Abstract
INTRODUCTION:
In 2021, the U.S. Preventive Services Task Force (USPSTF) recommended prediabetes and diabetes screening for asymptomatic adults aged 35-70 years with overweight/obesity, lowering the age from 40 in its 2015 recommendation. USPSTF suggested considering earlier screening in racial and ethnic groups with high diabetes risk at younger ages or lower body mass index (BMI). The current study examined the clinical performance of these USPSTF screening recommendations, as well as alternative age and BMI cutoffs, in the U.S. adult population overall and separately by race and ethnicity.
METHODS:
Nationally representative data were collected from 3,243 non-pregnant adults without diagnosed diabetes in January 2017-March 2020 and analyzed from 2021-2022. Screening eligibility was based on age and measured BMI. Collectively, prediabetes and undiagnosed diabetes were defined by fasting plasma glucose ≥100mg/dL or hemoglobin A1c ≥5.7%. The sensitivity, specificity and predictive values of alternate screening criteria were examined overall, and by race and ethnicity.
RESULTS:
Compared with 2015 criteria, the 2021 criteria exhibited marginally higher sensitivity [58.6% (95% CI: 55.5-61.6) vs. 52.9% (49.7-56.0)] and lower specificity [69.3% (65.7-72.2) vs. 76.4% (73.3-79.2)] overall, and within each racial and ethnic group. Screening at lower age and BMI thresholds resulted in even greater sensitivity and lower specificity, especially among Hispanic, non-Hispanic Black, and Asian adults. Screening all adults aged 35-70 regardless of BMI yielded the most equitable performance across all racial and ethnic groups.
CONCLUSIONS:
The 2021 USPSTF screening criteria will identify more adults with prediabetes and diabetes in all racial and ethnic groups than the 2015 criteria. Screening all 35-70 year-old adults exhibited even higher sensitivity and performed most similarly by race and ethnicity, which may further improve early detection of prediabetes and diabetes in diverse populations.
INTRODUCTION
Approximately half of U.S. adults have type 2 diabetes (hereafter, diabetes) or prediabetes, representing a major public health concern.1 Overall, 81% of adults with prediabetes are not aware of having the condition and 23% of diabetes cases remain undiagnosed.1 Screening for prediabetes and diabetes is the primary method for detecting these conditions and enabling early intervention or treatment, which is associated with improved clinical outcomes.2–4
Certain racial and ethnic groups experience a disproportionate burden of diabetes, with a nearly two-fold higher prevalence among Hispanic, non-Hispanic Black (hereafter, Black), and non-Hispanic Asian (hereafter, Asian) adults compared with non-Hispanic White (hereafter, White) individuals,5 as well as higher rates of poor glycemic control, complications, and diabetes-related mortality.6–10 There is evidence that Black and Hispanic adults develop diabetes at younger ages than White individuals.11,12 Asian Americans, as well as Hispanic and Black individuals, exhibit high diabetes risk at lower levels of body mass index (BMI).13–15 These findings have implications for prediabetes and diabetes screening in the same racial and ethnic groups, where earlier detection could accelerate evidence-based care and improve outcomes, thereby promoting health equity.
In 2021, the United States Preventive Services Task Force (USPSTF) recommended screening for prediabetes and diabetes among asymptomatic adults aged 35-70 years with overweight or obesity, defined by BMI ≥25kg/m2.16 This guideline lowered the screening age from 40 years in the 2015 USPSTF recommendation,17 a change that expands eligibility from 36.3% to 43.0% of U.S. adults.18 The current USPSTF recommendation also suggests that clinicians consider screening at a lower BMI or at even younger ages among racial and ethnic minority groups. There are tradeoffs associated with any age or BMI threshold chosen, resulting in some individuals with prediabetes or undiagnosed diabetes who will not be eligible for screening, and others without either condition who will be eligible. These tradeoffs, evaluated by comparing the sensitivity, specificity, and predictive values of alternate screening criteria, represent critical data needed by clinicians and other stakeholders that have not been examined previously.
These clinical performance characteristics of the 2015 and 2021 USPSTF screening criteria were compared in a nationally representative sample of U.S. adults, stratifying by race and ethnicity to examine health equity implications of the recent guideline change. To explore fully the potential of the current recommendation to promote health equity, the authors also examined lowering thresholds for age and BMI suggested by USPSTF to promote early detection of prediabetes and diabetes among racial and ethnic minority groups. Screening performance was estimated separately based on age 30-70 years and 18-70 years, as well as BMI ≥23kg/m2 and any BMI.
METHODS
Study Sample
The data source was the National Health and Nutrition Examination Surveys (NHANES) from January 2017-March 2020, with a response rate of 46.9%.19 All examined participants completed a blood sample that included hemoglobin A1c (A1c), of which approximately half completed a fasting sample that included fasting plasma glucose (FPG).20 Protocols for collecting interview data and biologic specimens in NHANES are described in-depth elsewhere.21 NHANES uses a complex, multistage probability sample design to describe statistics related to the health of the non-institutionalized U.S. civilian population. The study participants were non-pregnant adults aged ≥18 years without a self-reported diabetes diagnosis who underwent fasting blood collection. Those missing data for self-report of clinician-diagnosed diabetes, BMI, or glycemic measures were excluded (n=57). The final analytic sample included 3,243 participants.
Measures
Screening eligibility according to the 2015 and 2021 USPSTF recommendations was based on NHANES variables for age and measured BMI. Following the suggestion to consider screening at younger ages and lower BMI among racial and ethnic minority groups, the authors examined screening criteria that incorporate alternate age cutoffs (i.e., 30-70 years and 18-70 years), as well as alternate BMI thresholds (i.e., ≥23kg/m2 and any BMI). Because USPSTF provided no specific guidance about lower age cutoffs, the authors chose 18 years, representing all younger adults, and 30 years. The latter cutoff represents a 5-year age reduction from the current screening criteria, which is equivalent to the recent guideline change from 40 years to 35 years. Race and ethnicity were self-reported by participants. Racial and ethnic groups sampled in NHANES include Asian (comprising those with origins in the Far East, Southeast Asia, or the Indian subcontinent), Black, Hispanic, and White adults. Prediabetes was defined by FPG 100-125mg/dL or A1c 5.7-6.4%.22 Diabetes was defined by FPG ≥126mg/dL or A1c ≥6.5%.22 The authors also analyzed available data on the following participant characteristics that are associated with an elevated risk of prediabetes and diabetes: sex, waist circumference, educational attainment, household income, insurance status, and having a usual source of care.22,23
Statistical analysis
Descriptive statistics were used to characterize the subsamples of adults eligible for screening according to the 2015 and 2021 USPSTF screening criteria, as well as those newly eligible according to the 2021 recommendation. The authors also described characteristics of U.S. adults with prediabetes and undiagnosed diabetes, whom screening efforts aim to identify. In the full sample, the following performance characteristics of all screening criteria were examined: sensitivity (the proportion of adults with prediabetes or diabetes who meet the criteria), specificity (the proportion of those without prediabetes or diabetes who do not meet criteria), positive predictive value (PPV; the proportion of adults meeting the criteria who have prediabetes or diabetes), and negative predictive value (NPV; the proportion of those not meeting the criteria who are free of prediabetes and diabetes). The performance of all screening criteria was determined in the overall sample and separately within each racial and ethnic group; and the significance of differences between groups were examined using Chi-square tests. Sensitivity analyses estimated performance characteristics of the 2015 and 2021 USPSTF screening criteria using FPG alone and A1c alone to define prediabetes and undiagnosed diabetes. A P-value of <.05 was considered significant for all statistical testing. SAS-callable SUDAAN version 9.4 (Research Triangle Park, NC) was used to conduct statistical analysis using fasting sample weights. All analyses of these de-identified, publicly available data were conducted from November 2021-November 2022.
RESULTS
Of the 3,243 participants, 1,261 [weighted 37.3% (95% CI: 35.0%-39.7%)] were eligible for screening according to the 2015 USPSTF criteria and 1,451 [weighted 43.8% (41.2%-46.3%)] were eligible according to the 2021 criteria. These subsamples represent 80.4 million and 94.3 million U.S. adults respectively, which corresponds to a 14.8% proportionate increase in the total population eligible for screening. The numbers of newly eligible adults expanded across all racial and ethnic groups, corresponding to the following proportionate increases in eligible participants from 2015: Asian (17.9%); Black (13.9%); Hispanic (30.6%); and White (14.0%) (data not shown). The sociodemographic characteristics and clinical measurements of those eligible in 2015 and 2021 were similar. Those newly eligible in 2021 were more likely to self-report Hispanic ethnicity, lower socioeconomic status, and lack of health insurance or a usual source of care. Newly eligible adults also exhibited lower levels of fasting glucose and A1c. (Table 1)
Table 1:
Characteristics of U.S. Adults without Diagnosed Diabetes Eligible for Prediabetes and Diabetes Screening
| Characteristica | USPSTF 2015b | USPSTF 202lc | Newly Eligibled |
|---|---|---|---|
| % (95% CI) | % (95% CI) | % (95% CI) | |
| Unweighted n | 1,261 | 1,451 | 190 |
| Weighted n, millionse | 80.4 | 94.3 | 13.9 |
| Female sex | 49.7 (46.6-52.8) | 49.9 (46.8-52.9) | 50.9 (41.9-59.8) |
| Age, years | 54.6 (53.5-55.6) | 52.0 (50.9-53.1) | 37.1 (36.8-37.4) |
| Age categories, years | |||
| 18-34 | 0.0 (. -.) | 0.0 (. -.) | 0.0 (. -.) |
| 35-39 | 0.0 (. -.) | 14.7 (12.3-17.6) | 100.0 (. -. ) |
| 40-70 | 100.0 (. -. ) | 85.3 (82.4-87.7) | 0.0 (. -.) |
| ≥71 | 0.0 (. -.) | 0.0 (. -.) | 0.0 (. -.) |
| Race and ethnicity | |||
| Asian | 4.0 (2.4-6.3) | 4.0 (2.6-6.0) | 4.1 (2.6-6.5) |
| Black | 12.5 (9.4-16.4) | 12.1 (9.3-15.7) | 10.1 (6.9-14.5) |
| Hispanic | 15.5 (12.5-19.0) | 17.2 (14.0-21.1) | 27.5 (19.6-37.1) |
| White | 65.0 (60.2-69.5) | 63.2 (58.4-67.8) | 52.8 (42.3-63.1) |
| Educational attainment <High School | 12.0 (10.1-14.1) | 12.4 (10.8-14.2) | 14.8 (10.7-20.2) |
| Income below the federal poverty level | 13.2 (9.4-18.1) | 13.9 (10.6-18.0) | 17.9 (10.6-28.7) |
| Uninsured | 12.7 (9.7-16.5) | 13.7 (11.0-16.8) | 19.4 (13.7-26.7) |
| Usual source of care | 87.4 (82.2-91.2) | 86.1 (81.2-89.8) | 78.3 (70.9-84.2) |
| Weight statusf | |||
| Normal | 0.0 (. -.) | 0.0 (. -.) | 0.0 (. -.) |
| Overweight | 50.8 (46.4-55.2) | 49.5 (45.5-53.5) | 41.9 (35.5-48.6) |
| Obesity | 49.2 (44.8-53.6) | 50.5 (46.5-54.5) | 58.1 (51.4-64.5) |
| Body mass index, kg/m2 | 31.6 (31.1-32.2) | 31.9 (31.4-32.4) | 33.5 (32.2-34.8) |
| Waist circumference, cm | 106.0 (104.8-107.3) | 106.1 (104.8-107.4) | 106.6 (103.6-109.5) |
| Fasting plasma glucose, mg/dLg,h | |||
| 100-109 | 40.9 (36.0-46.0) | 39.5 (34.9-44.3) | 33.9 (25.5-43.3) |
| 110-125 | 18.3 (14.3-23.2) | 16.6 (13.0-21.0) | 7.0 (3.1-14.9) |
| ≥126 | 5.6 (4.1-7.5) | 5.4 (4.2-7.0) | 4.6 (1.9-10.7) |
| Mean | 106.9 (105.8-107.9) | 106.7 (105.6-107.8) | 105.7 (99.7-111.7) |
| Hemoglobin A1c, %g,h | |||
| 5.7-5.9 | 25.1 (21.9-28.5) | 23.4 (20.2-26.9) | 14.0 (9.1-20.9) |
| 6.0-6.4 | 12.5 (10.0-15.5) | 11.7 (9.6-14.2) | 7.1 (3.1-15.4) |
| ≥6.5 | 4.0 (3.2-5.0) | 3.9 (3.2-4.7) | 3.1 (1.2-8.0) |
| Mean | 5.62 (5.58-5.66) | 5.60 (5.56-5.64) | 5.40 (5.32-5.57) |
USPSTF = United States Preventive Services Task Force
Categorical variables are expressed as weighted column percentages (95% CI). Body mass index, waist circumference, and glycemic testing results are expressed as mean (95% CI).
The 2015 USPSTF screening criteria were age 40-70 years and BMI ≥25kg/m2.
The 2021 USPSTF screening criteria are age 35-70 years and BMI ≥25kg/m2.
Those newly eligible are aged 35-39 years with BMI ≥25kg/m2.
These numbers represent the population size among all U.S. adults.
Overweight and obesity status were determined based on a measured BMI of ≥25kg/m2 and ≥30kg/m2, respectively, for all racial and ethnic groups. Normal weight status was defined by a BMI <25kg/m2.
Glycemic measures are expressed as weighted column percentages (95% CI) within each specified range of test values, followed by the mean value for each glycemic test.
Some estimates within strata of glycemic values may not be statistically reliable due to small sample sizes.
Overall, 46.8% of U.S. adults without diagnosed diabetes had prediabetes or undiagnosed diabetes, representing 100.7 million U.S. adults. Of those with prediabetes, 18.8% were aged less than 35 years and 14.4% were greater than 70 years old. A smaller proportion of participants with undiagnosed diabetes were under 35 years old (6.4%), and a comparable share were older than 70 (14.4%). (Table 2) While the racial and ethnic composition of participants with prediabetes reflected the overall U.S. adult population, minority groups were overrepresented among those with undiagnosed diabetes. The characteristics of U.S. adults with prediabetes and undiagnosed diabetes were similar when defined by FPG alone and A1c alone (Appendix Table 1).
Table 2:
Characteristics of U.S. Adults with Prediabetes and Undiagnosed Diabetes
| Characteristica | Prediabetes | Undiagnosed Diabetes | ||
|---|---|---|---|---|
| Population sizeb | % (95% CI) | Population sizeb | % (95% CI) | |
| Unweighted n | 1,478 | 176 | ||
| Weighted n | 92.5 | 43.0 (40.2-45.8) | 8.2 | 3.8 (3.1-4.7) |
| Female sex | 42.8 | 46.2 (42.5-50.0) | 4.8 | 59.2 (45.5-71.6) |
| Mean age, years | 92.5 | 52.2 (50.6-53.8) | 8.2 | 55.7 (52.4-59.0) |
| Age categories, years | ||||
| 18-29 | 11.1 | 12.1 (9.9-14.7) | 0.4 | 4.3 (1.4-11.9) |
| 30-34 | 6.2 | 6.7 (4.3-10.4) | 0.2 | 2.2 (0.7-6.5) |
| 35-39 | 6.3 | 6.8 (5.4-8.5) | 0.6 | 7.5 (3.4-15.9) |
| 40-70 | 55.5 | 60.0 (55.8-64.0) | 5.9 | 71.3 (60.6-80.0) |
| ≥71 | 13.3 | 14.4 (11.5-17.8) | 1.2 | 14.4 (8.0-24.6) |
| Race and ethnicity | ||||
| Asian | 5.2 | 5.7 (3.9-8.1) | 0.8 | 9.2 (4.5-17.7) |
| Black | 10.8 | 11.7 (8.7-15.6) | 1.3 | 15.9 (9.4-25.5) |
| Hispanic | 13.4 | 14.5 (11.4-18.1) | 1.7 | 20.6 (14.4-28.6) |
| White | 58.9 | 63.7 (59.6-67.5) | 4.1 | 49.9 (35.1-64.7) |
| Educational attainment <High School | 10.8 | 11.8 (10.2-13.7) | 1.4 | 16.8 (10.9-25.0) |
| Income below the federal poverty level | 10.3 | 12.4 (9.7-15.8) | 1.1 | 15.2 (8.6-25.5) |
| Uninsured | 11.3 | 12.3 (9.4-15.8) | 1.7 | 21.2 (12.7-33.2) |
| Usual source of care | 78.6 | 84.9 (81.4-87.9) | 7.3 | 89.1 (82.5-93.4) |
| Weight statusc | ||||
| Normal | 16.3 | 17.6 (14.5-21.3) | 0.8 | 9.6 (4.0-21.3) |
| Overweight | 33.2 | 35.9 (32.2-39.8) | 1.5 | 18.9 (11.8-29.0) |
| Obesity | 43.0 | 46.5 (42.5-50.5) | 5.9 | 71.5 (60.6-80.4) |
| Body mass index, kg/m2 | 92.5 | 30.7 (30.2-31.3) | 8.2 | 34.9 (32.3-37.5) |
| Waist circumference, cm | 90.5 | 103.5 (102.2-104.8) | 8.2 | 112.4 (107.4-117.4) |
| Fasting plasma glucose, mg/dL | 92.5 | 108.4 (107.6-109.1) | 8.2 | 151.2 (140.4-162.0) |
| Hemoglobin A1c, % | 92.5 | 5.63 (5.59-5.67) | 8.2 | 6.92 (6.64-7.19) |
Categorical variables are expressed as weighted column percentages (95% CI). Body mass index, waist circumference, and glycemic testing results are expressed as mean (95% CI).
These numbers represent the population size in millions among all U.S. adults.
Overweight and obesity status were determined based on a measured BMI of ≥25kg/m2 and ≥30kg/m2, respectively, for all racial and ethnic groups. Normal weight status was defined by a BMI <25kg/m2.
The 2021 USPSTF screening criteria exhibited marginally higher sensitivity and lower specificity than the 2015 criteria, with similar PPV and NPV. The absolute increases in sensitivity from the 2015 to 2021 criteria were as follows: total population (5.7%); Asian (5.5%); Black (6.6%); Hispanic (11.1%); and White (4.0%) (Table 3). Using the standard recommended BMI cutoff of ≥25kg/m2, sensitivity of both the 2015 and 2021 screening criteria was significantly lower and the specificity significantly higher in Asian individuals than among other racial and ethnic subgroups. The sensitivity and specificity of both sets of criteria were similar in White, Black, and Hispanic adults. Similar trends in guideline performance were observed when using FPG alone (Appendix Table 2) and A1c alone (Appendix Table 3) to define prediabetes and undiagnosed diabetes.
Table 3:
Performance of Screening Criteria among U.S. Adults without Diagnosed Diabetes by Self-Reported Race and Ethnicity
| Screening Criteria and Population group | Sensitivity, % (95% CI) | Specificity, % (95% CI) | Positive predictive value, % (95% CI) | Negative predictive value, % (95% CI) |
|---|---|---|---|---|
| 2015 USPSTF Criteria a | ||||
| Total population | 52.9 (49.7-56.0) | 76.4 (73.3-79.2) | 66.3 (63.0-69.4) | 64.8 (61.0-68.5) |
| Asian | 38.0 (29.1-47.8) | 86.0 (79.9-90.5) | 71.6 (64.0-78.2) | 60.0 (53.5-66.2) |
| Black | 57.0 (50.8-62.9) | 73.5 (67.4-78.8) | 68.9 (62.5-74.6) | 62.4 (56.8-67.6) |
| Hispanic | 50.9 (44.6-57.2) | 75.3 (70.0-80.0) | 61.7 (55.5-67.6) | 66.3 (61.0-71.2) |
| White | 55.1 (51.1-59.0) | 75.8 (70.4-80.5) | 66.4 (61.1-71.3) | 66.0 (60.9-70.8) |
| P-valueb | <0.01 | 0.03 | 0.37 | 0.40 |
| 2021 USPSTF Criteria c | ||||
| Total population | 58.6 (55.5-61.6) | 69.3 (65.7-72.7) | 62.7 (59.3-65.9) | 65.6 (61.7-69.3) |
| Asian | 43.5 (34.1-53.3) | 82.3 (76.7-86.7) | 69.4 (62.8-75.3) | 61.1 (53.8-67.9) |
| Black | 63.6 (57.8-68.9) | 68.4 (62.6-73.8) | 67.5 (62.2-72.4) | 64.6 (59.2-69.7) |
| Hispanic | 62.0 (55.0-68.5) | 64.2 (58.8-69.3) | 57.5 (51.9-62.9) | 68.4 (61.8-74.3) |
| White | 59.1 (55.4-62.7) | 69.3 (63.3-74.6) | 62.5 (57.2-67.6) | 66.1 (61.1-70.8) |
| P-valueb | 0.02 | 0.02 | 0.08 | 0.57 |
USPSTF = USPSTF = United States Preventive Services Task Force
The 2015 USPSTF screening criteria were age 40-70 years and BMI ≥25kg/m2.
P-values for racial and ethnic differences in performance characteristics of the USPSTF screening criteria were determined using Chi-square tests. Boldface indicates statistical significance.
The 2021 USPSTF screening criteria are age 35-70 years and BMI ≥25kg/m2.
Table 4 displays the performance characteristics of alternate screening criteria. Lowering age and BMI cutoffs yielded progressively higher sensitivities and lower specificities in the overall adult population, and within each racial and ethnic subgroup. Lowering the age threshold for screening resulted in the largest increase in sensitivity and greatest decrease in specificity among Hispanic participants. A similar pattern was observed for lowering BMI thresholds, with the highest gain in sensitivity and largest decrease in specificity observed in Asian participants. Screening all adults aged 35-70 years exhibited the most similar sensitivity and specificity across racial and ethnic groups. The characteristics of 35-70-year-old adults, who would be eligible based on this approach, are presented in Appendix Table 4.
Table 4:
Performance of Alternate Screening Criteria by Self-Reported Race and Ethnicity
| Screening Criteria and Population group | Sensitivity, % (95% CI) | Specificity, % (95% CI) | Positive predictive value, % (95% CI) | Negative predictive value, % (95% CI) |
|---|---|---|---|---|
| Lowering Age Cutoff | ||||
| Age 30-70 years + BMI ≥25kg/m2 | ||||
| Total population | 64.0 (60.8-67.1) | 61.7 (58.3-65.0) | 59.5 (55.9-63.0) | 66.1 (62.3-69.8) |
| Asian | 50.7 (40.8-60.5) | 72.2 (63.5-79.4) | 62.8 (55.7-69.4) | 61.2 (53.9-68.1) |
| Black | 68.2 (62.6-73.3) | 59.4 (53.1-65.5) | 63.4 (58.1-68.4) | 64.4 (58.3-70.2) |
| Hispanic | 68.6 (60.9-75.4) | 54.6 (48.4-60.7) | 54.1 (48.7-59.5) | 69.1 (62.1-75.2) |
| White | 63.4 (59.2-67.5) | 62.9 (56.3-69.0) | 59.7 (54.0-65.1) | 66.5 (61.2-71.3) |
| P-valuea | 0.02 | 0.04 | 0.14 | 0.63 |
| Age 18-70 years + BMI ≥25kg/m2 | ||||
| Total population | 72.5 (69.4-75.4) | 45.1 (40.9-49.3) | 53.7 (49.8-57.5) | 65.1 (60.7-69.3) |
| Asian | 58.4 (49.4-67.0) | 59.8 (51.5-67.6) | 57.4 (49.9-64.5) | 60.8 (52.5-68.5) |
| Black | 77.6 (71.4-82.8) | 33.1 (27.3-39.6) | 54.5 (49.2-59.6) | 59.0 (51.7-65.9) |
| Hispanic | 83.4 (76.4-88.7) | 31.5 (25.2-38.6) | 48.8 (44.7-52.8) | 70.9 (60.7-79.4) |
| White | 70.1 (66.3-73.5) | 49.4 (43.5-55.4) | 54.6 (49.2-59.9) | 65.5 (60.0-70.7) |
| P-valuea | <0.001 | <0.001 | 0.17 | 0.39 |
| Lowering BMI Cutoff | ||||
| Age 35-70 years + BMI ≥23kg/m2 | ||||
| Total population | 63.8 (60.2-67.2) | 60.8 (56.6-64.9) | 58.8 (55.8-61.8) | 65.7 (61.0-70.0) |
| Asian | 57.3 (46.6-67.3) | 69.6 (61.3-76.8) | 63.5 (55.4-71.0) | 63.7 (55.4-71.3) |
| Black | 66.7 (61.8-71.2) | 65.4 (58.9-71.4) | 66.5 (60.9-71.7) | 65.6 (60.3-70.5) |
| Hispanic | 66.8 (59.3-73.5) | 60.7 (55.7-65.5) | 57.0 (51.6-62.3) | 70.1 (63.7-75.7) |
| White | 64.3 (59.3-69.0) | 58.5 (51.7-65.1) | 57.3 (52.5-62.0) | 65.4 (58.8-71.4) |
| P-valuea | 0.16 | 0.10 | 0.12 | 0.47 |
| Age 35-70 years + Any BMI | ||||
| Total population | 67.8 (63.8-71.6) | 52.1 (47.2-57.0) | 55.4 (52.8-58.1) | 64.8 (59.2-70.1) |
| Asian | 70.1 (62.0-77.2) | 53.8 (42.6-64.5) | 58.4 (49.0-67.2) | 66.0 (55.3-75.3) |
| Black | 70.4 (65.8-74.6) | 59.8 (51.4-67.7) | 64.4 (58.3-70.0) | 66.2 (60.2-71.8) |
| Hispanic | 68.4 (61.2-74.9) | 56.2 (50.1-62.1) | 54.9 (49.4-60.4) | 69.5 (62.9-75.3) |
| White | 67.6 (61.9-72.8) | 48.9 (41.6-56.2) | 53.5 (49.6-57.3) | 63.5 (55.1-71.2) |
| P-valuea | 0.67 | 0.09 | 0.09 | 0.60 |
BMI = Body mass index
P-values for racial and ethnic differences in performance characteristics of the USPSTF screening criteria were determined using Chi-square tests. Boldface indicates statistical significance.
DISCUSSION
This is the first study examining health equity implications of the recent USPSTF recommendation for prediabetes and diabetes screening by quantifying its clinical performance characteristics. Lowering the screening age from 40 years to 35 years in 2021 results in 13.9 million U.S. adults newly eligible. The greatest proportionate gains in eligibility were observed among Hispanic adults. The 2021 criteria were associated with marginally higher sensitivity and lower specificity than the 2015 criteria across the entire U.S. adult population, and within all racial and ethnic groups examined. Screening based on lower age and BMI thresholds resulted in even greater sensitivity and lower specificity among all U.S. adults, and especially in Hispanic and Asian individuals. These findings demonstrate that the 2021 USPSTF recommendation will identify a greater proportion of adults with prediabetes and diabetes than the 2015 criteria in all race and ethnicity groups studied. Among the alternate age and BMI thresholds examined, screening all adults aged 35-70 years regardless of BMI achieved the most similar performance by race and ethnicity.
Choosing the optimal screening approach to promote health equity requires evaluating tradeoffs in sensitivity and specificity that occur when lowering BMI or age cutoffs. Adherence to screening criteria that maximize sensitivity would result in identifying the greatest proportion of adults with prediabetes and diabetes. Maximizing sensitivity could be considered most appropriate because glycemic testing is inexpensive, the harms of false positive results are low,17 and evidence-based treatment options for prediabetes and diabetes are available.24,25 Adopting criteria that maximize specificity would cause the highest proportion of adults without prediabetes or diagnosed diabetes to be ineligible. This approach might be favored if the diagnosis or treatment were not time-sensitive, screening tests were expensive, or the harms of a false-positive screening result were significant. Overall, the 2021 USPSTF recommendation to screen adults aged 35-70 years with a BMI ≥25kg/m2 is associated with higher specificity than sensitivity.
Epidemiologic studies report that racial and ethnic minority groups experience greater risk of diabetes at younger ages than White individuals. One recent analysis of U.S. adults with diabetes found that 9.3% of White adults were diagnosed before age 35, compared with the following proportions of adults from other racial and ethnic groups: Asian (13.2%); Black (15.5%); and Hispanic (21.0%).12 The prevalence of type 2 diabetes in children and adolescents, although less common than among adults, demonstrates even larger racial and ethnic disparities.26 The current study found higher sensitivity and lower specificity when screening adults aged younger than 35 years, both in the total adult population and within each racial and ethnic subgroup. Gains in sensitivity were larger among racial and ethnic minority groups than White adults. Screening criteria that use lower age cutoffs exhibited the largest increase in sensitivity among Hispanic adults, which is likely related to this group’s high diabetes risk and younger age distribution.27,28
USPSTF’s suggestion to consider screening Asian individuals at a BMI ≥23kg/m2 is based on research demonstrating greater visceral fat accumulation and elevated diabetes risk at lower BMI values in Asian adults than White adults.13,16,29 Therefore, using a universal BMI cutoff of ≥25kg/m2 to define screening eligibility will miss Asian adults with prediabetes or undiagnosed diabetes who would not be screened according to their lower BMI. Indeed, the current analysis demonstrates substantially lower sensitivity of screening criteria that use BMI ≥25kg/m2 to define overweight/obesity in this group, compared with using the Asian-specific BMI ≥23kg/m2 cutoff. Recent analyses report that Hispanic and Black adults also experience increased diabetes risk at lower BMI values than White individuals.14,15 The current study’s findings suggest that screening below a BMI of 25kg/m2 may also increase detection of prediabetes and undiagnosed diabetes in these racial and ethnic groups. Lowering BMI thresholds was associated with higher sensitivity but decreased specificity across all racial and ethnic groups, highlighting the tradeoffs associated with this screening approach.
With the renewed imperative to promote health equity, some experts have suggested revising or discontinuing clinical algorithms that incorporate race as a basis for clinical decision-making.30–32 For example, widely followed equations that use serum creatinine levels to estimate glomerular filtration rate (GFR) include a correction for Black race that results in a higher estimated GFR in this group. Ending this race-based GFR correction, as some health systems have recently done, could result in earlier referrals for specialty care and transplantation, thereby offering potential to promote health equity in a clinical area where stark disparities are consistently documented among Black patients.33 Eliminating race-based clinical algorithms is further supported by scientific consensus that race represents a social construct and not a biologic one.34
However, this approach will not maximize equity in detecting prediabetes and undiagnosed diabetes if a BMI cutoff of ≥25kg/m2 is used across the entire adult population. A recent study reporting greater diabetes prevalence at younger ages and lower BMI values among racial and ethnic minority groups concluded that using race- and ethnicity-specific thresholds may better promote health equity.15 The current analysis demonstrates that screening all 35-70 year-old adults regardless of BMI exhibits greater sensitivity than using a BMI threshold, thereby maximizing detection of prediabetes and diabetes. Universal screening for all adults aged 35-70 years also yields the most equitable performance across racial and ethnic groups, while applying the same screening criteria across the entire adult population. Therefore, this approach for promoting health equity may be preferable to using race- and ethnicity-specific screening criteria. It may also be easier to implement by not requiring that clinicians remember alternate BMI or age thresholds when screening patients from different racial and ethnic groups. While lowering the screening age also increased sensitivity, this alternative was associated with less equitable performance across racial and ethnic groups.
The American Diabetes Association (ADA) currently recommends screening all adults aged ≥35 years, or at any age if they have overweight/obesity and an additional diabetes risk factor.22 The ADA criteria differ from the USPSTF recommendation by routinely including adults younger than 35, which exhibited significant variability in performance across racial and ethnic groups in the current analysis, and those older than 70 years. Substantial debate exists about the value of screening older adults, who have a relatively short time horizon for developing diabetes and its complications, while experiencing a higher risk of all-cause mortality.35 This analysis examined adults younger than 35 because USPSTF suggested considering this approach as a potential mechanism for promoting health equity. Because USPSTF did not include a similar suggestion for adults older than 70 years, that group is not examined here. USPSTF did not provide specific guidance on younger age cutoffs. This study demonstrates increased sensitivity and decreased specificity as age limits are lowered, highlighting the opportunity to identify a larger number of cases at the expense of screening more adults without prediabetes or diabetes.
This study is timely given its investigation of the recent USPSTF update in screening criteria for prediabetes and diabetes. The analysis focuses on health equity impacts of this guideline change, responding to urgent calls to eliminate diabetes-related disparities among racial and ethnic minority groups. Analyzing nationally representative data is another strength of this study. Examining guideline performance separately by A1c and FPG, and finding similar results, also represents a study strength. These findings provide further support for screening with A1c given its low cost and easy interpretability, regardless of fasting status.36
Limitations
Due to the cross-sectional design of NHANES, subsequent glycemic tests that are recommended for confirming a diagnosis of diabetes could not be examined.22 The glycemic definitions used here did not include 2-hour postload glucose, which was last collected for NHANES in 2015-2016. This limitation may have underestimated the prevalence of prediabetes and diabetes. Some age-stratified estimates of undiagnosed diabetes may be statistically unreliable due to small sample sizes in narrow age categories. Finally, clinician-diagnosed diabetes was determined based on participants’ self-report.
CONCLUSIONS
Following the 2021 USPSTF screening criteria will identify a greater proportion of adults with prediabetes and undiagnosed diabetes who are now eligible for screening, compared with 2015 criteria. Gains in sensitivity associated with the new criteria were greater among Black, Hispanic, and Asian individuals than White adults, especially when following USPSTF’s suggestion to consider screening at younger ages and at lower BMI values. The study findings suggest that screening all 35-70-year-old adults, regardless of BMI, results in the most similar performance across racial and ethnic groups and may therefore promote equity in detecting prediabetes and undiagnosed diabetes. Achieving equity in diagnosing these conditions also requires addressing structural barriers, which include not having a usual source of primary care, lacking health insurance, or having copays for screening tests based on insurance coverage.37,38 These barriers to receiving prediabetes and diabetes screening, which are especially prevalent among those newly eligible in 2021, may be best addressed through policy efforts. Expanding screening eligibility will likely increase healthcare costs, which highlights a need to study the costs and cost-effectiveness of any approach that is chosen. It could also be valuable for future research to examine the use of prediabetes and diabetes screening criteria in practice, studying their impact on diagnosis, treatment, and outcomes in diverse populations.
Supplementary Material
Funding/Support:
Drs. O’Brien and Ackermann were supported by a grant from the National Institutes of Health (P30-DK092949).
Disclosure of Financial Conflicts of Interest:
Dr. Bailey has received consultant fees from Lundbeck, Luto, Pfizer, and Sanofi. Dr. Bailey has received research support from Eli Lilly, Gordon and Betty Moore Foundation, Lundbeck, Merk, Pfizer, and Retirement Research Foundation for Aging. Dr. Ackerman has received consultant fees from UnitedHealth Group. Dr. Ali has received advisory panel support from Bayer AG and research support from Merk. All potential financial conflicts of interest reported here were unrelated to the current study. The other authors have no financial disclosures.
Role of the Funder/Sponsor:
The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer:
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Centers for Disease Control and Prevention. National Diabetes Statistics Report, Estimates of Diabetes and Its Burden in the United States. Centers for Disease Control and Prevention; U.S. Department of Health and Human Services. Updated January 18, 2022. Accessed April 8, 2022. https://www.cdc.gov/diabetes/data/statistics-report/index.html [Google Scholar]
- 2.Knowler W, Barrett-Connor E, Fowler S, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346(6):393–403. doi: DOI: 10.1056/NEJMoa012512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.UK Prospective Diabetes Study Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352(9131):837–53. [PubMed] [Google Scholar]
- 4.Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HAW. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359(15):1577–1589. doi: doi: 10.1056/NEJMoa0806470 [DOI] [PubMed] [Google Scholar]
- 5.Cheng YJ, Kanaya AM, Araneta MRG, et al. Prevalence of Diabetes by Race and Ethnicity in the United States, 2011-2016. JAMA. 2019;322(24):2389–2398. doi: 10.1001/jama.2019.19365 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Preston SH, Choi D, Elo IT, Stokes A. Effect of Diabetes on Life Expectancy in the United States by Race and Ethnicity. Biodemography Soc Biol. 2018;64(2):139–151. doi: 10.1080/19485565.2018.1542291 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Egede LE, Gebregziabher M, Hunt KJ, et al. Regional, geographic, and racial/ethnic variation in glycemic control in a national sample of veterans with diabetes. Diabetes Care. 2011;34(4):938–43. doi: 10.2337/dc10-1504 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Walker RJ, Strom Williams J, Egede LE. Influence of Race, Ethnicity and Social Determinants of Health on Diabetes Outcomes. Am J Med Sci. 2016;351(4):366–73. doi: 10.1016/j.amjms.2016.01.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Spanakis EK, Golden SH. Race/ethnic difference in diabetes and diabetic complications. Curr Diab Rep. 2013;13(6):814–23. doi: 10.1007/s11892-013-0421-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Miech RA, Kim J, McConnell C, Hamman RF. A growing disparity in diabetes-related mortality U.S. trends, 1989-2005. Am J Prev Med. 2009;36(2):126–32. doi: 10.1016/j.amepre.2008.09.041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.O’Brien MJ, Lee JY, Carnethon MR, et al. Detecting Dysglycemia Using the 2015 United States Preventive Services Task Force Screening Criteria: A Cohort Analysis of Community Health Center Patients. PLoS Med. 2016;13(7):e1002074. doi: 10.1371/journal.pmed.1002074 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wang MC, Shah NS, Carnethon MR, O’Brien MJ, Khan SS. Age at Diagnosis of Diabetes by Race and Ethnicity in the United States From 2011 to 2018. JAMA Intern Med. 2021;181(11):1537–1539. doi: 10.1001/jamainternmed.2021.4945 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hsu WC, Araneta MR, Kanaya AM, Chiang JL, Fujimoto W. BMI cut points to identify at-risk Asian Americans for type 2 diabetes screening. Diabetes Care. 2015;38(1):150–8. doi: 10.2337/dc14-2391 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gujral UP, Vittinghoff E, Mongraw-Chaffin M, et al. Cardiometabolic Abnormalities Among Normal-Weight Persons From Five Racial/Ethnic Groups in the United States: A Cross-sectional Analysis of Two Cohort Studies. Ann Intern Med. 2017;166(9):628–636. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Aggarwal R, Bibbins-Domingo K, Yeh RW, et al. Diabetes Screening by Race and Ethnicity in the United States: Equivalent Body Mass Index and Age Thresholds. Ann Intern Med. 2022;176(6):765–773. [DOI] [PubMed] [Google Scholar]
- 16.Davidson KW, Barry MJ, Mangione CM, et al. Screening for Prediabetes and Type 2 Diabetes: US Preventive Services Task Force Recommendation Statement. JAMA. 2021;326(8):736–743. doi: 10.1001/jama.2021.12531 [DOI] [PubMed] [Google Scholar]
- 17.Siu AL. Screening for Abnormal Blood Glucose and Type 2 Diabetes Mellitus: U.S. Preventive Services Task Force Recommendation Statement. Ann Intern Med. 2015;163(11):861–8. doi: 10.7326/m15-2345 [DOI] [PubMed] [Google Scholar]
- 18.Fang M, Wang D, Echouffo-Tcheugui JB, Selvin E. Prediabetes and Diabetes Screening Eligibility and Detection in US Adults After Changes to US Preventive Services Task Force and American Diabetes Association Recommendations. JAMA. 2022;327(19):1924–1925. doi: 10.1001/jama.2022.5185 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Centers for Disease Control and Prevention. Unweighted Response Rates for NHANES 2017-2020 by Age and Gender. Centers for Disease Control and Prevention, National Center for Health Statistics. Accessed April 20, 2022. https://wwwn.cdc.gov/nchs/data/nhanes3/ResponseRates/NHANES-2017-2020-Response%20Rates-2017-March2020-508.pdf [Google Scholar]
- 20.Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey: 2017-March 2020 Data Documentation, Codebook, and Frequencies; Plasma Fasting Glucose (P_GLU). Accessed 11/22/22, https://wwwn.cdc.gov/Nchs/Nhanes/2017-2018/P_GLU.htm
- 21.Centers for Disease Control and Prevention. NHANES Questionnaires, Datasets, and Related Documentation. National Center for Health Statistics. Accessed April 20, 2022. https://wwwn.cdc.gov/nchs/nhanes/Default.aspx [Google Scholar]
- 22.American Diabetes Association Professional Practice Committee. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes—2022. Diabetes Care. 2021;45(Supplement 1):S17–S38. doi: 10.2337/dc22-S002 [DOI] [PubMed] [Google Scholar]
- 23.Haire-Joshu D, Hill-Briggs F. The Next Generation of Diabetes Translation: A Path to Health Equity. Anna Rev Public Health. 2019;40:391–410. doi: 10.1146/annurev-publhealth-040218-044158 [DOI] [PubMed] [Google Scholar]
- 24.Jayapaul-Philip B, Dai S, Kirtland K, Haslam A, Nhim K. Availability of the National Diabetes Prevention Program in United States Counties, March 2017. Prev Chronic Dis. 2018;15:E109. doi: 10.5888/pcdl5.180063 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Armato JP, DeFronzo RA, Abdul-Ghani M, Ruby RJ. Successful treatment of prediabetes in clinical practice using physiological assessment (STOP DIABETES). Lancet Diabetes Endocrinol. 2018;6(10):781–789. [DOI] [PubMed] [Google Scholar]
- 26.Lawrence JM, Divers J, Isom S, et al. Trends in Prevalence of Type 1 and Type 2 Diabetes in Children and Adolescents in the US, 2001-2017. JAMA. 2021;326(8):717–727. doi: 10.1001/jama.2021.11165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pew Research Center. Age and Gender Distributions for Nativity Groups: 2013. Pew Research Center. Accessed April 15, 2022. https://www.pewresearch.org/hispanic/ph_2015-06_statistical-portrait-of-the-foreign-born_current-11/ [Google Scholar]
- 28.Schneiderman N, Llabre M, Cowie CC, et al. Prevalence of diabetes among Hispanics/Latinos from diverse backgrounds: the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). Diabetes Care. 2014;37(8):2233–2239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Araneta MR, Kanaya AM, Hsu WC, et al. Optimum BMI cut points to screen Asian Americans for type 2 diabetes. Diabetes Care. 2015;38(5):814–20. doi: 10.2337/dc14-2071 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Nabi J, Adam A, Kostelanetz S, Syed S. Updating Race-Based Risk Assessment Algorithms in Clinical Practice: Time for a Systems Approach. Am J Bioeth. 2021;21(2):82–85. doi: 10.1080/15265161.2020.1861365 [DOI] [PubMed] [Google Scholar]
- 31.Richmond SP, Grubbs V. How Abolition of Race-Based Medicine Is Necessary to American Health Justice. AMA J Ethics. 2022;24(3):E226–232. doi: 10.1001/amajethics.2022.226 [DOI] [PubMed] [Google Scholar]
- 32.Vyas DA, Eisenstein LG, Jones DS. Hidden in Plain Sight - Reconsidering the Use of Race Correction in Clinical Algorithms. N Engl J Med. 2020;383(9):874–882. doi: 10.1056/NEJMms2004740 [DOI] [PubMed] [Google Scholar]
- 33.Eneanya ND, Yang W, Reese PP. Reconsidering the Consequences of Using Race to Estimate Kidney Function. JAMA. 2019;322(2):113–114. doi: 10.1001/jama.2019.5774 [DOI] [PubMed] [Google Scholar]
- 34.Witzig R The medicalization of race: scientific legitimization of a flawed social construct. Ann Intern Med. 1996;125(8):675–9. doi: 10.7326/0003-4819-125-8-199610150-00008 [DOI] [PubMed] [Google Scholar]
- 35.Rooney MR, Rawlings AM, Pankow JS, et al. Risk of Progression to Diabetes Among Older Adults With Prediabetes. JAMA Intern Med. 2021;181(4):511–519. doi: 10.1001/jamainternmed.2020.8774 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Little RR, Rohlfing C, Sacks DB. The National Glycohemoglobin Standardization Program: Over 20 Years of Improving Hemoglobin A(1c) Measurement. Clin Chem. 2019;65(7):839–848. doi: 10.1373/clinchem.2018.296962 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Hill-Briggs F, Adler NE, Berkowitz SA, et al. Social Determinants of Health and Diabetes: A Scientific Review. Diabetes Care. 2020;44(1):258–79. doi: 10.2337/dci20-0053 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.O’Brien MJ, Kirley KA, Ackermann RT. Reducing Health Disparities Through Prevention: Role of the U.S. Preventive Services Task Force. Am J Prev Med. 2020;58(5):724–727. doi: 10.1016/j.amepre.2019.11.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.