Examining Diabetes Status by the Social Determinants of Health Among Adults in Hawai‘i

Abstract

The social determinants of health (SDoH) influence health outcomes based on conditions from birth, growth, living, and age factors. Diabetes is a chronic condition, impacted by race, education, and income, which may lead to serious health consequences. In Hawai‘i, approximately 11.2% of adults have been diagnosed with diabetes. The objective of this secondary cross-sectional study is to assess the relationship between the prevalence of diabetes and the social determinants of health among Hawai‘i adults who participated in the Behavioral Risk Factor Surveillance System between 2018–2020. The prevalence of diabetes among adults was 11.0% (CI: 10.4–11.5%). Filipino, Japanese and Native Hawaiian adults had the highest prevalence of diabetes at 14.4% (CI: 12.7–16.2%), 14.2% (CI: 12.7–15.7%), and 13.2% (CI: 12.0–14.4%), respectively. Poverty level and education were significantly associated with diabetes status. Within employment categories, the adjusted odds ratio (AOR) for retired and unable to work adults were large at AOR: 1.51 (CI: 1.26–1.81) and AOR: 2.91 (CI: 2.28–3.72), respectively. SDoH can impact the development and management of diabetes. Understanding the role SDoH plays on diabetes status is crucial for promoting health equity, building community capacity, and improving diabetes management.

Introduction

Social determinants of health (SDoH) influence health outcomes based on conditions from birth, growth, living, and age factors.¹ As the SDoH encompass a variety of factors including both social and environmental factors, multiple models exist for operationalizing this term for scientific inquiry. One model developed by Healthy People 2030 separates the SDoH into 5 different domains: (1) economic stability, (2) education, (3) health care, (4) neighborhood and the built environment, and (5) social and community context.² While the Healthy People 2030 model divides SDoH into categories, it is important to note that social and environmental factors have a complex interconnected relationship.³ The reciprocal interactions of the SDoH impact an individual’s health rather than the independent effect of a singular domain.³

SDoH play a critical role in health and contribute to health inequities including disparities in diabetes prevalence.⁴ Diabetes is a serious chronic health condition, and a leading cause of death in the US.⁵ Furthermore it may lead to a variety of adverse health outcomes including blindness, and cardiovascular disease.⁶ Type II diabetes, a preventable condition, accounts for 90–95% of all diabetes cases in the US.⁷ Recent data estimates that 14.7% of adults in the US have diabetes, of which 11.3% have been diagnosed with diabetes, and 3.4% are undiagnosed.⁷ In Hawai‘i, an estimated 11.2% of the adult population have been diagnosed with diabetes, and approximately 10 000 adults are newly diagnosed with diabetes every year.⁸

The prevalence of diabetes is unequally distributed among the adult population, especially among racial and ethnic minority groups.⁶ In 2019, a study in Hawai‘i found a higher prevalence of diabetes in Native Hawaiians and Pacific Islanders (NHPIs) and Filipino residents compared to White residents.⁹ This finding is consistent with previous research examining a multiethnic cohort that found Native Hawaiians and Asian populations had a higher risk for diabetes compared to their White counterparts.¹⁰ This disparity among the Indigenous and racial minority populations in Hawai‘i is similar across the US with American Indian, Alaskan Native, Black, and Hispanic populations compared to White populations.⁶ Historically, NHPI and Asian populations have been under-represented in research and placed into aggregate groupings masking subgroup differences and limiting generalizability of results.¹¹ Efforts to improve NHPI and Asian representation in clinical research have been made. Within the National Institutes of Health (NIH), the availability of funding for inclusion of NHPI and Asian populations in research has increased, yet underrepresentation remains an issue.¹² Inequities within minority populations go beyond clinical representation in research. Limited understanding of ethnic and cultural backgrounds may inhibit care due to language barriers and cultural differences.¹³ Considering these racial inequities that may be tied to historical injustices, additional factors such as the SDoH are needed to truly understand the intersectionality with race and ethnicity. Systemic racism and discrimination have driven social and economic inequities hindering people of color from achieving optimal health.^14–15

Education has been reported as inversely associated with the prevalence of diabetes among adults.⁶ A recent systematic review has indicated that low socioeconomic status (SES) may be a risk factor for diabetes-related complications.¹⁶ Diabetes has an economic impact which may affect an individual’s economic stability. Individuals diagnosed with diabetes have an estimated 130% higher medical expenditure compared to individuals living without diabetes.⁸ In 2017, the estimated total economic burden of diagnosed diabetes in the US was upwards of $300 billion.¹⁷ Health care access and quality are also related to diabetes prevalence, as health care access is directly related to diabetes diagnosis, surveillance, and treatment.^18–19 In neighborhood and built environment studies, green space and safe neighborhood walkability have been associated with a reduced risk for type II diabetes.²⁰

Research examining the independent effects and the complex interconnected relationships between the SDoH domains can help identify nonmedical factors that influence health outcomes. The objective of this study is to assess the association between diabetes and the SDoH domains among adults in Hawai‘i. By evaluating these associations, health care and public health professionals can gain a better understanding of diabetes and improve intervention efforts in at-risk populations that may be experiencing health inequities.

Methods

Data Source

The Behavioral Risk Factor Surveillance System (BRFSS) is a state-based Centers for Disease Control and Prevention (CDC)-funded survey that collects adult health-related data from randomly selected adults within each state and US territory.²¹ The survey includes questions focused on health and lifestyle data pertaining to demographic characteristics, health conditions, preventive services, and health related behaviors that are collected via landline and mobile telephones using disproportionate stratified sampling.²¹

Study Population

This secondary cross-sectional study consists of Hawai‘i residents who participated in BRFSS between 2018–2020. A total population of 23 338 adults, 7901 adults from 2018, 7683 adults from 2019, and 7754 adults from 2020 were included in the study. These adults represented a total weighted population of 3 348 355 adults with a weighted average population of 1 116 118 adults per year. Adults with missing information on diabetes status (n=42) were excluded from the study.

Measures

Adults with diabetes were defined based on respondents reporting “yes” to the question “have you ever been told by a doctor, nurse, or healthcare professional that you have diabetes?” Adults who responded “no,” and those with gestational diabetes or pre-diabetes were defined as not having diabetes as they are distinct conditions and are separately identified through other BRFSS questions. Responses with “don’t know, not sure, and refused” were classified as missing. Demographic variables included age, sex assigned at birth, race and ethnicity, marital status, home ownership, education, employment status, health care coverage, and county information.

Measures for Healthy People 2030 SDoH domains:

• (1)Social and community context were not directly measured as they involve a variety of concepts such as social cohesion, support, and capital which are not easily operationalized and present within BRFSS. Age, sex assigned at birth, and race, key factors that influence a person’s health outcome and overall quality of life, were used as indirect measurements and subsequently adjusted for.
• (2)County and neighborhood support were used to measure the neighborhood and the built environment domain. The counties in the state of Hawai‘i include Hawai‘i, Maui, Honolulu, and Kaua‘i county. Counties provide information regarding the resources and development of the built environment while neighborhood support was chosen as it provides information about existing physical activity related infrastructure.
• (3)Employment status and poverty level were chosen to measure the economic stability domain based on precedent within existing literature.²²
• (4)Educational attainment was used to measure the education access and quality domain.
• (5)Health care coverage and insurance type were chosen to measure the health care access and quality domain as they provide information related to both the access and quality components of this domain.

Age was categorized into 3 age groups including: younger adults (18–44 years), middle aged adults (45–64 years), and older adults (65 years and older). Race was categorized into White, Native Hawaiian, Filipino, Japanese, Other Asian, Pacific Islander (excludes Native Hawaiians), and Other Race groups. The Native Hawaiian category includes full and part Native Hawaiian individuals. Education included 5 categories: never attended school/only kindergarten to 8th grade, some high school (grade 9–11), high school graduate (Grade 12, or GED), some college/technical school (college 1-3 years), and college graduate (college 4 years, or more). Neighborhood support was determined by the question on if physical activity supporting infrastructure was available within a neighborhood. Poverty levels (0–130%, 131–185%, and 186% or more) were based on the number of children less than 18 years old and adults in a household and annual household income.²³ Health insurance included private, Medicaid, and Medicare coverage. All missing data were reported.

Statistical Analysis

All analyses accounted for the complex survey design using survey weights, clustering, and design strata to represent the state’s population.^24–26 Demographic characteristics (age, sex assigned at birth, race/ethnicity, home ownership, household income, employment, education level, health care coverage, and county) were examined by diabetes status and presented as weighted frequencies and 95% confidence intervals (CIs). Chi-square tests were performed to compare the frequency distribution of categorial measures in Table 1. Missing responses were included as a missing category but excluded from the multivariable model analysis.

Table 1.

Sociodemographic Characteristics and Prevalence (% [95% Confidence Interval, CI]) of Diabetes Among Adults in Hawai‘i from the Hawai‘i Behavioral Risk Factor Surveillance System (BRFSS) 2018–2020

Characteristics	Observed Frequency n	Weighted Prevalence % (95% CI)	Diabetes Prevalence % (95% CI)	P-valuea
Total Population	23 338	N/A	11.0 (10.4–11.5)	N/A
Age Group, years
Young Adult (18–44)	7 494	45.0 (44.1–45.9)	3.5 (2.9–4.0)	<.001
Middle Aged Adult (45–64)	7 779	30.2 (29.5–31.0)	14.3 (13.2–15.4)
Older Adult (65+)	7 783	23.7 (23.0–24.4)	20.7 (19.3–22.1)
Missing	282	1.2 (1.1–1.4)
Sex Assigned at Birth
Female	12 136	49.6 (48.7–50.5)	10.3 (9.5–11.0)	.002
Male	10 979	48.9 (48.0–49.8)	12.0 (11.2–12.8)
Missing	223	1.5 (1.3–1.8)
Race/Ethnicity
White	8 306	25.7 (25.0–26.4)	6.0 (5.3–6.7)	<.001
Native Hawaiian	4 261	17.6 (17–18.3)	13.2 (12.0–14.4)
Filipino	2 686	14.9 (14.3–15.6)	14.4 (12.7–16.2)
Japanese	3 307	17.1 (16.4–17.8)	14.2 (12.7–15.7)
Other	851	4.7 (4.2–5.1)	10.5 (8.8–12.1)
Other Asianb	2 065	11.9 (11.3–12.5)	12.5 (10.1–14.9)
Pacific Islanderc	1 108	5.6 (5.1–6.0)	7.4 (5.0–9.8)
Missing	754	2.9 (2.6–3.2)
Marital Status
Divorced	3 102	9.6 (9.2–10.1)	12.6 (11.1–14.2)	<.001
Married	11 627	51.9 (51.0–52.7)	12.1 (11.3–12.9)
Never Married	5 034	25.4 (24.6–26.2)	6.5 (5.7–7.4)
Separated	397	1.5 (1.3–1.7)	14.8 (8.4–21.2)
Unmarried Couple	992	4.5 (4.2–4.9)	6.6 (4.6–8.5)
Widowed	2 064	6.6 (6.1–7.0)	20.0 (17.2–22.7)
Missing	122	0.5 (0.4–0.6)
Home Ownership
Other arrangement	2 513	10.3 (9.8–10.8)	6.7 (5.5–7.9)	<.001
Own	13 230	62.7 (61.9–63.5)	12.5 (11.8–13.3)
Rent	7 485	26.4 (25.7–27.1)	9.0 (8.2–9.8)
Missing	110	0.7 (0.5–0.8)
Employment Status
Employed for wages	10 331	50.3 (49.4–51.2)	7.7 (7.1–8.4)	<.001
Homemaker	687	3.3 (2.9–3.6)	7.0 (4.9–9.2)
Not Employed	1 371	6.2 (5.7–6.6)	9.0 (7.1–10.9)
Retired	6 445	20.9 (20.3–21.6)	21.3 (19.8–22.8)
Self-Employed	2 646	10.4 (9.9–10.9)	7.6 (6.2–8.9)
Student	565	3.9 (3.5–4.3)	1.7 (0.5–2.9)
Unable to work	1 013	3.6 (3.3–3.9)	24.1 (20.4–27.7)
Missing	280	1.4 (1.2–1.7)
Health Coverage
No	1 415	6.8 (6.3–7.2)	7.0 (5.4–8.6)	<.001
Yes	21 868	92.9 (92.4–93.4)	11.3 (10.7–11.9)
Missing	55	0.3 (0.2–0.5)
Education Level
Never attended school/Less than Grade 8	218	2.2 (1.8–2.6)	18.2 (11.2–25.3)	<.001
Some High School	640	6.1 (5.5–6.7)	15.5 (12–18.9)
High School Graduate	6 050	28.9 (28.1–29.7)	12.0 (11.0–13.0)
Some College/ Technical School	6 571	32.8 (32.0–33.7)	11.1 (10.1–12.0)
College Graduate	9 804	29.7 (29–30.4)	8.4 (7.7–9.1)
Missing	55	0.2 (0.2–0.3)
County
Hawai‘i	4 549	13.5 (13.1–13.9)	10.8 (9.7–12)	.154
Honolulu	10 994	65.0 (64.4–65.6)	11.4 (10.7–12.2)
Kaua‘i	2 675	4.7 (4.5–4.8)	11.2 (9.7–12.7)
Maui	3 986	11.2 (10.9–11.6)	9.9 (8.8–11.1)
Missing	1 134	5.6 (5.2–6.0)

^aP-values are based on a chi-square test for Diabetes prevalence

^bExcludes Filipino and Japanese racial groups

^cExcludes Native Hawaiian racial group

Unadjusted and adjusted multivariable logistic regression models examined the association of diabetes status with each SDoH indicator. Reference groups for each SDoH indicator were based on the normative or largest category. Models were adjusted for race, sex assigned at birth, and age. Results generated from the regression models are presented as odds ratios (ORs) with 95% CIs. All statistical significance was based on P<.05. Statistical analysis was performed in SAS Studio version: Release 3.8 Enterprise Edition (SAS Institute, Inc., Cary, NC).

Results

The prevalence of diabetes was 11.0% (CI: 10.4–11.5%) for the total population (Table 1). Among age groups, older adults (65+) had the highest diabetes prevalence at 20.7% (CI: 19.3–22.1%). Males had higher diabetes prevalence compared to females at 12.0% (CI: 11.2–12.8%) and 10.3% (CI: 9.5–11.0%). Filipino, Japanese, and Native Hawaiian adults had the highest prevalence of diabetes at 14.4% (CI: 12.7–16.2%), 14.2% (CI: 12.7–15.7%), and 13.2% (CI: 12.0–14.4%), respectively. Diabetes prevalence was higher among adults with health care coverage (11.3%, CI: 10.7–11.9%) compared to those without health care coverage (7.0%, CI: 5.4–8.6%). Education status followed a negative step-wise pattern with fewer adults having diabetes with increasing education levels. Adults who never attended school, or only completed kindergarten to 8th grade had a diabetes prevalence of 18.2% (CI: 11.2–25.3%) compared to 8.4% (CI: 7.7–9.1%) among college graduates.

All SDoH indicators except for county and neighborhood support were associated with diabetes in the unadjusted model and adjusted model for age, sex assigned at birth, and race/ethnicity (Table 2). Within the unadjusted model, all household income groups had higher OR’s compared to the $75 000 or more reference group. However, in the adjusted model, the $20 000 to $24 999, $25 000 to $34 999, and $50 000 to $74 999 household income groups were not associated with odds of diabetes.

Table 2.

Unadjusted and Adjusted^a Odds Ratio (OR) and Confidence Intervals (CI) of Diabetes Status by the Social Determinants of Health (SDoH) Domains, Hawai‘i Behavioral Risk Factor Surveillance System (BRFSS), 2018–2020

SDoH Domain	Unadjusted OR [95% CI]	P-value	Adjusteda OR [95% CI]	P-value
SDoH Domain: Neighborhood and the Built Environment
County
Hawai‘i	0.94 [0.82–1.09]	.411	0.91 [0.78–1.06]	.211
Honolulu	Ref		Ref
Kaua‘i	0.98 [0.83–1.16]	.820	0.93 [0.78–1.11]	.439
Maui	0.85 [0.74–0.99]	.037	0.86 [0.73–1.01]	.062
Neighborhood Support
Yes	Ref		Ref
No	1.06 [0.93–1.19]	.395	0.95 [0.83–1.08]	.399
SDoH Domain: Economic Stability
Household Income
Less than $10,000	2.06 [1.54–2.76]	<.001	2.02 [1.43–2.85]	<.001
$10,000 to $14,999	2.41 [1.80–3.23]	<.001	2.01 [1.46–2.76]	<.001
$15,000 to $19,999	1.78 [1.42–2.24]	<.001	1.56 [1.21–2.00]	.001
$20,000 to $24,999	1.37 [1.09–1.72]	.006	1.25 [0.98–1.59]	.075
$25,000 to $34,999	1.47 [1.18–1.82]	.001	1.23 [0.99–1.53]	.057
$35,000 to $49,999	1.68 [1.40–2.02]	<.001	1.48 [1.22–1.80]	<.001
$50,000 to $74,999	1.32 [1.11–1.58]	.002	1.19 [0.98–1.43]	.077
$75,000 or more	Ref		Ref
Employment Status
Employed for wages	Ref		Ref
Homemaker	0.91 [0.64–1.28]	.583	1.04 [0.73–1.50]	.82
Not Employed	1.18 [0.92–1.52]	.193	1.22 [0.93–1.60]	.151
Retired	3.25 [2.85–3.70]	<.001	1.51 [1.26–1.81]	<.001
Self-Employed	0.98 [0.79–1.22]	.840	0.87 [0.70–1.09]	.234
Student	0.20 [0.10–0.43]	<.001	0.47 [0.22–1.00]	.051
Unable to work	3.82 [3.07–4.76]	<.001	2.91 [2.28–3.72]	<.001
Poverty Level
0–100%	1.42 [1.21–1.66]	<.001	1.64 [1.36–1.97]	<.001
101–185%	1.31 [1.11–1.54]	.001	1.30 [1.10–1.55]	.003
186–300%	0.93 [0.80–1.10]	.394	1.05 [0.89–1.25]	.558
301% or more	Ref		Ref
SDoH Domain: Education Access and Agency
Education Level
Never attended school/ Grade 8 or less	2.48 [1.53–4.02]	<.001	1.31 [0.75–2.29]	.337
Some High School	2.01 [1.52–2.66]	<.001	1.83 [1.34–2.51]	<.001
High School Graduate	1.49 [1.31–1.70]	<.001	1.49 [1.29–1.72]	<.001
Some College/Technical school	1.36 [1.19–1.54]	.002	1.24 [1.08–1.42]	.003
College Graduate	Ref		Ref
SDoH Domain: Healthcare Access and Agency
Health Coverage
No	0.59 [0.46–0.76]	<.001	0.78 [0.59–1.02]	.073
Yes	Ref		Ref
Health Insurance Type
Private	Ref		Ref
Medicaid	1.42 [1.13–1.78]	.002	1.58 [1.23–2.02]	<.001
Medicare	2.08 [1.83–2.36]	<.001	1.22 [1.04–1.44]	.017

^aAdjusted for age, sex assigned at birth, and race/ethnicity

Diabetes status was statistically associated with poverty level and employment status, after adjusting for age, sex and race/ethnicity. However, the employment statuses of homemaker, not employed, and self-employed, and poverty level 186–300% were not associated with diabetes in either model. Adults with diabetes were more likely to be in the 0–100% poverty level (adjusted odds ratio [AOR]: 1.64, CI: 1.36–1.97), and the 101–185% poverty level (AOR: 1.30, CI: 1.10–1.55) compared to the 301% or more poverty group. Adults who are retired (AOR: 1.51, CI: 1.26–1.81) or unable to work (AOR: 2.91, CI: 2.28–3.72) were more likely to have diabetes compared to the individuals employed for wages.

Among education level, diabetes was inversely associated with higher levels of educational attainment. Never attended school/grade 8 or less was not associated with diabetes in the adjusted model but was associated with diabetes in the unadjusted model (AOR: 2.48, CI: 1.53–4.02). Diabetes status was inversely associated with no health care coverage for the unadjusted model (OR: 0.59, CI: 0.46–0.76) but this was not statistically significant in the adjusted model (AOR: 0.78, CI: 0.59–1.02). Among those with health insurance, individuals with Medicare and Medicaid were 1.22 (CI: 1.04–1.44), and 1.58 (CI: 1.23–2.02) times as likely to have diabetes compared to those with private health insurance in the adjusted model.

Discussion

Diabetes status was associated with age, sex assigned at birth, race, education, poverty level, and health coverage. Previous studies have reported associations with diabetes and income, education, the built environment, and race.^27–39 In this study, diabetes status was also associated with economic stability, and other researchers have found similar associations with income and poverty level.^28–29 Research in Hawai‘i has demonstrated increased diabetes risk with lower household income.⁹ One study utilizing national health data found a stepwise association gradient with increasing diabetes prevalence among the lower income group.³⁰ Hawai‘i has a high cost of living and subsequently a high median household income of $83 173.⁴⁰For this reason, poverty level may be a better indicator of SES than income because it incorporates the cost of living and number of people in the household. SES is a combined measure known to be inversely associated with diabetes. A 2006 study conducted by Borell et al found an association between education and diabetes prevalence, similar to the results of this study.⁴¹

Having health care is the strongest predictor for diabetes screening among adults,⁴² and uninsured adults are more likely to have undiagnosed diabetes because of limited access to medical care.⁴³ This may explain why not having health care coverage was inversely associated with diabetes status in this study. Health care is important in the diagnosis and monitoring of diabetes by medical professionals. Increased access to adequate health care and resources in remote communities living in rural areas are needed. Limited internet accessibility, public transportation, food, and economic insecurity, far distance from health facility, and social isolation present challenges for at-risk populations from receiving optimal care.^44–46 The COVID-19 pandemic highlighted the persistent challenge populations in remote areas face, especially older adults, in receiving health care and social services.

This study’s findings differed from previous findings that found associations between diabetes and the neighborhood and built environment.^27,31–39 As a multiple-island state, the geographical composition of Hawai‘i is different from previously studied locations. The natural environment and wet and dry season climate of Hawai‘i promote outdoor recreation that supports more physical activity compared to other geographic locations. The islands in Hawai‘i are relatively small compared to other US states, and the short distance between hiking trails and beaches may increase the use of these spaces for physical activity. This may explain why physical activity promotion infrastructure, a measure of neighborhood support, was not statistically associated with diabetes. Overall, the residents of Hawai‘i have the highest life expectancy in the US at 80.7 years which may be attributed to an active lifestyle among many other potential factors.⁴⁷ Previous studies have reported large differences in life expectancy by race groups in Hawai‘i.⁴⁸ Associations between diabetes and the built environment may exist by race groups, which this study did not assess.

In Hawai‘i, the majority of adults living with diabetes were among minority racial groups with low socioeconomic status, less education, and no health care coverage compared to adults living without diabetes. This study did not stratify by race when examining the SDoH by diabetes. However, a 2019 Hawai‘i BRFSS study reported an association with diabetes and SDoH inequities among those who identified as Japanese, Filipino, Native Hawaiian and Pacific Islander, or Chinese compared to White adults.⁹ Another study reported income and education disparities in minority racial groups and diabetes status.²⁹ Native Hawaiian populations from low SES are at an increased risk of social and health inequities, such as limited access to care and substandard living conditions.⁴⁹

The relationship between the SDoH and diabetes is complex as the SDoH may not only contribute to the development of diabetes but also impact diabetes management. Research on a national level has shown that houselessness, food insecurity, lack of health insurance, and low economic stability all negatively impact diabetes management.^50–53 As many of these factors are similarly associated with diabetes prevalence, more research is needed to tease out the role the SDoH play in the development and management of diabetes. Through new research, targeted interventions may be developed to help improve both diabetes prevention and management.

This study has several limitations. The SDoH among adults living with diabetes were assessed, yet the cross-sectional methodology inhibits causal inferences from being made. It is unclear the directionality of the associations, if the SDoH influenced a person’s diabetes status, or if the diabetes status affected the SDoH. The BRFSS collects self-reported information that could lead to information bias. Individuals without a phone, and incarcerated are not included in BRFSS which is a potential source of selection bias. Houselessness is associated with low income, less education, and limited access to care that may lead to undiagnosed and untreated diabetes prevalence. Although this study did not examine houselessness by race, the authors did assess home ownership that included adults who own, rent, or have other living arrangements. Data regarding undiagnosed diabetes were not available, and thus could not be evaluated in the context of the SDoH. It is likely that the prevalence of undiagnosed diabetes is associated and impacted by the various SDoH indicators assessed in this study. The term SDoH is very broad and complex with numerous levels of measures. While this study was able to examine the SDoH in the context of the five domains from the Healthy People 2030 model, other facets such as food, community safety, and more were not assessed in this study. Commonly studied disparities with diabetes such as food insecurity, health literacy, and health care provider availability were not evaluated. Furthermore, confounding factors such as smoking, and obesity were not adjusted for in the multivariable model as the focus of this study was to evaluate the SDoH.

This study adjusted for race, sex assigned at birth, and age. While it is common practice to adjust for race, it is important to note that race is a social construct and a primary driver for SDoH and inequities in income, living conditions, and education especially among communities of color. Race is not an appropriate predictor for various health outcomes and conditions.⁵⁴ Instead, the causal effect of race may be better understood through evaluation of social factors such as the SDoH.⁵⁵ Previous studies have found that SDoH indicators are better predictors for adverse health outcomes compared to race.^56–57 This study did not compare the predictive power of SDoH indicators versus race. Further research on diabetes and SDoH inequities among race groups is needed.

This study has many strengths. The BRFSS is conducted every year throughout the state of Hawai‘i and uses validated methodology to collect information. The survey also accounts for population weights to be a representative sample of the population. Unlike many studies, Hawai‘i has a diverse minority race/ethnic population and a high representation of NHPI and Asian groups. The NHPI group can be disaggregated into two groups to further assess and better understand each group specific health needs. Key aspects of the SDoH domains were evaluated from the Healthy People 2030 model.

Conclusion

These results add to the growing literature on SDoH, health disparities, and diabetes status in Hawai‘i. A variety of SDoH indicators were statistically associated with lifetime prevalence of health care provider-diagnosed diabetes in Hawai‘i. The study’s findings are consistent with other studies, except for neighborhood environment, which evaluated SDoH and diabetes in other populations, supporting that SDoH are associated with general health outcomes and diabetes. Additional research is needed to evaluate the impact of SDoH on diabetes prevalence and incidence within Hawai‘i. Understanding the role SDoH plays on diabetes status is crucial for promoting health equity, building community capacity, and improving diabetes management.

Glossary

Abbreviations

AOR

adjusted odds ratio

BRFSS

Behavioral Risk Factor Surveillance System

CDC

Centers for Disease Control and Prevention

NHPI

Native Hawaiian and Pacific Islander

OR

odds ratio

SDoH

social determinants of health

SES

socioeconomic status

Conflict of Interest

None of the authors identify a conflict of interest.