Pooling five randomized trials, one Mendelian randomization study and thirteen conventional multivariable regression studies, a new quantitative triangulation framework finds no causal effect of beta-carotene on coronary heart disease. Trials gave RR=1.00 (95% CI 0.98-1.01) and MR gave RR=1.02 (0.98-1.07), while observational studies suggested a protective RR=0.83 (0.76-0.91). After the authors adjusted each design for estimated direction and magnitude of bias and for relevance to a shared target question, the combined estimate was RR=0.99 (0.82-1.21). Cardiovascular disease results were similar.
The beta-carotene case is a well-chosen test. Observational data once fueled enthusiasm for antioxidant supplements, until large trials found no cardiovascular benefit and even harm for lung cancer in smokers. This framework reproduces that verdict formally rather than by narrative judgment, and the widened confidence interval honestly reflects the uncertainty that bias adjustment introduces. For readers, the practical message is that dietary carotenoid associations in cohort studies likely reflect healthy-user confounding, not a supplement effect; beta-carotene pills are not a heart-protection strategy. The main limitation is that bias adjustments rest on expert-specified assumptions, so different priors could shift results, and the method has been demonstrated on one exposure where the answer was already known. This is a preprint awaiting peer review, and conclusions may change. As methodology, it is a meaningful incremental advance for evidence synthesis; as nutrition news, it is confirmatory.