Abstract
Introduction/Objectives:
Little is known about the prevalence of patient-reported social risk factors and the use of the HOUSES Index, a simple, reliable method of assessing socioeconomic status (SES) based on publicly available housing data, in a predominantly rural, primary care population.
Methods:
We conducted a cross-sectional analysis of adult patients paneled to family medicine clinicians in a US Midwest health system as of December 31, 2022. Patients’ listed address determined HOUSES Index as quartile rank (Q1 lowest SES) and rural/urban status. Social risk data including housing, food, transportation, finances, and violence were collected from health record questionnaires. A mixed effect model was used to assess associations between social risk, HOUSES Index, and rurality.
Results:
Of the 352 355 patients included, rural patients were more likely than urban patients to report all social risk factors and had lower SES as measured by HOUSES quartiles. In the mixed effects analysis, HOUSES quartile was independently predictive of reporting an at-risk social risk factor (Q1 vs Q4 OR = 2.27, 95% CI = 2.19-2.37), but rurality was not (OR = 1.02, 95% CI = 0.97-1.07) after adjusting for HOUSES.
Conclusions:
The increased prevalence of social risk factors among rural residents is largely explained by individual SES measured by HOUSES Index.
Keywords: rural health, primary care, social risk factors, social determinants of health, socioeconomic status, HOUSES Index
Introduction
Social and economic environments play an immense role in health outcomes and disparities and these social determinants of health are recognized as world-wide public health concerns.1-3 While social determinants of health act on multiple levels with both direct and complex effects on health, social risk factors are the adverse individual-level social conditions that negatively impact health, such as low educational attainment, homelessness, and social isolation. 4 The National Academies of Sciences, Engineering, and Medicine and many professional medical associations have endorsed and encouraged the assessment of social risk factors in healthcare settings to support patients’ social needs as part of their overall health and wellbeing.5,6 As healthcare systems are increasingly held accountable for patient outcomes and health equity,7,8 screening for social risk factors such as housing insecurity, transportation risk, food insecurity, and interpersonal violence is being widely implemented, but current methods lack standardization and accuracy.9-12
A novel, validated, and objective method of quantifying individual-level socioeconomic status (SES) based on housing/residential factors (referred to as the HOUSES Index), may help clinicians better understand their patients’ social environments.13,14 The HOUSES Index does not rely on patient questionnaires and can be used when patient-reported social data are lacking.13,15 The HOUSES Index is formulated by linking United States (US) address information to readily available and annually updated housing information from counties’ assessors, including individual housing unit size, estimated value, number of bedrooms, and number of bathrooms.13,14 Those with lower SES scores based on HOUSES Index have been found to have higher rates of diabetes, coronary artery disease, asthma, hypertension, mood disorders, and post-kidney transplant graft failure.16-22
Little is known about the prevalence of patient-reported social risk factors and the use of the HOUSES Index in rural populations. Currently, more than 60 million Americans, or 20% of the US population, live in what are classified as rural areas. 23 Geography, along with race and socioeconomic status, are key determinants of health for rural populations.23,24 Rural areas have older populations, higher poverty levels, lower per capita incomes, higher risk health behaviors, higher prevalence of chronic health conditions, and a lower life expectancy than urban areas.25-28 Racial and ethnic health disparities persist in rural areas, especially among Black populations.23,28 In addition, transportation costs, food insecurity, and rates of uninsured are higher among rural counties. 25
We sought to quantify the prevalence of patient-reported social risk factors and explore the associations between these social risks and the HOUSES Index among rural and urban patient populations in a US Midwest multi-state health system. We hypothesized that rural residents carried a higher burden of social risks and a lower SES, as measured by HOUSES Index, compared to urban residents.
Methods
The primary aim of this cross-sectional analysis was to study the distribution of and assess the strength of the association between patient-reported social risks and HOUSES quartiles across rural and urban populations. The Mayo Clinic Institutional Review Board reviewed the protocol and deemed the study exempt from full review due to negligible risk to patients. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 29
The study population included all adult individuals who received primary health care services from the Department of Family Medicine at Mayo Clinic Rochester and Mayo Clinic Health System. Mayo Clinic Health System is a group of 45 regional, multispecialty practices with clinics, hospitals, and other health care facilities in US Midwest, primarily serving rural communities through a community-based practice network. Patients were selected for inclusion if they were an adult (age 18 years or older) residing in a county with a Mayo Clinic Rochester or Mayo Clinic Health System primary care clinic and actively paneled to a family medicine primary care clinician (physician or advanced practice provider) as of December 31, 2022. Actively paneled patients include any patient assigned to a primary care clinician who had at least 1 clinical encounter within the prior 3 years. Patients were excluded if they did not give prior authorization for use of health record data for research purposes.
The primary outcome of interest was patient-reported social risk data collected via electronic health record (EHR) questionnaires administered with a clinic encounter through the Mayo Clinic patient online portal, an electronic tablet upon arrival, or by clinical staff during a patient visit. Patients were invited to update their social risk questionnaire as often as every 6 months and the most recently completed survey data collected prior to December 31, 2022 were used if answered any time in the previous 3 years. The EHR questionnaires assessed 5 separate social risk domains including financial resource risk, food insecurity, housing risk, transportation risk, and intimate partner violence (see Supplemental Appendix 1 for questionnaire and scoring matrix). Patient answers were categorized as a binary variable for the regression analysis, “at-risk” (moderate and high risk were combined for financial resource domain) or “low risk” for all 5 domains.
The primary predictor or independent variable for each individual patient was a HOUSES Index quartile assigned based on the patient address in the EHR as of December 31, 2022. Patient addresses were geocoded to match geographic reference data and property data available from local government assessors’ offices. 14 Each residential property item (square footage of housing unit, estimated housing unit value, number of bedrooms, and number of bathrooms) was ranked and combined into a composite z-score. 14 An individual’s HOUSES Index is normalized within each county, that is, a higher HOUSES Index indicates higher SES compared to neighbors. 14 Each patient’s HOUSES Index z-score was reported and analyzed as a quartile rank with Q1 representing the underserved population with the lowest SES and Q4 representing the highest SES (referred to as HOUSES quartile). 14
Demographic variables including age, self-identified gender, and self-identified race were obtained from EHR data. Rural and urban status of the home address reported in the EHR as of December 31, 2022 was defined by rural-urban commuting area (RUCA) codes. The RUCA codes classify U.S. census tracts using measures of population density, urbanization, and daily commuting. 30 RUCA codes 1 to 3 were considered urban (representing living in or near a city population of 50 000 or more) and RUCA codes 4 to 10 were considered rural (representing living in or near a city population of 49 999 or less). 31 Additional healthcare-related data collected from the EHR included insurance payer, total family medicine outpatient visits for the year 2022, and Adjusted Clinical Group (ACG) risk score (patient medical complexity measure). 32
Patient characteristics were reported using frequencies and percentages for categorical variables and means and standard deviation (SD) for continuous variables. Demographic, healthcare-related, and social risk data among rural and urban cohorts were compared using either Chi-squared (for categorical data) or Kruskal-Wallis test (for continuous data). Two mixed effects logistic regression models were used to assess the strength of the rural and urban associations. First, a mixed effects model that excluded the HOUSES quartile fixed effect was analyzed and reported. A second mixed effects model adjusted for the fixed effects of age, gender, race, ACG score, insurance payer, rurality, HOUSES quartile, and the interaction between rurality and HOUSES to predict the likelihood of reporting any “at-risk” social risk domain. For both generalized linear mixed effects models, county was used as the random effect to account for unmeasured, county-specific contextual factors (eg, local health infrastructure and social services availability) that could influence the outcome independently of individual SES. A variance components structure was used as the random effects covariance matrix. Non-binary and unreported genders were excluded in the mixed models due to the very small sample size (N = 52 patients). Multicollinearity between the fixed effects was checked and were found to not be in violation of this assumption (the largest correlation between rurality and another variable was = −0.11). Model fit was compared with AIC and BIC. Results are reported as odds ratios (OR) with 95% confidence intervals (95% CI). Patients were excluded from the mixed effect model if all social risk domain data or other fixed effects were missing. All analyses were conducted with SAS software (SAS Institute, Cary, NC).
Results
Of the 376 028 patients who gave research consent to use EHR data, 23 673 (6.3%) were missing the HOUSES variable and 352 355 patients had a valid address to obtain HOUSES quartile. Among 204 305 patients categorized in the urban cohort, 10 092 (4.9%) were missing a HOUSES quartile. Among the 171 723 patients categorized to the rural cohort, 13 581 (7.9%) were missing a HOUSES quartile. Reasons for missing HOUSES included use of post office box (10 567 or 44.6%), “apt” or “unit” in address (2440 or 10.3%), or unknown (10 666 or 45.1%). Patients missing all social risk data (151 253 individuals or 42.9% of the analyzed population) or other covariates (1 individual missing ACG risk score and 24 individuals missing gender) were excluded from the mixed effects model analysis. Patients with missing data were more likely to be younger age, male gender, non-White race, have fewer clinic visits, lower medical complexity, and unknown insurance, assigned lower HOUSES quartile, and report any social risk factor (Supplemental Table 1).
Rural study participants had a mean ± SD age of 52.1 ± 19.9 years, and urban participants were slightly younger with a mean ± SD of 49.6 ± 19.5 years (Table 1). Gender distribution was similar between rural and urban patients, with a predominance of females in both cohorts. A greater proportion of rural patients compared to urban patients were white (94.4% and 89.8%, respectively). Patient complexity was higher among rural patients (ACG = 1.3) than urban patients (ACG = 1.2) but mean annual outpatient visits were lower among rural patients (1.9 visits for rural patients vs 2.3 visits for urban patients). Rural participants were more likely to be insured through governmental insurance than urban participants (47.2% and 36.6%, respectively).
Table 1.
Demographic and Healthcare-Related Data of Study Population by Rural/Urban Status.
| Rural | Urban | Total | P | |
|---|---|---|---|---|
| Demographic and Healthcare-Related Variables | N = 158 142 | N = 194 213 | N = 352 355 | |
| Age | <.001 a | |||
| Mean (SD) | 52.1 (19.85) | 49.6 (19.48) | 50.7 (19.69) | |
| Gender, n (%) | <.001 b | |||
| Female | 86 237 (54.5) | 104 632 (53.9) | 190 869 (54.2) | |
| Male | 71 893 (45.5) | 89 540 (46.1) | 161 433 (45.8) | |
| Non-binary | 5 (0.0) | 7 (0.0) | 12 (0.0) | |
| Unknown | 7 (0.0) | 34 (0.0) | 41 (0.0) | |
| Race, n (%) | <.001 b | |||
| American Indian/Alaskan Native | 469 (0.3) | 598 (0.3) | 1067 (0.3) | |
| Asian | 1944 (1.2) | 7221 (3.7) | 9165 (2.6) | |
| Black | 2833 (1.8) | 6707 (3.5) | 9540 (2.7) | |
| Other | 2106 (1.3) | 3168 (1.6) | 5274 (1.5) | |
| Pacific Islander | 223 (0.1) | 195 (0.1) | 418 (0.1) | |
| Unknown | 1255 (0.8) | 2008 (1.0) | 3263 (0.9) | |
| White | 149 312 (94.4) | 174 316 (89.8) | 323 628 (91.8) | |
| ACG Risk score | <.001 b | |||
| Mean (SD) | 1.3 (1.95) | 1.2 (1.88) | 1.3 (1.91) | |
| Median | 0.7 | 0.6 | 0.7 | |
| Range | 0.0, 11.9 | 0.0, 11.9 | 0.0, 11.9 | |
| Outpatient Visits 2022 | <.001 b | |||
| Mean (SD) | 1.9 (3.49) | 2.3 (4.12) | 2.1 (3.86) | |
| Median | 1.0 | 1.0 | 1.0 | |
| Range | 0.0, 166.0 | 0.0, 144.0 | 0.0, 166.0 | |
| Payer type, n (%) | <.001 b | |||
| Commercial | 78 738 (49.8) | 118 959 (61.3) | 197 697 (56.1) | |
| Government | 74 622 (47.2) | 71 173 (36.6) | 145 795 (41.4) | |
| Unknown | 4782 (3.0) | 4081 (2.1) | 8863 (2.5) |
Kruskal-Wallis P-value.
Chi-square P-value.
Across all 5 patient-reported social risk domains, the proportion of at-risk responses was significantly higher among rural patients compared to urban patients (Figure 1(a)). The prevalence of at-risk social factors in the rural cohort ranged from 15.6% for financial resource risk to 3.3% for transportation risk and intimate partner violence. Rural participants were more likely to have missing social risk data than urban participants, ranging from 48.5% for transportation risk to 52.7% for housing risk (Figure 1(b)). The distribution of patients among HOUSES quartiles demonstrated a higher proportion of rural patients in HOUSES Q1, Q2, and Q3 and a lower proportion of rural patients in HOUSES Q4 (Figure 1(c)).
Figure 1.
Patient-reported social risk domains, missing social risk data, and HOUSES index by rural/urban status among study population: (a) proportion of patients reporting at-risk responses for 5 social risk domains, (b) proportion of patients missing social risk data for 5 social risk domains, and (c) proportion of rural and urban residents by HOUSES Index Quartile.
The proportion of at-risk patient-reported social risk answers among all study patients stratified by HOUSES quartiles and rural/urban status is reported in Table 2. Within each social risk domain, rural patients consistently reported higher proportions of at-risk answers across each HOUSES quartile. Regardless of rural/urban status, the lower the HOUSES quartile (lower SES), the higher the percentage of patients reporting an at-risk answer. Higher percentages of missing data were noted in lower HOUSES quartiles across both rural and urban cohorts.
Table 2.
Patient-Reported Social Risk Factors among Study Population stratified by HOUSES Quartiles and Rural/Urban Status.
| HOUSES quartile | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Total | P | ||||||
| Social Risk Factors | Rural N = 32 982 | Urban N = 39 667 | Rural N = 37 990 | Urban N = 41 531 | Rural N = 42 170 | Urban N = 50 046 | Rural N = 45 000 | Urban N = 62 969 | Rural N = 158 142 | Urban N = 194 213 | |
| Financial resource risk, n (%) b | <.001 a | ||||||||||
| Low risk | 10 525 (71.9) | 16 232 (77.8) | 15 217 (81.3) | 20 471 (85.3) | 19 127 (87.1) | 26 626 (89.9) | 22 388 (92.1) | 35 585 (93.5) | 67 257 (84.5) | 98 914 (87.9) | |
| Medium risk | 2885 (19.7) | 3234 (15.5) | 2541 (13.6) | 2556 (10.7) | 2153 (9.8) | 2220 (7.5) | 1470 (6.0) | 1869 (4.9) | 9049 (11.4) | 9879 (8.8) | |
| High risk | 1238 (8.5) | 1403 (6.7) | 949 (5.1) | 965 (4.0) | 683 (3.1) | 775 (2.6) | 456 (1.9) | 608 (1.6) | 3326 (4.2) | 3751 (3.3) | |
| Missing c | 18 334 (55.6) | 18 798 (47.4) | 19 283 (50.8) | 17 539 (42.2) | 20 207 (47.9) | 20 425 (40.8) | 20 686 (46.0) | 24 907 (39.6) | 78 510 (49.6) | 81 669 (42.1) | |
| Food insecurity, n (%) b | <.001 a | ||||||||||
| Low risk | 11 850 (82.7) | 17 750 (86.7) | 16 612 (90.6) | 21 822 (92.5) | 20 265 (93.9) | 27 924 (95.3) | 23 233 (96.6) | 36 652 (97.1) | 71 960 (91.9) | 104 148 (93.7) | |
| At risk | 2486 (17.3) | 2713 (13.3) | 1726 (9.4) | 1781 (7.5) | 1312 (6.1) | 1375 (4.7) | 814 (3.4) | 1097 (2.9) | 6338 (8.1) | 6966 (6.3) | |
| Missing c | 18 646 (56.5) | 19 204 (48.4) | 19 652 (51.7) | 17 928 (43.2) | 20 593 (48.8) | 20 747 (41.5) | 20 953 (46.6) | 25 220 (40.1) | 79 844 (50.5) | 83 099 (42.8) | |
| Housing risk, n (%) b | <.001 a | ||||||||||
| Low risk | 11 098 (82.1) | 16 817 (85.4) | 15 565 (89.1) | 20 530 (90.1) | 18 934 (91.7) | 26 205 (92.9) | 21 734 (94.0) | 34 420 (94.2) | 67 331 (90.1) | 97 972 (91.4) | |
| At risk | 2420 (17.9) | 2881 (14.6) | 1899 (10.9) | 2252 (9.9) | 1715 (8.3) | 2003 (7.1) | 1385 (6.0) | 2113 (5.8) | 7419 (9.9) | 9249 (8.6) | |
| Missing c | 19 464 (59.0) | 19 969 (50.3) | 20 526 (54.0) | 18 749 (45.1) | 21 521 (51.0) | 21 838 (43.6) | 21 881 (48.6) | 26 436 (42.0) | 83 392 (52.7) | 86 992 (44.8) | |
| Transportation risk, n (%) b | <.001 a | ||||||||||
| Low risk | 13 931 (92.9) | 19 995 (94.5) | 18 418 (96.2) | 23 616 (96.9) | 21 936 (97.6) | 29 577 (98.0) | 24 453 (98.5) | 38 257 (98.6) | 78 738 (96.7) | 11 1445 (97.3) | |
| At risk | 1058 (7.1) | 1159 (5.5) | 720 (3.8) | 749 (3.1) | 528 (2.4) | 595 (2.0) | 373 (1.5) | 548 (1.4) | 2679 (3.3) | 3051 (2.7) | |
| Missing c | 17 993 (54.6) | 18 513 (46.7) | 18 852 (49.6) | 17 166 (41.3) | 19 706 (46.7) | 19 874 (39.7) | 20 174 (44.8) | 24 164 (38.4) | 76 725 (48.5) | 79 717 (41.0) | |
| Intimate partner violence risk, n (%) b | <.001 a | ||||||||||
| Low risk | 13 267 (94.8) | 19 122 (95.3) | 17 399 (96.4) | 22 505 (96.6) | 20 651 (97.1) | 28 110 (97.5) | 23 132 (97.9) | 36 109 (97.8) | 74 449 (96.7) | 105 846 (97.0) | |
| At risk | 735 (5.2) | 938 (4.7) | 656 (3.6) | 782 (3.4) | 616 (2.9) | 731 (2.5) | 499 (2.1) | 794 (2.2) | 2506 (3.3) | 3245 (3.0) | |
| Missing c | 18 980 (57.5) | 19 607 (49.4) | 19 935 (52.5) | 18 244 (43.9) | 20 903 (49.6) | 21 205 (42.4) | 21 369 (47.5) | 26 066 (41.4) | 81 187 (51.3) | 85 122 (43.8) | |
Chi-Square P-value.
Percentages reported for columns of completed responses within each social risk domain and HOUSES quartile.
Missing data in italics reported as percentage of quartile population.
There were 201 077 patients included in the mixed effects logistic regression model predicting the likelihood of reporting an at-risk answer to any of the 5 social risk factors and rural versus urban status after excluding patients with missing data (Table 3). For the first model, after controlling for demographics (age, gender, and race) and healthcare-related data (ACG score and insurance payer), those in the rural cohort had a slightly higher odds of reporting an at-risk social risk factor than their urban counterparts (OR = 1.09, 95% CI = 1.02-1.12). In the second model including HOUSES quartile and the interaction between HOUSES and rurality, a strong relationship between HOUSES quartile and social risk factors was noted, but not with rurality (OR = 1.02, 95% CI = 0.97-1.07). There was an increased likelihood of reporting any at-risk social factor with lower HOUSES in a dose response manner, with HOUSES Q1 having the greatest odds of reporting at least 1 social risk (OR = 2.27, 95% CI = 2.19-2.37). The strength of these associations was more pronounced in the rural cohort compared to the urban cohort for HOUSES Q1 (rural vs urban OR = 1.09, 95% CI = 1.02-1.17), but there were no significant differences in rurality for any of the other 3 quartiles.
Table 3.
Mixed Effects Model a of Likelihood of Reporting Any At-Risk Social Risk Domain by HOUSES Quartile among Study Population (N = 201 077).
| Rural/urban model a | Rural/urban + HOUSES model a | |||
|---|---|---|---|---|
| Independent Variables | Odds ratio (95% CI) | P-value | Odds ratio (95% CI) | P-value |
| Rural/Urban Status and HOUSES Quartile | ||||
| Rural (RUCA 4-10) | 1.07 (1.02, 1.12) | .005 | 1.02 (0.97, 1.07) | .446 |
| Urban (RUCA 1-3) | Reference | Reference | ||
| HOUSES Q1 (Lowest SES) | — | 2.27 (2.19-2.37) | <.001 | |
| HOUSES Q2 | — | 1.62 (1.55-1.68) | <.001 | |
| HOUSES Q3 | — | 1.26 (1.21-1.31) | <.001 | |
| HOUSES Q4 (Highest SES) | — | Reference | ||
| Interaction | — | .002 | ||
| Rural vs Urban (HOUSES Q1) | — | 1.09 (1.02, 1.17) | ||
| Rural vs Urban (HOUSES Q2) | — | 1.01 (0.95, 1.08) | ||
| Rural vs Urban (HOUSES Q3) | — | 1.04 (0.97, 1.11) | ||
| Rural vs Urban (HOUSES Q4) | — | 0.94 (0.88, 1.01) | ||
Abbreviations: CI, confidence interval; RUCA, rural urban commuting area; SES, socioeconomic status.
Additional fixed effects adjusted for include age, gender, race, ACG score, insurance payer type, and number of outpatient visits; county was included as a random effect; variance components was used as the covariance structure.
Discussion
Results from this large, cross-sectional study show that social risks are common in our primary care population, with a higher prevalence among rural patients. Additionally, socioeconomic risk based on the HOUSES quartile was independently associated with an increased likelihood of reporting any elevated social risk in both rural and urban populations. This correlation was strong enough that rurality is not an independent predictor of social risk in the mixed effects model analysis.
Our results also revealed significant differences between rural and urban patients. Rural patients in this study tended to be older, less racially diverse, have more chronic conditions, and have less interaction with the healthcare system. Rural patients were more likely to have government insurance and made fewer visits to primary care than urban patients. These findings are consistent with previously reported research on rural populations.23,33 Rural patients in our study were more likely to report financial resource risk, housing risk, food insecurity, transportation risk, and intimate partner violence and the prevalence of these social risk factors suggest they are common among primary care patients. Rural patients were also of lower SES by HOUSES quartile than their urban counterparts. Unexpectedly, we found that when controlling for covariates and HOUSES quartile, rurality defined by RUCA codes did not independently predict social risk concerns except for an interaction between rurality and HOUSES Q1. This data suggests individual-level HOUSES Index may be a surrogate marker for identifying patients with social risk concerns among both rural and urban populations.
Another important finding of this study is that the pattern of missing data is non-random. Rural patients were more likely to be missing social risk data, suggesting that the rural-urban disparities revealed here may be underestimated. This is consistent with previous research showing that rural healthcare professionals experience significant organizational, community, and policy barriers to screening for social risks.34,35 We found a similar issue among patients with a low SES/HOUSES quartile. This supports prior research suggesting that patients who are at the highest socioeconomic risk may be the least likely to seek health care or to engage in social risk assessments.36-38
With increasing emphasis directed at assessing and intervening on social risk factors, there is a greater need to understand the prevalence, distribution, and associations of social risk factors in underserved communities to inform effective interventions. The problem of missing social risk data in healthcare social risk screenings can introduce structural inequities that further disadvantage vulnerable populations. Identifying at-risk populations such as those individuals with lower HOUSES quartiles may enhance implementation of social needs interventions. The HOUSES Index is widely available for US addresses, standardized, passively collected, and significantly associated with patient-reported social risks. Measuring individual SES using the HOUSES Index can also simplify assessing SES over time and space for healthcare institutions and researchers, enabling private and public institutions to direct social interventions toward the highest-risk populations.
The findings in this study emphasize the importance of understanding SES in rural communities when addressing social risk factors. Rural communities are heterogeneous, the binary division between rural and urban used in this and similar studies is an inadequate marker for the variability and complexity of these communities.39-41 Within this study, the most rural areas had variable proportions of patients with social risk concerns (from 11.9% to 17.1%) and HOUSES Q1 (from 14.1% to 40.4%). These differences in social risk vulnerability help to explain why solutions to rural health socioeconomic disparities may work in some areas but not in others. Rural areas are challenged by geographic spread, low population density, workforce challenges, and limited infrastructure requiring a more nuanced approach to socioeconomic interventions.42,43 Identification of the most vulnerable rural residents through use of HOUSES quartiles may assist with targeting interventions to improve effectiveness.
Our study has several strengths. First, our comprehensive sample across the entire community-based health system increased our ability to detect significant differences and report prevalences of at-risk social risk factors. Second, our primary care population had a high percentage (44.9%) of patients living rurally allowing for greater statistical power to assess relationships between our rural and urban populations. Lastly, we measured individual SES passively with HOUSES Index, which enhances our knowledge of the socioeconomic environment of our primary care population beyond patient-reported social risks.
Our results should be interpreted with the following precautions. As mentioned, almost 43% of our population were missing social risk data, which replicates real-world clinical experience but may underestimate the social risks of rural and disadvantaged populations as missing data is more common in rural and lower HOUSES quartiles. Unfortunately, non-response bias is widespread in social risk data, but our findings among individuals missing data is similar to previous studies. 37 Our data is from a subset of the population served by our primary care network and may not apply to the entire population of this region; however, the descriptions of our rural patients are similar to reported literature. These results may not generalize to other populations who are more geographically remote, racially diverse, or socially disadvantaged than our US Midwest residents, but the study findings can be replicated in other populations with the broad US availability of the HOUSES Index. And finally, due to the nature of cross-sectional analyses, causation cannot be determined.
Improving knowledge of social risk factors and SES in rural communities is essential to the development of interventions to reduce rural socioeconomic disparities. Additional study is needed to understand the associations of rurality, patient-reported social risks, and the HOUSES Index with quality health outcomes such as cancer screening rates, chronic disease management, and healthcare delivery/utilization. The HOUSES Index is a valuable tool to assess individual-level SES in the absence of or to augment patient-reported social risk data in healthcare settings. As healthcare systems continue to expand social needs assessment and interventions in rural areas, further study of the potential applications of HOUSES Index to identify and target the most disadvantaged populations to address rural health disparities is needed.
Conclusion
We found that the increased prevalence of social risk factors among rural residents is largely explained by individual SES as measured by the HOUSES Index. The HOUSES Index as a measure of SES can be a useful proxy when social risk factor or other SES data is missing, which occurs more frequently among patients with lower SES and in rural areas. This research will help researchers, government entities, community-based organizations, and healthcare organizations better understand how SES and social risk drive the health disparities between rural and urban populations.
Supplemental Material
Supplemental material, sj-docx-1-jpc-10.1177_21501319251369673 for The Association Between Patient-Reported Social Risks and the HOUSES Index: A Rural-Urban Comparison by Jessica L. Sosso, Karen M. Fischer, Chung-Il Wi, Dominika A. Jegen, Marc Matthews, Julie Maxson, Matthew E. Bernard, Stephen K. Stacey, Randy M. Foss, Brandon Hidaka, Rachael Passmore, Gregory M. Garrison and Tom D. Thacher in Journal of Primary Care & Community Health
Supplemental material, sj-docx-2-jpc-10.1177_21501319251369673 for The Association Between Patient-Reported Social Risks and the HOUSES Index: A Rural-Urban Comparison by Jessica L. Sosso, Karen M. Fischer, Chung-Il Wi, Dominika A. Jegen, Marc Matthews, Julie Maxson, Matthew E. Bernard, Stephen K. Stacey, Randy M. Foss, Brandon Hidaka, Rachael Passmore, Gregory M. Garrison and Tom D. Thacher in Journal of Primary Care & Community Health
Acknowledgments
This study was made possible using the resources of the HOUSES Program, part of the Mayo Clinic Precision Population Science Lab. The content of this article is solely the responsibility of the authors and does not represent the official views of the HOUSES Program.
Footnotes
ORCID iDs: Jessica L. Sosso 
https://orcid.org/0009-0000-6088-6091
Chung-Il Wi
https://orcid.org/0000-0001-8938-2997
Dominika A. Jegen
https://orcid.org/0000-0003-2075-8778
Randy M. Foss
https://orcid.org/0000-0003-3898-7317
Brandon Hidaka
https://orcid.org/0000-0002-2378-3621
Rachael Passmore
https://orcid.org/0009-0005-2905-0217
Gregory M. Garrison
https://orcid.org/0000-0001-6903-4610
Tom D. Thacher
https://orcid.org/0000-0002-7644-8173
Ethical Considerations: The Mayo Clinic Institutional Review Board reviewed the protocol and deemed the study exempt from full review due to negligible risk to patients. Participants who did not give prior authorization for research use of health record data were excluded from the study.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Mayo Clinic Health System Rural Health Seed Grant and the Mayo Clinic Department of Family Medicine supported this project. The Mayo Clinic and Mayo Clinic Health System 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.
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Drs. Wi is affiliated with the HOUSES Program of the Mayo Clinic Precision Population Science Lab. Dr. Sosso is a sub-awardee on NIH grant R56HL173775 – The Role of the Urban/Rural Divide and Socioeconomic Factors in the Incidence, Outcomes, and Post-partum Care of Hypertensive Disorders of Pregnancy. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement: The datasets generated and/or analyzed for this study are not publicly available as datasets contain protected health information. De-identified data may be available upon request from the authors.
Supplemental Material: Supplemental material for this article is available online.
References
- 1. Hood CM, Gennuso KP, Swain GR, Catlin BB. County health rankings: relationships between determinant factors and health outcomes. Am J Prev Med. 2016;50(2):129-135. doi: 10.1016/j.amepre.2015.08.024 [DOI] [PubMed] [Google Scholar]
- 2. Magnan S. Social determinants of health 101 for health care: five plus five. NAM Perspectives. Discussion Paper, National Academy of Medicine. 2017. Accessed March 3, 2025. https://nam.edu/wp-content/uploads/2017/10/Social-Determinants-of-Health-101.pdf [Google Scholar]
- 3. Social Determinants of Health. United States Department of Health and Human Services. Accessed May 2, 2024. https://health.gov/healthypeople/priority-areas/social-determinants-health
- 4. Alderwick H, Gottlieb LM. Meanings and misunderstandings: a social determinants of health lexicon for health care systems. Milbank Q. 2019;97(2):407-419. doi: 10.1111/1468-0009.12390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. National Academies of Sciences, Engineering, and Medicine. 2019. Integrating social care into the delivery of health care: Moving upstream to improve the Nation’s health. The National Academies Press. doi: 10.17226/2546 [DOI] [PubMed] [Google Scholar]
- 6. Gusoff G, Fichtenberg C, Gottlieb L. Professional Medical Association policy statements on social health assessments and interventions. Perm J. 2018;22:18-092. doi: 10.7812/TPP/18-092 [DOI] [Google Scholar]
- 7. Medicare Program; Hospital Inpatient Prospective Payment Systems for Acute Care Hospitals and the Long Term Care Hospital Prospective Payment System and Policy Changes and Fiscal Year 2023. Rates; Quality Programs and Medicare Promoting Interoperability Program Requirements for Eligible Hospitals and Critical Access Hospitals; Costs Incurred for Qualified and Non-Qualified Deferred Compensation Plans; and Changes to Hospital and Critical Access Hospital Conditions of Participation In: Department of Health and Human Services CMS, editor. Federal Register; 2022:35934-36649. Accessed July 17, 2024. https://www.govinfo.gov/content/pkg/FR-2024-05-02/pdf/2024-07567.pdf [Google Scholar]
- 8. The Joint Commission. R3 Report: Requirement, Rationale, Reference: New Requirements to Reduce Health Care Disparities. 2022. Accessed July 17, 2024. https://www.jointcommission.org/en-us/standards/r3-report/r3-report-36
- 9. Lofters AK, Schuler A, Slater M, et al. Using self-reported data on the social determinants of health in primary care to identify cancer screening disparities: opportunities and challenges. Multicenter Study. BMC Family Practice. 2017;18(1):31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Cottrell EK, Dambrun K, Cowburn S, et al. Variation in electronic health record documentation of Social Determinants of Health across a National Network of Community Health Centers. Am J Prev Med. 2019;57(6 Suppl 1):S65-S73. doi: 10.1016/j.amepre.2019.07.014 [DOI] [PubMed] [Google Scholar]
- 11. Stewart de, Ramirez S, Shallat J, McClure K, Foulger R, Barenblat L. Screening for Social Determinants of Health: active and passive information retrieval methods. Popul Health Manag. 2022;25(6):781-788. doi: 10.1089/pop.2022.0228 [DOI] [PubMed] [Google Scholar]
- 12. Vest JR, Wu W, Mendonca EA. Sensitivity and specificity of real-world social factor screening approaches. J Med Syst. 2021;45(12):111. doi: 10.1007/s10916-021-01788-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Juhn YJ, Beebe TJ, Finnie DM, et al. Development and initial testing of a new socioeconomic status measure based on housing data. J Urban Health. 2011;88(5):933-944. doi: 10.1007/s11524-011-9572-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Mayo Clinic. Mayo Clinic Housing-Based Socioeconomic Status (HOUSES) Program. Accessed July 14, 2024. https://www.mayo.edu/research/centers-programs/mayo-clinic-houses-program/about/methodology
- 15. Juhn YJ, Ryu E, Wi CI, et al. Assessing socioeconomic bias in machine learning algorithms in health care: a case study of the HOUSES index. J Am Med Inform Assoc. 2022;29(7):1142-1151. doi: 10.1093/jamia/ocac052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Greenwood J, Zurek KI, Grimm JM, Wi CI, Vogel JT, Garrison GM. Association of a housing based individual socioeconomic status measure with diabetic control in primary care practices. Prim Care Diabetes. 2022;16(1):78-83. doi: 10.1016/j.pcd.2021.10.001 [DOI] [PubMed] [Google Scholar]
- 17. Angstman KB, Wi CI, Williams MD, Bohn BA, Garrison GM. Impact of socioeconomic status on depression clinical outcomes at six months in a Midwestern, United States community. J Affect Disord. 2021;292:751-756. doi: 10.1016/j.jad.2021.05.098 [DOI] [PubMed] [Google Scholar]
- 18. Patten CA, Juhn YJ, Ryu E, et al. Rural-urban health disparities for mood disorders and obesity in a midwestern community. J Clin Transl Sci. 2020;4(5):408-415. doi: 10.1017/cts.2020.27 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Takahashi PY, Ryu E, Hathcock MA, et al. A novel housing-based socioeconomic measure predicts hospitalisation and multiple chronic conditions in a community population. J Epidemiol Community Health. 2016;70(3):286-291. doi: 10.1136/jech-2015-205925 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Wi CI, St, Sauver JL, Jacobson DJ, et al. Ethnicity, socioeconomic status, and health disparities in a mixed rural-urban US Community-Olmsted County, Minnesota. Mayo Clin Proc. 2016;91(5):612-622. doi: 10.1016/j.mayocp.2016.02.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Stevens MA, Beebe TJ, Wi CI, Taler SJ, St, Sauver JL, Juhn YJ. HOUSES Index as an innovative socioeconomic measure predicts graft failure among kidney transplant recipients. Transplantation. 2020;104(11):2383-2392. doi: 10.1097/TP.0000000000003131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Kizilbash S, Wi CI, Roy M, et al. Socioeconomic inequities in preemptive kidney transplantation and graft survival: an innovative approach to identifying disparities in kidney transplantation. Transplant Direct. 2025;11(1):e1734. doi: 10.1097/TXD.0000000000001734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Agency for Healthcare Research and Quality. National Healthcare Quality and Disparities Report Chartbook on Rural Healthcare. Agency for Healthcare Research and Quality. AHRQ Pub. No. 22-0010; 2021 [PubMed] [Google Scholar]
- 24. Erwin PC, Fitzhugh EC, Brown KC, Looney S, Forde T. Health disparities in rural areas: the interaction of race, socioeconomic status, and geography. J Health Care Poor Underserved. 2010;21(3):931-945. doi: 10.1353/hpu.0.0336 [DOI] [PubMed] [Google Scholar]
- 25. Randolph R, Thomas S, Holmes M, et al. Rural population health in the United States: a Chartbook. North Carolina Rural Health Research Program, Cecil G. Sheps Center for Health Services Research, The University of North Carolina at Chapel Hill. 2023. Accessed February 4, 2024. https://www.ruralhealthresearch.org/publications/1572 [Google Scholar]
- 26. Ricketts TC. The changing nature of rural health care. Annu Rev Public Health. 2000;21:639-657. doi: 10.1146/annurev.publhealth.21.1.639 [DOI] [PubMed] [Google Scholar]
- 27. Curtin SC, Spencer MR. Trends in death rates in urban and rural areas: United States, 1999-2019. NCHS Data Brief. 2021;417:1-8. [PubMed] [Google Scholar]
- 28. Probst JC, Bellinger JD, Walsemann KM, Hardin J, Glover SH. Higher risk of death in rural blacks and whites than urbanites is related to lower incomes, education, and health coverage. Health Aff. 2011;30(10):1872-1879. doi: 10.1377/hlthaff.2011.0668 [DOI] [PubMed] [Google Scholar]
- 29. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-1457. doi: 10.1016/S0140-6736(07)61602-X [DOI] [PubMed] [Google Scholar]
- 30. USDA. Rural-Urban Commuting Area Codes. 2023. Accessed July 10, 2023. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes/documentation/
- 31. Health Resources and Services Administration. How We Define Rural. Health Resources & Services Administration. 2025. Accessed April 21, 2025. https://www.hrsa.gov/rural-health/about-us/what-is-rural
- 32. Johns Hopkins University Bloomberg School of Public Health. The Johns Hopkins ACG Case-Mix System Documentation & Application. Johns Hopkins University Bloomberg School of Public Health; 2001. [Google Scholar]
- 33. Harrington RA, Califf RM, Balamurugan A, et al. Call to action: rural health: a presidential advisory from the American Heart Association and American Stroke Association. Circulation. 2020;141(10):e615-e644. doi: 10.1161/CIR.0000000000000753 [DOI] [PubMed] [Google Scholar]
- 34. Dauner KN, Loomer L. A qualitative assessment of barriers and facilitators associated with addressing social determinants of health among members of a health collaborative in the rural Midwest. BMC Health Serv Res. 2021;21(1):867. doi: 10.1186/s12913-021-06859-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Maganty A, Byrnes ME, Hamm M, et al. Barriers to rural health care from the provider perspective. Rural Remote Health. 2023;23(2):7769. doi: 10.22605/RRH7769 [DOI] [PubMed] [Google Scholar]
- 36. Lazar M, Davenport L. Barriers to health care access for low income families: a review of literature. J Community Health Nurs. 2018;35(1):28-37. doi: 10.1080/07370016.2018.1404832 [DOI] [PubMed] [Google Scholar]
- 37. Bharmal N, Rennick A, Shideler A, et al. Health-related social needs: which patients respond to screening and who receives resources? J Gen Intern Med. 2023;38(12):2695-2702. doi: 10.1007/s11606-023-08135-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Vest JR, Mazurenko O. Non-response bias in social risk factor screening among adult emergency department patients. J Med Syst. 2023;47(1):78. doi: 10.1007/s10916-023-01975-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Bennett KJ, Borders TF, Holmes GM, Kozhimannil KB, Ziller E. What is rural? Challenges and implications of definitions that inadequately encompass rural people and places. Health Aff. 2019;38(12):1985-1992. doi: 10.1377/hlthaff.2019.00910 [DOI] [PubMed] [Google Scholar]
- 40. Wang M, Garshick Kleit R, Cover J, Fowler C. Spatial variations in US poverty: beyond metropolitan and non-metropolitan. Urban Stud. 2011;49(3):563-585. doi: 10.1177/0042098011404932 [DOI] [PubMed] [Google Scholar]
- 41. Cordes SM. Biopsychosocial imperatives from the rural perspective. Soc Sci Med. 1985;21(12):1373-1379. doi: 10.1016/0277-9536(85)90446-0 [DOI] [PubMed] [Google Scholar]
- 42. Chapman A, Buccheri A, Mohotti D, et al. Staff-reported barriers and facilitators to the implementation of healthcare interventions within regional and rural areas: a rapid review. BMC Health Serv Res. 2025;25(1):331. doi: 10.1186/s12913-025-12480-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Mullan L, Armstrong K, Job J. Barriers and enablers to structured care delivery in Australian rural primary care. Aust J Rural Health. 2023;31(3):361-384. doi: 10.1111/ajr.12963 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material, sj-docx-1-jpc-10.1177_21501319251369673 for The Association Between Patient-Reported Social Risks and the HOUSES Index: A Rural-Urban Comparison by Jessica L. Sosso, Karen M. Fischer, Chung-Il Wi, Dominika A. Jegen, Marc Matthews, Julie Maxson, Matthew E. Bernard, Stephen K. Stacey, Randy M. Foss, Brandon Hidaka, Rachael Passmore, Gregory M. Garrison and Tom D. Thacher in Journal of Primary Care & Community Health
Supplemental material, sj-docx-2-jpc-10.1177_21501319251369673 for The Association Between Patient-Reported Social Risks and the HOUSES Index: A Rural-Urban Comparison by Jessica L. Sosso, Karen M. Fischer, Chung-Il Wi, Dominika A. Jegen, Marc Matthews, Julie Maxson, Matthew E. Bernard, Stephen K. Stacey, Randy M. Foss, Brandon Hidaka, Rachael Passmore, Gregory M. Garrison and Tom D. Thacher in Journal of Primary Care & Community Health