Bacon rashers, statistics, and controversy

Tim Spector and Christopher Gardner discuss the recent recommendations that there is no need to cut red meat

“No Ifs or ButtiesBacon is safe to eat after all,” said the headline in The Sun. The latest of summary meta-analysis reports was published in the Annals of Internal Medicine last week on the evidence for the effects of eating meat eating. The Canadian authors conclude that we have grossly over-estimated the harmful effects of both red meat and processed meat on cancer and mortality and that we should not alter our current habits. This comes soon after a succession of other high-profile reports which reached the opposite conclusion, with some even suggesting that eating bacon was as dangerous as smoking cigarettes. Current NHS and US guidelines are that we should modestly reduce intakes of meat to less than 70 grams a day, which is equivalent to a rasher of bacon. No wonder the public is confused, not to mention the health professionals who are trying to advise them.

An important clarification here is that there are no new data being considered. The publication is from health service researchers who applied a stringent set of evaluation criteria (called GRADE) they created at McMaster University to the existing studies. The criteria are well suited to studies of pharmaceuticals when there are randomized, double-blind, placebo-controlled trials available, but not for studies of meat, eggs, vegetables or any specific food.

Investigation of dietary risk factors has to rely on a range of evidence from test-tube studies, animal studies, observational studies and clinical trials of different durations and follow up. These clinical trials can have hard outcomes like death or heart attacks or changes in risk factors like lipids or blood pressure. Meta-analysis involves the summation of evidence in a systematic way from a variety of sources, which is intended to be objective.  However, the investigator is responsible for pre-selecting criteria to include and exclude studies, which introduces subjectivity into the process. The latest Canadian meta-analysis, used the same observational data as previous reports, yet came to radically different conclusions, primarily because they dismissed them as merely associations, not causation. Given that RCTs with foods or food groups are not possible, dismissing the next best available evidence such as recent data from UK Biobank is tantamount to concluding “we know nothing about diet,” which isn’t true. The authors also chose to leave out all artificial ex-vivo lab studies and animal studies, included in other meta-analyses, that have, for example, implicated nitrites or acrylamides as carcinogens. Many people however agree that drawing any human generalisations from these data is difficult.

The Canadian team did identify 12 RCTs to include, and excluded many others with meat data using questionable logic. For example, the Lyon Heart study was dismissed for being untrustworthy because it was halted early by a data monitoring group when benefits were seen. The PREDIMED study, that specifically included red meat reduction, was also not included. However, the authors did include the large Women’s Health Initiative with over 40,000 women, even though red meat restriction was not part of the Low-Fat design. And yet the authors report that this is essentially the only RCT they found trustworthy, and the trial that dominates the analysis of hard outcomes. For risk factor outcome RCTs (e.g., blood pressure, blood lipids) they excluded all RCTs shorter than 6 months, which eliminated many available and relevant studies. Previous summaries, which have been more inclusive, have shown modest health risks for red meat, but clear links between processed meat, heart disease and early death. The evidence for cancer causation is less clear-cut, and several trials of meat reduction have not resulted in lowering of cancer risks. 

The debate around the conflicting results and conclusions highlights that the meta-analysis can no longer be assumed to be the gold standard method of assessing many of the problems of nutritional research. We have only limited data to make important decisions crucial to our health. Despite diet being the number one modifiable factor in most chronic diseases, proper funding for research on any decent scale is still woeful. We have been distracted for too long from properly assessing the potential harms of ultra-processed foods, particularly those with low quality meats—notably, the first RCT of ultra-processed food was only published this year.  

Summarising the evidence of all the studies is not easy, but in our view, we should all be eating less meat and more high fibre, nutrient-dense foods that include many vegetables, legumes, fruits and nuts and seeds. The evidence of cancer causation from eating red or  processed meat is weak, and the previous comparisons with cigarettes are highly misleading. But for other diseases, the case against processed meats having a significant risk is still relatively strong. The case against red meat is weaker and may depend on quantity and quality or the emerging influence of our individualised gut microbiomes that could be crucial in determining risk and explain some of the inconsistencies in previous studies. No food item should ever be analysed in isolation; whatever is removed has to be substituted with something else. Cutting out meat allows more space on the plate for vegetables and fibre, which come with a large side-serving of evidence supporting their benefits for good health.

Over and above the scientific arguments, the media coverage highlights the increasing polarisation of views and suspicion of the motives behind research. When previous studies suggested meat (especially processed meat) was dangerous, critics attacked the vegan and low-fat lobbies for being behind them. When this latest report emerged, social media was rife with stories of how the powerful US cattle lobby influenced the conclusions.  It emerged that in 2017 the lead author had published a similar meta-analysis in the same journal concluding the evidence for adverse health effects of sugar were weak; that study was funded by ILSI, known to be a front group for many of the Big Food corporations. In the modern era of fake news, rich industry lobbyists and the sad history of the cigarette and pharmaceutical industries, impartiality is hard to prove.

Whichever experts we choose to believe on the health benefits or risks of red or processed meat we can no longer ignore the fact that around a quarter of global warming is due to agriculture, and the greatest single factor and most of our land is used in raising highly inefficient beef. Apart from giving up a car and aeroplanes, dramatically reducing our meat intake is probably the most important act we can do for the planet.

Tim Spector, Professor of Genetic Epidemiology, Kings College London. Twitter: @timspector

Competing interests: TS is author of The Diet Myth and eats meat once or twice a month.

Christopher Gardner, Professor of Medicine, Stanford. Twitter: @GardnerPhD

Competing interests: CG is conducting a study comparing a meat analog to red meat for CVD risk factors impact, and does not eat meat.