PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • This study involved a cross-sectional analysis of baseline data from the Canadian Longitudinal Study on Aging (CLSA) [59,60]. Baseline data were used, as the CLSA has only recently been launched and follow-up data were, at the time the present study was conducted, not yet available. CLSA consists of two cohorts: The Comprehensive Cohort involves participants who were randomly selected within age/sex strata from among individuals residing within 25 km of a data collection site (or 50 km in lower density cities) in ten sites across Canada. These participants were interviewed in their own homes with computer-assisted interview instruments. They also came to data collection sites for additional computer-assisted interviews and comprehensive assessments (e.g., physical measures, biological samples). The Tracking Cohort consists of a randomly (within age/sex strata) selected sample from the 10 Canadian provinces that completed computer-assisted telephone interviews. Participant exclusion criteria were: could not communicate in one of the two national languages, English or French; cognitive impairment at time of contact; resident of the three territories; full-time member of the Canadian Armed Forces; resident in a long-term care institution; and living on Federal First Nations reserves or other First Nations settlements. CLSA participants provided written consent before participating in the study. Public access census data from 2016 were used to derive geographic variables. CLSA questionnaire data were linked to census data via the first three digits of participants’ postal code (Forward Sortation Area, FSA). Postal codes are used by Canada Post for the purpose of sorting and delivering mail, with FSAs representing geographic areas. As of 2011, there were 1638 FSAs in Canada [61]. The present study received ethics approval from the University of Manitoba’s Health Research Ethics Board. A total of 51,338 participants living in 1558 FSAs participated in the CLSA baseline. Of these, FSAs with less than 10 participants were excluded from analyses, as estimates of social isolation and loneliness are less meaningful for these FSAs. Our final sample included 48,330 participants aged 45 to 85 from 977 FSAs. Because of missing values on some variables, the unweighted sample size in the analyses for social isolation was 47,752; for loneliness the final sample size was 47,818. A comparison of FSAs that were included in the analyses to those excluded indicated that they were comparable (e.g., 51.0% versus 50.4% for the % women in the FSAs). CLSA participants were also similar. For example, 51% of CLSA participants in the included FSAs were female and 5.6% had a household income of less than $20,000, compared to 50.2% and 6.3%, respectively in the excluded FSAs. Social isolation has been defined in different ways in previous literature [1]. Conceptually, our definition was guided by the Convoy Model of Social Relationships, according to which individuals are surrounded by a series of social network ties that range from closest to less close [62]. Spouses tend to play a key role in people’s social networks, followed by children, siblings (or other relatives), and friends. Consistent with the finding that a variety of social network members play an important role in people’s lives [19,20], contact with these network members has been used to define social isolation in previous research [16–18]. A social isolation index was derived based on five sets of questions: 1) marital status (married or living in a common-law relationship; never married or never lived with a partner; divorced; separated; widowed); 2) living arrangements (number of people currently living in household); 3) when participants last got together with each of the following social network members living outside of their household: children, siblings, close friends, and neighbors (1 = within the last day or two; 2 = within the last week or two; 3 = within the past month; 4 = within the past 6 months; 5 = within the past year; 6 = more than 1 year ago); 4) retirement status (retired, working part-time/full-time); and, social participation in eight activities in the past 12 months (e.g., family or friendship based activities, church or religious activities, sports or physical activities, and educational and cultural activities; 1 = at least once a day; 2 = at least once a week; 3 = at least once a month; 4 = at least once a year; 5 = never)(see https://datapreview.clsa-elcv.ca/ for the entire questionnaire). The social participation questions were recoded such that response categories 4 and 5 were coded as 0 (i.e. less than once a month), and response categories 1–3 were coded as 1 (i.e. at least once a month or more often). The re-coded variables were then summed to create a new score ranging from 0–8. Similar to previous research [16–18], we allocated one point when each of the following conditions applied: 1) living alone and not married or in a common law relationship; 2) got together with friends or neighbours “within the past 6 months” or less frequently, or reported having no friends or neighbors; 3) got together with relatives/siblings “within the past 6 months” or less frequently, or reported having no relatives or siblings; 4) got together with children “within the past 6 months” or less frequently, or had no children. A fifth criterion was that one point was allocated if participants were retired and had little social participation (scores 0 or 1 on our re-coded social participation scale; i.e. they participated in none or only one of the activities at least once a month or more often). This resulted in a social isolation index ranging from 0–5, with higher scores reflecting greater social isolation. As we were interested in identifying socially isolated groups of adults and their associated characteristics, we subsequently dichotomized the social isolation index. There are no established cut-offs in the literature to define individuals who are socially isolated. For the present purposes, we classified individuals with scores 3–5 on the index as socially isolated (coded as 1) and those with scores 0–2 as not socially isolated (coded as 0). This cut-off was chosen, as it classifies people with at least half of the criteria that make up the social isolation index as being socially isolated. As contact with social network members is only measured in terms of in-person contact in the CLSA, the cut-off, in part, also ensures that individuals who may have had contact with some social network members via other means only (e.g., telephone, internet) are not identified as being socially isolated. For example, individuals with no family (relatives/siblings, children) living close enough to allow frequent direct contact would not be considered socially isolated, unless they also met one other criterion. Sensitivity analyses were also conducted with a 2–5 (vs. 0/1) cut-off. Given that CLSA contains two cohorts (Tracking and Comprehensive) we compared the percent socially isolated in each cohort. The (unweighted) percentage of participants identified as socially isolated using a 3+ cut-off on our social isolation index was relatively similar in the Tracking cohort, compared to the Comprehensive cohort (6.7% vs. 5.5%). As the CLSA baseline questionnaire does not contain a loneliness scale, a single-item loneliness question that is part of the CESD depression scale [63] was used. Questions focus on the past week and participant were asked: “How often did you feel lonely?” (1 = all of the time [5-7days]; 2 = occasionally [3–4 days]; 3 = some of the time [1–2 days]; 4 = rarely or never [less than 1 day]. Similar single-item measures are commonly used in the literature [1]. The item was dichotomized, with “all of the time” and “occasionally” responses considered lonely (coded as 1) and the remaining categories as not lonely (coded as 0). The (unweighted) percent of participants classified as being lonely was similar in the Tracking versus Comprehensive cohort (11.2% vs. 10.8%). Personal factors included: age, sex, education, household income, functional status, and chronic conditions. Age was categorized into four categories (ages 45–54, 55–64, 65–74, and 75–85). Sex was coded as 0 = women and 1 = men. Education was dichotomized as: 0 = secondary school or less, and 1 = at least some post-secondary education. Household income was included as an overall measure of the financial resources available to an individual. It was measured by asking participants to give the best estimate of the total household income received by all household members, from all sources, before taxes and deductions, in the past 12 months (1 = less than $20,000; 2 = $20,000 or more, but less than $50,000; 3 = $50,000 or more, but less than $100,000; 4 = $100,000 or more, but less than $150,000; and 5 = $150,000 or more). As a considerable number of participants did not answer the question, a “missing” category was also included in order not to lose these individuals from analyses. This also allowed us to determine if individuals who did not answer the question differed systematically from those who did. Functional status was assessed using the Older Americans’ Resources and Services (OARS) Multidimensional Functional Assessment Questionnaire [64]. The scale includes seven questions related to Activities of Daily Living (e.g., getting out of bed, dressing, and eating) and seven questions related to Instrumental Activities of Daily Living (e.g., using the telephone, shopping, and preparing meals). For each question, participants responded whether they can complete the task without help, with some help, or are completely unable to perform it. The items can be used to categorize individuals into: no functional impairment; mild impairment; moderate impairment; severe impairment; and total impairment. As most participants had no functional impairment, responses were dichotomized: 0 = no functional impairment”, and 1 = at least some functional impairment. Chronic conditions were measured with a list of 33 conditions, such as osteoarthritis, respiratory conditions, and cardiac/cardiovascular conditions, with participants asked if a doctor had diagnosed them with the condition. An index was created by summing affirmative responses. A rural/urban variable is available in the CLSA data. The variable was added by CLSA based on a Statistics Canada definition and postal code conversion file [65,66]. “Urban core” refers to census metropolitan areas (CMA) with a population of at least 100,000 (50,000 or more of which live in the core) or census agglomerations (CA) with a core population of at least 10,000. “Secondary core” refers to a population center within a CMA with at least 10,000 residents that was the core of a CA but has now been merged with an adjacent CMA. Urban and secondary cores were combined in the present study. “Urban fringe” refers to population centers within a CMA or CA that are not contiguous with the core or secondary core with fewer than 10,000 residents. “Urban population centres outside CMA and CA” are defined as settlements outside CMA and CA with a population of at least 1,000 and a population density of 400 persons or more per square kilometer. “Rural” is defined as areas within CMA or CA not classified as core or fringe, or areas not defined as population centres. A “not defined” category was also included for which no urban/rural information was available. Although the rural/urban variable was considered a geographic variable in this study, it should be noted that it was not measured at the FSA-level, unlike the other geographic variables. FSAs are not necessarily aligned with the Statistics Canada rural/urban definition. This means that some FSAs contain both rural and urban areas, as defined by Statistics Canada. In the analyses, rural/urban was treated as an individual-level variable. Other geographic-based variables were derived from 2016 public access census data for each FSA. We selected census variables that were similar to individual predictors of social isolation and loneliness, including: percent of women; percent of the population aged 65+; and percent of the population living alone. Census data further contain the percent of the population aged 65 or older with low income based on the after-tax low-income cut-offs (LICO-AT). Statistics Canada defines the LICO-AT as an after-tax income threshold below which a family is expected to spend a larger share of its income on food, shelter and clothing (20 percentage points more) than the average family [67]. The percent of the population whose first language was not one of the two official languages (English, French), was also included as a way to assess the ethnic diversity of areas. Data were analyzed using multilevel analyses, given the nested nature of the data (individuals within FSA) using SAS version 9.4. Given that social isolation and loneliness were defined as dichotomous outcomes, logistic regressions were conducted using Proc Glimmix. A series of analyses were conducted for the two outcome variables. First, in model 1, variables measured at the individual-level were entered into the regression model. This included personal characteristics, as well as the rural/urban variable which, as noted above, was not available at the FSA level; second, in model 2, variables measured at the FSA-level were added. This approach was taken for both the full sample and, subsequently, for women and men, respectively. For loneliness, all variables were included in the analyses; for social isolation, variables that were included in its definition (marital status and living alone) were excluded in the analyses, as was the FSA-level variable percent living alone, given that it was strongly correlated with living alone. Effects were evaluated for significance at p < .01.
rdf:type