Sunday, March 23, 2014

Don't swallow 'fats are fine' line

Mar 23, 2014

Potential flaws in study that concludes saturated fats do not raise heart risks

By Andy Ho Senior Writer

For years, we have all been told to avoid "bad" saturated fats, especially trans fats, because they increase heart disease risks. Solid at room temperature, these are found in butter, cheese, ice cream, fatty cuts of meat and processed foods.

We have also been told to substitute them with "good" polyunsaturated fats to lower heart risks. These fats are liquid at room temperature like olive oil and they are also found in nuts, seeds and some fish.

But last week brought news of a study published in a top journal, the Annals of Internal Medicine, which found there is no relationship between fats and heart risks.

The researchers from Cambridge, Oxford, London, Bristol and Erasmus universities declared that saturated fats are not the main driver of heart risks, and polyunsaturated fats have no or little impact on reducing heart risks.

What's a heart patient to do?

Before anyone heads out to gorge on steak, fat and all, hold on. The study, though comprehensive, was a review of several published studies on the link between the different types of dietary fats and heart risks. The researchers looked at data from separate studies covering more than 600,000 participants in 18 countries.

This kind of study is called a meta-analysis, where data from separate, individual studies is pooled and statistically analysed to bring together their disparate findings.

Some experts now allege serious flaws in the study. The researchers acknowledge that some minor statistical errors may have crept in, but maintain that these do not affect their conclusions.

What should the layman believe?

First, consider some of the potential flaws of this study. For starters, it lumped together studies of widely varying quality.

Only when you bring together studies of equivalent quality - say, those published in top journals - can you ensure that the resulting analysis is reliable. Conversely, one poorly conducted study can jeo-pardise the whole meta-analysis.

In this case, studies of different designs were used. There were 49 "observational studies" where the amount of fat consumed was estimated by recall of dietary intake or measurements of blood levels of fatty acids. Then there were 27 "randomised, controlled trials" (RCTs) where participants were given fish oil supplements.

The sources of errors in these two types of studies differ. In observational nutritional studies, human recall is used to provide the data on how much saturated fats were consumed in the past 24 hours, for example. Apart from faulty memories, there are other specific systematic sources of human errors that creep into such observational studies (but not RCTs).

When there are very real possibilities of error, the study's meta-analysis of observational studies may result in spurious conclusions being made - precisely like "saturated fats do not raise heart risks".

That is likely to be spurious because we know that, biologically, saturated fats lead to raised levels of "bad" LDL cholesterol that gives rise to artery cloggers.

Statistical findings must be consistent with established biology. So if observational studies - or their meta-analyses - say otherwise, it is likely they were flawed.

Another fundamental flaw in the observational studies was that they focused only on fats, without looking at carbohydrates too.

They should have, because people who avoid saturated fats tend to eat more carbs.

Saturated fats lead to bad LDL cholesterol, mainly in the form of big, buoyant and fluffy particles which make artery cloggers. But these are "weaker" than another form of bad LDL cholesterol that exists in small, dense particles which can make even more cloggers.

And it is carbs and refined sugars that cause more of the latter to predominate while also reducing "good" HDL cholesterol.

People who cut their saturated fat intake tend to eat more carbs because of hunger.

For them, the latter type of LDL predominates, generating more cloggers, raising heart risks.

But the study did not consider if a reduction of saturated fats was replaced with carbs that may heighten heart risks. Without acknowledging this, the researchers did tell the media that it is "the high-carbohydrate or sugary diet that is driving your LDL in a more adverse way".

Next, consider the study's meta-analysis of RCTs. Here, the investigator randomly assigns groups to supplements of omega-3 (O-3) polyunsaturated fats found naturally in oily fish like salmon and mackerel, or to omega-6 (O-6) polyunsaturated fats, like that in sunflower, soyabean or corn oil.

In this way, any effects detected would not come from the human errors that plague observational studies.

In the part of this study meta-analysing RCTs of O-3 or O-6 supplementation, heart risks were found to be unaltered. As RCTs set the gold standard for clinical research, surely these findings are kosher?

Unfortunately, polyunsaturated fats are not all the same. Experts now know if we increase our intake of O-6 but not also O-3, our heart risks rise. That is, the O-6:O-3 ratio matters.

Sadly, the modern diet has too much O-6 from cheap peanut or corn oil, chicken, pork and eggs. But it has too little 0-3 in costlier olive oil, salmon or mackerel.

This means that study subjects in these RCTs likely had too much O-6 and not enough 0-3 in their diets to begin with. But most RCTs were done earlier not knowing that the O-6:O-3 ratio matters.

If so, these RCTs were biased, so this meta-analysis of such RCTs was also biased.

How is it then that an apparently flawed study coming out of top institutions can get published in a reputable journal?

Dr John Ioannidis, Stanford University's Rehnborg Chair in Disease Prevention, who studies how scientific studies are done, has proven mathematically that most conclusions drawn from published medical research are misleading, inflated or erroneous: 80 per cent of observational studies and 25 per cent of RCTs are wrong. His calculated error rates match real world rates at which medical research findings are subsequently refuted.

No one disputes the mathematical model in his 2005 paper headlined "Most published research findings are probably false". And that model assumes only modest levels of researcher bias and flaws in their research methods.

He showed that if you want to prove an idea that is probably wrong to be right, you can - by choosing what to study, manipulating your data analyses and reporting media-attracting findings.

And entrenched, powerful academics can influence the peer-review process to suppress objections and get their papers published. Scientists want fame too, for along with it comes funding.

So if anyone read the news last week and thought it's time to chew the fat and dive back into butter, sorry. This new study is unlikely to invalidate what doctors have been telling us for years.

Other experts taken aback by the findings were quick to advise everyone to stick with what we've been told for years.

So we should still consume as little fat as possible, as fats contain twice the calories of proteins or carbohydrates.

We should still have a diet that is rich in oily fish, olive oil and fresh vegetables with a minimum of saturated animal fats - butter, cheese, lard, ghee, cream - and very little sweetened soft drinks, pizza, chips, cake, cookies, white flour, sugar and other junk foods.

And while we're at it, also cut the salt, stop smoking and start exercising regularly.

No comments: