The Hierarchy of Evidence

The hierarchy of evidence ranks study types by their reliability. Systematic reviews are at the top, while expert opinions and case studies are at the bottom.

Science is not a democracy. One good study outweighs 1,000 bad ones.

The Symptom: "My Doctor Said..."

People often cite "a study" to prove a point.

"I read a study that said coffee causes cancer."
"My doctor said this supplement works."

But what kind of study? Was it on mice? Was it one person?

The Mechanism: The Pyramid

Imagine a pyramid. The higher you go, the more bias is removed.

1. Systematic Reviews & Meta-Analyses (The Peak)

This is the study of studies. It combines data from many RCTs to find the average truth. It filters out the noise of individual papers. (See Meta-Analysis).

2. Randomized Controlled Trials (RCTs)

The Gold Standard for proving causation. You test a drug against a placebo in humans. (See Randomized Controlled Trials).

3. Cohort Studies

You watch a group of people over time (e.g., smokers vs. non-smokers). Good for finding links, but prone to Confounding Variables.

4. Case-Control Studies

You compare sick people to healthy people and ask about their past. Prone to Recall Bias.

5. Case Reports / Series

"We gave the drug to Bob and he got better." This is an anecdote. It proves nothing. Bob might have gotten better anyway.

6. Animal / Lab Studies

"It killed cancer in a petri dish." Humans are not petri dishes. 90% of drugs that work in mice fail in humans.

7. Expert Opinion

The lowest form of evidence. Even Nobel Prize winners can be wrong.

The Prescription: Climb the Pyramid

When you see a claim, ask where it sits on the pyramid.

If it is a mouse study: Ignore it. It is interesting for scientists, not for patients.
If it is an RCT: Pay attention.
If it is a Meta-Analysis: Trust it (mostly).

At PaperScores, we weight the Methodology score based on this hierarchy. An RCT starts with a higher baseline score than a case report. We do not let anecdotes masquerade as data.

The Hierarchy of Evidence: Not All Papers Are Equal