An observational study measures an exposure (taking a supplement, eating a diet, having a condition) and an outcome without randomly assigning who gets the exposure. The two most common forms in supplementation research are cohort studies (follow a group over time and compare outcomes) and case-control studies (compare people with an outcome to those without and look back at exposures).
Why it matters
Many long-horizon questions in supplementation — does high omega-3 intake predict fewer cardiovascular events over 20 years, does vitamin D status predict all-cause mortality — cannot ethically or practically be answered with a randomized controlled trial. Observational evidence is the only evidence available for these questions, and it is often what podcast hosts cite.
What observational evidence can and cannot tell you
Observational studies can show:
- The direction and size of an association between exposure and outcome.
- Whether an effect persists after statistically adjusting for measured confounders (age, sex, BMI, smoking, income).
- Whether effects are consistent across large populations and long durations.
Observational studies cannot cleanly show:
- That the exposure caused the outcome. Unmeasured confounders (health consciousness, genetics, income, sleep, stress) travel with supplement use. People who take a daily multivitamin also tend to exercise more, sleep better, and drink less.
- Whether reverse causation is at play. Low vitamin D predicts many bad outcomes, but many of those outcomes themselves lower vitamin D.
This is why Unfair treats observational evidence as directional, not diagnostic.
Features that strengthen the read
Observational studies exist on a quality spectrum. Features that push a study toward reliable:
- Large sample — tens of thousands of participants reduces random error.
- Prospective design — exposure is measured before the outcome occurs, not recalled after.
- Mendelian randomization — uses genetic variants as instrumental variables; approximates randomization for lifetime-average exposures.
- Dose-response relationship — higher exposure predicts proportionally more (or less) outcome.
- Consistency across cohorts — the association holds across countries, decades, and populations.
- Biological plausibility — there is a known mechanism of action consistent with the observed effect.
Reading headlines
Most "study shows supplement X reduces risk of Y" headlines are observational. The useful questions are: was this prospective or retrospective, how large was the cohort, what confounders were adjusted for, and does the effect persist in RCTs where they exist. If RCTs contradict the observational signal, weight the RCT evidence more heavily — this is a common supplement stack mistake to avoid.
How this appears in Unfair
Compounds with observational support but no RCT evidence sit at the "preliminary" or "mechanistic" tier in evidence tier. They can still be part of a stack, but the recommendation confidence is lower and the rationale snippet makes the evidence level explicit.
Clinical safety note
Observational evidence cannot rule out harm in the way an RCT can, particularly for rare events and drug interactions. Treat any new compound in your stack with the same caution regardless of whether its support is observational or experimental.