How well does the evidence used to support the use of margarine stand up?

Last week one of my blog posts highlighted a piece I had written for the Times on the relative merits of butter and margarine from a health perspective. I was limited to 1,300 words, and in the space allowed attempted to summarise all the most important prevailing nutritional themes here. I came out heavily in favour of butter.

I know this can be a controversial view, and it’s inevitable that some professionals will not agree with it. Several years ago, for instance, I ‘bumped into’ someone on the internet who was in my year at medical school. Whilst polite, she seemed aghast at my views on butter and margarine. I (also politely) asked her to get together all of the evidence she could find that supports the idea that margarine is healthier than butter and send it my way. She emailed me back to say she couldn’t find any.

Not all people are open enough to take on new information and move on in their thinking. For example, dietician Reijo Laatikainen took me to task over my failure to report evidence which he felt invalidated my case.

In my piece, I had mentioned a huge meta-analysis that failed to find any benefits from reducing saturated fat in the diet and/or replacing it with something supposedly healthier [1]. Reijo was keen to point out to me the claim by the authors that they found an 18 per cent reduction in risk of ‘cardiovascular events’.

I pointed out to Reijo, though, that some of the studies used in the analysis involved changes other than those to do with fat. When the analysis was confined to studies in which fat intake was the only change, absolutely no benefits were found at all.

I’ve asked Reijo to comment on this and whether he feels his original assertion was correct. He has steadfastly refused to comment.

Reijo was also keen to point out a study included in the meta-analysis, the so-called The Los Angeles Veterans Administration Diet Study (LA Veterans study). This study tested the effects of a diet in which vegetable oils replaced animal fat in the diet (details below). I recalled this study when Reijo raised it, but could not remember the detail. One detail I did recall, though, was that while the experimental diet improved some ‘cardiovascular’ outcomes,overall risk of death was not improved.

My exchanges with Reijo around this study caused me to go back to read the original study [2], was well as a later report on the study [3] with some critical detail not included in the original study report. This report [3] was specifically cited by Reijo.

I’m going to describe the study and its findings now. Just bear in mind that this is the study that Reijo is pinning his hopes on as the saviour of the ‘butter is bad, margarine is healthy’ paradigm.

The study involved 846 middle-aged and elderly men (54-88 years) living in a veterans institution in the US. Each man was allocated to eat one of two diets:

1.    a diet low in animal fat but enriched with polyunsaturated vegetable oils (corn oil, safflower oil, soya bean oil, and cottonseed oil)

2.    a normal diet

The food was supplied in the centre’s canteen under-double blind conditions (neither the men nor the researchers knew which diet each participant was eating). This is good, but what was not so good is that, once assigned to the diet, men were considered as part of the study however long they lived at the centre subsequently. What this means is that it was possible for men who only ate the prescribed diet for a relatively short period of time (even a few days or weeks) to be included in the data analysis, and that these results would have equal weight of those obtained from people who had been in the study for several years.

But another major failing of the study was that the diets of the men were not strictly controlled at all. Dining room records reveal that, overall, only about half the meals the men ate were at the centre and therefore controlled. We have no idea what they ate the rest of the time (good, bad or indifferent).

Each participant was allotted his diet on a random basis. This is done to ensure that the two groups being compared are equivalent in terms of characteristics such as age and general health. While the two groups were well matched in many areas, one critical area that the men were not was cigarette smoking.

For example, 45 men in the ‘healthy diet’ group smoked more than a packet of cigarettes a day, compared to 70 in the control group. Also, there were 99 non-smokers in the ‘healthy diet’ group, but only 86 in the control. The generally higher levels of smoking in the intervention group would bias the results in favour of this group of course.

The ‘primary outcome’ (pre-determined main measured outcomes) for this study was the number of cases of ‘sudden death’ and heart attacks. The differences between the groups were not significant. In other words, the researchers failed to find any benefit in terms of the main outcome they measured.

In situations such as this, it is not uncommon for researchers to start ‘shifting the goalposts’, and dicing and splicing the data in a way that increases the chances of finding a positive result. The outcome that Reijo has been keen to draw our attention to relates to ‘cardiovascular fatalities’. This is a ‘composite’ outcome of a wide range of things including death from sudden death, heart attack, stroke, amputation and rupture of the aorta (the main artery in the body).

Lumping all these things together did produce a statistically significant result. However, as I mentioned earlier, the LA Veterans study failed to find any reduction in overall risk of death.
Now, the fact that cardiovascular deaths were reduced butoverall risk of death was not points to the possibility of an increased risk of death from non-cardiovascular causes. In this study, the ‘healthy’ diet appeared to increase the risk of cancer.

I put this to Reijobut he dismissed it. I suggest one needs to take the rough with the smooth. Is it good science to take a study and laud findings you like, and just wave away those you don’t? Not usually.

And on this point, remember that the LA Veterans study was included in the meta-analysis that found that this was the only study to find benefits in terms of cardiovascular mortality. No other suitable study found this, and overall there was no benefit [1].

And a few other home truths are that the meta-analysis also found that in studies were fat had been reduced or modified in a ‘healthy’ way, there was no benefit in terms of risk in any of these things: heart attack, stroke, diabetes, cancer. And life was not extended by a single day, either.

Here’s a summary of the evidence Reijo has put up to question my conclusions regarding butter and margarine:

1. A ‘positive’ finding from a meta-analysis (Cochrane review) of fat reduction/modification trials that does not stand up to scrutiny (a fact he won’t comment on)
2. The cardiovascular benefit of a study (LA Veterans) which also found an increased risk of cancer (and no reduced risk of death)
3. That the original intended measured outcome of the study was not found to be improved, and that benefits were only found once the authors started to ‘massage the data’
4. That this study was ‘double-blind’, but in no way ‘controlled’ in that the men were free to eat what they liked about half the time
5. That in this study, results could be included from men who participated in the study for a short time only
6. That higher smoking rates in the control group biased the results in favour of the ‘healthy diet’
7. That this is the only study listed in the Cochrane review that found benefits in terms of cardiovascular deaths (none of several others found benefit here)
   
8. That overall, fat reduction/modification has not been found to reduce risk of cardiovascular death
9. That overall, fat reduction/modification has not been found to reduce the risk of heart disease or stroke, and has not been found to extend life either

Reijo can ignore these facts but he can’t change them.

I don’t know Reijo and perhaps he’s a very well-meaning chap. Even if this is true, I think he’s suffering from a terrible case of bias. My sense is, for whatever reason, he is stuck in an old paradigm where saturated fat is ‘artery clogging’ and highly refined and heavily processed vegetable oils (margarine) somehow have (miraculous) health-giving properties.

Even in the face of overwhelming evidence to the contrary he appears to cling to this paradigm for grim life. Let’s hope it won’t be the death of him.

References:

1.  Hooper L, et al. (2012) Reduced or modified dietary fat for preventing cardiovascular disease. The Cochrane Library.

2.   Dayton S, et al. Controlled trial of a diet high in unsaturated fat for prevention of atherosclerotic complications. Lancet. 1968;2(7577):1060-2.

3.  Dayton S, et al. Los Angeles Veterans Administration Diet Study. Nutrition Reviews 1969;27(11):311-316.