Here we go again – is a saturated fat tax back on the table?
The other day a new paper modelling a saturated fat tax came out in the journal PlosOne.
In order to improve the population’s diet, the World Health Organization has called for the taxation of foods that are high in SFA.
Oh no not this again! There are very strong socio-economic reasons why such a tax makes no sense, and would probably have the opposite effect of a sugar tax, but before we get to discussing that we’ll look at what the authors did, what they got right, and where they went wrong.
First up, they (all 3 authors are public health analysts, none has published an experiment) are modelling the predicted effect of a tax on fat consumption, using data from Denmark, which had a short-lived saturated fat tax. it was a stupid move for two reasons. First, higher LDL cholesterol is associated with longevity in Denmark – you’ll see how they missed this next. And second, the revulsion against the saturated fat tax and its effect on traditional foods saw both the fat tax and Denmark’s longer-lived sugar tax being thrown out at the next election. So this folly cost them any benefits of their sugar tax. Then, they are modelling the expected rate of CVD (heart attack and stroke) mortality from the predicted changes in LDL cholesterol from substituting polyunsaturated fats (PUFA) for saturated fats (SFA), based on feeding studies.
But hang on – what about the odds of dying from other conditions, like cancer, diabetes, etc? This – all-cause mortality – is not mentioned at all. For all we know we’re shifting the disease burden around, or making it worse overall. Or there may be no benefit for CVD at all either (see admissions below).
The authors get a couple of things right. They acknowledge that there’s no evidence that SFA, by itself, is associated with CVD. Then they turn to the type of computerised remodelling of epidemiological data invented about 10 years ago that looks at replacing one fat with another. In this, it usually (but not always) looks as if replacing SFA with PUFA lowers CVD risk. That’s a hypothesis – so has anyone tested the hypothesis?
Yes they have – and to their credit these authors cite Steve Hamley’s meta-analysis, which corrects errors of data extraction found in others and discusses the confounders in various studies.
A recent meta-analysis, which based their findings on randomised controlled trials, concluded that replacing SFA with PUFA would have no effect on major IHD events . Nevertheless, it is argued that prospective cohort studies find the replacement of SFA with PUFA to be beneficial for the prevention of IHD.
Um, this is epidemiology in reverse. RCTs are the opportunity to test the hypothesis. The Cochrane Collaboration recently summarised the RCT evidence thus, in substantial agreement with Hamley: So are we really to base such big public health changes on no real data of efficacy and a simple hypothesis? How did that end last time with the low fat revolution?
Increasing PUFA probably makes little or no difference (neither benefit nor harm) to our risk of death (moderate-quality evidence), and may make little or no difference to our risk of dying from cardiovascular disease (low-quality evidence). However, increasing PUFA probably slightly reduces our risk of heart disease events and of combined heart and stroke events (moderate-quality evidence). Fifty three people would need to eat more PUFA to prevent one person experiencing a heart disease event, and 63 people to prevent one person experiencing a heart or stroke event. Increasing PUFA may very slightly reduce risk of death due to heart disease, as well as stroke, but harm is possible (low-quality evidence). PUFA probably slightly reduces fats circulating in the blood (triglycerides, moderate-quality evidence but without effects on other lipids or adiposity). The evidence mainly comes from dietary-advice trials of men living in high-income countries.
There’s a very important point here – although PUFA lowers LDL-cholesterol in the short term feeding studies that the tax modelling paper relies on, Cochrane haven’t highlighted that effect in the RCTs. Recently we looked at a study co-authored by Jim Mann where Chinese subjects were fed 40% of calories as soybean oil for a year – a very high PUFA intake. Did that lower their LDL? At the end of the year they had changes to their gut bacteria that Mann thought compatible with an increased risk for bowel cancer and heart disease – but their LDL cholesterol was the same, and they had higher triglycerides.
“The high fat group had a very different profile of bacteria in their gut, one more compatible with an increased risk of bowel cancer and also a much higher risk of inflammation leading to cardiovascular disease, heart disease and possibly diabetes,” Prof Mann says.
So we see that increasing PUFA in real life, which is the intended effect of the tax (or the only way its authors think it might benefit us) does not always produce the results they are predicting from short-term feeding studies done in healthy young white male subjects.
The feeding studies are interesting because they show that some saturated fats (lauric and myristic acids) raise LDL much more than other SFAs. So if the SFA -> LDL -> CVD hypothesis were true, these fatty acids should be the ones most associated with CVD, right?
But no – in every epidemiological study that separates different fatty acids, the cholesterol-raising SFAs are those least associated with CVD. Sometimes, they have a “protective” association.[6,7,8,9]
So what does this mean? Obviously, the cholesterol-raising effect of SFA does not have the same effect on cardiometabolic risk as the cholesterol-raising effect of the genes for familial hypercholesterolaemia. Why anyone still thinks that it can is beyond comprehension – almost every pathway is being influenced in a different way. Yet this crudest of analogies is what we seem to be stuck with as a replacement for the scientific process.
Harm is possible
Cochrane said of PUFA in the RCTs, “but harm is possible (low-quality evidence)”. What does this harm look like in epidemiology?
An increased risk of severe pre-eclampsia (a condition of high blood pressure and proteinuria in pregnancy which can result in seizures) is associated with higher ALA intake in Denmark (canola oil is the main source of ALA there) – RR 1.71
Note that DHA and EPA (fatty fish) have a protective association with severe pre-eclampsia; saturated fat has a protective association with pre-eclampsia which shows a clear dose-response in minimally adjusted (0.63) and multivariable adjusted (0.73) models, but is non-significant in the most adjusted model. Dairy fat is the main source of SFA in Denmark.
This is the country where they actually imposed a saturated fat tax, remember? It’s unlikely they did any modelling around complications of pregnancy beforehand.
Increased risk of melanoma and other skin cancers is associated with higher LA intake in the USA. This is in the NHS/HPFS database, where PUFA usually tends to look better than it does in most other cohort studies; the mechanistic explanation laid out in the paper is a strong one.
Higher intake of omega-6 fat was associated with risk of melanoma (pooled multivariate HR, 1.20; 95% CI, 1.02–1.41; Ptrend=0.03) although the association was attenuated and no longer significant after adjusting for other types of fat simultaneously.
For SCC [squamous cell carcinoma], although total fat intake was not associated with the risk, polyunsaturated fat intake was associated with SCC risk (pooled multivariate HR, 1.16; 95% CI, 1.05–1.28; Ptrend=0.001) (Table 3). Among types of polyunsaturated fat, higher intake of omega-6 fat was associated with SCC risk [HR, 1.23 (1.08, 1.41)]. We also found that cholesterol intake was associated with lower risk of SCC.
Similar associations were also seen for basal cell carcinoma. Co-incidentally or not, the skin cancer death rate in NZ doubled soon after margarine was “legalised” in 1972 and has not declined since despite the slip-slop-slap campaigns and introduction of high-SPF sunscreens. And this is a serious problem, because avoiding sunlight is associated with increased CVD risk – another unintended consequence that modelling will overlook.
How does this compare with a tax on sugary drinks?
The proposed saturated fat tax is not just a different version of a sugar tax, but its opposite. We support a tax on sugary drinks because it targets an unfair price difference – Coca Cola is cheaper than “healthy” alternatives like bottled water, and this price point difference drives people, especially those with less money, towards the drink that will cost them more in the long run. A tax on sugar content goes a small way towards correcting this imbalance. Despite the small extra “penalty” cost to those poor people who still drink SSBs, which some understandably object to, it will tend to make their habits a little more like those of the rich.
The proposed saturated fat tax will do the opposite. Highly saturated fats like coconut oil, butter, and beef and lamb dripping are already about 4x as expensive as the refined seed oils that are high in PUFA. This makes them foods of the healthier, better-off classes. This pricing difference already drives disadvantaged people towards seed oils and the foods made with them – KFC (chicken, higher in PUFA, is also cheaper than other meats), Best Foods mayonnaise, and margarine. If these high-PUFA foods were truly healthier, these populations would be better off as a result – but this is so, so not the case. A saturated fat tax just puts real foods and healthy fats (remember, dairy fat and coconut oil are not even associated with increased disease risk) further out of reach of the poor. It increases the gap between the dietary habits of the rich and poor.
At the very bottom of their paper, the saturated fat tax authors admit that the Danish saturated fat tax (20% on the price of butter etc.) was counter-productive, even according to their own logic.
Finally, a fundamental assumption in our study was that PUFA intake would increase accordingly. This assumption is in line with the dietary recommendations that SFA should be replaced with PUFA [21, 53], and other modelling studies that have assumed the SFA reduction would be replaced by PUFA [14, 16, 17, 20] (see S1 Table). A SFA decrease without a simultaneous increase in PUFA does not seem to be beneficial for health [55, 56]. In Denmark, the decreased SFA intake (ranging from 1.6% in males above age 85 to 4.9% in females aged 55–69) seems to have been accompanied by a changed PUFA intake ranging from a 3.1% decrease in males above age 85 to a 0.3% increase in females aged 60–74 . Therefore, our assumption may lead to an overestimation of the health benefits. Foods contain a mix of SFA, MUFA and PUFA. Therefore, further research needs to examine how this corresponding substitution of PUFA can be assured, rather than a replacement by MUFA or carbohydrates. For instance, a price increase for products high in SFA might need to be coupled with a corresponding price decrease for products high in PUFA.
(If you lowered the price of soybean oil any more you’d be giving it away for free, but OK, whatever you say). Despite this awareness, they still went ahead and published a paper claiming via computerized projections that a saturated fat tax would prevent CVD deaths, even though they had evidence the tax as applied in its only real-life example hadn’t worked as designed (even if you don’t factor in the extinction of the sugar tax it caused). Meanwhile, across the bridge in Sweden an increase in butter consumption, with a drop in carbohydrate intake, has coincided with a halving of the heart attack death rate over the past 15 years.
We’ve said it before and we’ll say it again – public health analysts need to get out among the people they are trying to protect; they need to shop alongside them for a start. The saturated fat tax authors don’t have a clinical intervention to their name. They have no idea what will work in the real world. You can build all the models you like; you can build the Millennial Falcon, for example, but you are never going to see one fly in real life.
 Substituting polyunsaturated fat for saturated fat: A health impact assessment of a fat tax in seven European countries
Johanna-Katharina Schönbach , Wilma Nusselder, Stefan K. Lhachimi
Published: July 10, 2019https://doi.org/10.1371/journal.pone.0218464
 Bathum L, Depont Christensen R, Engers Pedersen L, Lyngsie Pedersen P, Larsen J, Nexøe J. Association of lipoprotein levels with mortality in subjects aged 50 + without previous diabetes or cardiovascular disease: a population-based register study. Scand J Prim Health Care. 2013;31(3):172–180. doi:10.3109/02813432.2013.824157
 Hamley S. The effect of replacing saturated fat with mostly n-6 polyunsaturated fat on coronary heart disease: a meta-analysis of randomised controlled trials. Nutr J. 2017;16(1):30. pmid:28526025
 Abdelhamid AS, Martin N, Bridges C, Brainard JS, Wang X, Brown TJ, Hanson S, Jimoh OF, Ajabnoor SM, Deane KHO, Song F, Hooper L. Polyunsaturated fatty acids for the primary and secondary prevention of cardiovascular disease. Cochrane Database of Systematic Reviews 2018, Issue 11. Art. No.: CD012345. DOI: 10.1002/14651858.CD012345.pub3.
 , et al
Effects of dietary fat on gut microbiota and faecal metabolites, and their relationship with cardiometabolic risk factors: a 6-month randomised controlled-feeding trial.
 Praagman J, Beulens JW, Alssema M et al. The association between dietary saturated fatty acids and ischemic heart disease depends on the type and source of fatty acid in the European Prospective Investigation into Cancer and Nutrition-Netherlands cohort. Am J Clin Nutr. 2016; 103(2): 356-365.
 Praagman J, de Jonge EAL, Kiefte-de Jong JC et al. Dietary Saturated Fatty Acids and Coronary Heart Disease Risk in a Dutch Middle-Aged and Elderly Population. Arteriosclerosis, Thrombosis, and Vascular Biology. 2016;ATVBAHA.116.307578. https://www.ahajournals.org/doi/10.1161/ATVBAHA.116.307578
 Praagman, J., Vissers, L. E., Mulligan et al. Consumption of individual saturated fatty acids and the
risk of myocardial infarction in a UK and a Danish cohort. International journal of cardiology. 2018;10:064. https://doi.org/10.1016/j.ijcard.2018.10.064
 Wang DD, Li Y, Chiuve SE, Stampfer MJ, Manson JE, Rimm EB, Willett WC, Hu FB. Association of Specific Dietary Fats With Total and Cause-Specific Mortality. JAMA Intern Med. 2016;176(8):1134-1145.
 Jeppesen J, Hein HO, Suadicani P, Gyntelberg F. Low triglycerides-high high-density lipoprotein cholesterol and risk of ischemic heart disease. Arch Intern Med. 2001 Feb 12;161(3):361-6.
 Arvizu, M., Afeiche, M. C., Hansen, S., Halldorsson, T. F., Olsen, S. F., & Chavarro, J. E. (2018). Fat intake during pregnancy and risk of preeclampsia: a prospective cohort study in Denmark. European Journal of Clinical Nutrition. doi:10.1038/s41430-018-0290-
 Park MK, Li WQ, Qureshi AA, Cho E. Fat Intake and Risk of Skin Cancer in U.S. Adults. Cancer Epidemiol Biomarkers Prev. 2018;27(7):776–782. doi:10.1158/1055-9965.EPI-17-0782
 Martin Feelisch, Victoria Kolb-Bachofen, Donald Liu, Jon O. Lundberg, Lucia P. Revelo, Christoph V. Suschek, Richard B. Weller, Is sunlight good for our heart?, European Heart Journal, Volume 31, Issue 9, May 2010, Pages 1041–1045, https://doi.org/10.1093/eurheartj/ehq069