Nutri-Score 2023: Diet Quality and Mortality Risk

Nutrient profiling models translate complex dietary patterns into single scores that can guide public health messaging and front-of-pack labeling. In 2026, Khoury and colleagues published a prospective analysis in the European Journal of Nutrition using the updated 2023 Nutri-Score model on 7,212 high-cardiovascular-risk adults from the PREDIMED cohort, and found that worse diet quality predicted higher mortality from all causes and from cardiovascular disease.

Why This Study Matters

Nutri-Score is a front-of-pack labeling system already used in several European countries to summarize food healthfulness on a five-color scale from A (green, healthiest) to E (red, least healthy). The original model launched in 2017. In 2023, the underlying algorithm was updated to better align with current dietary guidelines -- the changes reclassified some foods, adjusted weighting for fats and oils, and recalibrated thresholds for sugar, sodium, and fiber.

Whenever a public health tool is updated, the new version needs validation. Does the revised algorithm actually predict the outcomes it is meant to predict? Does worse Nutri-Score correspond to worse health, prospectively, in real populations? The Khoury 2026 analysis was designed to test exactly that, using one of the most well-characterized European cohorts available: PREDIMED.

The PREDIMED cohort is older, Spanish, and at elevated cardiovascular risk by design. That makes it a useful proving ground for nutrient profiling systems intended to identify dietary patterns that affect mortality.

How It Was Designed

The study analyzed 7,212 participants aged 55 to 80 enrolled in PREDIMED, all at high cardiovascular risk at baseline. Diet was assessed using validated food frequency questionnaires. The updated 2023 Nutri-Score nutrient profiling model was applied to compute a dietary index (uNS-NPM DI) for each participant, summarizing overall diet quality based on intake of foods scoring favorably and unfavorably under the new algorithm.

The exposure was the cumulative average uNS-NPM DI across the follow-up period, divided into quintiles. Outcomes were all-cause mortality, cardiovascular mortality, cancer mortality, and death from other causes. Statistical analysis used time-dependent Cox regression with hazard ratios and 95% confidence intervals, adjusted for relevant confounders including age, sex, BMI, smoking status, physical activity, and other cardiovascular risk factors.

Median follow-up was 6 years. A total of 425 deaths were recorded -- 103 from cardiovascular causes, 169 from cancer, 153 from other causes. The analysis compared the highest quintile (worst diet quality) with the lowest quintile (best diet quality) and tested for linear trend across quintiles.

What They Found

Worse diet quality, as captured by the updated Nutri-Score index, was prospectively associated with significantly higher mortality from multiple causes.

Mortality Outcome	HR Q5 vs Q1 (95% CI)	Risk Direction	p-trend	Significant?
All-cause mortality	1.64 (1.19 to 2.28)	64% higher	0.007	Yes
Cardiovascular mortality	3.21 (1.29 to 7.95)	3.2x higher	0.002	Yes
Other-cause mortality	1.84 (1.11 to 3.07)	84% higher	n.s.	Q5 yes, trend no
Cancer mortality	1.16 (0.69 to 1.92)	16% higher	0.695	No

Red indicates higher risk in the worst-quality diet quintile compared with the best-quality diet quintile. HR values above 1.0 indicate higher risk. Models adjusted for age, sex, BMI, smoking, physical activity, and other confounders.

Reading the Results

The cardiovascular signal is the most striking. Participants in the worst quintile of diet quality had more than three times the cardiovascular mortality of those in the best quintile (HR 3.21, 95% CI 1.29 to 7.95). The confidence interval is wide because cardiovascular deaths were the smallest subgroup (103 events), but the point estimate and the p-trend (0.002) are consistent and statistically robust.

All-cause mortality scaled with diet quality. A 64% increase in death from any cause across the worst-versus-best diet quintile is a large effect for a single dietary exposure in an already-elderly, already-high-risk population. The p-trend of 0.007 indicates a graded relationship rather than a threshold effect.

The Nutri-Score 2023 algorithm validates as a mortality predictor. That is the headline methodological finding. A nutrient profiling tool intended for food labels actually correlates with downstream death rates in a real cohort. That gives the regulatory and public health communities a stronger empirical basis for using the updated algorithm in front-of-pack labeling.

Cancer mortality did not move. The 16% increase in cancer mortality was not statistically significant and showed no trend across quintiles. This is consistent with a broader literature in which short-term dietary patterns predict cardiovascular outcomes more reliably than cancer outcomes, which often reflect exposures over longer time horizons.

What Didn't Change

This is an observational analysis, not an intervention. The study cannot prove that adopting a higher-quality diet causes lower mortality -- only that the two are associated after adjustment for measured confounders. Residual confounding from unmeasured variables such as socioeconomic status, healthcare access, or psychosocial factors remains a real possibility.

The PREDIMED cohort is older and at high cardiovascular risk. Effects of similar magnitude may not generalize to younger or lower-risk populations. The cohort is also Spanish, with dietary patterns reflecting a Mediterranean baseline -- the Nutri-Score's ability to discriminate diet quality in this population may differ in cohorts with very different baseline diets.

Cancer mortality findings were null. The wide confidence interval (0.69 to 1.92) cannot rule out either a modest harm or a modest benefit. Diet-cancer relationships are likely heterogeneous across cancer types and may require larger samples, longer follow-up, or site-specific analyses.

Broader Context

Nutri-Score is one of several front-of-pack nutrient profiling systems used internationally. It is the official voluntary labeling system in France, Belgium, Spain, Germany, the Netherlands, Luxembourg, and Switzerland. The updated 2023 algorithm changed how fats are scored -- particularly oils, where the version 2 model now better distinguishes oils high in monounsaturated and polyunsaturated fats (such as extra virgin olive oil) from oils high in saturated fats.

That fat reclassification matters for olive oil specifically. The 2023 algorithm gives olive oil a more favorable score than the original 2017 version did, recognizing the cardiovascular evidence base for olive oil and other unsaturated cooking fats. The Khoury 2026 analysis supports the recalibrated model: a diet rich in foods scored favorably by Nutri-Score 2023 -- which includes high-polyphenol olive oil -- predicts lower mortality.

The PREDIMED main trial (2013) and its many sub-analyses have repeatedly shown that the Mediterranean dietary pattern, anchored by extra virgin olive oil, reduces cardiovascular events. The current analysis approaches the same question from a different angle -- using a standardized scoring algorithm rather than a randomized intervention -- and arrives at a consistent conclusion: diet quality, as captured by foods that align with cardiovascular best practice, predicts mortality.

Related Research

Continue exploring olive oil and polyphenol science:

• Olive Oil and 19% Lower Mortality in 92,383 U.S. Adults Over 28 Years
• Italian Cohort: Olive Oil Linked to Lower Cancer, CV and All-Cause Mortality
• American Heart Association 2021 Dietary Guidance: Olive Oil's Role

Source: View the original study on PubMed

Olivea's Dosage

This study did not test a supplement or a specific dose. It compared habitual dietary patterns scored by the updated Nutri-Score algorithm. Extra virgin olive oil is among the foods scored favorably under the 2023 update. Each Olivea capsule delivers concentrated olive polyphenols for those seeking a measured daily intake without consuming larger volumes of oil. Our most recent third-party certificate of analysis confirmed 23.5 mg of hydroxytyrosol per capsule.

According to PubMed, this study is indexed as PMID 41874696 (DOI: 10.1007/s00394-026-03946-4).

We share this research for transparency. This is an independent study -- we did not fund it, design it, or conduct it.

Editorial Information

Research note. This article summarizes third-party research published in a peer-reviewed journal. Olivea did not conduct or fund the study. Findings reflect the cited paper only and do not establish efficacy of Olivea products.

.