Background: Epidemiologic studies have identified numerous dietary and lifestyle risk factors for Crohn’s disease (CD). However, the causal relationship between these factors and CD is unknown. Randomized trials of lifestyle interventions are unlikely to be performed due to limited feasibility. We thus used observational data to emulate a target trial, a hypothetical, pragmatic randomized trial, to estimate the effect of a guideline-based healthy diet and lifestyle on CD risk.
Methods: We emulated a target trial using data from the prospective Nurses’ Health Study (NHS: 1990-2010), NHSII (1995-2015), and Health Professionals Follow-up Study (HPFS: 1990-2010) cohorts. At baseline, eligible individuals had no inflammatory bowel disease, complete covariate data, plausible dietary intake (females: 600-3500 kcal/d; males: 800-4200 kcal/d) and body mass index (BMI) ≥10kg/m². We used the parametric g-formula to estimate 20-year risks of CD, risk differences (RD) and risk ratios (RR) under a sustained, American Heart Association guideline-based healthy diet and lifestyle intervention (Table 1) vs no intervention (usual diet and lifestyle). To emulate randomization, we adjusted for age, sex, appendectomy, non-steroidal use, BMI, and prior values of smoking, physical activity, total energy intake, and intake of fruit, vegetables, whole grains, fish, sugary beverages, and processed meat. We also explored the effect of individual components of the intervention, adherence to the guideline-based intervention vs non-adherence, and varying durations of the intervention on CD risk.
Results: At baseline, 212,458 individuals were eligible (NHS 79,458, NHSII 93,504, HPFS 39,496). In follow-up, 292 participants developed CD. The 20-year risk of CD was 0.13% (95%CI 0.02-0.36) under the guideline-based intervention and 0.16% (0.14-0.18) under usual diet and lifestyle (RD: -0.03 [-0.14-0.19], RR: 0.81 [0.12-2.26]; Table 1). The effect of the guideline-based intervention was more pronounced when compared to non-adherence, though estimates were imprecise (RD: -0.12 [-0.49-0.18], RR: 0.52 [0.06-3.30]). Among individual components, fruit intake (≥2/d) was associated with reduced CD risk (RD: -0.02 [-0.04-0.00], RR: 0.87 [0.74-1.00]) with evidence of a dose-response relationship. Vegetable intake and smoking abstinence also showed modest reductions in CD risk. Finally, increasing duration of the guideline-based and fruit interventions was associated with lower CD risk (Table 2).
Conclusion: Adherence to some components of a guideline-based healthy diet and lifestyle intervention, including fruit and vegetable intake and smoking abstinence, may reduce risk of CD, with fruit intake showing the strongest, dose-dependent effect. If replicated in younger populations and high-risk individuals, these modifications may be attractive and low-risk targets for future efforts to prevent CD.

Emma McGee

Appearances