Ordered logit model of drinking frequency of ELSA participants in waves 0, 4 and 5 controlling for transitions in self-rated health: N=4741
| Variable | Model 1a | Model 1b | Variable | Model 1b | |||
|---|---|---|---|---|---|---|---|
| Coefficient | p Value | Coefficient | p Value | Coefficient | p Value | ||
| Transitions in self-rated health×wave | Marital status wave 0 | ||||||
| Wave 0 | Married | 0.00 | |||||
| Continual good health | 0.00 | 0.00 | Single | −0.28 | 0.25 | ||
| Health deteriorates | −0.48 | 0.01 | −0.11 | 0.52 | Separated/divorced | 0.13 | 0.45 |
| Continual poor health | −1.63 | <0.01 | −0.99 | <0.01 | Widowed | 0.04 | 0.80 |
| Health improves | −1.27 | <0.01 | −0.93 | <0.01 | Employment wave 0 | ||
| Wave 4 | Employed | 0.00 | |||||
| Continual good health | 0.00 | 0.00 | Economically inactive | <0.01 | 0.99 | ||
| Health deteriorates | −1.09 | <0.01 | −0.81 | <0.01 | Retired | 0.12 | 0.43 |
| Continual poor health | −2.29 | <0.01 | −1.81 | <0.01 | Wealth wave 1 | ||
| Health improves | −1.47 | <0.01 | −1.20 | <0.01 | Bottom quintile | 0.00 | |
| Wave 5 | Second quintile | 0.73 | <0.01 | ||||
| Continual good health | 0.00 | 0.00 | Third quintile | 1.00 | <0.01 | ||
| Health deteriorates | −1.09 | <0.01 | −0.81 | <0.01 | Fourth quintile | 1.61 | <0.01 |
| Continual poor health | −2.41 | <0.01 | −1.99 | <0.01 | Top quintile | 2.34 | <0.01 |
| Health improves | −1.55 | <0.01 | −1.31 | <0.01 | Education wave 0 | ||
| Continual good health | No education | 0.00 | |||||
| Wave 0 | 0.00 | 0.00 | Compulsory education | 0.63 | <0.01 | ||
| Wave 4 | −0.81 | <0.01 | −0.90 | <0.01 | Postcompulsory Education | 0.80 | <0.01 |
| Wave 5 | −0.90 | <0.01 | −1.00 | <0.01 | Degree or higher | 1.75 | <0.01 |
| Health deteriorates between waves 0 and 5 | Smoking wave 0 | ||||||
| Wave 0 | 0.00 | 0.00 | Non-smoker | 0.00 | |||
| Wave 4 | −1.41 | <0.01 | −1.60 | <0.01 | Former occasional smoker | 0.38 | 0.08 |
| Wave 5 | −1.65 | <0.01 | −1.89 | <0.01 | Former regular smoker | 1.27 | <0.01 |
| Continual poor health | Current smoker | 0.81 | <0.01 | ||||
| Wave 0 | 0.00 | 0.00 | BMI wave 0 | ||||
| Wave 4 | −1.47 | <0.01 | −1.72 | <0.01 | 25–30 | 0.00 | |
| Wave 5 | −1.68 | <0.01 | −1.99 | <0.01 | <20 | −0.62 | 0.11 |
| Health improves between waves 0 and 5 | 20–25 | 0.28 | 0.03 | ||||
| Wave 0 | 0.00 | 0.00 | >30 | −0.51 | <0.01 | ||
| Wave 4 | −1.02 | <0.01 | −1.16 | <0.01 | |||
| Wave 5 | −1.18 | <0.01 | −1.37 | <0.01 | |||
| Gender ref: Male | |||||||
| Female | −1.54 | <0.01 | −1.18 | <0.01 | |||
| Age at wave 0 (confounding variable) | −0.04 | <0.01 | −0.03 | <0.01 | |||
| Likelihood ratio test* | <0.01 | <0.01 | |||||
Model 1a includes SRH×wave, age wave 0 and gender.
Model 1b includes SRH×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; SRH, self-rated health.