Associations between socioeconomic factors and visit adherence among patients with glaucoma in the nationwide All of Us Research Program

Abstract

Purpose:

To identify socioeconomic factors associated with visit adherence among glaucoma patients in a nationwide cohort.

Design:

Cross-sectional study.

Subjects:

All subjects were participants in the NIH All of Us Research Program. This study cohort consists of participants who were diagnosed with glaucoma and who answered the question on the Health Care Access and Utilization Survey regarding whether they have seen an eye care provider in the last 12 months.

Methods:

Descriptive analyses were conducted based on participant age, gender, race/ethnicity, insurance status, level of education, and income bracket. Multivariable logistic regression adjusting for these factors was used to generate odds ratios for the association between socioeconomic factors and visit adherence.

Main Outcome Measure:

Visit adherence, defined as reporting seeing an eye care provider in the last 12 months.

Results:

Among 4517 glaucoma patients, 730 (16.3%) indicated they had not seen or spoken to an eye doctor in the last 12 months. In multivariable models, those with some college education (OR: 1.91; 95% CI: 1.19 – 3.04) and those with a college degree or advanced degree (OR: 2.25; 95% CI: 1.39 – 3.60) and those with the highest annual income of $200000 or more (OR: 1.64; 95% CI: 1.10 – 2.45) were more likely to have seen an eye doctor in the past year compared to those in the lowest education and income categories, respectively.

Conclusion:

Lower income and education levels were significantly associated with lower odds of seeing an eye doctor in the past year among all glaucoma patients in All of Us. This highlights an important health disparity and may inform subsequent interventions to promote improved adherence to clinical guidelines regarding eye care for glaucoma monitoring and management.

Among individuals in the United States living with blindness, prior research has shown that women and individuals of Black and Hispanic descent are disproportionately affected with glaucoma, and these individuals are at particular risk for visit non-adherence.^1-4 Recent research regarding visit adherence has been done largely at single-institutions, and many resulting publications concluded that study results are limited by the size of their cohort and the makeup of their local patient population.^3,4 While there have been data reviews at a larger scale, most of these are from the 1990s to early 2010s.^1,4,5

A recent study published by Stagg et. al used claims data and found that 75% of patients living with glaucoma in the United States underwent < 1 visual field test per year, which places them below established monitoring standards.⁶ While these data are consistent with lower visit adherence, claims data do not include factors such as income, race and ethnicity, education, or other socioeconomic characteristics which could greatly influence these patients’ ability to maintain regular visits with their eye care provider. This may contribute to the under-recognition of underrepresented and underserved communities. Furthermore, these factors are important for understanding how to design strategies for improving visit adherence. The current study seeks to fill a gap in research by utilizing a large-scale database provided by the National Institutes of Health (NIH) All of Us Research Program. This national database focuses on including participation from communities traditionally underrepresented in biomedical research, which allows for a large-scale data analysis with broader representation. In addition, the inclusion of data regarding demographics, socioeconomic characteristics, and barriers to healthcare utilization can provide more granular characterization of challenges with accessing eye care and inform future interventions for improving visit adherence.⁷

In this study, we examine the associations between demographic/socioeconomic characteristics and eye care visit adherence among glaucoma patients enrolled in the NIH All of Us Research Program to identify barriers to visit adherence.

METHODS

We extracted data from the NIH All of Us Research Program, a national database focused on enrolling individuals from communities traditionally underrepresented in biomedical research. At the time of our analysis, over 522000 participants were enrolled in version 6 of the dataset (the most recent version available at the time of analysis) with over 80% of participants being from these communities.⁸ Institutional Review Board approval was obtained, and participants provide written informed consent upon enrollment in the study. Patient information is de-identified through a series of data transformations prior to being available to researchers. The project adhered to the tenets of the Declaration of Helsinki.

We created the cohort for this study by identifying patients with a diagnosis of Glaucoma (n = 17387) using Systematized Nomenclature of Medicine Clinical Terms (SNOMED) concept coding (detailed in Table S1). We then removed participants who had missing information in any of the covariates being analyzed, including age, self-identified race and ethnicity, gender, level of education, level of income, and insurance status. From the resulting 14348 participants, we further narrowed our analysis to those who completely answered the Health Care Access and Utilization Survey (HCAUS), a 57-question survey available to all participants where patients could elect to provide information regarding their access and use of care in the past 12 months, before focusing our analysis on the question which indicates whether they have seen an optometrist, ophthalmologist, or eye doctor (someone who prescribes eyeglasses) in the last 12 months (n = 4517).

We calculated visit adherence by dividing the number of patients who responded “yes” to the eye care provider question by the total number of respondents. Age was categorized as 18 – 40, 40 – 64, 65 – 74, 75 – 84, ≥ 85 years. These categories were chosen prior to analysis and reflect categories we have used in previous studies.⁹ Annual income in U.S. dollars was categorized as $0 – $25000, $25000 – $50000, $50000 – $100000, $100000 – $200000, > $200000, and categories reflect reporting used by the All of Us database. Insurance status was categorized as Medicaid, Other Insured (Employer-provided, Private, Medicare, Military, Veteran Affairs, or other), and Uninsured. Education was categorized as less than a high school degree, high school degree/GED, some college, and college graduate and above. We then performed a cross-sectional analysis looking at how these factors were associated with the answer on visit adherence. We compared the findings using Pearson’s Chi-square tests to generate unadjusted p-values and applied the Holm-Bonferroni adjustment for multiple comparisons. We generated adjusted odds ratios (OR) and 95% confidence intervals (CIs) using univariable logistic regression to characterize responses between groups, with control groups chosen prior to analysis. We used the group corresponding to the lowest value as the control group for categorical values age, income, and education. We used females as the control group for gender as they are better represented in the database. We used non-Hispanic Whites as the control group for racial and ethnic groups as we were interested in disparities affecting minority populations. We used Medicaid as the control group for insurance as we were interested in disparities affecting this population versus other insurances. We then used multivariable logistic regression using the studied socioeconomic factors as covariates. Statistical tests were two-sided, and p-values were considered statistically significant at the α ≤ 0.05 level. All statistical analyses were performed in R notebooks available in the All of Us Researcher Workbench platform.

RESULTS

Of the 4517 participants in our study population, 3787 (83.7%) reported seeing an eye care provider within the last 12 months, and 730 (16.3%) reported not seeing an eye care provider within the last 12 months. The mean (SD) age of the cohort was 67.19 (12.47), ranging from 22 to 89 years. Participants included 2629 (58%) women. 3452 (76%) of participants were Non-Hispanic White, 542 (12%) were Non-Hispanic Black/African American, 145 (3.2%) were Non-Hispanic Asian, and 378 (8.4%) were Hispanic or Latino.

Table 2 outlines the number and proportion of participants in each category answering “yes” versus “no” to having seen an eye care provider in the last 12 months separated by characteristics including age, gender, race/ethnicity, education level, annual income, and insurance status. Across all categories, participants were more likely to have seen an eye care provider in the last 12 months than not, with proportions ranging from 67% to 90% answering yes.

Table 2.

Distribution of self-reported eye care visit adherence by selected demographic and socioeconomic characteristics among All of Us participants diagnosed with glaucoma

	Total (N=4517)	Spoken to an optometrist, ophthalmologist, or eye doctor (someone who prescribes eyeglasses) in the last 12 months (N=3787)	Has not spoken to an optometrist, ophthalmologist, or eye doctor (someone who prescribes eyeglasses) in the last 12 months (n=730)	P- values*
Age
18 – 40	200 (4.4%)	153 (77%)	47 (24%)	< 0.001
40 – 64	1281 (28%)	991 (77%)	290 (23%)
65 – 74	1643 (36%)	1393 (85%)	250 (0.15)
75+**	1393 (31%)	1250 (90%)	143 (10%)
Gender***
Female	2629 (58%)	2196 (84%)	433 (16%)	< 0.001
Male	1834 (41%)	1551 (85%)	283 (15%)
Race/Ethnicity
Non-Hispanic White	3452 (76%)	2957 (86%)	495 (14%)	< 0.001
Non-Hispanic Black/African American	542 (12%)	421 (78%)	121 (22%)
Hispanic or Latino	378 (8.4%)	297 (79%)	81 (21%)
Non-Hispanic Asian	145 (3.2%)	112 (77%)	33 (23%)
Education
Less than a high school degree or equivalent	111 (2.4%)	77 (69%)	34 (31%)	< 0.001
High school degree or GED	413 (9.1%)	319 (77%)	94 (23%)
Some college	1128 (25%)	930 (82%)	198 (18%)
College graduate or advanced degree	2865 (63%)	2461 (86%)	404 (14%)
Annual Income
$25000 or less	696 (15%)	546 (78%)	150 (22%)	< 0.001
$25000 – $50000	807 (18%)	660 (82%)	147 (18%)
$50000 – $100000	1413 (31%)	1214 (86%)	199 (14%)
$100000 – $200000	1113 (25%)	937 (84%)	176 (16%)
$200000 or more	488 (11%)	430 (88%)	58 (12%)
Insurance***
Insured	3974 (88%)	3363 (85%)	611 (84%)	< 0.001
Medicaid	521 (12%)	406 (78%)	115 (22%)

^*P-values represent α values given by Pearson’s Chi Square test

^**Original age categories included 75-84 and 85+, but these categories were collapsed together to avoid secondary calculation of cells <20.

^***Counts less than 20 (and corresponding percentages) cannot be displayed due to NIH All of Us Research Program Data and Statistics Dissemination Policy. In some cases, additional data may be obscured to prevent secondary calculation of these values.

With unadjusted logistic regression (Table 3), older patients age 65 – 74 (OR: 1.71; 95% CI: 1.02 – 2.42), age 75 – 84 (OR: 2.67; 95% CI: 1.83 – 3.87), and 85 and older (OR: 2.79; 95% CI: 1.52 – 5.36) were significantly more likely to have reported seeing their eye care provider in the last 12 months than patients 18 – 40 years of age. Participants who identified as Non-Hispanic Black/African American (OR: 0.58; 95% CI: 0.47 – 0.73), Non-Hispanic Asian (OR: 0.57; 95% CI: 0.39 – 0.86), and Hispanic or Latino (OR: 0.61; 95% CI: 0.47 – 0.80) were significantly less likely to report having seen an eye care provider in the last 12 months compared to non-Hispanic Whites. Participants with a high school degree or GED (OR: 1.50; 95% CI: 0.93 – 2.37) and or a college or advanced degree (OR: 2.69; 95% CI: 1.75 – 4.05) were significantly more likely to report having seen an eye care provider in the last 12 months than participants with less than a high school degree or equivalent. Participants with an annual income of $25000 or below were significantly less likely to report having seen an eye care provider in the last 12 months compared to all other income groups (Table 3). Participants with Medicaid insurance (OR: 0.64; 95% CI: 0.51 – 0.81) were also less likely to report having seen an eye care provider in the last 12 months than participants with other insurance; there was no significant difference for patients who were uninsured.

Table 3.

Univariable and multivariable logistic regression analysis of self-reported eye care visit adherence by selected demographic and socioeconomic characteristics among All of Us participants diagnosed with glaucoma

	Univariable Regression		Multivariable Regression
	Unadjusted OR (95% CI)	P Values	Unadjusted OR (95% CI)	P Values
Age
18 – 40	Ref		Ref
40 – 64	1.05 (0.73 – 1.48)	0.787	1.03 (0.71 – 1.46)	0.879
65 – 74	1.71 (1.02 – 2.42)	0.003	1.63 (1.12 – 2.33)	0.009
75 – 84	2.67 (1.83 – 3.87)	< 0.001	2.51 (1.69 – 3.68)	< 0.001
85 +	2.79 (1.52 – 5.36)	0.001	2.74 (1.48 – 5.32)	0.002
Gender
Female	Ref		Ref
Male	1.08 (0.92 – 1.27)	0.352	0.94 (0.79 – 1.22)	0.461
Race/Ethnicity
Non-Hispanic White	Ref		Ref
Non-Hispanic Black/African American	0.58 (0.47 – 0.73)	< 0.001	0.84 (0.66–1.09)	0.179
Hispanic or Latino	0.61 (0.47 – 0.80)	< 0.001	0.93 (0.70 – 1.25)	0.617
Non-Hispanic Asian	0.57 (0.39 – 0.86)	0.006	0.68 (0.45 – 1.04)	0.068
Education
Less than a high school degree or equivalent	Ref		Ref
High school degree or GED	1.50 (0.93 – 2.37)	< 0.001	1.41 (0.86 – 2.30)	0.166
Some college	2.07 (1.33 – 3.17)	0.088	1.91 (1.19 – 3.04)	< 0.001
College graduate or advanced degree	2.69 (1.75 – 4.05)	< 0.001	2.25 (1.39 – 3.60)	< 0.001
Annual Income
$25000 or less	Ref		Ref
$25000 – $50000	1.23 (0.96 – 1.59)	0.106	1.02 (0.76 – 1.36)	0.919
$50000 – $100000	1.68 (1.32 – 2.12)	< 0.001	1.24 (0.92 – 1.66)	0.156
$100000 – $200000	1.46 (1.15 – 1.86)	0.002	1.09 (0.79 – 1.49)	0.6046
$200000 or more	2.04 (1.47 – 2.85)	< 0.001	1.64 (1.10 – 2.45)	0.015
Insurance
Insured	Ref		Ref
Medicaid	0.64 (0.51 – 0.81)	< 0.001	1.18 (0.89 – 1.58)	0.260
Uninsured*	0.36 (0.07 – 2.63)	0.243	0.55 (0.10 – 4.04)	0.497

^*This was included in the model but not depicted in Table 1 due to counts less than 20 being able to be derived based on other information, so data were obscured in alignment with the All of Us Data and Statistics Dissemination Policy.

With multivariable logistic regression to adjust for demographic and socioeconomic characteristics, older patients age 65 – 74 (OR: 1.63; 95% CI: 1.12 – 2.33), age 75 – 84 (OR: 2.51; 95% CI: 1.69 – 3.68), and 85 and older (OR: 2.74; 95% CI: 1.48 – 5.32) remained significantly more likely to have reported seeing their eye care provider in the last 12 months than patients 18 – 40 years of age. There were no significant differences based on gender or race/ethnicity in the multivariable model. Participants with some college education (OR: 1.91; 95% CI: 1.19 – 3.04) and those with a college degree or advanced degree (OR: 2.25; 95% CI: 1.39 – 3.60) were significantly more likely to report seeing an eye care provider than participants with less than a high school degree or equivalent (Figure 1). In addition, participants with an annual income of $200000 or more (OR: 1.64; 95% CI: 1.10 – 2.45) were more likely to report seeing an eye care provider in the last 12 months than participants with an annual income of $25000 or less (OR 1.00, Figure 2). Insurance type was no longer significant.

Figure 1.

Percentage of participants who reported seeing an eye care provider in the last 12 months by level of education

Figure 2.

Percentage of participants who reported seeing an eye care provider in the last 12 months by annual income

DISCUSSION

In this study, we examined the associations between demographic/socioeconomic characteristics and eye care visit adherence among participants enrolled in the NIH All of Us Research Program with a diagnosis of glaucoma. Our key findings were: (1) 16% of participants overall reported not seeing an eye doctor in the last year despite having been diagnosed with glaucoma, (2) income and education were significantly associated with visit adherence, and (3) visit adherence among racial/ethnic minorities was significantly lower than Non-Hispanic Whites in unadjusted analyses but was no longer significant after controlling for other factors.

It is concerning that 16% of participants reported not having seen an eye care provider in the last year despite carrying a diagnosis of glaucoma. According to the American Academy of Ophthalmology Primary Open-Angle Glaucoma Preferred Practice Pattern Guidelines, follow-up evaluation should be carried out roughly every 12 months in patients who have achieved target intraocular pressure control for at least 6 months with no progression of disease.⁹ This interval decreases to 6 months and as little as 1 – 2 months in accordance with severity of disease. In a recent analysis of claims data, more than 75% of participants received less than the recommended 1 visual field test per year, suggesting an even larger gap in visit adherence. This difference between our study results could in part be explained by patients attending eye care visits without undergoing a visual field. However, our study may have underestimated how many patients are not following up regularly with their eye care provider due to recall bias, where participants may have reported seeing an eye care provider in the past year when it had in fact been longer. Likewise, there may be selection bias among participants, as those who are engaged in the All of Us Research Program and completing optional surveys may be more involved in their eye care and therefore more likely to follow up as advised. Of note, All of Us does try to mitigate this effect through community outreach at non-health related locations, working with community engagement partners, and creating a network of community advocacy groups to promote enrollment and retention of participants who may not be following up regularly with healthcare providers.¹⁰ Despite these efforts, visit adherence has been recognized as an increasing problem among ophthalmology patients as recent studies have shown that the COVID-19 pandemic had a large impact on missed ophthalmology appointments. In a retrospective observational study of a large tertiary care academic institution, Brant et al reported less than 60% of ophthalmology clinic patients completed all of their scheduled appointments compared to 82% prior to the pandemic.¹¹

Our findings also suggest that education and income are independent risk factors for poor visit adherence among glaucoma patients. Patients earning a college degree were 2.25 times more likely to report seeing an eye care provider than their counterparts who have not received a high school degree or equivalent, and patients earning >$200000 annually were 1.64 times more likely than those earning <$25000 annually. One challenge in identifying high-risk patients is that income, education, and other social determinants of health have not been traditionally well-documented in electronic health record systems.¹² Recently there have been more proactive efforts to gather this information, such as the updated version of the United States Core Data for Interoperability (USCDI), which is the national standard for interoperable health data.^13,14 The newest version of the USCDI includes a patient’s health insurance information, their health status such as disability and functional status, and broadened demographic information such as tribal affiliation in addition to race and ethnicity, sex as opposed to sex at birth, and occupation and occupation industry as health data information that should be included in medical records.¹⁵ In addition, EHRs are implementing more robust screening tools which allow providers to document socioeconomic characteristics with the goal of tracking and enhancing population health on a regional and national level.^13,16 While documentation of these characteristics is an important step, further considerations could include adapting patient-oriented information, visit summaries, and patient education resources to utilize lay language that is understandable to patients from all backgrounds, including those without a formal education or understanding of medical terminology who may have low health literacy. With regard to income, we must consider strategies to mitigate economic barriers to healthcare access, including cost of transportation, consequences of taking time off from work, access to childcare, and impact on home and family life. On a case-by-case basis, it may be beneficial to schedule multiple healthcare visits on the same day, provide transportation vouchers, and be flexible with appointment times to account for unreliable transportation. Telemedicine may also offer a potential solution to some of these barriers, and recent reports of telemedicine-based management for glaucoma are promising.^17-21

Finally, individuals from all studied minority racial and ethnic groups were less likely to report seeing an eye care provider in the past year when compared to their Non-Hispanic White counterparts. However, these differences were not significant when adjusting for socioeconomic factors in the multivariable model. This is to be expected as race/ethnicity is invariably associated with income and education in the US, which are factors known in our study to be linked to visit adherence. This emphasizes that intersectionality plays an important role in healthcare access, as each patient ultimately has multiple identities—each of which can be fluid and factor into their socioeconomic status and ability to access healthcare—and all of which must be considered when implementing protocols designed to address disparities affecting one group.²² Likewise, the use of these socioeconomic characteristics to formulate interventions must be careful to not inadvertently worsen disparities. One such example is the increased interest in the use of predictive no-show models embedded in EHRs to assess who is likely to miss their appointment in order to increase clinic productivity.^23-25 These models calculate the risk of a patient not showing up to their appointment, with the option to overbook those time slots to increase clinic efficiency and minimize financial loss. When race has been used as a factor in the model, there has been concern that this ultimately leads to racial minorities being more likely to be overbooked and having less time with the physician.^26,27 The implications of this type of intervention is especially prevalent in current discussions surrounding systemic racism and the role that our healthcare system plays in this. While race may not be an independent risk factor for reduced visit adherence based on our analysis, interventions aimed at addressing disparities must be careful to avoid perpetuating these inequities as we strive towards improving healthcare access for all of our patients. Conversely, interventions aimed at addressing the other socioeconomic characteristics which are equally part of the patient’s identity may be beneficial in improving these disparities while also limiting the negative consequences of interventions which factor race into their protocol in a way that may penalize minorities for recognized healthcare trends.

Limitations of this study include the reliance on patients to recall whether they had seen an eye care provider in the last 12 months as a metric for visit adherence. This may have over- or under-estimated visit adherence, but we would not expect there to be a differential bias across groups. Though datasets such as the NIH All of Us Research Program provide an array of benefits, such as a large number of participants and a wide variety of variables crossing many disciplines, the de-identification of the data means that we cannot solicit additional information or clarification from study participants, which can result in missed information and misclassification of information. This limitation also applies to other national datasets and claims databases. Similarly, we cannot seek out more participants for cohorts that are underrepresented in the data, such as Pacific Islanders or Native Americans/American Indians, which can result in lack of generalizability of the data to these communities. One such example is the small number of uninsured participants in this study (n < 20), which makes it difficult to accurately draw conclusions based on insurance status and requires further recruitment of both patients of an insured and uninsured status. The lack of uninsured participants may have also led to over-estimation of visit adherence among individuals with glaucoma, as we assume those without insurance coverage would be less likely to access eye care. Therefore, the gaps in visit adherence may be even greater. As mentioned in our discussion above, there may also be bias created by excluding patients who did not answer the optional HCAUS, though All of Us attempts to mitigate this through extensive recruitment efforts. This may also apply to participants who did not provide complete demographic information and were therefore removed from the cohort, though these patients are also less likely to have completed the optional HCAUS. The use of diagnosis codes for inclusion may also confer a bias as patients may have been incorrectly assigned a diagnosis of glaucoma. However, this is an issue of incorrect use of diagnosis codes and warrants further inspection through future studies.

In conclusion, lower levels of income and education were associated with significant variations in eye care visit adherence. These findings may be helpful in guiding the development of interventions to improve overall visit adherence and clinic efficiency, while avoiding interventions that can perpetuate racial healthcare inequities.