# Does a high-protein diet shorten your lifespan? Two readings of one survey

October 9, 2026 · Nutrition & Macros · 7 min read · https://burnweek.fit/blog/does-high-protein-shorten-lifespan/

> The famous protein-mortality study found the opposite effect after 65. Then an industry-funded team reanalyzed the same survey and found no higher risk.

**Key takeaways**

- Levine 2014 linked ≥20% calories from protein to HR 1.74 all-cause mortality at ages 50–65, but HR 0.72 over 65.
- A 2025 reanalysis of the same NHANES III survey found no association of animal or plant protein with higher mortality (animal protein had a small inverse cancer link).
- The 2025 study was industry-funded per the authors' disclosure; the 2014 senior author, Valter Longo, held equity in a medical-food company.
- A 2020 meta-analysis of 31 cohorts (715,128 people) tied higher total protein to 6% lower all-cause mortality.

## These studies do not establish that higher protein shortens human life

Will a high-protein diet shorten your lifespan? The evidence does not show it. The worry traces mainly to one 2014 analysis that linked high protein intake in people aged 50 to 65 with higher death rates. A 2025 reanalysis of the same national survey, using a different way of estimating what people habitually eat, found no association with higher mortality, and a 2020 meta-analysis of 31 cohort studies found higher total protein associated with slightly *lower* mortality. Every one of these studies is observational, so none can prove that protein caused or prevented anything.

That leaves a reasonable reading: eating enough protein to hit a sensible daily target, as described in [how much protein per day](/blog/how-much-protein-per-day), is not a known trade-off against longevity. It also leaves one caveat worth stating plainly: finding no association in survey data is not the same as proving long-term safety at any intake.

## The 2014 alarm: a signal confined to ages 50 to 65

The study behind the headlines is Levine and colleagues, published in *Cell Metabolism* in 2014. They were testing an idea from animal biology: mice and humans with deficient growth-hormone/IGF-1 signaling show fewer age-related diseases, and protein restriction lowers IGF-1, so perhaps high protein intake speeds aging ([Levine et al., 2014](https://pubmed.ncbi.nlm.nih.gov/24606898/)).

To test it in people, they used the third US National Health and Nutrition Examination Survey (NHANES III, 1988 to 1994): 6,381 adults aged 50 and over, followed for up to 18 years. "High" protein meant 20% or more of calories from protein; "low" meant under 10% ([Levine et al., 2014](https://pmc.ncbi.nlm.nih.gov/articles/PMC3988204/)).

The results split sharply by age:

| Age group (Levine 2014) | High vs low protein, all-cause mortality | High vs low protein, cancer mortality |
| --- | --- | --- |
| 50–65 (n = 3,039) | HR 1.74 (95% CI 1.02–2.97) | HR 4.33 (95% CI 1.96–9.56) |
| 66 and older (n = 3,342) | HR 0.72 (95% CI 0.55–0.94) | HR 0.40 (95% CI 0.23–0.71) |

In the middle-aged group, high protein went with a 74% higher hazard of overall mortality and roughly a fourfold increase in cancer death. Adjusting for animal protein removed the all-cause association and weakened the cancer association. In the over-65 group, the direction flipped: high protein went with *lower* overall and cancer mortality. High protein was also associated with a fivefold increase in diabetes mortality across the combined sample aged 50 and older ([Levine et al., 2014](https://pubmed.ncbi.nlm.nih.gov/24606898/)).

Notice what this means for the headline version. Even taken at face value, the 2014 paper did not say "high protein shortens life." It said something closer to "high protein in late middle age looked harmful, and high protein in older age looked protective." The authors' own suggestion was low protein in middle age followed by moderate to high intake later.

> The paper that launched the worry found that high protein went with *lower* mortality after 65.

The authors also listed the limits. Diet came from a single 24-hour recall, which they acknowledged could misclassify people's usual eating; some subgroup confidence intervals were wide; and there were no repeated diet measurements over the 18 years. The senior author, Valter Longo, disclosed an equity interest in a company that develops medical foods ([Levine et al., 2014](https://pubmed.ncbi.nlm.nih.gov/24606898/)), a conflict in the direction of the paper's thesis.

## The 2025 reanalysis: same survey, different analysis

In 2025, Papanikolaou, Phillips and Fulgoni went back to NHANES III with a larger sample: 15,937 adults (3,843 deaths per the abstract), mortality followed through 2006. Instead of a single day's recall sorted into percent-of-calorie bands, they modeled each person's *usual* intake with statistical methods designed to smooth out day-to-day noise, and they tested protein per gram as well as by intake tertile. Several analytical choices changed, so the disagreement cannot be attributed to intake measurement alone ([Papanikolaou et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40418846/)).

They found no association between animal protein (hazard ratio, HR, 0.99 per gram, 95% confidence interval, CI, 0.98–1.01) or plant protein (HR 1.02, 95% CI 0.95–1.10) and higher all-cause mortality, and none with cardiovascular death. Animal protein showed a small inverse association with cancer mortality (HR 0.95 per additional gram a day, 95% CI 0.91–1.00, p = 0.04). Blood insulin-like growth factor 1 (IGF-1), measured in 5,753 people, was not associated with any mortality outcome, and the results held when the authors split the sample into under-65, 50-to-65 and over-65 groups ([Papanikolaou et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40418846/)).

Two disclosures belong right next to those numbers. The work was industry-funded (per the authors' disclosure): two authors run consultancies serving food and beverage companies, and the third reported grants from dairy and food industry bodies among others ([Papanikolaou et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40418846/)). Independent replication would strengthen confidence in either result.

## Where the two analyses actually disagree

This is not two surveys of different people reaching different numbers. It is the same survey analyzed two ways, so the disagreement can be located:

- **How intake was measured.** Levine used one day of recall per person; the 2025 paper modeled usual intake. A single day is a noisy picture of someone's diet, which is the problem the 2025 authors set out to fix.
- **How protein was expressed.** Levine compared bands of percent of calories (under 10%, 10–19%, 20% and up); the 2025 analysis tested grams per day and tertiles. Percent of calories also moves when carbohydrate or fat intake changes, so it is not a pure protein measure.
- **Who was included.** Levine studied adults 50 and older; the 2025 paper included all adults and then checked age subgroups. Adjustment variables also differed, and the 2025 authors could not fully explain the gap.

Neither approach is free of confounding. People who eat very different amounts of protein also differ in what else they eat, how much they weigh and how healthy they already are. That is why the most defensible summary is that the 2014 signal did not replicate under a different method, not that the 2025 analysis proved safety.

## The wider cohort evidence points the other way

A single dataset, however well analyzed, is still one dataset. The broader picture comes from a 2020 BMJ meta-analysis of prospective cohort studies: 31 studies, 715,128 participants, 113,039 deaths over follow-up periods of 3.5 to 32 years ([Naghshi et al., 2020](https://pubmed.ncbi.nlm.nih.gov/32699048/)).

- Highest versus lowest total protein intake was associated with a **6% lower** risk of all-cause death (pooled effect 0.94, 95% CI 0.89–0.99).
- Plant protein was associated with lower all-cause (0.92) and cardiovascular (0.88) mortality.
- Animal protein was not significantly associated with cardiovascular or cancer death.
- Each additional 3% of energy from plant protein was associated with a 5% lower risk of death from all causes.

The authors reported no relevant financial conflicts. Heterogeneity for total protein and all-cause mortality was moderate (I² = 58.4%), and the same limit applies: cohorts cannot separate protein from the foods and lifestyles that come with it. The plant-protein association may reflect the wider diet that plant protein arrives in, a question taken up in [plant vs animal protein](/blog/plant-vs-animal-protein).

## What this means if you are tracking protein

Put the three sources side by side and the picture is fairly stable. The one analysis that raised alarm found the all-cause and cancer signal only in ages 50 to 65, found the reverse after 65, and did not replicate when the same data were measured differently. The pooled cohort evidence leans slightly toward benefit for total protein and more clearly toward plant protein.

What remains uncertain:

- **Very high, sustained intakes.** Few people in these surveys ate far above typical amounts for decades, so survey data say little about the extremes. The practical upper limits are covered in [how much protein is too much](/blog/how-much-protein-is-too-much).
- **Age as a moderator.** The one consistent thread in Levine's data is that age changed the direction of the association. After 65, the case for adequate protein is stronger, for reasons set out in [protein for older adults](/blog/protein-for-older-adults).
- **Source.** If you want to hedge, the cohort data give more support to shifting some protein toward plant sources than to eating less protein overall.

None of this is a reason to fear a protein target set in the usual ranges. It is a reason to hold both the 2014 alarm and the 2025 reassurance at arm's length, and to weight the larger body of evidence over either single paper.

## FAQ

### Did the 2014 Levine study find high protein harmful at every age?

No. High protein (20% or more of calories) was associated with higher all-cause and cancer mortality only in people aged 50 to 65. In those over 65 it was associated with lower mortality (all-cause HR 0.72) ([Levine et al., 2014](https://pubmed.ncbi.nlm.nih.gov/24606898/)).

### Who funded the 2025 study that found no link between protein and higher mortality?

It was industry-funded (per the authors' disclosure), and the authors reported food-industry consulting and grants from dairy and food bodies ([Papanikolaou et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40418846/)).

### Were blood IGF-1 levels associated with mortality in this analysis?

In the 2025 NHANES III analysis, blood IGF-1 levels in 5,753 adults were not associated with all-cause, cardiovascular or cancer mortality in any age group ([Papanikolaou et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40418846/)). That is one observational dataset, not a final answer on the IGF-1 hypothesis.

### Does plant protein look better than animal protein for longevity?

In pooled cohort data, the highest versus lowest plant-protein intake was associated with 8% lower all-cause mortality and 12% lower cardiovascular mortality, while animal protein showed no significant association either way ([Naghshi et al., 2020](https://pubmed.ncbi.nlm.nih.gov/32699048/)). Cohorts cannot rule out that the surrounding diet explains the difference.

## Sources

- [Levine ME, Suarez JA, Brandhorst S, et al. Low protein intake is associated with a major reduction in IGF-1, cancer, and overall mortality in the 65 and younger but not older population. Cell Metab. 2014;19(3):407-417.](https://pubmed.ncbi.nlm.nih.gov/24606898/)
- [Levine ME et al. 2014. Full text: PMC3988204.](https://pmc.ncbi.nlm.nih.gov/articles/PMC3988204/)
- [Papanikolaou Y, et al. Animal and plant protein usual intakes are not adversely associated with all-cause, cardiovascular disease-, or cancer-related mortality risk: an NHANES III analysis. Appl Physiol Nutr Metab. 2025.](https://pubmed.ncbi.nlm.nih.gov/40418846/)
- [Naghshi S, Sadeghi O, Willett WC, Esmaillzadeh A. Dietary intake of total, animal, and plant proteins and risk of all cause, cardiovascular, and cancer mortality: systematic review and dose-response meta-analysis of prospective cohort studies. BMJ. 2020;370:m2412.](https://pubmed.ncbi.nlm.nih.gov/32699048/)

Source: BurnWeek — "Does a high-protein diet shorten your lifespan? Two readings of one survey", https://burnweek.fit/blog/does-high-protein-shorten-lifespan/. Licensed CC BY 4.0: free to quote or reuse with a link to this page.