Diet determines cholesterol concentrations, and cholesterol concentrations determine the prevalence of ischaemic heart disease. This paper explores the potential effects of fiscal measures on diet and ischaemic heart disease. There is a clear economic rationale for this approach: the correction of market failure caused by externalities. Externalities are said to occur when some of the costs of consumption are not borne by the consumer. When ischaemic heart disease strikes, there are costs to the community (productivity losses or indirect costs) and to the health service (direct costs). A case can therefore be made for using taxation to compensate for the external costs of an atherogenic diet.
Summary points
Current dietary patterns are partly responsible for the high risk of ischaemic heart disease in Britain, in particular among low income groups; these dietary patterns are reinforced by the material constraints of poverty
Pricing of foodstuffs encourages the purchase and consumption of a cholesterol raising diet, particularly among people with tight food budgets
By extending value added tax to the main sources of dietary saturated fat, between 900 and 1000 premature deaths a year might be avoided
The additional tax revenue could finance compensatory measures to raise income for low income groups
Econometric and health policy research should investigate the effects of price changes on diet and health
Diet, cholesterol concentrations, and ischaemic heart disease
The relation between diet, serum cholesterol concentrations, and ischaemic heart disease is relatively well understood. In individuals, serum cholesterol concentrations—or more specifically, the ratio of low density lipoprotein to high density lipoprotein—are a major determinant of the risk of ischaemic heart disease. Serum cholesterol concentrations are largely determined by the proportion of dietary energy derived from saturated or polyunsaturated fats and by dietary intake of cholesterol. The Keys equation (box), which has recently been corroborated, describes this in a simple mathematical relationship.1–3
The Keys equation
Change in serum cholesterol concentration(mmol/l)=0.031×(2Dsf−Dpuf)+1.5√Dch
Dsf=change in percentage of dietary energy from saturated fats
Dpuf=change in percentage of dietary energy from polyunsaturated fats
Dch=change in dietary cholesterol intake
In populations, average cholesterol concentration predicts the incidence of ischaemic heart disease. A rise of 0.6 mmol/l is associated with 38% increase in ischaemic heart disease mortality; an equivalent fall results in a 25-30% fall in the incidence of ischaemic heart disease within five years.4,5 In a meta-analysis, 80% of international variation in ischaemic heart disease was attributed to variation in serum cholesterol concentrations.4 Where average cholesterol concentrations have changed, the incidence of ischaemic heart disease has fallen by the predicted amount.6 The fall in relative risk of ischaemic heart disease is greater in younger age groups (an estimated 54% at age 40, 39% at age 50, 27% at age 60, 20% at age 70, and 19% at age 80 for an 0.6mmol/l fall in cholesterol concentration), and the effects in women seem to be broadly similar to those in men.4 Because lowering cholesterol from any initial value has benefits, all income groups would experience some benefits.7
Sources of saturated fat in the British diet
The main sources of saturated fat in the diet in the United Kingdom are whole milk, butter, and cheese (table 1).8 Isocaloric substitution of these foodstuffs—substitution with alternatives providing the same dietary energy in the form of carbohydrate, monounsaturated fats, or polyunsaturated fats—would lower cholesterol concentrations. Table 2 illustrates the effects of replacing half of the intake of these foodstuffs with alternatives containing monounsaturated fats or carbohydrates. The effect would be to lower cholesterol concentrations by 0.2 mmol/l and the incidence of ischaemic heart disease by between 7.6% and 10.9%.
Table 1.
Main sources of saturated fat in the British diet
Foodstuff | Contribution to total dietary saturated fat (%)
|
||
---|---|---|---|
Men | Women | All | |
Whole milk | 11 | 12 | 11 |
Cheese | 9 | 9 | 9 |
Butter | 9 | 11 | 10 |
Biscuits | 4 | 4 | 4 |
Buns, cakes, and pastries | 5 | 6 | 6 |
Puddings and ice cream | 4 | 4 | 4 |
Total from these six sources | 42 | 46 | 44 |
Table 2.
Effect of a 50% isocaloric substitution of key foodstuffs with alternatives containing no saturated fat8
Foodstuff | % Dietary saturates* | % Dietary calories† | % Reduction in saturates‡ | Fall in serum cholesterol§
|
% Reduction in ischaemic heart disease¶
|
|||
---|---|---|---|---|---|---|---|---|
Clarke | Keys | Minimum | Maximum | |||||
Whole milk | 11 | 1.75 | 0.88 | 0.046 | 0.054 | 1.9 | 2.7 | |
Cheese | 9 | 1.44 | 0.72 | 0.037 | 0.045 | 1.6 | 2.2 | |
Butter | 10 | 1.60 | 0.80 | 0.041 | 0.049 | 1.7 | 2.5 | |
Biscuits | 4 | 0.64 | 0.32 | 0.017 | 0.020 | 0.7 | 1.0 | |
Buns, cakes, and pastries | 6 | 0.96 | 0.48 | 0.025 | 0.030 | 1.0 | 1.5 | |
Puddings and ice cream | 4 | 0.64 | 0.32 | 0.017 | 0.020 | 0.7 | 1.0 | |
Total of above | 44 | 7.02 | 3.51 | 0.182 | 0.218 | 7.6 | 10.9 |
Percentage of saturated fat in the average UK diet accounted for by this foodstuff.
Percentage of dietary energy derived from saturated fat of this origin.
Reduction in saturated fat from substituting half of this foodstuff with an alternative.
Clarke=fall in cholesterol (mmol/l) resulting from this change in diet (Clarke coefficients)3; Keys=fall in cholesterol (mmol/l) resulting from this change in diet (Keys coefficients).1
Minimum calculated using Clarke's coefficients and Law's lower estimate of effect on ischaemic heart disease3 4; maximum calculated using Keys' equation and Law's upper estimate of effect on ischaemic heart disease.1 4
The limits of informed consumer choice
In recent years the dietary gap between the rich and poor has widened.9 This is hardly surprising. Important financial, material, and cultural constraints prevent people on low incomes from acting on dietary information. There is also considerable disinformation concerning diet and ischaemic heart disease.10 Relying on informed consumer choice alone is therefore likely to widen differentials in nutrition between the rich (who have the means to act on information) and the poor (who do not). Is there another way?
Economic models of demand for foodstuffs
As dietary foodstuffs are purchased, it follows that the total contribution of a foodstuff to diet is likely to be proportional to consumer demand. In economic models, demand for a good is a function of its own price, the price of other goods, overall purchasing power (income), and “consumer taste.” One model of demand (which is supported by empirical data) says that the demand for certain broad categories of consumption goods is “separable.”11 This means that a change in the price of goods unrelated to foodstuffs (such as housing or clothes) will affect demand for foodstuffs in much the same way as a change in overall income. For example, if housing became cheaper, it would affect demand for foodstuffs in the same way as an increase in income. This means that demand for different kinds of foodstuffs is affected mainly by their relative prices and by the “food budget,” the amount of household income available to be spent on food.
Two strategies for improving nutrition emerge from this discussion. Firstly, increasing the food budget will improve nutrition, as higher income groups in Britain typically have more nutritious food consumption patterns.9 Secondly, systematically altering the relative prices of different foodstuffs will affect food consumption.
Estimating the effects of price changes
Consumer goods that have near substitutes have a high price elasticity of demand—that is, a small proportionate increase in the price leads to a large proportionate change in demand. Small changes in the relative prices of near substitutes can lead to large changes in consumption patterns. For example, a small price difference (about 10%) between leaded and unleaded petrol was sufficient to cause a considerable shift to unleaded petrol and to encourage manufacturers to produce cars that could use the cheaper fuel. Using price changes to alter food consumption is therefore most likely to be effective where foodstuffs have a high price elasticity of demand. It may even stimulate manufacturers to produce cholesterol lowering or cholesterol neutral foodstuffs.
There is little information in the public domain on the specific price elasticities of whole milk, butter, cheese, biscuits, buns, cakes and pastries, puddings, and ice cream. In the absence of empirical data we have to make some judicious estimates. In general, foodstuffs tend to have price elasticities of a magnitude of less than one.12 Price elasticities are likely to be larger where there are near substitutes. Given that reduced fat milks are near substitutes for whole milk, let us assume that the price elasticity of demand for whole milk is near to −1.0. This means that a 1% increase in price would lead to a 1% fall in consumption. Margarine is an acceptable substitute for butter: we can assume a smaller price elasticity for butter, perhaps −0.7. Reasonable substitutes—with more polyunsaturated fat and less saturated fat—for biscuits, buns, cakes and pastries, puddings, and ice cream can be manufactured. We can assume a high price elasticity of demand, perhaps −1.0. There is, however, no cholesterol neutral substitute for cheese, and it is likely that the price elasticity is low, perhaps −0.5.
At present most foodstuffs are exempt from value added tax. The simplest way of changing prices within the existing taxation framework would be to extend value added tax (currently 17.5%) to the principal sources of dietary saturated fat while exempting cholesterol neutral foods that are currently taxed (such as orange juice and low fat frozen yoghurt). Whole milk is likely to be substituted with semiskimmed milk, so that saturated fat intake will fall by half the overall reduction in consumption. Margarines rich in polyunsaturates have a neutral effect on cholesterol, so substitution reduces cholesterol concentrations proportionately. Biscuits, buns, cakes and pastries, puddings, and ice cream could be taxed if they raised cholesterol concentrations but exempt if the ratio of polyunsaturates to saturates (and trans fatty acids) were more favourable. Realistically, substitutes would also be likely to be cholesterol raising, so intake of saturated fat would fall by half the overall reduction in consumption. As table 3 shows, with these substitutions the incidence of ischaemic heart disease falls by between 1.8% and 2.6%. Given a proportionate fall in ischaemic heart disease mortality, this would prevent between 1800 and 2500 deaths a year, between 900 and 1300 of these in people aged under 75.13 Using the age specific reductions in ischaemic heart disease given above (a proxy for age specific mortality reductions) gives a similar estimate: about 900 to 1000 deaths are avoided in people under 75 (table 4).4
Table 3.
Effect of extending value added tax at 17.5% to the main sources of dietary saturated fat on the incidence of ischaemic heart disease
Foodstuff | Price elasticity of demand | % Reduction in dietary calories from saturates | Fall in serum cholesterol
|
% Reduction in ischaemic heart disease*
|
|||
---|---|---|---|---|---|---|---|
Clarke | Keys | Minimum | Maximum | ||||
Whole milk† | −1.0 | 0.15 | 0.010 | 0.012 | 0.4 | 0.6 | |
Cheese | −0.5 | 0.13 | 0.008 | 0.010 | 0.3 | 0.5 | |
Butter | −0.7 | 0.20 | 0.013 | 0.015 | 0.5 | 0.8 | |
Biscuits | −1.0 | 0.06 | 0.004 | 0.004 | 0.2 | 0.2 | |
Buns, cakes, and pastries† | −1.0 | 0.08 | 0.005 | 0.006 | 0.2 | 0.3 | |
Puddings and ice cream† | −1.0 | 0.06 | 0.004 | 0.004 | 0.2 | 0.2 | |
Total | — | 0.67 | 0.044 | 0.052 | 1.8 | 2.6 |
Minimum calculated using Clarke's coefficients and Law's lower estimate of effect on ischaemic heart disease3 4; maximum calculated using Keys' equation and Law's upper estimate of effect on ischaemic heart disease.1 4
Substitutes for these foodstuffs are assumed to contain half as much saturated fat, eg whole milk is substituted with semiskimmed milk.
Table 4.
Estimated numbers of deaths due to ischaemic heart disease avoided by extending value added tax to sources of dietary saturated fat
Age band | % Fall in risk per 0.6 mmol/l fall in cholesterol | Total No of deaths*
|
Estimated No of deaths avoided†
|
||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Minimum
|
Maximum
|
||||||||||
Men | Women | % Fall in risk | Men | Women | % Fall in risk | Men | Women | ||||
35-44 | 54 | 865 | 157 | 4.0 | 34 | 6 | 4.7 | 40 | 7 | ||
45-54 | 39 | 3 662 | 715 | 2.9 | 104 | 20 | 3.4 | 124 | 24 | ||
55-64 | 27 | 9 332 | 2 900 | 2.0 | 184 | 57 | 2.3 | 218 | 68 | ||
65-74 | 20 | 21 507 | 10 915 | 1.5 | 315 | 160 | 1.7 | 372 | 189 | ||
75-84 | 19 | 24 788 | 23 018 | 1.4 | 345 | 320 | 1.6 | 407 | 378 | ||
Total | 60 154 | 37 705 | 982 | 564 | 1161 | 666 |
England and Wales, 1996.
Minimum calculated using fall in cholesterol derived from Clarke's coefficients; maximum calculated using fall in cholesterol derived from Keys' equation.
Equity
Taxation is a blunt instrument. Because poor people spend a greater proportion of their income on food than rich people, they are likely to be more sensitive to price changes. They are also at higher risk of ischaemic heart disease. The health benefits of such a policy are therefore likely to be progressive. Paradoxically, this also means that the effects on material wellbeing are likely to be regressive. Most consumers will end up by spending more on food and this will disproportionately affect the poor. An important part of such a strategy should therefore be to compensate low income groups by raising their incomes. The most directly targeted approach would be to simultaneously raise the value of welfare benefits, particularly those intended for children in low income groups, who are in any case not the target of this policy. Since food taxation would raise revenue, the overall effect on government finances would be neutral.
Conclusions
The assumptions in this paper are somewhat conservative. No account is taken of the effect of raising low incomes or the likelihood that people at highest risk would benefit most. Price changes might increase the consumption of fruit and vegetables or reduce levels of obesity, but these considerations are beyond the scope of this paper. Given that there are potential benefits to a fiscal food policy, how politically feasible would such a policy be? In the short term the answer is probably “not at all,” as the present government has pledged not to extend value added tax to foodstuffs. But in the longer term? The notion that taxation might be used to adjust for externalities is gaining currency among policymakers—for example, the “polluter pays” principle and road pricing to reduce congestion. Nor is the use of taxation to influence health new: since 1993 it has been policy to increase the real level of tobacco duties by 3% every year.14 Cigarette taxation raises the same dilemma regarding equity. Low income groups tend to smoke more and are more price sensitive than high income groups. They therefore benefit the most from taxation, but disproportionately bear the tax burden.15
The nutritional and physiological parts of this model are relatively robust. The assumed relation between purchase of foodstuffs and food consumption is probably reasonable. The weak link is undoubtedly the assumed effects of price changes on purchase of foodstuffs. How reasonable are these assumptions? Could the impact of price changes be even greater? Neither searching economic research databases (EconLit, BIDS IBSS) nor writing to major supermarkets produced estimates of price elasticities of demand for these foodstuffs. One supermarket hinted that the data existed but were commercially sensitive. Yet this information is essential. If we are serious about improving nutrition a fiscal food policy is worth exploring. Interdisciplinary collaboration is needed between econometricians and nutritionists to investigate empirically the effects of price changes on the purchase of foodstuffs.
Footnotes
Competing interests: None declared.
References
- 1.Keys A, Anderson J, Grande R. Serum cholesterol response to changes in the diet. IV. Particular saturated fats in the diet. Metabolism. 1965;14:776–786. doi: 10.1016/0026-0495(65)90004-1. [DOI] [PubMed] [Google Scholar]
- 2.Tang JL, Armitage JM, Lancaster T, Silagy CA, Fowler GH, Neil HAW. Systematic review of dietary intervention trials to lower blood total cholesterol in free-living subjects. BMJ. 1998;316:1213–1220. doi: 10.1136/bmj.316.7139.1213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Clarke R, Frost C, Collins R, Appleby P, Peto R. Dietary lipids and blood cholesterol: quantitative meta-analysis of metabolic ward studies. BMJ. 1997;314:112–117. doi: 10.1136/bmj.314.7074.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Law MR, Wald NJ, Thompson SG. By how much and how quickly does reduction in serum cholesterol concentration lower risk of ischaemic heart disease? BMJ. 1994;308:367–373. doi: 10.1136/bmj.308.6925.367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Holme I. An analysis of randomized trials evaluating the effect of cholesterol reduction on total mortality and coronary heart disease incidence. Circulation. 1990;82:1916–1924. doi: 10.1161/01.cir.82.6.1916. [DOI] [PubMed] [Google Scholar]
- 6.Jousilahti P, Vartiainen E, Pekkanen J, Tuomilehto J, Sundvall J, Puska P. Serum cholesterol distribution and coronary heart disease risk: observations and predictions among middle-aged population in eastern Finland. Circulation. 1998;97:1087–1094. doi: 10.1161/01.cir.97.11.1087. [DOI] [PubMed] [Google Scholar]
- 7.Law MR, Thompson SG, Wald NJ. Assessing possible hazards of reducing serum cholesterol. BMJ. 1994;308:373–379. doi: 10.1136/bmj.308.6925.373. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gregory J, Foster K, Tyler H, Wisemann M. The dietary and nutritional survey of British adults. London: HMSO; 1990. . (Office of Population Censuses and Surveys, Social Survey Division.) [Google Scholar]
- 9.James WPT, Ralph A, Leather S. Socioeconomic determinants of health: the contribution of nutrition to inequalities in health. BMJ. 1997;314:1545–1548. doi: 10.1136/bmj.314.7093.1545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Tunstall-Pedoe H. Did MONICA really say that? BMJ. 1998;317:102. doi: 10.1136/bmj.317.7164.1023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Moschini G, Moro D, Green RD. Maintaining and testing separability in demand systems. Am J Agric Econ. 1994;76:61–73. [Google Scholar]
- 12.Van Driel H, Nadall V, Zeelenberg K. The demand for food in the United States and the Netherlands: a systems approach with the CBS model. J App Econometrics. 1997;12:509–523. [Google Scholar]
- 13.Office for National Statistics. Mortality statistics: cause. England and Wales, 1996. London: Stationery Office; 1998. [Google Scholar]
- 14.Townsend J. Policies to halve smoking deaths. Addiction. 1993;88:37–46. doi: 10.1111/j.1360-0443.1993.tb02762.x. [DOI] [PubMed] [Google Scholar]
- 15.Townsend J, Roderick P, Cooper J. Cigarette smoking by socioeconomic group, sex, and age: effects of price, income, and health publicity. BMJ. 1994;309:923–927. doi: 10.1136/bmj.309.6959.923. [DOI] [PMC free article] [PubMed] [Google Scholar]