A research paper may report several different counts: the number planned, the number enrolled, the number completing follow-up and the number included in an analysis. Before using a figure to describe a study’s size, check which of those it represents.
Follow the numbers
Look for a participant flow diagram or a clear account of exclusions and withdrawals. Compare the total with the numbers shown for each group and outcome. Differences can have legitimate explanations, but those explanations should remain in your reading record.
Keep the denominator with the percentage
A percentage is incomplete without knowing what it is a percentage of. Record the population or group alongside it. When different outcomes use different analysis sets, avoid presenting them as if every participant contributed to every measure.
Size is one part of interpretation
A larger study is not automatically a better match for a particular question. Design, measurement, missing information and the population studied also matter. Read the authors’ discussion of uncertainty and limitations instead of using one enrolment number as a quality score.
Avoid designing a study from a summary
A reader’s checklist cannot provide a suitable sample-size calculation for an experiment. Such planning depends on the actual question, analysis, assumptions and specialist input. Keep your task narrower: identify the counts the report used and where its conclusions may be limited.
Add the analysis population and source location to your reading notes. This guide does not calculate statistical power or offer clinical conclusions.
Sources
NCCIH: size of the study. NC3Rs: group and sample size.
Top comments (0)