Predictors of mortality in the prospective Dubbo study of Australian elderly

Background: A prospective study in non-institutionalised Australian elderly 60 years and over commenced in Dubbo, NSW in 1988. Aim: To examine clinical and socio-demographic predictors of all-causes mortality Methods: The data were derived from a community-based sample comprising 1236 men and 1569 women followed for a median period of 62 months. Results: Two hundred and thirty five men (19%) and 184 women (12%) died, 46% of male and 53% of female deaths respectively related to cardiovascular disease. In a proportional hazards model, the significant predictors of mortality were: older age, being married (relative risk [RR]=0.71 for men, 0.74 for women), current smoking for men (RR=3.11), taking more than three alcoholic drinks per day for men (RR=0.37), prior coronary heart disease for men (RR=1.36), severe hypertension for women (RR=1.99), use of anti-hypertensive medication for men (RR=1.74), diabetes for men (RR=1.62), poor-fair self-rated health for women (RR=1.74) and physical disability for men (RR=1.72). Serum cholesterol was associated with mortality in a ‘J-shaped’ relationship in men and in a reciprocal relationship in women. Blood pressure predicted mortality in an incremental fashion below 75 years, but in older subjects lower pressure was associated with excess mortality. Conclusion: Some predictors of mortality in the well elderly have been identified and a more extended period of follow-up will possibly resolve contradictory findings in some areas.