Estimating the effect of cardiovascular risk factors on all-cause mortality and incidence of coronary heart disease using G-estimation: The Atherosclerosis Risk in Communities study

Kate Tilling, Jonathan A.C. Sterne, Moyses Szklo

Research output: Contribution to journalArticlepeer-review

Abstract

Standard methods for analysis of cohort studies may give biased estimates of exposure effects in the presence of time-varying confounding. Such effects may instead be estimated by using G-estimation. This study aimed to examine the relations between important cardiovascular risk factors and all-cause mortality and risk of coronary heart disease (CHD), accounting for confounding between exposures over time using G-estimation. Results were compared with those from standard survival analyses (e.g., Weibull regression) with time-updated covariates. The dataset consisted of all participants in the Atherosclerosis Risk in Communities cohort study who had complete data on the first two of four visits, giving a sample of 13,898 people at baseline. Death and occurrence of CHD or stroke were recorded. G-estimated associations between several risk factors and mortality/CHD incidence differed from those estimated using standard survival analysis. The associations between mortality/CHD incidence and smoking, presence of diabetes, and use of antihypertensives were stronger than the standard survival estimates, while the G-estimated effect of low density lipoprotein and high density lipoprotein cholesterol on CHD incidence were more linear than the standard estimate. Complex relations between exposures over time may lead to biased exposure effect estimates in standard survival analyses. G-Estimation can be used to overcome such biases, and thus may have important implications for the analysis of observational studies.

Original languageEnglish (US)
Pages (from-to)710-718
Number of pages9
JournalAmerican journal of epidemiology
Volume155
Issue number8
DOIs
StatePublished - Apr 15 2002

Keywords

  • Cardiovascular diseases
  • Epidemiologic methods
  • Heart diseases
  • Mortality

ASJC Scopus subject areas

  • Epidemiology

Fingerprint Dive into the research topics of 'Estimating the effect of cardiovascular risk factors on all-cause mortality and incidence of coronary heart disease using G-estimation: The Atherosclerosis Risk in Communities study'. Together they form a unique fingerprint.

Cite this