Should identical CVD risks in young and old patients be managed identically? Results from two models

Su May Liew*, Rod Jackson, David Mant, Paul Glasziou

*Corresponding author for this work

Research output: Contribution to journalArticleResearchpeer-review

7 Citations (Scopus)
160 Downloads (Pure)

Abstract

Objectives: To assess whether delaying risk reduction treatment has a different impact on potential life years lost in younger compared with older patients at the same baseline short-term cardiovascular risk. Design: Modelling based on population data. Methods: Potential years of life lost from a 5-year treatment delay were estimated for patients of different ages but with the same cardiovascular risk (either 5% or 10% 5-year risk). Two models were used: an agebased residual life expectancy model and a Markov simulation model. Age-specific case fatality rates and time preferences were applied to both models, and competing mortality risks were incorporated into the Markov model. Results: Younger patients had more potential life years to lose if untreated, but the maximum difference between 35 and 85 years was <1 year, when models were unadjusted for time preferences or competing risk. When these adjusters were included, the maximum difference fell to about 1 month, although the direction was reversed with older people having more to lose. Conclusions: Surprisingly, age at onset of treatment has little impact on the likely benefits of interventions that reduce cardiovascular risk because of the opposing effects of life expectancy, case fatality, time preferences and competing risks. These findings challenge the appropriateness of recommendations to use lower risk-based treatment thresholds in younger patients.

Original languageEnglish
Article numberA22
JournalBMJ Open
Volume2
Issue number2
DOIs
Publication statusPublished - 2012

Fingerprint

Dive into the research topics of 'Should identical CVD risks in young and old patients be managed identically? Results from two models'. Together they form a unique fingerprint.

Cite this