Abstract
Background
Social connections impact morbidity and mortality, yet their influence on waist circumference (WC) and potential gender differences remains unclear.
Methods
Prospective study of the Canadian Longitudinal Study on Aging (CLSA, 2011–2021) on changes in marital status, living arrangements, and social participation (baseline to follow-up 1) in relation to changes in WC (follow-up 1 to follow-up 2) used stratified, multivariable random coefficient regression with post-estimated predicted means. Sample sizes varied: 13,941 participants for marital and living arrangement transitions and 13,921 for changes in social participation.
Results
Mean WC declined over time in aging cisgender heterosexual women and men in Canada. We found genders-specific associations between changes in different social connections changes in mean WC. Among women, decreased social participation was significantly associated with changes in WC (0.47 cm; 95 confidence interval [CI], 0.11 to 0.83), compared to women with stable social participation. Among men, becoming divorced was significantly associated with changes in WC (1.45 cm; 95% CI, 0.14 to 2.75), compared to men remained partnered. Women who decreased their social participation had a predicted mean WC of 87.61 cm (95% CI, 87.14 to 88.09), and men who became divorced had a predicted mean WC of 100.54 cm (95% CI, 99.17 to 101.91). Results for living arrangement changes were not significant, though patterned appeared gendered. All social connection transitions were associated with clinically meaningful predicted mean WC, placing both women and men with altered connections in the ‘high risk’ category for poor health outcomes.
Conclusion
Decreased social participation increased WC among cisgender heterosexual women, whereas becoming divorced increased WC among cisgender heterosexual men in Canada.
Keywords: Interpersonal relations, Waist circumference, Visceral adiposity, Aging, Gender, Longitudinal, Canadian Longitudinal Study on Aging
INTRODUCTION
About one in five Canadians are aged 65 years and older1 and a quarter of the population will be older adults by 2036 with a majority being women.2,3 Overweight and obesity affect about two-thirds of older adults in Canada.4 Moreover, visceral adiposity has clinical significance as a critical component of metabolic syndrome (subclinical heart disease) and also predicts multiple chronic diseases, including cardiovascular diseases—the leading cause of early death among Canadian women.5,6 Social isolation as a composite measure is associated with higher odds of both visceral adiposity and general obesity, especially in middle- and older-aged women.7 Older adults are known to be especially vulnerable to social isolation8 and often lack structural social connections, such as an intimate partner, a coresident, or social contacts.8 Thus, it is a public health and policy priority for health aging and obesity prevention to understand the role of adverse changes in the social context of aging adults.
Some literature on structured social connections indicates that ‘unmarried’ older adults typically have worse health and poorer survival than married counterparts.9 But this literature does not differentiate this heterogeneous group of ‘unmarried’ individuals (single, widowed, or divorced/separated) who have diverse social resources that contribute to healthy aging.10 Cohabitation is also an important factor in older adults’ wellbeing, and living alone could negatively affect their lifestyle and health.11 Moreover, participation in social activities can also influence older adults’ health-promoting behaviours and overall wellbeing.7 Importantly, each social connection appears to matter for obesity: older women who were single, widowed, or divorced/separated, who lived alone, or had limited social participation had worse anthropometric outcomes but only loneliving older men had reduced likelihood of obesity, compared to their socially connected counterparts.7 However, the size and nature of structured social connections changes as women and men age,12 and different social connection transitions could have different impacts on health and obesity.13
Published work on changes in the social context of aging adults and obesity has been limited by several factors. A handful of obesity studies examined the role of marital transitions only, all studies were based in the United Statement and, with three exceptions,14-16 did not include both women and men.17 This evidence review illustrated that transitions into and out of marriage have opposing metabolic consequences among aging adults.17 Great scope therefore exists to examine different types of social connections as broader determinants of obesity, and especially to understand gender differences to address women’s health inequities. Thus, in a departure from existing knowledge, this prospective study aimed to examine genderspecific association of three types of social connection transitions and measured waist circumference (WC) over 6 years in middleage and older adults in Canada.
METHODS
Study population
We used three waves of data (2011 to 2021) from 30,097 predominantly White, non-institutionalized, middle- and older-aged participants (45 years and above) in the Canadian Longitudinal Study on Aging (CLSA) comprehensive cohort18 that included English- or French-speaking people living within 25 to 50 km of 11 data collection sites;18 it excluded residents of three territories, Indian reserves or Crown lands, institutions, remote regions, and full-time members of the Canadian Forces.18 Our eligible sample of participants with information on each of the social connections, measured WC and covariables included: 13,941 participants for marital and living arrangement transitions and 13,921 for changes in social participation (Supplementary Fig. 1). Participants in the CLSA cohort voluntarily enrolled in the study and provided written informed consent.18 The secondary data analysis received approval from the (blinded for review) Behavioural Research Ethics Board of The University of British Columbia (certificate # H19-00971).
Independent variables
Transitions in marital status, living arrangement, and social participation were assessed between baseline (2011 to 2015) and follow-up 1 (FU1; 2015 to 2018). The marital transition variable had six categories: remained partnered (reference), remained non-partnered (single, widowed, or divorced/separated), became widowed, became divorced/separated, became partnered, and uncertain transitions (Supplementary Table 1). The accommodation transition variable included four categories: remained co-living (reference), remained lone-living, became lone-living, and became co-living. Participants were classified as stable (reference), increased or decreased social participation using a summary score of 0 to 8 different social activities (once/month or more), baseline subtracted from FU1. Details are given in Supplementary Methods.
Dependent variable
Our outcome of interest was change in mean WC between FU1 and follow-up 2 (FU2). We modeled WC at FU2, adjusting for WC at FU1 and we used the CLSA data on WC in centimeters (cm) that was clinically assessed halfway between the last rib and the iliac crest bone.18 For descriptive analysis, we used a binary variable for very high risk WC defined as ≥88 cm for women and ≥102 cm for men.19
Covariables
Covariable selection was informed by a directed acyclic graph (DAG) based on existing literature,9,14-16,20-22 biological plausibility and content knowledge of known confounders to reduce variable selection bias (Supplementary Figs. 2-4). DAGitty software version 3.0 (Nijmegen) was used to create the DAGs. Details of the eight individual-level and three province-level covariables included in main models are given in Supplementary Methods.
Sex and gender considerations
The CLSA asked participations at baseline ‘are you male or female?,’ which reflects responses based on sex assigned at birth, gender identity or both. The separate CLSA ‘gender identity’ question at FU1 (male, female, transgender, transman, transwoman, genderqueer, and other) was highly correlated with the male/female variable (Cramer’s V=0.998). We therefore used baseline male/female to stratify models, and interpreted results for (cis)women and (cis) men through a lens of gender as a social construct: gender roles/norms, gender relations, and institutionalized gender.23
Analytic approach
Descriptive statistics (means±standard error [SE] or frequencies and proportions) summarised socio-demographic characteristics and crude WC changes and visceral adiposity across three social connection transition variables. Survey inflation weights from CLSA were used for descriptive analysis to generate estimates representative of Canadians aged 45 to 85 living near CLSA data collection sites. The a priori strategy for main analyses was to examine genderspecific prospective associations of change in all three types of social connection in relation to change in WC over 6-year study period (Supplementary Fig. 5). A series of stratified, multivariable mixedeffects linear regression models were conditioned on key individual (age, age squared, WC at FU1, study duration, education, household income, home-ownership, smoking, and chronic conditions [including anxiety and depression]) and provincial covariables (provincial gross domestic product, % spending on social protection, and obesity prevalence). Mixed regression incorporated the natural nesting of CLSA data and used a two-level (individuals nested within geographical areas) random intercept model that assumed a linear trend in the coefficients of the outcome over the study period, with a fixed effect for time (a month variable). Post-estimation calculated predicted means (95% confidence interval [CI]) of WC at FU2 for all transition categories of each social connection variable in women separately from men (STATA’s ‘margins’ command), with results graphed for ease of interpretation (‘marginsplot’). Postcalculated omnibus test evaluated a statistically significant overall effect of different social tie transitions (STATS’s ‘testparm’). Main results are reported as beta-coefficients and 95% CIs. Analysis was conducted using STATA/MP 18 software (StataCorp LP). Final analytic samples for women were 7,328 to 7,341, and for men were 6,593 to 6,601, depending on the exposure studied.
Sensitivity analyses tested the robustness of main results to the inclusion of other potential confounders (alcohol frequency, physical activity, sleep deprivation, urban living, body mass index [BMI], and fruits and vegetables intake).24 For women, models further adjusted for reproductive variables (pregnancy, menopause status, and hormone replacement therapy usage) given their known influence on adiposity in later life.25 We also considered the independent effects of each transition variable by mutually adjusting each model for other baseline social connections (e.g., living arrangement and social participation score for marital transition).
RESULTS
Over time, women and men in the CLSA had a decline in mean WC: baseline mean±SE WC was 88.24±0.23 cm for women and 100.68±0.25 cm for men; FU1 was 88.05±0.25 cm for women and 100.51±0.25 cm for men; and FU2 was 87.58±0.33 cm for women and 100.26±0.45 cm for men. Supplementary Table 2 provides baseline sociodemographic characteristics of all CLSA participants. Our weighted analytic sample averaged 58.7±0.17 years at baseline and was comprised of 52% women. Study duration averaged 5.7±0.003 years, with 2.9±0.002 years between baseline and FU1. Table 1 shows about two-thirds of women remained partnered (68.8%), and most remained co-living (78.8%). Nearly a third remained or became non-partnered and about a fifth remained or became lone-living. Greater proportions of men remained partnered (78.7%) or co-living (85.8%), and smaller proportions reported adverse changes in these close social connections (Table 2). Transitions in women’s and men’s social participation were evenly distributed. Changes in all three types of social connection were closely related to several socio-demographic measures and visceral adiposity at follow-up (Tables 1 and 2). Overall, the change in mean WC was negative (meaning a decline) among women for all social connection transitions and among men for most except becoming divorced or co-living. However, the prevalence of visceral adiposity at study follow-up varied across social connection transitions, particularly for men. Those who were lost to follow-up had similar characteristics to those who completed the study (Supplementary Tables 3 and 4).
Table 1.
Sample characteristics across changes in social connections among women in the Canadian Longitudinal Study on Aging (2011–2021)
| Characteristic | Age (yr) | Middle age (45–64 yr) | University educated | Highest household income* | Home-owner | Urban | Non-smoker | No chronic condition† | Δ Mean WC (cm) | Visceral adiposity‡ |
|---|---|---|---|---|---|---|---|---|---|---|
| Transition in marital status (n=7,340) | ||||||||||
| Remained partnered (n=4,428) | 57.47±0.27 | 3,132 (78.7) | 3,550 (66.5) | 889 (18.5) | 4,214 (94.9) | 3,844 (90.1) | 2,384 (48.9) | 259 (6.0) | –0.50±0.18 | 1,883 (43.7) |
| Remained non-partnered (n=2,370) | 62.55±0.46 | 1,113 (58.5) | 1,775 (59.0) | 45 (1.6) | 1,707 (65.9) | 2,220 (96.1) | 1,123 (41.9) | 65 (4.6) | –0.48±0.26 | 1,192 (49.8) |
| Became divorced (n=87) | 53.54±1.56 | 79 (92.4) | 72 (74.9) | 19 (20.6) | 80 (95.5) | 77 (94.4) | 44 (45.1) | - | –0.52±1.56 | 28 (26.8) |
| Became widowed (n=134) | 67.76±1.43 | 35 (34.0) | 98 (60.1) | - | 115 (84.2) | 119 (88.0) | 72 (52.2) | 6 (11.8) | –1.00±0.77 | 61 (45.7) |
| Became partnered (n=130) | 56.06±1.61 | 102 (83.4) | 105 (75.8) | 5 (3.2) | 97 (84.1) | 115 (83.1) | 64 (55.2) | - | –0.03±0.61 | 56 (34.9) |
| Uncertain (n=191) | 58.82±0.93 | 126 (78.5) | 141 (57.0) | 12 (9.6) | 131 (69.3) | 184 (98.7) | 69 (29.9) | 12 (3.4) | –0.19±0.81 | 95 (51.0) |
| Transition in living arrangement (n=7,341) | ||||||||||
| Remained co-living (n=5,032) | 57.29±0.24 | 3,578 (79.2) | 4,021 (66.9) | 922 (17.0) | 4,678 (92.4) | 4,413 (90.6) | 2,669 (47.8) | 295 (6.2) | –0.46±0.17 | 2,170 (44.4) |
| Remained lone-living (n=1,735) | 65.72±0.41 | 693 (47.0) | 1,277 (55.5) | 27 (1.1) | 1,223 (65.1) | 1,631 (96.0) | 813 (43.4) | 39 (2.5) | –0.64±0.27 | 887 (49.1) |
| Became co-living (n=207) | 62.01±1.02 | 108 (58.9) | 157 (64.1) | 6 (1.8) | 151 (70.8) | 193 (96.7) | 90 (34.0) | 6 (3.4) | –0.76±0.73 | 98 (44.8) |
| Became lone-living (n=367) | 59.89±1.27 | 211 (71.5) | 286 (58.2) | 19 (4.2) | 293 (69.1) | 322 (92.6) | 184 (44.1) | 10 (12.4) | –0.51±0.72 | 169 (40.5) |
| Transition in social participation (n=7,328) | ||||||||||
| Stable (n=2,230) | 58.72±0.35 | 1,363 (74.6) | 1,718 (64.8) | 334 (15.2) | 1,945 (88.2) | 2,004 (91.8) | 1,133 (47.0) | 109 (6.2) | –0.63±0.23 | 959 (42.1) |
| Increased (n=2,499) | 59.41±0.39 | 1,610 (71.6) | 1,963 (64.3) | 307 (12.9) | 2,157 (84.8) | 2,226 (91.6) | 1,260 (47.1) | 115 (6.8) | –0.32±0.23 | 1,110 (42.8) |
| Decreased (n=2,599) | 58.13±0.38 | 1,614 (75.0) | 2,055 (65.3) | 333 (13.6) | 2,234 (88.1) | 2,318 (91.5) | 1,356 (46.2) | 126 (4.9) | –0.56±0.27 | 1,235 (49.1) |
Values are presented as mean±standard error or number (%) using inflation weight provided by and as guided by Canadian Longitudinal Study on Aging. All the variables were reported at baseline, except for outcome variables: visceral adiposity measured at follow-up 2 (FU2) and mean change in WC from follow-up 1 (FU1) to FU2. ‘-’ represents a cell size less than 5 in accordance with Statistics Canada privacy requirements.
*Highest income was defined as ≥C$150,000; †Chronic conditions was a derived variable of several health measures including anxiety and depression; ‡Visceral adiposity was defined as WC≥ 88 cm for women.19
WC, waist circumference.
Table 2.
Sample characteristics across changes in social connections among men in the Canadian Longitudinal Study on Aging (2011–2021)
| Characteristic | Age (yr) | Middle age (45–64 yr) | University educated | Highest household income* | Home-owner | Urban | Non-smoker | No chronic condition† | Δ Mean WC (cm) | Visceral adiposity‡ |
|---|---|---|---|---|---|---|---|---|---|---|
| Transition in marital status (n=6,601) | ||||||||||
| Remained partnered (n=5,143) | 58.75±0.30 | 3,015 (73.3) | 4,262 (67.0) | 1,128 (22.5) | 4,819 (91.0) | 4,617 (93.2) | 2,412 (43.1) | 368 (8.6) | –0.90±0.23 | 1,892 (38.7) |
| Remained non-partnered (n=1,016) | 58.93±0.59 | 583 (75.7) | 771 (55.0) | 38 (2.2) | 699 (57.3) | 932 (96.4) | 425 (36.3) | 64 (7.8) | –1.00±0.42 | 408 (42.0) |
| Became divorced (n=81) | 52.33±1.08 | 67 (95.7) | 66 (54.5) | 19 (14.7) | 73 (76.4) | 70 (92.5) | 45 (36.2) | 5 (12.7) | 0.49±1.15 | 30 (39.3) |
| Became widowed (n=67) | 65.50±2.51 | 20 (53.7) | 54 (66.9) | 8 (16.0) | 58 (85.3) | 64 (98.1) | 26 (34.7) | - | –0.91±1.29 | 24 (45.5) |
| Became partnered (n=137) | 54.24±0.75 | 98 (90.6) | 120 (73.9) | 18 (9.5) | 111 (66.0) | 126 (97.9) | 54 (26.9) | 13 (7.7) | –0.27±0.73 | 42 (37.5) |
| Uncertain (n=157) | 58.60±1.23 | 105 (77.4) | 121 (54.2) | 20 (21.7) | 119 (74.2) | 145 (95.5) | 53 (34.2) | 10 (19.1) | –1.20±0.55 | 58 (46.5) |
| Transition in living arrangement (n=6,600) | ||||||||||
| Remained co-living (n=5,472) | 58.43±0.27 | 3,284 (74.8) | 4,532 (66.4) | 1,177 (21.7) | 5,082 (88.6) | 4,918 (93.4) | 2,572 (42.3) | 394 (9.0) | –0.92±0.21 | 2,005 (39.4) |
| Remained lone-living (n=809) | 60.90±0.74 | 411 (67.5) | 608 (57.1) | 27 (1.6) | 538 (58.4) | 748 (96.0) | 307 (34.1) | 45 (6.6) | –0.82±0.56 | 328 (38.8) |
| Became co-living (n=105) | 57.02±1.39 | 66 (78.4) | 89 (70.2) | 5 (3.7) | 85 (75.6) | 93 (97.6) | 49 (50.3) | 8 (20.7) | 0.15±0.68 | 37 (27.2) |
| Became lone-living (n=214) | 56.64±1.35 | 128 (82.1) | 165 (52.3) | 24 (11.4) | 174 (74.0) | 194 (96.2) | 85 (33.3) | 13 (5.1) | –0.46±0.75 | 84 (46.6) |
| Transition in social participation (n=6,593) | ||||||||||
| Stable (n=1,976) | 58.74±0.42 | 1,174 (74.3) | 1,639 (68.6) | 391 (20.9) | 1,757 (85.7) | 1,758 (91.8) | 920 (42.0) | 143 (9.6) | –1.16±0.21 | 722 (37.2) |
| Increased (n=2,277) | 58.56±0.39 | 1,344 (74.5) | 1,831 (64.4) | 408 (17.9) | 2,026 (83.9) | 2,086 (95.0) | 1,013 (41.7) | 159 (8.1) | –0.63±0.24 | 840 (40.2) |
| Decreased (n=2,340) | 58.43±0.47 | 1,367 (74.6) | 1,920 (63.1) | 433 (19.4) | 2,089 (85.9) | 2,102 (94.1) | 1,078 (40.7) | 157 (8.9) | –0.92±0.42 | 890 (40.2) |
Values are presented as mean±standard error or number (%) using inflation weight provided by and as guided by Canadian Longitudinal Study on Aging. All the variables were reported at baseline, except for outcome variables: visceral adiposity measured at follow-up 2 (FU2) and mean change in WC from follow-up 1 (FU1) to FU2.
*Highest income was defined as ≥C$150,000; †Chronic conditions was a derived variable of several health measures including anxiety and depression; ‡Visceral adiposity was defined as WC ≥102 cm for men.19
WC, waist circumference.
Association of marital transitions with changes in WC among aging women and men
Table 3 shows the main results of multivariable-adjusted mixed linear regression, showing differences by gender. Overall, marital transitions were not significantly associated with changes in mean WC among women (omnibus test: P=0.62). However, distinct marital transitions were associated with unique changes in mean WC among men (omnibus test: P=0.03). Specifically, men who became divorced had significantly less decrease in mean WC (1.45 cm; 95% CI, 0.14 to 2.75), relative to men remaining partnered (model D).
Table 3.
Beta-coefficients and 95% CIs of changes in mean waist circumference (cm) associated with changes in social connections among aging women and men in the Canadian Longitudinal Study on Aging (2011–2021)
| Variable | Model A | Model B | Model C | Model D | ||||
|---|---|---|---|---|---|---|---|---|
| β | 95% CI | β | 95% CI | β | 95% CI | β | 95% CI | |
| Women | ||||||||
| Marital transitions (n) | 7,387 | 7,373 | 7,340 | 7,340 | ||||
| Remained partnered | Ref | Ref | Ref | Ref | ||||
| Remained non-partnered | 0.42* | 0.09 to 0.75 | 0.16 | –0.21 to 0.54 | 0.17 | –0.20 to 0.55 | 0.18 | –0.19 to 0.55 |
| Became divorced | 0.55 | –0.79 to 1.89 | 0.58 | –0.77 to 1.92 | 0.56 | –0.78 to 1.90 | 0.57 | –0.77 to 1.91 |
| Became widowed | 0.68 | –0.42 to 1.77 | 0.58 | –0.52 to 1.68 | 0.58 | –0.52 to 1.68 | 0.58 | –0.52 to 1.68 |
| Became partnered | 0.46 | –0.63 to 1.56 | 0.27 | –0.83 to 1.38 | 0.31 | –0.80 to 1.42 | 0.31 | –0.80 to 1.42 |
| Uncertain | 0.85 | –0.07 to 1.76 | 0.55 | –0.38 to 1.48 | 0.58 | –0.35 to 1.51 | 0.59 | –0.34 to 1.53 |
| Living arrangement transitions (n) | 7,388 | 7,374 | 7,341 | 7,341 | ||||
| Remained co-living | Ref | Ref | Ref | Ref | ||||
| Remained lone-living | 0.44* | 0.07 to 0.80 | 0.19 | –0.21 to 0.59 | 0.20 | –0.20 to 0.60 | 0.21 | –0.19 to 0.61 |
| Became co-living | 0.08 | –0.79 to 0.96 | –0.18 | –1.07 to 0.71 | –0.18 | –1.07 to 0.72 | –0.19 | –1.08 to 0.70 |
| Became lone-living | 0.06 | –0.61 to 0.73 | –0.10 | –0.78 to 0.58 | –0.10 | –0.78 to 0.58 | –0.10 | –0.78 to 0.58 |
| SP transitions (n) | 7,374 | 7,360 | 7,328 | 7,328 | ||||
| Stable SP | Ref | Ref | Ref | Ref | ||||
| Increased SP | 0.33 | –0.03 to 0.69 | 0.31 | –0.05 to 0.67 | 0.33 | –0.04 to 0.69 | 0.32 | –0.04 to 0.68 |
| Decreased SP | 0.47* | 0.11 to 0.83 | 0.46* | 0.10 to 0.82 | 0.47* | 0.11 to 0.83 | 0.47* | 0.11 to 0.83 |
| Men | ||||||||
| Marital transitions (n) | 6,670 | 6,654 | 6,601 | 6,601 | ||||
| Remained partnered | Ref | Ref | Ref | Ref | ||||
| Remained non-partnered | 0.34 | –0.10 to 0.70 | 0.16 | –0.28 to 0.60 | 0.18 | –0.26 to 0.63 | 0.19 | –0.25 to 0.63 |
| Became divorced | 1.45* | 0.15 to 2.75 | 1.43* | 0.13 to 2.73 | 1.44* | 0.14 to 2.74 | 1.45* | 0.14 to 2.75 |
| Became widowed | –1.41 | –2.84 to 0.03 | –1.40 | –2.84 to 0.03 | –1.39 | –2.83 to 0.04 | –1.41 | –2.85 to 0.03 |
| Became partnered | 0.90 | –0.10 to 1.91 | 0.81 | –0.19 to 1.81 | 0.78 | –0.23 to 1.79 | 0.78 | –0.23 to 1.79 |
| Uncertain | 0.60 | –0.31 to 1.59 | 0.51 | –0.44 to 1.46 | 0.52 | –0.43 to 1.48 | 0.55 | –0.41 to 1.50 |
| Living arrangement transitions (n) | 6,669 | 6,652 | 6,600 | 6,600 | ||||
| Remained co-living | Ref | Ref | Ref | Ref | ||||
| Remained lone-living | 0.25 | –0.19 to 0.69 | 0.09 | –0.38 to 0.57 | 0.11 | –0.37 to 0.59 | 0.12 | –0.36 to 0.60 |
| Became co-living | 0.65 | –0.48 to 1.78 | 0.52 | –0.62 to 1.66 | 0.53 | –0.62 to 1.68 | 0.53 | –0.62 to 1.68 |
| Became lone-living | 0.86* | 0.05 to 1.66 | 0.75 | –0.06 to 1.57 | 0.74 | –0.07 to 1.56 | 0.74 | –0.07 to 1.56 |
| SP transitions (n) | 6,662 | 6,645 | 6,593 | 6,593 | ||||
| Stable SP | Ref | Ref | Ref | Ref | ||||
| Increased SP | 0.35 | –0.01 to 0.70 | 0.36 | –0.001 to 0.71 | 0.33 | –0.03 to 0.68 | 0.33 | –0.03 to 0.69 |
| Decreased SP | 0.15 | –0.20 to 0.50 | 0.15 | –0.21 to 0.50 | 0.15 | –0.21 to 0.50 | 0.15 | –0.21 to 0.50 |
Gender-specific beta-coefficients and 95% CI obtained by stratified, random coefficient linear regression analysis adjusting for age, age squared, waist circumference at follow-up 1, study duration (model A), then education, household income, and home-ownership (model B), then smoking and chronic conditions (including anxiety and depression) (model C), and finally gross domestic product, provincial spending on social protection, provincial prevalence of obesity (model D).
*P<0.05.
CI, confidence interval; SP, social participation.
Fig. 1 displays the predicted mean WC at FU2 associated with each marital transition to show the variation in mean WC and the differences between women (Fig. 1A) and men (Fig. 1B) (Supplementary Table 5). Men showed the greatest variation between the highest mean WC for men who became divorced (100.54 cm; 95% CI, 99.17 to 101.91) and the lowest mean WC for men who became widowed (97.68 cm; 95% CI, 96.18 to 99.18).
Figure 1.
Predicted mean (95% confidence interval [CI]) waist circumference (WC) at 6-year follow-up associated with marital transitions in the Canadian Longitudinal Study on Aging (CLSA) (2011–2021). (A) Women and (B) men.
Association of changes in living arrangement with changes in WC among aging women and men
The final multivariable-adjusted mixed model showed that, overall, living arrangement transitions were not significantly associated with changes in mean WC in women (omnibus test, P=0.65) and men (P=0.27), although the directions of association appeared unique to women and men (Table 3). The biggest difference in mean WC for women was between remaining lone-living and becoming co-living, and for men between becoming lone-living and remaining co-living (Fig. 2; Supplementary Table 5).
Figure 2.
Predicted mean (95% confidence interval [CI]) waist circumference (WC) at 6-year follow-up associated with living arrangement transitions in the Canadian Longitudinal Study on Aging (CLSA) (2011–2021). (A) Women and (B) men.
Association of changes in social participation with changes in WC among aging women and men
Across models, there was a significant association between changes in social participation and changes in mean WC among women (omnibus test: P=0.03) but not men (omnibus test: P=0.20). Women who decreased their social participation between baseline and FU1 had significantly less decline in mean WC (0.47 cm; 95% CI, 0.11 to 0.83), compared to WC changes in women with stable social participation (Table 3). Fig. 3 illustrates the predicted mean WC at FU2 associated with stable, increased and decreased social, with unique patterns seen for women (Fig. 3A) and men (Fig. 3B) (Supplementary Table 5). Women who decreased their social participation had the highest predicted mean WC of 87.61 cm (95% CI, 87.14 to 88.09) while women with stable social participation had the lowest mean WC (87.15 cm; 95% CI, 86.66 to 87.63).
Figure 3.
Predicted mean (95% confidence interval [CI]) waist circumference (WC) at 6-year follow-up associated with social participation (SP) transitions in the Canadian Longitudinal Study on Aging (CLSA) (2011–2021). (A) Women and (B) men.
Overall, sensitivity analyses showed that main results for women and men remained the same or amplified after further adjustment for physical activity level, urban living, sleep deprivation, fruits and vegetables intake, reproductive factors (for women), and baseline social connections (Supplementary Table 6). Some adjustments are noted. Adjustment for baseline BMI attenuated results for women who became divorced, while adjustment for alcohol intake amplified results for men who become lone-living (0.86 cm; 95% CI, 0.04 to 1.68). Inclusion of baseline social connections also attenuated main results for women remaining non-partnered.
DISCUSSION
This prospective study examined changes in three different social connections in relation to changes in WC among aging women and men in Canada. Overall, changes in social connections were associated with subsequent changes in WC, with gender differences as hypothesized. Specifically, we found declines in social participation were significantly associated with changes in visceral adiposity among women. Loss of marriage from divorce was significantly associated with changes in visceral adiposity among men. Contrary to our hypothesis, we found null results for changes in living arrangement and changes in WC. Key findings were robust to alternative model specifications, and remained after mutually adjusting for other social connections (independent effects). Although predicted mean levels of WC at 6-year follow-up were relatively similar across different social connection changes, men showed the greatest difference in predicted mean WC for marital transitions of 2.86 cm and women showed the greatest difference in predicted mean WC for social participation of 0.46 cm. Results for all social connection changes were clinically meaningful as predicted mean WC levels reached the ‘high risk’ category for diabetes and coronary heart disease for both women (80 to 88 cm) and men (94 to 102 cm).19
Social connections, both the existence and lack of, have been linked to longevity in adults.26,27 Recent cross-sectional data in Canada suggest that different social connections are linked to BMI and WC in aging adults.7 Gender is also known to moderate how social factors contribute to differences in adiposity.28 There is little longitudinal evidence on the prospective association of changes in social connections and changes in mean WC,17 and no research in Canada. Current longitudinal obesity research, however, is restricted to marital transitions and rarely provides sex-disaggregated data.17 This prospective study is the first, to our knowledge, to prospectively assess changes in three different social connections in relation to changes in visceral adiposity among aging adults in Canada from a gender perspective. Importantly, this study assessed several distinct marital transitions, such as entering marriage or experiencing unclear transitions, that are not typically examined in the current literature.
Prior research showed that transition out of marriage either through divorce or widowhood is linked to decreases in general and visceral adiposity.17 Our study builds on this literature by distinguishing between widowhood and divorce/separation, and reveals the importance of these transitions for men specifically. Notably, men becoming divorced/separated had significantly less relative decline in mean WC and the highest predicted mean WC at 6-year follow-up. The biological mechanisms underlying this association may involve several pathways, which are outlined in Cohen’s model of how social relationships influence health.29 The first pathway involves a direct effect of social connection transitions on health-related biological factors, such as metabolic dysregulation, that lead to health status.30 A second pathway involves an indirect effect of social relationships on health via social influence, services and information altering health-promoting behaviours (e.g., diet, exercise, or sleep) that affect health-related biological factors and ultimately health. Adverse changes in social connections may reduce adherence to healthy eating patterns, leading to lower fruit and vegetable intake particularly among men,31 and thereby increase WC. A third pathway is the indirect effect of psychological states (e.g., perceived stress) on both the neuro-endocrine response (e.g., cortisol levels) and health-promoting behaviours (e.g., diet) that lead to metabolic dysregulation such as increased visceral adiposity.32 Notably, becoming separated/divorced in later life can be a sudden and emotionally taxing experience that activates the stress response to elevate cortisol levels and promote hedonically-driven intake of energy-dense foods, thereby leading to abdominal fat accumulation.32
Men often rely heavily on their spouses for support and so transitioning out of marriage, particularly through divorce, involves losing a key source of emotional and social support as well as potentially losing the gender identity of head of household.33,34 Such losses can be psychologically stressful leading to elevated cortisol levels and/or declining health-promoting behaviours that in turn increase men’s waistlines. Conversely, women’s broader social networks may buffer divorce-related stress as other non-marital relationships can provide emotional, tangible or other support that mitigate adverse WC changes.35 Our finding for divorce aligns with prior research on adults aged 15 to 90 showing increased WC among men who transitioned out of marriage compared to staying married.36 However, we observed a contrasting result for widowhood, where men experienced a non-significant greater decline in WC that may still be clinically meaningful as this finding might suggest men who become divorced have inadequate or irregular caloric intake due to gender differences in food preparation and meal planning skills, and/or reduced appetite in the absence of shared meals with a wife.15 Notably, the opposite effect of less decline in WC was observed for women becoming widowed who may occur because of metabolic consequences of cortisol from poverty-related stress due to gender differences in employment and cumulative wealth.37
Research has linked social participation to health outcomes, including adiposity. Engaging in social activities not only boosts physical activity but also reduces feelings of loneliness, depression, and psychological stress, all of which promote healthful behaviours that can influence adiposity.38 Specifically, a Canadian study focusing on older adults found that women with limited social participation (defined as fewer than two regular social activities) had greater levels of WC by 4.19 cm than women with five or more activities.7 This finding is supported by a review showing higher odds of visceral adiposity among those less involved in social activities, particularly in women.39 Our findings align with these observations that decreased social participation over 3 years resulted in significantly less decline in WC and the highest predicted mean WC among women in the subsequent 3 years. visceral adiposity. This gendered pattern is consistent with broader evidence that social isolation and reduced participation disproportionately impact women’s health behaviours, including diet.31 Women tend to have larger and more diverse social networks than men, which serve various functions beyond marriage and provide opportunities for sociability, and meaningful roles.31 When women experience reduced social participation or isolation, they lose access to these roles, disrupting dietary habits31 and potentially affecting their visceral adiposity. These findings highlight the need for further research on how changes in social participation could impact changes in visceral adiposity among women and men.
This study has several limitations of note. There is potential for misclassification bias stemming from self-reported exposures and covariables, particularly in categorizing individuals into uncertain marital transitions when they may belong to another category. Additionally, the study was confined to individuals whose WC measurements were taken at FU2 which excludes a large number of participants who were unable to visit collection sites due to coronavirus disease 2019 (COVID-19) restrictions, and this could introduce selection bias. Despite incorporating a wide range of covariables in our sensitivity analyses, residual confounding from measurement error likely remains. Another limitation to internal validity is unmeasured confounding due to the lack of CLSA data on energy intake that could not be included in our models. Finally, findings cannot be generalised to more diverse populations in other settings due to inherent limitations of Canada’s aging cohort representing a heteronormative (97%), cisgender (99%), and White (94%) population of aging adults. Thus, future prospective studies on social connections and metabolic health need to reproduce findings in similar populations and also replicate them in other populations of diverse ethnicity, sexual orientation, and gender identity.
Notwithstanding, there are several strengths of this study. WC was objectively measured rather than being self-reported, potentially avoiding underreporting due to desirability bias. Standardized measurement procedures across all participants increased the accuracy and consistency of the results, reducing potential for measurement bias. The study included multiple, repeated measures of three different social connection and several known confounders, including three indicators of socioeconomic status. In addition, our study distinguished different transitions out of marriage-like partnership that the current literature lacks and we included an uncertain marital transition to minimise misclassification bias. The cohort comprises a large sample of approximately between 6,000 to 7,000 middle-aged and older men and women, representative of the population in Canada based on age and sex. The greatest strength and novelty of this study was the application of a sex- and gender-based perspective to produce data disaggregated between women and men to inform more targeted healthy aging strategies. Finally, this study employed a more robust causal-effect estimation approach of (1) change in visceral adiposity by adjusting for WC at FU1 in models of WC at FU2, rather than employing a change score, and (2) temporality between the exposure and the outcome.
In conclusion, this prospective study of aging Canadians found that specific transitions in three different social connections were associated with changes in visceral adiposity among women and men, independent of known confounders and other social connections. Gender-specific associations were found and underscores the importance of disaggregated data for creating tailored supportive environments as a healthy aging strategy. More research is needed to replicate novel findings for understudied social connections, and future studies could also consider subpopulations of women and men to promote better health equity. Intervention research is needed to explore factors that could protect aging adults from visceral adiposity during social life transitions to improve obesity prevention.
SUPPLEMENTARY MATERIALS
Supplementary materials can be found online at https://doi.org/10.7570/jomes25012.
ACKNOWLEDGMENTS
This research was supported by the Canadian Institutes of Health Research (CIHR) (grant # AWD-020822). Annalijn I. Conklin is also supported by a Michael Smith Health Research BC Award (SCH-2020-0581). This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the CLSA is provided by the Government of Canada through the CIHR under grant reference: LSA94473 and the Canada Foundation for Innovation, as well as the following provinces, Newfoundland, Nova Scotia, Quebec, Ontario, Manitoba, Alberta, and British Columbia. This research was conducted using the CLSA Baseline Comprehensive version 7.0; Follow-up 1 Comprehensive version 3.2; and Follow-up 2 Comprehensive version 2.0 under application number 2206013. The CLSA is led by Drs Parminder Raina, Christina Wolfson and Susan Kirkland. The opinions expressed in this manuscript are the author’s own and do not reflect the views of the CLSA.
Data are available from the CLSA (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.
Footnotes
CONFLICTS OF INTEREST
The authors declare no conflict of interest.
AUTHOR CONTRIBUTIONS
Study concept and design: GV, WZ, NAK, and AIC; acquisition of data: AIC; analysis and interpretation of data: RMC, GC, SM, GV, WZ, NAK, and AIC; drafting of the manuscript: RMC and AIC; critical revision of the manuscript: GC, SM, GV, WZ, NAK, and AIC; statistical analysis: RMC; obtained funding: GV, NAK, and AIC; administrative, technical, or material support: AIC; and study supervision: AIC.
References
- 1.Statistics Canada. Canada’s population estimates: age and sex, July 1, 2022 [Internet]. Statistics Canada; 2022 [cited 2025 Dec 17]. Available from: https://www150.statcan.gc.ca/n1/daily-quotidien/220928/dq220928c-eng.htm
- 2.Statistics Canada. Population count and population growth in Canada [Internet]. Statistics Canada; 2015 [cited 2025 Dec 17]. Available from: https://www150.statcan.gc.ca/n1/pub/91-520-x/2010001/aftertoc-aprestdm1-eng.htm
- 3.Statistics Canada. A portrait of Canada’s growing population aged 85 and older from the 2021 Census [Internet]. Statistics Canada; 2022 [cited 2025 Dec 17]. Available from: https://www12.statcan.gc.ca/census-recensement/2021/as-sa/98-200-X/2021004/98-200-X2021004-eng.cfm
- 4.Twells LK, Gregory DM, Reddigan J, Midodzi WK. Current and predicted prevalence of obesity in Canada: a trend analysis. CMAJ Open. 2014;2:E18–26. doi: 10.9778/cmajo.20130016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jaffer S, Foulds HJA, Parry M, Gonsalves CA, Pacheco C, Clavel MA, et al. The Canadian Women's Heart Health Alliance ATLAS on the epidemiology, diagnosis, and management of cardiovascular disease in women: Chapter 2: scope of the problem. CJC Open. 2021;3:1–11. doi: 10.1016/j.cjco.2020.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Koparkar G, Biswas DA. Adiposity and cardiac defects: pathophysiology and etiology. Cureus. 2023;15:e34026. doi: 10.7759/cureus.34026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hosseini Z, Veenstra G, Khan NA, Conklin AI. Associations between social connections, their interactions, and obesity differ by gender: a population-based, cross-sectional analysis of the Canadian Longitudinal Study on Aging. PLoS One. 2020;15:e0235977. doi: 10.1371/journal.pone.0235977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ge L, Yap CW, Heng BH. Associations of social isolation, social participation, and loneliness with frailty in older adults in Singapore: a panel data analysis. BMC Geriatr. 2022;22:26. doi: 10.1186/s12877-021-02745-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wilcox S, Evenson KR, Aragaki A, Wassertheil-Smoller S, Mouton CP, Loevinger BL. The effects of widowhood on physical and mental health, health behaviors, and health outcomes: the Women's Health Initiative. Health Psychol. 2003;22:513–22. doi: 10.1037/0278-6133.22.5.513. [DOI] [PubMed] [Google Scholar]
- 10.Michael YL, Berkman LF, Colditz GA, Kawachi I. Living arrangements, social integration, and change in functional health status. Am J Epidemiol. 2001;153:123–31. doi: 10.1093/aje/153.2.123. [DOI] [PubMed] [Google Scholar]
- 11.Wang J, Zhang L, Wang S, Zhang L. Living arrangements, health lifestyles, and health outcomes among Chinese oldest-old. Front Public Health. 2023;11:1235768. doi: 10.3389/fpubh.2023.1235768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Cornwell B, Laumann EO, Schumm LP. The social connectedness of older adults: a national profile. Am Sociol Rev. 2008;73:185–203. doi: 10.1177/000312240807300201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Teas E, Marceau K, Friedman E. Life-course social connectedness: comparing data-driven and theoretical classifications as predictors of functional limitations in adulthood. Adv Life Course Res. 2023;55:100529. doi: 10.1016/j.alcr.2023.100529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wilson SE. Marriage, gender and obesity in later life. Econ Hum Biol. 2012;10:431–53. doi: 10.1016/j.ehb.2012.04.012. [DOI] [PubMed] [Google Scholar]
- 15.Shahar DR, Schultz R, Shahar A, Wing RR. The effect of widowhood on weight change, dietary intake, and eating behavior in the elderly population. J Aging Health. 2001;13:189–99. doi: 10.1177/089826430101300202. [DOI] [PubMed] [Google Scholar]
- 16.Rosenbloom CA, Whittington FJ. The effects of bereavement on eating behaviors and nutrient intakes in elderly widowed persons. J Gerontol. 1993;48:S223–9. doi: 10.1093/geronj/48.4.S223. [DOI] [PubMed] [Google Scholar]
- 17.Madani Civi R, Mehranfar S, Plunkett R, Veenstra G, Conklin AI. A systematic review of social connections as determinants of obesity: longitudinal evidence limited to marital transitions. Obes Rev. 2024;25:e13819. doi: 10.1111/obr.13819. [DOI] [PubMed] [Google Scholar]
- 18.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;48:1753a–j. doi: 10.1093/ije/dyz173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.International Diabetes Federation. The IDF consensus worldwide definition of the metabolic syndrome. IDF; 2006 [cited 2025 Dec 17]. Available from: https://idf.org/media/uploads/2023/05/attachments-30.pdf
- 20.Lee S, Cho E, Grodstein F, Kawachi I, Hu FB, Colditz GA. Effects of marital transitions on changes in dietary and other health behaviours in US women. Int J Epidemiol. 2005;34:69–78. doi: 10.1093/ije/dyh258. [DOI] [PubMed] [Google Scholar]
- 21.Eng PM, Kawachi I, Fitzmaurice G, Rimm EB. Effects of marital transitions on changes in dietary and other health behaviours in US male health professionals. J Epidemiol Community Health. 2005;59:56–62. doi: 10.1136/jech.2004.020073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kutob RM, Yuan NP, Wertheim BC, Sbarra DA, Loucks EB, Nassir R, et al. Relationship between marital transitions, health behaviors, and health indicators of postmenopausal women: results from the Women's Health Initiative. J Womens Health (Larchmt) 2017;26:313–20. doi: 10.1089/jwh.2016.5925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tannenbaum C, Greaves L, Graham ID. Why sex and gender matter in implementation research. BMC Med Res Methodol. 2016;16:145. doi: 10.1186/s12874-016-0247-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rassy N, Van Straaten A, Carette C, Hamer M, Rives-Lange C, Czernichow S. Association of healthy lifestyle factors and obesity-related diseases in adults in the UK. JAMA Netw Open. 2023;6:e2314741. doi: 10.1001/jamanetworkopen.2023.14741. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Amiri M, Mousavi M, Azizi F, Ramezani Tehrani F. The relationship of reproductive factors with adiposity and body shape indices changes overtime: findings from a community-based study. J Transl Med. 2023;21:137. doi: 10.1186/s12967-023-04000-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review. PLoS Med. 2010;7:e1000316. doi: 10.1371/journal.pmed.1000316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Holt-Lunstad J, Smith TB, Baker M, Harris T, Stephenson D. Loneliness and social isolation as risk factors for mortality: a meta-analytic review. Perspect Psychol Sci. 2015;10:227–37. doi: 10.1177/1745691614568352. [DOI] [PubMed] [Google Scholar]
- 28.Cooper AJ, Gupta SR, Moustafa AF, Chao AM. Sex/gender differences in obesity prevalence, comorbidities, and treatment. Curr Obes Rep. 2021;10:458–66. doi: 10.1007/s13679-021-00453-x. [DOI] [PubMed] [Google Scholar]
- 29.Cohen S. Social relationships and health. Am Psychol. 2004;59:676–84. doi: 10.1037/0003-066X.59.8.676. [DOI] [PubMed] [Google Scholar]
- 30.Holt-Lunstad J. Social connection as a critical factor for mental and physical health: evidence, trends, challenges, and future implications. World Psychiatry. 2024;23:312–32. doi: 10.1002/wps.21224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mehranfar S, Ceolin G, Madani Civi R, Keller H, Murphy RA, Cohen TR, et al. Gender, adverse changes in social engagement and risk of unhealthy eating: a prospective cohort study of the Canadian Longitudinal Study on Aging (2011-2021) Nutrients. 2025;17:1005. doi: 10.3390/nu17061005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chin B, Murphy MLM, Janicki-Deverts D, Cohen S. Marital status as a predictor of diurnal salivary cortisol levels and slopes in a community sample of healthy adults. Psychoneuroendocrinology. 2017;78:68–75. doi: 10.1016/j.psyneuen.2017.01.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Saito T, Murata C, Aida J, Kondo K. Cohort study on living arrangements of older men and women and risk for basic activities of daily living disability: findings from the AGES project. BMC Geriatr. 2017;17:183. doi: 10.1186/s12877-017-0580-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.McKenzie SK, Collings S, Jenkin G, River J. Masculinity, social connectedness, and mental health: men's diverse patterns of practice. Am J Mens Health. 2018;12:1247–61. doi: 10.1177/1557988318772732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bedrov A, Gable SL. Thriving together: the benefits of women's social ties for physical, psychological and relationship health. Philos Trans R Soc Lond B Biol Sci. 2023;378:20210441. doi: 10.1098/rstb.2021.0441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hosseinpour-Niazi S, Mirmiran P, Hosseinpanah F, Fallah-Ghohroudy A, Azizi F. Association of marital status and marital transition with metabolic syndrome: Tehran lipid and glucose study. Int J Endocrinol Metab. 2014;12:e18980. doi: 10.5812/ijem.18980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Holden KC, Smock PJ. The economic costs of marital dissolution: why do women bear a disproportionate cost? Annu Rev Sociol. 1991;17:51–78. doi: 10.1146/annurev.so.17.080191.000411. [DOI] [PubMed] [Google Scholar]
- 38.Dehi Aroogh M, Mohammadi Shahboulaghi F. Social participation of older adults: a concept analysis. Int J Community Based Nurs Midwifery. 2020;8:55–72. doi: 10.30476/IJCBNM.2019.82222.1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Lee XY, Yusof NW, Pillai NK, Yap CG, Jahan NK. Association between social network and obesity among adult population. Open J Endocr Metab Dis. 2022;12:20–46. doi: 10.4236/ojemd.2022.122003. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.