Observational evidence is uncertain and inconsistent: 4 arguments for more nuance
- Introduction: what is the case against animal source foods?
- Weak associations are only indicative
- Beware of problematic input data
- Lifestyle confounding and healthy-user bias
- Cherry picking - dismissal of neutral outcomes
- Cherry picking - dismissal of protective outcomes
Additional commentary on the specific case of the Blue Zone argument can be found elsewhere on this website.
Introduction: what is the case against animal source foods?
Observational evidence has linked 'plant-based' eating to a lower risk of an impressive series of health issues, including obesity, cardiovascular disease, diabetes, diverticular disease, eye cataract, cancer, and vulnerability to coronavirus infection. However, such data are largely insufficient from a methodological perspective to support dietary recommendations away from animal source foods. This applies to the case against red and/or processed meats as well, despite reported associations between their intake and various illnesses, including intestinal disease, asthma, pneumonia, kidney disease, type-2 diabetes, cardiometabolic disease, and cancer. These associations do not necessarily imply that animal source foods are intrinsically unhealthy, and more rigorous research is needed to draw definitive conclusions.
Weak associations are only indicative
Methodological quality assessments have revealed that the evidence currently used to relate the intake of animal source foods, such as red meat, to an increase in mortality or morbidities is uncertain. Such low-certainty evidence is suggestive at best. For the general population, it does not strongly support the dietary recommendation to reduce consumption beyond current levels. In contrast, other risk factors linked to the Western diet, like visceral fat and the metabolic syndrome, show much stronger and plausible associations with certain health outcomes now allegedlyascribed to ASFs. The use of Bradford-Hill criteria and GRADE-based assessments concludes that the evidence incriminating red meat for increased mortality and morbidities is of low to very-low certainty. It may as well be classified as merely suggestive. For the general population, the evidence does not strongly support the recommendation to reduce consumption beyond current levels.
Beware of problematic input data
Epidemiological research usually relies on surveys and food frequency questionnaires – imperfect, and sometimes fatally flawed methods of data collection. The input data suffer from categorization issues, errors due to memory-based reporting, social desirability bias, and survivorship bias (including individuals who fare well on certain diets, while overlooking those who do not). Studies on vegetarian diets may also be problematic because of misclassifications, combining true vegetarians with semi-vegetarians, leading to potential errors in the results.
Lifestyle confounding and healthy-user bias
Disentangling specific dietary effects from the complexity of overall diets and lifestyles in observational data is challenging. Context is crucial when interpreting statistical associations between food intake levels and health outcomes. The relationships are obscured by the overall diet quality, lifestyles, socio-demography, etc. The healthy user bias poses specific challenges. In high-income countries, associations tend to capture what the generally healthier middle-classes believe is healthy, rather than revealing the true biological effects. Studies indicate that both healthier vegetarians and meat eaters have lower risks compared to less healthy meat eaters. Population-specific factors need to be considered when interpreting the results of observational studies on meat consumption.
Cherry picking - dismissal of neutral outcomes
Developing a case against animal source foods is often based on cherry-picked evidence from observational studies. In many studies there is no unambiguous association between their avoidance and improved health. The evidence on unprocessed red meat, for instance, is conflicting, some large studies showing no association. Poultry, eggs, and dairy display mostly neutral or even protective associations with health. Butter's associations with mortality, cardiovascular disease, and diabetes are relatively small or neutral. The overall picture is complex, and context plays a crucial role in interpreting the results from observational studies on animal source foods.
Cherry picking – dismissal of protective outcomes
Embracing observational data to criticize omnivore diets would undermine the notion that vegan diets are always superior for avoiding chronic diseases and death. Those who rely on nutritional epidemiology to argue that we need more ‘plant-based’ eating need to acknowledge that some animal source foods also come with protective associations – implying that their underconsumption may be suboptimal for certain aspects of health. The global PURE study, for instance, has associated the intake of fish, eggs, full-fat dairy, and white meat, along with plant foods, to protective health outcomes. In the PURE cohort, red meat is mostly considered neutral in terms of health outcomes. In some studies, red meat consumption has even been associated with benefits, including lower morbidity, increased telomere length, and decreased risk of depression. Also, some studies have linked vegetarian and vegan diets to poorer health outcomes and lower quality of life. However, it is essential to recognize that both pro-meat and pro-plant studies may have limitations in establishing causal relationships, and they should be considered together in the overall debate about dietary choices.