Black–White Disparities in Social and Behavioral Determinants of Health Index and Their Associations with Self-rated Health and Functional Limitations in Older Adults

Abstract

Background

To characterize cumulative risk scores of social and behavioral determinants of health (SDoH) and examine their association with self-rated general health and functional limitations between non-Hispanic black and white older adults in the United States.

Method

We used data from the 2013–2014 National Health and Nutrition Examination Survey (NHANES), with a nationally representative sample of black and white adults aged 65 or older (n = 954 unweighted). We quantified the cumulative risk scores of SDoH (eg, food insecurity, education and poverty), ranging from 0 (no risk at all) to 8 (highest risk), and used multivariable-adjusted logistic and Poisson regression analyses to assess the association of SDoH by racial group with self-rated health and functional limitations, adjusting for other covariates.

Results

Black older adults had a higher mean cumulative risk score than white counterparts (2.3 ± 2.1 vs 1.5 ± 1.0; p < .001). Black older adults were more likely to report lower self-rated health than white older adults in each of SDoH domains (p < .01 for each). In multivariable-adjusted analyses, black older adults were more likely to report lower self-rated health than white older adults (p < .01 for all) regardless of SDoH risk factors. However, those with high SDoH risk factors (ie, ≥3 risk factors) were more likely to report functional limitations than those in the low-risk group (ie, <3 risk factors) in both racial groups (p < .01 for all).

Conclusion

SDoH-related black–white disparities remain persistent in older age. In particular, SDoH index scores for black and white older adults were differentially associated with functional limitations. Addressing SDoH should be an important consideration in reducing gaps in black–white disparities of functioning.

In 2015, the National Academy of Medicine (NAM) convened an expert panel to conduct a consensus study on identifying a set of valid self-rated measures that capture the most critical factors of social and behavioral determinants of health (SDoH) for morbidity and mortality (1–4). In the final report, 12 key SDoH risk factors from multiple domains (eg, sociodemographic, psychological, behavioral, and individual, neighborhood/community domains) were identified (1). At least 3 validation or feasibility studies of these SDoH measures have been reported (2,3,5), concluding that the NAM’s proposed measures are valid and feasible.

Black–white disparities in access to care (6,7), quality of care (8,9), and health outcomes (10–14) are well documented among older adults. Non-Hispanic black older adults, for example, often face unique barriers to the health care system (eg, perceived racial discrimination (7)), receive lower quality of care (8,9), and have higher prevalence and early onset of multimorbidities (10,13) than non-Hispanic white older adults. To date, however, no study has yet examined the prevalence and patterns of SDoH risk factors using NAM’s model to compare non-Hispanic black and white older adults, or examine their association with self-rated general health and functional limitations. These outcome measures are critically important as they are often considered proxy measures or indicators for disability (15), morbidity (16), or mortality (17–19). In addition, some gaps remain from previous studies (10,13,15), which were outdated or based in a local clinic setting, limiting generalizability.

Furthermore, existing studies suggest that health status improves with increased social advantage. For example, a study using census data from the United States found that functional limitations in Americans between the age of 55 and 84 are inversely related to socioeconomic status (eg, income and educational levels) (20). Based on these earlier studies, we propose 4 comparison groups to examine the associations of SDoH, a broader index of cumulative disparity than SES alone, with perceived health status and functional limitations: (i) non-Hispanic white older adults with low SDoH risk factors (Group 1); (ii) non-Hispanic black older adults with low SDoH risk factors (Group 2); (iii) non-Hispanic white older adults with high SDoH risk factors (Group 3); and (iv) non-Hispanic black older adults with high SDoH risk factors (Group 4) (4).

Using the 2013–2014 data from the National Health and Nutrition Examination Survey (NHANES), which capture the most of SDoH risk factors and include other valid measures for self-rated general health and functional limitations using nationally representative samples, we sought to address the following questions: (i) Are prevalence of SDoH index scores and individual patterns of SDoH risk factors different across the 4 different groups? (ii) Are self-rated general health and functional limitations different across the 4 comparison groups? And finally, (iii) How does SDoH vary in relation to self-rated general health and functional limitations across these 4 groups, controlling for other covariates? To our knowledge, this is the first study examining the associations of black–white disparities in SDoH with self-rated general health and functional limitations in a population of older adults.

Method

Data Source and Study Sample

We used 2013–2014 data from National Health and Examination Survey (NHANES), which is administered by the National Center for Health Statistics (NCHS), Centers for Disease Control and Prevention (CDC) (21). The NCHS of CDC use collected information to produce and report vital and health statistics of the U.S. populations regularly (21). The NHANES is a nationally representative cross-sectional survey, which collects systematic information about sociodemographic, dietary, and health-related inquiries of about 5000 noninstitutionalized U.S. civilians nationwide each year. In this study, we limited our sample to adults aged 65 or older (n = 1306 unweighted). Of these, we only included those who self-identified as either non-Hispanic white or non-Hispanic black, with complete covariate data (n = 954 unweighted), which was equivalent to 37.3 million older adults when data were extrapolated to the U.S. population. The overall survey response rate of 2013–2014 NHANES was 68.5% (22). While more recent data are available at the time of study conducted, we chose 2013–2014 survey data because they capture the most comprehensive information about SDoH. As we used publicly available, de-identified data, this study was exempted from review by the Institutional Review Board of the University of Connecticut School of Medicine. Further details of the survey, including study descriptions, questionnaires, sampling methodology, and other technical reports, are available on the NHANES website (23).

Measures

SDoH risk score

The SDoH index score was developed by the Institute of Medicine’s Committee on Recommended Social and Behavioral Domains and Measures for Electronic Health Records (1). One of main goals in this committee was to identify social and behavioral factors capturing SDoH and able to be readily used in electronic health records (1). Based on NAM’s report (1) and its related validation studies (2,24), 8 risk factors corresponding to SDoH domains were identified as follow: (i) low educational attainment (ie, less than a high school diploma or equivalent); (ii) imputed poverty–income ratio (ie, ≤200% federal poverty level); (iii) food insecurity using a Kendall/Cornell scale (ie, low or very low food security) (25–27); (iv) depression (ie, PHQ-9 score ≥ 10) (28,29); (v) recent tobacco use (yes/no); (vi) potential alcohol abuse using an AUDIT-C screening tool (ie, ≥4 for males and ≥3 for females for moderate risk of alcohol abuse) (30); (vii) low physical activity (no = moderate-to-vigorous physical activities, such as fitness, for at least 10 minutes continuously in a typical week; yes = otherwise) (31); and (viii) lack of a partner (a proxy measure for social connection) (24). Although physical activity was assumed to be highly correlated with one of our health outcomes, functional limitations, our preliminary analyses showed the lack of correlation between these 2 measures (Pearson’s correlation, r < .25). We did not include race/ethnicity in the SDoH index scoring because we used it in defining the population of interest in our study (24). We organized each risk factor as binary (1 = “yes” and 0 = “no”), and constructed a count variable, indicating cumulative risk factors ranging from 0 (no risk at all) to 8 (highest risk), indicating that individuals with an additional SDoH risk factor is more socially and behaviorally disadvantaged. We also created a binary variable of SDoH risk scores (ie, <3 vs ≥3 risk factors) based on a previous study (4,24) to compare self-rated general health and functional limitations.

Self-rated general health and functional limitations

For general health, survey participants were asked to self-rate their overall health as: “poor,” “fair,” “good,” “very good,” or “excellent.” This single-item questionnaire is considered a validated measure to predict mortality (17–19). We dichotomized this variable as 1= “very good” or “excellent” and 0 = otherwise based on the previous study (18). For functional limitations, 19 standardized questionnaire items were used (32). We grouped these items into 5 domains based on the CDC’s technical report: (32): (i) activities of daily living (ADLs) (ie, getting in and out of bed; using fork, knife, or drinking from cup; walking between rooms on the same floor; and dressing yourself); (ii) instrumental activities of daily living (IADLs) (ie, house chores; managing money; and preparing meals); (iii) functional limitations related to leisure and social activities (ie, going out to movies or events; leisure activity at home; and attending social events); (iv) lower extremity mobility (ie, walking up 10 steps; and walking a quarter-mile); and (v) general physical activities (ie, grasping or holding small objects; lifting or carrying; reaching up over head; sitting for long periods; standing for long periods; standing up from armless chair; and stooping, crouching or kneeling). For each item, we coded 1 if responses were “some difficulty,” much difficulty,’ or “unable to do” and 0 if the response was “no difficulty” (32). Our coding was consistent with the CDC’s technical report, so that findings from this study would be comparable to those in the CDC’s report (32). We also created an indicator variable for each domain (1 = yes to any; 0 = otherwise) and a count variable indicating the number of functional limitations, ranging from 0 to 19.

Covariates

We included a number of covariates in the survey questionnaires that are potentially relevant to self-rated general health or functional limitations (4,33,34): age, sex, insurance coverage (yes/no), and self-reported life-time chronic medical conditions (ie, asthma, arthritis, cancer, chronic obstructive pulmonary disease, diabetes, heart disease, health failure, and stroke) (32). These chronic conditions were coded “yes” or “no,” and we constructed a multimorbidity summed score, ranging from 0 (none) to 8 (all of them).

Analytical Plan

First, we estimated prevalence of individual and cumulative SDoH risk factors among U.S. older adults by race. We also compared how demographic and clinical characteristics differed among older adults by race. We used a weight-corrected Pearson’s chi-squared statistic when examining the group differences.

We then created 4 comparison groups: (i) non-Hispanic whites with low SDoH risk factors (Group 1); (ii) non-Hispanic blacks with low SDoH risk factors (Group 2); (iii) non-Hispanic whites with high SDoH risk factors (Group 3); and (iv) non-Hispanic blacks with high SDoH risk factors (Group 4). A low-risk group (<3 SDoH risk factors) and a high-risk group (≥3 SDoH risk factors) were defined based on a previous study (4). We compared self-rated general health and each domain of functional limitations across these groups. We used weight-corrected, Bonferroni-adjusted pairwise comparison in these analyses.

Finally, we performed multivariable-adjusted analyses. We used a multivariable-adjusted logistic regression model for general health, and multivariable-adjusted Poisson regression model for cumulative functional limitations. The key independent variable was the aforementioned comparison groups and we controlled for other covariates.

A p-value <.05 was used to indicate statistical significance for all comparisons, and the statistical software, Stata MP/6-Core 15.1 (College Station, TX) (35), was used for all analyses. The svy and related commands in Stata were used to account for the complex survey sampling design of the NHANES (eg, unequal probability of selection, clustering and stratification) (21,23). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (36).

Results

Characteristics of the Study Sample

Of the survey participants, representative of 37.3 million older adults nationwide, the mean age was 73.2 (SD = 4.7) and 55.7% were female (Table 1). On average, older adults had a cumulative SDoH risk score of 1.7 (SD = 1.1). The most common SDoH risk factors reported were physical inactivity (47.0%), lack of a partner (38.9%), and poverty (36.2%). Non-Hispanic blacks had a higher cumulative SDoH risk scores (2.3 ± 2.1) than non-Hispanic whites (1.5 ± 1.0) (p < .001). Non-Hispanic blacks had a higher prevalence rate in each domain of SDoH risk factors than non-Hispanic whites, though only some of them were statistically significant (eg, educational attainment, poverty, food insecurity, and tobacco use).

Table 1.

Demographic and Clinical Characteristics (column %) of U.S. Older Adults by Race, 2013–2014

	Non-Hispanic White	Non-Hispanic Black	Total	p-Value
Sample size
Unweighted sample	707	247	954
Weighted population	33 645 009	3 657 682	37 302 691
Social and behavioral determinants of health
Lower educational attainment	13.0	31.5	14.8	<.001
Poverty	34.1	56.0	36.2	<.001
Food insecurity	4.7	12.0	5.4	.031
Depressive symptoms	6.8	10.6	7.1	.173
Tobacco use	7.6	17.6	8.6	<.001
Potential alcohol abuse	3.5	6.4	3.8	.289
Physical inactivity	46.6	50.8	47.0	.300
Lack of a partner	36.9	57.9	38.9	<.001
Mean cumulative risk ± SD	1.5 ± 1.0	2.3 ± 2.1	1.6 ± 1.2	<.001
Age, mean ± SD	73.2 ± 4.3	72.5 ± 7.4	73.2 ± 4.7	.211
Female	55.5	57.9	55.7	.345
Uninsured	2.1	3.8	2.2	.316
Self-reported medical condition
Arthritis	56.3	57.5	56.4	.675
Cancer	32.8	15.5	31.1	<.001
Diabetes	18.7	35.9	20.4	<.001
Heart disease	21.0	15.6	20.5	.074
Asthma	12.2	18.5	12.8	.006
Stroke	9.6	9.5	9.6	.957
Heart failure	8.7	10.7	8.9	.346
COPD	9.3	1.8	8.5	.003
Multimorbidity, mean ± SD	1.7 ± 1.0	1.6 ± 1.8	1.7 ± 1.1	.474

Note: COPD = chronic obstructive pulmonary disease. Data are from National Health and Nutrition Examination Survey.

Self-rated General Health and Functional Limitations

About 45.4% of Group 1 (ie, non-Hispanic whites in the low-risk SDoH group) reported their health as “very good” or excellent.’ Groups 2, 3, and 4 reported 27.1%, 28.0%, and 15.5%, respectively (data not shown). In Bonferroni-adjusted pairwise comparison, Group 1 rated their general health significantly higher than the other 3 groups. Table 2 presents functional limitations by comparison group. Across all domains of functional limitations, Group 1 had the lowest functional limitations across all groups (p < .05 for each). Groups 3 and 4, on average, reported more functional limitations than Groups 1 and 2, and the Bonferroni-adjusted pairwise comparison suggests that Group 3 (ie, non-Hispanic whites in the high-risk SDoH group) reported the most functional limitations than the other groups (p < .05 for each).

Table 2.

Functional Limitations of U.S. Older Adults by Cumulative Social Determinants of Health Index Score and Race, 2013–2014

	<3 Risk Factors		≥3 Risk Factors
	Non-Hispanic White (1)	Non-Hispanic Black (2)	Non-Hispanic White (3)	Non-Hispanic Black (4)	Pairwise Comparison
Sample size
Unweighted sample	523	140	184	107
Weighted population	26 508 534	2 030 318	7 136 475	1 627 364
ADLs	13.6% (9.7%–18.9%)	22.8% (14.0%–35.0%)	30.3% (22.9%–38.8%)	35.5% (28.3%–43.5%)	1 < 3, 4
IADLs	18.9% (13.2%–26.3%)	27.1% (19.0%–36.9%)	42.6% (33.4%–52.4%)	33.2% (25.6%–41.8%)	1 < 3
Leisure and social activities	15.3% (11.0%–20.9%)	18.8% (11.5%–29.2%)	37.0% (29.1%–45.7%)	29.0% (22.6%–36.4%)	1 < 3, 4; 2 < 3
Lower extremity mobility	19.8% (13.7%–27.7%)	25.0% (18.3%–33.1%)	52.3% (42.4%–61.9%)	29.0% (19.5%–40.9%)	1, 2, 4 < 3
General physical activities	50.8% (42.8%–58.8%)	53.8% (42.4%–64.9%)	74.4% (63.0%–83.3%)	50.5% (40.4%–60.4%)	1 < 3; 4 < 3
Global	57.5% (49.0%–65.6%)	63.7% (54.6%–72.0%)	84.0% (76.7%–89.4%)	67.6% (60.9%–73.7%)	1, 2, 4 < 3

Note: ADLs = activities of daily living; IADLs = instrumental activities of daily living. Data are from National Health and Nutrition Examination Survey.

Multivariable-Adjusted Analyses

Table 3 presents multivariable-adjusted analyses to examine the associations of SDoH risk factors by race with 2 dependent variables: self-rated general health and cumulative functional limitations. Groups 2 and 4 were associated with a lower likelihood of reporting their health as “very good” or “excellent” when compared to Group 1 (adjusted odds ratio [AOR] = 0.42; 95% confidence intervals [CI] 0.27–0.66 for Group 2, and AOR = 0.25; 95% CI 0.13–0.47 for Group 4). Women were more likely than men to report their health as “very good” or “excellent” (AOR = 1.40; 95% CI 1.04–1.89). In addition, having an additional comorbid condition (AOR = 0.68; 95% CI 0.56–0.82) and functional limitations (AOR = 0.85; 95% CI 0.81–0.89) were less likely to report their health as “very good” or “excellent,” respectively.

Table 3.

Multivariable-Adjusted Association of Social Determinants of Health Index Score by Race With Self-rated Health and Functional Limitations in Older Adults, 2013–2014

	General Health		Functional Limitations
	AOR	95% CI, p	IRR	95% CI, p
Social and behavioral determinant of health index by race
Non-Hispanic white with <3 risk factors	Ref.	–	Ref.	–
Non-Hispanic black with <3 risk factors	0.42	0.27–0.66, .001	1.19	0.93–1.53, .167
Non-Hispanic white with ≥3 risk factors	0.71	0.47–1.07, .096	1.77	1.37–2.29, <.001
Non-Hispanic black with ≥3 risk factors	0.25	0.13–0.47, <.001	1.50	1.17–1.93, .003
Age	1.00	0.96–1.05, .846	1.02	1.00–1.05, .053
Female	1.40	1.04–1.89, .029	1.15	0.98–1.36, .085
Multimorbidity	0.68	0.56–0.82, <.001	1.25	1.18–1.33, <.001
General health	–	–	0.53	0.44–0.63, <.001
Functional limitations	0.85	0.81 - 0.89, <.001	–	–

Note: AOR = adjusted odds ratio; CI = confidence interval; IRR = incidence rate ratio . Data are from National Health and Nutrition Examination Survey.

Turning to functional limitations, Groups 3 and 4 were associated with having more functional limitations when compared to Group 1 (incidence rate ratio [IRR] = 1.77; 95% CI 1.37–2.29 for Group 3, and IRR = 1.50; 95% CI 1.17–1.93 for Group 4). Among covariates, one additional chronic condition was also independently associated with having more functional limitations (IRR = 1.25; 95% CI 1.18–1.33), whereas those who self-reported their health as “very good” or “excellent” were less likely to have functional limitations (IRR = 0.53; 95% CI 0.44–0.63).

Discussion

This is the first study to use NAM’s SDoH measures to quantify prevalence and disparities of SDoH risk factors between non-Hispanic black and white older adults, and whether SDoH index scores by racial group were associated with self-rated health and functional limitations among older adults. Our findings highlight that non-Hispanic black older adults are at greater risk of social and behavioral disadvantage (eg, food insecurity and tobacco use) than non-Hispanic white older adults. In addition, overall, non-Hispanic black older adults were more likely to report lower self-rated health than non-Hispanic white older adults, and those in the high-risk SDoH group were more likely to report functional limitations than those in the low-risk SDoH group across both racial groups.

Our findings are consistent with previous studies, suggesting that blacks have reported more SDoH risk factors than their white counterparts. For example, blacks are more likely to live in disinvested neighborhoods and to have lower income and educational levels than whites (37). The findings from our study suggest that disparities between non-Hispanic black and white groups as measured with SDoH indicators are accumulated and remain persistent in older age.

Another important finding from the current study is that overall, non-Hispanic blacks reported lower self-rated health than non-Hispanic whites. Given that the self-rated general health measure is a valid proxy measure to predict mortality (17–19), this finding implies that black older adults may face unique bio-psycho-social issues, compared to non-Hispanic white older adults, that may have not been fully addressed. To reduce disparities in self-rated health between these 2 racial groups, health care providers and other advocates should address unmet needs among black older adults using a bio-psycho-social perspective guided by the SDoH index.

Those in the high-risk SDoH group reported more functional limitations than those in the low-risk SDoH group across both non-Hispanic black and white older adult groups. Consistent with a previous study conducted in England (38), our findings imply that aging alone may not explain functional limitations, and social and behavioral factors may also explain key roles. Thus, further research is needed as to whether SDoH risk factors mediate or moderate functional ability by racial group among older adults.

There are several implications from this study. First, the SDoH index score was originally developed for use in electronic health records to better understand the roles of SDoH in access to care, quality of care, and outcomes of care (1). Our findings provide evidence that the SDoH index measure is useful to understand the associations of social/behavioral risk factors by racial group with health statuses in older adults, especially when adjusted for multimorbidity and health insurance coverage. The findings support suggestions from previous validation and feasibility studies (2,3,5), that the SDoH index score should be applied in electronic health records and survey research more widely and more research should be done to better understand the roles of SDoH measures on older adults’ treatment plans, quality of care, and health outcomes.

Second, health care providers should consider SDoH risk factors in their clinical practices to identify patients’ social or behavioral needs. When such information is gathered, it must be taken into consideration in the process of care coordination and continuation (39). Community health workers and treatment specialists in the area of diet, exercise, and substance use can assist in facilitating improved quality of care and quality of life among older patients who have diverse needs (40). In addition, given that black–white disparities remain persistent in older age, SDoH risk factors should be considered when developing psychosocial interventions for functional improvement specifically for older adults.

Several methodological limitations deserve comment in this study. First, the racial categories used in this study are simplified and socially constructed. While we acknowledge that there are heterogeneous groups within non-Hispanic black individuals (eg, African Americans and black individuals from elsewhere), we could not identify further information on such heterogeneity in our NHANES data. Second, some SDoH measures (eg, intimate partner violence and geo-codable residential address) were not captured in NHANES data (4). Third, due in part to the cross-sectional nature of data collection, no causal relationships can be drawn from our study. Finally, our data are only representative of community-dwelling older adults, and thus, findings from this study may not be generalized to those who are institutionalized (eg, nursing homes and assisted living facilities).

Despite these limitations, strengths of the study include: the use of valid measures of SDoH, self-rated general health, and functional limitations, and the use of nationally representative data. Overall, our findings highlight that black–white disparities remain persistent when assessing SDoH, and the interaction of these racial groups with SDoH plays key roles on self-rated general health and functional limitations in older adults. Addressing SDoH risk factors should be in parallel with reducing black–white disparities among older adults.

Funding

In the past 3 years, T.G.R. was supported in part by the National Institute on Aging (#T32AG019134). The funding agency had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Conflict of Interest

None declared.

Author Contributions

Study concept and design: T.G.R.; Data acquisition and statistical analyses: T.G.R.; Interpretation of data: All authors; Drafting of manuscript: T.G.R. and K.L.; Critical revision of manuscript for important intellectual content: All authors.