Skip to main content
. 2021 Jan 11;11:465. doi: 10.1038/s41598-020-80684-5

Figure 2.

Figure 2

The hazard ratios of loneliness, social disengagement and group-level segregation from Cox PH regression models and Cox MSM models (n = 679). Model 1. Cox PH unadjusted model; no control variables included; Model 2. Cox PH w/o TV: education, sex, loneliness (baseline), social disengagement (baseline), segregation (baseline), living with spouse (baseline), yearly income (baseline), smoking (baseline), drinking (baseline), depression (baseline), physical health (baseline), cognitive health (baseline), comorbidity (baseline); Model 3. Cox PH with TV: Model 2 covariates + loneliness, social disengagement, segregation, living with spouse, yearly income, smoking, drinking, depression, physical health, cognitive health, comorbidity – covariates changing over time; Model 4. Cox MSM (ATE): education, sex, loneliness (baseline), social disengagement (baseline), segregation (baseline), living with spouse (baseline), yearly income (baseline), smoking (baseline), drinking (baseline), depression (baseline), physical health (baseline), cognitive health (baseline), comorbidity (baseline); Model 5. Cox MSM (ATT): education, sex, loneliness (baseline), social disengagement (baseline), segregation (baseline), living with spouse (baseline), yearly income (baseline), smoking (baseline), drinking (baseline), depression (baseline), physical health (baseline), cognitive health (baseline), comorbidity (baseline). Exponentiated coefficients; HR-hazard ratios; TV-time-varying covariates; ATE-average treatment effects; ATT-average treatment effects of treated; 95% confidence intervals in parentheses.