Racial Disparities in Barriers to Care for Patients With Diabetic Retinopathy in a Nationwide Cohort

Abstract

Purpose: To ascertain specific barriers of care among patients with diabetic retinopathy (DR) from different racial/ethnic groups.

Methods: In this cross-sectional study, we included adult participants in the National Institutes of Health All of Us Research Program with DR who answered questions in the Healthcare Access & Utilization survey and Social Determinants of Health (SDoH) survey. Logistic regression was used to study the association between barriers to care and race/ethnicity.

Results: Our cohort included 885 DR patients who answered the Healthcare Access & Utilization survey and 385 DR patients who responded to the SDoH survey. After adjusting for confounders, Hispanic individuals were more likely than non-Hispanic White individuals to report delaying getting medical care due to not being able to get child care (odds ratio [OR] = 6.57 [95% confidence interval {CI}, 1.67–27.8]). Furthermore, compared to non-Hispanic White individuals, non-Hispanic Black individuals were significantly more likely to report being treated with less respect (OR = 2.62 [95% CI, 1.15–5.80]), treated with less courtesy (OR = 2.51 [95% CI, 1.01–5.92]), and receive poorer service than other people (OR = 2.85 [95% CI, 1.25–6.34]) when they go to a doctor's office or other healthcare provider.

Conclusions: We found that Hispanic and non-Hispanic Black individuals with DR reported greater delays/barriers to care compared to non-Hispanic White individuals even after controlling for individualized socioeconomic factors.

Translational Relevance: This study highlights the importance of taking steps to promote health equity, such as increasing access to child care resources and reducing implicit bias among eye care providers, to increase access to care and prevent vision loss from DR.

Introduction

Diabetic retinopathy (DR) is a microvascular complication of diabetes and the leading cause of blindness among working-age adults in the United States.¹ The prevalence of DR among adults with diabetes in the United States is estimated to be around 28%.¹ Worldwide, DR accounts for roughly 2.6% of all blindness, and this proportion is increasing.² Moreover, there are disparities in the prevalence of DR among different racial groups; one study found that non-Hispanic Black and Hispanic adults over 40 years of age with diabetes had a 46% and 84% higher DR prevalence, respectively, compared with non-Hispanic White adults.³ Black patients are also six times more likely to develop visual impairment caused by DR compared to White patients.⁴ 

Optimizing glycemic control and blood pressure control is recommended to slow the progression of retinopathy.⁵ Furthermore, treatment for DR is important because interventions such as laser photocoagulation and vitrectomy when necessary can reduce severe vision loss by 90%.⁶ Thus the American Diabetes Association recommends annual eye examinations if DR is present and more frequent examinations if retinopathy is progressing or sight threatening.⁵ The American Academy of Ophthalmology also recommends yearly eye examinations for patients with diabetes and no DR or minimal nonproliferative DR, and more frequent examinations for individuals with proliferative DR or more severe nonproliferative DR.⁷ However, adherence to annual eye examinations for patients with diabetes is poor, with studies reporting anywhere from 23.5%⁸ to 65%⁹ nonadherence. There are also racial/ethnic differences in completing eye examinations for patients with diabetes, with minorities being less likely to receive care.¹⁰ 

Previous studies have investigated factors associated with poor adherence to DR screening for patients with diabetes.^9,11–16 Factors identified have included younger age, lower income/education, Black race,¹¹ cholesterol levels, duration of diabetes,¹² neighborhood-level deprivation,¹³ not having a regular provider, and having poor housing conditions.¹⁶ However, there has been a paucity of studies focused on barriers to care among patients with established/diagnosed DR because previous studies (such as the ones detailed above) have focused on initial screening for DR among patients with diabetes. Moreover, although previous studies have identified that minority races have poorer adherence to DR screening, to our knowledge, there is a lack of studies on why this may be the case from patients’ perspectives. Studies on DR screening adherence have rarely incorporated patient-reported factors. Thus, in this study, we investigated racial disparities in eye care access for patients with DR using a diverse nationwide dataset incorporating patient perspectives regarding barriers to care. 

Methods

Study Population

Data was obtained from the National Institutes of Health (NIH) All of Us Research Program, a database that aims to enroll a diverse group of at least one million people in the United States.¹⁷ The All of Us program collects a wide range of data from participants, including electronic health record data, physical measurements, and biospecimen collection.¹⁷ Furthermore, after completing the core surveys (The Basics, Lifestyle, and Overall Health), all participants are invited to complete additional optional surveys, including surveys on Health Care Access & Utilization and Social Determinants of Health.¹⁷ Participants provided written informed consent at enrollment in the study, which was approved by the NIH All of Us Institutional Review Board. Identifying information was removed before participant data was available to researchers.¹⁷ The study adhered to the tenets of the Declaration of Helsinki. At the time of our analysis, 372,380 participants were contained in the latest version of the dataset (version 6). The overall response rate for the Healthcare Access & Utilization survey was 43% (160,880 participants who completed at least part of the survey/372,380 total participants) and 15% for the Social Determinants of Health survey (57,620/372, 380). 

A total of 4996 individuals were identified in the All of Us database with a DR diagnosis based on electronic health record diagnosis codes (Supplementary Table S1). Out of those, 1315 individuals (26.3%) answered the nine questions from the Health Care Access & Utilization survey questions examined in this study. Furthermore, we dropped observations with missing values for select covariates used in our analysis (age, race and ethnicity, income, insurance status, and education), yielding 885 individuals (68.0%) in our final data set. We effectively restricted our study population to non-Hispanic Black, Hispanic, and non-Hispanic White individuals since other racial categories had insufficient enrollment to provide statistical power. Patients without health insurance and patients who did not identify as either male or female were excluded because these groups had insufficient enrollment as well. For the Social Determinants of Health (SDoH) survey, 521 (10.4%) individuals answered the seven questions examined in this study, with 385 (74.2%) individuals remaining after excluding for missing values for select covariates (Fig.). 

Figure.

 

Flowchart of exclusion criteria leading to the final study population.

Variables

The Health Care Access & Utilization survey includes information about a participant's access to and use of health care, whereas the SDoH survey includes information about the social determinants of health, including a participant's neighborhood, social life, stress, and feelings about everyday life.¹⁸ 

We studied the following nine questions in the Health Care Access & Utilization survey: There are many reasons people delay getting medical care. In the past 12 months, have you delayed getting care for any of the following reasons? (1) Didn't have transportation, (2) You live in a rural area where distance to the health care provider is too far, (3) You were nervous about seeing a health care provider, (4) Couldn't get time off work, (5) You couldn't get child care, (6) You provide care to an adult and could not leave him/her, (7) Couldn't afford the copay, (8) Your deductible was too high or you could not afford the deductible, (9) You had to pay out of pocket for some or all of the procedure. Question 1 was a validated survey question derived from the National Health Interview Survey.¹⁹ The rest of the questions were developed specifically for use within the All of Us Research Program. 

For the SDoH survey, we studied the following seven questions: (1) How often can you find someone to take you to the doctor if you needed it? How often do any of these happen to you when you go to a doctor's office or other health care provider? (2) You are treated with less courtesy than other people, (3) You are treated with less respect than other people, (4) You receive poorer service than others, (5) A doctor or nurse acts as if he or she thinks you are not smart, (6) A doctor or nurse acts as if he or she is better than you, (7) You feel like a doctor or nurse is not listening to what you were saying. Question 1 was derived from the RAND MOS Social Support Survey Instrument.²⁰ The rest of the 6 questions examined were derived from the Discrimination in Medical Settings Scale.²¹ For both surveys, the answer choices to the questions were always, most of the time, sometimes, rarely, or never. For analysis, we dichotomized the survey answers by grouping always/most of the time/sometimes as yes, and rarely/never as no. 

Information regarding additional covariates (age, race and ethnicity, income, insurance status, and education) were extracted from participants’ survey responses in the All of Us The Basics survey.¹⁸ Age in years was categorized as less than 40 years, 40 to 64 years, 65 to 74 years, and 75 years or older. Annual household income was categorized as less than $25,000, greater than or equal to $25,000 and less than $50,000, greater than or equal to $50,000 and less than $100,000, and more than $100,000. Insurance status was categorized as Medicaid or other insured (employer or union provided, Medicare, military provided, privately purchased, Veterans Administration provided, Indian Health Service, or other). Education was categorized as no high school diploma, high school diploma/GED, some college, and college and greater. 

Statistical Analysis

Pearson χ² tests were used to generate unadjusted P values to compare various patient characteristics and survey responses by race and ethnicity. Logistic regression was used to generate odds ratios (ORs) and 95% confidence intervals (CIs) to compare the proportion responding yes to each survey question between different races and ethnicities, where non-Hispanic White individuals were used as the reference group. We performed both univariable and multivariable logistic regression adjusting for age, gender, insurance status, education, and income. For some survey questions, certain covariates did not fit the model because of zero counts in certain categories. In these cases, we excluded those covariates from those particular models. Of note, any counts less than 20 (and their corresponding percentages) are not allowed to be shared due to NIH All of Us policies aimed at reducing risk of re-identification of study participants. Statistical tests were two-sided, and P values were considered statistically significant at α = .05. The NIH All of Us Researcher Workbench (using R software; The R Foundation) was used to conduct the analyses, which are available online in our project workspace.^22,23 

Results

Of the 885 patients with DR who responded to the Health Care Access & Utilization survey questions examined, 169 (19.1%) were non-Hispanic Black, 117 (13.2%) were Hispanic, and 599 (67.7%) were non-Hispanic White. The median (interquartile range [IQR]) age was 65 (55–73) years, and 461 (52.1%) were female. Non-Hispanic White individuals had the highest median (IQR) age (67 [58–74] years) compared with non-Hispanic Black individuals (62 [52–69] years) and Hispanic individuals (61 [50–69] years) (Table 1). See Table 2 for a similar distribution of selected characteristics for patients who responded to the SDoH survey questions examined. For the Health Care Access & Utilization survey, only the child care and elderly care questions significantly varied by race and ethnicity in the Pearson χ² tests (Table 1). For the SDoH survey, three questions (treated with less respect than other people, treated with less courtesy than other people, receive poorer service than others) significantly varied by race and ethnicity (Table 2). 

Table 1.

 

Distribution of Selected Characteristics by Race and Ethnicity for Patients with Diabetic Retinopathy who Responded to the Health Care Access & Utilization Survey^*

 

Table 2.

 

Distribution of Selected Characteristics by Race and Ethnicity for Patients with Diabetic Retinopathy who Responded to the Social Determinants of Health Survey^*

 

Overall, 65 non-Hispanic Black individuals (38.5%), 41 Hispanic individuals (35.0%), and 191 non-Hispanic White individuals (31.9%) answered yes to at least one Health Care Access & Utilization survey question. In unadjusted univariable logistic regression models, Hispanic individuals were significantly more likely than non-Hispanic White individuals to report delayed medical care due to not being able to get child care (OR = 8.04; 95% CI, 2.26–31.9) and providing care to an adult and not being able to leave them (OR = 3.78; 95% CI, 1.10–12.0). Non-Hispanic Black individuals were more likely to report delayed medical care due to not being able to afford co-pay (OR = 1.86; 95% CI, 1.01–3.33) than non-Hispanic White individuals (Table 3). In multivariable logistic regression models adjusting for age, gender, insurance status, education, and income, Hispanic individuals (OR = 6.57; 95% CI, 1.67–27.8) were still more likely than non-Hispanic White individuals to report delaying getting medical care due to not being able to get child care (Table 3). The trends for elderly care and not being able to afford co-pay were attenuated after controlling for the covariates (Table 3). 

Table 3.

 

Univariable and Multivariable Logistic Regression for the Association between Health Care Access and Utilization-Related Barriers to Care by Race and Ethnicity

 

For the SDoH survey, in both univariable and multivariable logistic regression models, compared to non-Hispanic White individuals, non-Hispanic Black individuals were significantly more likely to report being treated with less respect (multivariable logistic regression OR = 2.62; 95% CI, 1.15–5.80), treated with less courtesy (OR = 2.51; 95% CI, 1.01–5.92), and receive poorer service than other people (OR = 2.85; 95% CI, 1.25–6.34) when they go to a doctor's office or other health care provider (Table 4). 

Table 4.

 

Univariable and Multivariable Logistic Regression for the Association Between Social Determinants of Health-Related Barriers to Care by Race and Ethnicity

 

Discussion

In this nationwide study, we analyzed 885 individuals with DR who responded to the nine Health Care Access & Utilization survey questions examined and 385 who responded to the seven SDoH survey questions examined and found that race and ethnicity confer additional risk of reporting greater delays or barriers to care, even when adjusting for socioeconomic characteristics. Compared to non-Hispanic White individuals, Hispanic individuals were more likely to report delaying getting medical care because of not being able to get child care, whereas non-Hispanic Black individuals were more likely to report being treated with less respect, treated with less courtesy, and receive poorer service than other people when they go to a doctor's office. These trends persisted even after adjusting for age, gender, insurance status, income, and education. Interestingly, certain factors were associated with nonadherence to eye examinations in previous studies, but we did not find racial disparities for these factors in our study. For example, An et al.¹¹ found that copayment was associated with nonadherence to retinal screening guidelines. Although we found that non-Hispanic Black individuals were more likely to report not being able to afford copays on univariable regression, this did not remain statistically significant in the multivariable model adjusting for income and other factors. 

Overall, this study used a different approach compared to previous studies that mainly identified factors associated with nonadherence to eye examinations across all races. Although we found that race was an independent risk factor for barriers to care among patients diagnosed with DR, prior studies did not focus on identifying the underlying factors leading to differences in adherence across races. For example, Eppley et al.¹² found that among patients with diabetes, greater age, higher education, insurance status, income, lower total cholesterol, higher levels of high-density lipoproteins, longer duration of diabetes, and higher self-report of DR were associated with greater adherence with annual eye examinations. However, when they stratified by race, none of the variables were newly associated with adherence.¹² Cai et al.¹⁶ found that not having a regular provider and having poor housing conditions were associated with decreased adherence to diabetic eye examinations but did not examine variations by race or ethnicity. Furthermore, An et al.¹¹ found that nonadherent patients were more likely to be younger, male, smokers, and have a lower income and less education. Although An et al.¹¹ did also find that nonadherent patients were more likely to be Black, they did not identify which variables (e.g., socioeconomic, clinical, etc.) were the underlying cause of this racial disparity. 

Moreover, although previous studies focused more on socioeconomic variables (such as age, gender, education level, and insurance) or clinical factors (such as smoking, cholesterol levels, and duration of diabetes), here we analyzed qualitative factors through survey questions related to healthcare utilization and access, as well as social determinants. This was made possible because of the recent launch in November 2021 of The All of Us Research Program's SDoH Survey,¹⁸ which allows for a better understanding of how SDOH impact health outcomes based on social and living settings. This data is not routinely available in electronic health records and provides unique insight into patient perspectives. In particular, there has been an increased emphasis on patient-reported outcomes and patient-centered care in medicine in general. In 2001, the Institute of Medicine Committee on Quality Health Care in America developed six aims to improve the health care system, one of which was patient-centered care.²⁴ The committee highlighted the importance of providing care that is respectful of individual patient preferences and is guided by patient values.²⁴ In patient-centered care, patients and health care providers collaborate through shared decision-making according to patients’ goals, and care focuses not only on patients’ physical health but also their emotional well-being.²⁵ In other words, patients should be listened to, respected, and involved in their care.²⁶ Moreover, patient-centered care has been increasingly emphasized in ophthalmology specifically as well. Sleath et al.²⁷ found that provider-patient communication is associated with glaucoma medication adherence. Furthermore, for DR treatment in particular, Marahrens et al.²⁸ studied patient preferences for involvement in the decision-making process for treating DR and found that the majority of patients preferred a shared decision-making process with the ophthalmologist rather than an ophthalmologist-dominant or patient-dominant decision-making style, underscoring the importance of patient communication and satisfaction. 

Our findings regarding barriers to care for Hispanic individuals revolved around child care. Previous studies have found similar barriers to care for Hispanic individuals. Blanco et al.²⁹ studied ethnic disparities in mental health care and found that lack of childcare, transportation costs, and inflexible work schedules were important barriers for Hispanic patients. Moreover, Coronado et al.³⁰ studied Mexican Americans and type 2 diabetes and described child care as an important environmental factor that contributes to limited access to health care for Hispanic individuals. 

In contrast, barriers to care for non-Hispanic Black individuals revolved around treatment in healthcare settings. Although our findings regarding Black patients feeling biased treatment have not previously been reported for patients with DR, this has been widely studied in other contexts. Black patients are more likely than White patients to report issues with communication with clinicians.³¹ Other barriers for Black patients that have been identified include perceived discrimination³² and medical mistrust.³³ Focus groups have also dove deeper into patient experiences, including the study by Griffith et al.³⁴ that found that Black men reported they were uncomfortable with the tone physicians used when talking to them. Other focus groups found that Black patients felt that their symptoms were discredited and that clinicians did not convey respect and did not acknowledge their perspectives.³⁵ These studies point to the contribution of implicit bias to disparities in health care. In light of this, our finding of non-Hispanic Black individuals with DR being more likely to report being treated with less respect and courtesy, and receiving poorer service is concerning and highlights an important area of intervention. 

It is critical to reduce the impact of implicit bias – some approaches include increasing ophthalmologists’ awareness of implicit bias, individuating (focus on specific information about individuals), and perspective-taking.³⁶ Efforts to increase diversity in the ophthalmology workforce are also important since patients may feel more comfortable receiving treatment from physicians who come from similar backgrounds.³⁷ Although there is a notable lack of underrepresented minorities (URMs) in the ophthalmology workforce (7.7% of ophthalmology residents and 6% of practicing ophthalmologists are URM compared with 33% of the U.S. population),³⁸ efforts are underway to increase URM representation, including the American Academy of Ophthalmology Minority Ophthalmology Mentorship Program for URM students³⁹ on a national scale, as well as outreach interventions by individual residency and fellowship programs.⁴⁰ 

In conclusion, the barriers to care identified in this study highlight the challenges patients with DR face when it comes to preserving their vision against this progressive sight-threatening condition. For DR, regular ophthalmology visits are essential because treatment can substantially improve outcomes and prevent severe vision loss. The racial and ethnic disparities observed in this study highlight the importance of developing interventions and targeting vulnerable populations to prevent visual impairment from untreated DR. Eye care clinicians can take a proactive role in addressing these disparities in barriers to care and promoting health equity by working to reduce implicit bias and implementing SDoH assessments to identify patients at risk and connect them with resources. On a larger systemic level, other strategies to promote health equity include improving representation in clinical trials (a recent study found that participants identifying as White were overrepresented in diabetic macular edema and retinal vein occlusion clinical trials)⁴¹ and diversifying the ophthalmology workforce. Broader social policy changes, beyond the realm of medicine, may be needed for issues such as affordable child care access. 

Limitations

One limitation was that this study only included adults and not adolescents or children. This is an important area of future research because adherence to DR screening is especially low among minority and socioeconomically disadvantaged youth.⁴² Moreover, we excluded patients because of missing covariates for logistic regression analysis, which resulted in a smaller sample size and reduced the power to detect racial disparities for certain survey questions. We performed a subanalysis comparing patient characteristics for those who were included and those who were excluded (e.g., people who did not fill out survey/did not have all covariates). For the Healthcare Access & Utilization Survey, age and gender were not statistically significantly different between the two groups, highlighting the generalizability of the study results. For the Social Determinants of Health Survey, those who were included were generally older and more likely to identify as male. Thus it is possible that the results for this survey may not generalize to younger populations or those who identify as female. Future efforts should be made to encourage broad participation in surveys. Finally, we relied on patient responses to survey questions, which could have been influenced by factors such as recall bias or concerns about how responses may be perceived. However, we do not expect this to significantly impact our results because the patients were aware that survey results would be deidentified. The extent of recall bias is also presumably similar across racial groups. Another limitation was that the surveys included in All of Us consisted of discrete or structured/closed-ended items. Future investigations of these issues may be enhanced by qualitative approaches such as open-ended survey items, interviews, or focus groups. 

Conclusion

Overall, we found that Hispanic and non-Hispanic Black individuals with DR reported greater delays/barriers to care compared to non-Hispanic White individuals even after controlling for individualized socioeconomic factors. This highlights the importance of taking steps to promote health equity, such as reducing implicit bias among eye care providers or increasing access to child care, to improve access to care and prevent vision loss from DR. 

Acknowledgments

In addition to the funded partners, the All of Us Research Program would not be possible without the contributions made by its participants. 

Supported by the National Institutes of Health Grants DP5OD029610 and P30EY022589 (Bethesda, MD, USA) and an unrestricted departmental grant from Research to Prevent Blindness (New York, NY). The All of Us Research Program is supported (or funded) by grants through the National Institutes of Health, Office of the Director: Regional Medical Centers: 1 OT2 OD026549; 1 OT2 OD026554; 1 OT2 OD026557; 1 OT2 OD026556; 1 OT2 OD026550; 1 OT2 OD 026552; 1 OT2 OD026553; 1 OT2 OD026548; 1 OT2 OD026551; 1 OT2 OD026555; IAA #: AOD 16037; Federally Qualified Health Centers: HHSN 263201600085U; Data and Research Center: 5 U2C OD023196; Biobank: 1 U24 OD023121; The Participant Center: U24 OD023176; Participant Technology Systems Center: 1 U24 OD023163; Communications and Engagement: 3 OT2 OD023205; 3 OT2 OD023206; and Community Partners: 1 OT2 OD025277; 3 OT2 OD025315; 1 OT2 OD025337; 1 OT2 OD025276. A.D. is a recipient of the Research to Prevent Blindness Medical Student Eye Research Fellowship. 

Disclosure: B.B. Huang, None; B. Radha Saseendrakumar, None; A. Delavar, None; S.L. Baxter, voxelcloud (C), iVista Medical Education (R), and equipment support from Optomed (F), Topcon (F)