In this contemporary nationally representative analysis, self-reported DM prevalence in an aggregate NHA category masked important heterogeneity in the burden of DM among Asian Americans. NHA in aggregate reported a lower DM prevalence compared to NHB and NHW; however, after adjusting for sex, education, and BMI, NHA had an approximately 20% higher prevalence of DM compared to NHW. Filipino Americans had the highest prevalence of DM compared to all other NHA subgroups and compared to NHB and NHW. Asian Indians and Japanese Americans also reported a higher prevalence of DM compared to NHW, but Chinese and Korean Americans reported a lower prevalence. DM prevalence in NHA was generally higher in older individuals, men, individuals with a history of CVD, current or former smokers, those who were not physically active, and those categorized as having obesity.
These data provide a more detailed understanding of self-reported DM prevalence at the self-identified NHA subgroup level and identify important differences in known (i.e., diagnosed) DM between NHA subgroups in the USA indicating that health status in the NHA population is likely heterogenous. Prevalence estimates from the National Health and Nutrition Examination Survey (NHANES) from 2011 to 2016 showed a similar relative prevalence of DM, higher in South and Southeast Asian Americans compared to that in East Asian Americans. These NHANES data notably identified a 27% overall adjusted prevalence (self-reported plus undiagnosed DM) and 17% adjusted prevalence of self-reported DM in NHA adults (adjusted for age, sex, and BMI).2 Our findings of self-reported DM prevalence in BRFSS of 8.9 percentage points lower than found in NHANES may represent a sampled population with a higher rate of undiagnosed DM. Identifying and addressing undiagnosed DM is an important public health strategy to prevent DM-related morbidity and mortality, particularly for NHA subgroups that are disproportionately affected in the context of projected increase of diabetes prevalence by 54% to approximately 55 million in the US by 2030.8 Prior analysis in NHANES suggests that over 50% of DM in NHA (in aggregate) is undiagnosed, indicating substantial opportunity for improvement.9
The findings in BRFSS confirm important differences in DM prevalence between NHA subgroups that may be related to differences in sociocultural factors and social determinants of health, dietary patterns, physical activity norms, metabolic physiology, and demographics including age distribution of individual subgroups (e.g., Japanese Americans). For example, prior research has shown Asian Indians may have lower beta cell function and lower ability to compensate for higher glucose levels compared to individuals in other race/ethnic groups, leading to relatively higher DM prevalence.10,11 High rates of DM have also previously been described in the Filipino American population in non-nationally representative samples,12,13 and behavioral factors were hypothesized as a primary contributing factor.14 There are also differences in diabetes prevalence in NHA subgroups in the US and their respective counterpart populations in Asia. In 2019, estimated overall age-adjusted diabetes prevalence was 10.4% in India, 9.2% in China, 7.1% in the Philippines, 5.6% in Japan, and 6.9% in South Korea.15 Whether genetic differences in susceptibility to DM exist in individuals of Asian ancestry and between Asian ancestry subgroups remains an area of active study, although currently identified type 2 DM genetic variants do not appear to explain disproportionate susceptibility to type 2 DM compared to populations of European ancestry.16 Overall, however, a paucity of data exist on contributors to variations in DM among Asian American subgroups, and additional research is needed to appreciate how multi-level factors—including environment, social determinants, behavioral factors such as nutrition and physical activity, and physiologic differences—may account for these differences between subgroups, and differences between NHA subgroups and their counterpart populations in Asia.
DM is a strong risk factor for CVD,17 which is responsible for significant premature mortality in NHA.18,19 Differences in DM may contribute to the disproportionate burden of CVD in specific NHA subgroups, such as in Asian Indians.20 We demonstrate that in NHA in aggregate, adjusted DM prevalence is similar to that of the NHW population in the US. However, disaggregation reveals wide differences in the prevalence of diagnosed DM that are masked in aggregated data, with particularly high prevalence in Asian Indian and Filipino Americans. Undiagnosed DM may show similar variability, as certain Asian American subgroups may experience barriers to health care utilization related to health literacy, English language fluency, and cultural differences in attitudes and acceptance of health care services, which may lead to lower awareness of, and less frequent identification and a lower likelihood of, screening for DM.21,22 These findings have important implications for targeting early prevention of type 2 DM, timely diagnosis, and adequate DM management strategies particularly to populations experiencing both high prevalence of DM and excess CVD in order to reduce disparities. Our findings may therefore inform efforts to reduce the burden of DM in the US at the individual, health system, community, and broader population levels.
The main strength of these data is that prevalence estimates are disaggregated into the five largest NHA subgroups in the US, which is necessary for accurate representation of the NHA population in health research.23 There are, however, several limitations. First, our results are based on self-report information, which may be subject to recall and reporting bias. Second, our estimates reflect the prevalence of diagnosed DM, as we are not able to assess undiagnosed DM in this sample since laboratory measures are not collected in BRFSS. The state-level BRFSS data were not weighted to be representative of NHA subcategories, so our findings provide an estimate of the relative burden of self-reported DM. Additionally, because of the limitations of the data collected in BRFSS and lack of survey administration in Asian-specific languages, we are unable to study in detail the roles of healthcare utilization, health literacy, English language fluency, environmental factors, immigration-related factors including time since immigration or immigrant generation, and social and cultural contributors in the observed prevalence distributions. Furthermore, these data are unable to identify differences in rates of undiagnosed diabetes that may exist between NHA subgroups, since only self-reported data are available. Given relatively small sample sizes of NHA subgroups, we also could not reliably estimate DM prevalence in smaller age groups, and we acknowledge that age-related DM prevalence differences may exist within our identified age groups.
This report provides a valuable description of the burden of DM among Asian American subgroups in the US. Overall, from 2013 to 2019, DM prevalence in NHA in aggregate (8.7%) was approximately similar to that of NHW populations (10.0%). However, disaggregated data showed significant variability in DM prevalence in NHA subgroups, with Chinese and Korean Americans having the lowest DM prevalence and Filipino, Japanese, and Asian Indian Americans having the highest. These data can inform public health practitioners and clinicians to more directly tailor programs and interventions to specific populations.