Scoping Review: mapping a field before reviewing it

The choice is about the question, not about ambition or rigour.

If you find yourself wanting to conclude that an intervention works, you have outgrown the scoping review — the design does not license that claim, because it never weighed the evidence.

The method was formalised by Arksey and O’Malley in Scoping studies: towards a methodological framework, which named the five stages still in use: identify the question, identify relevant studies, select studies, chart the data, and collate, summarise and report.

Levac, Colquhoun and O’Brien tightened each stage — team-based screening, iterative criteria, an analytic rather than descriptive charting step — in Scoping studies: advancing the methodology.

Reporting has its own standard: PRISMA-ScR, published by Tricco and colleagues as PRISMA Extension for Scoping Reviews. Most journals that publish scoping reviews now expect the checklist.

A health-law team scoped how digital consent has been studied in biobank research. The question was deliberately broad; the PCC anchors were biobank participants (population), electronic or dynamic consent (concept), and any jurisdiction (context).

Searches across five databases and two regulatory repositories returned 1,916 records; 87 were charted. The charting table carried twelve fields, including consent model, study design, whether participants were actually recruited (or the study was conceptual), and which ethical concern each study foregrounded.

The map showed a field three-quarters conceptual: 64 of 87 papers proposed or debated consent models, while only 23 studied real participants — and of those, 19 measured comprehension, none measured long-term withdrawal behaviour. That single sentence — visible only because “empirical vs conceptual” was a charting dimension — became the review’s contribution and the justification for the empirical study the team ran next.

• Running a scoping review to smuggle in an effectiveness conclusion. The design cannot support it; reviewers know this.
• Charting only bibliographic trivia. Year and country tell a reader nothing they could not get from the reference list.
• Treating iterative criteria as no criteria. Refinement is documented evolution, not improvisation.
• Skipping PRISMA-ScR. The checklist exists; omitting it reads as not knowing it exists.
• Appraising quality half-heartedly. Either appraise with a named tool and use the result, or state clearly that appraisal was out of scope and why — the halfway version satisfies nobody.

A scoping review inherits the visibility biases of publication: what is mapped is what got published, indexed, and written in the searched languages. The map is of the literature, not of the practice.

Because quality is usually not appraised, the map weights a weak study and a strong one identically — fine for “what exists”, misleading the moment counts get read as evidence.

And breadth costs depth: 87 charted studies cannot each be read closely. The method trades interpretive depth for coverage, which is exactly why it pairs well with a follow-on systematic review of the densest cluster.

A systematic literature review is the follow-on when the map reveals a cluster mature enough to synthesise.

A rapid review serves a decision deadline rather than a mapping goal.

An integrative review admits diverse designs like a scoping review but synthesises their content rather than mapping their existence.

Charting is a codebook applied to PDFs — which is precisely the shape of work AI-assisted analysis does well. Define the charting dimensions as questions, run them across the included full texts, and verify from the linked quotes rather than re-reading 87 papers; Evidano’s content analysis works exactly this way, with every extracted answer traceable to its source passage.

The dimensions themselves, the iterative criteria decisions, and the reading of the map remain researcher judgement — a tool can fill the chart, not decide what the chart should measure.