[Durgical](https://durgical.com/en/) › [DurgiPedia](https://durgical.com/en/durgipedia/) › [Epidemiology](https://durgical.com/en/epidemiology/) › [Evidence-based medicine](https://durgical.com/en/epidemiology/areas/ebm/) › Critical appraisal of a paper

# Critical appraisal of a paper

- Concept

Appraising a study means asking three things: is it valid, what did it find and how precisely, and does it apply to my patient?

Updated Nov 5, 2026

## Key points

- Is it valid? Randomisation, allocation concealment, comparable groups at baseline, blinding, complete follow-up and intention to treat.
- What are the results? Relative and absolute effect (RR, ARR, NNT) and its 95 % CI.
- Does it apply to my patient? Similar population, important outcomes, benefits versus harms, patient values.
- Loss to follow-up above 20 %, or different between groups, threatens validity.
- Checklists (CONSORT, STROBE) show what is missing from the report; risk-of-bias tools judge validity.

## What it is

Critical appraisal judges a paper in three steps. Internal validity: whether design and conduct avoid bias (randomisation, allocation concealment, blinding, losses, intention-to-treat analysis). Results: the size of the effect, its precision (CI) and its translation into absolute numbers. Applicability (external validity): whether the patients, intervention and outcomes resemble one's own setting and whether benefit outweighs harms and costs.

## Worked example

- Data (made up): a trial assigns 1 200 adults aged 45 to 60 to a rehabilitation programme or usual care; outcome, readmission at 1 year.
- Validity: randomisation and concealed allocation; losses of 18 % in one arm and 6 % in the other: a warning, intention to treat is a must.
- Results: readmission 9 % with the programme and 12 % with usual care.
- RR = 9 / 12 = 0.75; relative reduction 25 %; ARR = 12 % − 9 % = 3 %.
- NNT = 1 / 0.03 = 33.3; rounded up to 34 patients for 1 year.
- Reported 95 % CI for the RR: 0.60 to 0.94; it excludes 1, a benefit with acceptable precision.
- Applicability: your patient is 82 and lives alone, outside the studied range: the effect may differ.
- Reading: a probably valid study with a warning about losses, a moderate benefit (NNT 34) and doubtful applicability to this patient.

## Three questions for appraising a trial

| Question | What to check | Warning sign |
| --- | --- | --- |
| Is it valid? | Randomisation, concealment, blinding, losses, intention to treat | Groups differ at baseline, unequal losses |
| What did it find? | RR, ARR, NNT, 95 % CI | Very wide CI, relative effects only |
| Does it apply here? | Population, intervention, outcome, benefit-harm balance | Very different patients, surrogate outcomes |

## Markers

- Validity
- Results
- Applicability
- ITT

## In practice

- With practice, appraising a trial takes a few minutes if you go straight to methods, table 1, the flow diagram and the absolute results.
- An abstract usually highlights the relative effect: computing the ARR and NNT changes the impression of benefit.

## Common pitfalls

- Reading only the abstract and the authors' conclusions.
- Confusing statistical significance with clinical importance.
- Applying the result to patients outside the inclusion criteria without thinking about it.

## Memory aid

Is it true, is it big, is it for me? Three questions that cover validity, results and applicability.

## Test yourself

Think of the answer, then tap the question to see it.

## What are the three basic questions of critical appraisal?

Are the results valid? What are they, and how precise? Can they be applied to my patient?

## A trial loses 18 % of one arm and 6 % of the other. Why does this matter?

Because unequal losses can break the comparability created by randomisation and bias the result.

## A treatment lowers an event from 12 % to 9 %. What is the NNT?

ARR = 3 %; NNT = 1 / 0.03 = 33.3, rounded up to 34.

[Add to my cards (3)](https://durgical.app/estudiar/?add=/en/epidemiology/critical-appraisal-of-a-paper/)

## Connections

## Related 6

- [Evidence-based medicine](https://durgical.com/en/epidemiology/evidence-based-medicine/)
- [Randomised controlled trial](https://durgical.com/en/epidemiology/randomised-controlled-trial/)
- [Intention-to-treat analysis](https://durgical.com/en/epidemiology/intention-to-treat-analysis/)
- [Number needed to treat](https://durgical.com/en/epidemiology/number-needed-to-treat/)
- [Reporting guidelines: CONSORT, STROBE, PRISMA, STARD](https://durgical.com/en/epidemiology/reporting-guidelines-consort-strobe-prisma-stard/)
- [Statistical vs clinical significance](https://durgical.com/en/epidemiology/statistical-vs-clinical-significance/)

[← All of epidemiology](https://durgical.com/en/epidemiology/)

Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.

Sources: [Chapter 15: Interpreting results and drawing conclusions](https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-15) (Cochrane, 2026); [CONSORT 2025 Statement: updated guideline for reporting randomised trials](https://www.equator-network.org/reporting-guidelines/consort/) (EQUATOR Network, 2025).

## **Spotted a mistake?** We are always improving for the community. Tell us what is wrong and our team will review it. Report a mistake

Representative model, for educational purposes only.

---
Source: https://durgical.com/en/epidemiology/critical-appraisal-of-a-paper/
License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Cite as: DurgiPedia (durgical.com), https://durgical.com/en/epidemiology/critical-appraisal-of-a-paper/