The hierarchy of evidence is a ranking of study designs by how well each one guards against bias and chance. Systematic reviews and meta-analyses usually sit at the top, followed by randomized controlled trials, then cohort and case-control studies, case reports and expert opinion, with laboratory and animal research placed at or near the base because it does not measure effects in people.
The ranking is often drawn as a pyramid. Designs near the top are generally better at showing whether one thing causes another, while designs near the bottom are better at raising questions than answering them.
What are the levels of the hierarchy of evidence?
Versions differ in detail, but the usual ordering, from strongest to weakest for questions about cause and effect, looks like this:
- Systematic reviews and meta-analyses: structured summaries of all the studies on a question, sometimes with the results pooled statistically.
- Randomized controlled trials: studies that assign people to a treatment or a comparison group by chance.
- Cohort studies: studies that follow groups of people with and without an exposure over time.
- Case-control studies: studies that start with people who have a condition and compare their past exposures with those of people who do not.
- Case reports and case series: detailed descriptions of individual patients.
- Expert opinion: the judgment of experienced clinicians or researchers, without new data.
- Laboratory and animal studies: research on cells, tissues or animals that explores how something works.
Some versions put laboratory and animal work beside the pyramid instead of beneath it, because it answers a different kind of question.
Why are systematic reviews and meta-analyses at the top?
A single study can mislead, by chance or because of how it was run. A systematic review searches for all the relevant studies using a method fixed in advance, judges their quality, and summarizes what they show together. A meta-analysis goes a step further and combines the numerical results into one estimate.
Their strength depends on what goes in. A review of small, poorly designed studies inherits their weaknesses. The difference between the two is explained in systematic review versus meta-analysis.
Why do randomized controlled trials rank so high?
Random assignment means the groups being compared should be alike, on average, in everything except the treatment, including factors no one thought to measure. If the groups end up with different outcomes, the treatment is the most likely explanation. Blinding, where participants or researchers do not know who received which treatment, reduces bias further.
Trials have limits. They can be small, short, or run in carefully selected people who differ from those seen in everyday care. Read more in what a randomized controlled trial is, and why it matters.
Where do cohort and case-control studies fit?
These are observational studies: researchers watch what happens instead of assigning a treatment. They can include large groups of people and follow them for years, and they are sometimes the only ethical option. No one would randomly assign people to smoke, so the evidence that smoking causes lung cancer rests heavily on observational research.
The main weakness is confounding. People with a given exposure may differ in other ways that also affect the outcome, so an association on its own does not prove cause. See what observational studies can and cannot tell you.
What about case reports and expert opinion?
A case report describes what happened to one patient, and a case series describes a small group. They are useful for spotting something new, such as an unexpected side effect, and they can prompt larger studies. They cannot show how often something happens or prove that a treatment caused the outcome.
Expert opinion draws on experience and can be the only guidance available for rare situations. It sits low because experience is shaped by memorable cases and personal judgment, and experts can disagree with one another.
Where do laboratory and animal studies belong?
Studies in cells and animals explain how something might work. They are essential for developing treatments, but a result in a dish or a mouse does not always carry over to people. Headlines that report a lab finding as a health benefit skip that gap, as explained in why animal and lab studies do not prove effects in people.
What are the limits of the hierarchy of evidence?
The pyramid is a starting point for judging evidence, and it breaks down when treated as a strict rule.
- Design is not the same as quality. A well-run cohort study can be more trustworthy than a small, flawed trial.
- The best design depends on the question. Rare side effects and long-term harms are often studied with observational designs, because trials can be too small or too short to see them.
- Directness matters. A strong study in a different population, or one measuring a different outcome, may say little about your question.
- Consistency matters. Several studies pointing the same way carry more weight than one striking result.
- Reviews can be out of date. A review reflects the studies available when its search was run.
How can you use the hierarchy when reading a health claim?
Start by asking what kind of study stands behind the claim. A news story about a mouse study, a single case report or one doctor's view is a reason to look for stronger evidence before drawing conclusions. A claim backed by a systematic review of trials deserves more confidence, though you should still check that the review asked the question you care about.
The same habit helps with AI-generated health answers, which can cite a weak study for a strong claim or describe an early finding as settled. If you are a patient, use the hierarchy to ask better questions, and talk through what any study means for you with your clinician.
Frequently asked questions
What is at the top of the hierarchy of evidence?
Systematic reviews and meta-analyses usually sit at the top, because they bring together the results of many studies on the same question using a method set in advance. Their conclusions are only as strong as the studies they include.
Is a randomized controlled trial always better than an observational study?
No. A well-designed observational study can be more reliable than a small or poorly run randomized controlled trial, and some questions, such as rare or long-term harms, are hard to study with trials at all.
Why is expert opinion near the bottom of the hierarchy of evidence?
Expert opinion relies on experience instead of a study designed to reduce bias. It can be valuable when evidence is thin, but memorable cases and personal judgment can shape it.
Is there one official hierarchy of evidence?
No. Different groups publish different versions, and they vary in details such as where laboratory studies belong. The broad ordering, from reviews and trials down to case reports and opinion, is similar across them.