Mediterranean diet indices are not interchangeable in Crohn's disease: Caution when using indices to measure adherence to dietary patterns

Abstract

Background: A Mediterranean diet is characterised by high intakes of fruits, vegetables, nuts, legumes, wholegrains, olive oil and fish and is of interest in inflammatory bowel disease due to its potential to positively impact gut microbiota and disease. Numerous indices are used to assess alignment with a Mediterranean diet but differ in foods included and thresholds of consumption. Additionally, scores are applied either using absolute cut-offs (e.g. score 1 if legume intake >=450g per week) or using relative cut-offs based on the median of the sample (e.g. score 1 if above median intake). The use of different indices may impact study comparability and cause inconsistent results when examining the impact of Mediterranean dietary pattern on IBD outcomes. This study aimed to examine the agreement between two commonly used Mediterranean diet indices in patients with active Crohn's disease. Method(s): Patients with mild-To-moderately active Crohn's disease (CDAI 150-250) provided 7-day food diaries that were coded using Nutritics software. Adherence to Mediterranean diet was calculated using two common indices: The Mediterranean Diet Adherence Screener (MEDAS, 0-14) based on absolute cut-offs and the Alternate Mediterranean Diet Score (aMED, 0-9) based on relative cut-offs. Association between the indices was evaluated using Spearman's correlation and agreement between tertiles of adherence and pre-defined adherence categories (low/moderate/high) was assessed using percentage agreement and weighted Cohen's Kappa (kappa). Result(s): 102 participants with active Crohn's disease were included (mean CDAI 186.1, SD 31.1; 63 female). Median MEDAS and aMED scores were 4 (IQR 2) and 4 (IQR 3). A moderate positive correlation between MEDAS and aMED scores was observed (rs = 0.61, p <0.001) (Figure 1). Agreement in assigning adherence was moderate when dividing participants into tertiles of score (56.9% agreement, weighted K = 0.54, 95% CI 0.40-0.69, p < 0.001) but only fair when divided into standard adherence categories (45.1% agreement, weighted K = 0.24, 95% CI 0.14-0.34, p = 0.028) (Figure 2). 27% of patients were assigned to high adherence category using aMED whereas none (0%) were using MEDAS (Figure 2). Conclusion(s): Despite correlation in scores between MEDAS (absolute cut-offs) and aMED (relative cut-offs), there was only fair agreement in categorising patients to low, moderate and high adherence to Mediterranean diet. These scoring systems for Mediterranean diet are not interchangeable. In active Crohn's disease, indices that use relative cut-offs to score may overestimate Mediterranean diet adherence, whereas absolute cut-offs may better reflect diet pattern regardless of the population.