Fruit and vegetable intake and the risk of overall cancer in Japanese | Hacker News Reader

Fruit and vegetable intake and the risk of overall cancer in Japanese

sciencedirect.com

Did anyone else read the statistical analysis section? There were some "interesting"/unclear choices made, but I'm going to pick on one in particular that could affect the outcome: the decision on how to weight the individual cohort studies on which this study is based (pg 6).

> A random-effects model was used to obtain a single pooled estimate of the HRs from the individual studies for each category. The study-specific HRs were weighted by the inverse of the sum of their variance and the estimated between-studies variance component.

First note that there is no citation which motivates this weighting scheme. There may be a valid reason for doing it this way, but that's not clear from the paper.

Second, "the sum of their variance" I take to mean the sum of the variance components of the regression model. That's interesting because the components mean very different things. In the simple case of least squares, there's the "explained variance": the "strength" of the relationship between Y and X, and there's also the residual variance: the variance in X that isn't explained by Y. Aside from the fact that they mean different things, their sum is not a constant and so are not directly comparable across models/studies.

Third, weighted by the "between-studies variance component" has the effect of down-weighting studies that show a large difference from the mean study result and up-weighting studies which are similar. Now, what would you say if I conducted an experiment and transformed the raw data points according to this same procedure? "That's not fair because that's not the data we collected!"[0] And herein lies the problem: the purpose of this meta-analysis is to assess whether we can draw any consistent conclusions from the results of the individual cohort studies, but by weighting studies toward the mean the authors are taking it as a forgone conclusion that the mean study is ground truth.

I'm by no means an expert on meta-analysis so someone please chime in if you think I'm off base, but this all strikes me as a very peculiar and unjustified set of steps being taken before the model is even fit.

[0] Yes, there are cases for data transformation, motivated either by theory or meeting regression assumptions, but this is neither.

This is strange. Even if vegetable intake is not beneficial against cancer - should higher veg consumption not crowd out red meat (I think you eat less meat when eating more vegetables), leading to less cases of colon cancer? Also, what about vegetables with high fiber that gets converted to butyrates, and is thus anti-cancer? Is it all BS? Or is "overall cancer" a metric where high fiber might decrease colon cancer but leads to more cancers in other areas?

Have always been wondering about the supposed actual mechanism of how any meat would lead to colon cancer more/easier/likelier than fibrous foodstuffs, given that it is fibers that tend to end up in the colon whereas meat tends to be entirely liquefied in the stomach (not just softened and mashed as most plant foods are) and the resulting liquid near-on fully absorbed in the small intestines. Especially as the colon's entire purpose is to hold refuse and foreign substances. So somehow a few drops of leftover meat-liquid micro/macronutrients create more damage than roughage and seeds and such? Curious to hear from the better-informed non-epidemiologists out there ;)

>There are several potential reasons to explain how red and processed meat may cause bowel cancer. Red meat contains a compound which gives it’s red colour, haem, which promotes the formation of potentially carcinogenic N-nitroso compounds. In addition when red meat is cooked at high temperatures, this results in the production of compounds (heterocyclic amines and polycyclic aromatic hydrocarbons) that can cause bowel cancer in people with a genetic predisposition. In addition processed red meat has nitrites, as well as nitrates added to it as preservatives and these are thought to cause cancer. Processing also changes the nature of the meat, which may play a role in its link to cancer.

http://www.wcrf.org/int/blog/articles/2015/10/red-meat-and-b...

Interesting theories! And appropriately cautiously formulated, too.. a lot more to actually investigate then:

> how red and processed meat may cause bowel cancer

> compounds that can cause bowel cancer in people with a genetic predisposition

> nitrites, as well as nitrates and these are thought to cause cancer

> which may play a role in its link to cancer.

Do you know if steamed or boiled meats generate HAs and PAH?

I've know about both for a while and switched to only eating meat I've steamed or boiled.

http://cancerpreventionresearch.aacrjournals.org/content/4/2...

>The mechanism is not known, but heme iron has a catalytic effect on (i) the endogenous formation of carcinogenic N-nitroso compounds and (ii) the formation of cytotoxic and genotoxic aldehydes by lipoperoxidation. A review of evidence supporting these hypotheses suggests that both pathways are involved in heme iron toxicity

The irony is, of course, heme iron is the highest bioavailability form of iron, and is one of the main reasons to consume meat in the first place:

http://europepmc.org/abstract/med/3290310

> Food iron is classified as heme or nonheme. Approximately half of the iron in meat, fish, and poultry is heme iron. Depending on an individual's iron stores, 15% to 35% of heme iron is absorbed. Food contains more nonheme iron and, thus, it makes the larger contribution to the body's iron pool despite its lower absorption rate of 2% to 20%.

possibly acids making it from the stomach to the colon?

AFAIK is red meat not causing cancer per se. While there has been a correlation between red meat intake and colon cancer shown, the studies that have shown this didn't differentiate between highly processed products made out of red meat (think of sausage) and rather natural meat (think of steak) itself. The processed stuff is packed with all kinds of sugar, preservatives, flavor enhancer and colouration which might be the cause for the higher rates of colon cancer. To my knowledge these highly processed meat products are not common in Japan.

i think the most important take away with nutrition is eat what your ancestors ate. you're probably not japanese so its less likely that the results apply to you.

I myself switched from drinking soda to OJ. my health declined slowly for a long time until I found out my body doesnt like oranges.

fruit can be healthy but i doubt we are supposed to eat all fruit diets. Astin Kutcher took on Steve Job's all fruit diet and ended up in the hospital with pancreatitis. And we all know what happened to Steve Jobs.

OJ is just sugar water though, the value of fruit is (I think) the whole fruit.

true, but i have an even stronger alergic reaction to the fruit :)

its also worth noting that fruit is sugar water. its 98-99% water + 2% sugar and fiber

> its also worth noting that fruit is sugar water. its 98-99% water + 2% sugar and fiber

Oranges, apples, etc are only 85% water. Grapes 80%. The rest is fiber and is likely why your body tells you to stop eating them. Two and a half apples is a lot less appealing than a can of soda (actual sugar water).

The whole percent water thing is completely meaningless though, human brains are 85% water and babies are 80%.

The proportion of water is certainly not meaningless when considering the expected caloric intake of a food administered ad libitum. Satiety is partially controlled by stretch receptors in the gut lining and common sense will tell you it's much harder to consume the same amount of calories from a 2g/100mL sugar solution as from a 10g/100mL solution.

ah, i grossly over estimated the % water in those fruits. interestingly though, fruit juices fall in the 80-90% range I just read.

Hm. I've heard lots of different angles on not eating fruit but never got these numbers. Thanks!

And the sugar is the wrong one - fructose which is metabolized mainly by the liver whereas glucose is used by all cells in the body.

What about the ones that include the pulp?

the ones with pulp are still providing you with a massive amount of sugar relative to pulp

you can easily drink the equivalent of 12 oranges, in one drink but very very few people can eat 12 oranges in one sitting

there is wisdom in that nature packages sugar with fibre, at times lots and lots of fibre (cane sugar for example)

Cane sugars good and/or not too bad? We used to munch on that as kids, although more often than not we just sucked on it, so probably not so good.

Sucked from cane is not too bad; in that form it has a lot of other nutrients that make it easier for the body to handle.

When it's refined into crystals, is when it does more damage, particularly when consumed to excess.

No difference chemically between cane sugar and sugar from beet. It's sucrose i.e., glucose and fructose joined together.

Depending on how many years you want to go back in your diet, it might be quite hard to find it, artificial selection has rendered most of the food our ancestors ate non-existent.

Dude, OJ isn't healthy.

I tried OJ once, first breakfast in US, OJ instead of milk, puked exactly one hour later, never drank it again. So much for the health benefits of fruit juice.

Any thoughts on what happened? Health benefits aside, vomiting after drinking orange juice is hardly a typical reaction.

Which ancestors? When?

By some accounts my ancestors have been eating potatoes and roast meat with quite a lot of beer and / or apple cider, for quite some time.

I really liked this. I wish more negative results such as this were published and discussed, especially when they go against general intuition and feel-good narratives.

How are they controlling for deaths from other causes, e.g. cardiovascular disease? If you're not dead from some other cause then you're going to be more likely to die from cancer, even if vegetables are actually conferring a significant anti-cancer benefit.

I see that they are using people years to track followup, but I'm not clear from the methodology whether this is enough to fully control for this issue.

Are you stating that death from cardiovascular disease would be higher (or lower) from people who don't eat fruits/vegetables which means they have a "lower" chance of getting cancer?

My guess would be that they didn't take that into account and that the deaths from cardiovascular (or other diseases) would be ignored in the assumption they are "similar" between the two sides.

To rephrase, I'm suggesting that eating vegetables may lower the risk of cardiovascular disease.

If you don't die from cardiovascular disease, then it's more likely that cancer is the thing you do die from. So even if eating vegetables reduces your risk of cancer, you could actually be more likely to die from cancer if you eat vegetables.

Ahh thanks for the clarification. Your logic makes a lot of sense. I believe there was a study that showed pesticides may cause cancer (it was probably done in mice or something) from a long time ago but this is just the top of my head so I don't know if there's any relation to fruit/veggie consumption.

Anyone, is there also a study like this where the consumption of red meat (or meat in general) affects the risk of overall cancer?

The WHO's International Agency for Research on Cancer (IARC) released a report evaluating the link between the consumption of red and processed meat and cancer in 2015.

Here is a good overview of the report from the NHS which also looks at how the report was covered by the UK press:

http://www.nhs.uk/news/2015/10October/Pages/Processed-meat-c...

We know that processed red meat causes cancer.

The quality of the evidence is very very good.

The strength of the effect is weak.

Because the quality of the evidence is now very good (as good as for smoking) the WHO information was widely misreported as "red meat as bad as smoking".

This is likely to make web searches less useful.

Here's something from the English NHS about it: http://www.nhs.uk/news/2015/10October/Pages/Processed-meat-c...

> A key statistic provided by the infographic is that if everyone stopped smoking, there would be 64,500 fewer cases of cancer a year in the UK, compared with 8,800 fewer cases if everyone stopped eating processed or red meat.

> If everyone stopped smoking, there would be 64,500 fewer cases of cancer a year in the UK, compared with 8,800 fewer cases if everyone stopped eating processed or red meat.

Even that statistic greatly overstates the risk of red meat as compared to smoking. Because it's not adjusted for the percentage of the population that smoke vs the percentage that eat processed or red meat.

In the UK, only 16.9% currently smoke[1] but I believe that the great majority eat meat[2]. (I'm surprised how few people smoke now -- I remember a time when most people smoked.)

The statistic makes it sound like you're 7 times more likely to die from smoking than from processed or red meat (64500/8800 = 7). But if my math is correct, the death ratio should be more like smoking being 39 times worse (64500/16.9%) / (8800/90%) = 39).

[1] https://en.wikipedia.org/wiki/Prevalence_of_tobacco_consumpt...

[2] I didn't find a statistic specifically about red meat or processed meat, but only 10% of the population are "meat-avoiders" according to this: https://en.wikipedia.org/wiki/Vegetarianism_by_country#Unite...

Forgive my ignorance, but what exactly is the definition of processed red meat? I assume it lies somewhere between gnawing on a freshly slaughtered animal carcass and buying beef jerky form the corner store, but I'm not sure where, exactly.

"Methods of meat processing include salting, curing, fermentation, and smoking. Meat processing includes all the processes that change fresh meat with the exception of simple mechanical processes such as cutting, grinding or mixing."

https://en.wikipedia.org/wiki/Processed_meat

This is for Japan, doesn't they already eat a lot better than the Western world, that is, less red meat, less processed foods, more high oily fish etc.?

In my experience, this is generally true, but the Japanese diet is changing rapidly. I came to Japan almost 10 years ago and I remember seeing quite traditional lunch boxes (called bento) in the convenience stores. Now everything is deep fried, or comes with fairly large portions of meat. Unless you go to a specialised bento store, it's pretty difficult to get a bento with a piece of fish, vegetables and rice.

Also, with the reduced restrictions on imports, I've seen a veritable explosion of steak restaurants and yaki niku places (Korean barbeque). Now that they can get cheap pork from Canada and cheap beef from Australia, meat has gotten very popular. The meat sections of the grocery stores in my area are about the same size as the fish section. And even the small town I live in has the same number of butchers as they do specialty fish stores.

Having said that, it's still not difficult to eat a traditional Japanese diet and certainly older people tend to eat that way. Not sure what will happen in the next generation or two, as the culture is shifting (to my eyes, anyway). It may be for the best, though, as global fish stocks are shrinking in an alarming way.

Having seen the giant ice cream-stuffed crepes they sell in Harajuku, I'm not sure how true that is anymore...

Studies that invalidate common beliefs are really interesting.

Keep in mind that's the only reason we're hearing about this study. If they found what would be expected it wouldn't be as interning. Man bites dog and all that.

There are probably lots of studies that say the opposite if you look for them.

The study didn't invalidate anything.

It shows how unreliable scientific consensus about a theory that cannot be subject to controlled experimentation, like any medical theory, is.

Forks over knives (https://en.wikipedia.org/wiki/Forks_Over_Knives) quite clearly argues that meat increases cancer risks manyfold.

T. Colin Campbell and Caldwell Esselstyn are no quacks either and the document makes a solid argument for minimizing meat intake, at least for adults.

I thought we were under the consensus that these large scale associative studies were not accurate/effective in predicting causal effects, especially in nutrition.

I can't find any studies right now, but I remember reading about mechanistic explanations for plant compounds such as isothiocyanate and resveratrol in reducing and even completely suppressing tumour growth.

This study doesn't account for strict plant-based diets. There isn't a single reference to animal product portion size. It just lumps everyone together. Some groups had larger fruit and vegetable portions. But all had animals.

The study wasn't trying to account for strict plant-based diets. It was attempting to find a correlation between fruit/plant intake and cancer rates.

It would be unfortunate if someone reads the headline and thinks this means that this study negates the meat consumption-cancer risk correlation studies.

Though, when considered in regards to the large body of plant based diet studies, this study does add more evidence to the notion that to benifit from plants we probably need to omit the animal products altogether.

But imagine if you did a study about lungs of runners, but did not account for who smoked.

Perhaps pesticide consumption offsets fruit and vegetable benefits.

Every day, the case for eating plants gets weaker. I'm starting to believe that plants aren't necessary for human health, and are possibly harmful.

This website has been eye-opening: https://zerocarbzen.com

I think you are suffering from confirmation bias.

Doesn't change the facts.

Why should I believe that plants are necessary for health?

Confirmation bias means they aren't strictly facts you're believing

Thank you for posting this link. Interesting stuff that parallels my own experiences with vegetarianism and veganism, the adoption of which I consider one of the great mistakes of my life. I've never gone 100% meat and water though I do eat red meat 4+ times a week.

Reply on news.ycombinator.com