Literature Review on Differences in Care Provided in Urban and Rural Nursing Homes in the United States

Abstract

Despite evidence acknowledging disadvantages in care provided to older adults in rural nursing homes (NHs) in the United States (U.S.), since 2010 no literature review has focused on differences in care provided in U.S. urban versus rural NHs. We examined these differences, by searching U.S. English-language peer-reviewed articles published after 2010 on differences in care quality in urban and rural NHs. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines and used Newcastle-Ottawa Scale for quality appraisal. We conducted full-text abstraction of 56 (of 286) articles, identifying ten relevant studies. Metric specification of urban/rural location varied and care quality measures were wide-ranging, making it hard to interpret the evidence. The limited evidence did support that rural NHs, compared to urban, provided sparse mental health support and limited access to hospice care after controlling for facility and resident characteristics. Our review highlights the need for more research examining urban and rural differences in the quality of NH care and raises several issues in current research examining urban/rural NH differences, where future work is needed.

Introduction

As people age in the United States (U.S.), their needs for care evolve. Long-term care facilities like nursing homes (NH), skilled nursing facilities (SNF), and assisted living facilities (ALF) provide a variety of services, both medical and personal care, to people who are unable to manage independently in the community. NHs offer extensive care outside of hospitals, offering help with custodial care -- like bathing, getting dressed, and eating – as well as skilled care (Medicare.gov, 2020), which provides continuous monitoring and medical assistance by licensed nurses for individuals with long-term healthcare needs. Over 4 million Americans are admitted to or reside in NHs (including SNF) each year and nearly one million adults reside in ALFs.

Given the current aging of the population and growth in various chronic conditions among older adults, the demand for NH services was expected to increase substantially from 2010 until now and disproportionately for care in rural NHs (Seeman et al., 2010). Currently, an estimated 60 million U.S. residents (19%) (Symens Smith & Trevelyan, 2019b) reside in rural areas. In the last decade, the population of U.S. older adults has grown 34.2%, which equates to about 53 million, and expected to exceed 88 million by 2050 (United States Census Bureau, 2011, 2020a). Since 2000, the rural older adult population has accounted for about 22% of the older adult population (Barnes, 1997; Coburn et al., 2002; Symens Smith & Trevelyan, 2019b). Rural counties as a whole have a higher share of older Americans, roughly 25% (Symens Smith & Trevelyan, 2019a), compared to urban or suburban areas (Parker et al., 2018; West et al., 2014).

Prior research, including the National Advisory Committee on Rural Health and Human Services’ 2010 report (Dalton & al., 2002; The National Advisory Committee on Rural Health and Human Services, 2010) and the National Healthcare Disparities Report (Agency for Healthcare Research and Quality, 2010) released by the Agency for Healthcare Research and Quality (AHRQ) in 2010, documented differences in quality of rural and urban care for the elderly specific to NHs; the authors of the report used the Institute of Medicine’s (IOM) quality framework with 5 components of quality care (effectiveness, safety, timeliness, patient/family centeredness, access, efficiency) (Institute of Medicine, 2010). These past national trends pointed to a lack of home and community-based options and fewer rural NHs, but also a 35 percent higher ratio of NH beds in rural areas (Nelson & Stover Gingerich, 2010). This was coupled with limited access to services within rural NHs. Compared with residents in urban facilities and controlling for demographic and health status measures, residents in rural NHs were less likely to receive preventive services and visited health care providers less often (Agency for Healthcare Research and Quality, 2010; Larson & Fleishman, 2003). Rural NHs were also less likely to provide mental health services (Li, 2010).

Other past evidence demonstrates differences in the quality of care provided in urban and rural NHs, but the results are mixed. Among post-acute admissions, the risks for death and mood problems were higher in NHs in large rural towns (i.e., population of 10,000–49,999), relative to urban areas (i.e., population >50,000) (Phillips et al., 2004). (Coburn et al., 2003)Other researchers reported higher likelihood of multiple hospital admissions (Coburn et al., 2002) and higher rates of feeding tube use near the end of life in urban NHs compared to rural NHs. (Gessert & Calkins, 2001). This mixed evidence identifies the need to understand the current state of the science.

Rural NHs care for a higher proportion of older adults with long-term care needs (Fennell & Campbell, 2007; Phillips et al., 2004) who are characteristically older, in a lower socioeconomic status, have poorer health, and suffer from more chronic conditions (Agency for Healthcare Research and Quality, 2010). For example, higher rates of NH use among rural older adults aged 75 years and older are found compared to urban adults of the same age (12% vs. 8.2%) (Phillips et al., 2003). Besides caring for a population of older adults with poorer health and more chronic conditions, rural NHs face additional challenges related to the rurality of their facility location. Compared to urban NHs, rural facilities were smaller, had lower nursing staff ratios, and were less likely to offer specialized services, such as Alzheimer’s units (Phillips et al., 2004). Similarly, Caffrey (2005) reported rural NHs offered fewer services, had fewer health personnel and fewer nurses held a bachelor of science in nursing (Caffrey, 2005). However, rural NHs had less nursing turnover. For residents receiving posthospital care, another study of NHs found staff in rural NHs were less likely to document and plan for these residents to be discharged compared to staff in urban NHs (Barnes, 1997; Coburn et al., 2002; Parker et al., 2018; Seeman et al., 2010; Symens Smith & Trevelyan, 2019a, 2019b; United States Census Bureau, 2011, 2020a; West et al., 2014).

Despite the evidence acknowledging disadvantages in care provided for older adults in rural NHs, since 2010, no literature review has focused specifically on the differences in care provided in urban vs rural NHs. We sought to examine and update current evidence on differences in care provided in urban and rural NHs.

Methods

We reviewed articles on differences in care provided in urban and rural U.S. NHs. We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009; Moher et al., 2009).

Search strategy.

We used a structured search strategy in PubMed and the Cumulative Index of Nursing and Allied Health Literature (CINAHL) to identify U.S. English-language peer-reviewed articles published from January 2010 to March 2021 on care provided in NHs and compared care in urban and rural NHs. Table 1 lists the search strategy. Briefly, we identified articles using Medical subject headings (MeSH), keywords and Boolean operators with at least (1) one NH keyword, (2) one urban/rural keyword, and (3) one quality of care keyword e.g., quality of care and care experiences. We identified 276 articles.

Table 1.

Search Strategy in PubMed and Cumulative Index of Nursing and Allied Health Literature (CINHAL)

Terms Included	Key Words Used per Term
Nursing Home	“nursing homes”[MeSH Terms] OR (“nursing home”[Title/Abstract] OR “nursing homes”[Title/Abstract]) OR (“skilled nursing facility”[Title/Abstract] OR “skilled nursing facilities”[Title/Abstract])
AND
Rural/Urban	“rural population”[MeSH Terms] OR “rural”[Title/Abstract] OR “urban population”[MeSH Terms]) OR “urban”[Title/Abstract] OR “urbanicity”[Title/Abstract] OR “rurality”[Title/Abstract]
AND
Quality of care	((“shortage”[Title/Abstract] OR “shortages”[Title/Abstract]) AND ((“service”[Title/Abstract] OR “services”[Title/Abstract]) OR (“agency”[Title/Abstract] OR “agencies”[Title/Abstract]))) OR ((“lack”[Title/Abstract] OR “lacking”[Title/Abstract]) AND “capacity”[Title/Abstract]) OR “nursing homes/supply and distribution”[MeSH Terms] OR “quality of health care”[MeSH Terms] OR “quality”[Title/Abstract] OR “experience”[Title/Abstract] OR (“care experience”[Title/Abstract] OR “care experiences”[Title/Abstract])
Filters:	English Adult: 19+ years January 2010 – April 2020 Excluded: Commentaries/editorials

NOTE: MeSH = Medical subject headings.

Article Screening.

We reviewed titles and abstracts of identified articles. After an initial period of double coding to establish consistency across reviewers (DQ, LE), remaining abstracts were independently reviewed to determine eligibility. Reviewers discussed discrepancies during regular meetings and resolved disagreement to reach consensus on inclusion. Included abstracts were reviewed by both authors. Both reviewers confirmed inclusion of articles at title and abstract stage.

As shown in Figure 1, articles were initially excluded if the title and abstract indicated the research: did not study care provided in NHs (N=26); included only urban or only rural NH care (N=99); were not U.S. (N=102); or were not empirical studies (e.g., commentaries; N=3).

Figure 1.

PRISMA Flow Diagram

Abstraction.

We undertook full review of 56 articles. Two researchers abstracted specific information into a form developed a priori: study aim; time frame; type of study; study design; statistical approach; variables controlled for in analysis; type of care; care settings; sample size; sample description; study population descriptions; main outcome(s) measured; secondary outcome(s) (if any); description of urban/rural variable; description of NH characteristic variables; relevant rural vs urban findings; relevant NH findings; issues reported as possibly affecting NH care; issues reported related to urbanicity/rurality; study limitations related to urbanicity/rurality or NH care; policy and regulatory environment; facility characteristics and capabilities; and provider characteristics. To ensure both reviewers employed a similar approach, articles (N=7) were selected for double-review and discussion. Each reviewer then abstracted half the remaining articles. After abstraction, each article was reviewed by the other reviewer to ensure accuracy of abstracted content and discussed if needed to gain consensus.

Risk of bias, Quality Assessment, and Data Synthesis.

During full-text review, we assessed study quality of the included articles using the Newcastle-Ottawa Scale (NOS). The NOS is an 8-item checklist that examines cohort and case-control studies (nonrandomized studies) in three areas awarding stars for ‘high study quality’ characteristics: 1) selection (4 items each worth 1 star); 2) comparability (1 item worth maximum 2 stars); and 3) assessment of outcomes/ascertainment of exposure (3 items each worth 1 star). Scores range from 0–9 and with AHRQ-developed thresholds translate into an overall rating of good, fair or poor study quality. Good study quality includes three to four stars for selection, one to two stars for comparability, and two to three stars for outcome/exposure. Fair study quality includes two stars for selection, one to two stars for comparability, and two or three stars for outcome/exposure. Poor study quality includes zero to one star for selection, zero stars for comparability, or zero to one star for outcome/exposure. Two researchers (DQ and LE) independently scored each article and results discussed to gain consensus. Supplemental Table 1 reports study quality scores. Four articles were excluded as “poor”. Seven were rated as “good” and three articles as “fair”.

We excluded 46 studies after reading the full article because they: did not specifically examine care provided in a NH (N=36); included only urban or only rural care (N=6); or were rated as “poor” study quality (N=4). Altogether, ten studies were included (Bowblis et al., 2013; Clement et al., 2018; Hutt et al., 2011; Kang-Yi et al., 2011; Kang et al., 2011; Li, 2010; Lutfiyya et al., 2013; Niznik et al., 2019; Pu et al., 2016; Temkin-Greener et al., 2012).

Results

Of the ten studies, four were cohort studies (Bowblis et al., 2013; Clement et al., 2018; Hutt et al., 2011; Li, 2010) and the remaining six were cross-sectional studies (Kang-Yi et al., 2011; Kang et al., 2011; Lutfiyya et al., 2013; Niznik et al., 2019; Pu et al., 2016; Temkin-Greener et al., 2012). All studies were facility-based (Hutt et al., 2011; Kang et al., 2011; Li, 2010; Lutfiyya et al., 2013; Pu et al., 2016; Temkin-Greener et al., 2012) and in a single setting, with eight in NHs (Bowblis et al., 2013; Hutt et al., 2011; Kang-Yi et al., 2011; Kang et al., 2011; Li, 2010; Lutfiyya et al., 2013; Niznik et al., 2019; Temkin-Greener et al., 2012) and two in non-hospital-based SNFs (Clement et al., 2018; Pu et al., 2016). We describe the methods, population and measures in Table 2 for cohort studies and in Table 3 for cross-sectional studies. Table 4 summarizes the urban/rural results.

Table 2.

Methods, Population and Measures for Included Cohort Studies

Design & Statistical Approach	Population Characteristics	Urban/Rural Measures*	Nursing Home (NH) Quality Measures
Bowblis 2013
Regression and Blinder-Oaxaca decomposition techniques to attribute differences to facility and aggregate resident characteristics (ie, NH (NH) structure, facility characteristics, special care units, case-mix, and staffing levels), controlling for year, state fixed effects and clustering of facilitates within a state. P-value < 0.01	152,356 Online Survey Certification and Reporting (OSCAR) surveys from 17,953 unique NHs for the 48 contiguous states from January 1, 1999 to December 31, 2008 11,677 (65%) are urban, 2,576 (14%) micropolitan, 2,068 (12%) small rural town, and 1,662 (9%) isolated small rural town	Rural-Urban Commuting Area (RUCA) designation: Urban/large rural town/small rural town/isolated rural town	Proportion of NH residents with facility-acquired contractures (i.e., facility-based contracture rate) A contracture is an abnormal muscle shortening and joint fixation commonly seen among persons with immobility or central nervous system disorders
Clement 2018
Multiple linear regression to compare skilled nursing facility (SNF) means by urban/rural categories and used Wald differences test among coefficients for rurality variables. Controlled for community- and SNF-level resources P-value < 0.05	12,261 Medicare and Medicaid-participating non-hospital based SNFs Sample sizes by urban-rural codes not presented	Rural-Urban Commuting Area (RUCA) designation: Urban/ Suburban/ large rural town/rural area (based on 2010 Census)	Percent of all-cause unplanned 30-day risk adjusted rehospital-ization rates and admission to SNF averaged over Q3/Q4 2014 and Q1/Q2 2015 Center for Medicare and Medicaid Services (CMS) NH Compare data
Hutt 2011
Unadjusted odds ratios Logistic regression controlling for resident and facility characteristics and interaction between region and urban/rural P-value < 0.05	Using a 10% national random sample of NHs (n =1,840) 1999 to 2005, data for NH residents with heart failure (HF) for 2003–2004 were included 500,322 unique NH residents of which 187,408 (37.4%) had Heart Failure. Most were elderly women with multiple comorbidities, fair cognition and function; ~11% African American, 1% Asian, and 3% Hispanic	Facility characteristics included urban versus rural NH location and U.S. region (South, Northeast, West, or Midwest)	Mortality and subsequent first heart failure (HF) hospitalization and any HF hospitalization at 1 year CMS database called the Linked NH/SNF Stay File
Li 2010
Longitudinal trend of on-site mental health service provision in NHs Multivariate regression determined facility and geographic correlates in 2004 P-value < 0.05	1,409 NH in 1995, 1,406 in 1997, 1,423 in 1999, and 1,174 in 2004 Between 1995 and 2004, approximately 50–60% of NHs were chain affiliated, over 30% were nonprofit, and over 60% were located in a metropolitan statistical area (MSA)	NH located in a MSA	On-site mental health service provision National NH Surveys between 1995 and 2004

Table 3.

Methods, Population and Measures for Included Cross-sectional Studies

Design & Statistical Approach	Population Characteristics	Urban/Rural Measures*	Nursing Home (NH) Quality Measures
Kang 2011
Multilevel nonlinear mixed models accounting for the clustering of individual residents within the same facility were used to compare the four quality measures among residents in the three geographic areas, controlling for the individual and facility characteristics P-value < 0.05	12,507 residents aged 50 years or older in 1,174 NH 6,959 (56%) of the residents were in NH located in metropolitan areas, 2,851 (23%) in micropolitan areas, and 2,697 (22%) in rural areas	Office of Management and Budget definition of metropolitan, micro-politan, and rural areas Each metropolitan area and micropolitan area must have at least one urbanized area of 50,000 or more population and 10,000 to 50,000 population, respectively. A rural area has 10,000 or less population	Hospital admission in past 90 days after a NH admission, experience of moderate to severe pain in past 7 days, resident’s documented status of ‘Ever had a pneumococcal vaccination’ or ‘Ever had a flu shot in the past 12 months 2004 National NH Survey
Kang-Yi 2011
Multivariate linear regression models accounting for NH resident characteristics and NH characteristics (including location) used to examine relationships between Medicaid census and other important nursing home factors (including urban/rural location) and its impact on psychosocial well-being of NH residents P-value < 0.05	90,001 residents aged 65 years and older from 565 NHs in rural and urban areas of New York State, excluding NHs with bed sizes less than 25.	Urban indicator for NH location (No detail on how it is determined.)	Psychosocial well-being defined with three Minimum Data Set (MDS)-based measures: percentage of NH residents with low social engagement (i.e. 0–2 activities of 6 social activities), percentage of residents with depressive symptoms (i.e., prevalence of one or more of 16 depressive symptoms), and percentage of residents with interpersonal relationship problems (i.e., prevalence of one or more of 6 personal relationship issues). Combined New York NHs’ OSCAR and MDS data
Lutfiyya 2013
Independent-sample t-tests to compare the mean rural/urban NH quality ratings Linear mixed model binary logistic regression using covariates of NH ownership, number of beds, and geographic locale of NH, controlling for the random effects of state The measures were adjusted by case-mix based on the distribution of Minimum Data Set (MDS) assessments by resource utilization groups, version III (RUG-III)	15,695 Medicaid and Medicare NH included on the NH Compare Web site in 2010 15,177 mapped to an urban-rural code 69.2% NH were located in urban areas and 30.8% in rural areas In 16 states, at least half (50%) of NH were in rural areas and in 7 states more than 70% of NH were in rural counties	RUCA designation: urban (i.e., city with a population >50,000 and its commuting area), large rural town (i.e., population of 10,000–49,999 and its commuting area), small rural town (i.e., population of 2,500–9,999 with some commuting to an urban cluster), and isolated rural (i.e., fewer than 2,500 residents, primarily commuting to a tract outside an urban area or cluster)	NH-level overall 5-star quality rating, health inspections rating, composite of 10 facility performance quality measures, and a staffing measure 5-star quality ratings were from CMS 2010 NH Compare data Health inspection data were collected during state health inspections. Ratings depended on the number, scope, and severity of identified deficiencies reported during the 3 most recent annual inspections. Staff ratings included both registered nurse (RN) hours per resident day and total staffing (RN, licensed practical nurse, nurse aide hours) per resident day Staffing derived from CMS OSCAR data
Niznik 2019
Cox-proportional hazards models accounting for the clustering of individual residents within the same facility with time-varying covariates were used to identify patient-, provider- and system-level factors associated with acetylcholinesterase inhibitors (AChEIs)  discontinuation P-value < 0.001	37,106 non-skilled NH residents aged 65 and older with severe dementia receiving AChEIs within the first 14 days of an MDS assessment in a Medicare-certified NH	NH Compare data provided zip codes used to define rurality (urban, rural, highly rural) by linking to rural-urban continuum codes in the Area Health Resource File	Rurality was associated with decreased likelihood of discontinuation of AChEIs for NH residents aged 65 years and older with severe dementia (i.e., continuation of AChEIs). Rural NHs (adjusted hazards ratio (aHR) = 0.82; 95%CI=0.76–0.87) or highly rural NHs (aHR = 0.77; 95% CI=0.66–0.89) were associated with decreased likelihood of discontinuation (i.e., lower quality), compared to urban NHs 2015–2016 Medicare claims, Part D prescriptions, MDS version 3.0, Area Health Resource File, and NH Compare data
Pu 2016
Ordinary least square models controlled for overall rating, geographic locale, ownership type, and certificate type P-value < 0.001	15,639 SNFs; with 10,107 in urban areas (64.6%), while 5532 were considered rural	2010 Census Urban and Rural Classification and Urban Area Criteria for U.S. counties by state identify Urbanized Areas (UAs) as 50,000 or more people, used to categorize SNFs. “Rural” included population, housing and territories not included within an urban area	Overall quality 5-star rating, influenza vaccination or pneumococcal vaccination 2013 CMS NH Compare data
Temkin-Greener 2012
Retrospective Separate risk adjustment models for each outcome: pain, in-hospital death, hospice. Logistic regression models were fit at the individual resident level with random facility effects to account for resident clustering at the facility level. Models predict, for each resident, the probability of each outcome conditional on the individual risk factors. Each model included urban–rural status, facility-specific factors, and environmental characteristics P-value < 0.05	963,313 Medicare eligible, aged 65+, decedent long-term NH residents who died between Jan 1, 2005 and Dec 31, 2007, who resided in a Medicare and/or Medicaid certified facility (n = 15,954). Long-term residents were those whose stay was not Medicare reimbursable or who stayed longer than 90 days 915,688 decedent long-term residents in 13,206 NH facilities. 8,519 urban NH (65%), 1,938 large rural NH (15%), 1,567 small rural NH (12%) and 1,182 isolated rural NHs (9%)	RUCA designation: urban (i.e., city with a population >50,000 and its commuting area), large rural town (i.e., population of 10,000–49,999 and its commuting area), small rural town (i.e., population of 2,500–9,999 with some people commuting to an urban cluster), and isolated rural (i.e., fewer than 2,500 residents, primarily commuting to a tract outside an urban area or cluster)	Place of death was defined as dichotomous (1 if death occurred in hospital, 0 otherwise). Use of hospice (1 if decedent used NH hospice within last 100 days of life,0 otherwise). Pain (1 if resident experienced moderate pain daily or excruciating pain at any frequency, based on the last MDS assessment prior to death; 0 otherwise) Used CY2005–2007 100% Minimum Data Set, Medicare beneficiary file, and inpatient and hospice claims

Table 4.

Relevant Results from the Included Studies

Study	Urban/Rural Differences
Bowblis 2013	The more isolated the NH, the higher the contracture rates (i.e., lower quality) and the greater the increase in rates over the study period. Urban NHs had contracture rates of 7% in 1999 and 10% in 2008, corresponding to a 2-percentage point increase. NHs in isolated rural towns had an increase of 3 percentage points (11%–14%; unadjusted results). Compared to urban NH, micropolitan, small rural towns, and isolated rural towns had 1, 3, and 3 percentage point higher facility acquired contracture rates
Clement 2018	After controlling for community- and SNF-resources, risk-adjusted rehospitalization rates for SNFs were lowest in rural areas and large rural towns, followed by SNFs in suburban and urban areas. Mean average risk-adjusted rehospitalization rates were lowest in rural areas (19.8%) and significantly higher for SNFs in all other locations. After controlling for community and organizational variables, incl. extensive risk adjustment, urban SNFs re-hospitalize an additional 1.2% of their admissions from hospitals within 30 days compared to rural SNFs. Rural SNFs face less competition from other SNFs and have a higher proportion of elderly and poor adults but with lower Medicare Advantage enrollment. The number of hospital beds per 1,000 population is significantly lower in rural vs urban areas and large rural towns, although it is higher than in suburban SNFs. Rural SNFs are smaller and less reliant upon hospitals for admissions and Medicare reimbursement.
Hutt 2011	NH residents with heart failure (HF) were hospitalized were more likely to reside in larger, rural NHs. Residing in the South was associated with nearly one-third less the odds of dying compared with residing in other U.S, regions for the rural NHs, whereas the regional disparity became smaller for urban NHs.
Kang 2011	After controlling for individual- and facility-characteristics, residents in rural NHs were more likely to experience hospitalization (odds ratio [OR] = 1.50, 95% confidence interval [CI] = 1.16–1.94) and moderate to severe pain (OR = 1.68, 95% CI = 1.35–2.09). Individuals in micropolitan NHs were more likely to experience moderate to severe pain (OR = 1.67, 95% CI = 1.37–2.04). No differences were found for influenza and pneumococcal vaccination rates across residents in urban and rural NHs in multivariate analysis. There were differences in the non-adjusted analyses with rural NHs having higher vaccination rates (65.8% versus 76.0% influenza; 54.5% versus 58.8% pneumococcal, both p values <0.001).
Kang-Yi 2011	No statistically significant urban versus rural difference. Urban and rural NHs in the state of New York had similar levels of resident depressive symptoms, social engagement, and interpersonal relationship problems.
Li 2010	Facilities in a metropolitan statistical area (MSA) were more likely than their counterparts to provide on-site mental health services. Compared to non-MSA facilities, the increased rate of providing on-site mental health services for MSA facilities was 11% in 1995 (78% versus 67%), 7% in 1997 (84% versus 77%), 8% in 1999 (83% versus 75%), and 21% in 2004 (84% versus 63%). The enlarged difference in 2004 resulted mainly from the reduced proportion of non-MSA facilities providing mental health services from 1999 to 2004. The deficit in mental health care access is more pronounced in smaller or rural NHs. For overall availability (yes/no) of mental health services, the odds ratio was 2.60 (P = .000) for MSA vs non-MSA facilities. In the analyses of available mental health services, bed size, Medicare and Medicaid populations, census region, and MSA status were found to be significant predicators.
Lutfiyya 2013	After controlling for state effects and adjusting for size and ownership, rural NHs were less likely to have a 4 or higher rating compared to urban NHs (odds ratio = .901, 95% CI=0.824–0.986). Unadjusted independent sample t tests comparing the overall rating, health inspection rating, staffing measure, and quality ratings of rural vs urban NHs yielded statistically significant differences for the overall rating and health inspections. Rural NHs had higher overall ratings and better health inspection ratings. In both locales, one-fourth of NHs were owned by not-for-profit entities. Rural NHs tended to be smaller, with a higher proportion of rural NH having fewer than 100 beds and a greater proportion of urban NHs having more than 100 beds. NHs with an overall quality rating of 4 stars or higher were more likely to be owned by a not-for-profit entity rather than a government agency and have fewer than 100 beds.
Niznik 2019	Rurality was associated with decreased likelihood of discontinuation (i.e., continuation) of acetylcholinesterase inhibitors (AChEIs) for NH residents aged 65 years and older with severe dementia who were receiving AChEIs within the first 14 days of an MDS assessment (n = 37,106). Specifically, rural facilities (aHR = 0.82; 95%CI = 0.76–0.87) or highly rural facilities (aHR = 0.77; 95% CI = 0.66–0.89), compared to an urban facility, was associated with decreased likelihood of discontinuation (i.e., lower quality).
Pu 2016	Overall quality ratings were associated with increased vaccination rates. Urban SNFs had lower rates of influenza and pneumococcal vaccination compared to rural SNFs ( P < 0.001). Mean overall five-star ratings were significantly different at 3.5 for urban SNFs and 3.4 for rural SNFs (P < 0.001). Each star increase represented a 25 % increase with the lowest facility rating=1. The difference in these means (0.11) was equivalent to a 2.75 percentage point increase, which is a meaningful difference. Over 20% of rural SNFs had five stars compared to ~30 % of urban SNFs, suggesting urban SNFs received higher overall ratings on average. Urban locale was statistically significant and negative for influenza (−2.25) and for pneumococcal vaccination (−1.45), suggesting that for urban SNFs the rate of influenza vaccination was 2.25 percentage points lower and 1.45 percentage points lower for pneumococcal vaccinations.
Temkin-Greener 2012	NHs in more rural areas had significantly worse hospice use compared with urban NHs, even after adjusting for covariates and state fixed effects. Hospice use increased from 19% in isolated rural NHs to 22% in small rural town NHs and to 24% and 37% in large rural town and urban NHs. In-hospital deaths was highest in small rural town NHs (20%) and lowest in urban NHs (17%). Severe pain was lowest (13%) in urban NHs compared to isolated rural NHs (14%) and both small and large rural town NHs (15%). No statistically significant differences in pain experienced by residents in rural and urban NHs were found. In-hospital death results suggest rural NHs are more likely to hospitalize residents prior to death compared to urban NHs, even after controlling for facility and resident characteristics, including the presence of do-not-hospitalize orders. Urban NHs had more nursing staff and more highly skilled staff (i.e., higher RN ratio) compared with rural NHs. Compared to NHs in isolated rural areas, urban NHs were more likely to be for-profit (73% vs. 54%), in a chain (56% vs. 48%), larger (111 residents vs. 61), and have fewer Medicaid residents (61% vs. 64%). The number of hospice providers (10 vs. 0.6) and the availability of hospital beds (5 vs. 3) differed significantly in urban NHs compared to isolated rural NHs. Statistically significant higher quality of end of life hospice use was found in urban NHs (average = 0.12; SD = 0.21) compared to rural NHs which had increasingly poorer quality (e.g., isolated rural NHs = −0.06; SD = 0.21, P < .001).

Data used ranged from 1995 to 2016 (using 1995–2004 National NH Survey data and 2015–2016 Centers for Medicare and Medicaid Services NH Compare (CMS-NHC) data linked to Medicare claims, Part D prescriptions, Minimum Data Set (MDS) version 3.0 data and the Area Health Resource File). Three studies used CMS-NHC; one using 2010 data (Lutfiyya et al., 2013), another 2013 data (Pu et al., 2016), and a third 2014–2015 data (Clement et al., 2018). Two studies used the Centers for Disease Control and Prevention’s (CDC) National NH Surveys data; one using 1995–2004 data (Li, 2010) and one using 2004 data (Kang et al., 2011). Three studies used: CMS’s Online Survey and Certification Report System (OSCAR) (Bowblis et al., 2013; Lutfiyya et al., 2013) data, CMS’s Linked NH/SNF Stay File (Hutt et al., 2011), and Minimum Data Set (MDS) Medicare beneficiary file and inpatient and hospice claims data (Temkin-Greener et al., 2012). One article combined OSCAR and MDS (Kang-Yi et al., 2011). The one remaining study used combined CMS-NHC data with Part D prescriptions, MDS data version 3.0 and Area Health Resource File (Niznik et al., 2019).

Studies used different classifications to determine location of urban and rural NHs, with most studies defining multiple categories for rural locations. Three studies used the 4-category Rural-Urban Commuting Area (RUCA) Codes (Bowblis et al., 2013; Clement et al., 2018; Temkin-Greener et al., 2012) (United States Department of Agriculture, 2019) to identify the urban/rural continuum (based on census tract and zip code): 1) urban, 2) large rural town, 3) small rural town, and 4) isolated rural town. One study used the Office of Management and Budget’s (OMB) 3-category rural variable (based on county) (Kang et al., 2011): 1) metropolitan (population > 50,000), 2) micropolitan (10,000–50,000), and 3) rural (<10,000). Another study used CMS-NHC zip code data linked to the rural-urban continuum codes in the AHRF to define rurality (urban, rural, highly rural) (Niznik et al., 2019). Four studies used a binary urban/rural designation based on facility location (Hutt et al., 2011; Li, 2010; Lutfiyya et al., 2013; Pu et al., 2016). Two (Hutt et al., 2011; Li, 2010) of these four indicated whether a NH was in a metropolitan statistical area (MSA)(United States Census Bureau, 2020b) and a third study (Lutfiyya et al., 2013) used the 2003 urban-rural county continuum codes developed by the Economic Research Service of the U.S. Department of Agriculture (United States Department of Agriculture, 2019) (sometimes called the Beale codes; based on county employing OMB metropolitan status, county urban population, and commuting). The fourth (Pu et al., 2016) study used 2010 Census Bureau Urban and Rural Classification and Urban Area Criteria for counties that identifies urban (population > 50,000), based on census tract population and density (United States Census Bureau, 2019). Two studies used a binary urban/rural indicator combined with 4-category U.S. region codes (South, Northeast, Midwest, West) based on Census Bureau regions (Hutt et al., 2011; Li, 2010). One study used an urban/rural indicator, not describing its source (Kang-Yi et al, 2011).

Most (n=6) employed multivariate regression analysis of outcomes and urban/rural location (Bowblis et al., 2013; Clement et al., 2018; Kang-Yi et al., 2011; Kang et al., 2011; Li, 2010; Pu et al., 2016) and three used logistic regression (Hutt et al., 2011; Lutfiyya et al., 2013; Temkin-Greener et al., 2012), while one employed cox-proportional hazards models accounting for the clustering of individual residents within the same facility with time-varying covariates (Niznik et al., 2019). Six studies controlled for both resident and facility-level characteristics (Bowblis et al., 2013; Hutt et al., 2011; Kang-Yi et al., 2011; Kang et al., 2011; Niznik et al., 2019; Temkin-Greener et al., 2012) and the remaining four studies controlled for only facility-level characteristics (Clement et al., 2018; Li, 2010; Lutfiyya et al., 2013; Pu et al., 2016). Four studies employed risk-adjustments, six did not (Bowblis et al., 2013; Clement et al., 2018; Kang-Yi et al., 2011; Kang et al., 2011; Niznik et al., 2019; Temkin-Greener et al., 2012). Significance was determined using P-value <0.05 or less. Study samples ranged from 565 facilities to 17, 953 facilities and were roughly 65% urban.

Quality Metrics

Table 5 lists the quality of care metrics for NHs identified in our review and indicates how high-quality is defined. Four outcome measures were similar across studies: 1) experience of moderate-to-severe pain in past 7 days (n=2), 2) resident’s documented status of ever having had an influenza vaccination (n=2), 3) ever having had pneumococcal vaccination (n=2), and 4) CMS overall quality 5-star rating (n=2). The remaining outcome measures assessed were wide-ranging: proportion of NH residents with facility-acquired contractures (facility-based contracture rate) (Bowblis et al., 2013); percent of all-cause unplanned 30-day risk adjusted NH admission and rehospitalization rates (Clement et al., 2018); mortality and subsequent hospitalization for heart failure (HF) [i.e., following the first heart failure hospitalization was there any HF hospitalization after 1 year] (Hutt et al., 2011); on-site mental health service provision (Li, 2010); psychosocial well-being comprised of fewer depressive symptoms, more social engagement and fewer interpersonal relationship problems (Kang-Yi et al , 2011), hospital admission in past 90 days after NH admission (Kang et al., 2011); health inspection rating (i.e., number, scope, and severity of identified deficiencies) (Lutfiyya et al., 2013) and staffing ratings (i.e., registered nurse (RN) hours per resident day and total staffing that includes RN, licensed practical nurse, and nurse aide hours per resident day (Lutfiyya et al., 2013); discontinuation of acetylcholinesterase inhibitors (AChEIs) for NH residents age 65+ years with severe dementia (Niznik et al, 2019) and place of death and use of hospice (Temkin-Greener et al., 2012).

Table 5.

Identified Quality of Care Metrics for Nursing Homes

High Quality is defined by:	Aspect of Quality measured
High CMS 5-star quality rating	Global measure
Low rates of hospitalizations*	Global measure
Low rehospitalization rates*	Global measure
Low prevalence of in-hospital deaths*	Patient centeredness
Low contracture rates*	Patient centeredness
Psychosocial wellbeing defined by fewer depressive symptoms, more social engagement and fewer interpersonal relationship problems	Patient centeredness
Less residents experiencing moderate-to-severe pain*	Patient centeredness
Less prevalence of severe pain	Patient centeredness
More likely to have on-site mental health	Access
More prevalence of hospice use*	Access
More nursing staff	Timeliness
More highly skilled staff (i.e., higher RN ratio)	Safety
High rates for influenza vaccination	Safety
High rates for pneumococcal vaccination	Safety
Discontinuation of acetylcholinesterase inhibitors (AChEIs) for nursing home residents age 65+ years with severe dementia	Safety

NOTE:

^*Metrics are generally risk-adjusted.

The majority of included articles (n=9) consistently found that after controlling for individual and facility-level covariates, rural NHs have poorer access to care and provide lower quality of care. Specifically, after controlling for resident and facility-level covariates, rural NHs are less likely to have a 4 or higher CMS 5-star quality rating (Lutfiyya et al., 2013) (i.e., lower quality) when compared with urban NHs. For example, in Lutfiyya (2013) authors found that the mean overall five-star ratings were significantly different for urban NHs with a mean 3.5 rating compared to rural NHs with a mean 3.4 (P < 0.001) (Lutfiyya et al., 2013). Each star increase represented a 25% increase because the lowest rating a facility could receive is 1. Accordingly, the difference in these means (0.1) was equivalent to a 2.75 percentage point increase, which may not be a substantively meaningful difference in quality.

Furthermore, after controlling for NH and resident characteristics, NHs differed by location. The more rural the NH, the higher the NH risk-adjusted contracture rates (2 percentage points) (i.e., lower quality) and the greater the increase in rates overtime (Bowblis et al., 2013). Also the more rural, the more likely NH residents were hospitalized (i.e., lower quality; OR = 1.50, 95% CI = 1.16–1.94; risk-adjusted) (Kang et al., 2011) and the more likely NH residents with HF were hospitalized (Hutt et al., 2011) (i.e., lower quality). Rural NHs had higher prevalence of risk-adjusted in-hospital deaths (Temkin-Greener et al., 2012) (i.e., lower quality). For example, the highest prevalence of risk-adjusted in-hospital deaths were in small rural town facilities (20%) and lowest in urban areas (17%). Rurality was associated with decreased likelihood of discontinuation of AChEIs for NH residents aged 65 years and older with severe dementia (i.e., continuation of AChEIs) (Niznik et al., 2019). Compared to urban facilities, both rural and highly rural facilities were associated with decreased likelihood of discontinuation of AChEIs for NH residents aged 65 years and older with severe dementia (i.e., lower quality), as stopping AChEIs for those with severe dementia does not lead to increased risks but their continuation has other negative side effects (including polypharmacy) (Niznik et al., 2020). Also, the more rural, the less likely the NH had on-site mental health services (Li, 2010) (i.e., lower quality), and the lower the prevalence of hospice use (i.e., lower quality). For example, in Temkin-Greener (2012) risk-adjusted hospice use increased from 19% in isolated rural facilities to 22% in small rural town facilities and to 24% and 37% in large rural town and urban areas, respectively (Temkin-Greener et al., 2012). They also found that rural NHs were found to have less nursing staff, less highly skilled staff (i.e., lower RN ratio), more Medicaid residents, and fewer hospice providers.

Also, after controlling for NH and resident characteristics, NH resident experiences differed by location. The more rural the NH, the more likely NH residents experienced risk-adjusted moderate-to-severe pain (i.e., lower quality; OR = 1.68, 95% CI = 1.35–2.09) (Kang et al., 2011) and the higher the prevalence of severe pain (Temkin-Greener et al., 2012) (i.e., lower quality). For example, 14% of residents in isolated rural NHs experienced severe pain and 15% of residents in NHs located in small rural towns and large rural towns experienced severe pain, with the lowest percent (13%) experiencing severe pain for NH residents in urban NHs.

The one exception was Clement (2018) who found that rural NHs have lower NH risk-adjusted rehospitalization rates (i.e., higher quality) (Clement et al., 2018). This team found that urban NHs re-hospitalize an additional 1.2 percent of their NH admissions from hospitals within 30 days compared to rural NHs. They also found that rural NHs have a higher proportion of older adults and populations living in poverty. Rural NHs were smaller, less reliant upon hospitals, and their average residents had lower average activities of daily living scores (i.e., less dependent/more independent). Also, differences were most apparent with more refined measures of rurality.

In the two articles that studied the rates for influenza and pneumococcal vaccination across residents in urban and rural NHs, authors found mixed results. No significant differences were found for urban and rural NH residents in Kang (2011) using the 2004 National NH Survey (Kang et al., 2011); however, Pu (2016) found rural NHs had higher rates of influenza and pneumococcal vaccination using the 2013 CMS NH Compare data (Pu et al., 2016).

We excluded 4 articles based on a “poor” rating, which was primarily applied because the study was about rural NH only (i.e., no comparison to urban NHs) (Meyer et al., 2014; Palumbo et al., 2011; Theeke et al., 2014; Towsley et al., 2013). These studies targeted rural states, identified rurality via self-report of NH residents, or defined rurality using three-category RUCA codes. Without comparison to urban NHs, it was not possible with the NOS for a rural NH study to receive a fair or good study quality rating given that ‘exposure’ was defined as urban-to-rural. Despite exclusion, these 4 studies made some interesting observations about rural NHs, which we delineate in Supplemental Table 2 that describes the methods, measures, and results of these rural NH only studies.

Discussion

Our review found limited and emerging descriptive evidence supporting rural NHs, compared to urban, provided sparse mental health support and limited access to hospice care after controlling for facility and resident characteristics. This is consistent with previous evidence on NH access and quality of NH care, which indicated overall that the more rural the NH, the lower the quality of care provided. Our more recent findings point to several main areas of needed improvement that NHs could continue to focus on, such as improving access to hospice use, education/resources for pain management, working on alternatives and strategies that provide on-site mental health for residents; our findings also point to continued work on hiring strategies to obtain and maintain nursing staff and highly-skilled staff and strategies for lowering hospitalization rates and contracture rates.

Importantly, we also found that quality outcomes across the literature were measured using different indicators of quality. Previous evidence to our review indicated that the more rural the NH the less likely the facility was to provide mental health services, the less likely residents receive preventive services and less likely to be visited by health care providers, and more likely to experience multiple hospital admissions (Agency for Healthcare Research and Quality, 2010; Coburn et al., 2002; Li, 2010; Phillips et al., 2004). Our review of the evidence found that the more rural the facility the less likely it was rated a 4 or higher using the CMS 5-star quality rating. Also, the more rural the facility the higher the facility contracture rate and the greater the increase in these rates overtime (Bowblis et al., 2013). In addition, the more rural the NH, the more likely the residents overall and the residents with HF were hospitalized (Temkin-Greener et al., 2012), and the more likely residents were to experience moderate-to-severe pain (Kang et al., 2011) and have a higher prevalence of severe pain (Temkin-Greener et al., 2012).

Additionally, our review found that compared to urban facilities, small rural town NHs had a higher prevalence of in-hospital deaths (20% small rural town vs 17% urban) and both isolated rural and small rural town NHs had lower prevalence of hospice use (19%, 22% vs 37%) (Temkin-Greener et al., 2012) (i.e., lower quality) (Saliba et al., 2005a, 2005b), suggesting that rural NHs were more likely to hospitalize their residents prior to death compared to urban facilities, even after controlling for facility and resident characteristics, including the presence of do-not-hospitalize orders. This may be due to having less palliative care resources in the more rural NHs (e.g., access to advanced clinicians such as physicians, nurse practitioners and/or chaplains) (America’s Care of Serious Illness: A State-By-State Report Card on Access to Palliative Care in Our Nation’s Hospitals, 2019). Another possible reason for low hospice and palliative care use could be cultural factors (McDermott & Selman, 2018) at the end of life. For example, Southern NHs house a large proportion of Black and African American NH residents (Travers et al., 2021), who have been found to have lower hospice use (Estrada et al., 2021; Housing Assistance Council, 2012) related to spirituality and clinician (McDermott & Selman, 2018) trust. It also could result from the lack of education on palliative care for NH staff and/or residents (Huff, 2019) that likely has resulted in wide palliative care variations across the U.S. (Tark et al., 2020)(Estrada et al., 2021; Housing Assistance Council, 2012).

Furthermore, the evidence available for the access and quality of care provided in rural vs urban NHs is hard to interpret because of the varied use and specificity of metrics for “rural.” Rural definitions can be based on administrative, land-use, or economic concepts, exhibiting considerable variation in the characteristics of the measured population (Cromartie, 2019; Cromartie & Bucholtz, 2008; Economic Research Service, 2019). In our review, three studies measured “rurality” using RUCA codes, most used a binary urban/rural variable, making it difficult to determine how rural a location is and compare findings related to similar types of rural NHs (Bennett et al., 2019). A more refined definition of rurality, for example using the six rural categories of the RUCA codes, allows for more targeted quality improvement strategies to address the quality of NH care based on the specific location of the NH, providing an opportunity to examine how facilities in isolated or small rural towns compared to facilities in large or less remote rural towns.

In two studies (Kang et al., 2011; Pu et al., 2016) rural NHs had higher influenza and pneumococcal vaccination rates in unadjusted analyses, although this finding was not robust in the multivariate analysis conducted by Kang (2011). Pu (2016) suggested that this result may be due to the higher number of government owned NHs in rural areas and government owned NHs having higher vaccination rates. Kang (2011) found that the primary source of payment (i.e., Medicare versus Medicaid) was related to vaccination rates with NHs with higher proportion of Medicaid patients having higher vaccination rates. The reason for the paradoxical findings of rural NHs having higher vaccination rates is not clear. In both studies, the data used were over a decade old. It is not clear if COVID-19 vaccination rates in NHs will have similar patterns.

Our study has limitations. It relies on manuscripts written in the last ten years that examined urban rural issues up through 2020. As is common in systematic reviews that use a search range including year, our review includes articles based on some data prior to 2010 and up through data in 2016. Furthermore, the identified literature is emerging and demonstrates gaps in the literature spanning the last ten years. Our search strategy is driven by the preset database filters and identifies articles that have specific terms such as urban/rural in the title or abstract or as a MeSH term, which are the National Library of Medicine’s Medical Subject Headings. This may result in some misalignment of the search terms and the AHRQ quality definition and also may exclude studies that do not align with the database filters; however, the purpose of using MeSH terms is to expand the search skills and identify articles that may not have been found only by title or abstract. We may have missed studies in which important urban rural comparisons were not the main focus, and therefore not apparent in either the title or abstract. Our study, however, importantly identifies articles that focus on urban and rural comparisons and have relevant urban and rural findings. Most notably, the number of relevant studies identified overall is small and these studies used older data, which limits the strength of any conclusions. However, our review identifies several gaps in the evidence that need further study. Furthermore, the variables used to measure the access and quality of care in NH varied so no consensus could be found for any given quality of NH care metric. Instead, it helped paint a descriptive detailed picture of care quality in rural NH. The variables used to assess urban vs. rural NH location were also not consistent, but these differences indicated that more refined rural measures of location revealed differences.

Implications for Research

Evidence regarding access and quality of care provided in rural vs urban NHs is hard to interpret because of the wide range of quality outcomes measured and the varied use and specificity of metrics defining “rural.” Evidence on why and how there are differences in the access and quality of care provided in urban and rural NHs is still emerging, with limited but only general results given the variety of outcome measures studied. Despite the data limitations, our review provides additional descriptive evidence to this picture that after controlling for resident and facility characteristics, rural NHs, compared to urban, provided sparse mental health support and limited access to hospice care.

Our review importantly highlights the need for more research examining urban and rural differences in the quality of NH care and raises several issues and gaps in the current evidence examining urban and rural differences in NH care. First, this review highlights issues of how the rurality metric is defined given that the more-finely defined measures of rural NH locations were able to show more apparent differences across the urban/rural location. Research is needed on how rurality is and should be defined. Future research on differences in urban and rural NH care should also focus on a common set of access and quality of care metrics for NHs, such as the Nursing Home Quality Metrics CMS database or Nursing Home Compare data. Evidence on urban-rural NH care differences based on relevant up-to-date data is needed. Additionally, most studies controlled for NH structural characteristics, thus providing a fair comparison of the access and quality of care in NHs across the urban and rural location but interpreting the significance of these facility characteristics and interacting them with urban/rural variables could also provide additional information on important differences across urban and rural NH care. Research is needed on how urban/rural is considered alongside other structural facility-based components of care, such as for-profit status or NH use of technology, or telehealth, or electronic remote order entry used by providers/pharmacists to manage complex drug regimens experienced by older adults in NHs (Alexander et al., 2020). Importantly, many of the quality metrics are measured in large data sets such as the Minimum Data Set (MDS), which is based on patient assessments completed and submitted to Center for Medicare and Medicaid Systems that are required for NH that receive payment from Medicare and Medicaid. Given this, researchers could include in their analyses a focus on examining and publishing urban and rural differences or disparities in these quality metrics.

Lastly, the use of telehealth during the COVID-19 pandemic may be mechanism that could potentially increase access for rural NHs. A recent study indicates larger metropolitan NHs reported greater telehealth use and ownership had little effect on telehealth use (Alexander et al., 2020); however further research is needed.

Conclusion

In sum, more research is needed examining urban and rural differences in the quality of NH care. Such research on urban-rural NH care differences should focus on a common set of quality of care metrics and future work is needed to specifically understand how and why quality differs across non-urban/rural NHs. Evidence is needed using up-to-date data on a common set of quality of care metrics across multiple dimensions of quality (e.g., access, safety, timeliness, patient/family centeredness), with finely defined measures of rurality, and analyses that examine structural, facility-based characteristics along with urban/rural variables.

Supplementary Material

SupplCopy

Supplemental Table 2. Description of Methods, Measures and Results for Excluded Rural only Nursing Home Studies (N=4)

Suppl

Supplemental Table 1. Quality Ratings of Included Articles using Newcastle-Ottawa Quality Scale (NOS) with AHRQ Thresholds