The Contrary Evidence
Every robust health intervention has a body of research that appears to contradict the mainstream consensus, and engaging with that contrary evidence rather than dismissing it is how confident claims get replaced by accurate ones. In this case, the contrary evidence comes from a combination of null results in well-designed trials, mechanistic arguments about confounding variables in positive studies, and the observation that populations that do not follow the recommended behaviour appear to achieve comparable health outcomes through different pathways.
Engaging seriously with contrary evidence does not mean concluding that the mainstream position is wrong. It means holding the mainstream position with appropriate uncertainty, understanding the specific conditions under which the evidence is most and least reliable, and recognising the difference between "the evidence supports this" and "this is definitely correct." The former is achievable; the latter rarely is in complex biological systems where multiple variables interact in ways that controlled studies cannot fully capture.
The Confounding Problem
Confounding is the primary methodological challenge in health research, and it is more pervasive and more difficult to correct than most popular health communication acknowledges. People who adopt health-promoting behaviours differ from people who do not on dozens of variables simultaneously — socioeconomic status, health consciousness, access to healthcare, stress levels, social support networks — and statistical adjustment for these variables can only go so far when many of them are correlated with each other and with the outcome of interest.
The randomised controlled trial was designed to solve the confounding problem by randomly assigning the intervention rather than letting people self-select into it. But RCTs have their own limitations: they are expensive, they cannot blind participants to behavioural interventions, they tend to recruit atypical populations who volunteer for research, and their follow-up periods are constrained by funding cycles rather than by the timescales over which health effects actually operate. Understanding the hierarchy of evidence — and the specific limitations of each level — is prerequisite to correctly weighting any health research finding.
The Practical Bottom Line
Given the evidence as it currently stands — with its genuine uncertainties and its genuine signal — the practical approach for an intelligent person is to implement the intervention at the dose with the most consistent evidence, measure the outcomes that matter to you personally rather than the surrogate endpoints that are easy to measure, and update your approach when new evidence changes the picture. This is not hedging; it is the correct epistemic response to a genuine state of uncertainty that exists at the research frontier in almost every domain of health science.