Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Inequities in home care use among older Canadian adults: Are they corrected by public funding?

  • Afshin Vafaei,

    Roles Conceptualization, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

    Affiliations Department of Family Medicine, Queen’s University, Kingston, Ontario, Canada, Department of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada

  • Ricardo Rodrigues,

    Roles Conceptualization, Writing – review & editing

    Affiliations ISEG Lisbon School of Economics and Management, University of Lisbon, Lisbon, Portugal, SOCIUS, Research Centre in Economic and Organizational Sociology, Lisbon, Portugal

  • Stefania Ilinca,

    Roles Conceptualization, Writing – review & editing

    Affiliation European Centre for Social Welfare Policy and Research, Vienna, Austria

  • Stefan Fors,

    Roles Conceptualization, Writing – review & editing

    Affiliations Aging Research Center, Karolinska Institutet & Stockholm University, Stockholm, Sweden, Centre for Epidemiology and Community Medicine, Region Stockholm, Stockholm, Sweden

  • Selma Kadi,

    Roles Writing – review & editing

    Affiliation European Centre for Social Welfare Policy and Research, Vienna, Austria

  • Eszter Zolyomi,

    Roles Writing – review & editing

    Affiliation European Centre for Social Welfare Policy and Research, Vienna, Austria

  • Susan P. Phillips

    Roles Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing

    phillip@queensu.ca

    Affiliations Department of Family Medicine, Queen’s University, Kingston, Ontario, Canada, Department of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada

Abstract

Background

Although care use should parallel needs, enabling and predisposing circumstances including the socio-demographic inequities of socioeconomic status (SES), gender, or isolation often intervene to diminish care. We examine whether availability of state-funded medical and support services at home can rebalance these individual and social inequities, and do this by identifying if and how intersecting social identities predict homecare use among older Canadian adults.

Methods

Using the Canadian Longitudinal Study on Aging (CLSA) of 30,097 community-dwelling adults aged 45 to 85, we performed recursive partitioning regression tree analysis using Chi-Squared automatic interaction detection (CHAID). Combinations of individual and social characteristics including sociodemographic, family-related, physical and psychological measures and contextual indicators of material and social deprivation were explored as possible predictors of formal and informal care use.

Results

Diminished function i.e. increased need, indicated by Activities of Daily Living, was most strongly aligned with formal care use while age, living arrangement, having no partner, depression, self-rated health and chronic medical conditions playing a lesser role in the pathway to use. Notably, sex/gender, were not determinants. Characteristics aligned with informal care were first—need, then country of birth and years since immigration. Both ‘trees’ showed high validity with low risk of misclassification (4.6% and 10.8% for formal and informal care, respectively).

Conclusions

Although often considered marginalised, women, immigrants, or those of lower SES utilised formal care equitably. Formal care was also differentially available to those without the financial or human resources to receive informal care. Need, primarily medical but also arising from living arrangement, rather than SES or gender predicted formal care, indicating that universal government-funded services may rebalance social and individual inequities in formal care use.

Background

Access to formal homecare tends to align with medical need, but vary with socio-demographics such as income, gender, widowhood, or living arrangement, even when medical need is comparable [1]. In theory, universally available public funding could eliminate socially mediated care inequities in older adults’ access to and use of homecare. This has, however, seldom been directly tested [2]. Testing matters for several reasons. Social circumstances and their sequelae are among the strongest known determinants of health but are also perplexingly resistant to change. Evidence that government-funded universal medical care could rebalance the health impact of social inequities would make a persuasive case for such funding. We will look at whether intersecting social characteristics that are known to undermine use of formal homecare persist in a setting (Canada) with longstanding government funded medical and formal homecare systems.

Medicare in Canada

A guiding principle of the Canadian universal care system (subsequently referred to as Medicare) is that access to all essential medical care should be unrelated to income, age, sex, location, or other sociodemographic characteristics. However, because of perceived gaps between essential formal homecare paid for by the state, and optimal care, even in Canada accessing a breadth of formal care may confer disproportionate financial burdens upon those with lower incomes [3]. In the Canadian context formal homecare refers to clinical and support services arising from medical need and provided to a person living in the community (and not in an institutional care setting) by a paid worker rather than a family member or a friend. In 2015–16, 91% of in-home nursing care involved no payment by the recipient. However this was true for only 44% of support services [4]. Further, 64.6% of those receiving formal home care had the costs paid for entirely by government as did 36.5% of the recipients of support services [4]. It should be noted that this represents the national average and does not acknowledge variations across provincial jurisdictions. Funding for informal care is not built into the Canadian system (i.e. Canadians receive no direct government payments to provide care to family or friends), making it reasonable to expect that those with limited human or financial resources might require and use formal government-funded care more than would others who, for example, are able to offset a friend’s time commitment to provide informal home care with material compensation. Having access to informal or self-funded care might, therefore, interact with, and diminish the need for formal care and, in so doing, alter the association between social characteristics and formal care receiving. For example, women who lack informal caregivers might appear to utilise more formal care if, as should be the case in countries with Medicare (e.g. Canada), such care can be accessed without personal financial burden.

Care needs of older adults

As populations age the prevalence of chronic diseases and functional limitations increases, precipitating greater need for homecare [5]. Generally, these needs arise from limitations in function, Activities of Daily Living (ADL) or Instrumental Activities of Daily Living (IADL), and with multimorbidity [6]. However, patterns of care use seem to also be related to the complex interplay of individual characteristics, social networks, and contextual/policy level circumstances. Theoretical models referred to in the literature tend to describe patterns of care use well, but not those characteristics that lead to use. Understanding these possible predictors is a necessary first step in modifying social and healthcare systems to better address the needs of aging populations.

Defining formal and informal care.

There are no standard definitions of formal and informal care. In general, formal care refers to ‘paid’ services based on or arising from medical need, whereas care delivered by family, relatives, friends, neighbours, or community members is considered to be informal care [7]. Informal and formal care often address different needs and are likely inversely related, with access to the former diminishing the need for the latter [8, 9]. Informal and formal care use will be influenced differently by enabling and predisposing factors (see below for more information on these terms). For example, in settings without state funded care, persons of higher socio-economic status (SES) and greater income are more able to pay for formal services while those of lower SES may need to cobble together informal, cost-free supports [10].

Need for, and use of care.

Care use can be conceptualised using three dimensions: need for care, predisposing characteristics, and enabling circumstances [11]. Need arising from functional, physical and cognitive health limitations represents the most immediate reason for seeking care [1214]. Individual-level factors such as age, sex, and health behaviours can predispose to functional limitations that trigger need. Enablers tend to be structural or contextual, and shape care access and use. They include income [10, 15, 16], availability of health insurance, local policies [17, 18], and social support networks [18, 19]. Enabling and predisposing factors sometimes overlap. For instance, gender can be both. Women, in general, have more established social networks that can enable informal care. On the other hand, older women more often live alone with no household informal caregiver [20, 21]. In addition, female sex is a recognized risk factor for a variety of chronic diseases and functional deficits that precipitate a need for care [22]. Economic considerations and cultural norms of women’s lower SES relative to men and of men’s stereotypic stoic behaviour add more complexity to gendered patterns of care receiving. When determining the relationship between social characteristics and use of formal care it is therefore important to consider not only single determinants such as sex, income, or living situation, but also the interplay or intersections among these as they shape need, enabling and predisposing dimensions of care.

Examining intersectionality and care.

With roots in the study of social inequalities and, particularly, gender inequity [23, 24] intersectionality theory assumes that membership in marginalized social groups generates interlocking rather than additive or multiplicative systems of oppression and harm [25]. These “social locations” defined by interconnections or intersections of various social factors create both power and marginalization [24]. Numerous methods for studying intersectionality exist although, as would be expected with a novel framework, none has yet emerged as ‘the best’ [26]. One of these, recursive partitioning, groups individuals using stepwise division of social locations [2628]. Its output is a graphical ‘tree’ that illustrates how combinations of factors defining homogenous subgroups contribute to the probability of the outcome of interest. Of particular relevance to our research, recursive partitioning allows for identification of subgroups (i.e. those with combinations of characteristics) at higher risk within larger groups that are deemed, overall, to be of low risk for the outcome. This contrasts with traditional regression methods that quantify subgroups at ‘risk’ specifically within the high-risk group. To illustrate this—utilising recursive partitioning one could identify which characteristics precipitate care among a low risk group with no limitation in ADLs.

As stated previously, access to formal home care should align with need arising from cognitive impairment, mobility decline or multimorbidity [7, 29]. Research has also identified disparities in socio-demographics such as gender, widowhood, living arrangement and income that interact with medical circumstances to alter need for care [1]. Specifically, greater need has been documented among women [13, 30], those with lower levels of education [31], and racialized populations [32]. Few, however, have considered intersecting characteristics that predict access to and use of that care. None has examined whether availability of government funded, universal access to medically necessary formal care can correct for the impact of individual or intersecting socio-demographic inequities. Guided by an intersectionality framework, our aim was to identify circumstances that predict formal care use among middle-aged and older adults in a setting where access is decoupled from an individual’s ability to pay for that care. To disentangle the effects of need, enabling and predisposing factors, and the offsetting effect of availability of informal care, we apply an intersectional framework and operationalise this by constructing regression trees for both formal and informal care.

Methods

Settings and participants

Data used were from the baseline (2010 to 2015) Canadian Longitudinal Study on Aging (CLSA), a random sample of 30,097 community-dwelling adults aged 45 to 85. Participants resided within a 25- to 50-km radius of 1 of the 11 data collection sites (Victoria, Vancouver, Surrey, Calgary, Winnipeg, Hamilton, Ottawa, Montreal, Sherbrooke, Halifax, St. John’s) in 7 Canadian provinces [33]. Those living in a First Nations community, an institution or a care facility, full-time members of the Canadian Armed Forces, anyone unable to speak French or English, or with cognitive impairment that made them unable to understand the study or answer basic personal questions were excluded.

Sampling strategy and data collection

CLSA employs two sampling strategies for the Comprehensive Cohort used. Recruitment from provincial health registries (14% of the sample) followed random selection of eligible persons with 86% of the sample recruited through random digit dialing of landline telephone numbers for a given geographic area. After establishing eligibility of the respondent, consent was obtained either verbally or in writing. To ensure adequate representation of diverse demographic groups, the CLSA sample was stratified within each province by age group, sex, and distance from the data collection site [33]. Trained interviewers administered questionnaires in participants’ homes or a collection site. The CLSA reports a participation rate of 45% and an overall response rate of 10% [33].

Assessment of formal and informal homecare use

Use of homecare was self-reported via answers to the following: "During the past 12 months, did you receive short-term or long-term professional assistance at home because of a health condition or limitation that affects your daily life, for any of the following activities?" and "During the past 12 months, did you receive short-term or long-term assistance from family, friends, or neighbours because of a health condition or limitation that affects your daily life, for any of the following activities?" Listed activities included personal care such as assistance with eating, dressing, bathing, or toileting, medical care such as help taking medicine or help with nursing care (for example, dressing changes or foot care), managing care such as making appointments, help with activities such as housework, home maintenance, or outdoor work, transportation, including trips to the doctor or for shopping and meal preparation or delivery.

Potential predictors

Sociodemographic characteristics.

Information used included sex, highest educational attainment (less than secondary school, secondary school graduation, some post-secondary, post-secondary graduation) and total personal income from all sources, before taxes and deductions, in the past 12 months as a categorical variable (<$20,000, $20,000-$49,999, $50,000-$99,999, $100,000-$149,999, ≥$150,000, Don’t know/No answer/Refused). By linking participants’ residence postal codes to dissemination areas, their place of residence was classified into ‘urban core’, ‘other urban’, and ‘rural’, a standard proxy measure for place of residence in Canada as per precedents [34]. Additional socio-demographic characteristics included age in years, country of birth, and time, in years, since immigration.

Family related variables.

These included marital status (with partner, widowed, divorced, separated, single) and living arrangement (number of generations in the household) as they are previously documented family-related predictors of receiving formal and informal care at home [19] and in the community [15, 35].

Physical and mental health variables.

To measure functional status, we used a composite variable derived by the CLSA that calculates the total number of times the respondent needed help with or was completely unable to do an activity in the Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL). This derived variable is a modification of the Duke Older Americans Resources and Services Multidimensional Assessment Questionnaire [36] that excludes questions regarding meal preparation ability. The final score (range 0–13) was then categorized into four groups: ‘0’ = no activities of daily living problems, ‘1–3’, ‘4–5’, and ‘more than 6’ as ‘mild’, ‘moderate’, and ‘severe’ activities of daily living problems, respectively.

Perception of health was evaluated via two direct questions: “Would you say your health is excellent, very good, good, fair, poor?” and “Would you say your mental health is excellent, very good, good, fair, poor?” For analysis, we collapsed the first three categories into ‘good’ and the last two into ‘poor’ Self-Rated Health (SRH) and Self-Rated Mental Health (SRMH) categories. The 10-item version of the Centre for Epidemiologic Studies Depression (CESD) scale was used to measure depression with a cut-off of 10 or more indicating a positive screen for depression [37], and number of chronic conditions was entered into the analysis as a dichotomous variable (‘no self-report of chronic condition’ vs ‘at least one’).

Contextual variables.

To measure contextual indicators of deprivation, Material and Social Derivation Indices (MSDI) for all Canadian census dissemination areas were retrieved from the ‘Institut national de santé publique du Québec’ website and linked to participants’ postal codes [38]. Material indicators included 1) the proportion of the population aged 15 years and over with ‘education less than high school’, 2) with ‘no employment’ and 3) average income of the area. Social deprivation implies being separated, divorced or widowed, living alone, or in a single-parent family. The distribution of deprivation scores was then divided into quintiles with quintile 1 representing the population living in the most privileged area and quintile 5 living in the most deprived one.

Ethics approval and consent to participate

The study was approved by the Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board (HSREB). Tools for Research at Queen’s (TRAQ) number: 6020994 in accordance with the ethical standards of the Declaration of Helsinki (1964) and its subsequent amendments. Informed consent was obtained by the CLSA from all participants before data collection.

Statistical analysis

Descriptive analyses were performed to estimate the distributions of covariates, overall, and across the four care groups: 1) receiving formal care only, 2) receiving informal care only, 3) receiving a mix of formal and informal care, 4) receiving no care. Differences in distributions were compared statistically by Chi-square test for contingency tables.

To identify combinations of significant variables that predict care use we performed recursive partitioning regression tree analysis using Chi-Squared Automatic Interaction Detection (CHAID) analysis [39] as the growing method. Tree-Structured modeling techniques were developed more than four decades ago [40] mostly for data mining purposes, and are being used in health research with increasing frequency in recent years [41, 42]. CHAID is a classification method that examines the relative importance of each precursor in explaining the occurrence of a specified level of the outcome. It chooses the optimal partition based on χ2 calculated statistical significance adjusted using the Bonferroni correction. In order to allow more precursors in the models we set the tree depth limitation to five (instead of three levels which is the default for CHAID) and relaxed stopping criteria for splitting to a maximum size of 50 for parent nodes and of 20 for child nodes. This approach resulted in large trees with some nodes sized as small as 3. To generate more interpretable trees and meaningful nodes we then modified these criteria. In the final trees no parent node smaller than 100 was allowed to split and the minimum size for end nodes was set to 40. Applying this restriction decreased the final number of nodes of the formal care model from 41 to 35 and of informal care model from 38 to 36. To test trees’ stability and confirm the validity of the prediction accuracy (i.e., the correct classifications) of the models we ran a 10-fold cross-validation. This standard method of validation divides the sample randomly into 10 mutually exclusive subgroups, then each of the 10% folds serves once as a test sample while the rest of the sample (90%) is used to validate findings. The cross-validation risk, which is the proportion of cases incorrectly classified after adjustment for prior probabilities and misclassification, is the average of risk estimated across the 10 test samples. CHAID algorithm only prints the full-sample classification table. Separate algorithms were generated for ‘any informal care use’ and ‘any formal care use’ as the outcomes.

Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and SPSS version 27.

Results

Approximately half of those surveyed (50.9%) were female, 67% lived with a partner and more than 77% had postsecondary education. Overall, participants were high functioning: almost 90% reported no ADL or IADL limitation, or perceived their health as good (Table 1). Women received both formal and informal homecare more frequently (2.7% vs 1.9% among men for formal care and 10.1% vs 7.0% for informal care).

Need, as indicated by limitations in ADL/IADL, was the primary determinant for receiving care, whether formal or informal; using regression trees language, need was the first ‘splitter’ or branch of the tree. The next level of possible predictor was different for formal and informal care. We, therefore, report detailed results separately for formal and informal care outcomes.

Receiving formal care

By generating a total of 35 nodes and 19 terminal nodes CHAID identified that 28,611 (95.1%) participants did not receive formal care and 1,459 (4.9%) did. The overall risk of misclassification was 4.5% and remained almost the same (4.6%) after 10-fold cross-validation.

As stated above, the primary precursor or possible predictor (the first splitter) of formal care use was limitation in ADL and IADL. Among those with moderate to severe functional limitations 57% received formal care. The comparable proportion among those with mild limitation was 22.4%. For those with any level of ADL limitation the next ‘branch’ in the tree, that is, the next indicator was having no partner. Among the moderate to severely limited group, 9.5% of those with a partner received formal care compared with 76.4% of those with no partner. Comparable proportions for those with mild limitation were 15.5% and 29.9%, respectively. Further dissecting the branches of the tree to find intersecting characteristics that aligned with receiving care (i.e. beyond partner status) identified subsequent possible predictors for formal care to be: SRH (for moderate/severe ADL limitation + partner: 32.2% with good SRH received care vs 50.0% in the poor SRH group), living arrangement + SRH for the mild limitation with no partner group, and SRH + age for mild limitation with partner group (Fig 1).

Among those with no functional limitations (89.9%), 2.4% received formal care. This allocation of formal care was predicted best by age (45–64 Y– 1.6%, 65–74 Y– 2.6%, 75+ Y– 5.5%). For those younger than age 75, SRH and family-related variables such as partner status and living arrangement were important, whereas in the older than 75 group, self-reported depression also played a role.

Overall, sex, that is, being a man or a woman, was not a branch point for formal care. Of particular note measures of SES such as income or education also did not have any impact on receiving formal care.

Informal care

As with formal care, the main precursors of informal care, were ADL/IADL limitations, SRH, partner status and depression. In contrast to receiving formal care, however, sex, and immigration status also emerged as important (Fig 2). With moderate to severe ADL limitations, partner status remained important but only among women. In this group almost 86% of women with a partner received informal care compared with only 60% of women with no partner. Age was a possible predictor of receiving informal care regardless of SRH status in those with mild functional limitations. Country of birth was predictive for those with no depression symptoms. Canadian-born participants were somewhat more likely to receive informal care than were foreign-born individuals (28% vs 22%).

The pattern of receiving informal care was more complex among those (7.3%) with no ADL limitations, showing intersections among sex, immigration status and health. Country of birth aligned with informal care use for men with poor SRH, but not for women. In those with good SRH and no chronic condition years since immigration predicted informal care use for women.

The regression tree generated by the CHAID algorithm for informal care use had almost twice the risk of misclassification relative to the tree for formal care use. The final tree, which included a total of 36 nodes and 19 terminal nodes, correctly classified 26,470 (99.1%) of those who did not receive informal care and 427 (12.7%) of those who received this level of care. Overall risk of misclassification, therefore, was calculated at 10.6%. The risk increased slightly to 10.8% after 10-fold cross-validation.

Discussion

While need, as measured by ADL/IADL, was the most important antecedent of both formal and informal care use, other possible precursors varied considerably across the two types of care. This is consistent with theoretical conceptualisations that consider formal and informal care use as different entities [9, 43]. In contrast to findings that socio-economic status (SES) measured in various ways predicted formal care use in other countries [15, 16, 44], in the Canadian context we have studied this did not emerge. While variability in SES definitely exists in Canada, Medicare seems to ameliorate SES related inequalities in home care utilisation. With government funding for essential although not always optimal formal homecare, previously identified barriers of sex or SES are no longer apparent as sources of inequity. Those whose lives were marked by the isolation of having no partner or living alone inevitably had less access to informal care, but this was offset by greater use of formal care. We cannot, however, determine whether living with someone rather than alone acted as a barrier to receiving formal care. Although individual circumstances are not amenable to change via social policies, perhaps in a setting such as Canada, where, without financial penalty, formal care can offset a lack of informal care, government funded formal care may minimise the harms of socioeconomic or individual deficits.

Our findings diverge somewhat from those of Gilmour [4] whose multivariate analysis of formal home care use among Canadians demonstrated that those of lower socioeconomic status received more care. There are several possible explanations for this divergence. Although the timeframe studied was similar, Gilmour included populations age 12+ whereas our study was limited to adults aged 45+. Perhaps more important is that multivariate analyses [4] examine the impact of single characteristics while controlling for all others, while recursive partitioning regression tree analysis (our research design) begins to tease out intersections of various characteristics. This has the potential to enrich findings. Both papers do demonstrate that with government funding for home care, the disadvantages of economic deprivation disappear.

We examined whether precursors of care use changed with different levels of ADL/IADL limitation. Starting from need and dividing the study population into those who did and did not have limitations the next characteristics that aligned with any care use were marital status, SRH, age, number of chronic conditions, and depression. We explored whether intersections among these factors explained variability in care use. Of note, SES measured either at the individual level by income and education, or at a contextual level by deprivation indices and residence location was not of significance. Sex, country of birth, and years since immigration were aligned with informal but not formal care. Family supports were, however, important sources of care for those with no ADL limitations. Informal care use was also a function of an individual’s immigration status. These findings are consistent with existing precedents that consider informal and formal care receiving as different ‘care’ behaviours [9].

Results support the construct that deep patterns of care use will be uncovered when an intersectional approach is adopted. Utilizing recursive partitioning methods revealed complex interactions for high risk subpopulations at specific, intersecting social locations. We were able to identify factors that intersect as possible predictors of formal and informal care use. Of note, in the Canadian context of government funded formal care, sex was not a predictive characteristic for access to it, while living arrangement was. However, as living arrangements in old age are profoundly gendered (men are more likely to live with spouses while women live alone), the finding that both formal and informal care depended on living arrangements, raises relevant gender issues [45]. Older men are more likely to provide informal care to their spouses [4648], adding to the interplay of the social factor, gender, in shaping care use patterns in old age. Our exploration of intersecting circumstances demonstrates that it was not sex, alone, that predicted care utilisation but rather the effect of living alone, a reality faced disproportionately by older women [49].

Our regression tree models hinted that potential unmet care needs remain. For example, 43% of individuals with moderate to severe ADL limitations did not receive formal homecare. This might be explained by marital status and associated availability of informal care, as those with partners were the ones less likely to receive formal care (40% vs 76% in those with no partner). Among participants with no ADL limitations 562 (2.4%) received formal and 2183 (8.1%) received informal care. ADL limitation, although a key precursor of care use, is not the sole medical or socio-demographic reason adults might seek care. We identified other possible and plausible antecedents such as living arrangement, age, and SRH in participants with no ADL limitation, however, results of recursive partitioning remain exploratory and should be confirmed via future etiological analyses.

Methodologic strengths of this study

The methods used are novel and have allowed for a deep ‘dissection’ of intersecting characteristics that predict use of formal care, and revealed how social safety nets such as Medicare can correct for previously reported inequities. Ordinary regression models are focused on the estimation of risk in high-risk groups. Identification of risk factors in lower-risk populations (for example in those with no or only mild ADL limitations) is an advantage of recursive partitioning models. By recursive and stepwise subgroupings, combinations of these subgroups can be interpreted as social locations and hence provide a quantification of intersectionality [23]. Within-group heterogeneity usually is overlooked when epidemiological studies use traditional analytic approaches, while regression trees reveal the sources of this heterogeneity. Most existing studies with similar research questions have explored data via regression analyses [23, 50]. The few that used regression trees [14, 41, 51] did not conceptualise determinants of care use based on intersectionality theory and therefore did not find the complex interplay between potential associated factors. We have demonstrated the shortcomings of relying on diminished function (e.g limitations in ADL/IADL) as the sole indicator of need for care and have identified factors (for example age and living arrangement in those with no ADL limitations) that predict care use in higher functioning individuals. Both models, constructed following the CHAID algorithm, produced stable trees after 10-fold cross-validation with 4.6% risk of misclassification for formal care use and 10.7% risk of misclassification for informal care use.

Limitations

The CLSA follows a very robust sampling strategy and is representative of middle-aged and older Canadians living in the community. Our aim was to identify social and individual characteristics aligned with use of homecare and, hence, exclusion of those living in institutions seems like a minor issue. Institutional care, however, may act as a replacement for homecare, especially for older adults who live alone and have significant needs. We cannot determine whether this exclusion had any impact on our results. The survey data available does not differentiate those whose formal care is state funded from those paying privately for such care. However, Canadian data cited above show that 91% of nursing care (the key component of what we have defined as formal care) required no funding from the recipient, suggesting that although private funding sometimes pays for home support (e.g. cleaning, food preparation, etc) it rarely is needed to cover medical care [4]. Other excluded groups such as those living in First Nations communities, or those unable to speak French or English, may represent different care needs and use behaviours that were not captured in the dataset. Homecare use variables were self-reported and binary rather than quantified, and do not distinguish between, for example, occasional or almost full-time care. The data used for this study were collected prior to the COVID pandemic. Its impact on labour supply, availability of formal care and access to it may have shifted in ways that are as yet to be identified. Finally, the results of this study can only be interpreted as exploratory. Cross-sectional data cannot identify causality or the direction of any associations found. Confirmatory analyses using longitudinal data and possibly causal mediation techniques remain to be undertaken.

Conclusions

Using a novel methodology we found that the impact of inequities associated with social locations or individual isolation on use of formal homecare diminishes or even disappears when government-funded medical care is universally available. Nevertheless, identification of vulnerable subgroups, both those who did and those who did not receive care, highlights subpopulations that might benefit from increased clinical attention or public health interventions. Future research might explore the interplay between formal and informal homecare in settings with varying levels of state funded formal care. With aging of populations throughout the world, a deeper understanding of care use patterns will be essential for designing delivery systems that are not wasteful and that ensure access to high-quality care for all who need it.

Acknowledgments

This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). This research has been conducted using the CLSA Baseline Comprehensive Dataset version 4.0, under Application Number 170309. The CLSA is led by Drs. Parminder Raina, Christina Wolfson and Susan Kirkland.

References

  1. 1. Mah JC, Stevens SJ, Keefe JM, Rockwood K, Andrew MK. Social factors influencing utilization of home care in community-dwelling older adults: a scoping review. BMC Geriatr. 2021;21(1):145. pmid:33639856
  2. 2. Rodrigues R, Ilinca S, Schmidt AE. Income-rich and wealth-poor? The impact of measures of socio-economic status in the analysis of the distribution of long-term care use among older people. Health Econ. 2018;27(3):637–46. pmid:29024158
  3. 3. Sanmartin C, Hennessy D, Lu Y, Law MR. Trends in out-of-pocket health care expenditures in Canada, by household income, 1997 to 2009. Health Rep. 2014;25(4):13–7. pmid:24744043
  4. 4. Gilmour H. Formal home care use in Canada. Health Rep. 2018;29(9):3–9. pmid:30230520
  5. 5. UN. World Population Ageing 2020 Highlights: Living arrangements of older persons In: Department of Economic and Social Affairs PD, editor. 2020.
  6. 6. Zhang W, Sun H. Formal and informal care received by middle-aged and older adults with chronic conditions in Canada: CLSA data. PLoS One. 2020;15(7):e0235774. pmid:32634161
  7. 7. Li J, Song Y. Formal and Informal Care. In: Gu D, Dupre ME, editors. Encyclopedia of Gerontology and Population Aging. Cham: Springer International Publishing; 2019. p. 1–8.
  8. 8. Greene VL. Substitution between formally and informally provided care for the impaired elderly in the community. Med Care. 1983;21(6):609–19. pmid:6865520
  9. 9. Lyons KS, Zarit SH. Formal and informal support: the great divide. Int J Geriatr Psychiatry. 1999;14(3):183–92; discussion 92–6. pmid:10202661
  10. 10. Bertogg A, Strauss S. Spousal care-giving arrangements in Europe. The role of gender, socio-economic status and the welfare state. Ageing and Society. 2020;40(4): 735–58.
  11. 11. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36(1):1–10. pmid:7738325
  12. 12. Bassetti T, Rebba V. 2015. "Getting to the Roots of Long-Term Care Needs: A Regression Tree Analysis," 66167,. University Library of Munich, Germany. 2015.
  13. 13. Paraponaris A, Davin B, Verger P. Formal and informal care for disabled elderly living in the community: an appraisal of French care composition and costs. Eur J Health Econ. 2012;13(3):327–36. pmid:21400197
  14. 14. Penkunas MJ, Eom KY, Chan AW. Classification trees for identifying non-use of community-based long-term care services among older adults. Health Policy. 2017;121(10):1093–9. pmid:28942091
  15. 15. Rodríguez M. Use of informal and formal care among community dwelling dependent elderly in Spain. European Journal of Public Health. 2013;24(4):668–73. pmid:23813712
  16. 16. Van Houtven CH, Norton EC. Informal care and health care use of older adults. J Health Econ. 2004;23(6):1159–80. pmid:15556241
  17. 17. Blomberg J, Breeze E, Koskinen S, Martikainen P. Help from spouse and from children among older people with functional limitations: Comparison of England and Finland. Ageing and Society. 2012;32(6): 905–33.
  18. 18. Suanet B, Van Groenou M, Van Tilburg T. Informal and formal home-care use among older adults in Europe: Can cross-national differences be explained by societal context and composition? Ageing and Society. 2012;32(3): 491–515.
  19. 19. Kjær A, Siren A. Formal and informal care: Trajectories of home care use among Danish older adults. Ageing and Society. 2020;40(11):2495–518.
  20. 20. Andersson MA, Monin JK. Informal Care Networks in the Context of Multimorbidity: Size, Composition, and Associations With Recipient Psychological Well-Being. J Aging Health. 2018;30(4):641–64. pmid:28553797
  21. 21. Pinquart M, Sörensen S. Spouses, adult children, and children-in-law as caregivers of older adults: a meta-analytic comparison. Psychol Aging. 2011;26(1):1–14. pmid:21417538
  22. 22. Arber S, Ginn J. Gender differences in informal caring. Health & Social Care in the Community. 1995;3(1):19–31.
  23. 23. Bauer GR, Scheim AI. Advancing quantitative intersectionality research methods: Intracategorical and intercategorical approaches to shared and differential constructs. Soc Sci Med. 2019;226:260–2. pmid:30914246
  24. 24. Crenshaw K. Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics. The University of Chicago Legal Forum. 1989;140:139–67.
  25. 25. Hankivsky O, Reid C, Cormier R, Varcoe C, Clark N, Benoit C, et al. Exploring the promises of intersectionality for advancing women’s health research. Int J Equity Health. 2010;9:5. pmid:20181225
  26. 26. Phillips S, Vafaei A, Yu S, Rodrigues R, Ilinca S, Zolyomi E, et al. Systematic review of methods used to study the intersecting impact of sex and social locations on health outcomes. SSM Popul Health. 2020;12:100705. pmid:33319029
  27. 27. Bi Q, Goodman KE, Kaminsky J, Lessler J. What is Machine Learning? A Primer for the Epidemiologist. Am J Epidemiol. 2019;188(12):2222–39. pmid:31509183
  28. 28. Cairney J, Veldhuizen S, Vigod S, Streiner DL, Wade TJ, Kurdyak P. Exploring the social determinants of mental health service use using intersectionality theory and CART analysis. J Epidemiol Community Health. 2014;68(2):145–50. pmid:24098046
  29. 29. Andersen R, Newman JF. Societal and Individual Determinants of Medical Care Utilization in the United States. The Milbank Quarterly 2005;83(4).
  30. 30. Larsson K, Kåreholt I, Thorslund M. Care utilisation in the last years of life in Sweden: the effects of gender and marital status differ by type of care. Eur J Ageing. 2014;11(4):349–59. pmid:28804339
  31. 31. Shih CM, Wang YH, Liu LF, Wu JH. Profile of Long-Term Care Recipients Receiving Home and Community-Based Services and the Factors That Influence Utilization in Taiwan. Int J Environ Res Public Health. 2020;17(8). pmid:32294909
  32. 32. Yeboah-Korang A, Kleppinger A, Fortinsky RH. Racial and ethnic group variations in service use in a national sample of Medicare home health care patients with type 2 diabetes mellitus. J Am Geriatr Soc. 2011;59(6):1123–9. pmid:21649625
  33. 33. Raina P, Wolfson C, Kirkland S, Griffith LE, Balion C, Cossette B, et al. Cohort Profile: The Canadian Longitudinal Study on Aging (CLSA). Int J Epidemiol. 2019;48(6):1752–3j. pmid:31633757
  34. 34. Pong RW, Desmeules M, Lagacé C. Rural-urban disparities in health: how does Canada fare and how does Canada compare with Australia? Aust J Rural Health. 2009;17(1):58–64. pmid:19161503
  35. 35. Solé-Auró A, Crimmins EM. Who cares? A comparison of informal and formal care provision in Spain, England and the USA. Ageing Soc. 2014;34(3):495–517. pmid:24550574
  36. 36. Fillenbaum GG. Multidimensional Functional Assessment of Older Adults: The Duke Older Americans Resources and Services Procedures: Taylor & Francis; 2013.
  37. 37. Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale). Am J Prev Med. 1994;10(2):77–84. pmid:8037935
  38. 38. Gamache P, Hamel D, Blaser C. Material and social deprivation index: A summary–INSPQ Website; 2019.
  39. 39. Ma X. Using Classification and Regression Trees: A Practical Primer: Information Age Publishing; 2018.
  40. 40. Kass G. An Exploratory Technique for Investigating Large Quantities of Categorical Data. Journal of the Royal Statistical Society Series C (Applied Statistics). 1980;92(2):119–27.
  41. 41. Igarashi A, Ishibashi T, Shinozaki T, Yamamoto-Mitani N. Combinations of long-term care insurance services and associated factors in Japan: a classification tree model. BMC Health Serv Res. 2014;14:382. pmid:25209623
  42. 42. Su X, Azuero A, Cho J, Kvale E, Meneses KM, McNees MP. An introduction to tree-structured modeling with application to quality of life data. Nursing research. 2011;60(4):247–55. pmid:21720217
  43. 43. Chappell N, Blandford A. Informal and Formal Care: Exploring the Complementarity. Ageing and Society. 1991;11(3): 299–317.
  44. 44. T. M. Measuring social protection for long-term care. OECD Health Working Papers. 2017;93.
  45. 45. Schmidt AE. Analysing the importance of older people’s resources for the use of home care in a cash-for-care scheme: evidence from Vienna. Health Soc Care Community. 2017;25(2):514–26. pmid:26919130
  46. 46. Dahlberg L, Demack S, Bambra C. Age and gender of informal carers: a population-based study in the UK. Health Soc Care Community. 2007;15(5):439–45. pmid:17685989
  47. 47. Kahn J, McGill BS, Bianchi S. Help to Family and Friends: Are There Gender Differences at Older Ages? Journal of marriage and the family. 2011;73 1:77–92. pmid:21738263
  48. 48. Patterson SE, Margolis R. The Demography of Multigenerational Caregiving: A Critical Aspect of the Gendered Life Course. Socius. 2019;5:2378023119862737.
  49. 49. Steinbeisser K, Schwarzkopf L, Grill E, Schwettmann L, Peters A, Seidl H. Gender-linked determinants for utilization of long-term care in community-dwelling adults 65+ in Germany: Results from the population-based KORA-Age study. Exp Gerontol. 2021;153:111500. pmid:34332058
  50. 50. Floridi G, Carrino L, Glaser K. Socioeconomic Inequalities in Home-Care Use Across Regional Long-term Care Systems in Europe. J Gerontol B Psychol Sci Soc Sci. 2021;76(1):121–32. pmid:32996570
  51. 51. Nakabe T, Sasaki N, Uematsu H, Kunisawa S, Wimo A, Imanaka Y. Classification tree model of the personal economic burden of dementia care by related factors of both people with dementia and caregivers in Japan: a cross-sectional online survey. BMJ Open. 2019;9(7):e026733. pmid:31289069