Predictors of long-term mortality in the elderly: The Dubbo Study

Background: This study examines the predictors of long-term all-causes mortality (ACM) in Australian senior citizens. Methods: We have analysed ACM in a cohort of 2805 citizens, 1233 men and 1572 women aged ≥60years, first examined in 1988 and followed for 20years. Hazard ratios and 95% confidence intervals for ACM were obtained from Cox models employing conventional predictors. Results: Over 20years 66% of men (815/1233) and 53% of women (833/1572) died. Constant proportional hazard over the 20years was demonstrated for all predictors, indicating similar relative hazard of ACM during long-term or short-term follow up. There was significant prediction of ACM by current smoking (hazard ratio 1.96, 95% confidence interval 1.57-2.43 in men; 1.67, 1.32-2.10 in women), high blood pressure (1.37, 1.03-1.81; 1.41, 1.07-1.86), diabetes (1.46, 1.17-1.82; 1.83, 1.43-2.34), impaired peak expiratory flow (1.39, 1.15-1.69; 1.80, 1.47-2.21), coronary heart disease at study entry in men (1.33, 1.13-1.57), physical disability (1.38, 1.13-1.68; 1.45, 1.17-1.79) and alcohol intake (0.82, 0.69-0.97; 0.77, 0.66-0.89 respectively). ACM was not significantly predicted by standard lipid parameters. Over the 20-year period smoking was associated with reduced survival of 41months in men and 25months in women, hypertension with reduced survival of 20 and 17months, and diabetes with reduced survival of 24 and 30months respectively. Conclusions: The findings confirm the contribution of cigarette smoking, hypertension and diabetes to ACM in senior citizens, conditions that are potentially amenable to intervention.