Table 1.

Data structure.

Variable	Levels	Application in analysis
Ds: disease	1 if incident disease (first time event of CHD, stroke, or lung cancer in 1999 or COPD hospitalisation in 2006)	Main explanatory variable
Age	1-year age group	Included in propensity score
5-year age intervals	5-year age group	Used for stratification of matching procedure
Gender	0 if man; 1 if woman	Used for stratification of matching procedure
CMI: comorbidities	Diseases weigh from 1 to 6 (modified Charlson index)	Included in propensity score
Married: marital status	1 if married or cohabiting, 0 otherwise	Included in propensity score
Origin: national origin	0 if Danish; 1 if not Danish	Included in propensity score
Edu: length of education	1–8, where 8 is university degree	Included in propensity score
Socio: socioeconomic status at baseline (1998) People outside the labour market were excluded from this analysis	(1) Self-employed	Included in propensity score
	(2) White collar
	(3) Blue collar
	(4) Unemployed
Residence	Regional code used as a proxy for urban/rural	Included in propensity score
LM: labour market affiliation	1 if employed or unemployed, 0 if retired or dead	Outcome variable
Wage	Continuous variable expressing the difference between wage in 2000 and wage in 1998	Outcome variable