In the last post, I talked about the Houstmuller trial and how its inclusion in Cochrane's saturated fat review is unjustified (1). But there are two other trials in Cochrane's review that should also be excluded. These are the Oslo Diet-Heart Study (ODHS) and the St. Thomas' Atherosclerosis Regression Study (STARS) (2,3). Both trials apparently showed a cardiovascular benefit for diets that involved reduced saturated fat intakes and increased omega-6 linoleic acid intakes.

Similar to the Houstmuller trial, removing Oslo and STARS from Cochrane's analysis changes the result from a 17% reduction in cardiovascular events to a statistically non-significant 11% reduction.

(Image: Cochrane's 2020 analysis (4) for CVD events after exclusion of Oslo and STARS)

Besides cardiovascular events, exclusion of Oslo and STARS would also affect other endpoints. The overall effect for cardiovascular mortality would go from a statistically non-significant 6% reduction to a 1% reduction. And heart disease mortality would change from a statistically non-significant 3% reduction to a 2% increased risk (for diets lower in saturated fat).

(Image: A statistically non-significant 2% increased risk of heart disease death after exclusion of Oslo)

Thus, if Oslo and STARS are removed, the evidence for reducing saturated fat intakes would essentially vanish.

Why should Oslo and STARS be excluded?

The most obvious reason is that both trials were "multifactorial" or involved multiple interventions. That is, the interventions involved much more than just saturated fat reductions. Because of this, neither trial had any hope of testing the effect of reduced saturated fat intakes.

For example, the extensive interventions in the Oslo study were well described by Ramsden and colleagues (5). First, they noted that the experimental group was provided with large quantities of important seafood nutrients, including omega-3 fatty acids and vitamin D3:

Experimental dieters were instructed to substitute fish, shellfish and ‘whale beef’ for meats and eggs, and were actually supplied with ‘considerable quantities of Norwegian sardines canned in cod liver oil, which proved to be popular as a bread spread’. These cold-water fish, sardines, cod liver oil, shellfish and whale provided an estimated 2·0 en % (about 5 g/d) of EPA and DHA . . . In addition, the fish and cod liver oil consumption provided Oslo dieters with 610 IU of daily vitamin D3.

Second, they pointed out that participants of the experimental group were encouraged to limit their intakes of refined grains/sugar and to consume more nuts, fruits, and vegetables:

Furthermore, experimental dieters were encouraged to eat more nuts, fruits, and vegetables . . . and to restrict their intake of refined grains and sugar.

And third, they emphasized that the experimental group had a "profound" reduction in trans-fat intake (TFA):

In the two decades before the ODHS, Oslo males had an alarming 7-fold increased incidence of first MI. This rapid rise coincided with pervasive use of partially hydrogenated fish and vegetable oil margarines, accounting for 65 g/person per d at study onset. These margarines provided the control group an estimated 6·9 g (2·6 en %) of unusual 20 and 22 carbon TFA produced from the partial hydrogenation of fish oil. Importantly, margarines were ‘entirely restricted’ and replaced with non-hydrogenated soybean and cod liver oils in the experimental group, whereas the control group continued consumption.

Therefore, the Oslo trial was clearly a multifactorial one.

But what about STARS? The following quote makes it clear that STARS, too, was multifactorial (6):

STARS also involved numerous dietary interventions: More fruits, vegetables and starchy complex carbohydrate foods, reduced consumption of processed foods (including "cookies, pastry, cakes"). While the dietary guidelines given to the patients called for "strictly limiting" intake of meat, fish and dairy products, dietary records showed 3-fold higher DHA intake among intervention subjects . . .

Also of key importance is the intervention group maintained lower caloric intake.

In other words, both trials involved so many dietary changes that the experimental groups were eating a new diet with less trans-fat, more marine fatty acids and nutrients, fewer "junk" foods, etc. Saturated fat reductions (and linoleic acid increases) were not the only changes (7-9). Not even close.

Thus, how are we to separate the cardiovascular effect of reducing saturated fat intakes from all these other changes? It is not possible. Oslo and STARS are not "controlled" trials capable of testing saturated fat.

Cochrane's Admissions

The dietary differences between the control group and intervention group in both trials did not go unnoticed by Cochrane. In 2012, Cochrane labeled the Oslo trial as "High risk" for dietary differences other than fat, citing "Differences in fruit and vegetables, fish etc." between the groups (10).

And not only that. In 2018, Cochrane conducted two reviews on polyunsaturated fat where they completely excluded both STARS and Oslo. Why? Because the trials were "multifactorial" (11,12).

In fact, take a look at the following image from Cochrane's polyunsaturated fatty acid (PUFA) review about why they excluded the Oslo trial (highlighted):

As you can see, Cochrane excluded the Oslo trial in their PUFA review because it is a "multifactorial dietary intervention" and therefore "cannot separate out the effects of PUFAs from other dietary interventions." But if this is the case, then the trial cannot also separate out the effects of saturated fat from other dietary interventions.

So the question now becomes: Why did Cochrane include Oslo and STARS in their saturated fat reviews but not their PUFA reviews? Well, when we questioned the Cochrane investigators after publication of their 2020 saturated fat review, they stated the following (13):

We did not include studies that were multifactorial in terms of smoking, exercise, drugs etc, but did allow concurrent dietary changes.

Therefore, the Cochrane investigators are apparently using two different definitions of "multifactorial" without any justification. They are implicitly saying that confounding by other dietary factors are allowed in their reviews of saturated fat, but not allowed in their reviews of polyunsaturated fat.

This is an unacceptable double-standard, which becomes even more egregious when we consider known dietary confounders that could explain most of the results, such as omega-3.

The Confounding Effect of Omega-3

Interestingly, in Cochrane's first review of saturated fat in 2001, the following was stated about the Oslo trial (14):

In this trial [Oslo] participants in the intervention group were supplied with oily fish, in addition to dietary advice. Oily fish, which is rich in omega 3 fatty acids, seems to have independent beneficial effects, so the benefits seen in this trial may have been due to the fish oils and not to cholesterol lowering or alterations in other dietary fats.

Indeed, Oslo involved massive doses of omega-3 marine fatty acids (EPA + DHA) in a secondary prevention population (with a prior heart attack), and participants in STARS (also secondary prevention) doubled their intakes (5).

Now, while changes in other dietary factors (e.g., less trans-fat consumption) may have played a beneficial role in Oslo and STARS, there are good reasons to suspect that increases in omega-3 marine fatty acids had a big influence. If we analyze older dietary trials (before the year 2000), we would see consistent benefits for trials that involved increased intakes of marine fatty acids:

(Image: An analysis of trials conducted before the year 2000 suggesting a mortality benefit for marine fatty acids. It shows that STARS and Oslo are consistent with the "omega-3 effect").

Oslo also showed a 57% reduction in fatal heart attacks and a 26% (non-significant) reduction in coronary heart disease death. But these effects were not seen in other dietary trials that replaced saturated fat with polyunsaturated fat without marine fatty acids (15). If anything, a pooled analysis of these trials shows an average effect in the opposite direction of what Oslo claimed (harmful direction):

(Image: When the confounding effects of marine fatty acids are removed, replacing saturated fat with polyunsaturated fat looks like a bad idea. Taken from reference 15)

In contrast, beneficial effects on fatal heart attacks and heart disease mortality can be produced by omega-3:

(Image: Despite being conducted under vastly different circumstances, the VITAL omega-3 trial of 2019 showed remarkably similar results to Oslo on heart disease mortality and fatal heart attacks. Taken from reference 16)

Furthermore, another indication that the benefits may be due to omega-3 is the early time to benefit, which was suggested in the Oslo trial (17):

It is notable that omega-3 fatty acid therapy shows a consistently early time to benefit in the 3- to 6-month range . . . the Oslo Diet Trial, evaluated 412 patients and published data on timing of benefit. By year 1, 11 CAD-relapse events occurred in the treatment group compared with 21 events in the control group.

Thus, changes in trans-fat and/or omega-3 fatty acids can plausibly explain all the results. There's no justification for blaming saturated fat.

Other Issues

For the STARS trial, about 71% of the clinical events were softer endpoints more susceptible to bias (13). The principal investigator (G.F. Watts) also admitted to other limitations, including the highly selected group of male patients, a drug run-in period prior to randomization, the lack of adequate blinding, and the fact that STARS "was not designed to test clinical events" (18).

As for randomization and allocation concealment in STARS, Cochrane states: "blinded random cards issued centrally by statistician advisor." But this is neither an adequate description of randomization nor allocation concealment. It tells us nothing about the type of randomization (simple randomization?) or what made the cards "random." Additionally, the word "blinded" is not a description of the specific method of allocation concealment (envelopes?).

There were also issues in the Oslo trial. First, there was no blinding, which could have introduced various biases (9). Second, there were systematic differences in care between the groups and an "optimistic attitude" among the dieters and their families (2,4,9). Third, there were 46 post-randomization exclusions without CVD data (no intention-to-treat analysis). And fourth, almost half of the subjects had cholesterol levels of at least 300 mg/dL, suggesting that many had genetic lipid disorders (not representative of most people).

Was Allocation Concealment in Oslo Adequate?

Cochrane judged the Oslo trial as "Low risk" for allocation concealment, stating (4):

Randomisation appears to have occurred before medical examination within the study, so was not affected by participant characteristics and was concealed.

I am not convinced.

Randomization was carried out by the "Institute" using "tables of random numbers" (2). But beyond this, no information was provided about the randomization process. What was preventing the Institute, for example, from allocating more older subjects or overweight subjects to the control group? Was the Institute really ignorant of participant characteristics? After all, it is the "Institute" that registered and excluded participants for having or not having certain characteristics.

The above example is relevant because they were actually more patients aged 60 years or older and more overweight patients in the control group (6). Also, the youngest subjects — who were under the age of 40 and possibly had genetic disorders to get a heart attack at such a young age — all ended up in the control group. Moreover, there was a clear imbalance in anticoagulant treatment at the start of the trial. Although these imbalances could be caused by post-randomization exclusions or "chance," selection bias cannot be excluded.

Whatever the case, poor reporting is itself a flaw, and Oslo should be labeled as "Unclear risk" for allocation concealment, at best.

Conclusion

Even ignoring their poor methodological quality, Oslo and STARS are disqualified as tests of saturated fat based on the dietary interventions alone. Therefore, neither trial should be included in any meta-analysis of saturated fat.


References

1) https://www.buymeacoffee.com/DietaryDiary/the-houtsmuller-trial-a-case-research-fraud

2) Leren P. The effect of plasma cholesterol lowering diet in male survivors of myocardial infarction. A controlled clinical trial. Acta Medica Scandanavica Supplement, 1966; 466: 1-92.

3) Watts, G. F., Lewis, B., Lewis, E. S., Coltart, D. J., Smith, L. D. R., Swan, A. V., ... & Mann, J. I. (1992). Effects on coronary artery disease of lipid-lowering diet, or diet plus cholestyramine, in the St Thomas' Atherosclerosis Regression Study (STARS). The Lancet, 339(8793), 563-569.

4) Hooper, L., Martin, N., Jimoh, O. F., Kirk, C., Foster, E., & Abdelhamid, A. S. (2020). Reduction in saturated fat intake for cardiovascular disease. Cochrane database of systematic reviews, (8). (https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD011737.pub3/epdf/full)

5) Ramsden, C. E., Hibbeln, J. R., Majchrzak, S. F., & Davis, J. M. (2010). n-6 fatty acid-specific and mixed polyunsaturate dietary interventions have different effects on CHD risk: a meta-analysis of randomised controlled trials. British Journal of Nutrition, 104(11), 1586-1600.

6) https://www.bmj.com/content/353/bmj.i1246/rr-13

7) Hoenselaar, R. (2012). Re: More discrepancies around saturated fat and cardiovascular diseases. Nutrition, 28(11/12), 1206.

8) Hamley, S. (2017). The effect of replacing saturated fat with mostly n-6 polyunsaturated fat on coronary heart disease: a meta-analysis of randomised controlled trials. Nutrition journal, 16(1), 1-16.

9) Heileson, J. L. (2020). Dietary saturated fat and heart disease: a narrative review. Nutrition reviews, 78(6), 474-485.

10) Hooper, L., Summerbell, C. D., Thompson, R., Sills, D., Roberts, F. G., Moore, H. J., & Smith, G. D. (2012). Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane database of systematic reviews, (5). - https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC6486029/

11) Hooper, L., Al‐Khudairy, L., Abdelhamid, A. S., Rees, K., Brainard, J. S., Brown, T. J., ... & Deane, K. H. (2018). Omega‐6 fats for the primary and secondary prevention of cardiovascular disease. Cochrane Database of Systematic Reviews, (7). - https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD011094.pub4/full

12) Abdelhamid, A. S., Martin, N., Bridges, C., Brainard, J. S., Wang, X., Brown, T. J., ... & Group, C. H. (2018). Polyunsaturated fatty acids for the primary and secondary prevention of cardiovascular disease. The Cochrane database of systematic reviews, 2018(11). - https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD012345.pub3/full

13) https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD011737.pub3/detailed-comment/en?messageId=320249873

14) Hooper, L., Summerbell, C. D., Higgins, J. P., Thompson, R. L., Capps, N. E., Smith, G. D., ... & Ebrahim, S. (2001). Dietary fat intake and prevention of cardiovascular disease: systematic review. Bmj, 322(7289), 757-763. - https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC30550/

15) Ramsden, C. E., Zamora, D., Majchrzak-Hong, S., Faurot, K. R., Broste, S. K., Frantz, R. P., ... & Hibbeln, J. R. (2016). Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968-73). bmj, 353.

16) Manson, J. E., Cook, N. R., Lee, I. M., Christen, W., Bassuk, S. S., Mora, S., ... & Buring, J. E. (2019). Marine n− 3 fatty acids and prevention of cardiovascular disease and cancer. New England Journal of Medicine, 380(1), 23-32.

17) Denke, M. A. (2005). Diet, lifestyle, and nonstatin trials: review of time to benefit. The American journal of cardiology, 96(5), 3-10.

18) Watts, G. F. (1996). The St Thomas’ Atherosclerosis Regression Study and an appraisal of its potential limitations. In Lipid-Lowering Therapy and Progression of Coronary Atherosclerosis (pp. 143-150). Springer, Dordrecht.