Once you adjust for confounders, the health benefits of plant-based eating disappear
The study usually cited, from the UK Biobank, measured how many vegetables people ate, not whether they ate a plant-based diet, and its results do not show plant eating to be harmful. One study on one food group cannot cancel the wider evidence. And the case for veganism has never depended on it being healthier, only on it being adequate.
Much of the case for plant-based health rests on observational studies, and people who eat more plants also tend to be wealthier, slimmer, less likely to smoke and more health-conscious. When a large, well-run cohort like the UK Biobank adjusts carefully for those factors and the apparent benefit largely vanishes, that should make us more sceptical of the headline claims. Healthy-user bias is real, and it may account for much of what gets credited to the plants.
The study does not show what it is being used to show. The UK Biobank analysis looked at how many tablespoons of cooked and raw vegetables people reported eating, and whether that tracked heart disease. It did not compare vegans or vegetarians with meat eaters. Most participants ate meat, and vegetable intake varied within a fairly narrow range. Finding that an extra spoonful of vegetables adds little once income and lifestyle are accounted for is a long way from finding that plant-based diets have no benefit.
The point about confounding is fair, and it cuts in every direction. People who eat more vegetables also tend to smoke less, exercise more and earn more. Good studies adjust for that, and the authors were right to. But adjustment has limits too. If you control for things like weight or blood pressure, you can remove part of the very effect you are trying to measure. The authors themselves did not claim that vegetables do nothing, and the raw-vegetable association was weakened rather than shown to be harmful.
The question a plant-based diet raises is also different. It is mostly about what replaces meat. Here the evidence is broader than one cohort: trials and pooled analyses on swapping red and processed meat for legumes, whole grains and nuts, and on LDL cholesterol, point the same way. The speaker in this debate conceded that the literature is mixed and that no single study settles it. That concession applies to this study as well.
Finally, the ethical case does not rest on veganism being a health intervention. It needs only to be adequate, and the major dietetic associations say a well-planned vegan diet is. If the health benefits turned out smaller than advocates claim, the question of whether the animals needed to be killed would be exactly where it was.
Drawing a general conclusion from a sample too small or too unrepresentative to support it.
How it applies here: A single cohort study measuring vegetable intake among mostly meat-eating participants is taken as grounds to dismiss the health benefits of plant-based diets as a whole. That pattern is defined by what replaces meat, which this study did not test.