PPO1 Predicting Future Coronary Heart Disease Mortality trends in England and Wales in 2020: a Bayesian age Period Cohort Approach. (10th September 2013)

• Record Type: Journal Article
• Authors: Castillo, M Guzman; O'Flaherty, M; Bandosz, P; Capewell, S
• Is Part Of: Journal of epidemiology and community health. Volume 67(2013)Supplement 1
• Page Start: A49
• Page End: A49
• Publication Date: 2013-09-10
• Subjects: Public health -- Periodicals; Epidemiology -- Periodicals; 614.4
• DOI: 10.1136/jech-2013-203126.100
• Languages: English
• ISSNs: 0143-005X

Abstract

Abstract : Background: Although Coronary Heart Disease (CHD) death rates have halved since the 1970s, CHD remains a major cause of mortality in the UK. Furthermore, population ageing plus recent increases in obesity and diabetes may soon increase total CHD deaths. Future CHD mortality predictions are thus potentially problematic. Our aim was therefore to explore future projections of CHD mortality in England & Wales under conventional but contrasting assumptions. Methods: In SCENARIO 1, we assumed a conventional counterfactual, that recent CHD mortality rates (ONS, ICD10 codes I20-I25) in 2006 would persist unchanged. The future number of deaths was then calculated by projecting those rates into 2020 population estimates. In SCENARIO 2, we assumed that recent changes in CHD mortality rates would continue. We used a well-established hierarchical Bayesian Age Period Cohort (APC) model, which works under the assumption that variability in incidence data can be explained by age, period and cohort effects. We fitted this model to observed CHD mortality rate trends from 1982 to 2006. We then used the model to predict mortality rates in 2020, and then these rates applied to the 2020 population to compute future deaths. Results: In SCENARIO 1 by assuming that 2006 mortality rates would continue at that level to 2020, the number of CHD deaths would substantially increase: +49% in men and +25% women (104, 500 total CHD deaths in 2020). In SCENARIO 2, assuming recent trends continued, the APC model suggested that number of deaths would decrease by -48% (95% CI -25%, -64% in men); and -59% (-39%, -73%) in women (36, 425 total deaths in 2020, despite a predicted slowing of the mortality decline in groups aged 35-64 years). Conclusion: The decline in CHD mortality has been reasonably continuous since 1982, and there is little reason to believe it will soon abruptly halt, (the dubious underlying assumption of a commonly used mortality counterfactual). By contrast, the Bayesian APC model offers a far more plausible prediction of future trends by simultaneously considering age, period and cohort effects. Therefore, despite population ageing, the number of CHD deaths will most likely halve again between 2006 and 2020. However, CHD mortality rates elsewhere have previously shifted rapidly due to changes in population risk factors. There is thus no room for complacency.