The number of people who entered a study may differ from the number with a recorded outcome. That difference is worth following through the report. Missing results can affect interpretation, particularly when their absence is related to what happened to participants.
The Cochrane Handbook’s discussion of missing outcome data explains why the amount, reasons, and potential relationship to outcomes require attention. A label such as intention-to-treat does not, by itself, remove every problem caused by missing data.
Follow the numbers through the study
In an invented example, one hundred participants enter each group, but outcome information is available for ninety-eight in one group and eighty in the other. The difference raises questions; it does not prove which result is correct or that anyone acted improperly.
Look for a participant flow diagram and explanations of loss to follow-up. Did people withdraw, miss an assessment, or have data excluded for a stated reason? Were reasons and amounts different between groups?
Then read how the analysis handled the missing information and whether the authors examined the effect of alternative assumptions. Do not fill missing outcomes with your own guesses.
A useful summary might say that the study reports a result but has substantial missing outcome data whose possible effect needs consideration. That communicates a specific limitation more clearly than dismissing the entire study because the final sample is smaller.
Keep absence separate from a negative result
In the illustrative groups above, “no recorded outcome” must not silently become “no symptoms” or “no benefit.” A missing assessment supplies neither answer. When making a reading worksheet, use separate rows for observed outcomes and unavailable outcomes, then copy the study’s stated handling method into a third row.
If the report is unclear, your summary can say exactly that: the number of missing observations is stated, but the available explanation does not resolve why the groups differ. You can identify a limitation without inventing the unobserved results or recalculating an analysis that requires information you do not have.

