Can fat taxes prevent heart disease?

Two terms don’t always mean what we believe they do: “research” and “evidence-based.” Take the research just published in the Journal of Epidemiology Community Health entitled: “Evidence-based Public Health Policy and Practice: Could targeted food taxes improve health?”

According to its authors, “yes.” They concluded: “A carefully targeted fat tax could produce modest but meaningful changes in food consumption and a reduction in cardiovascular disease.”

But a close look finds the evidence doesn't support these conclusions and that their findings had been derived from the most problematic tool used in epidemiology: a mathematical computer model.

Model making

Every stage in the creation of a computer model involves arbitrary decisions and assumptions, oftentimes hundreds, about what variables to use and which to omit, how to define them and to what degree variables affect things. These assumptions may not be what clinical evidence can support, but they can quickly become buried in the model. Models also somehow try to control or adjust for endless variables the creator believes might confound his/her findings.

“You can build in many assumptions that might well be unjustifiable under independent examination,” said Dr. John Brignell, former chair of Industrial Instrumentation at the University of Southampton. “In just a few hours you can create a model, just a computer program, which is so complex that no outsider can hope to unravel it,” he said. That can make it virtually impossible for the study results to be audited or tested. In his book, Epidemiologists — Have they got scares for you!, he outlined the main hazards of computer modelling, warning:

It is relevant...to comment on the touching faith that politicians and the media have in computer models. It is generally true to say that most large computer models are not worth the magnetic oxide they are written on, and I say that as one who has been computer modelling for over 40 years. Often they are written in the glaring absence of knowledge of the fundamental interactions on which they ought to depend... Yet they are used as the excuse for profound and often extremely damaging policies that affect everyone. That is why computer models are dangerous tools.

For their model, the authors in this study attempted to predict the number of lives saved by assuming that lowering salt and saturated fats in the diet (assuming low-fat diets work by lowering blood cholesterol levels) would reduce deaths from cardiovascular disease and strokes. For example, they estimated that every 3 gram/day reduction in salt intake would lower incidence of heart disease by 9-10% and strokes by 12-14%. Their estimated health benefits for low-fat diets were based on the assumption that every 1 mg/dl reduction in low-density lipoprotein (“bad cholesterol”) would reduce heart disease by 1%.

Without the need to go any further, it'a already clear these assumptions contradict and exceed the actual clinical evidence on the ability of “heart healthy” low-salt diets and low-fat diets to prevent deaths from cardiovascular disease and strokes. In fact, as we’ve examined, even the latest Cochrane review of 39 clinical trials conducted in multiple countries over the course of three decades on the ability of “heart healthy” dietary interventions (reducing saturated fats and salt) and lifestyle interventions found: “Contrary to expectations, these lifestyle changes had little or no impact on the risk of heart attack or death...”

But the study’s weaknesses didn’t end with these flawed assumptions about the ability of certain ingredients to prevent heart disease deaths. The authors took food consumption, expenditure and economic data on food elasticity values from Britain’s National Food Survey 2000, an annual survey using interviews and food diaries, to predict how increasing costs could decrease consumption of certain foods. They assumed that food purchases equaled consumption, and that no food was discarded or not eaten. They then used trial and error to find "the maximum potential gain with a taxation level of 17.5%," taking three approaches in their modelling: the effects of -

· taxes on “unhealthy” dairy products, like cheese and butter, and foods high in saturated fats

· taxes on foods according to their “healthfulness” on the SSCg3D score (it estimates how unhealthy a food is based on its calories, saturated fat, salt, and sugar content, subtracted by its produce content, iron, calcium and omega-3 fatty acids)

· and, for the “best outcomes,” taxes on a wider range of foods to reduce consumption of saturated fats, salt, sugar and calories.

Their first approach taxing foods high in saturated fats came up as unlikely to reduce cardiovascular disease; the second, taxing foods by their healthfulness, “might avert 2,300 deaths” per year in the UK, they said; and lastly, taxing the greatest variety of foods - all those popularly seen as “unhealthy” - yielded a prediction that 3,200 cardiovascular deaths could be prevented.

The media ran with the biggest estimate with news stories, headlining:

“Fat tax will save 3,000 lives a year”

A fat tax should be imposed on junk food, say doctors A fat tax should be imposed on unhealthy foods to fight the escalating obesity epidemic, doctors say. Up to 3,200 lives a year could be saved if VAT [value added tax] was added to a third of the food chain, including milk, bacon and junk food, a new paper claims....

This claim is well on its way to becoming popular wisdom and few consumers will ever know the faulty assumptions behind the predictions, the authors’ own warnings about interpreting their results, or the things that hadn’t even been included in their model which could more than wipe out any possible benefits.

Watch for the caveats and missing data

Their model didn’t look at the health effects of other nutrients in foods. Its predictions used population data and didn’t consider things that influence individual decisions in food purchases. Nor did it include things besides diet that affect health, such as socioeconomic situations, age, gender and lifestyles. It also assumed that higher food taxes would have no impact on the amount people dined out. And, of course, their data was based on British people and couldn’t credibly be applied to other populations.

For these reasons, the authors cautioned that their findings should be interpreted carefully. They admitted that food consumption was difficult to predict and that their “model suggests that there could be a variety of unintended potentially detrimental effects.” For example, they found that reducing saturated fat tended to increase salt consumption, reducing salt lowered the consumption of low-fat foods, and that taxes on milk and “unhealthy” foods also reduced fruit and vegetable consumption. In fact, they said “all three taxation estimates predicted a fall in fruit and vegetable consumption of approximately 2–4%.”

In their Results, they also acknowledged the limited effectiveness of taxation beyond the current rates of 17.5% value-added taxes already on confections, snacks and most drinks; yet, with no supportive evidence, went on to claim that higher taxation might achieve better results:

Food consumption is relatively insensitive to price changes, such that a taxation rate of 17.5% is likely to reduce the intake of nutrients such as salt and saturated fats by no more than 5–10%. So the scope for significantly altering the national diet by judicious use of VAT seems limited. Greater change could be achieved with a higher level of taxation, but this is unlikely for political and economic reasons.

Most disturbingly absent from their computer model was consideration of any potential harmful effects of compelling “heart healthy” diets.

Salt reduction, for example, doesn’t appear entirely benign, according to growing medical research. The European Society of Cardiology Guidelines for the Management of Arterial Hypertension, for instance, reported recent research showing low-salt diets can have negative effects: activating the renin-angiotensin system and the sympathetic nervous system, increasing insulin resistance and hypodehydration (especially with the elderly). This, they concluded, could lead to increased risks for cardiovascular disease. Similarly, people who might benefit from salt in their diets wouldn’t be helped, but weren’t included in their computer model, either. Salt also improves the flavor of many nutritious foods, helping to prevent nutritional deficiencies especially among vulnerable populations, such as children and elderly.

Nor did they consider the harmful affects of low-fat, low-calorie diets for vast segments of the population, especially growing children, teens, pregnant women and elderly. For instance, as we’ve seen here, a notable percentage of senior citizens are not getting enough calories, fats and proteins, and energy-dense foods and are suffering from poor nutritional status, according to NHANES and medical studies. The medically-documented consequences include delayed wound healing, increased risks of infection, damaged heart and intestinal functions, longer hospital stays and higher rates of complications and higher mortality rates, depression, apathy, functional decline, loss of muscle strength, falls and increased fractures. And children and teens trying to “eat healthy” are coming up increasingly deficient in calories and vital nutrients needed for optimum growth and development, and are developing dysfunctional eating that leaves them with life-long struggles with food. Far more young people are dying from anorexia than heart disease.

Among the “unhealthy” foods identified in this study’s computer model were “processed” foods like frozen vegetables and convenience foods which, for busy working people and families with young children, provide perfectly nutritious meals. The unintended consequences of taxing them to unaffordability, could be fewer vegetables consumed and turning to more fast food meals. And, as we saw in the InCHIANTI study, the elderly suffer from high rates of undernutrition, short on calories and almost all macro- and micronutrients. Advanced age makes it physically more difficult for them to shop, cook and chew foods and the InCHIANTI researchers noted that more processed and prepared foods could be the greatest help in reducing the percentage suffering from poor nutrition.

Worse, the impact of higher food costs on the poor wasn’t considered in their computer model, yet each of their taxation approaches raised the overall weekly food bills for consumers — by 3.2%, 4% and 4.6%, respectively. As we’ve seen here, more expensive food means less of it for low-income households. The latest government food insecurity report shows that the percentage of low-income households suffering “very low food security” (the new term for hunger) in the U.S., continues to increase, and reached 12.6% in 2005.

The study authors acknowledged that fat taxes would be regressive, meaning low-income households would pay a higher percentage of their income on food. But, rather than factor this into their computer model, they viewed this situation positively, maintaining: “[T]heoretically, those on low incomes should be more price sensitive in their pattern of demand and therefore may be more likely to change their consumption patterns and obtain larger proportional health benefit.”

The take home message is to never assume that a “study” actually conducted research as we may think of that term, and that claims of something being “evidence-based” doesn’t always mean it is. Whenever someone claims to tell us they know best how we, and everyone else, should eat and live, and proposes some costly initiative to compel compliance, we would be wise to critically look at the science.

© 2007 Sandy Szwarc