Table 3

Direct and indirect effects of social network properties on metabolic syndrome (n=10 103)*

Community-based low-risk participants (n=8097)
Social network propertiesMen (n=2808)Women (n=5289)
Effect size (95% CI)† (×10−2)P valueEffect size (95% CI)† (×10−2)P value
Size of social network
 Direct effect3.65 (1.74 to 7.07)0.0403.19 (1.15 to 5.20)0.002
 Indirect effect—physical activity0.52 (0.08 to 1.05)0.0240.31 (0.11 to 0.55)<0.001
 Indirect effect—cigarette smoking0.00 (−0.11 to 0.11)0.9770.00 (−0.06 to 0.48)0.910
 Indirect effect—alcohol consumption−0.01 (−0.16 to 0.14)0.933−0.03 (−0.12 to 0.03)0.420
 Indirect effect—depressive symptoms0.14 (−0.15 to 0.47)0.3450.00 (−0.15 to 0.14)0.994
Social network closeness
 Direct effect1.45 (−1.99 to 4.84)0.396−1.40 (−3.43 to 0.56)0.176
 Indirect effect—physical activity0.07 (−0.08 to 0.29)0.3620.05 (−0.05 to 0.17)0.314
 Indirect effect—cigarette smoking0.06 (−0.10 to 0.29)0.4450.02 (−0.04 to 0.09)0.554
 Indirect effect—alcohol consumption−0.03 (−0.20 to 0.11)0.697−0.03 (−0.12 to 0.03)0.358
 Indirect effect—depressive symptoms0.16 (−0.16 to 0.50)0.3400.00 (−0.23 to 0.23)0.991
Hospital-based high-risk participants (n=2006)
Social network propertiesMen (n=1168)Women (n=838)
Effect size (95% CI)† (×10−2)P valueEffect size (95% CI)† (×10−2)P value
Size of social network
 Direct effect−1.12 (−9.07 to 6.51)0.8028.40 (0.09 to 16.23)0.046
 Indirect effect—physical activity0.13 (−0.47 to 0.80)0.6490.07 (−0.61 to 0.82)0.824
 Indirect effect—cigarette smoking−0.01 (−1.02 to 0.92)0.9910.12 (−0.41 to 0.83)0.662
 Indirect effect—alcohol consumption−0.05 (−0.64 to 0.51)0.857−0.04 (−0.50 to 0.37)0.870
 Indirect effect—depressive symptoms0.12 (−0.30 to 0.67)0.6010.03 (−0.36 to 5.07)0.894
Social network closeness
 Direct effect−1.44 (−7.34 to 4.19)0.6301.33 (−5.69 to 8.18)0.705
 Indirect effect—physical activity0.04 (−0.18 to 0.33)0.7900.04 (−0.33 to 0.47)0.854
 Indirect effect—cigarette smoking0.42 (−0.29 to 1.33)0.2670.20 (−0.51 to 1.14)0.598
 Indirect effect—alcohol consumption−0.03 (−0.47 to 0.36)0.859−0.07 (−0.67 to 4.34)0.761
 Indirect effect—depressive symptoms0.66 (−0.09 to 1.58)0.0830.60 (−0.70 to 1.97)0.344
  • Estimated using the ‘mediation’ package. The quasi-Bayesian Monte Carlo method was applied with 5000 times of simulation each.

  • Significant values are in bold.

  • *All results are from the dataset after multiple imputation by chained equations, with a chain length of 10.

  • †Adjusted for age, household income, education level, comorbidities and social network closeness.