Diverging overall trends in diabetes prevalence and incidence among US adults have been documented over the past 12 years.¹ While prevalence increased from 1990–2009 and has since plateaued, incidence also reached a peak between 2008 and 2009 but has gradually declined into 2019.^{2, 3} Significant age, sex, racial and ethnic, and socioeconomic inequalities in diabetes prevalence and incidence have been documented in the US with a higher concentration of diabetes among men, adults aged 45 and older, people from racial and ethnic minority groups, and lower socioeconomic groups; however changes in these inequalities after the 2008 peak are not well understood.^3–5 There have been significant historical events since 2008 that may have affected trends in inequalities in diabetes prevalence and incidence. For instance, the enactment of the 2010 Affordable Care Act (ACA), may have improved access to diabetes preventive care. In addition, the COVID-19 pandemic may have led to higher hospitalization and mortality among racial and ethnic minority groups and communities with low incomes, with and without diabetes.^6–9 Thus, this study examined national trends in disparities for US adults during 2008–2021 by age, race and ethnicity, sex, and socioeconomic status (education and family income-poverty ratio) to further understand changing diabetes patterns among traditionally marginalized populations and to enhance monitoring of diabetes disparities.

Over time from 2008–2021, the US population became older (adults ≥65 years increased), more racially and ethnically diverse, and had greater education and PIR (p <0.001) (see Appendix Table 1). There were no statistically significant changes in the sex distribution over time (p=0.96).

Between 2008 and 2021, there were no statistically significant overall changes in diabetes prevalence while incidence declined over time (AAPC −3.33%; p<0.001) (see Figure 1 and Appendix Table 2). Diabetes prevalence and incidence trends across age, sex, education, PIR, and race and ethnicity are shown in Figure 2 and Appendix Table 2. Diabetes prevalence did not significantly change across age groups, while incidence significantly declined over time for those 45–64 (AAPC −3.28%; p<0.001) and ≥65 (AAPC −5.12%; p<0.001) years of age. Also, no significant changes by sex were found for diabetes prevalence while the incidence declined for both female (AAPC −4.02%; p<0.001) and male individuals (AAPC −2.67%; p<0.001). Across racial and ethnic groups, diabetes prevalence did not significantly change except for NH Asian adults who had significant increases in diabetes prevalence (AAPC 0.96%; p=0.042). Diabetes incidence significantly decreased over time for Hispanic (AAPC −4.78%; p<0.001), NH Black (AAPC −3.66%; p=0.025) and NH White (AAPC −3.54%; p=0.001) adults but did not significantly change for NH Asian adults. Diabetes prevalence significantly increased over time for those with < high school education (AAPC 0.60%; p=0.006) and those with some college/associates degree (AAPC; 0.76%; p=0.031) while diabetes incidence significantly decreased across all education groups. Diabetes prevalence significantly increased over time for those with a PIR of <1.0 (AAPC 1.13%; p=0.026) while the prevalence did not significantly change over time for other PIR groups. The incidence significantly decreased over time nearly across all PIR groups, except those with a PIR of 1.0–2.9 for whom the trend was not statistically significant.

Table 1 shows the BGV and SII across all years of the study period and the AAPC for these inequality measures for diabetes prevalence and incidence. Trends in BGV and SII plotted over time are shown in Figure 3. Based on the overall BGV, there were significant overall racial and ethnic inequalities in diabetes prevalence and incidence during 2008–2021 while sex inequalities were only statistically significant for prevalence. The positive SII value for age group indicates greater diabetes prevalence and incidence among the older age groups, while the negative SII values for both the education and PIR diabetes prevalence and incidence suggest a higher concentration of diabetes prevalence and incidence among the lower education and PIR groups. Significant worsening inequality trends for diabetes prevalence was found for PIR (AAPC 1.90%; p=0.003), but there were no statistically significant changes for age, race and ethnicity, sex, and education. In contrast, improving inequality trends for diabetes incidence was found for age (AAPC −6.47%; p<0.001), education (AAPC 8.27%; p=0.001), sex (AAPC −12.34%; p=0.002), and race and ethnicity (AAPC −10.59%; p=0.010). No statistically significant changes in incidence were observed for PIR.

Although age, sex, racial and ethnic, and SES inequalities in diagnosed diabetes prevalence and incidence persisted during 2008–2021, differing trends were observed in these inequalities over time. During 2008–2021, age, sex, racial and ethnic, and education inequalities in diagnosed diabetes prevalence did not significantly change, but incidence inequalities improved over time. Income-related inequalities in diagnosed diabetes prevalence worsened over time while no significant changes were found for incidence inequalities.

The improving age inequalities in diagnosed diabetes incidence may be primarily driven by the decline in incidence over time among those aged 45–64 and ≥65 years. Saturation may have led to the decline in the rate of newly diagnosed diabetes among these age groups. Increased screening efforts combined with implementing a lowered threshold in 1997 for fasting glucose resulted in most of the at-risk population being captured (i.e., depleting susceptible population by reducing the number of undiagnosed diabetes cases); subsequently incidence of diagnosed diabetes decreased over time. ²¹ The changes in recommendation in 2022 by the American Diabetes Association to screen all patients ≥35 years of age and the 2021 recommendation by the USPSTF to screen all asymptomatic adults aged 35–70 years who have overweight or obesity may increase the incidence of diagnosed diabetes among younger age groups in the upcoming years not captured in this study. It may be important for future studies to monitor how these recent changes in recommendations could affect the trends in age inequalities.^{22, 23}

A prior study examining national trends in prevalence and incidence using NHIS generally found no significant trends in prevalence by age, sex, race and ethnicity, and education except an increasing prevalence trend among those with > high school education; the study found declines in incidence only among those with a high school education and NH White adults from 2008–2017.² This study also found no significant trends across age and sex groups, but observed significant increasing trends in prevalence for those with < high school education and some college or an associate’s degree, and for NH Asian adults from 2008–2021. In contrast to the previous study, this study found significant declining trends in incidence during 2008–2021 among those 45–64 and 65–70 years of age, and generally across all sex, education, and racial and ethnic groups; however, there were no significant trends in incidence for NH Asian adults. The contrasting findings in trends between studies may be due to the prior study including data through 2017 while this study included newer data from 2008–2021. In addition, another study comparing disparities in diagnosed diabetes prevalence between two time periods, 1999–2002 to 2011–2014, found that the overall education and PIR prevalence disparities increased over time.²⁴ The differing results may be due to the two studies use of different methodology and time periods.

An important factor which may explain the declining racial and ethnic, and education inequalities in diabetes incidence could be the faster rate of decline in diabetes incidence among historically disadvantaged groups compared to the more advantaged groups (e.g., NH White adults and those with a bachelor’s degree or higher). The decline in diabetes incidence among these disadvantaged groups may be partly due to the passage of the ACA that increased access and use of healthcare by racial and ethnic minority groups and lower SES groups, which may be important for diabetes prevention.^{6, 8, 9} Conversely, there has also been an increase in use of high deductible health plans (HDHPs) among privately insured populations which may have led to decreased healthcare visits and thus decreased screening and lower case detection among this population.²⁵ Use of HDHPs have increased across all privately insured racial and ethnic and income groups. However, health savings accounts, which help pay for care in HDHPs, are less used by disadvantaged populations which makes these populations more vulnerable to suboptimal care.²⁶ Furthermore, these declines in inequalities may be explained by saturation in which racial and ethnic minority and lower education groups have a larger susceptible population of undiagnosed diabetes in earlier years, allowing for greater decline in incidence over time.²¹ Expanding access to lifestyle change interventions such as CDC’s National Diabetes Prevention Program that also focuses on racial and ethnic minority and lower SES groups, may be particularly important in helping to reduce type 2 diabetes risk factors, given that previous studies found lower participation of these groups.^27–29

These findings of an increasing prevalence trend only among those with a family income below the federal poverty level may be explained by their decline in diabetes mortality.³⁰ As described in prior studies, diabetes prevalence will continue to rise if the decline in mortality rate is greater than the decrease in incident cases.³¹ Although there may have been lag time between Medicaid expansion in 2014 and the observed prevalence, the expansion may partially explain the reductions in diabetes related mortality for lower income groups as it increased health insurance coverage allowing greater access to diabetes care.^{32, 33} Furthermore, interventions to improve obesity and physical activity may be important as a previous study found that these modifiable factors were major contributors to income-related diabetes prevalence inequalities.³⁴ Future studies could examine whether policy interventions (e.g., universal basic income and increasing minimum wage) to reduce income inequalities affect diabetes incidence and prevalence given that some research suggests these policies benefit health.^35,36

Between 2008–2021, age, sex, racial and ethnic, and education inequalities in diagnosed diabetes incidence declined while income-related inequalities in diabetes prevalence increased over time. Overall, diagnosed diabetes inequalities persisted with disadvantaged groups having higher prevalence and incidence. Monitoring changes in inequality trends could provide a benchmark for assessing future progress in addressing these inequalities and help inform programs and policies to improve these inequalities. To close these gaps, future research could focus on identifying factors driving these inequalities including morbidity and mortality as contributing factors.