A review of adjusted estimators of attributable risk.

摘要: This paper reviews adjusted methods of estimation attributable risk (AR), that is allow one to obtain estimates AR while controlling for other factors. Estimability and basic principles are first considered the rationale estimators discussed. Then, reviewed focusing on cross-sectional, cohort case-control studies. Two inconsistent briefly commented upon. Next, based stratification, namely weighted-sum Mantel-Haenszel (MH) approaches, contrasted. It appears approach, which allows full interaction between exposure adjustment factors, can be affected by small-sample bias. By contrast, MH rests assumption no factors may misleading if present. Model-based represent a more general flexible approach includes both stratification approaches as special cases offers intermediate options. Bruzzi et al.'s Greenland Drescher's Finally, problems considered, from case-cohort data, with multiple levels, recurrent events, prevented preventable fractions, generalized impact fraction. Comments future directions presented.