Comment on "Food Processing and Risk of Crohn’s Disease and Ulcerative Colitis: A European Prospective Cohort Study"

Abstract

We read with interest the study by Meyer et al,1Meyer A. et al.Clin Gastroenterol Hepatol.S1542-3565(22)00929-6Google Scholar which clarified the association between unprocessed/minimally processed food consumption and reduced risk of Crohn’s disease (CD) via a large cohort (the European Prospective Investigation into Cancer and Nutrition [EPIC]). This study was well-designed in terms of the measurement of exposures, because the authors precisely quantified the energy intake from both unprocessed/minimally processed and ultraprocessed food (UPF) using a validated questionnaire, and outcomes, because the diagnosis of CD or ulcerative colitis was based on physicians’ ascertainment. The result that not UPF but unprocessed/minimally processed food impacted the incidence of CD was quite novel and impressive. We believe this study will substantially contribute to patients’ nutritional care for preventing CD. However, the results were partially inconsistent with previous studies that the consumption of UPF was associated with the risk of CD2Lo CH et al.Clin Gastroenterol Hepatol. 2022; 20: e1323-e1337Abstract Full Text Full Text PDF PubMed Scopus (19) Google Scholar or ulcerative colitis.3Narula N. et al.BMJ. 2021; 374 (n1554)PubMed Google Scholar We would like to discuss why such inconsistency was observed, based on 2 possible biases. First, we would like to comment on confounding bias. Although this study adjusted for numerous important confounding factors including smoking or socioeconomic status, there may be some important residual ones that can be measured and should be adjusted for. For example, factors associated with gut microbiota, such as urban/rural residence or oral contraceptive use, can be possible residual confounding factors.4Piovani D. et al.Gastroenterology. 2019; 157: 647-659Abstract Full Text Full Text PDF PubMed Scopus (227) Google Scholar Additionally, race and ethnicity may be a potential and important confounding factor for the association between UPF consumption and the risk of inflammatory bowel disease (IBD).5Spekhorst L.M. et al.J Crohns Colitis. 2017; 11: 1463-1470Crossref PubMed Scopus (2) Google Scholar Without adjustment for such variables, the results might be based on reduced internal validity in the analysis of UPF. Second, we would like to comment on immortal time bias. Although the authors discussed possible selection bias because of the middle-aged individuals in this study, we can speculate this study may incorporate the risk of immortal time bias rather than selection bias.6Yadav K. et al.JAMA. 2021; 325: 686-687Crossref PubMed Scopus (53) Google Scholar Essentially, the peak age for the onset of IBD is approximately 15–25 years,7Johnston R.D. et al.Inflamm Bowel Dis. 2008; 14: S4-S5Crossref PubMed Google Scholar and this study only included participants who did not develop IBD during their young adulthood. Such a situation may have created an “immortal time” where all participants must be alive without IBD from their young adulthood to study enrollment. Immortal time bias may potentially underestimate the association between UPF consumption and the incidence of IBD.6Yadav K. et al.JAMA. 2021; 325: 686-687Crossref PubMed Scopus (53) Google Scholar Although immortal time bias may be inevitable because the study used the EPIC cohort and all participants were middle-aged, caution should be exercised when interpreting the results. We expect that future studies that include a wide age range of individuals can account for such an immortal time. In summary, this study may have some potential biases regarding internal validity, which could have contributed to a nonsignificant association between UPF use and the risk of IBD. However, we and the authors agree that the main results, namely, the association of unprocessed/minimally processed food consumption with the reduced risk of CD, exist. We would like to highlight the importance of further research that compensates for the issues mentioned previously.