How to develop rapid reviews of diagnostic tests according to experts: A qualitative exploration of researcher views

Rapid reviews (RRs) have been used to provide timely evidence for policymakers, health providers, and the public in several healthcare scenarios, most recently during the coronavirus disease 2019 pandemic. Despite the essential role of diagnosis in clinical management, data about how to perform RRs of diagnostic tests are scarce. We aimed to explore the views and perceptions of experts in evidence synthesis and diagnostic evidence about the value of methods used to accelerate the review process.

We performed semistructured interviews with a purposive sample of experts in evidence synthesis and diagnostic evidence. We carried out the interviews in English between July and December 2021. Initial reading and coding of the transcripts were performed using NVIVO qualitative data analysis software.

Of a total of 23 invited experts, 16 (70%) responded. We interviewed all 16 participants representing key roles in evidence synthesis. We identified 14 recurring themes including the review question, characteristics of the review team, and use of automation, as the topics with the highest number of quotes. Some participants considered several methodological "shortcuts" to be ineffective or risky, such as automating quality appraisal, using only one reviewer for diagnostic data extraction and only performing descriptive analysis. The introduction of limits might depend on whether the test being assessed is a new test, the availability of alternative tests, the needs of providers and patients, and the availability of high-quality systematic reviews.

Our findings suggest that organizational strategies (e.g., defining the review question, availability of a highly experienced team) may have a role in conducting RRs of diagnostic tests. Several methodological shortcuts were considered inadequate for accelerating the review process, though they need to be assessed in well-designed studies. Improved reporting of RRs would support evidence-based decision-making and help users of RRs understand their limitations.