Agreement among Mediterranean Diet Pattern Adherence Indexes: MCC-Spain Study

Olmedo Requena, Rocío
González-Donquiles, Carmen
Dávila-Batista, Verónica
Romaguera, Dora
Castelló, Adela
Amiano, Pilar
Dierssen Sotos, Trinidad
Guevara, Marcela
Lozano-Lorca, Macarena
Alguacil, Juan
Peiró Pérez, Rosana
Huerta, José María
Gracia Lavedan, Esther
Aragonés, Nuria
Fernández-Villa, Rosana
Solans Margalef, Marta
Gómez Acebo, Inés
Castaño-Vinyals, Gemma
Kogevinas, Manolis
Pollán, Marina
Martín, Vicente
There are many different methods used to measure the degree of adherence to a Mediterranean diet (MD), limiting comparison and interpretation of their results. The concordance between different methodologies has been questioned and their evaluation recommended. The aim of this study was to evaluate the agreement among five indexes that measure adherence to a Mediterranean dietary pattern. The study population included healthy adults selected in the Multi-Case Control Spain (MCC-Spain) study recruited in 12 provinces. A total of 3640 controls were matched to cases by age and sex. To reach the aim, the following scores of adherence to a Mediterranean dietary pattern were calculated: Mediterranean diet score (MDS), alternative Mediterranean diet (aMED), relative Mediterranean diet (rMED), dietary score (DS) and literature-based adherence score (LBAS). The relative frequency of subjects with a high level of adherence to a MD varied from 22% (aMED index) to 37.2% (DS index). Similarly, a high variability was observed for the prevalence of a low level of MD: from 24% (rMED) to 38.4% (aMED). The correlation among MDS, aMED and rMED indexes was moderate, except for MDS and aMED with a high coefficient of correlation 0.75 (95% CI 0.74–0.77). The Cohen’s Kappa coefficient among indexes showed a moderate–fair concordance, except for MDS and aMED with a 0.56 (95% CI 0.55–0.59) and 0.67 (95% CI 0.66–0.68) using linear and quadratic weighting, respectively. The existing MD adherence indexes measured the same, although they were based on different constructing algorithms and varied in the food groups included, leading to a different classification of subjects. Therefore, concordance between these indexes was moderate or low. ​
This document is licensed under a Creative Commons:Attribution (by) Creative Commons by4.0