It seems I’m back in a time loop, repeating the same old critiques of the same old flaws in studies on the same old topic. That’s right: this week has brought another round of a very familiar cycle – a cycle that starts with an observational study reporting health hazards with meat consumption, progresses to media hype, which leads to a flood of emails from concerned readers, which prompts my attempt to put the results in perspective.
So what kicked off this latest trip around the merry-go-round? This week, a publication by authors Li et al. described positive associations between meat intake and risk for type 2 diabetes (T2D) across various populations.1 The findings were quickly snapped up by The New York Times,2 and around we go…
About the study
The authors sought to investigate the potential link between consumption of meat and diabetes risk on a global scale. To do so, they conducted a meta-analysis using data from a total of 1,966,444 adults (≥18 years of age) across 31 independent cohorts included in the InterConnect project, an international research effort to explore genetic and environmental risk factors for obesity and diabetes. These cohorts represented a spectrum of global populations, with 12 of the included cohorts from the Americas (primarily the United States), nine from Europe, seven from the western Pacific/east Asia, two from the eastern Mediterranean, and one from southeast Asia.
Diet information was collected via self-report, with meat consumption being divided across three categories: unprocessed red meat, processed meat, and poultry. (Consumption of fish and seafood were not classified as meat consumption in this study.) Incident type 2 diabetes (T2D) was determined from medical records, use of medication for diabetes treatment, or self-report with additional validation such as through medication prescriptions or HbA1c assays. Participants who had had a diagnosis of diabetes at baseline were excluded, as were those who reported implausible daily calorie intakes or had not provided sufficient data on relevant exposures, outcomes, or confounding variables.
A total of 107,271 cases of incident T2D were recorded over the follow-up period (median: 10 years; interquartile range: 7-15 years). Increased intake of all three forms of meat was associated with statistically significant increases in diabetes risk after adjustment for covariates related to demographics, health (e.g., smoking, BMI, alcohol consumption, physical activity), and diet (e.g., consumption of fish, fiber, and sugar-sweetened beverages). Processed meat consumption showed the strongest association, with a 15% increase in T2D risk for every 50 g/day increase in intake (HR: 1.15; 95% CI: 1.11-1.20), followed by red meat consumption (HR: 1.10 per 100 g/day increase in intake; 95% CI: 1.06-1.15) and poultry consumption (HR: 1.08 per 100 g/day increase in intake; 95% CI: 1.02-1.14).
A familiar phrase to summarize familiar flaws
I’ve discussed studies like this one many times in the past, and the basic flaws remain the same. Yet far be it for me to be the one to stop the cycle, so as long as these studies keep getting published, I’ll keep repeating a phrase that sums up one of the critical ways in which they fall short: correlation is not causation.
The results presented by Li et al. indicate an association between meat intake and incident diabetes, but they tell us nothing about the potential causal relationships between these variables. Causality is everything in medicine (maybe life), so we must always ask ourselves this question when confronted with the data, “X correlates with Y,”: “Yes, but does X cause Y?” The notion that meat consumption causes diabetes is just one of many possible explanations for the association.
When it comes to diet and disease, we have virtually limitless other variables that might influence or covary with both meat intake and diabetes risk, thus presenting countless routes by which meat intake and T2D might be correlated despite lacking any true causal link between them. Diet and disease risk are linked with numerous socioeconomic factors and other variables relevant to health which might translate to differences in baseline risk between groups. For instance, relatively high intake of processed meats is often associated with lower income due to the cheaper, more shelf-stable nature of these products (e.g., hot dogs, cured meats), and low income is in turn typically associated with poorer access to quality healthcare (as might otherwise detect prediabetes and prevent progression to T2D), as well as with lower ability to engage in other healthy behaviors such as exercising regularly or ensuring adequate intake of micronutrients. While studies attempt to “correct” for these covariates, it is impossible to do so perfectly or to account for every factor which might play a role. Furthermore, avoidance of meat (through a belief that meat is harmful – cue the endless cycle of meat-is-harmful stories), might be proxy for a slew of behaviors that are, indeed, health-promoting (e.g., more exercise, more vegetable consumption, less calorie consumption, less alcohol consumption, less processed food consumption) and, by extension, T2D-avoiding.
In the specific case of the study by Li et al., we have very little insight on potential discrepancies in baseline risk, as the authors do not report baseline data on the confounding variables for which they evidently “adjusted.” However, they do share that hazard ratios associated with meat intake were substantially higher in models that excluded adjustments for BMI (HR: 1.18; 95% CI 1·07-1·29 for unprocessed red meat; HR: 1.23; 95% CI: 1.14-1.34 for processed meat; and HR: 1.21; 95% CI: 1.12-1.31 for poultry). This observation would agree with the alternative explanation that factors related to general health – rather than meat intake per se – are responsible for the apparent association between meat consumption and diabetes risk.
A tenuous association
As we’ve already seen, adding adjustment for BMI cuts the excess risk associated with meat consumption approximately in half, leaving us with a 10% excess risk for every 100 g/day of unprocessed red meat, 15% for every 50 g/day of processed meat, and 8% for every 100 g/day of poultry. Given that no study can account for every possible covariate, these numbers are small enough on their own to cause plenty of doubt about these correlations.
But beyond their limited magnitude, these associations were also found to be inconsistent. No significant associations between incident T2D and any form of meat consumption were observed in either the eastern Mediterranean or the southeast Asian populations, though these regions had the smallest representation in the study, so these analyses may have been underpowered. More tellingly, none of the regions except Europe showed significant positive association between T2D risk and poultry intake, and even when analyzing all 31 cohorts together, further adjustments for cooking method or comorbidities at baseline abolished the positive association for poultry consumption altogether.
Around and around
This is the latest in a long line of observational studies reporting a link between consumption of animal meat and a negative health outcome. It certainly won’t be the last. Despite the fact that, from a nutritional standpoint, the distinction between “animal-based” and “plant-based” is rather arbitrary, it seems that every few months brings a new report of a positive association between meat intake and some scary outcome – from diabetes and obesity to cancer and heart disease. I wouldn’t be surprised if the next paper links meat intake with the likelihood of getting struck by lighting.
But all of these studies have the same problem: they can’t tell us anything about causality. Diet is closely tied to innumerable cultural, socioeconomic, and behavioral variables with complicated (and not fully understood) relationships to disease risk. Li et al. made attempts to correct for other dietary variables and a few other factors related to general health, but even compared to many other observational studies in this space, the present investigation overlooked several key confounds, such as socioeconomic status or macronutrient ratios. (Indeed, even without these adjustments, the associations they report were weak at best.)
So around we go again in a never-ending effort to head off alarm and outcries over meat consumption resulting from flawed studies. While there may be cultural, environmental, or religious reasons why certain individuals might choose to abstain from meat consumption, we have yet to see any meaningful evidence to support adding “health” to that list.
For a list of all previous weekly emails, click here.
References
- Li C, Bishop TRP, Imamura F, et al. Meat consumption and incident type 2 diabetes: an individual-participant federated meta-analysis of 1·97 million adults with 100 000 incident cases from 31 cohorts in 20 countries. Lancet Diabetes Endocrinol. 2024;12(9):619-630. doi:10.1016/S2213-8587(24)00179-7
- Callahan A. Eating Meat Is Linked With Diabetes Risk, New Studies Suggest. The New York times. https://www.nytimes.com/2024/08/20/well/eat/red-meat-diabetes-study.html. Published August 20, 2024. Accessed August 22, 2024.
