Who Julie is for

Systematic review authors

You are running the review with fewer hands than the method assumes, and the second screener is the first thing to go. Julie assesses every record twice and brings you the ones the two assessments could not settle, so your methods section can state dual screening with human adjudication.

What stays yours: approving the criteria and search strategies, running the searches, and every call on a record the assessments disagreed about.

Guideline panels and their secretariat

You have to put something in front of a panel that can be voted on, and afterwards show that the vote was about that material. Julie prepares a dossier per question and a brief to circulate beforehand or, on the GRADE route, an Evidence-to-Decision framework with the certainty of the evidence already entered.

What you do: record what the panel decided, recommendation by recommendation: the final wording of anything it amended or, on GRADE, the direction, strength and rationale of each. The evidence work behind the recommendations waits for that record.

HTA and evidence synthesis units

You run several reviews at once, and the real risk is one that stalls where nobody can say what state it is in. Julie tells you which stage each review is in, what is missing, and which decision is waiting on a person.

What you get: no stage spends anything without an estimate you have seen and a ceiling you set, and every paid run is recorded under your approval.

Researchers with studies to pool

You already have the trials, from your own review or from someone else’s, and the question is whether and how they combine. Julie lays out what each study can contribute, computes the analysis you signed, and carries it into GRADE certainty and the Methods.

What stays yours: the analysis plan, which studies belong together, which are described rather than pooled, and the certainty judgements the evidence leaves open.

Methods editors and peer reviewers

You have to decide whether a submitted review's method is defensible. What you are reading is not prose written afterwards from memory: the methods report is produced from the run's own record, and where it cannot supply something the checklist asks for, it names that item instead of reading as complete.