Multivariable model for the relationship between population area size and postpartum depression (EPDS ≥ 13)
| Main exposure variable | Odds ratio (95% CI) |
|---|---|
| Unadjusted | |
| Population area size | |
| Urban | Ref |
| Semiurban | 0.56 (0.41–0.75) |
| Semirural | 0.74 (0.56–0.99) |
| Rural | 0.64 (0.48–0.85) |
| Adjusted* | |
| Population area size | |
| Urban | Ref |
| Semiurban | 0.60 (0.42–0.84) |
| Semirural | 0.90 (0.65–1.24) |
| Rural | 0.80 (0.57–1.11) |
| Parity, primiparous | 0.82 (0.64–1.05) |
| Country of origin (Canadian-born v. non-Canadian-born) | 0.51 (0.39–0.67) |
| Household income | |
| Low | Ref |
| Low-middle | 0.88 (0.57–1.20) |
| Middle-high | 0.82 (0.60–1.29) |
| High | 0.80 (0.54–1.17) |
| History of depression (v. no history) | 2.11 (1.60–2.79) |
| Life stressors | 1.29 (1.20–1.38) |
| Interpersonal abuse | 1.46 (1.06–2.00) |
| Maternal health | |
| Good/fair/poor | Ref |
| Excellent/very good | 0.40 (0.31–0.51) |
| Postpartum social support | |
| Low | Ref |
| Low-middle | 0.67 (0.45–1.00) |
| Middle-high | 0.36 (0.25–0.52) |
| High | 0.15 (0.10–0.23) |
Note: CI = confidence interval, EPDS = Edinburgh Postnatal Depression Scale, ref = reference.
↵* All variables were adjusted for all covariates in the model. We included variables that are known predictors of postpartum depression (history of depression, life stressors in past 2 yr, interpersonal abuse in last 2 yr, social support, maternal self-reported health) or those for which there were major differences between groups for possible prognostic variables (parity, country of origin, household income). Hosmer–Lemeshow for goodness of fit of the multivariable model X2 = 8.43, p = 0.4. All variance inflation factors < 1.30.