Blood-Based Diet Scores Also Predict Lower Mortality

Nutrition research has a familiar weak spot: food questionnaires. People forget, round down, and occasionally remember their diet as the person they meant to be on Monday. That error matters. But a Melbourne study found that diet-quality scores reconstructed from inflammatory blood markers still predicted all-cause mortality, suggesting the diet-survival association does not depend entirely on what participants report eating.

This does not prove that a Mediterranean-style diet causes longer life. The strongest mortality evidence remains observational, and the blood-marker scores were calibrated against questionnaire-based scores. Still, several lines of evidence now point in the same direction: better overall diet quality tracks with lower mortality across different scoring systems, while randomized trials support benefits for cardiovascular outcomes and risk factors.

Blood markers offered a different way to test the question

Food-frequency questionnaires have known limits. A validation study of questionnaire-derived diet-quality scores documented recall error, systematic underreporting, and social-desirability bias. Those problems can blur real associations or create misleading ones if healthier reporters also differ in smoking, exercise, income, medical care, or other behaviors.

In the Melbourne Collaborative Cohort Study, researchers examined a subcohort of 770 participants, among whom 249 died during a median 14.4 years of follow-up. They used plasma inflammatory-marker profiles to derive proxies for three dietary patterns: the Mediterranean Diet Score, the Alternative Healthy Eating Index, and the energy-adjusted Dietary Inflammatory Index.

Each biomarker-derived score predicted all-cause mortality in the expected direction. Per one-standard-deviation increase, the Mediterranean-pattern marker score had a hazard ratio of 0.84 (95% CI, 0.72-0.98), and the Alternative Healthy Eating Index marker score had a hazard ratio of 0.76 (95% CI, 0.66-0.89). Higher scores on the inflammatory-index marker profile predicted higher mortality: hazard ratio 1.18 (95% CI, 1.01-1.39).

Those confidence intervals exclude 1.0, so these associations were statistically significant. The study authors also reported that questionnaire-derived dietary patterns did not show effects of the same magnitude. That result is consistent with the possibility that blood markers captured a biologically relevant signal that self-report measured poorly.

It is not a clean escape from the questionnaire problem, however. The investigators derived the biomarker scores by regressing plasma markers against food-frequency-questionnaire scores. If the questionnaires contain systematic bias, some of that bias may shape the biomarker weights. The same cohort also supplied both score derivation and mortality testing, which raises the risk of overfitting. External validation in an independent cohort would make this finding more persuasive.

Three diet scores converged in a larger Australian cohort

A separate Australian Diabetes, Obesity and Lifestyle Study analysis addressed a related concern: perhaps results depend on how researchers construct a "healthy diet" score. The study followed 10,009 adults for a mean 17.7 years and recorded 1,955 deaths from any cause and 520 cardiovascular deaths.

Researchers applied three different indices: the Diet Guidelines Index, the MIND diet score, and the Dietary Inflammatory Index. After adjustment for age, sex, education, smoking, physical activity, energy intake, diabetes, and hypertension, all three remained associated with all-cause mortality. For each standard-deviation increase, the hazard ratios were 0.94 (95% CI, 0.89-0.99) for the Diet Guidelines Index, 0.93 (95% CI, 0.89-0.98) for the MIND score, and 1.08 (95% CI, 1.02-1.15) for the Dietary Inflammatory Index.

The direction is consistent across metrics: diets that better match dietary guidelines or MIND-style patterns tracked with lower mortality, while a more inflammatory dietary score tracked with higher mortality. The effect sizes were modest—roughly 6% to 8% per standard deviation—not a license to promise immortality through chickpeas. They are also relative associations, not absolute changes in an individual's chance of dying.

The study found limited evidence that diet quality predicted nonfatal cardiovascular events. That mismatch matters. If diet quality lowers cardiovascular mortality, one might expect fewer nonfatal heart attacks and strokes as well. The finding could reflect an effect on survival after an event, effects distributed across several causes of death, limited power for specific nonfatal outcomes, or residual confounding. It means the mortality association should not be reduced to a simple story about preventing plaque.

Possible pathways are measurable, but not proven causal

The Melbourne marker profiles repeatedly pointed toward the tryptophan-kynurenine pathway. Five of six inflammatory markers shared across the three blood-derived diet scores belonged to that pathway, and healthier dietary patterns correlated with lower concentrations of those markers, along with lower C-reactive protein and pro-inflammatory cytokines.

Kynurenine biology offers a plausible explanation, not a completed causal chain. Chronic inflammation can activate indoleamine 2,3-dioxygenase 1, or IDO1, which converts tryptophan to kynurenine. Experimental and mechanistic work links this pathway with immune activation, oxidative stress, and cellular senescence. Yet no dietary intervention trial has directly shown that changing diet lowers mortality through IDO1 or kynurenine metabolites.

A PREDIMED substudy adds relevant human evidence: tryptophan-kynurenine pathway metabolites were associated with incident heart failure and atrial fibrillation, and Mediterranean-diet adherence may have modified those associations. It connects the pathway to clinical cardiovascular endpoints, but it does not establish that the diet changed mortality through the pathway.

The larger Women's Health Study analysis followed 25,315 initially healthy women for 25 years and found that greater Mediterranean-diet adherence predicted lower all-cause mortality. Its mediation analysis identified inflammatory biomarkers, lipoprotein insulin resistance, and small-molecule metabolites including branched-chain amino acids as pathways that accounted for part of the association. Mediation analysis can quantify whether markers fit the observed statistical pathway; it cannot remove all confounding or prove that altering one marker will extend life.

The paper also cited a meta-analysis of 29 observational studies involving 1,676,901 participants, in which a two-point increase in Mediterranean Diet Score was associated with about a 10% lower all-cause mortality risk. That scale of evidence improves precision, but pooling observational studies does not turn association into a randomized trial.

What randomized trials can—and cannot—settle

Randomized evidence supports the direction of the cardiovascular story. A 2024 network meta-analysis of dietary interventions found that dietary interventions, with Mediterranean diets prominent, improved cardiovascular outcomes. A 2026 systematic review and meta-analysis likewise supported Mediterranean-diet efficacy for cardiovascular-disease prevention in patients.

Trials in people with type 2 diabetes also report improvements in glycemic control, lipid profiles, and blood pressure with Mediterranean dietary interventions, according to a 2025 meta-analysis of randomized trials. Another randomized-trial review found that dietary patterns can improve inflammatory biomarkers in adults with type 2 diabetes.

These are meaningful results, but they answer a narrower question than the mortality cohorts. The major diet trials were generally powered for cardiovascular outcomes or risk factors, not all-cause mortality. Biomarkers are useful clues, not patient outcomes. Better glucose control or lower inflammation does not by itself prove a longer life.

The practical verdict

The case for diet quality is stronger than "people who report eating vegetables also report being virtuous." Blood-derived inflammatory profiles predicted mortality in the Melbourne cohort, and three differently designed diet scores predicted mortality in AusDiab. Long-term data in women also link Mediterranean-diet adherence with lower mortality and identify plausible metabolic and inflammatory pathways.

Residual confounding remains the central problem. There is no Mendelian-randomization evidence here for anti-inflammatory dietary patterns and all-cause mortality, no randomized trial designed to establish an all-cause-mortality benefit, and no direct trial showing that kynurenine modulation causes the survival association.

For eating, the evidence supports steady improvement rather than dietary theater: more fruits, vegetables, legumes, whole grains, fish, and unsaturated fats; less red meat and ultra-processed food. No individual nutrient or supplement has reproduced the mortality signal of the overall pattern. A plasma kynurenine-to-tryptophan ratio or neopterin result may be an interesting inflammatory readout, but it is not a dietary report card or a longevity guarantee.