Aims: Self-rated general health (SRH) predicts future mortality. We examined all-cause mortality at 10, 20, and 29 years' follow-up and its association with SRH measured at the age of 40 years in a cohort of 1,198 healthy Danes born in 1936 and who were residents in suburban Copenhagen.
Methods: The association between SRH (dichotomized into good versus poor) and all-cause mortality was estimated in standard time-homogenous Cox regression models adjusting for covariates related to mortality, and in time-heterogeneous Cox regression models without covariate adjustment, where time-heterogeneity features as a separate risk assessment for each of the three follow-up periods defined by the follow-up examinations.
Results: At the age of 40 years, 153 (14.6%) of 1,045 participants reported poor and 85.4% good SRH. Dead participants totalled 36 at the 10-year, 96 at the 20-year, and 207 at the 29-year follow-up. For poor SRH, mortality hazard ratios (multivariate analysis) were persistently significant, but slowly declining with follow-up time. The time-heterogeneous models explain this feature: increased mortality risk was significant only in the first decade after assessment: 2.30 (95% CI 1.11-4.78) vs. 0.91 (95% CI 0.36-2.31) and 0.73 (95% CI 0.34-1.55).
Conclusions: The association between poor SRH and mortality emphasizes the importance of health personnel taking account of people's health rating, particularly when a recent assessment has been made. SRH is related to death, even when controlling for known covariates, but it is not a long-term effect.