Prevalence of Diabetes, Impaired Fasting Glucose ... - Diabetes Care

3 downloads 0 Views 1MB Size Report
Glucose, and Impaired Glucose Tolerance in U.S. Adults. The Third National Health and Nutrition Examination Survey, 1988-1994. MAUREEN I. HARRIS, PHD, ...
E p i d e m i o I o g y / H e a 11 h O R I G I N A L

Se r v ice s/ P sy ch o soc ia I

R e s e a r c h

A R T I C L E

Prevalence of Diabetes, Impaired Fasting Glucose, and Impaired Glucose Tolerance in U.S. Adults The Third National Health and Nutrition Examination Survey, 1988-1994 MAUREEN I. HARRIS, PHD, MPH KATHERINE M. FLEGAL, PHD CATHERINE C. COWIE, PHD, MPH MARK S. EBERHARDT, PHD

DAVID E. GOLDSTEIN, MD RANDIE R. LITTLE, PHD HSIAO-MEI WIEDMEYER, MS DANITA D . BYRD-HOLT

OBJECTIVE — To evaluate the prevalence and time trends for diagnosed and undiagnosed diabetes, impaired fasting glucose, and impaired glucose tolerance in U.S. adults by age, sex, and race or ethnic group, based on data from the Third National Health and Nutrition Examination Survey, 1988-1994 (NHANES 111) and prior Health and Nutrition Examination Surveys (HANESs). RESEARCH DESIGN A N D M E T H O D S — NHANES III contained a probability sample of 18,825 U.S. adults ^ 2 0 years of age who were interviewed to ascertain a medical history of diagnosed diabetes, a subsample of 6,587 adults for whom fasting plasma glucose values were obtained, and a subsample of 2,844 adults between 40 and 74 years of age who received an oral glucose tolerance test. The Second National Health and Nutrition Examination Survey, 1976-1980, and Hispanic HANES used similar procedures to ascertain diabetes. Prevalence was calculated using the 1997 American Diabetes Association fasting plasma glucose criteria and the 1980-1985 World Health Organization (WHO) oral glucose tolerance test criteria. RESULTS— Prevalence of diagnosed diabetes in 1988-1994 was estimated to be 5.1% for U.S. adults ^ 2 0 years of age (10.2 million people when extrapolated to the 1997 U.S. population). Using American Diabetes Association criteria, the prevalence of undiagnosed diabetes (fasting plasma glucose ^126 mg/dl) was 2.7% (5.4 million), and the prevalence of impaired fasting glucose (110 to 2 0 years of age (81% of all eligible subjects) completed a household interview in which information was obtained about demographic characteristics, race, Hispanic ethnicity, and medical history of diabetes. For the last item, questions were asked about a prior diagnosis of diabetes by a physician; women were also asked whether the diagnosis had been made during pregnancy and whether they had also been told after pregnancy that they had diabetes. Based on this information, 1,503 adults were classified as having previously diagnosed diabetes, including 658 men and 845 women. Of the women, 47 had been told of their diabetes during pregnancy but also had a diagnosis of diabetes made outside of pregnancy. Three subjects were excluded from analysis because of missing information. The remaining 17,319 subjects (8,158 men and 9,161 women) were categorized as not having diagnosed diabetes, including 105 women who appeared to have only gestational diabetes.

Fasting plasma glucose and the oral glucose tolerance test Each household was randomly assigned to either a morning or an afternoon/evening examination session. There were 8,608 people ^ 2 0 years of age who did not have diagnosed diabetes and who were assigned to a morning session, and plasma glucose values were obtained after an overnight fast of 9 to < 2 4 h for 6,587 people (77%). This included 2,798 non-Hispanic whites, 1,753 non-Hispanic blacks, 1,771 Mexican-Americans, and 265 people of other racial or ethnic groups. Plasma glucose values were not obtained for people who did not participate in the examination (n = 811), were exam-

DIABETES CARE, VOLUME 21, NUMBER 4, APRIL 1998

Table 1—Percentage of the U.S. population >20 years of age with physician-diagnosed diabetes, based on NHANES III, 1988-1994 i \ge

Race/sex All racest Both sexes Men Women Non-Hispanic white Both sexes Men Women Non-Hispanic black Both sexes Men

Women Mexican-American Both sexes Men Women

(years) 60-74 >75

20-39

40-49

50-59

1.1 1.1 1.1

3.9 3.3 4.4

8.0 9.6 6.6

12.6 11.8 13.3

1.0 1.2 0.9

3.3 3.0 3.5

7.5 9.9 5.3

1.6 1.6 1.6

6.2

0.8 0.5 1.2

>20

>20*

13.2 13.8 12.8

5.1 4.9 5.4

5.3 5.3 5.2

11.3 10.9 11.7

12.6 13.2 12.3

5.0 5.0 5.0

4.8 5.2 4.5

6.7

13.8 13.0 14.5

20.9 16.8 23.9

17.5 14.7 19.0

6.9 5.9 7.8

8.2 7.3 9.1

7.3 5.5 9.2

16.0 15.5 16.5

24.4 18.8 29.0

21.7 17.8 25.0

5.6 4.2 7.2

9.3 7.7 10.9

5.5

Data are % and are based on a medical history interview in which subjects were asked about prior diagnosis of diabetes by a physician. *Values are age- and sex-standardized. "•'Values include those of racial and ethnic groups not listed separately.

ined at home where fasting was not required (n = 209), were examined in the afternoon/evening (n = 331), had medical and safety reasons for exclusion (n = 10), became faint or ill (n = 2), refused the venipuncture (n = 28), had unsuccessful venipuncture (n = 27), fasted for < 9 h (n = 369) or >24 h (n = 8), or had an unknown fasting time (n = 89), and for other reasons (n = 137). For a subsample of subjects aged 40-74 years, a 75-g glucose-equivalent oral glucose challenge (Dextol or Trutol) was given and a blood sample was drawn 2 h (±15 min) later. The postchallenge glucose value was obtained for 2,844 (93%) of the 3,055 people aged 40-74 years who had no medical history of diabetes and for whom an overnight fasting plasma glucose value was obtained. This group consisted of 1,390 men and 1,454 women; 1,373 of the subjects were non-Hispanic whites, 690 were non-Hispanic blacks, 657 were MexicanAmericans, and 124 were of other racial or ethnic categories. There were no statistically significant differences in a variety of sociodemographic and clinical variables between subjects assigned to the morning session versus subjects assigned to the afternoon/evening session. In addition, for those in the morning session, there were no significant differences between subjects for whom plasma glucose values were obtained versus those for whom plasma glucose val-

ues were not obtained. Thus, data from the subsamples are considered to represent the entire NHANES III sample. The procedures for blood collection and processing have been described (12). Venous whole blood was collected into a vacuum tube containing the glycolytic inhibitors potassium oxalate and sodium fluoride and was centrifuged immediately at 1500g for 10 min. The plasma was frozen at —70°C, shipped on dry ice to the University of Missouri Diabetes Diagnostic Laboratory, and stored at — 70°C until analysis. Plasma glucose was measured using a modified hexokinase enzymatic method. Both withinassay and between-assay quality control procedures were used; the coefficient of variation of the method was 1.6-3.7% during the 6 years of the survey (12). People without previously diagnosed diabetes were categorized according to American Diabetes Association diagnostic criteria (8) as follows: undiagnosed diabetes, fasting plasma glucose >126 mg/dl; impaired fasting glucose, fasting plasma glucose 110 to 20* sample (subjects with diagnosed diabetes) and the plasma glucose samples (subjects Undiagnosed diabetes with undiagnosed diabetes, impaired fasting All racest glucose, or impaired glucose tolerance). 0.6 6.2 5.7 Both sexes 2.7 2.8 2.5 4.6 Prevalence based on fasting and 2-h glucose Men 0.5 8.4 3.0 3.6 7.3 3.1 3.3 in the subsamples of people without diagWomen 0.6 4.7 2.4 5.8 4.5 2.5 1.6 nosed diabetes was adjusted by the prevaNon-Hispanic white lence of diagnosed diabetes, so that the sum Both sexes 0.4 2.1 6.0 2.5 2.5 4.0 4.9 of all diagnostic categories added to the total Men 0.4 8.2 6.0 2.9 2.9 2.9 3.5 U.S. population. To compute the estimated Women 0.4 2.1 4.4 2.0 4.3 4.3 1.3 numbers of people in the diagnostic cateNon-Hispanic black gories in 1997, the NHANES III age-specific 7.7 3.4 6.1 4.9 Both sexes 3.9 3.6 1.4 rates were applied to U.S. Census Bureau Men 2.7 6.6 0.0 4.3 2.6 3.0 1.1 projections of the U.S. population of 1 July Women 1.7 3.7 4.0 8.5 8.5 7.6 4.5 1997 (17,18). Age-adjusted rates were comMexican-American puted by the direct method using the age 3.4 4.9 8.0 Both sexes 5.9 4.5 10.0 1.5 distribution of the 1980 U.S. population Men 5.4 12.9 6.3 10.1 6.8 3.8 1.5 (19). SE was calculated and tests of statisti3.5 6.2 Women 3.0 3.6 7.5 4.9 1.5 cal significance were performed using Impaired fasting glucose SUDAAN, a program that accounts for the All racest nonrandom sample design in computing 2.8 14.0 14.1 Both sexes 7.1 6.9 6.9 8.0 variance estimates. 9.2 16.2 Men 8.7 4.5 17.9 10.1 8.8 Women Non-Hispanic white Both sexes Men Women Non-Hispanic black Both sexes Men Women Mexican-American Both sexes Men Women

1.2

4.3

6.8

12.3

11.9

5.2

5.0

2.7 4.8 0.8

6.7 10.2 3.2

7.7 9.1 6.4

13.9 15.6 12.5

13.7 18.4 11.0

6.9 9.0 5.0

6.8 8.9 4.6

2.8 3.3 2.5

7.0 6.7 7.2

10.0 9.3 10.5

12.1 15.4 9.8

15.7 18.7 14.1

6.2 6.7 5.8

7.0 7.7 6.4

4.4 5.7 2.9

11.8 15.6 7.9

8.6 8.6 8.6

17.1 24.5 10.1

14.7 18.3 11.9

7.3 9.3 5.2

11.6

8.9 6.3

Data are %. Undiagnosed diabetes is defined as fasting plasma glucose ^126 mg/dl ; impaired fasting glucose is defined as fasting plasma glucose 110-125 mg/dl. *Values are age- and sex standardized. TValues include those of racial and ethnic groups not listed separately.

Prior surveys National time trends in the prevalence of diabetes are based on comparisons of data from NHANES III, NHANES II, and Hispanic HANES. Similar procedures were used in these surveys, including interview questions on previous diagnosis of diabetes, measurement of fasting plasma glucose, and administration of an oral glucose challenge to a subsample of participants. In NHANES II, a representative national sample of the U.S. population was examined (6,13). Hispanic HANES was not a national survey, but its Mexican-origin target population

520

included —84% of the 1980 Mexican-origin population of the U.S. (7,14), so comparisons between NHANES III and Hispanic HANES should provide reasonably valid estimates of trends for Mexican-Americans.

Statistical analysis Statistical analyses were carried out using SAS (15) and SUDAAN (16). To provide estimates that were representative of the U.S. population, analyses of each HANES incorporated sampling weights. These weights account for the unequal probabilities of selection resulting from the cluster design,

RESULTS — Prevalence of diabetes diagnosed by a physician before NHANES III is shown in Table 1. Prevalence was 5.1% (95% CI 4.7-5.6) for all people > 2 0 years of age, 5.0% (4.4-5.6) for non-Hispanic whites, 6.9% (6.2-7.7) for non-Hispanic blacks, and 5.6% (5.1-6.2) for MexicanAmericans. Prevalence rates increased with age, although there was a slight decline at age ^ 7 5 years for non-Hispanic blacks and Mexican-Americans. Rates were similar in men and women for non-Hispanic whites but were higher in women for non-Hispanic blacks and Mexican-Americans: the ratio of the age-standardized rate in women vs. men was 0.87 for non-Hispanic whites (NS), 1.25 for non-Hispanic blacks (P = 0.04), 1.41 for Mexican-Americans (P = 0.003), and 0.98 for all races combined (NS). Prevalence was lower in non-Hispanic whites: the age- and sex-standardized prevalence ratio for nonHispanic whites vs. non-Hispanic blacks was 1.70 (P < 0.0001) and vs. Mexican-Americans was 1.92 (P < 0.0001). The difference between non-Hispanic blacks and MexicanAmericans was not statistically significant. When the rates in Table 1 were applied to U.S. population projections for 1997, the number of people ^ 2 0 years of age who have diagnosed diabetes was estimated to be 7.5 million for non-Hispanic whites, 1.5

DIABETES CARE, VOLUME 21, NUMBER 4, APRIL 1998

Harris and Associates

and U.S. population projections for 1997, the number of people ^ 2 0 years of age who have diabetes (either diagnosed or 21 1 undiagnosed) was estimated to be 7.5 mil20.2 lion for men, 8.1 million for women, 11.3 20 17.8 million for non-Hispanic whites, 2.3 mil17 5 c lion for non-Hispanic blacks, 1.2 million o for Mexican-Americans, and 15.6 million CO for the total population. 315 Q. 12.9 The prevalence of impaired fasting gluO 12.4 Q. cose is shown in Table 2. Rates increased from age 20-39 years to age 60-74 years or CD age ^ 7 5 years. Prevalence was higher for 8.4 O 7.7 CD men than for women in each racial or eth6.8 Q_ nic group, and the ratio of the age-standardized rate in men vs. women was 1.8 for 5" all races combined. The age- and sex-standardized prevalence for Mexican-Ameri1.6 1.7 cans (8.9%) was somewhat higher than the rates for non-Hispanic whites (6.8%) and Men Women Men Women Men Women Men Women Men Women Men Women non-Hispanic blacks (7.0%). When the 50-59 40-49 60-74 75 + 20-39 age-std rates in Table 2 were applied to U.S. popuSex and Age (years) lation projections for 1997, the numbers of people ^ 2 0 years of age who have Figure 1—Prevalence of diabetes in men and women in the U.S. population age ^20 years, based on impaired fasting glucose were estimated to NHANES 111. Diabetes includes previously diagnosed and undiagnosed diabetes defined by fasting be 10.3 million for non-Hispanic whites, plasma glucose ^.126 mgldl. age-std, age-standardized. 1.3 million for non-Hispanic blacks, 0.9 million for Mexican-Americans, 8.1 million million for non-Hispanic blacks, 0.8 million ^75 years. Prevalence was similar for men for men, 5.3 million for women, and 13.4 for Mexican-Americans, 4.6 million for and women in each age group, and the age- million for the total population. men, 5.6 million for women, and 10.2 mil- standardized rates were 8.4 and 7.7%, Figure 2 compares prevalence in nonlion for the total population. respectively. Based on the prevalence rates Hispanic whites, non-Hispanic blacks, and The prevalence of undiagnosed diabetes using American Diabetes Association fasting plasma glucose criteria is shown in Table 2. Prevalence increased with age, although for some race/sex groups, there was a decline in the rate for those ^75 years of age. The agestandardized prevalence was slightly higher in men than in women for non-Hispanic whites and Mexican-Americans, but was lower in men than in women for non-Hispanic blacks. The age- and sex-standardized rate for non-Hispanic whites (2.5%) was somewhat lower than for non-Hispanic blacks (3.6%) and Mexican-Americans (4.5%). When the rates in Table 2 were applied to U.S. population projections for 1997, the number of people ^20 years of age who have undiagnosed diabetes was estimated to be 3.8 million for non-Hispanic whites, 0.7 million for non-Hispanic blacks, 0.4 million for Mexican Americans, 2.9 million for men, 2.5 million for women, Non-Hispanic Non-Hispanic white black and 5.4 million for the total population. Men Women Total prevalence of diabetes based on American Diabetes Association criteria is Figure 2—Age-standardized prevalence of diagnosed and undiagnosed diabetes and impaired fasting shown in Figure 1. Prevalence rates rose glucose in the U.S. population >20 years of age, presented according to sex and racial or ethnic group, from 1-2% at ages 20-39 years to 18-20% based on NHANES III Diagnosed diabetes was defined by medical history, undiagnosed diabetes by fastat ages 60-74 years and plateaued at ages ing plasma glucose 2:126 mg/dl, and impaired fasting glucose by fasting plasma glucose 110-125 mg/dl. 25 -i

n-

DIABETES CARE, VOLUME 21, NUMBER 4, APRIL 1998

521

U.S. diabetes prevalence

Table 3—Percentage of the U.S. population 40-74 years of age with undiagnosed diabetes and for a decline at 60-74 years of age for nonHispanic blacks and Mexican-Americans. impaired glucose tolerance, based on 1980-1985 World Health Organization fasting and Rates were somewhat greater for men than postchallenge plasma glucose diagnostic criteria and NHANES III (1988-1994) for women in each racial or ethnic group. The age- and sex-standardized prevalence Age (years) rate was higher in Mexican-Americans 50-59 60-74 40-74 Race/sex 40-49 40-74* (11.4%) than non-Hispanic whites (6.4%) and non-Hispanic blacks (7.5%). The prevaUndiagnosed diabetes lence of impaired glucose tolerance (Table 3) All races t increased with age for almost all sex and 6.4 Both sexes 2.3 6.7 10.8 6.3 racial or ethnic groups. The rates were simMen 12.4 7.1 6.8 2.9 7.6 ilar for men and women in each racial or Women 6.0 9.5 5.7 1.8 5.9 ethnic group, and the age- and sex-stanNon-Hispanic white dardized rate was higher for Mexican-Amer6.1 Both sexes 1.9 5.5 11.0 6.4 icans (20.2%) than for non-Hispanic whites Men 11.8 6.7 2.3 6.8 7.2 (15.3%) and non-Hispanic blacks (14.0%). 4.3 10.4 Women 1.5 5.6 5.6 Non-Hispanic black Both sexes Men Women Mexican-American Both sexes Men Women Impaired glucose tolerance All racest Both sexes Men Women Non-Hispanic white Both sexes Men

Women Non-Hispanic black Both sexes Men Women Mexican-American Both sexes Men Women

4.0 5.8 2.9

9.1 10.4 8.1

8.6 9.7 7.8

6.7 8.3 5.6

7.5 8.7 6.3

6.2 5.8 6.6

16.2 16.4 16.0

11.6 14.3 9.0

9.9 10.3 9.6

11.4 12.3 10.5

11.9 12.6 11.2

14.3 13.3 15.3

20.7 19.2 21.9

15.6 15.0 16.2

15.8 15.2 16.4

11.1 11.7 10.5

13.3 11.4 15.0

20.9 19.5 22.1

15.3 14.3 16.2

15.3 14.4 16.1

9.4 9.4 9.4

14.2 15.7 12.9

17.6 19.4 16.2

13.1 14.3 12.3

14.0 15.0 13.0

17.1 18.6 15.6

20.3 13.0 26.0

23.5 22.5 24.9

19.4 18.3 20.4

20.2 18.2 22.3

Data are %. Undiagnosed diabetes is defined as fasting plasma glucose ^140 mg/dl or 2-h postchallenge glucose ^200 mg/dl; impaired glucose tolerance is defined as fasting plasma glucose