americans (13,14). Home interviews, physical examinations, and laboratory examinations were performed on ∼34,000 persons aged ≥2 months. Adults were randomly assigned to either morning or afternoon sessions for physical and laboratory examinations. Only those who participated in the morning session were eligible for the oral glucose tolerance test (OGTT) and fasting glucose test; therefore, we limited our analyses to this population. Of the 6,974 participants aged ≥25 years who were randomly assigned to the morning, 572 reported a previous diagnosis of diabetes when not pregnant and 6,402 did not. After a 9- to 24-h fasting period, a blood sample was obtained from 5,830 (91.1%) of the participants without self-reported diabetes. Of the participants who gave a fasting blood sample, 2,426 were between ages 45 and 74 years, the age group to which OGTT was restricted. After a 2-h (±15 min) 75-g glucose challenge, a second blood sample was obtained from 2,269 (93.5%) participants in that age group.

We assessed other information from the participants, including demographics and risk factors for CVD. Being overweight was defined as having a BMI ≥25 kg/m² (12). Dyslipidemia was defined as an HDL concentration <45 mg/dl in men or <55 mg/dl in women, a total cholesterol concentration ≥200 mg/dl, an LDL concentration ≥100 mg/dl, or a fasting triglyceride concentration ≥200 mg/dl or as having ever been prescribed medication for high cholesterol (15,16). Hypertension was classified as having an elevated blood pressure (systolic ≥140 mmHg or diastolic ≥90 mmHg) or as having ever been prescribed medication for hypertension (17). Microalbuminuria was determined by a urinary albumin concentration ≥30 μg/ml (18). Smoking meant being a current smoker and having smoked ≥100 cigarettes during one’s lifetime.

Aside from adults with self-reported diabetes, all analyses were conducted among overweight adults aged 45–74 years who received both the fasting glucose test and a 2-h glucose challenge. Adults without self-reported diabetes were classified as having undiagnosed diabetes if their 2-h glucose concentration was ≥200 mg/dl and/or their fasting glucose concentration was ≥126 mg/dl (19,20). We used the term “prediabetes” to identify adults without diabetes who had IGT (2-h glucose concentration of 140–199 mg/dl) or IFG (fasting glucose concentration 110–125 mg/dl) (12,19,20). The prevalence of IGT, IFG, prediabetes, and IGT plus IFG was estimated among overweight adults aged 45–74 years. A comparison of people with prediabetes with people meeting DPP criteria for IGT (IGT plus fasting glucose of 95–125 mg/dl) (9) across several demographic and clinical characteristics was also performed. To project 1988–1994 NHANES III estimates to the year 2000, we applied the respective prevalences of self-reported diabetes and being overweight, undiagnosed diabetes and being overweight, and prediabetes and being overweight to year 2000 population estimates (21). SUDAAN software was used to account for the complex sampling design of the survey (22). All analyses accounted for the clustered sampling design and oversampling and were adjusted for differential noncoverage and nonresponse (13,14).

RESULTS

Among overweight adults aged 45–74 years, 12.5% had self-reported diabetes (the corresponding U.S. population projection for 2000 is 6.7 million persons), 10.8% (5.7 million) had undiagnosed diabetes, 22.6% (11.9 million) had prediabetes, 17.1% (9.1 million) had IGT, 11.1% (5.8 million) had IFG, 5.6% (3.0 million) had both IGT and IFG (Table 1), and 54.1% (38.4 million) had normal glucose metabolism. In all, 14.4% or 7.3 million people met DPP IGT criteria. Among the overweight adults aged 45–74 years with prediabetes, >50% had IGT only; <25% had IFG only or IGT and IFG combined (Fig. 1).

The prevalence of impaired metabolism states was similar by sex and by race/ ethnicity, with two exceptions (Table 1). Mexican-American men were more than twice as likely as Mexican-American women to have IFG (16.5 vs. 7.6%), and the prevalence of prediabetes was higher among Mexican-Americans than among non-Hispanic blacks (27.3 vs. 18.9%).

There were a few differences in characteristics across the metabolic states (Table 2). Among overweight adults aged 45–74 years, a higher proportion of men than women had IFG (55.4 vs. 44.6%). Compared with adults with IFG, adults with IGT had a lower mean weight (83.5 vs. 87.3 kg) and a lower mean waist circumference (103.3 vs. 106.1 cm). Among adults with prediabetes, the prevalence of CVD risk factors was high: 94.9% had dyslipidemia, 56.5% had hypertension, 13.9% had microalbuminuria, and 16.6% were current smokers. People with prediabetes had characteristics that were remarkably similar to those with DPP IGT (IGT plus fasting glucose 95–125 mg/dl) across several demographic and clinical variables; the only exception was that people with prediabetes had lower 2-h glucose concentrations (149.7 vs. 164.1 mg/dl).

CONCLUSIONS

Overall, ∼25% of overweight adults aged 45–74 years had prediabetes, which translates into about 12 million persons in the U.S. in the year 2000. Recent controlled trials on diabetes prevention have confirmed that lifestyle changes such as diet, weight loss, and exercise as well as the drug metformin can substantially delay or prevent the progression from impaired metabolism to type 2 diabetes (7–10). Thus, several million individuals could benefit from diabetes prevention intervention. Furthermore, in the process of identifying those with prediabetes, we estimated that an additional 6.5 million persons with undiagnosed diabetes would be detected.

We compared the characteristics of individuals with IGT in the population represented by NHANES III with those of individuals with IGT recruited in the DPP (10,23). As previously reported (23), the DPP sample was younger and had a lower proportion of men, a lower proportion of nonminority groups, a higher mean BMI, lower mean lipid concentrations, a lower proportion of smokers, but roughly similar mean fasting glucose (106.5 vs. 104.2 mg/dl) and mean 2-h glucose concentration (164.9 vs. 161.9 mg/dl). Although the DPP population had a higher proportion of minority groups, the DPP researchers found that lifestyle and metformin interventions worked equally well across all race/ethnic groups (10). Of interest is the fact that people with prediabetes had remarkably similar characteristics to those with IGT in the DPP trial (IGT plus fasting glucose 95–125 mg/dl) across several demographic and clinical variables.

To date, no diabetes prevention trial has been conducted solely among persons with IFG, although each of the recent trials included some persons with IGT who also had IFG. Both conditions indicate prediabetes (12), and the two states overlap somewhat, but are not identical. In our study, 25% of those with prediabetes had both IGT and IFG; yet we found that demographic and CVD risk factor profiles were similar between those with IGT and IFG except for some differences in gender distribution, mean weight, and mean waist circumference. In the Hoorn study (24), the conversion rate from IGT to diabetes and from IFG to diabetes was fairly similar (57.9/1,000 vs. 51.4/1,000 person-years).

The risk of diabetes is strongly linked to contemporary lifestyle, which is heavily influenced by socioeconomic and cultural factors; diabetogenic lifestyles are also ubiquitous (4,5). It is intuitive to argue that a diabetes prevention approach aimed at the entire population rather than one targeted at a high-risk group should be adopted. However, several factors support an aggressive, DPP-like approach targeting high-risk persons (defined by IGT or IFG in overweight adults). For example, in the total population serum glucose concentrations were bimodally distributed (19), and the threshold for diabetes risk coincided with the onset of prediabetes. Using data reported in the Hoorn study (24), we estimated that although 8% of that study population had IGT, 40% of incident cases of diabetes occurred among subjects with IGT, and although 10% had IFG, 42% of incident cases occurred among subjects with IFG. In addition, most, if not all, individuals who develop type 2 diabetes pass through IGT or IFG; this phase offers an opportunity to efficiently identify those at imminent risk of developing diabetes. Moreover, in our analysis, the prevalence of certain CVD risk factors (e.g., dyslipidemia and hypertension) among those with IGT or IFG was very high; lifestyle interventions in this group may help prevent CVD, in addition to delaying or preventing the development of diabetes.

The NHANES III data posed some advantages and disadvantages for our analyses. It was advantageous that the data were nationally representative and had standardized laboratory, clinical, and physical measurements. However, the fact that limited data were available on racial/ethnic groups at high risk for diabetes (e.g., Native Americans), and that no 2-h glucose data from adults ages <40 years and ≥75 years were available was disadvantageous. In addition, we might have underestimated the true size of the U.S. population at risk for diabetes because the proportions of overweight and diabetic adults have increased since the NHANES III was conducted (25). We have estimated that in 2000, among overweight Americans aged 45–74 years, 9.1 million had IGT, 5.8 million had IFG, and 11.9 million had prediabetes.

There is strong evidence for substantially delaying or preventing diabetes among overweight adults with IGT. The ADA has recommended that all overweight people aged ≥45 years with prediabetes are potential candidates for diabetes prevention (12). To what extent other people with abnormal glucose tol-erance (e.g., those with IFG) would respond similarly to subjects that participated in the clinical trials of prevention (7–11) is not clear. In any case, our estimates suggest that very large numbers of people at risk for diabetes may benefit from diabetes prevention interventions; our estimates may also be conservative, because data were not available for all age groups. The immediate challenge is to translate research evidence for diabetes prevention into practice and policy. Questions remain about which groups may benefit and by how much from diabetes interventions, the most efficient means to identify these groups, and the most effective ways to deliver and sustain interventions.