Health risk or resource? Gradual and independent association between self-rated health and mortality persists over 30 years

Abstract

Background: Poor self-rated health (SRH) is associated with increased mortality. However, most studies only adjust for few health risk factors and/or do not analyse whether this association is consistent also for intermediate categories of SRH and for follow-up periods exceeding 5-10 years. This study examined whether the SRH-mortality association remained significant 30 years after assessment when adjusting for a wide range of known clinical, behavioural and socio-demographic risk factors.

Methods: We followed-up 8,251 men and women aged ≥ 16 years who participated 1977-79 in a community based health study and were anonymously linked with the Swiss National Cohort (SNC) until the end of 2008. Covariates were measured at baseline and included education, marital status, smoking, medical history, medication, blood glucose and pressure.

Results: 92.8% of the original study participants could be linked to a census, mortality or emigration record of the SNC. Loss to follow-up 1980-2000 was 5.8%. Even after 30 years of follow-up and after adjustment for all covariates, the association between SRH and all-cause mortality remained strong and estimates almost linearly increased from "excellent" (reference: hazard ratio, HR 1) to "good" (men: HR 1.07 95% confidence interval 0.92-1.24, women: 1.22, 1.01-1.46) to "fair" (1.41, 1.18-1.68; 1.39, 1.14-1.70) to "poor"(1.61, 1.15-2.25; 1.49, 1.07-2.06) to "very poor" (2.85, 1.25-6.51; 1.30, 0.18-9.35). Persons answering the SRH question with "don't know" (1.87, 1.21-2.88; 1.26, 0.87-1.83) had also an increased mortality risk; this was pronounced in men and in the first years of follow-up.

Conclusions: SRH is a strong and "dose-dependent" predictor of mortality. The association was largely independent from covariates and remained significant after decades. This suggests that SRH provides relevant and sustained health information beyond classical risk factors or medical history and reflects salutogenetic rather than pathogenetic pathways.

Figure 1. NRP 1A 1977–79 and Swiss National Cohort (SNC) until 2008: chart of linked participants.

NRP: National Research Program. NA: not available, can only be determined when 2010 census will be linked.

Figure 2. Adjusted hazard ratios for all-cause mortality, by self-rated health category, sex and increasing length of follow-up (referent: excellent SRH, n = 7,959).

95% confidence intervals are given for hazard ratios after 10 years, 20 years and maximum follow-up. Adjustment for age, marital status, educational level, smoking status, medical history, medication status, fasting blood glucose and systolic blood pressure.

Figure 3. Survival of men and women by self-rated health category, Switzerland 1977–1979, followed up until 2008: Kaplan-Meier curves by sex (N = 8,008).

Number of persons at risk is shown in Table S3 (Supporting Information).