Calculating population attributable fraction for cardiovascular risk factors using different methods in a population based cohort study.

BACKGROUND

Population Attributable Fraction (PAF) is one of the most practical measures for estimating the burden of risk factors with some challenges in its calculation. Cardiovascular disease (CVD) is the first cause of death worldwide and the estimation of accurate PAFs for CVD risk factors is of great importance in conducting preventive strategies. Our aim was to estimate the PAFs of CVD risk factors via direct, i.e. based on regression models, and indirect, i.e. using related equations, methods.

METHODS

Participants (3200 males and 4245 females aged ≥30 yr) without history of CVD were selected from the population-based cohort of Tehran Lipid and Glucose Study (TLGS). Hazard ratio (HR) and Odds ratio (OR) of conventional risk factors were calculated for CVD events after ten yr of follow-up. Levin's and Miettinen's equations were applied to indirectly estimate the PAFs and average PAF was directly derived from logistic regression model.

RESULTS

The sum of PAFs resulted from indirect estimations reached to more than 100% (around 200% and 150% based on Levin's and Miettinen's formula respectively). The direct estimation attributed 80% and 86% of burden of CVD events to conventional risk factors in men and women respectively. The rank and pattern of PAFs of risk factors was somehow different among different methods.

CONCLUSIONS

Estimating priorities of risk factors may differ in different methods for calculating PAF. This study provides evidence on the more expediency of direct method over indirect ways when individual data is available through a population-based cohort.