Living arrangements are associated with people’s health behaviours. Living with families may influence people’s health behaviours through mutual interactions (14), but the direction could be complex. Contradictory findings have been found concerning the association between living arrangements and health behaviours, showing the complexity of this subject. Compared to those living with families, people who live alone may have higher likelihoods of eating unhealthily (15), smoking cigarettes, and consuming alcohol (16). They may also have worse physical and mental health (17), and lower levels of life satisfaction (18). However, some studies found no significant differences in some health dimensions between living alone versus living with families. For example, older adults living alone do not differ regarding time spent engaging in physical activities from their counterparts co-residing with families (19).

Canadians’ living arrangements have shifted remarkably over the past four to five decades, and are still in a consistent transition. Specifically, single-person households now make up the highest proportion of all household types (20). This transition is mainly due to compositional changes at the population level in Canada’s sex ratios, the rising levels of Canadians’ educational attainments, the increasing proportion of women’s labour force participation, people’s changing marriage and fertility behaviours, and its ageing population structure (17, 21). Under this context, however, to date, little scholarly attention has been paid to identifying whether living arrangements are associated with Canadians’ illicit drug use.

We apply the health lifestyle theory, which is widely used in sociology, to explore how essential others, usually family members, may have influenced people’s illicit drug use behaviours. As the main theoretical contributor, Cockerham (p.6) (22) defines health lifestyles as “collective patterns of health-related behaviour based on choices from options available to people according to their life chances.” Health lifestyle theory aims to explain how structural factors contribute to the production and reproduction of people’s health lifestyles. Specifically, Cockerham (23) points out that interactions and interplay between structural factors and people’s agency they learned and practised during past life experiences produce dispositions to act, which affect their daily actions and further shape health lifestyles. Structural factors proposed by Cockerham et al. (24) are factors at the levels of social institutions or conditions that exert influences on individuals’ health behaviours and health lifestyles, including social class, demographic background (e.g., age, gender), living conditions, and collectivities.

The current research does not address an individual’s agency but aims to investigate social structure as a potential influential facet to the production of health behaviours and health lifestyles. Our emphasis is to test living arrangements as a type of collectivity. From a theoretical lens, collectivities are “collections of actors linked together through particular social relationships, such as kinship, work, religion, and politics” (22). Living arrangements refer to whom an individual is living with. Living arrangements imply daily interactions and mutual influences among family members regarding health-promoting or health-harming behaviours and lifestyles. Norman and Ford (25) indicate that close ties with families or other essential others significantly increase people’s social involvement and decrease the possibility of deviant or unhealthy behaviours. For example, Tucker and Anders (26) contend that married couples are likely to promote changes in health behaviours for their partners, such as healthy eating and physical exercise. On the contrary, a recent Japanese study indicates that older men living alone are more likely to eat alone, which is associated with unhealthy eating and obesity (27).

Differences between men and women are also widely reported in terms of mutual influences on health behaviours within the household. Women are more likely to encourage healthy behaviours (28) and discourage unhealthy behaviours (28, 29), leading to men benefitting more from their female partners rather than the opposite. For example, Margolis (30) investigated the changes in American couples’ smoking behaviour when facing health shocks, and found that men are more likely to benefit from their partners rather than the other way around. We therefore test if any differences between men and women exist regarding the association between living arrangements and illicit drug use.

Taking living arrangements as a type of collectivity through the lens of the health lifestyle theory, our research aims to address two research questions. First, whether or not young, middle-aged, and older Canadians’ living arrangements (living alone vs. living with spouses/partners, children, both, or others) are associated with their illicit drug use, and if so, to what magnitude? The reason for dividing respondents into three age groups lies in the fact that living with parents, spouses/partners, children, or others may have different social, emotional, and health meanings depending on one’s age and different life stages (31, 32). To answer the first question, we propose Hypothesis 1 (H1) considering the complex associations between living arrangements and people’s health behaviours.

H1

Canadians living alone are more likely to use illicit drugs compared to those co-residing with core family members.

Second, are there any differences between men and women in the associations between living arrangements and illicit drug use? Prior studies indicated that men are more likely to receive positive influences from their female partners regarding changes in health behaviours (28, 30). Hypothesis 2 (H2) has thus been proposed in the current study to address whether sex difference exists in relevance to our core research interest.

H2

The difference between living alone and living with families regarding illicit drug use is more significant among women than men.

We use the 2015–2016 Canadian Community Health Survey (CCHS) Annual Component. Statistics Canada collected, administrated, and released the data. The CCHS is a series of repeated cross-sectional datasets focusing on the physical and mental health conditions of Canadian residents aged 12 and above. The 2015–2016 CCHS collected data on Canadians’ illicit drug use, living arrangements, demographic and socioeconomic backgrounds, and physical and mental health statuses, with an acceptable proportion of missing values in total. Such data richness ensures successful modelling to address research questions. The CCHS 2015 is also nationally representative (33), guaranteeing the generalizability of research findings.

This article focuses on Canadians aged 20 and above. The reason is that, according to CCHS 2015, Canadians aged 12–19 rarely lived by themselves (0.65%). We further divided Canadians aged 20 and above into three age groups: “young Canadians” or “young adults” who were aged 20–34 years, “middle-aged Canadians” or “middle-aged adults” aged 35–59 years, and “older Canadians” or “older adults” aged 60 and above. We deleted samples with missing values which are 3.22%, 3.84%, and 5.89% of the total sample size of the three age groups, respectively. The sample size of young Canadians is 19,091, and the sample sizes of middle-aged Canadians and older Canadians are 38,241 and 36,863, respectively. The final total sample size is 94,195.

Table 1 presents weighted sample characteristics with a 95% Confidence Interval (CI). The proportion of respondents living alone is 26.83% of older Canadians, the highest among all three, followed by middle-aged Canadians’ 13.17% and young Canadians’ 11.23%. Similarly, older Canadians have the highest proportion of living with spouses/partners (53.52%), which is about 32% higher than the proportion of middle-aged Canadians (21.04%) and 36% higher than that of young Canadians (17.47%) with statistical significance. Middle-aged Canadians have the largest proportion of living with spouses/partners and children (46.47%), which is more than twice higher compared to young Canadians (22.10%) and approximately six times higher than that of older adult Canadians (8.10%). These differences are statistically significant as indicated by their nonoverlapping 95% CIs. As for the proportion of living with children only, the middle-aged group has a significantly higher proportion (7.18%) than their young (2.89%) and older counterparts (2.93%). Lastly, the proportion of living in other types of arrangements is the highest among young adults (19.26%), which is about 7% higher in comparison with the middle-aged (12.14%) and about 10% higher than older adults (8.62%) with statistical significance. Thus, noticeably, that young, middle-aged, and older Canadians significantly differ in the distributions of living arrangement types, which demonstrates the reasonability of separate modelling for young, middle-aged, and older Canadians.

We applied two regression models. Binary logit regression was applied to predict whether using illicit drugs in the past 12 months for young and middle-aged adults because the dependent variable is dichotomously coded. But for predicting illicit drug use during the past year among older adults, the complementary log-log model was used because the proportion of illicit drug use is relatively small among the aged population (37). Our multivariable modelling logic includes two steps. In the first step, we began by adding the independent variable and all covariates to the model to identify the association between living arrangements and illicit drug use among young, middle-aged, and older Canadians. In the second step, we further added the interaction term between living arrangements and sex to the model to test possible sex differences in this vein.

The policy implications are apparent in the context that an increasing number of Canadians now live by themselves. We first recommend that the local communities and governments emphasize solo-living Canadians who are likely to be put into a risky situation regarding using illicit drugs, especially for young and middle-aged men.

Additionally, cohort differences should also be taken into consideration because the younger cohorts in this study, or the “Generation Y” (also known as Millennials), who were born in the 1980s and the first half of the 1990s, have significantly higher proportions and likelihoods of using illicit drugs in the past 12 months in comparison to the middle-aged and older cohorts. Importantly, if these young adults lack significant co-residing members (spouses/partners and children) who may play a positive role as a “supervisor” on their health behaviours and health lifestyles within the household, they have an even higher risk of engaging in these behaviours.

This study has a couple of limitations to report. First, the main limitation is that the dependent variable—whether a respondent used illicit drugs during the past year—is broad. It remains unclear what types of illicit drugs respondents have used and their frequency.

Second, endogeneity issues remain. Causality has not been identified due to relevant data limitations. However, two-way causality may exist. Specifically, using illicit drugs may have contributed to higher risks of marital disruption, which have led to the shifts in respondents’ living arrangements. Tracking people’s living arrangement transitions may help in addressing this issue.

Third, respondents’ interactions with families (within households) and peers (outside households) are unknown. However, stronger interactions within households may increase the likelihood of respondents’ recent illicit drug use if their family members were drug users. Similarly, having frequent interactions with peers who perform deviant behaviours significantly predict young adults’ drug use behaviours (25). Therefore, respondents’ interactions with families and peers deserve more exploration when investigating their illicit drug use possibilities.

The last limitation is that unattached respondents living with others were coded into the category of “other types.” Those unattached respondents have thus not been analysed in detail. Closer attention should address who they are and the underlying reasons for their illicit drug use, since this subpopulation reported high proportions of doing so, especially among the young and middle-aged.

Our research is the pioneer in investigating the multifaceted associations between living arrangements and Canadians’ recent illicit drug use behaviour. According to our findings, the first crucial take-home message demonstrates that living alone can be a risk factor for Canadians’ use of illicit drugs. Second, importantly, children and spouses/partners may have positively prevented young and older Canadians from using illicit drugs within the household. Living with spouses/partners works the most positively for middle-aged Canadians. Lastly, differences owing to sex have also been found regarding illicit drug use, in that living with families may have a more positive effect on young and middle-aged women compared to men. Policy recommendations are proposed accordingly that more health-related support should be provided to Canadians living by themselves.