Table 2. Overview of transition models.

Outcome	Outcome variable	Model type	Predictors
Risk factors
BMI	Continuous	OLS	Age, sex, education, BMI
Stop smoking	Binary; reversible	Probit	Age, sex, education
Diagnoses incidence
Diabetes	Binary; absorbing state	Probit	Age, sex, education, BMI
Cancer	Binary; absorbing state	Probit	Age, sex, education, current smoker, past smoker
Lung disease	Binary; absorbing state	Probit	Age, sex, education, current smoker, past smoker
Hypertension	Binary; absorbing state	Probit	Age, sex, education, diabetes
Heart disease	Binary; absorbing state	Probit	Age, sex, education, diabetes, hypertension, BMI
Stroke	Binary; absorbing state	Probit	Age, sex, education, diabetes, hypertension, cancer, heart disease
Functional limitations
ADLs	Ordered	Ordered probit	Age, sex, education, diabetes, hypertension, cancer, heart disease, lung disease, stroke, ADLs, IADLs
IADLs	Ordered	Ordered probit
Mortality incidence
Mortality	Binary; absorbing state	Probit	Age, sex, education, diabetes, hypertension, cancer, heart disease, lung disease, stroke, ADLs, IADLs
Health care utilisation
ED visits	Count	2-part; Poisson	Age, sex, education, diabetes, hypertension, cancer, heart disease, lung disease, stroke, ADLs, IADLs, prior use of the outcome variable
Inpatient admissions	Count	2-part; Poisson
Outpatient visits	Count	2-part; Poisson

Legend: (I)ADLs: (Instrumental) Activities of Daily Living. BMI: Body mass index. GP: general practitioner. ED: emergency department. OLS: Ordinary least squares. For details of how outcomes are calculated and ordered, see Table 1. Predictors are taken from the wave prior to outcome.