Abstract The application of dietary quality indices (DQIs) is limited by inconsistent methods and suboptimal reporting, but the effect of index composition heterogeneity on the strength of disease associations is not clear. A scoping review was carried out to identify evidence on the most common disease-related DQIs and assess the heterogeneity of their application and its possible implications. We systematically identified umbrella reviews, systematic reviews, and primary studies that investigated the association of DQIs with all-cause mortality, risk of or mortality from cardiovascular diseases, type 2 diabetes, and cancer among adults. Compositions of the prioritized DQIs were explored in primary studies and the degree of deviance from the original DQI versions was quantified. A meta-regression was conducted to investigate the association between the degree of score modification and disease risk estimates. From 175 eligible primary studies, 51 DQIs were identified and retrieved from 20 systematic reviews included in 2 umbrella reviews. The most common indices were the Mediterranean diet (MedDiet scores), Alternative Healthy Eating Index (AHEI), Dietary Approaches to Stop Hypertension (DASH), Healthy Eating Index (HEI), and Dietary Inflammatory Index (DII). Higher diet quality reflected by these indices was associated with beneficial health outcomes in 17 meta-analyses. Heterogeneity of MedDiet scores was not considered, because it has been addressed in previous reviews. Substantial heterogeneity was identified in the remaining DQI applications. Among studies that reported composition details, AHEI-2010 and DII were modified in almost all studies, and DASH and HEI-2015 in half of the studies. The underlying reasons were mainly related to population- and study-specific characteristics. DQI modifications did not appear to substantially influence the direction or strength of associations with mortality and disease risk. However, heterogeneity of index composition and its suboptimal reporting limit the reproducibility and comparability of results from studies on DQIs and health outcomes, and standard applications are preferred.
El-Khoury et al. (Wed,) studied this question.