# How to compare studies that seem to disagree

Check whether two papers answer the same question before weighing their conclusions.

Canonical URL: https://attendme.ai/guides/compare-conflicting-studies

Availability: live. Content review: editorial; Automated source and product-fact consistency checks; 2026-10-14.

## Define the common question

Write the population, intervention, comparator and outcome you want to compare.

## Separate reports from studies

A follow-up report or subgroup analysis is not a separate randomised comparator. Check whether reports describe the same trial.

## Compare like with like

Look at baseline risk, follow-up, endpoints and study setting before comparing effect estimates.

## Record unresolved differences

Use the comparison worksheet to show what differs and what remains uncertain. Preserve the source for each numerical result.

## A worked workflow example

Start with one paper and investigate why related studies might appear different. AttendMe Compare finds similar papers from a starting paper. Inspect the studies it actually uses before drawing a comparison.

### Your input

Start with the TEAM trial, PMID 36286256. Ask: 'Do these studies examine the same intervention, adult ICU population and outcome over the same follow-up?' Keep any specific second paper's identifier beside it in the comparison worksheet.

### In AttendMe

Open Library, choose Critical Care and find the saved starting paper in My Library. Select Compare. Comparative Analysis identifies the similar studies actually used. It does not guarantee a particular second paper. If that paper is absent, appraise it separately and compare both sources in the worksheet.

### What you inspect

Read Studies at a glance, How they fit together, Certainty across these studies and Gaps. Open each returned PMID or DOI. Check whether reports describe separate trials or the same trial. Compare endpoints and follow-up before weighing any difference. A similarity percentage is not an evidence-quality rating.

### Your reviewed result

Example comparison row: 'Study identity: verify each PMID. Endpoint: copy its source definition. Follow-up: record the period. Supporting passage: locate it. Comparability: explain the difference.' Leave unavailable details unresolved. The reviewed result states what differs and which evidence would resolve the question.

## Before you reuse the result

Use this final check to decide whether the work is ready to discuss, save or continue. A useful output preserves its evidence, its scope and the questions it could not answer.

- The comparator list identifies every paper the analysis used.

- Follow-up reports are not counted as independent trials.

- Each comparison preserves its endpoint, follow-up and source passage.

- The conclusion explains uncertainty without declaring a treatment winner.

## Sources

- [Illustrative paper identity: TEAM trial, PubMed 36286256](https://pubmed.ncbi.nlm.nih.gov/36286256/)

Continue: https://attendme.ai/auth/signup?intent=evidence-comparison