Food and drink account for around one third of total greenhouse gas (GHG) emissions attributable to the UK consumers (when contributions from land-use change for agriculture are included). Around half of these emissions are ‘embedded’ in imports.³ Livestock products are particularly GHG intensive, with the Food and Agriculture Organisation attributing 18% of total global GHG emissions to these (when contributions from land use and land-use change are included).⁴ Although emissions can be reduced by changing production methods, savings achieved will not be sufficient to offset the effects of rising global demand, and radical departures from ‘business as usual’ trajectories will be needed to prevent global GHG emissions from livestock production rising unsustainably.^{5 6} Even when food imports to the UK are ignored, failure to reduce domestic agricultural emissions will risk making the government's 2050 target for an 80% reduction in total UK-based GHG emissions ‘unattainable’.⁷ Considering only the final food products, the UK is approximately 50% to 90% self-sufficient in livestock production (see web appendix). However, total GHG emissions arising from the full life-cycles of livestock food products are much bigger because of the overseas emissions associated with the large quantity of cereals and soy imported to feed animals raised in the UK.

Recent large meta-analyses^8–10 have found significant increases in risk of coronary heart disease (CHD), type 2 diabetes and colorectal cancer with increased intake of processed meat (risk increases of 42%, 19% and 18%, respectively, per 50 g increase per day). A significant increase in colorectal cancer risk has also been shown with increased intake of red meat (17%/100 g increase per day).

We estimate that sustained dietary intakes at our counterfactual levels would materially reduce incidence of CHD, diabetes mellitus and colorectal cancer. Our method for calculating changes in population aggregate risks could not allow for confounding on the outcome.²¹ Our point estimates for these reductions have associated uncertainties, which we have estimated using Monte Carlo simulation, although the RR estimates may still be more uncertain than we have assumed. Our estimates have been based on meta-analyses of a limited number of reports of the association between intakes of different types of meat and the chronic diseases of interest and are therefore highly dependent on these results. A more recent meta-analysis indicates that our results may be conservative for diabetes.²² This research, including over 440 000 individuals, found a similar but statistically significant increase in risk of type 2 diabetes with unprocessed red meat intake (RR 1.19 (95% CI 1.04 to 1.37) per 100 g unprocessed red meat per day) but a far stronger association with processed meat than that used here (RR 1.51 (95% CI 1.25 to 1.83) per 50 g processed meat per day). A recent update of the WCRF/AICR meta-analysis does not change the RR estimates given in that report and used in our analyses.²³

Using meta-analyses of the association between intakes of different types of meat and the risks of vascular disease and diabetes, we have avoided simply regarding meat as a vehicle for dietary fats and assuming all associated risks to be mediated via effects on blood lipids.²⁴ This food-based approach to assessing the health effects of meat is supported by the failure of epidemiological studies to confirm expected associations between intakes of unprocessed red meat and risk of CHD, by the differing patterns of epidemiological association with unprocessed and processed red meat and by the evidence that RPM intakes are associated with other vascular risk factors, notably blood pressure.^{25 26} While we have only considered a limited range of diseases here, the incidence of stroke and a wider range of cancers could also be expected to decline.²⁷

Using 2004 Global Burden of Disease estimates for the UK,¹¹ the reduction in health losses under the counterfactual would be 50 960 disability-adjusted life-years (DALYs) per year for ischaemic heart disease, 5421 DALYs per year for diabetes and 13 761 DALYs per year for colorectal cancer. If effects on these diseases were independent of each other, total reduction in DALYs would be 70 142/year, equivalent to almost 1% of health losses from all causes in the UK in 2004. These calculations are based on the assumptions that effects on incidence-based disease burdens are proportional to effects on incidence and that the results based on the diets of 19–64 year-olds are applicable to the over-65 population, where the majority of the disease burden lies.

The predicted reduction in GHG emissions would equate to a total saving in UK food- and drink-associated emissions of 27.8 million tonnes CO₂-e/year across the 2009 UK population. To put this into context, the UK GHG ‘footprint’ has been estimated (using production-based accounts) at 10.16 tonnes CO₂-e/person/year.^{28 29} Total emissions attributable to the UK consumers will exceed this by perhaps 30%–40% due to large net imports of embedded GHG.³⁰ This implies that consumption-based emissions are over 14 tonnes CO₂-e/person/year. Emissions reductions under the counterfactual therefore represent a saving of over 3% of this figure, a worthwhile amount given that climate change mitigation is going to require contributions from diverse sources.

Recent work for the UK Committee on Climate Change (CCC) has modelled the reductions in GHG emissions both in the UK and in overseas resulting from three specified dietary change scenarios.^{7 31} Under a scenario in which UK intake of livestock products was reduced by 50% (with a two-third reduction in all meat and the deficit replaced with plant-based foods), the reduction in GHG emissions was estimated to be 15.0 million tonnes CO₂-e/year. In a second scenario, beef and sheep meat were replaced with pig and poultry, with no overall reduction in total meat intake, resulting in a reduction in GHG emissions of 6.3 million tonnes CO₂-e/year. The dietary changes in the CCC scenarios were more extreme than the counterfactual dietary pattern taken here, with either a greater total reduction in meat or total elimination of beef and sheep meat. However, the GHG reductions estimated in this work were greater due to inflation to account for wasted food. While we made no attempt to model the impact of a reduction in waste, this demonstrates the great potential to make GHG savings even without major dietary changes through reducing waste. This approach, however, would not bring co-benefits to health.

We found that around one quarter of the UK population had habitual intakes of RPM below 55 g/day and 27 g/day for men and women, respectively, representing around two thirds (62%) and one half (51%) of their sex-specific means. Examination of the rest of the diet revealed that some, but far from all, of this reduction was offset by increased white meat intake, and remaining dietary substitution for RPM came from a wide variety of other sources.¹⁰ We have not considered beneficial effects from compensatory increases in other dietary components, especially fruit and vegetables and dietary fibre. Other assessments of the health effects of broadly similar dietary changes have found these beneficial effects to be of even greater magnitude than the reductions in harms.²⁴ Recent estimation of the health effects of the CCC dietary scenarios has found that the greatest health gains were achieved when meat was replaced by fruits and vegetables.³² The influence of increases in these foods was far greater than health benefits attributable to reductions in salt consumption or changes in the fatty acid profile of the diets. When considering both the health and environmental effects of reducing RPM consumption therefore substitute foods are important, and clear advice should be given regarding these in order that benefits are maximised.

Intakes of RPM are socially patterned, especially in women. Forty-five per cent of low (F1) but only 29% of high (F5) RPM strata for women were in social class I or II and 41% versus 19% had formal education beyond A level.¹² Although mainly outside the scope of this paper, it may also be noted that inequalities in health outcomes are produced by inequalities in health determinants, so a downward convergence of RPM intakes would be expected to yield a third benefit: a reduction in health inequalities. This is illustrated by the large potential risk reductions available to high consumers were they to converge down to the intakes of the low consumers.