A priori-defined Mediterranean-like dietary pattern predicts cardiovascular events better in north Europe than in Mediterranean countries

Abstract

Background: The Mediterranean Diet (MD) is a model of healthy eating contributing to a favorable health status, but its clinical usefulness is still debated. The aim of this study was to relate the adherence to MD with the incidence of cardio/cerebro-vascular events (VEs) in north and south European participants of the IMPROVE study.

Methods: IMPROVE is an observational, longitudinal, prospective cohort study involving 3703 individuals from five European countries (Finland, Sweden, Netherlands, France and Italy). The study end-point was the incidence of the first combined cardio/cerebro-vascular event occurring during 36-months follow-up. At baseline, a dietary questionnaire about the usual intake during the year preceding enrollment was administered. Based on 7 nutritional items, a MD Score was constructed in which minimal adherence was 0 and maximal adherence was 7.

Results: Latitude was the strongest determinant of MD score (p < 0.001). VEs occurred in 215 participants. The incidence of VEs was the highest in subjects with MD score 0-1, lower in those with score 2-3 and the lowest in those with score ≥ 4. MD score remained significantly associated with subsequent VEs after adjustment for confounders (hazard ratio for one-point increment of the score = 0.75, p < 0.001) and the association was stronger in northern than in southern countries (p = 0.04 for MD Score × latitude interaction).

Conclusions: The MD adherence score based on a simple dietary questionnaire detects changes of risk of VEs. According to our findings north Europeans appear to benefit most from VE-prevention when their diet is altered to the MD diet.

Keywords: IMPROVE study; Mediterranean diet score; Vascular diseases.

Fig. 1

Hazard ratios for combined cardiovascular events associated with one point increase of MD score: subgroups analysis. Horizontal lines represents 95%CI, adjusted for covariates in Model 3 (see Methods), excluding the respective stratification variables.