Abstract
Objective:
Independent living is desirable for many older adults. Although several factors such as physical and cognitive functions are important predictors for nursing home placement (NHP), it is also reported that socioeconomic status (SES) affects the risk of NHP. In this study, we aimed to examine whether an individual-level measure of SES is associated with the risk of NHP after accounting for neighborhood characteristics.
Design:
A population-based study (Olmsted County, Minnesota, USA).
Setting and Participants:
Older adults (age 65+ years) with no prior history of NHP.
Methods:
Electronic health records (EHR) were used to identify individuals with any NHP between April 1, 2012 (baseline date) and April 30, 2019. Association between the (HOUsing-based index of SocioEconomic Status (HOUSES) index, an individual-level SES measure based on housing characteristics of current residence, and risk of NHP was tested using random effects Cox proportional hazard model adjusting for area deprivation index (ADI), an aggregated SES measure that captures neighborhood characteristics, and other pertinent confounders such as age and chronic disease burden.
Results:
Among 15,031 older adults, 3341 (22.2%) experienced NHP during follow-up period (median: 7.1 years). At baseline date, median age was 73 years old with 55% female persons, 91% non-Hispanic Whites, and median number of chronic conditions of 4. Accounting for pertinent confounders, the HOUSES index was strongly associated with risk of NHP (hazard ratio 1.89; 95% confidence interval 1.66–2.15 for comparing the lowest vs highest quartiles), which was not influenced by further accounting for ADI.
Conclusions and Implications:
This study demonstrates that an individual-level SES measure capturing current individual-specific socioeconomic circumstances plays a significant role for predicting NHP independent of neighborhood characteristics where they reside. This study suggests that older adults who are at higher risk of NHP can be identified by utilizing the HOUSES index and potential individual-level intervention strategies can be applied to reduce the risk for those with higher risk.
Keywords: Socioeconomic status, the HOUSES index, nursing home placement, area deprivation index
Introduction
Independent living among older adults is desirable, even those who are seriously ill, but often challenging and may not be an option for everyone as they age due to aging-related impairments and social or environmental circumstances (eg, living alone).1 At present, roughly 15% of people who are 85 years or older live in nursing homes with about 1% of those 65–75 years of age.2 As the US population is aging rapidly (the number of older adults age 65 years or order is projected to increase from 52 million in 2018 to 95 million by 2060),3 the number of seniors at risk of being a nursing home resident will increase drastically. For instance, admission to skilled nursing facilities (SNFs) is particularly high in patients with Alzheimer’s disease at 49.1% compared with 37.1% in patients without dementia having Medicare fee-for-service.4 For those admitted to SNF, there was a high mortality rate due to COVID-19 infection during the initial part of the pandemic; however, COVID-19 vaccination has improved care in SNF.5-7 Therefore, it will be crucial to identify those who are at high risk for nursing home placement (NHP) and then provide necessary interventions to mitigate the risk.8,9 With changes to reimbursement and acceptance of advanced payment models in Centers for Medicare and Medicaid Services such as accountable care organizations,10 hospitals and health organizations have increased interest in identifying those patients at highest risk of NHP.
While there exist several known risk factors for NHP such as age, cognitive impairment, and frailty,11-14 another important consideration for NHP involves social and environmental circumstances within the patient’s life [ie, social determinants of health (SDoH)]. For instance, caregiver support plays an important role, particularly after a hospital stay. Given that the presence of a caregiver can help reduce risk of NHP, living alone places a patient at risk for NHP.8 Additionally, in a study based on older adults receiving caregiver support, factors affecting long-term NHP extend patient’s individual characteristics to their social context, including caregiver’s circumstances such as physical, emotional, and financial strains.15 Individuals with significant problems related to SDoH tend to have worse health, are less able to finance their long-term care expense at home, and are more likely to feel lonely, which put them at a higher risk for NHP.16 Adverse characteristics linked to SDoH (eg, financial hardship and limited social support) may also influence the level of support within the home or access to needed medical care. Therefore, the importance of integration between SDoH and medical care is increasingly recognized for those who undergo medical transition after significant health problems such as hospitalization. Despite its importance, detailed SDoH information that can be used for identifying those at risk is typically not readily available in electronic health records (EHRs), which limits intervention efforts that can mitigate the risk of NHP. In particular, active attainment of information through questionnaires may be burdensome for patients and staff.
As a key element of SDoH, the role of socioeconomic status (SES) on NHP has been widely recognized. Along with educational attainment,17 neighborhood-level SES measures have been studied for their association with NHP risk. For example, in patients undergoing lower extremity surgery, zip code was used to predict NHP showing weak association.11 A recent study showed that residents in nursing homes located in socially deprived areas had a higher rate of COVID-19 infection, partly because of lack of resources and crowdedness, which emphasizes an importance of neighborhood-level intervention.18 However, little is known about whether additional individual-level SES improves an ability to predict NHP beyond what neighborhood-level SES offers.
Therefore, we aimed to test whether an individual-level SES contributes to risk of NHP after accounting for neighborhood characteristics by conducting a population-based study from an upper Midwest area where the health care environment is nearly self-contained. We utilized the HOUsing-based index of SocioEconomic Status (HOUSES) index, an individual-level SES measure, which uses housing characteristics of current residence that may reflect current patient-specific social and/or environmental circumstances. Given that many risk factors associated with NHP are patient-specific (eg, frailty and social isolation), we hypothesize that the HOUSES index will provide an additional ability to predict NHP beyond what neighborhood-level SES captures.
Materials and Methods
Study Individuals and Setting
This is a population-based study using data from older adults (age 65 years or older on April 1, 2012) who were residents of Olmsted County, Minnesota, USA where the health care environment is self-contained with unified medical records available through Rochester Epidemiology Project.19 The age, sex, and race/ethnic characteristics of Olmsted County residents are similar to those in the upper Midwest although a higher proportion of Olmsted County residents work in the health care industry.20 This study was reviewed and approved by both the Mayo Clinic Institutional Review Board and the Olmsted Medical Center Institutional Review Board.
Ascertaining Individuals with Nursing Home Placement
Starting from April 1, 2012 (baseline date), eligible individuals (with no prior history of NHP) were followed up passively through EHRs until April 30, 2019, to identify individuals who experienced NHP during the study period (ie, any admission regardless of the length of stay). Individuals with NHP were identified by using current procedural terminology (CPT) visit billing codes associated with NHP (99304, 99305, 99306, 99307, 99308, 99309, 99310, 99315, 99316, and 99318) and associated dates were used as the index date. Individuals who were nursing home residents prior to April 1, 2012, were excluded. For those with multiple records during study period, the date for the first CPT code was considered as the index date. Individuals were followed up until the last clinic visit, April 30, 2019, death date, or the index date, whichever came first. Although the main analysis was focused on any nursing home admission, we also conducted a subset analysis using individuals who had another CPT code at least 90 days after the date for the first CPT code (assuming their nursing home stay would be longer than 90 days; refer to as long-term stay) to see if length of nursing home stay affects the association results.
SES Measures
The primary SES measure is the HOUSES index, which is a validated individual-level housing-based SES measure based on 4 housing features (house value, size of house, the number of bedrooms, and the number of bathrooms).21 To formulate the HOUSES index, each housing feature was first transformed to a standardized z score, and then the z scores of the 4 housing features were summed to create a composite score. The HOUSES index was calculated based on residence at baseline date (April 1, 2012). Quartiles of the HOUSES index based on Olmsted County population in 2012 were used in the analysis with Q1 being the lowest SES and Q4 being the highest SES. As a neighborhood-level SES, we used the area deprivation index (ADI),22,23 a neighborhood-level SES measure within each Census block group that ranks neighborhoods with respect to socioeconomic disadvantage using 17 census variables such as income, education, household characteristics, and housing, normalized within the whole US [ranging from 1% (least deprived) to 100% (most deprived)]. Similar to the HOUSES index, ADI was also analyzed in quartiles (Q1: highest SES; Q4: lowest SES). As a secondary individual-level SES measure, educational attainment (high school or less, some college, 4-year college, or graduate/professional degree) was also extracted from EHR. Because educational attainment is a commonly used SES measure and is reliably available in EHR, we considered it as a secondary individual-level SES measure in the current study. However, the only analysis considered for educational attainment in this study was to compare its association magnitude for risk of NHP with that of the HOUSES index with an expectation that the HOUSES index would have stronger association with risk of NHP.
Other Covariates
We also retrieved other pertinent variables: age, sex, and race/ethnicity (non-Hispanic Whites vs others) from EHR. As a measure for the degree of multimorbidity (disease burden score), we used the sum of 20 chronic conditions defined by the United States Department of Health and Human Services (ranging 0–20).24,25 Activities of daily living (ADL; a marker for functional status) and living situation (living alone vs with others; a marker for caregiver support), 2 well-known risk factors affecting NHP, were also retrieved from patient-provided information in EHR that were collected within 2 years prior to the baseline date (ie, April 1, 2010–April 1, 2012). For ADL, we used 10 variables (yes/no for each variable) related to activity difficulties: getting in and out of bed, feeding yourself, dressing yourself, using the toilet, housekeeping, climbing stairs, bathing, walking, using transportation, and managing medications.26 ADL was analyzed as a continuous variable (the number of difficulties among the 10 activities) as well as a dichotomized variable (at least 1 vs none).
Statistical Analyses
Basic characteristics of the study individuals were summarized in percentages for categorical variables and median (25th–75th percentiles) for continuous variables. Univariate association between each variable and risk of NHP was tested using Cox proportional hazard models. Adjusting for age, sex, race/ethnicity, and disease burden score, association of each SES measure (the HOUSES index as the primary individual-level SES measure, ADI as a neighborhood-level SES measure, and educational attainment as a secondary individual-level SES measure) was first tested using Cox proportional hazard models. Given that ADI is an aggregated measure having the same score for a given census block group, we used random effect Cox model for testing ADI.
To see whether individual-level SES (the HOUSES index) plays a role on NHP beyond neighborhood characteristics, we adjusted for ADI in addition to variables previously adjusted using random effect Cox model. We also tested a potential interaction effect between the HOUSES index and ADI (to see whether the effect of individual-level SES on NHP differs by neighborhood characteristics). We repeated the same analysis for predicting long-term NHP.
To understand a potential mechanism for association between the HOUSES index and NHP, association of ADL and living situation (2 well-established risk factors for NHP) with the HOUSES index was tested after adjusting for age and sex, with (using random-effect logistic regression model) and without (standard logistic regression model) adjusting for ADI.
Results
Characteristics of Study Cohort
The initial cohort consisted of 16,841 older adults who were age 65 years or older, residents in Olmsted County, Minnesota, at baseline date (April 1, 2012), had a home address that could be geocoded, and provided research authorization to investigators for research. Among those, 1810 individuals (11.7%) were excluded due to a prior history of NHP, which left a total of 15,031individuals to be used in the analysis. In the study cohort with median follow-up duration of 7.1 years, 3341 (22.2%) experienced any nursing home admission (Table 1) with 45% (n = 1474) of those who stayed longer than 90 days. Median age of the cohort at baseline date is 73 years with 55% female adults, 91% non-Hispanic whites and median number of chronic conditions of 4. Older age, female, and non-Hispanic White were associated with a higher risk of NHP. Consistent with literature, increased risk of NHP was observed for individuals having more chronic conditions [hazard ratio (HR) 1.26; 95% confidence interval (CI) 1.24–1.27], having at least 1 ADLs (HR 3.05; 95% CI 2.82–3.31), and living alone (HR 2.03; 95% CI 1.87-2.20; Table 1).
Table 1.
Sociodemographic Characteristics of Study Individuals
| Overall Cohort (N = 15,031) | NHP During Follow-Up Duration | HR* (95% CI) | ||
|---|---|---|---|---|
| Yes (n = 3341) | No (n = 11690) | |||
| Age (y) at index date (April 1, 2012)† | ||||
| Median (25th–75th percentile) | 73 (68–79) | 79 (72–84) | 71 (68–77) | 1.11 (1.11–1.12) |
| Sex, n (%) | ||||
| Female | 8215 (54.7%) | 1997 (59.8%) | 6218 (53.2%) | REF |
| Male | 6816 (45.3%) | 1344 (40.2%) | 5472 (46.8%) | 0.80 (0.74–0.85) |
| Race/ethnicity group, n (%) | ||||
| Non-Hispanic Whites | 13681 (91.4%) | 3170 (94.9%) | 10511 (90.4%) | REF |
| Other | 1290 (8.6%) | 169 (5.1%) | 1121 (9.6%) | 0.57 (0.49-0.67) |
| Missing | 60 (0.4%) | 2 (0.1%) | 58 (0.5%) | 0.19 (0.05-0.75) |
| HOUSES, n (%) | ||||
| Q1 (lowest SES) | 3231 (23.6%) | 1055 (34.7%) | 2176 (20.4%) | 3.09 (2.72-3.50) |
| Q2 | 4515 (33.0%) | 1017 (33.4%) | 3498 (32.8%) | 1.83 (1.62-2.08) |
| Q3 | 3535 (25.8%) | 653 (21.5%) | 2882 (27.1%) | 1.46 (1.27-1.67) |
| Q4 (highest SES) | 2412 (17.6%) | 317 (10.4%) | 2095 (19.7%) | REF |
| Missing | 1338 (8.9%) | 299 (8.9%) | 1039 (8.9%) | 1.87 (1.60-2.20) |
| ADI‡ (national-level), n (%) | ||||
| Q1 (highest SES) | 3766 (25.1%) | 692 (20.7%) | 3074 (26.3%) | REF |
| Q2 | 6043 (40.2%) | 1272 (38.1%) | 4771 (40.8%) | 1.11 (0.91-1.34) |
| Q3 | 3378 (22.5%) | 939 (28.1%) | 2439 (20.9%) | 1.60 (1.30-1.97) |
| Q4 (lowest SES) | 795 (5.3%) | 208 (6.2%) | 587 (5.0%) | 1.49 (1.12-2.00) |
| Missing | 1049 (7.0%) | 230 (6.9%) | 819 (7.0%) | 1.45 (0.62-3.40) |
| Disease burden score (count)†,§ | ||||
| Median (25th—75th percentile) | 4 (3-6) | 6 (4-7) | 4 (2-6) | 1.26 (1.24-1.27) |
| Number of problematic ADLs (count)† | ||||
| Median (25th—75th percentile) | 0 (0–0) | 0 (0–1) | 0 (0–0) | 1.23 (1.21–1.26) |
| Problematic ADLs, n (%) | ||||
| None | 8336 (55.5%) | 1634 (48.9%) | 6702 (57.3%) | REF |
| At least 1 activity | 2248 (15.0%) | 918 (27.5%) | 1330 (11.4%) | 3.05 (2.82–3.31) |
| Missing | 4447 (29.6%) | 789 (23.6%) | 3658 (31.3%) | 0.98 (0.90–1.07) |
| Living situation, n (%) | ||||
| Living alone | 2415 (16.1%) | 855 (25.6%) | 1560 (13.3%) | 2.03 (1.87–2.20) |
| Others | 8457 (56.3%) | 1744 (52.2%) | 6713 (57.4%) | REF |
| Missing | 4159 (27.7%) | 742 (22.2%) | 3417 (29.2%) | 0.92 (0.85–1.01) |
HRs and corresponding 95%CIs were calculated using Cox proportional hazard model to test association between each characteristic and risk of NHP.
For continuous variables, the effect presented is for 1-unit change in the variable
HRs and 95% CIs were calculated using Cox proportional hazard model with a random-effect term for census block groups.
For disease burden score variable, there is no missing value because this variable was defined as the number of chronic conditions that a given individual had at least 1 International Classification of Diseases (ICD) code available in electronic health records as of index dates.
Association of SES Measures with Likelihood of NHP
Univariately, lower SES was associated with a higher risk of NHP for both HOUSES index and ADI. The association signal was stronger when using the HOUSES index: HR 3.09 (95% CI 2.72–3.50) for comparing the lowest SES (Q1) vs the highest SES (Q4) compared with HR 1.49 (95% CI 1.12–2.00) for comparing the lowest SES (Q4) vs the highest SES (Q1) when using ADI (Table 1).
Adjusting for age, sex, race/ethnicity, and disease burden score, lower SES measured by the HOUSES index was associated with higher risk of NHP in a dose-response manner [HR 1.89 (95% CI 1.66–2.15)] for comparing HOUSES Q1 vs Q4, HR 1.36 (95% CI 1.20–1.55) for comparing Q2 vs Q4, and HR = 1.22 (95% CI 1.07–1.40) for comparing Q3 vs Q4; Model 1 in (Table 2). A subset analysis predicting long-term NHP (expected to stay over 90 days) showed slightly stronger association results with the HOUSES index (eg, HR 2.10 (95% CI 1.68–2.63)] for comparing HOUSES Q1 vs Q4; Supplementary Table 1).
Table 2.
Association Between SES Measures (Model 1: HOUSES, Model 2: ADI, and Model 3: Both HOUSES and ADI) and Risk of NHP
| Characteristics | Model 1, HR (95% CI) | Model 2, HR (95% CI) | Model 3, HR (95% CI) |
|---|---|---|---|
| Age (y) | 1.09 (1.09–1.10) | 1.09 (1.09–1.10) | 1.09 (1.09–1.10) |
| Sex | |||
| Male | 0.87 (0.80–0.93) | 0.85 (0.79–0.91) | 0.87 (0.81–0.94) |
| Female | REF | REF | REF |
| Race/ethnicity | |||
| Non-Hispanic White | REF | REF | REF |
| Others | 0.63 (0.54–0.74) | 0.62 (0.53–0.73) | 0.61 (0.52–0.72) |
| Disease burden score (count) | 1.16 (1.15–1.18) | 1.16 (1.15–1.18) | 1.16 (1.15–1.18) |
| HOUSES (quartiles) | |||
| Q1 (lowest SES) | 1.89 (1.66–2.15) | – | 1.81 (1.57–2.10) |
| Q2 | 1.36 (1.20–1.55) | – | 1.35 (1.17–1.55) |
| Q3 | 1.22 (1.07–1.40) | – | 1.23 (1.07–1.41) |
| Q4 (highest SES) | REF | – | REF |
| ADI (quartiles) | |||
| Q1 (highest SES) | – | REF | REF |
| Q2 | – | 1.05 (0.92–1.20) | 0.96 (0.84–1.09) |
| Q3 | – | 1.30 (1.13–1.50) | 1.09 (0.95–1.25) |
| Q4 (lowest SES) | – | 1.40 (1.14–1.72) | 1.10 (0.90–1.34) |
Model 1 included age, sex, race/ethnicity, disease burden score, and HOUSES quartiles in Cox proportional hazard regression. This model was based on data from 13,649 individuals.
Model 2 included age, sex, race/ethnicity, disease burden score, and ADI quartiles in Cox proportional hazard regression with a random-effect term for census block groups. This model was based on data from 13918 individuals.
Model 3 included age, sex, race/ethnicity, disease burden score, HOUSES, and ADI quartiles in Cox proportional hazard regression with a random-effect term for census block groups. This model was based on data from 13277 individuals.
For continuous variables, the effect presented is for 1-unit change in the variable.
Lower SES measured by ADI was also associated with higher risk of NHP (eg, HR 1.40; 95% CI 1.14–1.72; model 2 in Table 2). Educational attainment, a secondary individual-level SES measure, was also associated with risk of NHP, but its association magnitude [HR 1.18 (95% CI 1.08–1.29) for comparing high school degree or less vs graduate/professional degree] was smaller than both the HOUSES index and ADI (Supplementary Table 2).
We observed no statistically significant interactive effects between SES based on the HOUSES index (individual-level SES measure) and ADI (neighborhood-level SES measure) on risk of NHP (interaction P value = .14). Adjusting for neighborhood characteristics made a very minor impact on association between the individual-level SES and risk of NHP [eg, HR 1.81 (95% CI 1.57–2.10)] for comparing Q1 vs Q4 when adjusting for ADI, compared to HR 1.89 (95% CI 1.66–2.15) without ADI adjustment; model 3 in Table 2).
Association Between Well-Established Risk Factors for NHP and HOUSES Index
We tested association of ADL (measuring the degree of troubles needed for daily activities) and living situation (living alone vs other living types) with HOUSES as they are two well-known risk factors for NHP. Patients with lower SES as measured by HOUSES index were more likely to live alone than those with higher SES [odds ratio 6.35; 95% CI 5.24–7.68 for comparing the lowest SES (Q1) vs the highest SES (Q4)] (Table 3). Adjusting for ADI, the association remained significant with slight attenuation of association magnitude. Similarly, individuals with lower SES were more likely to have troubles with at least 1 daily activity (odds ratio 2.38; 95% CI 2.00–2.84 for comparing HOUSES Q1 vs Q4) and ADI adjustment slightly attenuated the association.
Table 3.
Association of the HOUSES Index With Living Situation (Living Alone vs Others) and ADL, With and Without Adjusting for the ADI
| Living Situation | ||||
|---|---|---|---|---|
| HOUSES | Living Alone, n (%) | Others, n (%) | OR (95% CI), Model 1 (Without ADI Adjustment) |
OR (95% CI), Model 2 (With ADI Adjustment) |
| Q1 (lowest SES) | 938 (43.4%) | 1223 (56.6%) | 6.35 (5.24–7.68) | 5.75 (4.64–7.12) |
| Q2 | 717 (21.9%) | 2557 (78.1%) | 2.58 (2.14–3.12) | 2.46 (2.01–3.01) |
| Q3 | 362 (13.8%) | 2266 (86.2%) | 1.57 (1.29–1.92) | 1.50 (1.22–1.85) |
| Q4 (highest SES) | 157 (8.5%) | 1682 (91.5%) | REF | REF |
| ADL | ||||
| HOUSES | At Least One ADL Difficulty, n (%) |
No Difficulties With ADL, n (%) |
OR (95% CI), Model 1 (Without ADI Adjustment) |
OR (95% CI), Model 2 (With ADI Adjustment) |
| Q1 (lowest SES) | 714 (33.7%) | 1403 (66.3%) | 2.38 (2.00–2.84) | 2.17 (1.77–2.66) |
| Q2 | 664 (20.9%) | 2518 (79.1%) | 1.50 (1.26–1.77) | 1.40 (1.16–1.69) |
| Q3 | 396 (15.5%) | 2164 (84.5%) | 1.14 (0.95–1.37) | 1.08 (0.89–1.31) |
| Q4 (highest SES) | 221 (12.4%) | 1555 (87.6%) | REF | REF |
OR, odds ratio.
Model 1 included age, sex, and HOUSES quartiles in logistic regression. The number of individuals used was 9902 for analyzing living situation and 9635 for analyzing ADL, respectively.
Model 2 included age, sex, HOUSES quartiles, and ADI quartiles in logistic regression with a random-effect term for census block groups. The number of individuals used was 9639 for analyzing living situation and 9388 for analyzing ADL, respectively.
Discussion
We report that the HOUSES index, an individual-level SES measure reflecting current social and environmental circumstances, has additional predictive ability for future NHP among older adults beyond the impact of neighborhood-level SES of areas where they live. Our data also showed that HOUSES may reflect caregiver support and/or physical/cognitive functional status. Many older adults, even for those seriously ill, often desire to live in their own home, thus it is crucial to identify those at higher risk of NHP and then apply intervention strategies to mitigate the risk. This study suggests that additional intervention efforts at an individual-level may be beneficial to further mitigate the risk in addition to existing community-level interventions. Although solving challenges with SES is often difficult, many hospitals use care transition programs with health coaches or home visits to help older adults with higher risk.27,28
NHP involves 2 types of admission in terms of duration of nursing home stay (short-term vs long-term) with many patients receiving short-term nursing home stay after illness or procedure.29 These residents are expected to stay for a short period of time while they recover by using physical therapy and/or occupational therapy. Other patients are admitted to the nursing home for long-term placement and for residential use. Given that there is an increasing shift from residential long-term admission to short-term within health care setting,29 it is important to identify older adults with higher risk of NHP, and the role of SES on long-term placement in a nursing home has been documented. Literature shows that measures of lower SES are often predictive of long-term placement.30 Due to documented association between SES and risk of NHP, we expect SES at both neighborhood-level and individual-level will affect the risk of NHP. ADI is a commonly used neighborhood-level SES metric that measures neighborhood-level deprivation,22 and the HOUSES index is an individual-level SES measure that has been utilized successfully to demonstrate health disparities in many health outcomes.31-34 Our study shows that both measures are associated with risk of NHP, which is consistent with previous literature. We also observed that the association results were stronger with the HOUSES index compared to ADI, which may reflect heterogeneity in subject-level characteristics of older adults living in the same neighborhood.
More importantly, the association with the HOUSES index and NHP was independent of ADI. Therefore, knowing patient’s individual-level SES beyond characteristics of neighborhoods where they live is crucial to mitigate the risk of NHP. A plausible reason for strong association between the HOUSES index and NHP is that there is likely a high degree of heterogeneity in residents who live in the same neighborhood with respect to their social and environmental circumstances such as living arrangement (eg, caregiver support) and financial situations, which can be reflected through characteristics of their residence35 (thus measured by HOUSES index). The HOUSES index has been associated with aging-related outcomes such as hospitalization and multiple chronic conditions,36 which implies that the association with NHP may reflect functional decline from multiple health conditions and subsequent hospitalization. In our study, we observed that those with lower SES at an individual-level measured by the HOUSES index were more likely to live alone and were frailer (ie, higher degree of difficulty doing daily activities) independent of neighborhood-level SES. This observation may explain why individual-level SES plays a role in predicting NHP even after accounting for neighborhood SES.
This study has some considerable strengths. One major strength of this study is that we used an SES measure that is an individual-level not aggregated one (eg, ones based on census data), does not suffer from biases (eg, participation bias and recall bias) that are common for survey-based or self-reported SES (eg, self-reported income), and reflects current social and environmental circumstances compared with other measures that are more static over lifetime (eg, educational attainment level). Given that the HOUSES index requires address information and publicly available property data, reporting bias related to HOUSES will be minimal compared to other SES measures. Another strength is that this is a population-based study using patients from an integrated system of care in the hospital, community, and long-term care. The collection of both predictors and outcomes from EHR is comprehensive and near complete in Olmsted County residents where this study was conducted.
There are also limitations in this study. Similar to other studies using EHR data, it is possible that some outcomes of NHP were not captured if they were not seen by the provider team. However, the number of people that we missed would be relatively small because EHR system is near complete in this study setting, and thus, we do not believe this would have a biased effect. We also recognize that pertinent covariates relevant to NHP may not have been captured in EHR for some patients and its missingness may be affected by patient’s SES, partly because of health care access issues. Although this is a known issue for any study utilizing EHR data, not specific to our study, our data showed that missingness of the pertinent variables that were directly extracted from EHR was negligible. However, approximately 12% of the study cohort had at least 1 variable with missing information in this study, largely driven by missing SES measures (HOUSES and ADI) that required adequate residential address information in EHR. Individuals with the lowest SES by HOUSES were more likely to have incomplete data compared to those with higher SES (~5% for HOUSES Q1, ~2% for Q2 or Q3, and ~3% for Q4). Similar trend was observed for ADI with an exception that with the proportion of missingness for individuals with the highest SES by ADI was similar to those with the lowest SES (Supplementary Table 3). Lastly, our study is based on residents from Olmsted County, Minnesota, where individual characteristics (especially for older individuals) will be similar to those from other upper Midwest regions where the majority are white, and thus our study finding is generalizable to the upper Midwest. All outcomes of NHP can be geographic and regional in nature and may not generalize outside the United States.
Conclusions and Implications
We found that the HOUSES index as a measure for individual-level SES predicts NHP independent of neighborhood SES. We also found that 2 known risk factors for NHP (having difficulty doing daily activities and living alone) were strongly associated with the HOUSES index even after adjusting for neighborhood SES. This study provides further evidence of the importance of social determinants of health, especially at the individual-level, with NHP and a potential starting point for further research into using an individual-level SES measure such as the HOUSES index for clinical intervention models.
Supplementary Material
Acknowledgments
This study was funded by the Mayo Clinic Community Health: Assessment and Improvement Measures Program (CHAMP) research award.
This study used the resources of the Rochester Epidemiology Project (REP) medical records-linkage system, which is supported by the National Institute on Aging (NIA; AG 058738), by the Mayo Clinic Research Committee, and by fees paid annually by REP users. The content of this article is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health (NIH) or the Mayo Clinic.
Footnotes
The authors declare no conflicts of interest.
References
- 1.Mor V, Thomas KS, Rahman M. Defining and measuring nursing home placement. J Am Geriatr Soc. 2018;66:1866–1868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.HealthinAging.org. Nursing homes. Age-friendly healthcare & you web site. Accessed August 2, 2022. https://www.healthinaging.org/age-friendly-healthcare-you/care-settings/nursing-homes#:~:text=The%20chance%20of%20being%20admitted%20to%20a%20nursing,the%20older%20adult%20lacks%20a%20spouse%20or%20children [Google Scholar]
- 3.Mather M, Scommegna P, Kilduff L. Fact sheet: Aging in the United States. Population reference bureau. Accessed August 10, 2022. https://scorecard.prb.org/aging-unitedstates-fact-sheet/ [Google Scholar]
- 4.Bardenheier BH, Rahman M, Kosar C, Werner RM, Mor V. Successful discharge to community gap of FFS medicare beneficiaries with and without ADRD narrowed. J Am Geriatr Soc. 2021;69:972–978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kosar CM, White EM, Feifer RA, et al. COVID-19 mortality rates among nursing home residents declined from march to november 2020. Health Affair. 2021;40:655–663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.White EM, Yang XF, Blackman C, Feifer RA, Gravenstein S, Mor V. Incident SARS-CoV-2 infection among mRNA-vaccinated and unvaccinated nursing home residents. New Engl J Med. 2021;385:474–476. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bernardi F, Cozzani M, Zanasi F. Social inequality and the risk of living in a nursing home: implications for the COVID-19 pandemic. Genus. 2021;77:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Greiner MA, Qualls LG, Iwata I, et al. Predicting nursing home placement among home- and community-based services program participants. Am J Manag Care. 2014;20:e535–e536. [PubMed] [Google Scholar]
- 9.Valiyeva E, Russell LB, Miller JE, Safford MM. Lifestyle-related risk factors and risk of future nursing home admission. Arch Intern Med. 2006;166:985–990. [DOI] [PubMed] [Google Scholar]
- 10.Jacobs D, Fowler E, Fleisher L, Seshamani M. The medicare value-based care strategy: Alignment, growth, and equity. Health affairs forefront web site. Accessed July 21, 2022. https://www.healthaffairs.org/content/forefront/medicare-value-based-care-strategy-alignment-growth-and-equity [Google Scholar]
- 11.Schaefer MS, Hammer M, Platzbecker K, et al. What factors predict adverse discharge disposition in patients older than 60 years undergoing lower-extremity surgery? The adverse discharge in older patients after lower-extremity surgery (ADELES) risk score. Clin Orthop Relat Res. 2021;479:546–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gaugler JE, Duval S, Anderson KA, Kane RL. Predicting nursing home admission in the U.S: A meta-analysis. BMC Geriatr. 2007;7:13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Willink A, Davis K, Schoen C. Risks for nursing home placement and medicaid entry among older medicare beneficiaries with physical or cognitive impairment. Issue Brief (Public Policy Inst Am Assoc Retired Persons). 2016;37:1–14. [PubMed] [Google Scholar]
- 14.Hosking A, Hommel A, Lorenzl S, et al. Characteristics of patients with late-stage parkinsonism who are nursing home residents compared with those living at home. J Am Med Dir Assoc. 2021;22:440–445.e442. [DOI] [PubMed] [Google Scholar]
- 15.Wolff JL, Mulcahy J, Roth DL, et al. Long-term nursing home entry: A prognostic model for older adults with a family or unpaid caregiver. J Am Geriatr Soc. 2018; 66:1887–1894. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bonnet CJS, Laferrere A. Private financing of long term care: income, savings and reverse mortgates. Econ Stat/Economics and Statistics. 2019;507-508:5–24. [Google Scholar]
- 17.Angelelli J, Grabowski DC, Mor V. Effect of educational level and minority status on nursing home choice after hospital discharge. Am J Public Health. 2006;96:1249–1253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lord JDG, Ghiasi A, Weech-Maldonado R. Nursing homes located in socially deprived communities have been disproportionately affected by COVID-19. J Health Care Finance. 2021;48:1–12. [Google Scholar]
- 19.St Sauver JL, Grossardt BR, Finney Rutten LJ, et al. Rochester epidemiology project data exploration portal. Prev Chronic Dis. 2018;15:E42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.St Sauver JL, Grossardt BR, Leibson CL, Yawn BP, Melton LJ, Rocca WA. Generalizability of epidemiological findings and public health decisions: An illustration from the rochester epidemiology project. Mayo Clin Proc. 2012;87:151–160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.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;88:933–944. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Singh GK. Area deprivation and widening inequalities in US mortality, 1969-1998. Am J Public Health. 2003;93:1137–1143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kind AJH, Jencks S, Brock J, et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalization. Ann Intern Med. 2014;161:765–774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Goodman RA, Posner SF, Huang ES, Parekh AK, Koh HK. Defining and measuring chronic conditions: Imperatives for research, policy, program, and practice. Prev Chronic Dis. 2013;10:E66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Rocca WA, Grossardt BR, Boyd CM, Chamberlain AM, Bobo WV, St Sauver JL. Multimorbidity, ageing and mortality: Normative data and cohort study in an American population. BMJ Open. 2021;11:e042633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chamberlain AM, Rutten LJF, Jacobson DJ, et al. Multimorbidity, functional limitations, and outcomes: Interactions in a population-based cohort of older adults. J Comorb. 2019;9:2235042X19873486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Coleman EA, Parry C, Chalmers S, Min SJ. The care transitions intervention–results of a randomized controlled trial. Arch Intern Med. 2006;166:1822–1828. [DOI] [PubMed] [Google Scholar]
- 28.Naylor M, Brooten D, Jones R, Lavizzomourey R, Mezey M, Pauly M. Comprehensive discharge planning for the hospitalized elderly–a randomized clinical-trial. Ann Intern Med. 1994;120:999–1006. [DOI] [PubMed] [Google Scholar]
- 29.Tyler DA, Feng ZL, Leland NE, Gozalo P, Intrator O, Mor V. Trends in postacute care and staffing in US nursing homes, 2001-2010. J Am Med Dir Assoc. 2013;14:817–820. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Toot S, Swinson T, Devine M, Challis D, Orrell M. Causes of nursing home placement for older people with dementia: A systematic review and meta-analysis. Int Psychogeriatr. 2017;29:195–208. [DOI] [PubMed] [Google Scholar]
- 31.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;104:2383–2392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ryu E, Juhn YJ, Wheeler PH, et al. Individual housing-based socioeconomic status predicts risk of accidental falls among adults. Ann Epidemiol. 2017;27:415–420. [DOI] [PubMed] [Google Scholar]
- 33.Barwise A, Juhn YJ, Wi CI, et al. An individual housing-based socioeconomic status measure predicts advance care planning and nursing home utilization. Am J Hosp Palliat Me. 2019;36:362–369. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.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;16:78–83. [DOI] [PubMed] [Google Scholar]
- 35.Roy N, Dube R, Despres C, Freitas A, Legare F. Choosing between staying at home or moving: A systematic review of factors influencing housing decisions among frail older adults. PLoS One. 2018;13:e0189266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.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 Commun H. 2016;70:286–291. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.