Table 3.
Results of association between SS and PA and quality rating
| Primary Author (Year) | Type of SS1 | Theory2 | Type of analysis3 | Adjustments | Results of association between PA and SS or loneliness4 | Summary result5 | Paper quality rating6 |
|---|---|---|---|---|---|---|---|
| Booth [64] | SSPA | SCT [19] with comments on determinants for older adults from [112] | Forced entry logistic regression analysis | Age, sex, country of birth, marital status, employment status, living situation | Sig. greater number of active people had high social support (42.7% inactive Vs 55.6% active had high social support (P = 0.010). Partner or friends being active was sig. associated with being active. | + | Mod |
| Bopp [65] | SSPA | NS | Bivariate associations. Logistic regression analysis | nil | Sig. positive correlation between total social support (family) and strength training (ST) participation (β =1.10, p = 0.001) and also hours per week of ST training (β = 0.26, p = 0.003). No sig. association SS (friends) and ST | + (family) 0 friends) |
Mod |
| Carlson [31] | SSPA | SEM | Mixed Effect regression models | Total PA - age, ethnicity and gender, walk for transport : age ethnicity, months at address, number of vehicles per adult, walk for leisure: ethnicity and months at address | SS sig. associated with total MVPA (min/ week). B = 14.35, p < 0.01). SS sig. associated with min/ week walking for transport (B = 7.35, p < 0.05). SS sig. associated with min/ week walking for leisure (P < 0.05) |
+ | Mod |
| Gellert [66] | SSPA | NS | ANOVA. Regression analysis using MODPROBE macro | gender, age | Mean PA (F (2, 299) = 4.39, p < 0.05) as well as SS (F (2, 299) =5.49, p < 0.01) was higher in the group with individuals whose partners took part in the intervention, compared to the other two groups. | + | Weak |
| Hall [67] | SSPA | Socioecological model [113] | MANOVAs | age | No Sig. difference between SSPA (friends or family) on whether participants did <10,000 steps or > = 10,000 steps per day. SSE Friends: F = 0.02, p = 0.88 SSE Family: F = 0.02, p = 0.89. |
0 (friends or family) | Mod |
| Kaplan [80] | General SS | NS | Bivariate relationships | Gender, age, education, marital status, smoking, chronic conditions, BMI, injury, functional limitations, distress, region | Higher social support was sig. associated with greater odds of doing frequent PA in females. Females OR (95%CI) =1.08 (1.04-1.13), not significant for males. OR (95%CI) = 1.04 (0.99-1.09) | + female) 0 (males) |
Mod |
| Kim [68] | SSPA | SEM [114] | Correlation followed by Stepwise multiple regression analysis | gender, education level, living situation, self-efficacy | SS (family) pos. associated with amount of PA Fchange (2,279) =10.24, p = 0.012 (second most important contributor to PA after self-efficacy) No sig. effect of SS from friends. | + (family) +(friend) |
Mod |
| Kraithaworn [69] | SSPA | Health promotion model, socio-ecological model. | Path analysis using LISREL | Nil | SS did not significantly predict PA levels directly or indirectly. (Direct effect B = 0.1. Indirect effect B = 0.08, total effect B = 0.18) SS had an indirect effect on PA levels through sense of community. | 0 | Mod |
| Lian [70] | SSPA | PRECEDE health promotion framework | Multiple regression by stepwise method | Nil | More Family encouragement and higher proportion of family members exercising was significantly associated with greater frequency of at least 20 min of moderate to vigorous exercise per week: standardised β = 0.131 and standardised β = 0.108 respectively for men and standardised β = 0.154 and 0.138 for women. (For all P < 0.001) In addition, frequency of contact with people significantly associated with greater frequency of moderate to vigorous exercise per week in women. Standardised β = 0.052, p < 0.05). No association for friends | + (family) 0 (friends) |
Weak |
| Luo [58] | Loneliness | NS | Cross-lagged path analysis | Age, gender, ethnicity, residence (urban or rural), education, financial independence, relative economic status, number of visiting children in 2002. | Regular PA participation decreased odds of being lonely 3 years later and loneliness decreases odds of being active in 3 years. Lonely02-- > PE05 and lonely05-- > PE08 β = -0.028, P < 0.001. PE02-- > lonely05 and PE05-- > lonely08. B = -0.111, p > 0.001. | - | Mod |
| McAuley [25, 71] | SSPA. | SCT (but also testing theoretical models). | Structural equation modeling | Nil | Those who reported more frequent PA, had higher levels of SS, which influenced both a better exercise experience and directly and indirectly a higher self-efficacy, which predicted higher exercise participation at both 6 and 18 months.” Model fit statistics: χ 2 (6) = 5.20, P > .10; NNFI, 1.0; CFI, 1.0; RMSEA, 0.027 |
0 (Indirect + associations: SS → Affect → SE → PASE 6 months → PASE 18 months) |
Weak |
| Mowen [81] | General SS | Stress-buffering and main effect of SS on health. | Path analysis | Nil | Larger SS network size or SS satisfaction did not increase odds of having a moderate or vigorously active lifestyle. SS network β = 0.014, SS satisfaction β = 0.007 | 0 | Mod |
| Netz [59] | Loneliness | NS. | ANOVA with Chi2 test and Multinomial stepwise logistic regressions | BMI, being religious versus secular, Self-rated health and education | No assoc. between odds of feeling lonely and PA level in men. In women it explained 20% of variance. Greater loneliness was associated with lower odds of engaging in sufficient PA as compared to "inactive" OR (SE) = -0.52 (0.23). Adjusted OR (CI) 0.59 (0.38, 0.94). No significant association between living alone and activity levels. | -(female) 0 (males) |
Mod |
| Newall et al. [60] | Loneliness | Fredrickson 's Broaden and Build Theory [115] | Regression analysis | Age, gender, income satisfaction, marital status, functional status, health status | Loneliness was not significantly associated with mean everyday PA (β = 0.001, p > 0.05). Also no interaction between loneliness and happiness (B = 0.08, p > 0.05). However, greater loneliness was associated with subjectively feeling less physically active compared to peers. | 0 | weak |
| O'Brien Cousins [72] | SSPA | Theory of Planned behaviour [116] SCT [117] | Multiple regression analysis | Education, marital status, employment status, country of origin | Exercise level (more PA per week) was associated with a greater composite SSPA. B = 0.264, SE = 0.055 (P < 0.01) | + | Mod |
| Oka. [73] | SSPA. | NS. | Chi2 analysis and an independent group t-test | Age, gender, marital status, BMI, smoking status, alcohol consumption, self-efficacy for exercise, advice from HCP, perceived neighbourhood environment | Greater SS did not increase the likelihood of meeting PA guidelines in either males or females. Adjusted Odd Ratio (AOR) for meeting national PA guidelines and having higher SS for exercise: AOR (95%CI) = 0.82 (0.63-1.07) | 0 (males or females) | Mod |
| Orsega-Smith [33, 74] | SSPA | SCT | Correlation analysis, multiple regression analysis and ANCOVA | Age, physical health | More LTPA significantly associated with higher SSPA from both family and friends. SS (family) Adj. B = 0.72, p < 0.05, SS (friends) Adj. B = 0.113, p < 0.0001. Also, people who met the CDC recommended guidelines for PA were significantly more likely to have higher SS from friends and family. | + (friends) + (family) |
Mod |
| Park [75] | SSPA | NS | Multiple regression, independent 2 sample t-test for high vs low-active and SS | Nil | Multiple regression. No significant correlation between SS from friends and PA. Negative association between SS family and PA. SS family: B = -0.220, t = -3.107 p < 0.01. Both High and low active individuals scored low on SS from friends and family with no significant difference between them. | 0 (friends) - (family) |
Weak |
| Potts [83] | General SS | Health belief model (Becker, 1974 [118, 119] | Ordinary least squares regression | Demographic factors (gender, age, education, marriage, income), health status, perceived frailty | People with stronger social support networks more likely to exercise regularly. B = 0.11 (p < 0.01). | + | Mod |
| Sasidharan [76] | SSPA | SCT | Separate factor analyses for friends and family SS | Nil | Sig. positive association between SS (friends) and LTPA b (unstandardized) (SE) = 0.13 (0.09), p < 0.05 No significant association for family SSL. B (SE) = -0.05 (0.03). |
+(friends) 0 (family) |
Weak |
| Schuster [77] | SSPA | SCT | Hierarchical multiple regression | Perceived barriers | Perceived SS was significantly correlated with LTPA (r = 0.474, p < 0.0001). Perceived SS accounted for an additional 17.5% of the variance in intentional exercise (P < 0.001)' after perceived barriers had been entered into the model. | + | weak |
| Shankar [61] | Loneliness | NS | Multinomial logistic regression | Age, gender, limiting long-standing illness, depression, and marital status-adjusted wealth | Loneliness associated with a greater likelihood of being inactive. OR (95% CI) reference = risky behaviour. Loneliness OR = 1.08 (1.04-1.113) of being inactive vs active. Social isolation: 1.115 (1.11-1.19) of being active vs inactive | - | Mod |
| Shiovitz-Ezra [62] | Loneliness | NS | Multivariate logistic regression | Age, gender, education, income, ethnicity, self-rated health. Functional impairment | No sig. association between being lonely to some degree and doing any PA. PA OR (SE) [95% CI]: 0.8 (0.11) [0.6-1.07] | 0 | Mod |
| Theeke [63] | Loneliness | NS | Chi-square statistics and one-way analysis of variance | Marital status, self-reported health, education, functional impairment, number of chronic illnesses, age, annual household income, number of individuals in household. | Chi-squared testing showed significant difference in frequency of moderate activity in Never lonely, briefly lonely and chronically lonely groups. Chi2 = 438.347 (P < 0.005). The chronically lonely group did less average exercise than the briefly lonely or never lonely groups (no statistical test reported for this). | - | Mod |
| Vance [82] | General SS | NS | Correlation and step-wise regression | Nil | No significant association between social network and total PA (r = 0.02) | 0 | Weak |
| Wilcox [78] | SSPA | SCT | Hierarchical regression analysis | Sociodemographic measures (age, race, education, marital status), | Non-significant trend for greater social support from friends and family (total) to be associated with higher levels of PA. B = 0.16, p = 0.09. Qualitative discussion identified social support as being a very common motivator to PA. | 0 | Mod |
| Yeom [79] | SSPA | Wellness Motivation Theory [120] | Repeated measures ANOVA and Chi-Square test | Nil | Intervention group (IG) significantly increased support from family (F = 21.87, p < 0.01) and friends (F = 24.72, p < 0.001) compared to controls. IG more likely to engage in regular PA after the intervention, compared with controls. Chi-squared =25.01, p < 0.001. | + (friends) + (family) |
Mod |
1SSPA = Social Support for PA
2NS = Not specified. SCT = Social Cognitive Theory
3ANOVA = Analysis of Variance
4Sig. = Significant, CI = confidence interval
5+ = positive association, - = negative association, 0 = no association
6Mod = Moderate quality