What a systematic review is, and what you need before you start

A systematic review answers one clearly defined question by finding, screening, and summarizing the existing evidence using a method decided in advance and written down — so that someone else could repeat it and reach the same result. Littview manages that process, but it starts after some of the work is already done. This article covers what “systematic” means in practice and what you should have ready before you create your first project.

What makes a review systematic

Four things separate a systematic review from an ordinary literature review:

• The question comes first. It is defined before you look at any results, and it does not drift as you read.
• The method is fixed in advance. Where you search, what counts as eligible, how many people judge each paper, and how disagreements are settled are all decided before screening starts — so the papers you happen to find don’t reshape the rules you judge them by.
• The search aims to be exhaustive and repeatable. Several databases, a query you wrote down, and a result set you can export.
• Every decision is justified and reported. Papers are excluded for stated reasons, those reasons are counted, and the whole flow is reported — usually as a PRISMA diagram.

That last point is what Littview is built around: the criteria a reviewer selected, their notes, and the passages they highlighted in the PDF stay attached to the paper from screening through to the report, and your PRISMA counts and report drafts are generated from what your project actually did.

What you need before you start

1. A research question and the objective behind it. State it in one sentence, structured with whichever framework fits your field — PICO or PICOS (population, intervention, comparator, outcome, study design) for clinical questions, SPIDER for qualitative ones, or a plain scoped question elsewhere. In Littview this lives in the project Description, which asks you to describe the research question, scope, and objective of your review. It appears on the project card and at the top of the project, so it is what everyone joining the review reads first.

2. A protocol, written before you screen. Even a one-page document is enough: the question, the eligibility criteria, the databases and search strings, how many reviewers see each paper, how conflicts are resolved, and which variables you plan to extract. Health-related reviews are often registered on PROSPERO; other fields commonly use OSF. Everything below is essentially your protocol translated into Littview settings.

3. Inclusion and exclusion criteria, drafted. In Littview these are not decoration: reviewers justify every decision by selecting from exactly the criteria you enter, exclusion reasons are tallied per criterion for your PRISMA reporting, and the generated Methods draft quotes your wording. Have them settled before reviewers start, so decisions recorded early and late are made against the same rules.

4. A search you have already run — and exported. Littview does not search bibliographic databases for you. Run your search in PubMed, Scopus, Web of Science, IEEE Xplore, or whatever your field uses, then bring the results in as a file — BibTeX, RIS, PubMed (MEDLINE), or EndNote XML — or import directly from your Zotero library. Each upload becomes its own library and all libraries are screened together, so one library per database keeps the record clean. Copies of the same paper arriving from different databases are expected; Littview groups them for you at the deduplication stage.

Note

Keep a note of the query, the date you ran it, and the number of records returned for each database. Those are the numbers your PRISMA diagram and Methods section need for the identification stage, and no tool can reconstruct them after the fact.

5. A decision on who reviews, and how many opinions each paper gets. Reviewing alone is fully supported — the workflow is identical. If others review with you, decide how many independent reviews each paper should receive: one for single screening, two for classic double screening, or more. That number drives how Littview distributes papers and balances the workload, and it is also what the recommended agreement threshold is calculated from — the level of agreement at which a paper is decided automatically instead of being held for conflict resolution. A project has exactly three roles — Owner, Assistant, and Reviewer — and you can invite people at any time, including after screening has begun.

6. Access to the full texts. Screening on title and abstract works without PDFs, but reading the full paper before deciding, highlighting the passage that justifies a decision, and every part of data extraction need the PDF attached to the reference. Sort out your library access or document delivery early, so this doesn’t block the review halfway through.

What Littview does not do for you

• It does not run your searches. The search strategy and its execution stay yours; Littview begins at the exported results.
• It does not decide. AI can suggest a screening decision, recommend how to settle a conflict, or draft extraction answers with the evidence it used — always labeled, always with its reasoning, and switchable on or off per member. The decision stays yours.
• It does not write your review. The PRISMA diagram, evidence table, and Methods and Results drafts are generated from your project’s real state, but they are starting points you edit — not finished text.

Once those six pieces are in place, the next article maps the whole Littview workflow from import to report draft.