A review asking when an intervention works needs an explicit population, intervention, comparison, and outcome. Define eligible studies for each outcome. Record where and when you searched, the exact query, and the deduplication rule so the search can be refreshed.
Build a study matrix
For each study, capture the sample, measurement, result, uncertainty, limitations, and a pointer to the relevant passage. Organize rows by study and group them by disputed claims when drafting. When two findings disagree, compare measurement timing, population, and controls before explaining the difference.
Outline paragraphs around claims
Give each paragraph one claim and list supporting studies, counterexamples, and scope. Track unanswered questions separately. Check every numerical or directional claim against the original study and attach the citation to the sentence it supports.
Preserve a review trail
Use Notion Research Documentation to organize the search log, matrix, and open questions. Ask a reviewer to pick any claim and trace it to the study and the cited passage. On an update, check whether newer eligible evidence changes an existing claim.