TRIAGE is the rubric behind every score in your digest — three independent domains instead of one misleading "importance" number, built specifically to answer the question a busy trainee actually has: is this worth reading fully, reading critically, or just skimming?
A flawless, perfectly-run trial of a surrogate endpoint and a landmark trial that changes bedside practice can both look like "5 out of 5" under a single importance score — even though only one of them should actually change what you do tomorrow. TRIAGE keeps three separate questions separate, because they really are different questions:
How much can you trust the result? Starts from the study design itself, then adjusts for the specific limitations that actually weaken a finding.
Does the paper's own conclusion actually match what it reported? Catches the gap between what a study found and what it claims to have found.
Even a rock-solid finding may not apply to your patients. This domain asks whether the outcome studied is one that actually changes bedside care.
Each domain is scored independently from 0 to 5, in half-point increments. They're never averaged or summed — a weak score in any one domain matters on its own, which is exactly what determines your reading-priority tier.
This is the rule that makes TRIAGE actually useful at a glance: your reading priority is set by whichever of the three domains is weakest, not by an average. A flawless trial measuring the wrong outcome still only earns a skim.
| Tier | Rule | Suggested Approach |
|---|---|---|
| Priority 1 | E, I, and A all ≥ 4 | Likely worth reading in full — probably practice-relevant. |
| Priority 2 | E, I, and A all ≥ 3 | Probably worth a critical read — check the methods before weighing it. |
| Priority 3 | Lowest domain = 2 | Likely a skim — useful context more than a reason to act. |
| Priority 4 | Any domain ≤ 1 | Probably awareness-only. |
| Excluded | I = 0 | Not shown in your digest at all — see below. |
These are starting points for triage, not verdicts — every suggestion here can be, and sometimes should be, overridden by your own read of the abstract or paper.
That last row is a real, active safety behavior, not a theoretical one: if PubMed's own metadata indicates an article has been retracted, Chiron drops it from your digest entirely rather than showing it with a low score.
A genuinely rigorous appraisal needs the full paper — funding disclosures, registration details, and a real look at study conduct all live in the full text, not the abstract. Chiron handles this honestly with two distinct modes, applied automatically depending on what's actually available.
The abstract-only edition. Every article is scored this way the moment it's fetched, using only the abstract and PubMed's own indexed metadata — nothing is guessed or inferred beyond what's actually stated.
When a free version of the article is available on PubMed Central, Chiron automatically fetches the complete paper and applies the full, original TRIAGE rubric in place of the abstract edition.
If a full-text fetch or scoring attempt is unsuccessful for any reason, the article falls back to TRIAGE-AE automatically — you'll still get a properly scored article, it just won't carry the star.
Per the rubric's own rule, Chiron never shows a tier by itself — the three domain scores are always displayed alongside it, right in the Chiron Analysis panel of every article:
E 4 / I 5 / A 5 — Priority 1 ✦ Full Text Verified
In the digest itself, this appears as three compact rows of dots — one per domain — with the tier and star shown just beneath, in the same spot every article's score has always lived.
TRIAGE is a tool Chiron built to help you triage a long list of articles, not a validated, peer-reviewed scoring system. It has not been tested against expert appraisal or outcomes research the way instruments like RoB 2 or GRADE have. Use it to help decide what to read first, never as a substitute for reading the article yourself or as the sole basis for a clinical or practice decision. A low score is a prompt to look closer if something seems off, not a verdict on the work — and a high score is not a guarantee of quality.
TRIAGE-AE is a genuine, careful attempt to score what an abstract can honestly support — but it is still, unavoidably, working with less information than a full paper provides. Chiron is explicit about this rather than papering over it: the star is the signal that a score reflects real verification against the complete text, not just a confident-sounding estimate.