Table 3.
Adjusted RRs (95% CIs) according to presence of depressive symptoms in various cohort subgroups*
| Multivariate† RR | Interaction‡ P value |
|
|---|---|---|
| BMI (kg/m2) | 0.55 | |
| <25 (n§ = 34,379) | 0.87 (0.40–1.89) | |
| 25–29.9 (18,628) | 1.43 (0.97–2.11) | |
| >30 (14,812) | 1.24 (0.96–1.61) | |
| Smoking | 0.82 | |
| Yes (10,361) | 1.28 (0.77–2.12) | |
| No (61,817) | 1.21 (0.97–1.53) | |
| Parental history of diabetes | 0.47 | |
| Yes (14,474) | 1.11 (0.79–1.56) | |
| No (57,687) | 1.28 (0.99–1.66) | |
| Alcohol consumption | 0.62 | |
| Yes (47,888) | 1.24 (0.92–1.67) | |
| No (24,290) | 1.23 (0.92–1.63) |
Relative risk of developing type 2 diabetes for women with depressive symptoms compared with women without depressive symptoms;
calculated using stratified pooled logistic regression with same covariates as in Table 2;
calculated using likelihood ratio test by subtracting −2logL for multivariate model with interaction from −2logL for multivariate model without interaction. Test has a χ2 distribution with v degrees of freedom, where v = (dfmodel with interaction) − (dfmodel without interaction);
n = number of nurses in that category at baseline.