Ordered logit models of drinking frequency of ELSA participants in waves 0, 4 and 5 controlling for transitions in depressive symptoms: N=4741
| Variable | Model 2a | Model 2b | Variable | Model 2b | |||
|---|---|---|---|---|---|---|---|
| Coefficient | p Value | Coefficient | p Value | Coefficient | p Value | ||
| Transitions in depressive symptoms×wave | Marital status wave 0 | ||||||
| Wave 0 | Married | 0.00 | |||||
| No depression | 0.00 | 0.00 | Single | −0.27 | 0.26 | ||
| Becomes at risk of depression | −0.52 | 0.01 | −0.08 | 0.67 | Separated/divorced | 0.10 | 0.54 |
| Continual depression | −1.04 | <0.01 | −0.14 | 0.56 | Widowed | 0.08 | 0.62 |
| Depression improves | −0.98 | <0.01 | −0.34 | 0.09 | Employment wave 0 | ||
| Wave 4 | Employed | 0.00 | |||||
| No depression | 0.00 | 0.00 | Economically inactive | −0.31 | 0.05 | ||
| Becomes at risk of depression | −1.18 | <0.01 | −0.72 | <0.01 | Retired | −0.03 | 0.85 |
| Continual depression | −2.20 | <0.01 | −1.27 | <0.01 | Wealth wave 1 | ||
| Depression improves | −1.46 | <0.01 | −0.82 | <0.01 | Bottom quintile | 0.00 | |
| Wave 5 | Second quintile | 0.73 | <0.01 | ||||
| No depression | 0.00 | 0.00 | Third quintile | 1.09 | <0.01 | ||
| Becomes at risk of depression | −1.18 | <0.01 | −0.72 | <0.01 | Fourth quintile | 1.60 | <0.01 |
| Continual depression | −1.94 | <0.01 | −1.02 | <0.01 | Top quintile | 2.47 | <0.01 |
| Depression improves | −1.24 | <0.01 | −0.60 | 0.02 | Education wave 0 | ||
| No depression | No education | 0.00 | |||||
| Wave 0 | 0.00 | 0.00 | Compulsory education | 0.68 | <0.01 | ||
| Wave 4 | −0.96 | <0.01 | −0.95 | <0.01 | Postcompulsory education | 0.87 | <0.01 |
| Wave 5 | −1.14 | <0.01 | −1.13 | <0.01 | Degree or higher | 1.86 | <0.01 |
| Becomes at risk of depression between waves 0 and 5 | Smoking wave 0 | ||||||
| Wave 0 | 0.00 | 0.00 | Non-smoker | 0.00 | |||
| Wave 4 | −1.61 | <0.01 | −1.59 | <0.01 | Used to smoke occasionally | 0.35 | <0.01 |
| Wave 5 | −1.73 | <0.01 | −1.71 | <0.01 | Used to smoke regularly | 1.20 | <0.01 |
| Continual depression | Current smoker | 0.67 | <0.01 | ||||
| Wave 0 | 0.00 | 0.00 | BMI wave 0 | ||||
| Wave 4 | −2.12 | <0.01 | −2.08 | <0.01 | 25–30 | 0.00 | |
| Wave 5 | −2.04 | <0.01 | −2.01 | <0.01 | <20 | −0.60 | 0.13 |
| Depression improves between waves 0 and 5 | 20–25 | 0.34 | 0.01 | ||||
| Wave 0 | 0.00 | 0.00 | >30 | −0.61 | <0.01 | ||
| Wave 4 | −1.43 | <0.01 | −1.42 | <0.01 | |||
| Wave 5 | −1.39 | <0.01 | −1.39 | <0.01 | |||
| Gender | |||||||
| Ref: male | 0.00 | 0.00 | |||||
| Female | −1.55 | <0.01 | −1.18 | <0.01 | |||
| Age at wave 0 (confounding variable) | −0.04 | <0.01 | −0.03 | ||||
| Likelihood ratio test* | <0.01 | <0.01 | |||||
Model 2a includes depressive symptoms×wave, age wave 0 and gender.
Model 2b includes depressive symptoms×wave, age wave 0, gender, marital status, employment, net household wealth, education, smoking and BMI.
*The likelihood ratio test compares the meologit model with an ologit model, that is, if the mixed-effects ordered logit regression model is favoured over a standard ordered logit regression model.
BMI, body mass index; ELSA, English Longitudinal Study of Ageing.