Hill's Criteria is especially good at identifying causal relationships to problems

Claude AI - Sept 2026

Hill's criteria have a real track record, but mostly as a way to justify and defend a causal conclusion rather than to discover one. The history is more nuanced than "Hill's criteria led to breakthroughs."

Smoking came first, then the criteria. Doll and Hill's case-control study (1950) and British Doctors Study (from 1951) had largely established smoking as a cause of lung cancer before Hill wrote his famous 1965 lecture [Hill 1965, Proc R Soc Med]. The 1964 US Surgeon General's report had already used a similar, shorter list: consistency, strength, specificity, temporality, and coherence. Hill's nine "viewpoints" distilled the reasoning that won the smoking argument against heavy industry opposition, when an RCT of smoking was impossible. He also cautioned that none of the viewpoints was a checklist or proof, and that experimental evidence was the strongest of them.

Where Hill-style reasoning carried the day. The pattern is exposures that are harmful, where randomization is unethical: - Asbestos and mesothelioma, secondhand smoke, radon, and occupational carcinogens (IARC's classification process is essentially Hill-based) - Aspirin and Reye's syndrome, and DES and vaginal adenocarcinoma - Zika and microcephaly (2016), where CDC used a closely related teratogen framework to declare causation quickly - Epstein-Barr virus and multiple sclerosis (2022 military cohort), widely described as meeting Hill's criteria

Folic acid and neural tube defects is an interesting hybrid. Observational evidence was strong, but the 1991 MRC randomized trial is what settled it and drove fortification.

Where Hill-type observational evidence failed. These cases matter most for vitamin D, because they involve beneficial nutrient or hormone exposures: - Beta-carotene looked protective against lung cancer, but the ATBC and CARET trials showed harm in smokers. - Hormone replacement therapy looked cardioprotective, but WHI reversed that. - Vitamin E and cardiovascular disease showed the same pattern.

These observational associations scored well on strength, consistency, dose-response, and biological plausibility, yet were driven by healthy-user confounding. Retrospective reviews note that the criteria handle confounding and reverse causation poorly [Fedak et al. 2015].

Implications for vitamin D. Skeptics will point to the beta-carotene and HRT precedents, plus reverse causation (illness and inflammation lower 25(OH)D). Hill analyses carry more weight when paired with: - Mendelian randomization, which modern reviews treat as an update to Hill's "experiment" viewpoint - RCTs restricted to deficient participants or analyzed by achieved serum level, rather than intention-to-treat on replete populations - Strong dose-response and temporality data, especially from pre-diagnosis serum samples

The smoking precedent shows Hill's framework can win a causal argument against resistance. The nutrient precedents show why it rarely wins alone for a supplement.

Related in VitaminDwiki