First off - Epidemiology, weak hazard ratios, correlation is not causation, can only inform on future research, should not be used for personal health decision due to its weakness.

That doesn't stop other diet camps from making broad claims on even weaker evidence... but it really should.

• Who is doing the research: Institute of nutrition, Fudan University, Shanghai, China
• On the Basis of what: Chinese Longitudinal Healthy Longevity Survey - Epidemiology/FFQ
• In What Context: Normal people living their lives in China, eating the standard chinese diet, rich in carbohydrates and seed oils

Why do I find this interesting, It provides a contrast to the unending torrent of epidemiology pushing a purely plant based diet. And it does demonstrate a improvement in older age health span with the increased consumption of animal foods.

According to my own Standards of Nutritional Evidence this is just a big nothing burger, but its some variety

Notes

Really well written paper, thoughtful, and aware of the data's limitations. Curiously since they were focusing on frailty they threw out death as a hard outcome out of scope.

Given that vegetarians could be a mix of people who adopted vegetarian diet either by choice or due to lack of resources, we conducted prespecified subgroup analyses stratified by socioeconomic factors including years of education, household income, and financial support separately.

It's a really good point, many of the people in this population were economic vegetarians, which has a whole slew of problems

• Not well formulated plant base
• Unable to source imported plants and supplements to round out the diet.

vegetarian participants were older, more likely to be female, urban, unmarried, living alone, and of lower socioeconomic status (less educated, with lower household income, and more likely to experience insufficient financial support); less likely to be ever-smokers, less likely to exercise, drank less alcohol, consumed less energy, and had lower plant-based dietary quality (Table 1).

lots of confounders... they try to control for them... but epidemiology controls are just assumptions based on assumptions until the model spits out a acceptable correction... JUST LIKE THE LLM WORKFLOW TO SOLVE PROBLEMS.... you wouldn't trust a LLM would you?

During a median follow-up of 3.0 (IQR: 1.83–5.33) years, vegetarians at baseline had a higher risk (HR = 1.13, 95% CI: 1.07–1.20) of developing frailty compared with omnivores, after adjusting for age, sex, residence, years of education, living arrangement, household income, financial support, marital status, exercise status smoking status, alcohol consumption, BMI, and pTEE

Similar patterns were observed when categorizing the vegetarian diet into three subgroups: pesco-vegetarians (HR = 1.15, 95% CI: 1.05–1.26), ovo-lacto-vegetarians (HR = 1.11, 95% CI: 1.02–1.20), and vegans (HR = 1.12, 95% CI: 1.01–1.25) (Table 2), where vegetarian diets were positively associated with risk of frailty.

TLDR: The more animal foods people ate (eggs, fish, etc) the less frail they became over time.

The association between vegetarian diet and frailty risk was stronger in males (HR = 1.18, 95% CI: 1.07–1.30) after stratifying by sex.

That is interesting... but only low socioeconomic males...

Upon further stratification by baseline frailty status, the association remained statistically significant only in participants who were non-frail at baseline (HR = 1.14, 95% CI: 1.05–1.24). However, in those classified as prefrail at baseline, the association became non-significant (HR = 1.07, 95% CI: 0.99–1.16) (Additional file 1: Table S6).

Almost like there is some buffer the body is maintaining over time, when the buffer is eaten up your cooked... So if you started without a buffer you didn't get much worse, but if you lost your buffer you got worse.

Compared with participants who maintained an omnivorous diet, those who had ever adopted a vegetarian diet exhibited higher risks of frailty (Stick-To-Vegetarian: HR: 1.19, 95% CI: 1.03–1.38; Omnivorous-To-Vegetarian: HR: 1.16, 95% CI: 1.04–1.30; Vegetarian-To-Omnivorous: HR: 1.14, 95% CI: 1.02–1.27), with those who adhered to a vegetarian diet throughout both surveys showing the highest risk (Fig. 2).

Going back to the buffer analogy, eating into the buffer is the risk... the more exclusive the diet is of animal foods, the smaller that buffer is as exposed by increased risk.

Interestingly, previous studies based on the same cohort studies reported associations between dietary patterns rich in plant-based food, as typically assessed by plant-based diet indices, and lower frailty risk, suggesting a protective effect of a more plant-based diet [10, 11].

I really appreciate how thoughtful the authors are here.

However, unlike plant-based diet indices, which produce quantitative scores that negatively weigh all animal foods and positively weigh all plant foods [35], our research primarily focused on a vegetarian diet that represented extreme cases of excluding specific or all animal-derived foods.

The games you can play with epidemiology... pizza clearly is meat and not carbs!

An intervention study reported that older men gained more fat-free and skeletal muscle mass on a meat-containing diet compared to an ovo-lacto-vegetarian diet [37].

Less carbs, more protein, make sense given the insulin model of obesity

More studies are warranted, given the controversial findings of plant-based diet and vegetarian diet.

I know EVERY research paper says this... but its actually nice to see this outlined in a epidemiology paper.

With lower total and animal protein intake as well as less bioavailable protein sources [20, 45], vegetarians were reported to have lower lean mass, decreased muscle creatine and creatinine level, and a consequent reduction in muscle mass, strength and quality, and physical function [45,46,47,48], indicating sarcopenia and ensuing frailty [45, 49].

Lean muscle mass going into older age is a key to health span!

Moreover, vegetarians showed a lower diet quality compared with omnivores in our study, even though the diet quality was constructed based on plant-based food groups only, which could partially explain our results. Our findings could also be explained by deficiencies in calcium, vitamin D, and vitamin B12 in vegetarian diets [12, 36], which are crucial for muscle and bone health related to frailty [53].

Those economic vegetarians again.

dietary assessment via the FFQ, in which participants were asked about the food intake frequency at present instead of a given time range, was prone to information bias.

I love these authors.

we could not rule out the potential residual confounding despite adjusting for multiple confounders

OMFG, the first time I've EVER seen a epidemiology paper admit this. I REALLY LOVE these authors.

However, although the association was considered statistically significant, the effect size was relatively small, and given the limitations of this study, particularly the potential imprecision in capturing dietary exposure, the reliance on self-reported indicators to construct the FI, and the inability to adjust for energy intake, these findings should be interpreted with caution.

This should be on EVERY epidemiology paper!