Exploring diet categorizations and their influence on flare prediction in inflammatory bowel disease, using the...

Diet is an important environmental factor in inflammatory bowel disease (IBD) onset and disease course, but analyses are hindered by its complexity. We aim to explore the Sparse Grouped Least Absolute Shrinkage and Selection Operator (Sparse Grouped LASSO or SGL) method to study whether different food categorizations, representing different dietary patterns, can predict flares in IBD.

Baseline data on habitual dietary intake and longitudinal data on disease course were collected over a 24 month-period in two distinct cohorts. Food items were classified into 22 food groups. These were further classified into three diet categorizations: 1. Plant vs animal vs mixed; 2. Potentially healthy vs potentially unhealthy vs neutral; 3. Ultra-processed vs not ultra-processed. The SGL parameter 'lambda' identifies important groups using a-priori group information, while allowing for only a subset of variables within a group to be important predictors.

Of 724 eligible patients, 427 were in remission at baseline and were included in the SGL analyses. 106 (24.8 %) included patients developed a flare within 11.2 ± 6.6 months (65.1 % female, 34 % ulcerative colitis, mean age 43.3 ± 14.7 years). They had a higher crude food intake of red meat (p = 0.028) and vegetables (p = 0.027) than those who stayed in remission. Prediction models for flare development were moderate with AUC varying between 0.425 and 0.542 for model 1, 0.512 and 0.562 for model 2 and 0.451 and 0.612 for model 3. All models showed red meat, legumes and vegetables as the first selected predicting variables. However, female sex and energy intake had the highest predictive values in all 3 models.

Categorization of the same food groups in different ways influences the predictive value of the SGL method. The current exploration of the SGL method shows that food might not be the most important predictor of flares in IBD.