Social inequity in ageing in place among older adults in Organisation for Economic Cooperation and Development countries: a mixed studies systematic review

Abstract

Background

Most older adults wish to remain in their homes and communities as they age. Despite this widespread preference, disparities in health outcomes and access to healthcare and social support may create inequities in the ability to age in place. Our objectives were to synthesise evidence of social inequity in ageing in place among older adults using an intersectional lens and to evaluate the methods used to define and measure inequities.

Methods

We conducted a mixed studies systematic review. We searched MEDLINE, EMBASE, PsycINFO, CINAHL and AgeLine for quantitative or qualitative literature that examined social inequities in ageing in place among adults aged 65 and older in Organisation for Economic Co-operation and Development (OECD) member countries. Results of included studies were synthesised using qualitative content analysis guided by the PROGRESS-Plus framework.

Results

Of 4874 identified records, 55 studies were included. Rural residents, racial/ethnic minorities, immigrants and those with higher socioeconomic position and greater social resources are more likely to age in place. Women and those with higher educational attainment appear less likely to age in place. The influence of socioeconomic position, education and social resources differs by gender and race/ethnicity, indicating intersectional effects across social dimensions.

Conclusions

Social dimensions influence the ability to age in place in OECD settings, likely due to health inequalities across the lifespan, disparities in access to healthcare and support services, and different preferences regarding ageing in place. Our results can inform the development of policies and programmes to equitably support ageing in place in diverse populations.

Key Points

• Social dimensions influence the ability to age in place, but gaps remain in our understanding of this process.

• Those with greater socioeconomic and social resources, people of colour and rural dwellers are more likely to age in place.

• Those with higher educational attainment and women appear less likely to age in place.

• There are important intersectional effects across social dimensions that influence the ability to age in place.

• Additional research is needed on upstream determinants of social inequity in ageing in place and mechanisms to reduce inequities.

Introduction

The concept of ageing in place—the capacity of older adults to live in their own homes and communities as they age—has gained significant attention in the context of global population ageing [1]. An overwhelming majority of older adults wish to age in place as it fosters a sense of identity, autonomy and connectedness [2]. Policies that support ageing in place have emerged as priorities within the Organisation for Economic Cooperation and Development (OECD) [3]. Such policies have important systemic implications by offering a desirable and cost-effective alternative to residential long-term care (LTC) [4].

Ageing in place is a dynamic process influenced by health status, functional capacity, health services and support availability, environmental context and individual preferences [5]. However, as structural forces often create avoidable inequities in health outcomes, healthcare access and social support for underserved populations [6, 7], individual-level social determinants of health may also shape inequity in the ability to age in place [8]. Unmet health and support needs create barriers to remaining safely in one’s home, [9, 10] leading to increased utilisation of LTC or limited access to LTC if it becomes necessary [11].

A growing body of literature explores the concept of ageing in place [12], including a limited number of reviews that have examined lived experiences [13], decision-making processes [14], supportive technology [15] and the cost-effectiveness of ageing in place [16]. None of these reviews explored questions of social equity. Many studies have examined predictors of admission to LTC, which can be conceptualised as a fundamental disruption to ageing in place [17]. Three recent systematic reviews of LTC admission examined the influence of sociodemographic factors and found evidence that older adults who live with others, own their home and are people of colour have reduced LTC utilisation [18–20]. However, they found limited or inconclusive evidence regarding the role of other social dimensions, including gender, marital status, income and education [18–20].

Although previous literature serves as an important foundation for understanding ageing in place, gaps remain in our understanding of how social dimensions shape these processes. Existing studies typically focus on a single or a select few social dimensions and often do not consider their joint influence, resulting in a piecemeal understanding of the broader social context [6]. Recent calls to incorporate a wide range of social dimensions into health and social research emphasise the need for an integrated perspective. Tools, such as the PROGRESS-Plus framework, which outlines key social dimensions that influence health inequity [21], and the theoretical framework of intersectionality, which describes the potentially synergistic effect of social dimensions in shaping inequities [22], offer the means to structure nuanced evaluations of equity.

As the global population of older adults grows in size and diversity, the social determinants of health that create disparities across the lifespan likely influence the ability to age in place. It is therefore critical to understand how social dimensions shape inequity in ageing in place to inform health and social policies to support ageing in place [23].

Objectives

The primary aim of this review is to synthesise evidence of social inequity in ageing in place among older adults in OECD settings using an intersectional lens. A secondary aim is to examine how social inequities in ageing in place are measured in the literature, including how social dimensions are defined, what methods are used to quantify inequities, and to what extent studies account for intersectional effects.

Methods

This review follows the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) 2020 statement and guidelines [24], with modifications recommended by the PRISMA-Equity extension (Appendix A) [25]. The protocol was registered in PROSPERO (CRD42022332333) [26].

Search strategy and selection criteria

We developed a search strategy for MEDLINE, EMBASE, PsycINFO, CINAHL and AgeLine, using keywords and subject headings reflecting ‘social inequity’ and ‘ageing in place’ or ‘admissions to LTC’ (Appendix B). We included quantitative, qualitative or mixed-methods studies that examined inequity in ageing in place (defined as residence in or relocations between private dwellings) or admissions to LTC [17] among adults aged 65 or older along individual-level social determinants of health [8]. We included studies conducted in OECD settings to increase the generalizability of findings [27].

We restricted the search to peer-reviewed studies due to limited methodological transparency in grey literature, studies published since 2000 as ageing trajectories have evolved substantially since then, and studies published in English or French as no translation services were available. We excluded studies without original data, studies evaluating inequity in hospital discharge location, place of death, temporary admissions to LTC (e.g. post-acute care, rehabilitation) or discharges from LTC as these reflect fundamentally different trajectories of ageing in place, and studies reporting only unadjusted estimates as these do not inform evaluations of inequity.

Study selection

Identified records were imported into EndNote for deduplication [28]. Remaining references were uploaded into Covidence for screening [29]. Three reviewers (CBF, GB and GH) reviewed the titles and abstracts of the first 50 records and discussed discrepancies until consensus was achieved. Two reviewers (CBF, and GB or GH) then independently screened the titles and abstracts of all records. Potentially relevant records were retrieved for full-text review by two independent reviewers (CBF, and GB or GH) to determine eligibility. Disagreements were resolved by discussion among reviewers and consultation with a senior author (AQV) if necessary.

Data collection

We extracted estimates of social inequities in ageing in place, defined as relative or absolute differences between groups (quantitative studies) or determinants within groups (qualitative studies). We extracted details regarding the setting (country, follow-up period), population (sample size, age and characteristics) and methods (data sources, modelling strategy, confounders and methods accounting for intersectional effects) [30]. If studies reported results from multiple models adjusting for nested confounders, the most fully-adjusted model was extracted. If studies reported results from multiple modelling strategies, all results were extracted. Data was extracted by one reviewer (CBF) using a form that was designed and piloted for this review. A random audit of 20% was conducted by a second reviewer (GB), and discrepancies were resolved through consensus.

Assessment of study quality

Study quality was evaluated using the Mixed Methods Appraisal Tool (MMAT), a critical appraisal tool that facilitates the quality assessment of studies with diverse designs [31]. The MMAT evaluates study quality across five domains using design-specific questions evaluating the appropriateness of the design, data and analytical approach [32]. Each domain is scored yes, no or can’t tell. Yes scores are summed to generate an overall score. We considered scores of 0–2 as low quality, 3–4 as moderate quality and 5 as high quality. All studies were assessed by one reviewer (CBF). Studies were included regardless of their score to ensure a comprehensive synthesis of all identified social dimensions.

Review synthesis

Results were synthesised using a convergent design in which quantitative and qualitative data were synthesised separately and integrated into an overall synthesis using qualitative content analysis [33]. This approach supports a nuanced synthesis of diverse evidence. The synthesis was guided by the PROGRESS-Plus framework, which includes place of residence, race/ethnicity, occupation, gender, religion, education, socioeconomic position and social resources, plus other context-specific dimensions [21]. Results were grouped for synthesis within PROGRESS-Plus dimensions by including key indicators of each dimension and well-established proxies for dimensions with broad definitions: place of residence (e.g. rurality, region and environmental characteristics), socioeconomic position (e.g. income, home ownership and means-tested health insurance) and social resources (e.g. marital status and living alone) [8, 21]. Any identified social dimension that could not be classified within the pre-specified PROGRESS domains was included under other context-specific domains [21]. Due to expected substantive and methodological heterogeneity, no meta-analysis was planned.

All reported measures of effect (e.g. relative risks, absolute differences and themes) were included in the content synthesis. We generated forest plots of relative measures from quantitative studies to illustrate high-level trends. Results pertaining to inequity in admissions to LTC were transformed by calculating the reciprocal to facilitate interpretation in terms of ageing in place. For studies with multi-categorical measures, we plotted only the most extreme comparison. To facilitate the interpretation of results from studies that did not report confidence intervals, we denoted statistical significance using graphical indicators. Results of qualitative studies were synthesised separately by tabulating determinants within PROGRESS-Plus domains. We explored heterogeneity in the results by examining differences reported by studies using an intersectional approach.

Results

Of 4874 identified records, 55 publications were included (Fig. 1) [24].

Figure 1.

PRISMA flow diagram.

Study characteristics

Included studies spanned several OECD regions: North America (n = 39; the United States [n = 33], Canada [n = 6]), Europe (n = 8), the United Kingdom (n = 5) and Australia (n = 3) (Table 1). A large majority were quantitative (n = 51) and a minority were qualitative (n = 4).

Table 1.

Summary of included studies

Author, year	Country	Outcome	Study design	Study population	Sample size	Data period	Follow-up period	Age: mean (SD)/categorical (%)	Sex/Gender: % women
Quantitative studies—Longitudinal
Akamigbo 2007 [34]	United States	LTC	Secondary analysis of longitudinal cohort data	Asset and Health Dynamics Among the Oldest Old Survey (AHEAD) cohort	6242	1993–2004	11 years	Black: 77.46 (5.83) White: 77.33 (5.63)	Black: 66.80 White: 61.5
Aykan 2003 [35]	United States	LTC	Secondary analysis of longitudinal cohort data	Asset and Health Dynamics Among the Oldest Old Survey (AHEAD) cohort	6953	1999–1995	2 years	Women: 77.93 (6.05) Men: 77.14 (5.62)	60.79
Berridge 2018 [36]	United States	LTC	Secondary analysis of longitudinal cohort data	National Health and Aging Trends Study (NHATS) cohort	6459	2011–2012	1 year	Black: < 85: 89.5 85+: 10.5 White: < 85: 85.3 85+: 14.7	Black: 61.7 White: 57.8
Buys 2013 [37]	United States	LTC	Secondary analysis of longitudinal cohort data	University of Alabama at Birmingham (UAB) Study of Aging cohort	993	1999–2007	8.5 years	75.3 (7)	50
Cagney 2005 [38]	United States	LTC	Secondary analysis of longitudinal cohort data and linked Medicare claims data	National Long Term Care Survey (NLTCS) cohort	2603	1989–1993	5 years	Black: 65–74: 34.6 75–84: 45.0 85+: 20.5 White: 65–74: 38.7 75–84: 44.4 85+: 16.9	Black: 61.0 White: 64.0
Cai 2009 [39]	United States	LTC	Secondary analysis of longitudinal cohort data	Asset and Health Dynamics Among the Oldest Old Survey (AHEAD) cohort	5980	1995–2002	7 years	77.6 (5.8)	62.4
Cai 2015 [40]	United States	LTC	Secondary data analysis of administrative registry data	Home and community-based care (HCBS) enrolees	93 508	2006–2008	2 years	79.29 (8.34)	76.69
Casanova 2021 [41]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	66 248	2002–2014	2 years	Community-dwelling: 66–75: 59 76–85: 33 86+: 8 LTC: 66–75: 19 76–85: 42 86+: 39	Community-dwelling: 58 LTC: 69
Chyr 2020 [42]	United States	LTC	Secondary analysis of longitudinal cohort data	National Health and Aging Trends Study (NHATS) cohort	2725	2011–2018	8 years	Community-dwelling: 65–74: 67.6 75–84: 28.5 85+: 3.8 LTC: 65–74: 23.2 75–84: 41.5 85+: 35.3	Community-dwelling: 55.1 LTC: 68.0
Gandhi 2018 [43]	United States	LTC	Secondary data analysis of Medicare claims data	Medicare beneficiaries	84 212	2012	1 year	65–74: 48.5 75–84: 33.3 85+: 18.3	54.4
Goda 2011 [44]	United States	LTC	Secondary analysis of longitudinal cohort data	Asset and Health Dynamics Among the Oldest Old Survey (AHEAD) cohort	2283	1993–1995	2 years	77.7 (5.4)	NR
Gonzalez 2020 [45]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	8220	2006–2014	8 years	74.31 (range: 65–104)	57
Grundy 2007 [46]	England and Wales	LTC	Secondary analysis of longitudinal cohort data	Office for National Statistics Longitudinal Study (ONS-LS) cohort	36 647	1991–2001	10 years	65–69: 46.2 70–74: 29.4 75–79: 16.6 80+: 7.9	63
Hancock 2002 [47]	United Kingdom	LTC	Prospective cohort study	Patients in a single general practice	1425	1988–1999	11 years	Median: 80 (IQR: 77.0–83.5)	65
Hansen 2014 [48]	Denmark	LTC	Secondary analysis of administrative health data	Older adults in Copenhagen	37 109	2007	1 year	65–69: 27 70–74: 21 75–79: 18 80+: 35	62
Hays 2002 [49]	United States	AiP	Secondary analysis of longitudinal cohort data	Established Populations for Epidemiologic Studies of the Elderly (EPESE) cohort	4132	1986–1996	10 years	NR	NR
Hedinger 2015 [50]	Switzerland	LTC	Secondary analysis of administrative health data	Older adults in Switzerland	35 739	2000–2008	8 years	Men: 83.6 Women: 85.2	67.86
Himes 2000 [51]	Germany and United States	LTC	Secondary analysis of longitudinal cohort data	German Socio-Economic Panel (GSOEP) and Asset and Health Dynamics Among the Oldest Old Survey (AHEAD) cohorts	NR	GSOEP: 1984–1996, AHEAD: 1993–1995	GSOEP: 12 years, AHEAD: 2 years	GSOEP: 74.17 (NR) AHEAD: 77.42 (NR)	61.4
Jenkins Morales 2020 [52]	United States	LTC	Secondary analysis of longitudinal cohort data	National Health and Aging Trends Study (NHATS) cohort	3403	2015–2018	3 years	NR	61.3
Jenkins Morales 2020 [53]	United States	LTC	Secondary analysis of longitudinal cohort data	National Health and Aging Trends Study (NHATS) cohort	5212	2015–2017	2 years	65–79: 75.9 80+: 24.1	55.4
Kersting 2001 [54]	United States	LTC	Secondary analysis of longitudinal cohort data	Longitudinal Study of Aging (LSOA) cohort	7541	1984–1990	6 years	76.83 (5.59)	61.88
Kersting 2001 [55]	United States	LTC	Secondary analysis of longitudinal cohort data	Longitudinal Study of Aging (LSOA) cohort	7541 (Black: 555)	1984–1990	6 years	Black: 76.38 (5.70) Non-Black: 76.86 (5.58)	Black: 64.5 Non-Black: 61.80
Luppa 2012 [56]	Germany	LTC	Secondary analysis of longitudinal cohort data	German Study on Ageing, Cognition, and Dementia in Primary Care Patients (AgeCoDe) cohort	254	2003–2009	6 years	Community-dwelling: 83.5 (3.8) LTC: 84.6 (4.6)	Community-dwelling: 63.8 LTC: 74.0
Martikainen 2009 [57]	Finland	LTC	Secondary data analysis of population registry data	40% random sample of community-dwelling residents of Finland	280 722	1997–2003	6 years	Men: 65–69: 38.46 70–74: 29.45 75–79: 17.37 80–84: 9.53 85+: 5.20 Women: 65–69: 29.74 70–74: 27.16 75–79: 20.92 80–84: 13.53 85+: 8.74	61.36
Martikainen 2009 [57]	Finland	LTC	Secondary data analysis of population registry data	40% random sample of community-dwelling residents of Finland	280 722	1997–2003	6 years	Men: 65–69: 38.46 70–74: 29.45 75–79: 17.37 80–84: 9.53 85+: 5.20 Women: 65–69: 29.74 70–74: 27.16 75–79: 20.92 80–84: 13.53 85+: 8.74	61.36
McCann 2011 [58]	Ireland	LTC	Secondary analysis of longitudinal cohort data with linked administrative and Census data	Northern Ireland Longitudinal Study (NILS) cohort	51 619	2001–2007	6 years	65–74: 57.65 75–84: 34.27 85+: 8.08	58.26
McCann 2012 [59]	Ireland	LTC	Secondary analysis of longitudinal cohort data with linked administrative and Census data	Northern Ireland Longitudinal Study (NILS) cohort	51 619	2001–2007	6 years	NR	NR
Nihtilä 2008 [60]	Finland	LTC	Secondary analysis of population registry data	40% random sample of community-dwelling residents of Finland	140 902	1998–2002	5 years	Men: 72.0 (5.6) Women: 71.2 (5.0)	44.54
Noel-Miller 2010 [61]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	2116	1998–2006	8 years	Men: 75.24 (0.14) Women: 72.80 (0.14)	50
Opoku 2006 [62]	United States	LTC	Secondary analysis of longitudinal cohort data and linked Medicare claims and Census data	National Long Term Care Survey (NLTCS) cohort	6183	1982–1999	18 years	LTC: 65–74: 8.7 75–84: 35.3 85+: 56 Community-dwelling: 65–74: 28.4 75–84: 50 85+: 21.6	LTC: 77.7 Community-dwelling: 64.9
Pimouguet 2016 [63]	Sweden	LTC	Secondary analysis of longitudinal cohort data	Swedish National study on Aging and Care in Kungsholmen (SNAC-K) cohort	2404	2001–2007	6 years	77.8 (9.0)	66.1
Sabia 2008 [64]	United States	AiP	Secondary analysis of longitudinal cohort data	Panel Study of Income Dynamics (PSID) cohort	628	1972–1992	21 years	66–70: 7.44 71–85: 10.33	NR
Sarma 2009 [65]	Canada	LTC	Secondary analysis of longitudinal cohort data	National Population Health Survey (NPHS) cohort	2033	1994–2005	12 years	NR	60
Smith 2000 [66]	United States	LTC	Case–control study	Rochester Epidemiology Project (REP) cohort	Cases: 220 Controls: 296	1980-NR	NR (until death)	Cases: 80.8 (7.1) Controls: 81.6 (6.9)	64.54
Taylor 2022 [67]	Australia	LTC	Secondary analysis of longitudinal cohort data	Registry of Senior Australians (ROSA) cohort	16 864	2014-NR	2.4 years (median)	83.0 (7.3)	60.2
Thomeer 2015 [68]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	18 952	1998–2010	12 years	NR	Non-Hispanic White: 0.57 Non-Hispanic Black: 62 Hispanic: 58
Thomeer 2016 [69]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	21 564	1998–2012	14 years	NR	55.68
Tomiak 2000 [70]	Canada	LTC	Secondary analysis of administrative health data	Community-dwelling older adults	5153	1986–1990	5 years	Men: 65–74: 64.2 75–84: 30.5 85+: 5.3 Women: 65–74: 60.7 75–84: 31.9 85+: 7.4	55.9
Van den Bosch 2013 [71]	Belgium	LTC	Secondary analysis of administrative health data	Échantillon Permanent.e Steekproef (EPS) cohort	69 562	2004–2009	6 years	65–74: 68.28 75–84: 27.52 85+: 4.20	55.76
Willink 2016 [72]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	>10 000	1998–2012	14 years	65–74: 52 75–84: 33 85+: 15	58
Wu 2016 [73]	United States	LTC	Secondary analysis of administrative health data	Beneficiaries of the Michigan Choice Home and Community-Based Services (HCBS) Waiver Program	8172	2010–2014	4 years	65–74: 30.4 75–84: 37.0 85+: 32.7	74.2
Wu 2015 [74]	England	AiP	Secondary analysis of longitudinal cohort data	Medical Research Council Cognitive Function and Aging Study (MRC-CFAS) cohort	2424	1991–2001	10 years	74–79: 40.22 80–84: 31.52 85–89: 17.70 90+: 8.66	59.57
Yaffe 2002 [75]	United States	LTC	Randomised controlled trial	Medicare Alzheimer's Disease Demonstration and Evaluation (MADDE) trial	3859	1989–1994	3 years	78.9 (7.8)	59.78
Yu 2020 [76]	Australia	LTC	Secondary analysis of longitudinal cohort data linked with administrative data	Australian Longitudinal Study on Women's Health (ALSWH) cohort	4924	2003–2014	11 years	86.1 (NR)	100
Quantitative studies – Cross-sectional
Jenkins 2001 [77]	United States	LTC	Secondary analysis of longitudinal cohort data and linked Medicare claims data	National Long Term Care Survey (NLTCS) cohort	3180	1989	NA	82	83.1
Kang 2018 [78]	United States	LTC	Secondary analysis of administrative health data	Patients being discharged from hospitals	186 646	2007–2010	NA	65–79: 58.03 80+: 41.97	57.47
Liu 2000 [79]	Australia	LTC	Secondary analysis of population registry data	Residents in nursing homes	NR	1994–1995	NA	NA	NA
Raymo 2022 [80]	United States	LTC	Secondary analysis of longitudinal cohort data	Health and Retirement Study (HRS) cohort	95 641	2000–2016	NA	NA	NA
Richards 2008 [81]	Canada	AiP	Secondary analysis of survey data	Participation and Activity Limitation Survey (PALS)	722	2001	NA	NR	NR
Sharma 2016 [82]	United States	LTC	Secondary analysis of survey data	American Community Survey (ACS)	730 590	2009–2011	NA	84.69 (NR)	100
Smith 2008 [83]	United States	LTC	Secondary analysis of administrative health data	Residents in nursing homes	1 466 471	2000	NA	NR	NR
Trottier 2000 [84]	Canada	LTC	Secondary analysis of survey data	National Population Health Survey (NPHS) cohort	15 074	1996–1997	NA	Community-dwelling: 79.1 (NR) LTC: 84.0 (NR)	Community-dwelling: 57.8 LTC: 73.7
Qualitative studies
Burns 2016 [85]	Canada	AiP	Qualitative case study	Older adults residing in houseless shelters	15	2012–2015	NA	NR	46.7
Burns 2012 [86]	Canada	AiP	Qualitative case study	Older adults residing in two inner-city neighbourhoods	30	NR	NA	65–69: 3.3 70–74: 13.3 75–79: 36.7 80–84: 16.7 85+: 30.0	63.3
Pierce 2023 [87]	United States	AiP	Qualitative case study	Older adults from the LGBTQIA+ community	23	2019	NA	Median: 71 (NR)	NR
Prasad 2022 [88]	United States	AiP	NR	Older adults from the LGBTQIA+ community	31	NR	NA	50–64: 54.8 65–79: 45.2	35.5

^* Abbreviations: AiP, ageing in place; LGBTQIA+, lesbian, gay, bisexual, transgender, queer/questioning, intersex, asexual and others; LTC, long-term care; NA, not applicable; NR, not reported.

Methods used to examine inequity in ageing in place

Most studies used data from longitudinal survey-based cohorts or health administrative databases (Table 1). Among quantitative studies, most (n = 47) examined ageing in place indirectly through admissions to LTC as lifetime utilisation (n = 28) or time-to-admission (n = 20). Only four quantitative studies examined ageing in place directly, defining it as relocation between private dwellings (n = 2) [64, 74], continued residence in a private dwelling (n = 1) [81] or living alone in a private dwelling (n = 1) [49]. Qualitative studies examined experiences of ageing in place (n = 2) [85, 86] and navigating health and social services that support ageing in place (n = 2) [87, 88].

The most frequently-used methods to measure inequities were multivariable logistic regression and survival analysis (quantitative studies) and thematic analysis (qualitative studies). Quantitative studies typically adjusted for medical need (e.g. age, health status and functional capacity) and other social dimensions, with many noting a mitigation of inequities after adjusting for these covariates [41, 45, 53, 69]. However, some studies found greater gender- and race-based disparities after confounder adjustment [54, 68].

Study quality

Most studies were moderate to strong in quality (mean score 4.02/5). The main potential sources of bias among quantitative studies were a lack of representativeness of the study population and inadequate control for potential confounders (e.g. medical need and other social dimensions) (Appendix C). The main source of uncertainty regarding the risk of bias was a lack of clarity about the outcome definition, particularly whether short-term admissions to LTC were included.

Evidence of social inequity in ageing in place

Included studies covered 8 of the 9 PROGRESS-Plus dimensions: place of residence (n = 19), race/ethnicity (n = 28), occupation (n = 5), gender (n = 33), education (n = 23), socioeconomic position (n = 42), social resources (n = 45) and other context-specific domains (n = 8), but none examined religion.

Studies leveraged a broad range of indicators to measure PROGRESS-Plus dimensions (Table 2). Quantitative studies typically included only one measure of gender, race/ethnicity and place of residence in their models, but several included multiple indicators of socioeconomic position and social resources. Quantitative studies included dimensions or indicators that are typically absent from administrative data, offering a complementary view of social dimensions shaping inequity in ageing in place.

Table 2.

PROGRESS-Plus domains in literature on inequity in ageing in place

PROGRESS-Plus Domain	Studies (#)	Dimension or indicator measured (#)
Place of residence	19	Rurality/urbanicity (8) Rural-born (1) Geographic region (5) Residential continuity (1) Type of residence/dwelling (1) Age-friendliness of community (1) Gentrification (2) Neighbourhood safety & inclusivity (2) Experiences of houselessness (1)
Race/Ethnicity	28	Race (17) Race/Ethnicity (6) Ethnicity (3) Visible minority status (3) Neighbourhood ethnic diversity (1)
Occupation	5	Occupational class (e.g. manual, non-manual)/Hauser-Warren Index (3) Retirement status (2)
Gender	33	Gender (19) Sex (14)
Education	23	Highest completed degree (categorical) (14) Years of education completed (continuous) (7) Standardised classification (e.g. CASMIN, ISCED) (2) Highest completed education of father (1)
Socioeconomic position	42	Income—individual or household (continuous) (7) Income—individual or household (categorical) (8) Income relative to federal poverty level or low-income cut-off (7) Wealth (continuous) (2) Wealth (categorical) (5) Home ownership (19) Medicaid coverage (15) Area-level measures of deprivation (3) Housing conditions (1) Cost of living (1)
Social resources	45	Marital status (25) Bereavement/spousal death (3) Living arrangements (18) Household size (4) Number of children/childlessness (13) Number of siblings (4) Children moving out/in (3) Contact with others/community engagement (10)
Other context-specific domains	8	Immigrant status/Length of residence (5) LGBTQIA+ identity (2) Having a history of substance use (1)

^* Abbreviations: CASMIN, Comparative Analysis of Social Mobility in Industrial Nations; ISCED, International Standard Classification of Education; LGBTQIA+, Lesbian, gay, bisexual, transgender, queer/questioning, intersex, asexual, and others. Note: dimensions or indicators may sum to more than the number of studies evaluating each domain due to the inclusion of multiple indicators in several studies.

Place of residence

Nineteen studies evaluated inequity linked with place of residence, including 15 quantitative studies [34, 35, 40, 43, 49, 50, 59, 64–67, 70, 74, 76, 78] and 4 qualitative studies [85–88]. Rural residents are more likely to age in place than urban residents (Fig. 2a) [34, 43, 59, 70, 74]. Geographic trends in ageing in place appear to correlate with population density [50, 64, 78, 82], but may also reflect differences in regional policies (not shown) [50, 64, 65, 78, 85, 86, 88]. Qualitative studies identified residence in neighbourhoods undergoing gentrification [86, 88] or with safety issues [86, 87] and experiences of houselessness as barriers to ageing in place (Table 3) [85].

Figure 2.

Forest plots of relative measures of social inequity in aging in place identified by quantitative studies. Abbreviations: CI = confidence interval; HR = hazard ratio; OR = odds ratio; RR = rate ratio; SEP = socioeconomic position. Note for estimates lacking confidence intervals: ° indicates P > .05, * indicates P < .05.

Note for outliers: arrow indicates that the corresponding estimate and/or confidence interval falls outside the bounds of the plotted area.

Note for studies reporting multiple modelling strategies: Cai 2009-1: OR, Cai 2009-2: HR; Casanova 2021-1: model not correcting for missing data due to death, Casanova 2021-2: model correcting for missing data due to death using imputation.

Note for Smith 2008: did not report confidence interval or P-value.

Table 3.

Determinants of social inequity in ageing in place identified by qualitative studies

Study	PROGRESS-Plus Dimension
	Place of residence	Race/ethnicity	Socioeconomic position	Social resources	Other
Burns 2016 [85]	Residence in a shelter for populations experience houselessness		Housing conditions	Social exclusion Social insideness	Having a history of substance use
Burns 2012 [86]	Gentrification Overcrowding Neighbourhood safety	Ethnic diversity in neighbourhood (indirect displacement and social exclusion)	Cost of living Home ownership	Social inclusion Community
Pierce 2023 [87]	Neighbourhood safety & inclusivity		Home ownership Financial assets to navigate the cost of local gay-friendly LTC facilities	Discrimination Separation	LGBTQIA+ identity
Prasad 2022 [88]	Gentrification			Support from community members	LGBTQIA+ identity

^* Abbreviations: LGBTQIA+, lesbian, gay, bisexual, transgender, queer/questioning, intersex, asexual, and others.

Race and ethnicity

Twenty-eight studies evaluated racial and ethnic inequities, including 27 quantitative studies [34–40, 42, 43, 45, 49, 51–54, 61, 62, 64, 68, 69, 72, 73, 75, 77, 78, 80, 83] and 1 qualitative study [86]. Quantitative research almost unanimously found that racial/ethnic minorities are more likely to age in place than white older adults (Appendix D) [34, 37, 39, 40, 42, 43, 45, 52, 55, 61, 68, 69, 72, 73, 75, 77]. The magnitude of this inequity appears to depend on the racial/ethnic group, as studies reported greater differences among Hispanic than Black older adults [39, 42, 45, 61, 68, 69, 72, 75]. Qualitative research found that increasing ethnic diversity in neighbourhoods leads to perceived displacement among majority populations (Table 3) [86].

Occupation

Five quantitative studies examined occupational inequity [48, 49, 64, 74, 82]. Their results are inconclusive but suggest that manual workers are marginally slightly more likely to age in place than non-manual workers (Appendix D) [48, 74].

Gender

Thirty-three quantitative studies evaluated sex- and gender-based inequity [34, 36–43, 45, 46, 48, 49, 51–56, 59, 62, 64, 65, 68, 72–74, 77–81, 84]. Eight studies, including most studies using intersectional approaches, found that women are less likely to age in place than men (Fig. 2b) [34, 43, 45, 46, 59, 72, 78, 81]. However, four studies that did not account for joint effects found that women are more likely to age in place, suggesting that gender-based effects are highly intersectional [39, 41, 48, 54]. Gender-based inequity appears to be modified by socioeconomic position, as one study found that low-income women were more likely to age in place than higher-income women [37]. Race also appears to influence gender-based inequity, as three studies found that women of colour were more likely to age in place than white women [34, 55, 68].

Education

Twenty-three quantitative studies evaluated educational inequities [34–36, 38, 45, 48–50, 55, 56, 60–62, 64–66, 68–70, 74, 76, 82, 84]. Those with more education appear less likely to age in place than those with less education (Fig. 2c) [35, 36, 45, 56, 60, 68, 76], although two studies found the inverse [62, 84]. Several studies found that older adults with a secondary education were more likely to age in place than those with a primary education, suggesting a non-linear or threshold effect (not shown) [36, 38, 48, 70]. Educational inequities appear to differ by gender [35, 50, 69].

Socioeconomic position

Forty-two studies evaluated socioeconomic inequities, including 39 quantitative studies [34–39, 42–55, 57, 59–62, 64, 65, 68–72, 74, 76–78, 81, 82, 84] and 3 qualitative studies [85–87]. Higher income increases the likelihood of ageing in place (Fig. 2d) [38, 42, 44, 48, 51, 60, 65, 68, 81]. However, several studies found non-linear effects of income, indicating that those with moderate income are less likely to age in place than their low- or high-income counterparts (not shown) [53, 55, 57, 70, 72]. Income-based effects appear more pronounced among men [57, 70, 71]. There may also be important racial differences in the effect of income [34, 45, 55]. The effect of wealth is inconclusive but appears weakly associated with ageing in place among women [35]. Homeowners are more likely to age in place [39, 50, 53, 57, 59, 68–70, 72, 86, 87]. This effect is more pronounced among men [57, 69, 70]. In the United States, those covered by Medicaid were less likely to age in place [43, 45, 53, 68, 69]. Qualitative research identified socioeconomic inequities linked with housing conditions [85] and having sufficient assets to afford the cost of home care and LGBTQIA-friendly LTC facilities (Table 3) [87].

Social resources

Forty-five studies evaluated inequities linked with social resources, including 41 quantitative studies [34–39, 41, 42, 45–61, 63–66, 68–70, 72, 74–78, 81, 82, 84] and 4 qualitative studies [85–88]. Married older adults are more likely to age in place than their single, divorced or widowed counterparts (Appendix D) [34, 35, 45, 46, 48, 50, 51, 65, 68–70, 78, 81, 84]. This effect is magnified among men [35, 46, 50, 69, 70] and white populations [34, 68]. Those who live alone are less likely to age in place than those living with others [37, 39, 46, 47, 53, 54, 57, 59, 63, 72, 75, 76]. This effect is slightly more pronounced among men [46, 57] and white populations [55]. Those with children are more likely to age in place than those without children [34, 35, 38, 46, 50]. This effect is more pronounced among women [35, 69]. The effect of household size is inconclusive [65, 69]. Qualitative research identified social connectedness as a determinant of ageing in place, including the strength of community ties [85, 86, 88] and experiences of social exclusion or discrimination (Table 3) [85, 87].

Other context-specific dimensions

Seven studies revealed additional social inequities in ageing in place. Five quantitative studies found that immigrants are more likely to age in place than native-born populations (Appendix D) [45, 48, 50, 65, 68]. This effect appears stronger among immigrants from non-Western countries [48] and of Hispanic ethnicity [68]. Three qualitative studies identified barriers to ageing in place linked with LGBTQIA+ identity [87, 88] and having a history of substance use (Table 3) [85].

Intersectionality in studies of ageing in place

Less than half of quantitative studies (n = 21, 41%) used intersectional methods [34, 35, 37, 38, 45, 46, 49, 50, 55, 57, 58, 60, 61, 68–71, 74, 78, 80, 82]. The most commonly-used method was stratification along dimensions of interest (n = 17), including gender (n = 11) [35, 46, 50, 57, 59–61, 69–71, 80], race/ethnicity (n = 6) [34, 49, 55, 68, 80, 82], socioeconomic position (n = 2) [37, 60] and education (n = 1) [60]. Some studies evaluated intersectional effects using statistical interactions between two dimensions (n = 7), including race and social resources (n = 2) [38, 69], gender and social resources (n = 1) [58], race/ethnicity and socioeconomic position (n = 1) [45], place of residence and socioeconomic position (n = 1) [74], race/ethnicity and immigrant status [68] and between social resource indicators (n = 1) [61].

Nearly all studies accounting for intersectional effects found that social dimensions jointly influence ageing in place. The effect of several socioeconomic position and social resource indicators, including home ownership [57, 69, 70], income [57, 70, 71], marital status [35, 46, 50, 69, 70] and living alone [46, 57] were more pronounced among men, while wealth [35] and having children [35, 69] were more pronounced for women (Fig. 2e–g). Some social inequities (e.g. gender—Fig. 2d) were only evident when using intersectional approaches, suggesting that intersectional effects contribute to observed heterogeneity in studies modelling social dimensions discretely.

Discussion

Our findings indicate that social dimensions impact the likelihood of ageing in place. Older adults with greater social resources and higher socioeconomic position are more likely to age in place, potentially due to greater access to health services and social support to promote health and autonomy [89], as well as having sufficient resources to adapt their environment to enable them to age safely in place [9, 90]. The effect of income appears non-linear, which may indicate limited access to LTC among lower-income populations [91] or reflect complex interactions between income and other socioeconomic factors [92]. Older adults with higher educational attainment appear less likely to age in place. This counterintuitive finding may reflect differences in family structure [93] or result from bias due to exposure misclassification if an underlying threshold effect is incorrectly specified [94]. Women are less likely to age in place, potentially due to longer life expectancy [95], a higher burden of chronic disease [96] and functional limitations [97] and differences in spousal caregiving [98, 99]. Racial/ethnic minorities and immigrants are more likely to age in place, potentially due to differences in social norms surrounding caregiving or preferences for ageing in place [100], or disparities in access to LTC [101]. Rural residents are also more likely to age in place, possibly due to stronger community ties [102] or limited access to LTC [103]. We identified important intersectional effects across dimensions, highlighting gender-related effects of socioeconomic position and social resources in particular [104].

Social inequities in ageing in place stem from structural determinants that influence the unequal distribution of resources and opportunities [8]. These mechanisms shape differential health status and functional capacity across the lifespan [105] and disparities in access to and utilisation of health services and social support among underserved populations [106]. Individual-level social determinants of health therefore influence the degree of support needed and the availability of financial and social resources to adequately support ageing in place by adapting dwellings, relocating or mobilising support to accommodate changing needs. This context is critical to inform the careful interpretation of inequities as the unequal distribution of health and resources shapes both the ability to age safely in place and the capacity to exercise one’s choice between ageing in place or being admitted to LTC [9].

A key strength of this review is its broad scope, as our search strategy captured a wide range of social dimensions. Similarly, the inclusion of diverse outcomes to account for the absence of a widely-accepted definition of ageing in place resulted in a comprehensive overview [17]. The inclusion of both quantitative and qualitative research contributed to a rich synthesis. While quantitative evidence measures the magnitude of social inequity and identifies intersectional effects, qualitative research is better suited to capturing populations that are under-represented in traditional epidemiological data. Finally, our integration of the PROGRESS-Plus framework [21] with an intersectional lens [22] contributed to a structured narrative that revealed distinct and synergistic effects.

As with all reviews, it is possible that some relevant articles were missed by our search and that our results are influenced by publication bias. However, this is likely mitigated by the large number of included studies. Our results may also be skewed by the utilisation of the same cohorts across multiple studies. The restriction to studies conducted in OECD settings—a large proportion of which were conducted in the United States—may limit the generalizability of results. The absence of measures of ageing in place in administrative data and longitudinal cohorts limited the ability to examine ageing in place directly [17]. This resulted in significant methodological heterogeneity in included studies and prevented meta-analysis.

Our synthesis revealed several opportunities for future research on ageing in place. We identified social dimensions that have received limited attention, including occupation, immigrant status, LGBTQIA+ identity and religion. Additional research on these dimensions is needed to better understand how they influence ageing in place. Further research exploring the intersectional effects of gender, socioeconomic position and social resources is also needed. Research grounded in a life-course perspective and qualitative research exploring lived experiences of ageing in place within social dimensions is needed to illuminate the causes of inequity. More broadly, the development and inclusion of direct measures of ageing in place into prospective cohorts would enhance our understanding of complex longitudinal trajectories. Finally, research focusing on upstream determinants of inequity in ageing in place is needed, including differences in preferences for ageing in place [107], disparities in access to healthcare, unpaid caregiving and LTC [11, 108], as well as the influence of the health and social policy landscape.

Conclusions

This review offers a comprehensive synthesis of social inequity in ageing in place in OECD settings. Our results indicate that social dimensions play an important role in shaping the ability to age in place and reveal important intersectional effects across dimensions. These insights are critical for efforts to advance equity in ageing populations, as failure to consider inequities can exacerbate disparities [109]. Our findings can inform the development of policies, programmes and services to promote ageing in place among diverse populations and ensure that all older adults have the opportunity to remain in their homes and communities for as long as they wish and are able [110].

Declaration of Conflicts of Interest

None declared.

Declaration of Sources of Funding

This research was supported by a project grant from the Canadian Institutes of Health Research (FRN: 166208) held by A.Q.-V. and I.V. C.B.-F. is supported by a Fonds de recherche du Québec—Santé (FRQS) Doctoral Award. A.Q.-V. is the recipient of the Canada Research Chair in Policies and Health Inequalities.