Health Reports
Transitions to long-term and residential care among older Canadians

by Rochelle Garner, Peter Tanuseputro, Douglas G. Manuel and Claudia Sanmartin

Release date: May 16, 2018

In 2016, 16.9% of Canadians were aged 65 years or older, and 2.2% were aged 85 years or older, representing a 20.0% increase in these age groups since 2011.Note 1 The proportion of the Canadian population aged 65 years and older is expected to increase to 20.0% by 2024.Note 2 These demographic shifts raise concerns about the future need for nursing home (NH) care,Note 3Note 4Note 5 because age is a strong predictor of admission to an NH.Note 6Note 7Note 8 According to the 2016 Census, 6.8% of Canadians aged 65 years and older were living in an NH or residence for senior citizens (hereafter referred to as a seniors’ residence, SR): this proportion jumps to 30.0% among Canadians aged 85 years and older.Note 9Note 10

Published estimates of the future need for NH care in Canada typically rely on population projections of age and sex.Note 11Note 12 However, ratio-based approaches to the capacity planning of long-term care often over- or underestimate the number of beds (or units) needed to meet demand.Note 13 Furthermore, transitions to long-term care are associated with other factors, such as physical and cognitive limitations, acute health events, social support, household composition, and income—yet these factors are often not considered in projections.Note 14Note 15Note 16Note 17Note 18 While some Canadian studies have been conducted on population-based predictors for NH care,Note 7Note 19Note 20 few have considered a broad range of potential predictors, chiefly because of a lack of data on the range of factors reflected in the Canadian population. Furthermore, few studies have considered other competing outcomes, such as transitions to retirement homes, transitions to supportive living or mortality. Evidence suggests that controlling for mortality is important when estimating the potential for NH entry.Note 21

In Canada, NHs typically offer the highest level of support, and some are subsidized by publicly funded health care. The number of NH beds are limited and waitlists are often long. SRs typically provide less intensive services than NHs and are generally paid for out-of-pocket. Individuals with health care needs and their families base long-term care decisions on their preferences, level of need, ability to pay, and the availability of a bed or unit. Individuals who do not qualify for, who cannot afford, or who opt not to access NH or SR care may rely on community supports and/or care from family and friends to remain at home.

Population health surveys, such as Statistics Canada’s Canadian Community Health Survey (CCHS), include a range of questions on the health and health-related behaviours of Canadians. However, population health surveys typically exclude individuals living in facilities such as NHs and SRs, and are usually cross-sectional in nature. Other data sources, such as the Canadian Census of Population, capture less detailed health information than surveys, but provide a broader representation of the whole population. The 2011 Census enumerated individuals living in NHs and SRs among residents of collective dwellings. According to the 2011 Census, an NH was defined as a facility that provides 24/7 professional health monitoring and skilled nursing care to residents, often individuals who are elderly and not independent in most activities of daily living.Note 22 Assisted or supportive living, rest homes, retirement residences, and other facilities that provided personal support services but did not provide continuing health care services were classified in the 2011 Census as SRs.Note 23

By linking CCHS respondents to the 2011 Census, it was possible to identify individuals who moved from their private residence at the time of the CCHS interview to an NH or SR by Census Day. The purpose of this study is to estimate how a range of demographic, health and socioeconomic factors among older Canadians are associated with their transition from living in a private dwelling to living in an NH or SR.

Methods

Record linkage data sources

This study is part of a larger record linkage project that combined information from three different data sources: (1) the CCHS, (2) the Canadian Mortality Database (CMDB), and (3) the 2011 Census. The Executive Management Board of Statistics Canada approved this linkage projectNote 23 and the Directive on Record Linkage governs the use of the linked data.Note 24

The CCHS is a national, cross-sectional, population-based survey. It measures the health, behaviour, health care use and sociodemographic characteristics of the non-institutionalized household population aged 12 years or older. The survey excludes people living on reserves and other Aboriginal settlements in the provinces, full-time members of the Canadian Forces, the institutionalized population, and people living in selected Quebec health regions. Altogether, these exclusions represent less than 3% of the target population. The CCHS was conducted biennially between 2000 and 2006, then annually starting in 2007. Additional details regarding the CCHS sampling strategy and survey content are available on the Statistics Canada website.Note 25 Individuals who responded to the CCHS between 2000 and 2011 and agreed to share and link their data (n=701,877) were eligible for linkage with the other data sources.

The CMDB is a census of all deaths registered in Canada. It includes information such as cause and date of death, as well as the name, date of birth, and postal code of the deceased at the time of death. Deaths recorded in the CMDB between January 1, 2000, and December 31, 2011, among individuals aged 12 years or older (n=2.77 million) were eligible for record linkage in this study.

The 2011 Census of Population is an enumeration of the Canadian population on Census Day (May 10, 2011; n=33.5 million). The census provides individual, family and household-level information. In this study, it was used primarily to determine an individual’s place of residence.

Record linkage methodology

Record linkage was conducted in two steps. First, respondents to CCHS cycles from 2000 through 2011 (n=701,877) were linked to the CMDB using G-Link, a SAS-based record linkage software developed at Statistics Canada, and a probabilistic linkage methodology based on the Fellegi-Sunter theory of record linkage.Note 26Note 27 CCHS respondents were linked to death records based on given name, last name, date of birth, postal code and sex. Additional information from tax files was used to enhance the linkage with alternative postal codes and names (e.g., maiden name, father’s name).Note 28 Overall, 5.3% of CCHS respondents were linked to a mortality record. More information on this linkage is available elsewhere.Note 29

In the second step, the same CCHS respondents were linked to the 2011 Census. This linkage followed a hierarchical, deterministic, exact-match approach that compared the following linkage keys across the files: social insurance number (SIN), name (first and last), date of birth, postal code, and telephone number. This approach maximizes the discriminatory power of the linking variables and minimizes the influence of errors and missing values.Note 30 Separate linkages of the CCHS and the census to the tax data provided the SIN as a linking variable. Overall, 80.9% of CCHS respondents were linked to a unique individual in the census. Most links (82.6%) were created using the SIN. Among individuals who were not linked, 26.5% were known to have died. More information is available in an internal report.Note 31

Respecting respondent privacy

Statistics Canada ensures respondent privacy during the linkage process and subsequent use of linked files. Only employees directly involved in the linkage process had access to the unique identifying information required for linkage, such as names. Health-related information was not accessible to these individuals. Once the data linkage process was completed, all identifying information was removed from the analytical file.

Study sample

The sample for this study includes respondents from three CCHS survey cycles (Cycle 3.1, 2005/2006; Cycle 4.1, 2007/2008; and CCHS-Healthy Aging, 2008/2009) who agreed to share and link their data, who were believed to be alive and aged 60 years or older on Census Day 2011, and who were successfully linked to a unique census individual living in a private dwelling, NH or SR (see “Outcome Measure”). The final analytic sample included 81,411 individuals.

Outcome measure

The outcome measure is the dwelling place of an individual on Census Day. The census classifies dwellings as either private or collective. This study is interested in collective dwellings classified as an NH or SR. A key difference between an NH and an SR is the level of care provided, which is reflected in the collective dwelling classification used by the census.

Among individuals living in a private dwelling on Census Day, the majority (78.0%) were in the same living arrangement as at the time of the CCHS interview. Living arrangement at the time of the CCHS interview was based on the household relationship matrix, whereas living arrangement at Census Day was based on census family status. Respondents who changed living arrangements were classified as living in a private dwelling with additional family (PDAF) if they experienced the following changes: (1) lived alone at the time of the CCHS interview, but were living with other individuals at the time of the census; (2) did not live with a spouse/partner or children at the time of the CCHS interview, but were doing so at the time of the census; or (3) lived with their spouse/partner and no children at the time of the CCHS interview, but were living with their children (with or without their spouse present) at the time of the census. While family members are often the primary source of informal care,Note 32 the classification used in this study is only a proxy measure since the reason for changing living arrangements cannot be determined.

Covariate measures

Marital status and change in status

Marital status at the time of the CCHS interview and at the census were each dichotomized as married (including common-law) and not married (including separated, divorced, widowed and never married). Changes in marital status between the CCHS interview and Census Day were classified as: (1) remained married (i.e., married at the time of the CCHS interview and on Census Day), (2) lost spouse (i.e., married at the time of the CCHS interview but not married on Census Day), and (3) not married at the time of the CCHS interview (regardless of marital status on Census Day). Most individuals who lost a spouse were widowed by Census Day (78.1% of women compared with 60.5% of men) rather than divorced or separated. Among individuals who were not married at the time of the CCHS interview, a small proportion of women (3.2%) and men (9.0%) were married or living common-law on Census Day. Unfortunately, because of small cell counts, individuals who lost a spouse or were not married at the time of the CCHS interview could not be further disaggregated in analyses.

Sociodemographic and economic measures

To provide a common metric for the three cycles of the CCHS, age was calculated as of Census Day 2011 based on the date of birth provided at the CCHS interview. All other characteristics refer to respondents’ statuses at the time of the CCHS interview. Household income was divided by Statistics Canada’s low income cut-off (LICO) corresponding to the respondent’s household and community size.Note 33 These adjusted household income ratios were subsequently divided into quintiles at the provincial level, with a missing category included to retain residents of the territories (which have no LICO), as well as respondents who did not report household income (14.9%). Home ownership was a dichotomous variable that distinguished respondents living in a home owned by a member of the household from respondents living in a home that was rented. Living arrangements were dichotomized as respondents living alone and respondents not living alone. For province of residence, individuals living in the Atlantic provinces (i.e., Nova Scotia, New Brunswick, Prince Edward Island, and Newfoundland and Labrador) were grouped because of small sample sizes in these regions. Place of residence was also classified as either urban or rural based on postal code and census geography. Lastly, individuals were characterized as either immigrant or Canadian-born.

Health measures

Several measures of health status were considered. Each individual rated their general and mental health on five-point scales ranging from “excellent” to “poor.” Respondents also rated their health at the time of the interview compared with their health a year prior on a five-point scale ranging from “much better than a year ago” to “much worse than a year ago.” Furthermore, respondents were asked whether they had ever been diagnosed by a doctor with any of the following chronic conditions lasting more than six months: asthma, arthritis, back problems (not arthritis), high blood pressure, emphysema, chronic obstructive pulmonary disorder, diabetes, heart disease, cancer, ulcers, the effects of a stroke, urinary incontinence, bowel disease, Alzheimer’s disease or other dementia (referred to henceforth as dementia), a mood disorder, or an anxiety disorder.

Smoking status classified respondents as never smokers, former smokers or current smokers. The self-reported height and weight of individuals were used to derive their body mass index (BMI), which was subsequently categorized as underweight (BMI<18), acceptable weight (18≤BMI<25), overweight (25≤BMI<30) and obese (BMI≥30). Respondents were also asked if they had been a patient overnight in a hospital, an NH or a convalescent home in the year prior to the interview.

Analysis

The association between the characteristics of CCHS respondents and their place of residence on Census Day was examined crudely and after adjusting for age, separately for men and women. Sex-specific generalized multinomial logistic regression models were conducted to examine the association between the characteristics and dwelling location (private dwelling [reference group], PDAF, NH, or SR) of respondents. Because of the large number of potential covariates, only factors that were significantly associated (p<0.05) with dwelling location after age adjustment (results not shown) were considered as potential covariates in final models. Only covariates that were statistically significant (p<0.05) for one of the non-reference outcomes were retained in final sex-specific regression models.

All analyses were weighted using CCHS survey weights adjusted for respondents’ consent to link and share their information. Variation estimation was conducted using the bootstrap technique. Analyses were run in SAS-callable SUDAAN (version 11.0).

Results

In the analytic sample of CCHS respondents linked to the census, 1.4% of respondents were living in an NH at the time of the census, 1.2% of respondents were living in an SR, and 6.6% of respondents were living in a PDAF. Women were more likely than men to be living in an NH (1.8% of women vs. 0.9% of men), an SR (1.7% of women vs. 0.7% of men) or a PDAF (6.8% of women vs. 6.3% of men; Table 1 ). The proportion of respondents living in an NH or SR was relatively insignificant until respondents were aged 75 or older, after which living in an NH or SR became more prevalent (Figure 1 ).

Compared with individuals still living in a private dwelling on Census Day 2011, individuals living in an NH were older, were more likely to be unmarried at the time of the CCHS interview or to have lost their partner between the CCHS interview and the census, were in poorer health, and were more likely to have been born in Canada (Table 1 ). A similar pattern was also seen for individuals living in SRs. Fewer differences were found between individuals living in a PDAF and those in private dwellings. Individuals living in PDAFs were more likely to be unmarried and living alone at the time of the CCHS interview than those living in other private dwellings: the latter difference may result from the definition of PDAF.

Regression results

Final regression models for women and men are shown in Tables 2 and 3, respectively. On Census Day, individuals were significantly more likely to live in an NH than a private dwelling if they: had lost their spouse by Census Day (OR=4.3 for women, OR=3.7 for men) or were not married at their CCHS interview (OR=2.0 for women, OR=2.4 for men), compared with those who remained married; did not own their dwelling (OR=2.1 for women, OR=2.3 for men), compared with those who did; reported spending time in a hospital or a convalescent home in the year prior to the CCHS interview (OR=1.8 for women, OR=1.9 for men); or had been diagnosed with dementia (OR=6.7 for women, OR=6.2 for men).

Decreased self-perceived general health and mental health also significantly increased the likelihood of an individual living in an NH, showing a gradient in the effect across levels of each measure (Tables 2 and 3). In addition, living alone significantly increased the odds of women living in an NH (OR=1.5), as did having diabetes (OR=1.5), urinary incontinence (OR=1.3) or a diagnosed mood disorder (OR=1.7, Table 2). For men, these factors were not significantly associated with living in an NH (Table 3). Furthermore, immigrant women (OR=0.7) were significantly less likely to live in an NH compared with Canadian-born women (Table 2).

An individual was significantly more likely to live in an SR on Census Day if they: lost their spouse between the CCHS interview and Census Day (OR=4.2 for women, OR=3.5 for men), compared with those who remained married; did not own their home (OR=2.6 for women, OR=2.9 for men), compared with those who did; had diabetes (OR=1.4 for women, OR=1.8 for men); or lived in Quebec (OR=2.3 for women, OR=3.1 for men) or Alberta (OR=3.0 for women, OR=2.6 for men), compared with respondents from Ontario (Tables 2 and 3).

Men who were not married at the time of the CCHS were also significantly more likely to live in an SR on Census Day (OR=2.7, Table 3); this association was not significant for women (Table 2). Among women, a diagnosed mood disorder (OR=1.8) or urinary incontinence (OR=1.3) was associated with higher odds of living in an SR (Table 2 ). Being hospitalized or in a convalescent home in the year prior to the CCHS interview was associated with higher odds of living in an SR among men (OR=1.8, Table 3); this association was not significant among women (Table 2).

Compared with living in a private dwelling, few characteristics were significantly associated with living in a PDAF on Census Day. Although individuals who were living alone at the time of the CCHS interview were significantly more likely to live in a PDAF on Census Day compared with individuals who lived with others (OR=8.3 for women, Table 2; OR=16.4 for men, Table 3), this is an association by definition, since individuals who lived alone at the time of the CCHS interview had a greater chance of gaining family members than individuals who lived with others. Beyond living arrangement, individuals who were not married at the time of the interview were significantly less likely to live in a PDAF (OR=0.5 for women, Table 2; OR=0.3 for men, Table 3), compared with those who remained married. Among men, losing a spouse also significantly decreased the likelihood of living in a PDAF (OR=0.6, Table 3). Women with very good self-perceived health were significantly less likely (OR=0.8, Table 2) than women with excellent self-perceived health to live in a PDAF, whereas women with good self-perceived mental health relative to women with excellent self-perceived mental health had increased odds (OR=1.2, Table 2 ). Men living in British Columbia (OR=1.4, Table 3) were more likely to live in a PDAF compared with men living in Ontario.

Discussion

The availability of uniquely linked, population-based health survey and census data enabled a comprehensive, national look at factors associated with older Canadians’ transitions from living in a private dwelling to living in an NH or SR. This study simultaneously examined multiple settings that provide support, including SRs and PDAFs, while also accounting for mortality.

As expected, a diagnosis of dementia was strongly associated with transitions to NHs. Among women, it also significantly increased the odds of living in an SR. These findings are supported by existing evidence, both Canadian and international.Note 16Note 19Note 21 Current estimates suggest that over 66% of NH residents in Canada have a diagnosis of dementia.Note 34 Findings from this study contribute to our understanding of the role of dementia in the transitions of individuals to NHs and other supportive settings. Consistent with the literature, findings from this study show that other chronic conditions, namely diabetes, urinary incontinence and mood disorders among women, were also significantly associated with transitions to NHs.Note 20Note 35 Sub-optimal mental health also significantly increased the odds of both men and women living in an NH.

This study also found that losing a spouse was a significant predictor of an individual transitioning to either an NH or an SR. Compared with respondents who remained married, respondents who lost their spouse had more than four times the odds of living in an NH or SR on Census Day. Similar findings have been observed in international studies.Note 36Note 37

The strength of this study also lies in the ability to identify individuals living in PDAFs. In the study, approximately 7% of individuals were living in a PDAF, more than the proportion of individuals living in NHs and SRs combined. With the current focus on “aging in place,” alternate living arrangements are becoming more common. According to the 2016 Census, 11.6% of Canadians aged 65 years and older were living in private dwellings with people other than a spouse or child.Note 38 The high cost of SRs and NHs, along with the limited number of available beds, limited funding for residential care and rising out-of-pocket costs for individuals, may mean that alternate living arrangements will become necessary or preferable in the future, particularly given the progressive aging of the Canadian population.

This study also shows the protective effect of immigration status: immigrant respondents were much less likely to transition to an NH or SR by Census Day than Canadian-born respondents. According to the 2011 National Household Survey, immigrant seniors who had been in Canada for a relatively short time were both less likely to live alone and more likely to live in multigenerational households than Canadian-born seniors or immigrant seniors who had lived in the country for a longer period of time.Note 39 Furthermore, it is traditional among certain immigrant groups for elderly people to live with their children or other relatives.Note 40 This may explain why immigrant CCHS respondents were less likely to move into an SR or NH by the study follow-up. Given that future cohorts of Canadian seniors are likely to be more ethnoculturally diverse than current seniors,Note 41 the effect of ethnicity and immigrant status may significantly influence residential preferences in the future.

Certain interprovincial differences related to the likelihood of individuals transitioning to an SR were also highlighted in this study. Respondents from Alberta and Quebec were more likely than those from Ontario to have moved to an SR by Census Day 2011; however, there were few interprovincial differences related to NH entry. According to the Seniors’ Housing Survey conducted by the Canada Mortgage and Housing Corporation, Quebec has the largest pool of SR spaces in Canada, the lowest average rents for such spaces, and a high degree of targeted marketing in the seniors’ housing market.Note 42 The number of retirement spaces in Alberta has grown over time, increasing 5.9% from 2009 to 2010.Note 43 This number has continued to rise: in Alberta, the increase in the number of retirement spaces in 2015 outpaced the increase in the number of seniors aged 75 years and older.Note 44 The high availability of retirement spaces in Alberta and Quebec may contribute to the interprovincial differences seen in this study.

Limitations

Although there was a high overall linkage rate (90.2%), linkage rates were significantly lower among older respondents, who are most likely to reside in NHs or SRs. According to the 2011 Census, 3.1% of Canadians aged 60 years and older lived in an NH and 1.9% lived in an SR. In this study, the proportion of individuals living in an SR (1.2%) is comparable with census findings. In contrast, the proportion of the analytic sample residing in NHs is notably smaller in this study than in census findings. One reason for this may be that the study sample used individuals known to be living in private dwellings, thereby excluding individuals who were already living in an SR or NH.

This study was only able to examine place of residence at a single point in time, Census Day 2011. It is likely that some CCHS respondents who died before Census Day may have been living in an NH or SR at the time of their death. However, because these transitions could not be identified, they are missing from the analysis. Future studies may wish to examine data with a finer degree of precision in terms of entry dates into long-term and residential care.

Similarly, changes in respondent characteristics (apart from marital status) are also not included in this study. The effect of changes in health status after the interview, or of acute events occurring between the CCHS interview and the census, cannot be discerned from this analysis. Furthermore, characteristics were limited to self-reported measures. Use of direct measurements and other sources of information may have yielded different estimates.

Finally, although including PDAF as a dwelling variable was a strength and unique trait of this study, its measurement is imperfect. It is unknown whether individuals living in PDAFs were receiving additional support from family members or whether, in the case of adult children moving back into the family home, the respondent was the one offering additional assistance, financially or otherwise. Further examination of alternate private dwelling living arrangements should use other data sources to better understand this growing, and potentially important, residential transition.

Conclusion

Although age is a strong predictor of an individual transitioning from a private dwelling to an SR or NH, other factors such as loss of a marital partner or diagnosis of a chronic condition are also predictive of such residential transitions. Future projections of the demand and need for assisted living and institutional care could consider the influence of these other factors in their calculations.

References
Date modified: