A framework to address potential bias in colorectal cancer: Its implementation on a nutritional epidemiologic study in Argentina

Abstract

Colorectal cancer (CRC) is the third most incident cancer in Argentina and diet is widely recognized for being associatedwith CRC. The objective of this work was to construct a methodological framework to quantitatively assess systematicerrors in a case-control study, carried out in adult population of Córdoba province (2010-2016).ACRC case-control study (n=490, 161/329 cases/controls) was conducted. Confounding was analyzed applying regression models approach forobserved and unobserved variables. Selection bias was investigated utilizing deterministic scenarios, assigning a priori participationprobabilities. Information bias, derived from missing data in covariates, was handled applying multiple imputation by chained equations (MICE),considering the missing at random mechanism.Confounding effects of sex and socioeconomic status were found. In respect of selection bias, differences regarding conventional estimates werenegligible. MI approach showed a promoting effect of the South Cone dietary pattern. Significant associations with sex, BMI, family history ofCRC and socioeconomic status were observed.