- Email: [email protected]
Trends in the Prevalence of Diabetes Among U.S. Adults: 1999–2016
ARTICLE IN PRESS
RESEARCH ARTICLE
Trends in the Prevalence of Diabetes Among U.S. Adults: 1999 2016 D1X XMichael Fang, D2X XMA
Introduction: The prevalence of diabetes has increased substantially over the past three decades. This study sought to estimate recent trends in the prevalence of diabetes among U.S. adults.
Methods: This paper estimated trends in the prevalence of diagnosed, undiagnosed, and total diabetes among U.S. adults from 1999 2000 to 2015 2016 (analyzed in 2017). Data come from 42,554 respondents aged 20 years who participated in the National Health and Nutrition Examination Survey. Diagnosed diabetes was measured through self-report, undiagnosed diabetes was measured as never being diagnosed with diabetes but having glycated hemoglobin levels 6.5%, and total diabetes was measured as the sum of individuals with diagnosed and undiagnosed diabetes. Results: In the overall U.S. adult population, the unadjusted prevalence of total diabetes increased from 7.7% in 1999 2000 to 13.3% in 2015 2016 (p<0.001 for trend). Growth was observed for all subgroups, though the rate of change was higher in older adults, racial minorities, and those who were obese compared with their peers. Increasing prevalence among Mexican-American adults was particularly pronounced, rising by 10.1 percentage points during the study period (8.3% to 18.4%, p < 0.001). Roughly 40% of the increase in total diabetes was accounted for by changes in the age and rates of obesity in the U.S. population.
Conclusions: From 1999 to 2016, the prevalence of diabetes among U.S. adults increased at a substantial rate. This growth occurred differentially across subgroups, particularly impacting MexicanAmerican adults, and was driven in large part by population aging and increasing obesity rates. Am J Prev Med 2018;000(000):1 9. © 2018 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.
TAGEDH1INTRODUCTIONTAGEDN
D
iabetes is a group of metabolic disorders that stems from problems in insulin production or action and leads to hyperglycemia.1 Diabetes is associated with numerous negative health consequences, including heart disease, kidney failure, blindness, and premature death.2 7 The condition also carries a substantial economic burden, costing the U.S. an estimated $322 billion in lost productivity and medical costs in 2012.8 Within the U.S., the prevalence of diabetes has increased substantially over the past three decades among adults.3,9 13 Between 1988 and 2012, rates of total diabetes (diagnosed and undiagnosed) increased from 8.6 to 12.3 cases per 100 U.S. adults.9 The growth of diabetes has occurred unevenly across sociodemographic groups, increasing at a more pronounced rate among
disadvantaged populations, such as racial/ethnic minorities and less educated individuals.9,10,14 At the same time, two recent analyses found that the prevalence of total and diagnosed diabetes among adults may have stabilized between 2008 and 2012.9,10 Several studies also showed that the share of all diabetes cases that are undiagnosed has declined substantially since the 1980s.13 15 Taken together, existing trend analyses indicate that although important strides have been made, diabetes still remains an urgent public health concern, especially for disadvantaged populations. From the Population Studies Center, Institute for Social Research, University of Michigan, Ann Arbor, Michigan Address correspondence to: Michael Fang, MA, Population Studies Center, Institute for Social Research, University of Michigan, 426 Thompson Street, Ann Arbor MI 48106. E-mail: [email protected]. 0749-3797/$36.00 https://doi.org/10.1016/j.amepre.2018.05.018
© 2018 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.
Am J Prev Med 2018;000(000):1 9
1
ARTICLE IN PRESS 2
Fang / Am J Prev Med 2018;000(000):1 9
The primary aim of this study is to update national trends in diabetes among U.S. adults. Data were from the National Health and Nutrition Examination Survey (NHANES), collected from 1999 2000 to 2015 2016. The NHANES is the only national study that collects diabetes information through self-report and through blood samples, making it the only source available for estimating national trends in total diabetes. The analyses considered trends in undiagnosed, diagnosed, and total diabetes in the total population and across different sociodemographic groups.
TAGEDH1METHODSTAGEDN Study Sample Details about the NHANES are available elsewhere.16 18 In brief, the study consists of a series of nationally representative, crosssectional surveys designed to monitor the health of individuals in the U.S. Beginning in 1999, the NHANES began collecting data in continuous, 2-year cycles. During each survey cycle, a nationally representative sample of respondents are selected from the U.S. non-institutionalized, civilian population using a complex, stratified, multistage probability cluster sampling design. After obtaining written informed consent from the participants, data are collected through in-home interviews and visits to a mobile examination center, where health examinations are administered by trained medical staff members. All data collection received approval from the National Center for Health Statistics research ethics review board. For this study, all waves of data from the continuous NHANES were used to estimate trends in the prevalence of diabetes (1999 2000 to 2015 2016). Across these nine waves, response rates ranged from 61% to 84% for in-home interviews and 59% to 80% for the mobile center visits. Of the 92,062 respondents in these data, the author excluded respondents who were aged <20 years (n=42,550), women who were pregnant (n=1,486), and those with missing data for their diabetes status (n=4,654), weight or height (n=767), or sociodemographic characteristics (n=51). These restrictions yielded a final sample of 42,554 respondents.
Measures During visits to the mobile examination center, trained phlebotomists drew samples of blood from respondents following a standardized protocol. These samples were subsequently analyzed to determine respondents’ glycated hemoglobin (HbA1c) levels using high-performance liquid chromatography methods. The laboratory and tools used to measure HbA1c levels varied across survey years.19 HbA1c testing was performed at University of Minnesota from 2007 2008 to 2011 2012 and at the University of Missouri from 2013 2014 to 2015 2016 using the Tosoh A1C 2.2 Plus Glycohemoglobin Analyzer (2007 2008), Tosoh G7 Glycohemoglobin Analyzer (2009 2010 and 2011 12), or Tosoh G8 Glycohemoglobin Analyzer (2011 2012 to 2015 2016).20 A recent study suggests that these changes in laboratory methodology may have increased HbA1c levels,14 and NHANES analytic guidelines advise researchers to compare HbA1c levels across survey years with caution.19 As a sensitivity test, all prevalence rates
were re-estimated using calibrated HbA1c values that account for these laboratory changes (Appendix Table 1, available online). Three measures of diabetes were used for this study. The first was diagnosed diabetes. During the in-home interviews, respondents were asked if they had ever been told by a doctor or health professional that they had diabetes. Those who answered yes were coded as having diagnosed diabetes. The second was undiagnosed diabetes. Consistent with several prior studies,14,21 respondents with HbA1c levels 6.5%1 and who answered no to being diagnosed with diabetes by a health professional were coded as having undiagnosed diabetes. The third was total diabetes, measured as the total number of respondents who had either diagnosed or undiagnosed diabetes. The remaining participants (i.e., those who were never diagnosed with diabetes and who had HbA1c levels <6.5%) were coded as individuals without diabetes. Respondents with missing information for either diagnosed diabetes status or HbA1c were excluded from the analysis. Height and weight were measured using standardized techniques and equipment during mobile examination visits. These measures were used to calculate respondents’ BMI (calculated by dividing weight in kilograms by the square of height in meters). Those with BMI <25 were categorized as normal, those with BMI 25 but <30 were categorized as overweight, and those with BMI 30 were categorized as obese. A standardized questionnaire was administered during the inhome visits to collect sociodemographic information from respondents, including sex (male, female), race/ethnicity (nonHispanic white, non-Hispanic black, Mexican American), age (20 44, 45 64, 65 years), and education (less than high school, high school graduate, some college, and college graduate). All sociodemographic measures were self-reported.
Statistical Analysis This paper began by estimating the prevalence of diabetes across all nine waves of the continuous NHANES, from 1999 2000 to 2015 2016. Estimates were for the overall adult population and stratified by sex, age, race/ethnicity, educational level, and BMI category. Race-specific prevalence was only estimated for Mexican Americans, non-Hispanic blacks, and non-Hispanic whites due to small sample size for other racial/ethnic groups during the study period. To test for time trends between 1999 2000 and 2015 2016, logistic regression models were estimated. In these models, diabetes status was regressed onto survey year, which was entered as a continuous variable. A statistically significant coefficient for the survey year variable was interpreted as a change in the prevalence of diabetes. Quadratic and cubic time trends were tested and found to be nonsignificant for the overall population and for different subgroups, therefore, linear terms were used to assess temporal trends. The author then proceeded to investigate the factors that might explain changes in diabetes trends. To do this, a series of logistic regression models were estimated examining the relationship between survey cycle, entered as a continuous linear variable, and total diabetes. An unadjusted model was used and then potential mediators were added, including age, BMI category, sex, race/ethnicity, and education, in a stepwise fashion. The coefficient for the survey cycle in the adjusted models was compared with the coefficient in the www.ajpmonline.org
& 2018
Table 1. Weighted Characteristics of U.S. Adults Aged 20 Years or Older, by Diabetes Status: NHANES 1999 2000 to 2015 2016 Characteristics
2001 2002
2003 2004
2005 2006
2007 2008
2009 2010
2011 2012
2013 2014
2015 2016
51.6 48.4
52.6 47.4
48.7 51.3
46.6 53.4
49.8 50.2
52.8 47.2
50.5 49.5
50.7 49.3
54.1 45.9
16.2 47.4 36.5
21.3 45.4 33.3
14.9 46.2 38.9
17.3 45.6 37.1
13.3 49.5 37.2
13.8 45 41.1
16.2 49.1 34.7
13.6 48.1 38.2
15.9 46.2 37.9
42.8 27.6 22.5 7.2
31.1 23.5 27.1 18.3
30 23.6 28.2 18.1
26.5 29 27.4 17.2
31.2 27.5 28.1 13.2
28.9 21.5 30.8 18.8
26.5 25.5 28.3 19.8
22.3 24 36.2 17.5
21.9 21.3 33.6 23.3
8.1 74.3 17.6
8.8 72.3 18.9
9.7 74.8 15.5
10.4 70.2 19.5
9.2 70.7 20.1
12.8 69.2 17.9
10.9 67.5 21.6
11.6 72.1 16.3
14.8 67.2 18
14.3 32.1 53.6
14.4 31.7 54
14.3 27.8 58
13.4 28.5 58.1
12.7 24 63.4
9.7 24.7 65.6
14 23.4 62.6
9.8 27.7 62.6
10.6 25.3 64.1
48.9 51.1
48.6 51.4
49 51
49.1 50.9
48.3 51.7
48.1 51.9
48.3 51.7
48.4 51.6
47.6 52.4
56.8 29 14.2
54 33.3 12.7
52.5 32.6 14.9
49.9 34.8 15.2
49.9 35.8 14.3
50.3 34.9 14.8
48.6 36 15.4
49.5 34.4 16.1
47.4 34.6 18.1
22.6 26.1 27.8 23.5
17.7 25.5 29.9 26.9
16.6 27.5 31.9 23.9
16.3 24.6 31.8 27.4
18.8 24.7 29.3 27.1
17.7 23.1 30.3 28.8
15 19.5 32.5 33
14.3 21.4 32.1 32.2
13 20.7 32.7 33.6 (continued on next page)
ARTICLE IN PRESS
1999 2000
Fang / Am J Prev Med 2018;000(000):1 9
Adults with diabetesa Sex Male Female Age group 20 44 years 45 64 years 65 years Education Less than HS HS completed Some college College graduate Race/ethnicity Mexican-American NH white NH black BMI categoryb Normal Overweight Obese Adults without diabetes Sex Male Female Age group 20 44 years 45 64 years 65 years Education Less than HS HS completed Some college College graduate Race/ethnicity
3
ARTICLE IN PRESS 4 Note: Values are percentages. Boldface indicates statistical significance (p<0.05) x2 test. HS, high school; NH, non-Hispanic; NHANES, National Health and Nutrition Examination Survey. a Those with diabetes have either been diagnosed with diabetes by a doctor or other health professional or have glycated hemoglobin levels 6.5%. b Respondents with BMI (weight in kilograms divided by height in meters squared) <25 were coded as normal, those with BMI >25 and <30 were coded as overweight, and those with BMI >30 were coded as obese.
30.6 33.1 36.3 31.9 33.6 34.5 36.5 35.7 27.7 38 34 28
35.4 35.3 29.2
34.9 33.3 31.8
34.2 36.1 29.7
33.4 34.3 32.3
33.1 35.3 31.7
9.9 77.9 12.1 10.4 77.6 12 8.6 79.7 11.6 7.8 81.4 10.8
Mexican-American NH white NH black BMI categoryb Normal Overweight Obese
7.1 81.6 11.3
8.2 80.5 11.4
8.3 80.5 11.2
9.3 80.2 10.5
9.3 79.1 11.6
2013 2014 2011 2012 2009 2010 2007 2008 2005 2006 2003 2004 2001 2002 1999 2000 Characteristics
Table 1. Weighted Characteristics of U.S. Adults Aged 20 Years or Older, by Diabetes Status: NHANES 1999 2000 to 2015 2016 (continued)
2015 2016
Fang / Am J Prev Med 2018;000(000):1 9 unadjusted model to assess the extent to which these covariates explained trends in diabetes prevalence over time. All analyses were conducted using Stata, version 14.0. To adjust for complex survey design effects (e.g., oversampling of certain populations), nonresponse bias, and post-stratification, all analyses were weighted using recommended survey weights, making the results here representative of the adult, civilian, non-institutionalized U.S. population.22 A two-sided p-value of <0.05 was considered to indicate statistical significance. Because data from the NHANES are de-identified and publicly available, the University of Michigan’s IRB determined that this study was exempt from review. All analyses were performed in 2017.
TAGEDH1RESULTSTAGEDN Table 1 presents the characteristics of U.S. adults aged 20 years with and without diabetes. The composition of those with diabetes remained stable during the study period, with the exception of changes in education. The share of adults with diabetes that did not finish high school or only attained a high school degree declined by 21 and 6 percentage points, respectively, whereas those who had some college education or graduated from college increased by 11 and 16 percentage points (p<0.001). Among those without diabetes, a growing proportion of adults were older (p=0.001), had some college education or graduated from college (p<0.001), and were obese (p<0.001). Table 2 presents the prevalence of diabetes in the overall population and across subgroups. Between 1999 2000 and 2015 2016, the prevalence of total diabetes increased by 5.6 percentage points among the overall adult population, going from 7.7% to 13.3% (p<0.001 for trend; Figure 1). Expressed in relative terms, there was a 72.0% increase in the prevalence of diabetes among U.S. adults during this 18-year period. The increased prevalence of total diabetes was largely a function of growth in diagnosed diabetes. During the study period, the share of American adults diagnosed with diabetes grew 5.3 percentage points (6.0% to 11.3%, p<0.001 for trend; Figure 1), or 88.3%. On the other hand, the unadjusted prevalence of undiagnosed diabetes remained stable during this period (p=0.058). Because the prevalence of diagnosed diabetes increased while the prevalence of undiagnosed diabetes remained roughly unchanged, the share of total diabetes that was undiagnosed decreased significantly from 1999 2000 to 2015 2016 (22% to 15.3%, p<0.013 for trend). Estimates and trends were generally similar using calibrated values of HbA1c (Appendix Table 1, available online). Growth in the prevalence of total diabetes was observed for every subgroup examined, though the rate of change varied considerably. Notable disparities were observed across race/ethnicity, BMI category, and age www.ajpmonline.org
& 2018
Table 2. Weighted Prevalence of Total Diabetesa per 100 Adults Aged 20 Years or Older in the U.S.: NHANES 1999 2000, 2015 2016 Prevalence (95% CI) Characteristics Overall population Total diabetes Diagnosed diabetesb Undiagnosed diabetesc
Age group 20 44 years 45 64 years 65 years Education Less than HS HS graduate Some college College graduate Race/ethnicitye Mexican American NH white NH black BMI categoryf Non-obese
<0.001
7.7 (6.4, 9.1) 6 (4.9, 7.1) 1.7 (1.3, 2.1) 22 (18.6, 25.5)
8.1 (6.9, 9.3) 6.6 (5.7, 7.5) 1.5 (1.1, 2.0) 18.8 (14.5, 23.1)
9.7 (8.4, 11.0) 7.7 (6.4, 9.0) 2 (1.5, 2.4) 20.4 (15.6, 25.3)
9.6 (8.3, 10.8) 7.9 (6.9, 8.8) 1.7 (1.0, 2.3) 17.6 (11.7, 23.4)
10.9 (9.4, 12.4) 8.5 (7.0, 9.9) 2.4 (1.9, 2.9) 22.2 (17.7, 26.8)
10.8 (9.8, 11.8) 8.4 (7.4, 9.4) 2.4 (2.1, 2.7) 22.2 (19.0, 25.5)
11.4 (9.9, 12.9) 9.2 (8.0, 10.5) 2.2 (1.6, 2.7) 19 (15.0, 23.0)
12.1 (11.2, 13.1) 10.1 (9.1, 11.2) 2 (1.5, 2.5) 16.4 (12.4, 20.4)
13.3 (11.4, 15.2) 11.3 (9.7, 12.8) 2 (1.5, 2.6) 15.3 (12.4, 18.3)
8.1 (6.3, 9.9) 7.3 (5.9, 8.8)
8.7 (7.3, 10.1) 7.5 (6.3, 8.7)
9.7 (8.2, 11.1) 9.8 (8.4, 11.2)
9.1 (7.7, 10.5) 10 (8.1, 11.9)
11.2 (9.8, 12.6) 10.6 (8.5, 12.7)
11.7 (10.3, 13.2) 9.9 (8.7, 11.1)
11.9 (10.0, 13.7) 11 (9.2, 12.7)
12.6 (11.2, 14.1) 11.7 (10.5, 12.8)
14.9 (12.2, 17.5) 11.8 (10.0, 13.7)
<0.001
2.3 (1.6, 3.1) 12 (9.5, 14.6) 17.7 (14.4, 21.1)
3.4 (2.2, 4.5) 10.7 (8.3, 13.2) 18.7 (16.8, 20.7)
3 (2.2, 3.7) 13.2 (11.6, 14.8) 22 (17.7, 26.3)
3.5 (2.5, 4.5) 12.2 (10.1, 14.3) 20.4 (18.1, 22.7)
3.2 (2.3, 4.0) 14.5 (12.0, 17.0) 24.1 (20.0, 28.2)
3.2 (2.3, 4.1) 13.5 (11.8, 15.2) 25.2 (22.2, 28.2)
4.1 (3.1, 5.2) 15 (12.7, 17.2) 22.4 (19.4, 25.5)
3.7 (3.0, 4.3) 16.2 (13.6, 18.8) 24.7 (22.6, 26.9)
4.9 (3.8, 6.0) 17 (14.3, 19.7) 24.4 (19.8, 28.9)
<0.001
13.7 (10.7, 16.6) 8.1 (6.5, 9.8) 6.3 (3.6, 9.1) 2.5 (1.7, 3.3)
13.4 (11.4, 15.4) 7.5 (6.1, 8.9) 7.4 (5.7, 9.1) 5.6 (3.5, 7.7)
16.3 (13.5, 19.1) 8.5 (7.1, 9.9)
16.8 (14.9, 18.8) 12 (8.8, 15.2) 10.5 (8.9, 12.1) 5.6 (4.0, 7.2)
16.5 (14.5, 18.5) 10.1 (8.3, 11.9) 11 (9.2, 12.8) 7.3 (5.5, 9.2)
18.5 (15.5, 21.5) 14.4 (10.9, 17.9) 10.1 (8.4, 11.7) 7.2 (4.9, 9.4)
17.7 (15.5, 19.9) 13.4 (11.4, 15.5) 13.5 (11.2, 15.8) 7 (5.4, 8.6)
20.5 (17.6, 23.3) 13.6 (11.3, 15.9) 13.6 (10.7, 16.6) 9.6 (7.5, 11.8)
<0.001
8.7 (6.4, 11.0) 7.5 (5.6, 9.5)
14.7 (12.2, 17.1) 11.1 (8.1, 14.0) 8.3 (6.7, 10.0) 6.2 (4.4, 8.1)
8.3 (5.9, 10.7) 6.7 (5.2, 8.2) 10.9 (9.4, 12.5)
8.5 (7.6, 9.3) 6.8 (5.9, 7.6) 12.6 (10.1, 15.0)
10.8 (7.7, 14.0) 8.7 (7.1, 10.2) 12.2 (10.1, 14.4)
11.4 (9.6, 13.3) 8.2 (6.7, 9.8) 15.1 (12.6, 17.6)
10.7 (8.8, 12.7) 9.7 (7.3, 12.1) 18.9 (17.2, 20.5)
14 (10.7, 17.2) 9.3 (7.9, 10.7) 15.3 (13.2, 17.4)
13.2 (9.6, 16.8) 9.3 (7.6, 11.0) 18.4 (14.9, 21.9)
13.3 (10.7, 15.8) 11.3 (10.0, 12.6) 15.7 (14.2, 17.2)
18.4 (13.2, 23.6) 11.5 (9.5, 13.6) 18.3 (15.6, 21.0)
<0.001
<0.001 0.058 0.013
<0.001
<0.001 <0.001
<0.001 <0.001 <0.001
<0.001 <0.001
0.032
5 (continued on next page)
ARTICLE IN PRESS
Female
p for linear trend
Fang / Am J Prev Med 2018;000(000):1 9
Percentage of diabetes cases that were undiagnosedd Sex Male
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
ARTICLE IN PRESS 6
Obese
Overweight
Note: Boldface indicates statistical significance (p<0.05). HbA1c, glycated hemoglobin; HS, high school; NH, non-Hispanic; NHANES, National Health and Nutrition Examination Survey. a Total diabetes refers to respondents who have either been diagnosed with diabetes by a doctor or other health professional or have HbA1c levels 6.5%. b Diagnosed diabetes refers to respondents who have been told by a doctor or other health professional that they have diabetes. c Undiagnosed diabetes refers to respondents who have not been told by a doctor or other health professional that they have diabetes but have HbA1c levels 6.5%. d Estimates and CIs were computed using respondents who had any type of diabetes (i.e., diagnosed or undiagnosed) rather than the entire adult sample. Figures were weighted using the recommended sample weight. e Race-specific prevalence was not computed for respondents from other racial/ethnic backgrounds due to small sample size. However, their information was included for all other estimates. f Respondents with BMI (weight in kilograms divided by height in meters squared) <25 were coded as normal, those with BMI >25 and <30 were coded as overweight, and those with BMI >30 were coded as obese.
<0.001
0.006
5 (3.6, 6.5) 10.5 (8.2, 12.8) 21.3 (18.6, 24.0) 4 (2.9, 5.2) 10.2 (8.4, 12.1) 20 (18.5, 21.5) 5.2 (3.5, 6.8) 7.9 (6.9, 8.8) 20.3 (17.3, 23.3) 3.4 (2.6, 4.2) 8 (6.0, 10.0) 19.7 (17.8, 21.7) 4.3 (3.1, 5.6) 7.5 (6.5, 8.6) 20.7 (17.3, 24.1) 3.9 (3.0, 4.8) 8.3 (6.6, 10.0) 16.2 (13.9, 18.5) 4.2 (2.1, 6.2) 7.8 (5.9, 9.6) 17.6 (15.2, 19.9) 3.3 (2.2, 4.5) 7.2 (5.8, 8.6) 14.6 (12.2, 17.0) 3.1 (1.9, 4.2) 7.3 (5.0, 9.6) 13.8 (11.9, 15.8)
Prevalence (95% CI)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Characteristics
a
Table 2. Weighted Prevalence of Total Diabetes per 100 Adults Aged 20 Years or Older in the U.S.: NHANES 1999 2000, 2015 2016 (continued)
p for linear trend
Fang / Am J Prev Med 2018;000(000):1 9
Figure 1. Prevalence of diagnosed, undiagnosed, and total diabetes per 100 adults among U.S. adults aged 20 years and older from 1999 2000 to 2015 2016. Note: Diagnosed diabetes is measured through self-report, undiagnosed diabetes is measured as never being diagnosed with diabetes but having HbA1c levels 6.5%, and total diabetes is measured as the sum of diagnosed and undiagnosed diabetes. Data come from the National Health and Nutrition Examination Survey. HbA1c, glycated hemoglobin.
group. The prevalence of total diabetes grew by 4.8 percentage points for non-Hispanic whites (6.7% to 11.5%, p<0.001 for trend), compared with 7.4 percentage points for non-Hispanic blacks (10.9% to 18.3%, p<0.001 for trend) and 10.1 percentage points for Mexican Americans (8.3% to 18.4%, p<0.001 for trend). For non-obese adults, the prevalence of total diabetes rose by 1.9 percentage points (3.1% to 5.0%, p=0.032 for trend), compared with 7.5 percentage points for those who were obese (13.8% to 21.3%, p<0.001 for trend). Across age groups, the prevalence of total diabetes among adults aged 20 44 years increased by 2.6 percentage points (2.3% to 4.9%, p<0.001 for trend), 5 percentage points among adults aged 45 64 years (12% to 17%, p<0.001 for trend) and 6.7 percent points among adults aged 65 years (17.7% to 24.4%, p<0.001 for trend). On the other hand, differences across sex and educational level were less pronounced. For instance, the prevalence of total diabetes increased by 4.5 percentage points for females (7.3% to 11.8%, p<0.001 for trend) and 6.8 percent points for males (8.1% to 14.9%, p<0.001 for trend). Estimates and trends across subgroups were generally similar using calibrated values of HbA1c (Appendix Table 1, available online). Table 3 shows the results from logistic regression models examining the relationship between survey cycle and total diabetes among U.S. adults. The unadjusted model indicated that each one-unit increase in survey cycle (equivalent to 2 years) was associated with a 7.2% increase in the odds of having diabetes for U.S. adults (Model 1: OR=1.07, 95% CI=1.05, 1.09, p<0.001). After adjusting for changes in the age distribution of the www.ajpmonline.org
ARTICLE IN PRESS Fang / Am J Prev Med 2018;000(000):1 9
Table 3. Unadjusted and Adjusted Association Between Survey Cycle and Total Diabetesa: NHANES 1999 2016 Model Model 1b Model 2c Model 3d Model 4e
OR (95% CI) 1.072 (1.051, 1.094) 1.057 (1.035, 1.080) 1.043 (1.021, 1.065) 1.044 (1.025, 1.063)
Note: Boldface indicates statistical significance (p<0.05). HbA1c, glycated hemoglobin; NHANES, National Health and Nutrition Examination Survey. a Total diabetes refers to respondents who have either been diagnosed with diabetes by a doctor or other health professional or have HbA1c levels 6.5%. b Model 1 is unadjusted. c Model 2 adjusts for age. d Model 3 adjusts for age and BMI category. e Model 4 adjusts for age, BMI category, sex, race/ethnicity, and education.
population, the growth in diabetes over time was reduced by around 21% (Model 2: OR=1.06, 95% CI=1.04, 1.08, p<0.001). Adjusting for both changes in age and BMI in the population accounted for roughly 40% of the growth of diabetes during the study period (Model 3: OR=1.04, 95% CI=1.02, 1.07, p<0.001). Adjusting for changes in sex, race, and educational level account did not further explain the growth in the prevalence of total diabetes.
TAGEDH1DISCUSSIONTAGEDN From 1999 2000 to 2015 2016, the unadjusted prevalence of diabetes increased by 72.0% among adults in the U.S., reaching an all-time high of 13.3 diagnoses per every 100 adults in 2015 2016. The growth in diabetes was found across all subgroups, though the rate of change was notably higher among older adults, racial minorities, and those who were obese. The increase in the prevalence of total diabetes among the Mexican American population was particularly pronounced, rising by 10.1 percentage points during the 18-year study period. The increased prevalence of total diabetes was driven to a large extent by rising rates of obesity and the aging of the U.S. adult population. It is important to note that this study used a different measure of undiagnosed diabetes than the one commonly found in CDC reports and other studies using the NHANES. More specifically, although this paper relied on HbA1c, past studies have combined HbA1c readings with fasting blood glucose and oral glucose tolerance readings to more fully assess undiagnosed diabetes.9,11,13,14 Unfortunately, these latter two measures were not yet available for the 2015 2016 wave of the NHANES at the time of writing this paper. As a result, the exact estimates here are not directly comparable with these past studies, as the & 2018
7
prevalence of undiagnosed diabetes and by extension, total diabetes, are likely underestimated. Despite measurement differences, the overall growth in prevalence rates documented here are still substantively consistent with other trend analyses using the more comprehensive measure of total diabetes. For instance, one paper used the more comprehensive total diabetes measure and found that the prevalence of total diabetes increased by 3.5 percentage points from 1999 to 2012, while this paper documented a 3.7 percentage point increase during the same time frame.9 Moreover, the results here also align well with other studies that define total diabetes using selfreported information and HbA1c. For instance, the analyses by Cowie et al.23 of the NHANES data showed the prevalence of total diabetes grew 7.4% in 1988 1994 to 9.3% in 2003 2006. Similarly, using the NHANES, Selvin and colleagues14 show that the prevalence of total diabetes increased from 6.2% in 1988 1994 to 9.9% in 2005 2010. Considered together, these analyses indicate that diabetes remains a pervasive and growing public health concern among U.S. adults. The high growth of diabetes among Mexican-American adults found in this study is also consistent with the literature on racial/ethnic disparities in diabetes. One recent study found that the prevalence of diabetes among Mexican-American adults grew by 6 percentage points from 1999 to 2012, compared with 2 percentage points for white adults.9 A second study showed that from 1988 1994 to 2005 2010, the prevalence of diabetes increased 4.1 percentage points for Mexican-Americans adults and 2.8 for whites adults.14 A third paper found that between 2001 and 2009, the rate of growth in the prevalence of type 2 diabetes was ten times higher in Hispanic youth compared with non-Hispanic white youth.24 This suggests that the diabetes epidemic has—and may continue to have—a disproportionate impact on the Hispanic population. The possible mechanisms underlying the rising rates of diabetes among U.S. adults identified in this study, population aging and increased rates of obesity, are also consistent with prior studies. For instance, one study found that controlling for age, race, and education explained 22% 28% of the growth in diabetes from 1988 to 2010,14 whereas another found that age accounted for 9% and 22% of the increase in diabetes from 1976 to 2010 among American men and women, respectively.25 Similarly, BMI has been identified in multiple studies as being one of the most important risk factors for developing type 2 diabetes.26,27 A study using the NHANES showed that controlling for changes in BMI and social characteristics among U.S. adults completely attenuated the growth of diabetes from 1999 to 2010.14 Two community studies also showed that roughly half of new cases of type 2 diabetes among respondents could have been prevented if they were not
ARTICLE IN PRESS 8
Fang / Am J Prev Med 2018;000(000):1 9 28,29
obese or overweight. A longitudinal study based on roughly 85,000 U.S. women echoed this finding, showing that roughly 60% of new type 2 diabetes cases were attributable to respondents being overweight.27 Although the increasing prevalence of total diabetes is concerning from a public health perspective, it was encouraging to find that undiagnosed diabetes remained largely unchanged during the study period. Indeed, the growth in total diabetes was largely a function of increasing rates of diagnosed diabetes. Because of this, the share of diabetes cases that was undiagnosed declined significantly over time. This trend is consistent with prior studies and suggests that improvements in diabetes awareness and screening are likely resulting in better detection of diabetes among U.S. adults.9,13 15,21
Limitations The main strength of this study was the use of a large, nationally representative data set (NHANES), which allowed for accurate estimates of the prevalence of diabetes among U.S. adults over time. Nevertheless, this study had several important limitations. First, as mentioned, the paper was only able to assess undiagnosed diabetes using HbA1c information due to data limitations. Second, although this study found a large increase in the prevalence of diabetes for Mexican-American adults, it is important to note that the estimate for 2015 2016 had relatively wide CIs. The imprecision of this figure is important because the difference in prevalence rates in 2015 2016 and 1999 2000 were used to determine the rate of change over time. Third, the NHANES generally takes a single HbA1c reading from respondents. The American Diabetes Association recommends that individuals who have HbA1c levels at or above 6.5% be re-tested a second time to confirm the diagnosis.11 Thus, there may be some individuals who are misclassified as having undiagnosed diabetes in this paper. Fourth, although this paper included subgroup analyses, it important to note the NHANES data may have lacked the statistical power to detect differences among smaller groups. Therefore, it is important to interpret these analyses with some caution. Finally, because of limitations with the NHANES data, this paper could not differentiate between respondents with type 1 or type 2 diabetes.
TAGEDH1CONCLUSIONSTAGEDN This study found that the prevalence of diabetes among U.S. adults increased substantially from 1999 to 2016. This growth occurred differentially across populations, particularly impacting Hispanic adults. The increased prevalence of total diabetes was driven largely by population aging and growing obesity rates. These findings suggest that additional
interventions aimed at reducing obesity may be necessary to help slow growth of diabetes within the U.S. population.
TAGEDH1ACKNOWLEDGMENTSTAGEDN The author thanks Dr. Elizabeth Selvin for sharing her NHANES data set that contained calibrated values for glycated hemoglobin. The author also thanks Dr. Susan Fang for her comments on an early draft of this paper. The author confirms that there are no known conflicts of interest associated with this publication and there has been no financial support for this work that could have influenced its outcome. No financial disclosures were reported by the author of this paper.
TAGEDH1SUPPLEMENTAL MATERIALTAGEDN Supplemental materials associated with this article can be found, in the online version, at doi:10.1016/j.amepre.2018.05.018.
TAGEDH1REFERENCESTAGEDN 1. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 37(suppl 1):S81–S90. https://doi.org/ 10.2337/dc14-S081. 2. Emerging Risk Factors Collaboration. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet. 375(9733):2215– 2222. https://doi.org/10.1016/S0140-6736(10)60484-9. 3. Afkarian M, Zelnick LR, Hall YN, et al. Clinical manifestations of kidney disease among U.S. adults with diabetes, 1988 2014. JAMA. 2016;316(6):602–610. https://doi.org/10.1001/jama.2016.10924. 4. Rawshani A, Rawshani A, Franzen S, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med. 2017;376 (15):1407–1418. https://doi.org/10.1056/NEJMoa1608664. 5. Franco OH, Steyerberg EW, Hu FB, Mackenbach J, Nusselder W. Associations of diabetes mellitus with total life expectancy and life expectancy with and without cardiovascular disease. Arch Intern Med. 2007;167(11):1145– 1151. https://doi.org/10.1001/archinte.167.11.1145. 6. Kobrin Klein BE. Overview of epidemiologic studies of diabetic retinopathy. Ophthalmic Epidemiol. 2007;14(4):179–183. https://doi.org/ 10.1080/09286580701396720. 7. Tancredi M, Rosengren A, Svensson A-M, et al. Excess mortality among persons with type 2 diabetes. N Engl J Med. 2015;373 (18):1720–1732. https://doi.org/10.1056/NEJMoa1504347. 8. Dall TM, Yang W, Halder P, et al. The economic burden of elevated blood glucose levels in 2012: diagnosed and undiagnosed diabetes, gestational diabetes mellitus, and prediabetes. Diabetes Care. 2014;37 (12):3172–3179. https://doi.org/10.2337/dc14-1036. 9. Menke A, Casagrande S, Geiss L, Cowie CC. Prevalence of and trends in diabetes among adults in the United States, 1988 2012. JAMA. 2015;314(10):1021–1029. https://doi.org/10.1001/jama.2015.10029. 10. Geiss LS, Wang J, Cheng YJ, et al. Prevalence and incidence trends for diagnosed diabetes among adults aged 20 to 79 years, United States, 1980 2012. JAMA. 2014;312(12):1218–1226. https://doi.org/10.1001/ jama.2014.11494. 11. Cowie CC, Rust KF, Ford ES, et al. Full accounting of diabetes and prediabetes in the U.S. population in 1988 1994 and 2005 2006. Diabetes Care. 2009;32(2):287–294. https://doi.org/10.2337/dc08-1296. 12. Mokdad AH, Ford ES, Bowman BA, et al. Diabetes trends in the U.S.: 1990 1998. Diabetes Care. 2000;23(9):1278–1283. https://doi.org/ 10.2337/diacare.23.9.1278.
www.ajpmonline.org
ARTICLE IN PRESS Fang / Am J Prev Med 2018;000(000):1 9 13. Selvin E, Wang D, Lee AK, Bergenstal RM, Coresh J. Identifying trends in undiagnosed diabetes in U.S. adults by using a confirmatory definition: a cross-sectional study. Ann Intern Med. 2017;167(11):769– 776. https://doi.org/10.7326/M17-1272. 14. Selvin E, Parrinello CM, Sacks DB, Coresh J. Trends in prevalence and control of diabetes in the United States, 1988 1994 and 1999 2010. Ann Intern Med. 2014;160(8):517–525. https://doi.org/10.7326/M13-2411. 15. Gregg EW, Cadwell BL, Cheng YJ, et al. Trends in the prevalence and ratio of diagnosed to undiagnosed diabetes according to obesity levels in the U.S. Diabetes Care. 2004;27(12):2806–2812. https://doi.org/ 10.2337/diacare.27.12.2806. 16. Johnson CL, Dohrmann SM, Burt VL, Mohadjer LK. National Health and Nutrition Examination Survey: sample design, 2011 2014. Vital Health Stat 2. 2014(162):1–33. 17. Curtin LR, Mohadjer LK, Dohrmann SM, et al. National Health and Nutrition Examination Survey: sample design, 2007 2010. Vital Health Stat 2. 2013(160):1–23. 18. Curtin L, Mohadjer L, Dohrmann S, et al. The National Health and Nutrition Examination Survey: sample design, 1999 2006. Vital Health Stat 2. 2012(155):1–39. 19. National Center for Health Statistics. Updated Advisory for NHANES Hemoglobin A1c(Glycohemoglobin) Data. www.cdc.gov/nchs/data/ nhanes/A1c_webnotice.pdf. Published 2012. Accessed October, 2017. 20. CDC. National Health and Nutrition Examination Laboratory Protocol. Hyattsville, MD: U.S. HHS, CDC; 2015 2016. 21. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetes and impaired fasting glucose in adults in the U.S. population. Diabetes Care. 2006;29(6):1263–1268. https://doi.org/10.2337/dc06-0062.
& 2018
9
22. Johnson CL, Paulose-Ram R, Ogden CL, et al. National Health and Nutrition Examination Survey. Analytic guidelines, 1999 2010. Vital Health Stat 2. 2013(161):1–24. 23. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetes and high risk for diabetes using A1C criteria in the U.S. population in 1988 2006. Diabetes Care. 2010;33(3):562–568. https://doi.org/ 10.2337/dc09-1524. 24. Mayer-Davis EJ, Lawrence JM, Dabelea D, et al. Incidence trends of type 1 and type 2 diabetes among youths, 2002 2012. N Engl J Med. 2017;376(15):1419–1429. https://doi.org/10.1056/NEJMoa1610187. 25. Menke A, Rust KF, Fradkin J, Cheng YJ, Cowie CC. Associations between trends in race/ethnicity, aging, and body mass index with diabetes prevalence in the United States: a series of cross-sectional studies increase in diabetes prevalence over time. Ann Intern Med. 2014;161 (5):328–335. https://doi.org/10.7326/M14-0286. 26. Vazquez G, Duval S, Jacobs Jr. DR Jr., Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev. 2007;29(1):115– 128. https://doi.org/10.1093/epirev/mxm008. 27. Hu FB, Manson JE, Stampfer MJ, et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med. 2001;345(11):790– 797. https://doi.org/10.1056/NEJMoa010492. 28. Steinbrecher A, Morimoto Y, Heak S, et al. The preventable proportion of type 2 diabetes by ethnicity: the multiethnic cohort. Ann Epidemiol. 2011;21(7):526–535. https://doi.org/10.1016/j.annepidem.2011.03.009. 29. Jiang Y, Chen Y, Mao Y. CCDPC Obesity Working Group. The contribution of excess weight to prevalent diabetes in Canadian adults. Public Health. 2008;122(3):271–276. https://doi.org/10.1016/j.puhe.2007.06.002.
RESEARCH ARTICLE
Trends in the Prevalence of Diabetes Among U.S. Adults: 1999 2016 D1X XMichael Fang, D2X XMA
Introduction: The prevalence of diabetes has increased substantially over the past three decades. This study sought to estimate recent trends in the prevalence of diabetes among U.S. adults.
Methods: This paper estimated trends in the prevalence of diagnosed, undiagnosed, and total diabetes among U.S. adults from 1999 2000 to 2015 2016 (analyzed in 2017). Data come from 42,554 respondents aged 20 years who participated in the National Health and Nutrition Examination Survey. Diagnosed diabetes was measured through self-report, undiagnosed diabetes was measured as never being diagnosed with diabetes but having glycated hemoglobin levels 6.5%, and total diabetes was measured as the sum of individuals with diagnosed and undiagnosed diabetes. Results: In the overall U.S. adult population, the unadjusted prevalence of total diabetes increased from 7.7% in 1999 2000 to 13.3% in 2015 2016 (p<0.001 for trend). Growth was observed for all subgroups, though the rate of change was higher in older adults, racial minorities, and those who were obese compared with their peers. Increasing prevalence among Mexican-American adults was particularly pronounced, rising by 10.1 percentage points during the study period (8.3% to 18.4%, p < 0.001). Roughly 40% of the increase in total diabetes was accounted for by changes in the age and rates of obesity in the U.S. population.
Conclusions: From 1999 to 2016, the prevalence of diabetes among U.S. adults increased at a substantial rate. This growth occurred differentially across subgroups, particularly impacting MexicanAmerican adults, and was driven in large part by population aging and increasing obesity rates. Am J Prev Med 2018;000(000):1 9. © 2018 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.
TAGEDH1INTRODUCTIONTAGEDN
D
iabetes is a group of metabolic disorders that stems from problems in insulin production or action and leads to hyperglycemia.1 Diabetes is associated with numerous negative health consequences, including heart disease, kidney failure, blindness, and premature death.2 7 The condition also carries a substantial economic burden, costing the U.S. an estimated $322 billion in lost productivity and medical costs in 2012.8 Within the U.S., the prevalence of diabetes has increased substantially over the past three decades among adults.3,9 13 Between 1988 and 2012, rates of total diabetes (diagnosed and undiagnosed) increased from 8.6 to 12.3 cases per 100 U.S. adults.9 The growth of diabetes has occurred unevenly across sociodemographic groups, increasing at a more pronounced rate among
disadvantaged populations, such as racial/ethnic minorities and less educated individuals.9,10,14 At the same time, two recent analyses found that the prevalence of total and diagnosed diabetes among adults may have stabilized between 2008 and 2012.9,10 Several studies also showed that the share of all diabetes cases that are undiagnosed has declined substantially since the 1980s.13 15 Taken together, existing trend analyses indicate that although important strides have been made, diabetes still remains an urgent public health concern, especially for disadvantaged populations. From the Population Studies Center, Institute for Social Research, University of Michigan, Ann Arbor, Michigan Address correspondence to: Michael Fang, MA, Population Studies Center, Institute for Social Research, University of Michigan, 426 Thompson Street, Ann Arbor MI 48106. E-mail: [email protected]. 0749-3797/$36.00 https://doi.org/10.1016/j.amepre.2018.05.018
© 2018 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.
Am J Prev Med 2018;000(000):1 9
1
ARTICLE IN PRESS 2
Fang / Am J Prev Med 2018;000(000):1 9
The primary aim of this study is to update national trends in diabetes among U.S. adults. Data were from the National Health and Nutrition Examination Survey (NHANES), collected from 1999 2000 to 2015 2016. The NHANES is the only national study that collects diabetes information through self-report and through blood samples, making it the only source available for estimating national trends in total diabetes. The analyses considered trends in undiagnosed, diagnosed, and total diabetes in the total population and across different sociodemographic groups.
TAGEDH1METHODSTAGEDN Study Sample Details about the NHANES are available elsewhere.16 18 In brief, the study consists of a series of nationally representative, crosssectional surveys designed to monitor the health of individuals in the U.S. Beginning in 1999, the NHANES began collecting data in continuous, 2-year cycles. During each survey cycle, a nationally representative sample of respondents are selected from the U.S. non-institutionalized, civilian population using a complex, stratified, multistage probability cluster sampling design. After obtaining written informed consent from the participants, data are collected through in-home interviews and visits to a mobile examination center, where health examinations are administered by trained medical staff members. All data collection received approval from the National Center for Health Statistics research ethics review board. For this study, all waves of data from the continuous NHANES were used to estimate trends in the prevalence of diabetes (1999 2000 to 2015 2016). Across these nine waves, response rates ranged from 61% to 84% for in-home interviews and 59% to 80% for the mobile center visits. Of the 92,062 respondents in these data, the author excluded respondents who were aged <20 years (n=42,550), women who were pregnant (n=1,486), and those with missing data for their diabetes status (n=4,654), weight or height (n=767), or sociodemographic characteristics (n=51). These restrictions yielded a final sample of 42,554 respondents.
Measures During visits to the mobile examination center, trained phlebotomists drew samples of blood from respondents following a standardized protocol. These samples were subsequently analyzed to determine respondents’ glycated hemoglobin (HbA1c) levels using high-performance liquid chromatography methods. The laboratory and tools used to measure HbA1c levels varied across survey years.19 HbA1c testing was performed at University of Minnesota from 2007 2008 to 2011 2012 and at the University of Missouri from 2013 2014 to 2015 2016 using the Tosoh A1C 2.2 Plus Glycohemoglobin Analyzer (2007 2008), Tosoh G7 Glycohemoglobin Analyzer (2009 2010 and 2011 12), or Tosoh G8 Glycohemoglobin Analyzer (2011 2012 to 2015 2016).20 A recent study suggests that these changes in laboratory methodology may have increased HbA1c levels,14 and NHANES analytic guidelines advise researchers to compare HbA1c levels across survey years with caution.19 As a sensitivity test, all prevalence rates
were re-estimated using calibrated HbA1c values that account for these laboratory changes (Appendix Table 1, available online). Three measures of diabetes were used for this study. The first was diagnosed diabetes. During the in-home interviews, respondents were asked if they had ever been told by a doctor or health professional that they had diabetes. Those who answered yes were coded as having diagnosed diabetes. The second was undiagnosed diabetes. Consistent with several prior studies,14,21 respondents with HbA1c levels 6.5%1 and who answered no to being diagnosed with diabetes by a health professional were coded as having undiagnosed diabetes. The third was total diabetes, measured as the total number of respondents who had either diagnosed or undiagnosed diabetes. The remaining participants (i.e., those who were never diagnosed with diabetes and who had HbA1c levels <6.5%) were coded as individuals without diabetes. Respondents with missing information for either diagnosed diabetes status or HbA1c were excluded from the analysis. Height and weight were measured using standardized techniques and equipment during mobile examination visits. These measures were used to calculate respondents’ BMI (calculated by dividing weight in kilograms by the square of height in meters). Those with BMI <25 were categorized as normal, those with BMI 25 but <30 were categorized as overweight, and those with BMI 30 were categorized as obese. A standardized questionnaire was administered during the inhome visits to collect sociodemographic information from respondents, including sex (male, female), race/ethnicity (nonHispanic white, non-Hispanic black, Mexican American), age (20 44, 45 64, 65 years), and education (less than high school, high school graduate, some college, and college graduate). All sociodemographic measures were self-reported.
Statistical Analysis This paper began by estimating the prevalence of diabetes across all nine waves of the continuous NHANES, from 1999 2000 to 2015 2016. Estimates were for the overall adult population and stratified by sex, age, race/ethnicity, educational level, and BMI category. Race-specific prevalence was only estimated for Mexican Americans, non-Hispanic blacks, and non-Hispanic whites due to small sample size for other racial/ethnic groups during the study period. To test for time trends between 1999 2000 and 2015 2016, logistic regression models were estimated. In these models, diabetes status was regressed onto survey year, which was entered as a continuous variable. A statistically significant coefficient for the survey year variable was interpreted as a change in the prevalence of diabetes. Quadratic and cubic time trends were tested and found to be nonsignificant for the overall population and for different subgroups, therefore, linear terms were used to assess temporal trends. The author then proceeded to investigate the factors that might explain changes in diabetes trends. To do this, a series of logistic regression models were estimated examining the relationship between survey cycle, entered as a continuous linear variable, and total diabetes. An unadjusted model was used and then potential mediators were added, including age, BMI category, sex, race/ethnicity, and education, in a stepwise fashion. The coefficient for the survey cycle in the adjusted models was compared with the coefficient in the www.ajpmonline.org
& 2018
Table 1. Weighted Characteristics of U.S. Adults Aged 20 Years or Older, by Diabetes Status: NHANES 1999 2000 to 2015 2016 Characteristics
2001 2002
2003 2004
2005 2006
2007 2008
2009 2010
2011 2012
2013 2014
2015 2016
51.6 48.4
52.6 47.4
48.7 51.3
46.6 53.4
49.8 50.2
52.8 47.2
50.5 49.5
50.7 49.3
54.1 45.9
16.2 47.4 36.5
21.3 45.4 33.3
14.9 46.2 38.9
17.3 45.6 37.1
13.3 49.5 37.2
13.8 45 41.1
16.2 49.1 34.7
13.6 48.1 38.2
15.9 46.2 37.9
42.8 27.6 22.5 7.2
31.1 23.5 27.1 18.3
30 23.6 28.2 18.1
26.5 29 27.4 17.2
31.2 27.5 28.1 13.2
28.9 21.5 30.8 18.8
26.5 25.5 28.3 19.8
22.3 24 36.2 17.5
21.9 21.3 33.6 23.3
8.1 74.3 17.6
8.8 72.3 18.9
9.7 74.8 15.5
10.4 70.2 19.5
9.2 70.7 20.1
12.8 69.2 17.9
10.9 67.5 21.6
11.6 72.1 16.3
14.8 67.2 18
14.3 32.1 53.6
14.4 31.7 54
14.3 27.8 58
13.4 28.5 58.1
12.7 24 63.4
9.7 24.7 65.6
14 23.4 62.6
9.8 27.7 62.6
10.6 25.3 64.1
48.9 51.1
48.6 51.4
49 51
49.1 50.9
48.3 51.7
48.1 51.9
48.3 51.7
48.4 51.6
47.6 52.4
56.8 29 14.2
54 33.3 12.7
52.5 32.6 14.9
49.9 34.8 15.2
49.9 35.8 14.3
50.3 34.9 14.8
48.6 36 15.4
49.5 34.4 16.1
47.4 34.6 18.1
22.6 26.1 27.8 23.5
17.7 25.5 29.9 26.9
16.6 27.5 31.9 23.9
16.3 24.6 31.8 27.4
18.8 24.7 29.3 27.1
17.7 23.1 30.3 28.8
15 19.5 32.5 33
14.3 21.4 32.1 32.2
13 20.7 32.7 33.6 (continued on next page)
ARTICLE IN PRESS
1999 2000
Fang / Am J Prev Med 2018;000(000):1 9
Adults with diabetesa Sex Male Female Age group 20 44 years 45 64 years 65 years Education Less than HS HS completed Some college College graduate Race/ethnicity Mexican-American NH white NH black BMI categoryb Normal Overweight Obese Adults without diabetes Sex Male Female Age group 20 44 years 45 64 years 65 years Education Less than HS HS completed Some college College graduate Race/ethnicity
3
ARTICLE IN PRESS 4 Note: Values are percentages. Boldface indicates statistical significance (p<0.05) x2 test. HS, high school; NH, non-Hispanic; NHANES, National Health and Nutrition Examination Survey. a Those with diabetes have either been diagnosed with diabetes by a doctor or other health professional or have glycated hemoglobin levels 6.5%. b Respondents with BMI (weight in kilograms divided by height in meters squared) <25 were coded as normal, those with BMI >25 and <30 were coded as overweight, and those with BMI >30 were coded as obese.
30.6 33.1 36.3 31.9 33.6 34.5 36.5 35.7 27.7 38 34 28
35.4 35.3 29.2
34.9 33.3 31.8
34.2 36.1 29.7
33.4 34.3 32.3
33.1 35.3 31.7
9.9 77.9 12.1 10.4 77.6 12 8.6 79.7 11.6 7.8 81.4 10.8
Mexican-American NH white NH black BMI categoryb Normal Overweight Obese
7.1 81.6 11.3
8.2 80.5 11.4
8.3 80.5 11.2
9.3 80.2 10.5
9.3 79.1 11.6
2013 2014 2011 2012 2009 2010 2007 2008 2005 2006 2003 2004 2001 2002 1999 2000 Characteristics
Table 1. Weighted Characteristics of U.S. Adults Aged 20 Years or Older, by Diabetes Status: NHANES 1999 2000 to 2015 2016 (continued)
2015 2016
Fang / Am J Prev Med 2018;000(000):1 9 unadjusted model to assess the extent to which these covariates explained trends in diabetes prevalence over time. All analyses were conducted using Stata, version 14.0. To adjust for complex survey design effects (e.g., oversampling of certain populations), nonresponse bias, and post-stratification, all analyses were weighted using recommended survey weights, making the results here representative of the adult, civilian, non-institutionalized U.S. population.22 A two-sided p-value of <0.05 was considered to indicate statistical significance. Because data from the NHANES are de-identified and publicly available, the University of Michigan’s IRB determined that this study was exempt from review. All analyses were performed in 2017.
TAGEDH1RESULTSTAGEDN Table 1 presents the characteristics of U.S. adults aged 20 years with and without diabetes. The composition of those with diabetes remained stable during the study period, with the exception of changes in education. The share of adults with diabetes that did not finish high school or only attained a high school degree declined by 21 and 6 percentage points, respectively, whereas those who had some college education or graduated from college increased by 11 and 16 percentage points (p<0.001). Among those without diabetes, a growing proportion of adults were older (p=0.001), had some college education or graduated from college (p<0.001), and were obese (p<0.001). Table 2 presents the prevalence of diabetes in the overall population and across subgroups. Between 1999 2000 and 2015 2016, the prevalence of total diabetes increased by 5.6 percentage points among the overall adult population, going from 7.7% to 13.3% (p<0.001 for trend; Figure 1). Expressed in relative terms, there was a 72.0% increase in the prevalence of diabetes among U.S. adults during this 18-year period. The increased prevalence of total diabetes was largely a function of growth in diagnosed diabetes. During the study period, the share of American adults diagnosed with diabetes grew 5.3 percentage points (6.0% to 11.3%, p<0.001 for trend; Figure 1), or 88.3%. On the other hand, the unadjusted prevalence of undiagnosed diabetes remained stable during this period (p=0.058). Because the prevalence of diagnosed diabetes increased while the prevalence of undiagnosed diabetes remained roughly unchanged, the share of total diabetes that was undiagnosed decreased significantly from 1999 2000 to 2015 2016 (22% to 15.3%, p<0.013 for trend). Estimates and trends were generally similar using calibrated values of HbA1c (Appendix Table 1, available online). Growth in the prevalence of total diabetes was observed for every subgroup examined, though the rate of change varied considerably. Notable disparities were observed across race/ethnicity, BMI category, and age www.ajpmonline.org
& 2018
Table 2. Weighted Prevalence of Total Diabetesa per 100 Adults Aged 20 Years or Older in the U.S.: NHANES 1999 2000, 2015 2016 Prevalence (95% CI) Characteristics Overall population Total diabetes Diagnosed diabetesb Undiagnosed diabetesc
Age group 20 44 years 45 64 years 65 years Education Less than HS HS graduate Some college College graduate Race/ethnicitye Mexican American NH white NH black BMI categoryf Non-obese
<0.001
7.7 (6.4, 9.1) 6 (4.9, 7.1) 1.7 (1.3, 2.1) 22 (18.6, 25.5)
8.1 (6.9, 9.3) 6.6 (5.7, 7.5) 1.5 (1.1, 2.0) 18.8 (14.5, 23.1)
9.7 (8.4, 11.0) 7.7 (6.4, 9.0) 2 (1.5, 2.4) 20.4 (15.6, 25.3)
9.6 (8.3, 10.8) 7.9 (6.9, 8.8) 1.7 (1.0, 2.3) 17.6 (11.7, 23.4)
10.9 (9.4, 12.4) 8.5 (7.0, 9.9) 2.4 (1.9, 2.9) 22.2 (17.7, 26.8)
10.8 (9.8, 11.8) 8.4 (7.4, 9.4) 2.4 (2.1, 2.7) 22.2 (19.0, 25.5)
11.4 (9.9, 12.9) 9.2 (8.0, 10.5) 2.2 (1.6, 2.7) 19 (15.0, 23.0)
12.1 (11.2, 13.1) 10.1 (9.1, 11.2) 2 (1.5, 2.5) 16.4 (12.4, 20.4)
13.3 (11.4, 15.2) 11.3 (9.7, 12.8) 2 (1.5, 2.6) 15.3 (12.4, 18.3)
8.1 (6.3, 9.9) 7.3 (5.9, 8.8)
8.7 (7.3, 10.1) 7.5 (6.3, 8.7)
9.7 (8.2, 11.1) 9.8 (8.4, 11.2)
9.1 (7.7, 10.5) 10 (8.1, 11.9)
11.2 (9.8, 12.6) 10.6 (8.5, 12.7)
11.7 (10.3, 13.2) 9.9 (8.7, 11.1)
11.9 (10.0, 13.7) 11 (9.2, 12.7)
12.6 (11.2, 14.1) 11.7 (10.5, 12.8)
14.9 (12.2, 17.5) 11.8 (10.0, 13.7)
<0.001
2.3 (1.6, 3.1) 12 (9.5, 14.6) 17.7 (14.4, 21.1)
3.4 (2.2, 4.5) 10.7 (8.3, 13.2) 18.7 (16.8, 20.7)
3 (2.2, 3.7) 13.2 (11.6, 14.8) 22 (17.7, 26.3)
3.5 (2.5, 4.5) 12.2 (10.1, 14.3) 20.4 (18.1, 22.7)
3.2 (2.3, 4.0) 14.5 (12.0, 17.0) 24.1 (20.0, 28.2)
3.2 (2.3, 4.1) 13.5 (11.8, 15.2) 25.2 (22.2, 28.2)
4.1 (3.1, 5.2) 15 (12.7, 17.2) 22.4 (19.4, 25.5)
3.7 (3.0, 4.3) 16.2 (13.6, 18.8) 24.7 (22.6, 26.9)
4.9 (3.8, 6.0) 17 (14.3, 19.7) 24.4 (19.8, 28.9)
<0.001
13.7 (10.7, 16.6) 8.1 (6.5, 9.8) 6.3 (3.6, 9.1) 2.5 (1.7, 3.3)
13.4 (11.4, 15.4) 7.5 (6.1, 8.9) 7.4 (5.7, 9.1) 5.6 (3.5, 7.7)
16.3 (13.5, 19.1) 8.5 (7.1, 9.9)
16.8 (14.9, 18.8) 12 (8.8, 15.2) 10.5 (8.9, 12.1) 5.6 (4.0, 7.2)
16.5 (14.5, 18.5) 10.1 (8.3, 11.9) 11 (9.2, 12.8) 7.3 (5.5, 9.2)
18.5 (15.5, 21.5) 14.4 (10.9, 17.9) 10.1 (8.4, 11.7) 7.2 (4.9, 9.4)
17.7 (15.5, 19.9) 13.4 (11.4, 15.5) 13.5 (11.2, 15.8) 7 (5.4, 8.6)
20.5 (17.6, 23.3) 13.6 (11.3, 15.9) 13.6 (10.7, 16.6) 9.6 (7.5, 11.8)
<0.001
8.7 (6.4, 11.0) 7.5 (5.6, 9.5)
14.7 (12.2, 17.1) 11.1 (8.1, 14.0) 8.3 (6.7, 10.0) 6.2 (4.4, 8.1)
8.3 (5.9, 10.7) 6.7 (5.2, 8.2) 10.9 (9.4, 12.5)
8.5 (7.6, 9.3) 6.8 (5.9, 7.6) 12.6 (10.1, 15.0)
10.8 (7.7, 14.0) 8.7 (7.1, 10.2) 12.2 (10.1, 14.4)
11.4 (9.6, 13.3) 8.2 (6.7, 9.8) 15.1 (12.6, 17.6)
10.7 (8.8, 12.7) 9.7 (7.3, 12.1) 18.9 (17.2, 20.5)
14 (10.7, 17.2) 9.3 (7.9, 10.7) 15.3 (13.2, 17.4)
13.2 (9.6, 16.8) 9.3 (7.6, 11.0) 18.4 (14.9, 21.9)
13.3 (10.7, 15.8) 11.3 (10.0, 12.6) 15.7 (14.2, 17.2)
18.4 (13.2, 23.6) 11.5 (9.5, 13.6) 18.3 (15.6, 21.0)
<0.001
<0.001 0.058 0.013
<0.001
<0.001 <0.001
<0.001 <0.001 <0.001
<0.001 <0.001
0.032
5 (continued on next page)
ARTICLE IN PRESS
Female
p for linear trend
Fang / Am J Prev Med 2018;000(000):1 9
Percentage of diabetes cases that were undiagnosedd Sex Male
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
ARTICLE IN PRESS 6
Obese
Overweight
Note: Boldface indicates statistical significance (p<0.05). HbA1c, glycated hemoglobin; HS, high school; NH, non-Hispanic; NHANES, National Health and Nutrition Examination Survey. a Total diabetes refers to respondents who have either been diagnosed with diabetes by a doctor or other health professional or have HbA1c levels 6.5%. b Diagnosed diabetes refers to respondents who have been told by a doctor or other health professional that they have diabetes. c Undiagnosed diabetes refers to respondents who have not been told by a doctor or other health professional that they have diabetes but have HbA1c levels 6.5%. d Estimates and CIs were computed using respondents who had any type of diabetes (i.e., diagnosed or undiagnosed) rather than the entire adult sample. Figures were weighted using the recommended sample weight. e Race-specific prevalence was not computed for respondents from other racial/ethnic backgrounds due to small sample size. However, their information was included for all other estimates. f Respondents with BMI (weight in kilograms divided by height in meters squared) <25 were coded as normal, those with BMI >25 and <30 were coded as overweight, and those with BMI >30 were coded as obese.
<0.001
0.006
5 (3.6, 6.5) 10.5 (8.2, 12.8) 21.3 (18.6, 24.0) 4 (2.9, 5.2) 10.2 (8.4, 12.1) 20 (18.5, 21.5) 5.2 (3.5, 6.8) 7.9 (6.9, 8.8) 20.3 (17.3, 23.3) 3.4 (2.6, 4.2) 8 (6.0, 10.0) 19.7 (17.8, 21.7) 4.3 (3.1, 5.6) 7.5 (6.5, 8.6) 20.7 (17.3, 24.1) 3.9 (3.0, 4.8) 8.3 (6.6, 10.0) 16.2 (13.9, 18.5) 4.2 (2.1, 6.2) 7.8 (5.9, 9.6) 17.6 (15.2, 19.9) 3.3 (2.2, 4.5) 7.2 (5.8, 8.6) 14.6 (12.2, 17.0) 3.1 (1.9, 4.2) 7.3 (5.0, 9.6) 13.8 (11.9, 15.8)
Prevalence (95% CI)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Characteristics
a
Table 2. Weighted Prevalence of Total Diabetes per 100 Adults Aged 20 Years or Older in the U.S.: NHANES 1999 2000, 2015 2016 (continued)
p for linear trend
Fang / Am J Prev Med 2018;000(000):1 9
Figure 1. Prevalence of diagnosed, undiagnosed, and total diabetes per 100 adults among U.S. adults aged 20 years and older from 1999 2000 to 2015 2016. Note: Diagnosed diabetes is measured through self-report, undiagnosed diabetes is measured as never being diagnosed with diabetes but having HbA1c levels 6.5%, and total diabetes is measured as the sum of diagnosed and undiagnosed diabetes. Data come from the National Health and Nutrition Examination Survey. HbA1c, glycated hemoglobin.
group. The prevalence of total diabetes grew by 4.8 percentage points for non-Hispanic whites (6.7% to 11.5%, p<0.001 for trend), compared with 7.4 percentage points for non-Hispanic blacks (10.9% to 18.3%, p<0.001 for trend) and 10.1 percentage points for Mexican Americans (8.3% to 18.4%, p<0.001 for trend). For non-obese adults, the prevalence of total diabetes rose by 1.9 percentage points (3.1% to 5.0%, p=0.032 for trend), compared with 7.5 percentage points for those who were obese (13.8% to 21.3%, p<0.001 for trend). Across age groups, the prevalence of total diabetes among adults aged 20 44 years increased by 2.6 percentage points (2.3% to 4.9%, p<0.001 for trend), 5 percentage points among adults aged 45 64 years (12% to 17%, p<0.001 for trend) and 6.7 percent points among adults aged 65 years (17.7% to 24.4%, p<0.001 for trend). On the other hand, differences across sex and educational level were less pronounced. For instance, the prevalence of total diabetes increased by 4.5 percentage points for females (7.3% to 11.8%, p<0.001 for trend) and 6.8 percent points for males (8.1% to 14.9%, p<0.001 for trend). Estimates and trends across subgroups were generally similar using calibrated values of HbA1c (Appendix Table 1, available online). Table 3 shows the results from logistic regression models examining the relationship between survey cycle and total diabetes among U.S. adults. The unadjusted model indicated that each one-unit increase in survey cycle (equivalent to 2 years) was associated with a 7.2% increase in the odds of having diabetes for U.S. adults (Model 1: OR=1.07, 95% CI=1.05, 1.09, p<0.001). After adjusting for changes in the age distribution of the www.ajpmonline.org
ARTICLE IN PRESS Fang / Am J Prev Med 2018;000(000):1 9
Table 3. Unadjusted and Adjusted Association Between Survey Cycle and Total Diabetesa: NHANES 1999 2016 Model Model 1b Model 2c Model 3d Model 4e
OR (95% CI) 1.072 (1.051, 1.094) 1.057 (1.035, 1.080) 1.043 (1.021, 1.065) 1.044 (1.025, 1.063)
Note: Boldface indicates statistical significance (p<0.05). HbA1c, glycated hemoglobin; NHANES, National Health and Nutrition Examination Survey. a Total diabetes refers to respondents who have either been diagnosed with diabetes by a doctor or other health professional or have HbA1c levels 6.5%. b Model 1 is unadjusted. c Model 2 adjusts for age. d Model 3 adjusts for age and BMI category. e Model 4 adjusts for age, BMI category, sex, race/ethnicity, and education.
population, the growth in diabetes over time was reduced by around 21% (Model 2: OR=1.06, 95% CI=1.04, 1.08, p<0.001). Adjusting for both changes in age and BMI in the population accounted for roughly 40% of the growth of diabetes during the study period (Model 3: OR=1.04, 95% CI=1.02, 1.07, p<0.001). Adjusting for changes in sex, race, and educational level account did not further explain the growth in the prevalence of total diabetes.
TAGEDH1DISCUSSIONTAGEDN From 1999 2000 to 2015 2016, the unadjusted prevalence of diabetes increased by 72.0% among adults in the U.S., reaching an all-time high of 13.3 diagnoses per every 100 adults in 2015 2016. The growth in diabetes was found across all subgroups, though the rate of change was notably higher among older adults, racial minorities, and those who were obese. The increase in the prevalence of total diabetes among the Mexican American population was particularly pronounced, rising by 10.1 percentage points during the 18-year study period. The increased prevalence of total diabetes was driven to a large extent by rising rates of obesity and the aging of the U.S. adult population. It is important to note that this study used a different measure of undiagnosed diabetes than the one commonly found in CDC reports and other studies using the NHANES. More specifically, although this paper relied on HbA1c, past studies have combined HbA1c readings with fasting blood glucose and oral glucose tolerance readings to more fully assess undiagnosed diabetes.9,11,13,14 Unfortunately, these latter two measures were not yet available for the 2015 2016 wave of the NHANES at the time of writing this paper. As a result, the exact estimates here are not directly comparable with these past studies, as the & 2018
7
prevalence of undiagnosed diabetes and by extension, total diabetes, are likely underestimated. Despite measurement differences, the overall growth in prevalence rates documented here are still substantively consistent with other trend analyses using the more comprehensive measure of total diabetes. For instance, one paper used the more comprehensive total diabetes measure and found that the prevalence of total diabetes increased by 3.5 percentage points from 1999 to 2012, while this paper documented a 3.7 percentage point increase during the same time frame.9 Moreover, the results here also align well with other studies that define total diabetes using selfreported information and HbA1c. For instance, the analyses by Cowie et al.23 of the NHANES data showed the prevalence of total diabetes grew 7.4% in 1988 1994 to 9.3% in 2003 2006. Similarly, using the NHANES, Selvin and colleagues14 show that the prevalence of total diabetes increased from 6.2% in 1988 1994 to 9.9% in 2005 2010. Considered together, these analyses indicate that diabetes remains a pervasive and growing public health concern among U.S. adults. The high growth of diabetes among Mexican-American adults found in this study is also consistent with the literature on racial/ethnic disparities in diabetes. One recent study found that the prevalence of diabetes among Mexican-American adults grew by 6 percentage points from 1999 to 2012, compared with 2 percentage points for white adults.9 A second study showed that from 1988 1994 to 2005 2010, the prevalence of diabetes increased 4.1 percentage points for Mexican-Americans adults and 2.8 for whites adults.14 A third paper found that between 2001 and 2009, the rate of growth in the prevalence of type 2 diabetes was ten times higher in Hispanic youth compared with non-Hispanic white youth.24 This suggests that the diabetes epidemic has—and may continue to have—a disproportionate impact on the Hispanic population. The possible mechanisms underlying the rising rates of diabetes among U.S. adults identified in this study, population aging and increased rates of obesity, are also consistent with prior studies. For instance, one study found that controlling for age, race, and education explained 22% 28% of the growth in diabetes from 1988 to 2010,14 whereas another found that age accounted for 9% and 22% of the increase in diabetes from 1976 to 2010 among American men and women, respectively.25 Similarly, BMI has been identified in multiple studies as being one of the most important risk factors for developing type 2 diabetes.26,27 A study using the NHANES showed that controlling for changes in BMI and social characteristics among U.S. adults completely attenuated the growth of diabetes from 1999 to 2010.14 Two community studies also showed that roughly half of new cases of type 2 diabetes among respondents could have been prevented if they were not
ARTICLE IN PRESS 8
Fang / Am J Prev Med 2018;000(000):1 9 28,29
obese or overweight. A longitudinal study based on roughly 85,000 U.S. women echoed this finding, showing that roughly 60% of new type 2 diabetes cases were attributable to respondents being overweight.27 Although the increasing prevalence of total diabetes is concerning from a public health perspective, it was encouraging to find that undiagnosed diabetes remained largely unchanged during the study period. Indeed, the growth in total diabetes was largely a function of increasing rates of diagnosed diabetes. Because of this, the share of diabetes cases that was undiagnosed declined significantly over time. This trend is consistent with prior studies and suggests that improvements in diabetes awareness and screening are likely resulting in better detection of diabetes among U.S. adults.9,13 15,21
Limitations The main strength of this study was the use of a large, nationally representative data set (NHANES), which allowed for accurate estimates of the prevalence of diabetes among U.S. adults over time. Nevertheless, this study had several important limitations. First, as mentioned, the paper was only able to assess undiagnosed diabetes using HbA1c information due to data limitations. Second, although this study found a large increase in the prevalence of diabetes for Mexican-American adults, it is important to note that the estimate for 2015 2016 had relatively wide CIs. The imprecision of this figure is important because the difference in prevalence rates in 2015 2016 and 1999 2000 were used to determine the rate of change over time. Third, the NHANES generally takes a single HbA1c reading from respondents. The American Diabetes Association recommends that individuals who have HbA1c levels at or above 6.5% be re-tested a second time to confirm the diagnosis.11 Thus, there may be some individuals who are misclassified as having undiagnosed diabetes in this paper. Fourth, although this paper included subgroup analyses, it important to note the NHANES data may have lacked the statistical power to detect differences among smaller groups. Therefore, it is important to interpret these analyses with some caution. Finally, because of limitations with the NHANES data, this paper could not differentiate between respondents with type 1 or type 2 diabetes.
TAGEDH1CONCLUSIONSTAGEDN This study found that the prevalence of diabetes among U.S. adults increased substantially from 1999 to 2016. This growth occurred differentially across populations, particularly impacting Hispanic adults. The increased prevalence of total diabetes was driven largely by population aging and growing obesity rates. These findings suggest that additional
interventions aimed at reducing obesity may be necessary to help slow growth of diabetes within the U.S. population.
TAGEDH1ACKNOWLEDGMENTSTAGEDN The author thanks Dr. Elizabeth Selvin for sharing her NHANES data set that contained calibrated values for glycated hemoglobin. The author also thanks Dr. Susan Fang for her comments on an early draft of this paper. The author confirms that there are no known conflicts of interest associated with this publication and there has been no financial support for this work that could have influenced its outcome. No financial disclosures were reported by the author of this paper.
TAGEDH1SUPPLEMENTAL MATERIALTAGEDN Supplemental materials associated with this article can be found, in the online version, at doi:10.1016/j.amepre.2018.05.018.
TAGEDH1REFERENCESTAGEDN 1. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 37(suppl 1):S81–S90. https://doi.org/ 10.2337/dc14-S081. 2. Emerging Risk Factors Collaboration. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet. 375(9733):2215– 2222. https://doi.org/10.1016/S0140-6736(10)60484-9. 3. Afkarian M, Zelnick LR, Hall YN, et al. Clinical manifestations of kidney disease among U.S. adults with diabetes, 1988 2014. JAMA. 2016;316(6):602–610. https://doi.org/10.1001/jama.2016.10924. 4. Rawshani A, Rawshani A, Franzen S, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med. 2017;376 (15):1407–1418. https://doi.org/10.1056/NEJMoa1608664. 5. Franco OH, Steyerberg EW, Hu FB, Mackenbach J, Nusselder W. Associations of diabetes mellitus with total life expectancy and life expectancy with and without cardiovascular disease. Arch Intern Med. 2007;167(11):1145– 1151. https://doi.org/10.1001/archinte.167.11.1145. 6. Kobrin Klein BE. Overview of epidemiologic studies of diabetic retinopathy. Ophthalmic Epidemiol. 2007;14(4):179–183. https://doi.org/ 10.1080/09286580701396720. 7. Tancredi M, Rosengren A, Svensson A-M, et al. Excess mortality among persons with type 2 diabetes. N Engl J Med. 2015;373 (18):1720–1732. https://doi.org/10.1056/NEJMoa1504347. 8. Dall TM, Yang W, Halder P, et al. The economic burden of elevated blood glucose levels in 2012: diagnosed and undiagnosed diabetes, gestational diabetes mellitus, and prediabetes. Diabetes Care. 2014;37 (12):3172–3179. https://doi.org/10.2337/dc14-1036. 9. Menke A, Casagrande S, Geiss L, Cowie CC. Prevalence of and trends in diabetes among adults in the United States, 1988 2012. JAMA. 2015;314(10):1021–1029. https://doi.org/10.1001/jama.2015.10029. 10. Geiss LS, Wang J, Cheng YJ, et al. Prevalence and incidence trends for diagnosed diabetes among adults aged 20 to 79 years, United States, 1980 2012. JAMA. 2014;312(12):1218–1226. https://doi.org/10.1001/ jama.2014.11494. 11. Cowie CC, Rust KF, Ford ES, et al. Full accounting of diabetes and prediabetes in the U.S. population in 1988 1994 and 2005 2006. Diabetes Care. 2009;32(2):287–294. https://doi.org/10.2337/dc08-1296. 12. Mokdad AH, Ford ES, Bowman BA, et al. Diabetes trends in the U.S.: 1990 1998. Diabetes Care. 2000;23(9):1278–1283. https://doi.org/ 10.2337/diacare.23.9.1278.
www.ajpmonline.org
ARTICLE IN PRESS Fang / Am J Prev Med 2018;000(000):1 9 13. Selvin E, Wang D, Lee AK, Bergenstal RM, Coresh J. Identifying trends in undiagnosed diabetes in U.S. adults by using a confirmatory definition: a cross-sectional study. Ann Intern Med. 2017;167(11):769– 776. https://doi.org/10.7326/M17-1272. 14. Selvin E, Parrinello CM, Sacks DB, Coresh J. Trends in prevalence and control of diabetes in the United States, 1988 1994 and 1999 2010. Ann Intern Med. 2014;160(8):517–525. https://doi.org/10.7326/M13-2411. 15. Gregg EW, Cadwell BL, Cheng YJ, et al. Trends in the prevalence and ratio of diagnosed to undiagnosed diabetes according to obesity levels in the U.S. Diabetes Care. 2004;27(12):2806–2812. https://doi.org/ 10.2337/diacare.27.12.2806. 16. Johnson CL, Dohrmann SM, Burt VL, Mohadjer LK. National Health and Nutrition Examination Survey: sample design, 2011 2014. Vital Health Stat 2. 2014(162):1–33. 17. Curtin LR, Mohadjer LK, Dohrmann SM, et al. National Health and Nutrition Examination Survey: sample design, 2007 2010. Vital Health Stat 2. 2013(160):1–23. 18. Curtin L, Mohadjer L, Dohrmann S, et al. The National Health and Nutrition Examination Survey: sample design, 1999 2006. Vital Health Stat 2. 2012(155):1–39. 19. National Center for Health Statistics. Updated Advisory for NHANES Hemoglobin A1c(Glycohemoglobin) Data. www.cdc.gov/nchs/data/ nhanes/A1c_webnotice.pdf. Published 2012. Accessed October, 2017. 20. CDC. National Health and Nutrition Examination Laboratory Protocol. Hyattsville, MD: U.S. HHS, CDC; 2015 2016. 21. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetes and impaired fasting glucose in adults in the U.S. population. Diabetes Care. 2006;29(6):1263–1268. https://doi.org/10.2337/dc06-0062.
& 2018
9
22. Johnson CL, Paulose-Ram R, Ogden CL, et al. National Health and Nutrition Examination Survey. Analytic guidelines, 1999 2010. Vital Health Stat 2. 2013(161):1–24. 23. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetes and high risk for diabetes using A1C criteria in the U.S. population in 1988 2006. Diabetes Care. 2010;33(3):562–568. https://doi.org/ 10.2337/dc09-1524. 24. Mayer-Davis EJ, Lawrence JM, Dabelea D, et al. Incidence trends of type 1 and type 2 diabetes among youths, 2002 2012. N Engl J Med. 2017;376(15):1419–1429. https://doi.org/10.1056/NEJMoa1610187. 25. Menke A, Rust KF, Fradkin J, Cheng YJ, Cowie CC. Associations between trends in race/ethnicity, aging, and body mass index with diabetes prevalence in the United States: a series of cross-sectional studies increase in diabetes prevalence over time. Ann Intern Med. 2014;161 (5):328–335. https://doi.org/10.7326/M14-0286. 26. Vazquez G, Duval S, Jacobs Jr. DR Jr., Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev. 2007;29(1):115– 128. https://doi.org/10.1093/epirev/mxm008. 27. Hu FB, Manson JE, Stampfer MJ, et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med. 2001;345(11):790– 797. https://doi.org/10.1056/NEJMoa010492. 28. Steinbrecher A, Morimoto Y, Heak S, et al. The preventable proportion of type 2 diabetes by ethnicity: the multiethnic cohort. Ann Epidemiol. 2011;21(7):526–535. https://doi.org/10.1016/j.annepidem.2011.03.009. 29. Jiang Y, Chen Y, Mao Y. CCDPC Obesity Working Group. The contribution of excess weight to prevalent diabetes in Canadian adults. Public Health. 2008;122(3):271–276. https://doi.org/10.1016/j.puhe.2007.06.002.