Do studies on red meat consumption suffer from confounding and measurement bias?

Confounding is a real problem, but researchers try to control it

The core worry is that people who eat a lot of red meat also tend to smoke, drink more alcohol, exercise less, and eat fewer fruits and vegetables. Any of those factors could cause the disease, not the meat. This is called confounding, and it's a legitimate concern. Most large studies try to adjust for these factors statistically, but you can never be 100% sure you've accounted for everything. For example, a 2021 meta-analysis of 148 studies on red meat and cancer found that after adjusting for confounders, high red meat intake was still linked to a 9% higher risk of breast cancer and a 10% higher risk of colorectal cancer [3]. But the authors note that residual confounding (unmeasured factors) could still play a role.

A clever way around this is Mendelian randomization (MR), which uses genetic variants as a kind of natural random assignment. A 2023 MR study found that genetically predicted processed meat intake was linked to a 112% higher risk of colorectal cancer, but found no causal link for unprocessed red meat [5]. This suggests that processed meat may be the real culprit, and that some of the associations seen for 'red meat' in older studies might be driven by processed meat or by confounding. The MR approach is not perfect, but it's less vulnerable to confounding than standard observational studies.

Measurement bias is widespread and can dramatically skew results

Measurement bias comes from two main sources: how we measure what people eat, and how we define 'red meat.' People are notoriously bad at remembering and reporting their diet. Food frequency questionnaires (FFQs) are the most common tool, but they rely on memory and can be inaccurate. A 2022 study developed a blood-based 'metabolite score' for red meat intake—a set of 139 molecules in the blood that objectively reflect how much red meat someone ate. This score predicted a 17% higher risk of type 2 diabetes, and the authors argue that objective biomarkers like this could reduce measurement error in future studies [2].

Even more striking is the problem of definition. A 2021 study showed that depending on how you define 'red meat'—whether you include processed meat, processed poultry, or chicken nuggets—the estimated percentage of Americans who eat red meat can range from 47% to 75%, and the average weekly intake can vary by nearly 8 ounces [4]. That's a huge swing. If one study defines red meat as 'unprocessed beef and lamb' and another includes 'bacon, sausage, and chicken patties,' they are studying different exposures. This makes it hard to compare results across studies and can lead to conflicting findings. The authors explicitly call this 'bias introduced by heterogeneous meat food group methodology' [4].

Stronger methods are emerging to address these issues

Despite the problems, the evidence is not hopeless. Several lines of research now converge on the idea that processed red meat is more harmful than unprocessed red meat. A 2026 meta-analysis of 15 prospective studies found that processed red meat was linked to a 12% higher risk of stroke, while unprocessed red meat showed no significant association [1]. Similarly, a 2021 umbrella review of 72 meta-analyses found that both red and processed meat were linked to higher risks of several cancers, but the dose-response effects were stronger for processed meat: 50 grams per day of processed meat increased cancer risk by 8-72%, while 100 grams per day of red meat increased risk by 11-51% [6].

Randomized controlled trials (RCTs) are the gold standard for avoiding confounding, but they are rare for long-term outcomes like cancer. However, a 2022 randomized crossover trial (where people ate red meat for a period and then avoided it) showed that the same blood metabolites that predicted disease risk actually changed when people changed their meat intake [2]. This strengthens the case that the associations are causal. So while confounding and measurement bias are real, the combination of large prospective studies, Mendelian randomization, and biomarker validation is building a more reliable picture: high intake of red meat—especially processed red meat—is likely harmful, but the size of the effect is modest and varies by disease.

About These Sources

This answer is built on 6 peer-reviewed studies — published from 2021 to 2026, 1 from 2024 or later, 6 in Q1 journals, collectively cited 597 times — selected as the most relevant from 15 studies that passed quality screening, drawn from 64 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions