Study Scores
Summarizes published research. Not medical advice. Read the Disclaimer.
Every entry on a Studies page gets a score from 1 to 5. The score describes how much weight the study's design can carry. It says nothing about whether the result was positive, and it is separate from the effect grades on Evidence Levels.
Scores are applied by a fixed script to the PubMed record (publication type, abstract, funding statement). Nobody adjusts them by hand. That keeps them consistent, and it also means they can be wrong when an abstract is vague. Report errors through How to Contribute.
What each page includes
Human randomized controlled trials, systematic reviews and meta-analyses indexed in PubMed for the topic, found with MeSH and title searches plus publication-type filters. Sauna and cold exposure also include prospective cohort studies, because trials are few. Study protocols, retracted papers, comments and letters are left out. Where a search returns more than about 300 records, the list keeps every review but only the trials that measured an effect the article covers. Each page states its search date and counts.
Starting point by design
Design | Starts at |
|---|---|
Meta-analysis | 4 |
Systematic review without pooled analysis | 3 |
Randomized controlled trial | 3 |
Prospective cohort | 2 |
Adjustments
Code | Applies to | Change | Rule |
|---|---|---|---|
+large | Reviews | +1 | 15 or more studies pooled, or 5,000 or more participants |
-few | Reviews | -1 | Fewer than 5 studies |
-nonRCT | Reviews | -1 | Abstract describes observational or animal data and no randomized trials |
+blind | Trials | +1 | Double-blind, placebo or sham controlled |
open | Trials | 0 | No blinding or placebo described |
+n200 | Trials | +1 | 200 or more participants |
-n<30 | Trials | -1 | Fewer than 30 participants |
-n? | Trials | -1 | Sample size not stated in the abstract |
-pilot | Trials | -1 | Described as pilot, feasibility or preliminary |
-acute | Trials | -1 | Single dose or session, with no follow-up in weeks or months |
+n10k | Cohorts | +1 | 10,000 or more participants |
-small | Cohorts | -1 | Under 100 participants, or size not stated |
-industry | Any | -1 | Abstract states funding from a company |
NOABS | Any | Score 1 | No abstract in PubMed, so it can't be assessed |
Final scores are capped between 1 and 5. The reason code next to each score lists the starting design and every adjustment, for example "RCT +blind +n200" (3 + 1 + 1 = 5).
Limits
Replication isn't scored per study. It shows up in how many trials and reviews a page lists for the same effect.
Sample size and study counts are read from the abstract text and can be off for complex designs.
Summaries restate each abstract's own conclusion, so they carry the authors' framing.
A high score means a strong design, not a correct or applicable result.