NHANES is a cross-sectional survey of the health and nutrition of the noninstitutionalized, household-dwelling U.S. population conducted by the National Center for Health Statistics and by the Centers for Disease Control and Prevention (¹⁰,¹¹). The survey consists of two components: the in-home interview and the mobile exam center, which performs several laboratory tests including the fasting plasma glucose (FPG) test and the 2-hr oral glucose tolerance test (OGTT). The in-home survey collects demographic and clinical information, including the subject's age, race, sex, medical history, therapy, and lifestyle variables such as the frequency and duration of exercise and dietary habits.

NHANES has collected data in 2-year cohorts since 1999. The 2005–2006 sample from NHANES IV, the most recent publicly available version, was used in this study. We restricted our analyses to the subjects in the subsample that had morning examinations since they were the only subjects who had valid FPG and OGTT testing. We also excluded from our analyses subjects aged <18 years, those with diagnosed or undiagnosed diabetes, and those with a history of myocardial infarction (MI).

The current ADA definition for IFG is blood glucose 100–125 mg/dl after an 8-h fast (¹²). When the IFG category was initially introduced by the ADA, IFG was defined as 110–125 mg/dl after fasting. In 2003, the ADA lowered the threshold of IFG to better capture those who met the World Health Organization's (WHO) criteria for IGT. WHO defines IGT as a glucose level of 140–199 mg/dl 2 h after a glucose load (¹³). We compared the estimates of the prevalence of pre-diabetes using the old and the new fasting blood glucose criterion.

Subjects were asked if their physicians have ever told them that they have “borderline diabetes, pre-diabetes, impaired fasting glucose, or impaired glucose tolerance.” A prior diagnosis of pre-diabetes was considered to be a “yes” to any of the four terms.

The ADA recommends metformin alone as the antihyperglycemic of choice based upon the DPP results. However, because NHANES does not indicate the class of antihyperglycemic used for IFG/IGT, we could only determine whether any antihyperglycemic medication was given. Lifestyle modification included either exercise, diet modification, or both. Subjects who averaged at least 30 min of vigorous or moderate activity daily for the previous 30 days were considered “compliant” with ADA recommendations for treatment of IFG/IGT (⁴). Vigorous activity was defined as activity that causes “heavy sweating or large increases in breathing or heart rate.” Moderate activity was defined as activity that causes “moderate sweating or slight to moderate increases in breathing or heart rate.” Subjects were asked if they maintained their activity levels over the last year relative to the last 30 days. Provider recommendation for diet modification included reporting either counseling to reduce weight or counseling to reduce fat/calorie intake, or both. NHANES did not indicate the chronological order of the physician recommendations and the actual change in exercise or diet, so it was not possible to determine if lifestyle behaviors changed in response to the recommendations.

The associations between subject demographics and medical conditions and the presence of pre-diabetes were examined using the χ² statistic for categorical predictor variables and logit modeling for continuous predictor variables. Sampling weights were used to provide estimates that are representative of the U.S. population; all percentages presented are weighted. Most variables analyzed had little missing data (<4%), but 16.7% of subjects were missing data for either income, education, or both. Therefore, multiple imputation was used to more accurately account for the high level of missing data for all analyses involving these two variables (¹⁴). Multivariable logistic regression was used to examine independent predictors of treatment and adherence to lifestyle modification therapies. Independent variables in the model included sex, race, education, insurance status, and the percent of poverty level. All statistical analyses were performed using the STATA software (version 10.0; StataCorp LP, College Station, TX).

The NHANES 2005–2006 cohort included 2,425 subjects aged 18 years or older in the morning examination (fasting) sample. From this sample, 878 were excluded from our analysis: 87 because they had had a prior MI (since we were interested in pre-diabetes and diabetes as coronary artery disease risk factors); 659 because they had missing OGTT data; 16 because they had diagnosed diabetes; and 116 because they had undiagnosed diabetes (based on FPG and/or OGTT testing). Subjects who were on oral medications or receiving insulin for their diagnosed diabetes were excluded from OGTT testing; hence, only 16 subjects were excluded for diagnosed diabetes and 116 for undiagnosed diabetes. These exclusions resulted in a sample of 1,547 subjects in the study. By applying sample weights to make the results representative of the nondiabetic U.S. population without a history of MI, we estimate that about 34.6% (CI 30.3–38.9%) of nondiabetic U.S. adults had pre-diabetes. Of the pre-diabetic subjects, 84% met IFG criteria, 44% met IGT criteria, and 28% met both. Using the 1997 ADA criteria for IFG (110–125 mg/dl) resulted in 43% fewer subjects meeting the criteria for IFG, reducing the estimate of the prevalence of pre-diabetes in nondiabetic U.S. adults from 34.6 to 19.8% (Table 1).

The subjects' demographic characteristics and medical conditions are reported in Table 2. Those with pre-diabetes had substantially higher cardiovascular risk, with a mean Framingham 10-year risk for cardiovascular events of 8.5% (CI 6.0–10.6%), which was almost twice that of normoglycemic subjects (5.2% [CI 3.9–6.4%], P < 0.001) (¹⁵,¹⁶).

Only 3.4% of the entire study sample reported a prior diagnosis of “impaired fasting glucose, impaired glucose tolerance, borderline diabetes, or pre-diabetes” (Table 2). Of those reporting a diagnosis, 38.5% no longer met the pre-diabetes criteria (either due to resolution or misdiagnosis); 61.5% had unresolved pre-diabetes. No diagnosed pre-diabetic subjects reported receiving oral antihyperglycemic medications (CI 0–10.8%). Multivariable analysis found that subjects who had pre-diabetes tended to be older, male, and Mexican American (Table 3).

Of pre-diabetic subjects, 31.7% (CI 23.3–40.2%) reported receiving counseling for exercise, 33.4% (CI 26.4–40.5%) for diet, and 25.9% (CI 17.9–34.5%) for both (Table 4). Of those who reported exercising, only about half reported achieving the ADA IFG/IGT guidelines of at least 30 min daily. Rates of recommendations for and practice of diabetes prevention behaviors were similar when the 1997 ADA criteria for IFG (FPG of 110–125 mg/dl) were applied.