Grant Schofield and George Henderson

On the same day that our low carb and diabetes paper was published in the New Zealand Medical Journal, a new analysis of the correlation between New Zealanders’ fat intakes and cholesterol levels from the public health epidemiology team at Otago University was published in the Australian and New Zealand Journal of Public Health.

Trends in serum total cholesterol and dietary fat intakes in New Zealand between 1989 and 2009.  Jody C. Miller, Claire Smith, Sheila M. Williams,  Jim I. Mann, Rachel C. Brown, Winsome R. Parnell, C. Murray Skeaff.
Aust NZ J Public Health. 2016; Online; doi: 10.1111/1753-6405.12504

It contains this passage:

Reduction in saturated fat, a key component of public health nutrition measures aimed at IHD risk reduction, has in the past been achieved as a result of near universal acceptance of the nutrient disease link and by collaboration among health and nutrition educators, regulators, the media, the public and food providers. This consensus has recently been threatened by a movement advocating benefits of a diet low in carbohydrate and high in fat, including saturated fat (e.g. the ‘paleo’ diet). Despite the questionable scientific basis for such a nutritional approach, it appears to have widespread appeal – to the extent that it has been shown to be associated with an increase in mean population cholesterol levels in some areas where uptake of the advice has been high.(42)

Reference 42 is a Swedish paper from 2012 showing that increased uptake of low carb diets and an increase in butter (and oil) sales and decrease in sales of “cooking margarine” has indeed been associated with an increase in cholesterol there.[1]

However, the Swedish data on heart disease shows that its incidence continues to decline (most recently in 2014). Indeed, heart attack (AMI) mortality in women had plateaued, and only began to decline again following the increase in butter sales around 2008. Going by the official Swedish health data, available here in an easy-to-search interactive database, there doesn’t seem to be anything to worry about.

But what about New Zealand? The new study compares fat intakes from dietary surveys taken in 1989 (n=1,618), 1997 (n=4,636), and 2008/2009 (n=4,721) with the mean non-fasting cholesterol levels in those populations.

So it only looks at fat, and no other dietary factor, and only looks at total cholesterol. Which is a marker even your doctor is unlikely to find very revealing these days. There is a supplementary table about HDL, which increased in young males, but not those 55 and over between 1997 and 2008/2009, and also increased in women of both age groups. But no news about triglycerides, ApoB, small dense LDL particles or any of the other atherogenic and protective factors we see in other studies about heart disease.

There’s also no mention of blood glucose and HbA1c. We know the incidence of diabetes and pre-diabetes, significant risk factors for heart disease, have increased in the study period. Does this correlate with the decrease in saturated fat intake and reduction in cholesterol? You bet it does, but you’re not going to find it mentioned here.

This narrow focus on confirming one belief about the causes of heart disease and ignoring other evidence in the same data set is a flaw in this study.

What did they find? A decrease in saturated fat intake from 15.9% of energy in 1989 to 13.1%E in 2008/2009 correlated with a decrease in cholesterol from 6.15 mmol/L in 1989 to 5.39 mmol/L in 2008/09, some of which was due to statin use by older New Zealanders.

The paper states “Although some meta-analyses have not demonstrated a convincing reduction in cardiovascular mortality and saturated fat when macronutrient substitution is ignored,(40) cardiovascular risk is reduced when saturated fat is replaced with polyunsaturated fat.(41)”

Their ref 41 is the Hooper et al 2015 meta-analysis showing a small reduction in combined CHD events (after statistical adjustment because there is no such effect for the raw data), but no reduction in heart attacks or cardiovascular or total mortality, when saturated fat is replaced with polyunsaturated fat in RCTs (the people told to eat less SFA in these RCTs were told to change their diets to be “healthier” in multiple ways,  including eating less refined grains and sugar, exercising more etc, so it’s not that great a test of the hypothesis, but anyway).[2]

So if there is such an effect, was saturated fat replaced with polyunsaturated fat in New Zealand? According to this study polyunsaturated fat intake as a % of energy declined in all groups between 1989 and 2008/2009.  The effect of replacing saturated fat with polyunsaturated fat is irrelevant for this paper – it just didn’t happen in New Zealand.

This means that New Zealanders replaced saturated fat with something else. Maybe protein, maybe starch, maybe sugar – none of this is in the paper.

How that is supposed to have reduced heart disease we don’t know.

All the reduction in saturated fat intake in this study is from people eating less dairy – less butter, whole milk, and cheese. Of course, these are the very foods associated with lower incidence of diabetes, and even lower rates of cardiovascular disease, in recent studies.[3,4,5].Under these circumstances – less fat, less saturated fat, and less polyunsaturated fat – it is hard to see how dietary factors can account for the increase in HDL between 1997 and 2008/2009. However, cigarette smoking decreased significantly between 1997 and 2008/2009 (though it actually rose a bit between 1989 and 1997) and smoking is associated with lower HDL (and, indeed, higher LDL), so this might be the explanation.[6]

There’s another contradiction at the heart of this evidence. We’ve been told often that prospective cohort epidemiological studies don’t show a link between saturated fat and heart disease because the range of saturated fat intakes in most populations is too small – everyone is exposed to too much saturated fat to see the benefit from reducing it, and people in RCTs don’t usually reduce it enough.

The old “you’re just not trying hard enough” argument!

However, in this study, the range of saturated fat intake is very small – from ~16%E in 1989 to ~13%E in 2008/2009 – still “well above internationally recommended ranges of intake” as the authors state. This is a fraction of the variation seen in cohort epidemiology – it’s like the difference between two quintiles in many studies. Yet, now we’re supposed to accept that this small variation has produced a major benefit, one which is not seen in real outcomes such as heart attacks when we compare people in larger populations whose saturated fat intakes differ much more for long periods of time [4] or indeed in RCTS where this is achieved.

If the evidence from New Zealand nutritional surveys is to be used meaningfully, it needs to be analysed for all trends for which data was collected, not just fat and cholesterol. You’ll never see the correlations you don’t look for.
One interesting finding from 2008/2009 was that reported caloric intake had decreased since 1997, from 2866 calories for men to 2479, while women’s reported caloric intake had declined from 1911 calories to 1815, though this last change wasn’t significant. Yet BMI continues to increase.

This graph from the 1997 survey report by the same authors might still have some relevance.

The claim that if some people are now eating more saturated fat (but also more polyunsaturated fat from nuts, fish, lard, and olive oil) and less carbohydrate, and eating real foods instead of processed foods, because it improves their well-being, controls their weight, and improves their TG/HDL ratio and blood sugar, then this is therefore going to increase the risk of heart disease because of the prospect of a relatively small increase in total cholesterol in the population, needs a lot more evidence than is presented in this paper.

It’s certainly possible that there are people on the fringes of comprehension who might just use a paleo news item as an excuse to double up on their meat pie with coke. Just as there were many who bought into the high-sugar low-fat yoghurt craze, and still do.
It would be good if there weren’t conflicting messages out there, but it’s in the nature of science that hypotheses are easy to generate, but can be a lot harder to lay to rest.

This study has gone way beyond what the data say. In fact, they claim things their data do not say.  The data certainly say nothing about the possible benefits or harms of people eating paleo or LCHF.

References

[1] Johansson I, Nilsson LM, Stegmayr B, Boman K, Hallmans G, Winkvist A. Associations among 25-year trends in diet, cholesterol and BMI from 140,000 observations in men and women in Northern Sweden. Nutr J. 2012;11:40.

[2] Hoenselaar B. Letter to the Editor: Further response from Hoenselaar. Br J Nutr. 2012 Sep;108(5):939-42.

Click to access 2012Hoenselaar-4.pdf

[3] Ericson, U, Hellstrand, S, Brunkwall, L, Schulz, C-A, Sonestedt, E, Wallström, P, et al. Food sources of fat may clarify the inconsistent role of dietary fat intake for incidence of type 2 diabetes. AJCN 2015;114.103010v1

[4] Praagman J, Beulens JWJ, 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 ajcn122671

[5] Mohammad Y. Yakoob, Peilin Shi, Walter C. Willett, Kathryn M. Rexrode, Hannia Campos, E. John Orav, Frank B. Hu, Dariush Mozaffarian. Circulating Biomarkers of Dairy Fat and Risk of Incident Diabetes Mellitus Among US Men and Women in Two Large Prospective Cohorts. Circulation AHA.115.018410
Published online before print March 22, 2016

[6] He BM, Zhao SP, Peng ZY. Effects of cigarette smoking on HDL quantity and function: Implications for atherosclerosis. J Cell Biochem. 2013 Jul 15.