7

votes

Why don't large, well-designed studies find that increasing PUFA hurts your health?

Answered on August 19, 2014
Created July 27, 2013 at 3:28 PM

Large, well-powered, high quality meta-analyses of studies on lipid intake frequently find at least some cardiovascular benefits to reducing saturated fat (SFA) and replacing it with unsaturated fat (PUFA). When they don't, the finding is that it doesn't make a difference.

But most of these are in industrialized populations, and the dietary intervention studies usually include other CW advice. That means that the participants are very likely reducing meat, eggs, and dairy, and replacing them with "healthy" oils like canola and soybean.

According to paleo, these oils are oxidized and pro-inflammatory. Participants who follow this advice should have more severe atherosclerosis and more cardiac deaths. Why don't these studies consistently find that increasing PUFA is harmful?

Clarification

I'm not asking for single studies showing that SFA is good for you, or for tirades against Ancel Keys or food frequency questionnaires. The question is why a review of good studies doesn't conclusively show that SFA is healthier than (omega-6) PUFA.

References

  • The Cochrane Review (Hooper et al.) meta-analysis of well-designed, controlled dietary intervention studies with long-term followup. Finds a reduction in cardiovascular events when SFA is replaced with PUFA, but no overall change in mortality. abstract (can't find free text of the 2012 update)

  • A meta-analysis of metabolic ward studies finds that replacing SFA with PUFA lowers LDL and increases HDL. summary full text

  • The Siri-Tarino et al. meta-analysis of prospective observational studies finds no relationship between SFA vs. PUFA consumption and cardiovascular disease summary and full text

  • Wikipedia's Saturated fat and cardiovascular disease controversy page covers other large combined analyses. Some find pro-PUFA effects and some find no effect, but none of them find a combined negative effect of increased PUFA.

Common criticisms

  • "Food frequency questionnaires are unreliable and epidemiological studies are full of confounds."

The same finding has appeared many times in metabolic ward studies and studies of dietary interventions.

  • "Increased LDL and/or TC isn't actually bad for you."

Many of these studies use actual cardiovascular disease or death as an outcome, and still find benefits to replacing saturated fat with PUFA.

  • "The authors are just twisting the data to show what they want."

Then there shouldn't be so many studies with null results. Also, you'd expect them to say that low-fat diets help. The Cochrane meta-analysis reaches the anti-CW conclusion that you have to replace SFA with PUFA, not with carbs.

A few possible explanations:

  • People who replace SFA with PUFA are probably following the good CW advice too, like eating more vegetables and less sugar. But then, they're probably also eating more whole grains and legumes.

  • "PUFA" includes fat from fish, and some papers lump in MUFA as well. However, except for a few mediterranean diet studies, I don't think most studies focused on these fats as opposed to standard vegetable oils.

  • Combining results across studies, replacing SFA with PUFA reduces heart attacks, strokes, etc., but may not reduce deaths overall. Maybe standard PUFAs are good for you in some ways and bad for you in others.

I don't find any of these completely convincing, especially given how strongly we argue that reasonable levels of SFA are innocuous, and omega-6 PUFA hurts people's health. What do you think?

72cf727474b8bf815fdc505e58cadfea

on August 12, 2013
at 08:00 PM

@thbq: My understanding is that fully hydrogenated oil is not really the problem. By definition trans fats have to be only partially hydrogenated, which means still unsaturated. I was just thinking about how confusing the data would be if we were still lumping those in with other unsaturated fats.

Medium avatar

(10611)

on July 29, 2013
at 03:54 PM

Yeah there's a big health difference between trans-fat Crisco and lard, but it's not detected by taste buds. I've been thinking a lot about the loss of dietary diversity. The world is heading down the cheapest path, towards a hyperpalatable mix of sugar, starch and vegetable oil.

Medium avatar

(10611)

on July 29, 2013
at 03:13 PM

Not that I believe that soy oil is optimal, but it will take the long term, large population, multifactorial study to sort this out. Smoking, sedentism, obesity, cancer, accidental death, calories consumed, macro ratios and many more factors have to be controlled to pull out the effect of o6 versus sat fat with statistical confidence. 100,000 people over 40 years MIGHT be enough.

Medium avatar

(10611)

on July 29, 2013
at 01:39 PM

Phrased a different way, the veterans didn't live longer happier lives on the sole basis of the oil they consumed. This is not a convincing case for avoiding soy oil because it's toxic. Hydrogenated soy oil avoid for sure, along with the other more well documented ways to extend longevity.

A2c38be4c54c91a15071f82f14cac0b3

(12682)

on July 29, 2013
at 03:02 AM

thhq, 1st comment: I don't really agree, but as I said it's not a perfect study. The vegetable oil group had more cancer and I personally find it a bit unlikely that something could increase cancer deaths solely because it reduced heart disease deaths. But sure, I concede it's kinda weak.

Medium avatar

(10611)

on July 29, 2013
at 02:09 AM

Trans fats exist because vegetable oil is made into artificial lard by saturation in a hydrogenater. In the process the PUFA ceases to exist, so not quite a fair test. Other than to possibly show that synthetic vegetable lard is unhealthy.

Medium avatar

(10611)

on July 29, 2013
at 02:04 AM

The LA Veterans study doesn't make a case for anything other than for the non-significance of oil type on all-cause mortality, which is identical for both groups.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 05:27 PM

Good gosh, imagine what a mess it would be if we were trying to assess the effects of PUFA but hadn't figured out about trans-fats yet. Though that would certainly be a boon for the paleo camp...

32f5749fa6cf7adbeb0b0b031ba82b46

(41757)

on July 28, 2013
at 02:36 PM

Observational studies should't be so easily tossed aside. There's a goldmine of information in there, but the problem is when folks don't understand what the data is good for. Correlative data is fine for forming hypotheses, and often enough suggests causation, but when confounders exist, it's simply not enough.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 05:20 AM

Thanks, I hadn't heard of that. Though our community is very quick to discard population studies when they disagree with our ideas, so I'm not sure I'd uncritically share the authors' interpretation of Israeli Jews as "a population-based dietary experiment of the effect of a high omega-6 PUFA diet".

A2c38be4c54c91a15071f82f14cac0b3

(12682)

on July 28, 2013
at 01:41 AM

Sorry if it this didn't come across, but a number of those studies make SFA's look good. For example the Syndey Diet heart study found 70% higher rates of CVD death when heart disease patients replaced butter with safflower oil, but because the safflower oil was consumed partly in the form of margarine the study has been criticized for likely including higher trans fats in the PUFA group.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 01:17 AM

Your knowledge of the literature is impressive. In each case I think the argument you make is plausible, but overall it is highly improbable that a variety of unrelated confounds and biases would ALL push the results in the direction of making SFA look bad. Do you think there's some unifying factor that explains why the flaws in every study push in the same direction?

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 01:13 AM

On Paleohacks, people talk about excess omega-6 as toxic, full stop. I think most would predict that replacing even poor SFAs with corn oil would be a net harm.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 12:47 AM

The studies that get mortality also look at all-cause mortality, for just this reason (you don't want to approve a drug that prevents 5000 heart attacks but causes 10,000 fatal hemorrhages). In these huge, long-term cohorts, if PUFA were causing cancer or autoimmune disease, it should show up in the mortality data.

C45d7e96acd83d3a6f58193dbc140e86

on July 27, 2013
at 07:59 PM

Studies that substitute SFA with PUFA generally do not differentiate between eating coconut oil and grass fed steaks with eating hydrogenated oils and hotdogs. There is a big difference. To do a study whe the people went from good saturated fats to PUFA has not happened, and no one that tries to eat only good saturated fats would participate anyway.

Frontpage book

Get FREE instant access to our Paleo For Beginners Guide & 15 FREE Recipes!

6 Answers

best answer

6
A2c38be4c54c91a15071f82f14cac0b3

(12682)

on July 27, 2013
at 06:25 PM

I think a big part of it is that almost all the long term controlled studies on this issue have had (as folks on both side have pointed out) decent flaws that seriously weaken them as evidence. This includes a lot of confounding factors between groups like vegetables and omega-3's (the Oslo Diet Heart study and STARS trial), contraindicated medications (the Finnish Mental Hospital trial, the Helsinki Businessman trial), sugar (Finnish trial again), carbohydrate intake (Rose Corn Oil trial), or trans fats (Sydney Diet Heart study, Oslo again).

Then there's studies which did not provided statistically significant results for numerous possible reasons. This includes the Minnesota Coronary Survey, the famous MRFIT, DART, and the British MRC soybean oil trial.

So what are we left with?

The LA Veterans trial is interesting because it provides evidence that PUFA oils raise cancer rates compared to saturated fat. This study is often used by pro-PUFA people because heart disease was lower in the PUFA group, but so were the number of heavy and moderate smokers, which I think is a reasonable explanation for lower heart disease rates. So it's clearly not a perfect study, but I think it's good evidence that O6 PUFA predominating vegetable oils will increase one's risk of cancer.

Link: http://www.thelancet.com/journals/lancet/article/PIIS0140673671910865/abstract

However, most reviews of controlled studies have either been on only cardiovascular disease rates or total deaths (which draws from the problematic studies previously mentioned and the LA Veterans trial, despite the difference in smoking rates). Some reviews have been done of cancer rates, but once you get past the LA trial you run out of studies that are worth looking at (or in the words of one review "each of these trials had serious methodological limitations in terms of adherence, duration, and small number of cancer cases").

I've sort of ignored observational studies because that would make my answer even longer, but I definitely can understand your curiosity. I guess part of what I've come to believe is that large, well designed studies don't find increasing PUFA hurts your health because there are almost no large, well designed studies showing PUFA's effect on health*.

*Well, maybe there are studies on fish oil O3's, which are PUFAs, but I assumed we were mostly talking about vegetable oils.

Medium avatar

(10611)

on July 29, 2013
at 02:04 AM

The LA Veterans study doesn't make a case for anything other than for the non-significance of oil type on all-cause mortality, which is identical for both groups.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 01:17 AM

Your knowledge of the literature is impressive. In each case I think the argument you make is plausible, but overall it is highly improbable that a variety of unrelated confounds and biases would ALL push the results in the direction of making SFA look bad. Do you think there's some unifying factor that explains why the flaws in every study push in the same direction?

Medium avatar

(10611)

on July 29, 2013
at 01:39 PM

Phrased a different way, the veterans didn't live longer happier lives on the sole basis of the oil they consumed. This is not a convincing case for avoiding soy oil because it's toxic. Hydrogenated soy oil avoid for sure, along with the other more well documented ways to extend longevity.

A2c38be4c54c91a15071f82f14cac0b3

(12682)

on July 28, 2013
at 01:41 AM

Sorry if it this didn't come across, but a number of those studies make SFA's look good. For example the Syndey Diet heart study found 70% higher rates of CVD death when heart disease patients replaced butter with safflower oil, but because the safflower oil was consumed partly in the form of margarine the study has been criticized for likely including higher trans fats in the PUFA group.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 05:27 PM

Good gosh, imagine what a mess it would be if we were trying to assess the effects of PUFA but hadn't figured out about trans-fats yet. Though that would certainly be a boon for the paleo camp...

A2c38be4c54c91a15071f82f14cac0b3

(12682)

on July 29, 2013
at 03:02 AM

thhq, 1st comment: I don't really agree, but as I said it's not a perfect study. The vegetable oil group had more cancer and I personally find it a bit unlikely that something could increase cancer deaths solely because it reduced heart disease deaths. But sure, I concede it's kinda weak.

Medium avatar

(10611)

on July 29, 2013
at 02:09 AM

Trans fats exist because vegetable oil is made into artificial lard by saturation in a hydrogenater. In the process the PUFA ceases to exist, so not quite a fair test. Other than to possibly show that synthetic vegetable lard is unhealthy.

Medium avatar

(10611)

on July 29, 2013
at 03:13 PM

Not that I believe that soy oil is optimal, but it will take the long term, large population, multifactorial study to sort this out. Smoking, sedentism, obesity, cancer, accidental death, calories consumed, macro ratios and many more factors have to be controlled to pull out the effect of o6 versus sat fat with statistical confidence. 100,000 people over 40 years MIGHT be enough.

72cf727474b8bf815fdc505e58cadfea

on August 12, 2013
at 08:00 PM

@thbq: My understanding is that fully hydrogenated oil is not really the problem. By definition trans fats have to be only partially hydrogenated, which means still unsaturated. I was just thinking about how confusing the data would be if we were still lumping those in with other unsaturated fats.

3
F199416a9911e88f6a452371c9d6c04d

(30)

on July 27, 2013
at 05:07 PM

Interesting question. These studies seem to focus mainly on CVD as an effect of SFA vs. PUFA. It's possible the real harm of industrial seed oils is not so much in increasing heart disease but in causing auto-immune disease, cancer, chronic disease, mental illness - all of which seem to be increasing in epidemic proportions, especially amongst young people. Perhaps we get confused about nutrition and diet on health because we consider each disease symptom to be isolated and unrelated to each other. Maybe the question scientists should be evaluating is whether higher PUFA consumption leads to more generalized damaged health, as opposed to causing specific symptoms.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 12:47 AM

The studies that get mortality also look at all-cause mortality, for just this reason (you don't want to approve a drug that prevents 5000 heart attacks but causes 10,000 fatal hemorrhages). In these huge, long-term cohorts, if PUFA were causing cancer or autoimmune disease, it should show up in the mortality data.

2
32f5749fa6cf7adbeb0b0b031ba82b46

(41757)

on July 27, 2013
at 09:20 PM

There is the less well known Israeli Paradox, which is observational between omega-6s and disease.

http://www.ncbi.nlm.nih.gov/pubmed/8960090

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 05:20 AM

Thanks, I hadn't heard of that. Though our community is very quick to discard population studies when they disagree with our ideas, so I'm not sure I'd uncritically share the authors' interpretation of Israeli Jews as "a population-based dietary experiment of the effect of a high omega-6 PUFA diet".

32f5749fa6cf7adbeb0b0b031ba82b46

(41757)

on July 28, 2013
at 02:36 PM

Observational studies should't be so easily tossed aside. There's a goldmine of information in there, but the problem is when folks don't understand what the data is good for. Correlative data is fine for forming hypotheses, and often enough suggests causation, but when confounders exist, it's simply not enough.

1
56c28e3654d4dd8a8abdb2c1f525202e

(1822)

on July 27, 2013
at 11:12 PM

Good link by Matt. In my work I am in contact with many Muslims, who eat the same diet as Israeli Jews. And they, too, appear to have similar problems. Limited statistics of course, and let us keep in mind that these people were eating wheat while my ancestors at least were eating large mammals and tubers. They have had about twice as long to adapt to wheat, and they still have those problems.

It is interesting to browse the CIA Handbook, and see what life expectancies are around the world. Not too high across the Middle East, including Oman which has the best health care in the world according to the WHO.

1
56c28e3654d4dd8a8abdb2c1f525202e

(1822)

on July 27, 2013
at 08:48 PM

Sustained Fitness above nails it, and there is a post by Guyenet that says exactly the same thing, but right now I can not find it. I submit that there are studies of the kind: in Kitava, 90% of fat is saturated (coconut), therefore eating some 20% of calories in saturated fat, twice the American average, 77% of people smoke, average cholesterol is 247, zero heart attacks. There is also a study where heart patients were fed a Mediterranean and paleo diets. Both improved their health, but paleo improved it much more.

You can only believe tests where specific fats are compared. I believe in everything Kitava has shown. I do not believe tests where saturated fat may include margarine.

Medium avatar

(10611)

on July 29, 2013
at 03:54 PM

Yeah there's a big health difference between trans-fat Crisco and lard, but it's not detected by taste buds. I've been thinking a lot about the loss of dietary diversity. The world is heading down the cheapest path, towards a hyperpalatable mix of sugar, starch and vegetable oil.

1
Medium avatar

(10611)

on July 27, 2013
at 05:24 PM

The main health problems with soy, canola and corn oil - and the sat fats - are due to hyper consumption. To do a study such as you desire would take years with a large population. How could you control such a study to being constant fat type, never mind controlled calorie intake?

If you look at historical national data, maybe. Many of the Med diet countries have shifted from low calorie MUFA olive oil diets to high calorie PUFA diets in the last century. If you could decouple the fat change from overeating, CVD rates might allow you to sort out the effect of oil. I can't think of a similar way to sort sat fat from PUFA though.

Here's the type of national study I have in mind, but as you can see it's multifactorial and would need a lot more nuance to detect the specific effect of single factors. http://www.ncbi.nlm.nih.gov/m/pubmed/12826634/

C45d7e96acd83d3a6f58193dbc140e86

on July 27, 2013
at 07:59 PM

Studies that substitute SFA with PUFA generally do not differentiate between eating coconut oil and grass fed steaks with eating hydrogenated oils and hotdogs. There is a big difference. To do a study whe the people went from good saturated fats to PUFA has not happened, and no one that tries to eat only good saturated fats would participate anyway.

72cf727474b8bf815fdc505e58cadfea

on July 28, 2013
at 01:13 AM

On Paleohacks, people talk about excess omega-6 as toxic, full stop. I think most would predict that replacing even poor SFAs with corn oil would be a net harm.

Answer Question


Get FREE instant access to our
Paleo For Beginners Guide & 15 FREE Recipes!