The mediating role of combined lifestyle factors on the relationship between education and gastric cancer in the Stomach cancer Pooling (StoP) Project

Abstract

Background: The causal pathway between high education and reduced risk of gastric cancer (GC) has not been explained. The study aimed at evaluating the mediating role of lifestyle factors on the relationship between education and GC METHODS: Ten studies with complete data on education and five lifestyle factors (smoking, alcohol drinking, fruit and vegetable intake, processed meat intake and salt consumption) were selected from a consortium of studies on GC including 4349 GC cases and 8441 controls. We created an a priori score based on the five lifestyle factors, and we carried out a counterfactual-based mediation analysis to decompose the total effect of education on GC into natural direct effect and natural indirect effect mediated by the combined lifestyle factors. Effects were expressed as odds ratios (ORs) with a low level of education as the reference category.

Results: The natural direct and indirect effects of high versus low education were 0.69 (95% CI: 0.62-0.77) and 0.96 (95% CI: 0.95-0.97), respectively, corresponding to a mediated percentage of 10.1% (95% CI: 7.1-15.4%). The mediation effect was limited to men.

Conclusions: The mediation effect of the combined lifestyle factors on the relationship between education and GC is modest. Other potential pathways explaining that relationship warrants further investigation.

Fig. 1. Directed acyclic graph showing the relationship between education and gastric cancer risk and the decomposition of the effects.

Arrow A displays the natural direct effect (NDE) of education on gastric cancer risk, while path B + C displays the natural indirect effect (NIE) mediated by the healthy lifestyle score. The sum of NDE and NIE gives the total effect (TE). The last three arrows display the confounding variables.

Fig. 2. Lifestyle factors across levels of education.

Distribution of cigarette smoking (a), alcohol intake (b), fruit and vegetable intake (c), processed meat intake (d), salt intake (e) and healthy lifestyle score categories (f). Intakes of fruit and vegetable, processed meat and salt were categorized according to study-specific levels. Categories of lifestyle score were obtained from sex-specific tertiles.

Fig. 3. Results of the moderated mediation analysis: mediation effects of lifestyle score on the relationship between education and gastric cancer moderated by sex, age group and geographic area.

Total effect for intermediate (a) and high level (b) of education. Natural direct effect (NDE) for intermediate (c) and high level (d) of education. Natural indirect effect (NIE) for intermediate (e) and high level (f) of education. Test of interactions for moderated mediation effects: (1) for sex, p = 0.274 for NDE and p < 0.001 for NIE; (2) for age group: p = 0.034 for NDE and p = 0.623 for NIE; (3) for geographic area, p = 0.012 for NDE and p = 0.140 for NIE.