banner

Systematic literature reviews

Medextens systematic literature review systems aim at gathering evidence about the context of medical device use.

Information strategy

Strategy

Questions to be addressed and use of results

We help to identify the clinical and epidemiological context in which the medical device will be used, and translate them into benchmarks questions to be addressed in a systematic literature review.
Typical benchmarks are incidence or prevalence of the target condition, sensitivity and specificity of reference diagnostic systems and the relation between the two, differences or ratio in outcomes between two types treatments of the same condition.
Estimates are compared, and when applicable, stratified according to relevant subgroups.
Results may be used, for example to determine target population size, which influences budget impact, the effect achieved by medical technology in current use. Results may also serve as the the justification of a new clinical investigation or parameter input into a decision model.

Planning and conduct

Review protocol

The review protocol is developed prospectively. It fully describes:

data sources
  • The objectives and scope of the search. These lead to questions to be answered or hypotheses to be tested.
  • Source of evidence with the search engines, reference repositories to store, sort and retrieve documents (e.g., Reference Manager, EndNote, Zotero).
  • Screening methods to identify all relevant publications and filter out irrelevant publications in order to meet the scope and avoid publication bias.
  • Process to summarize and and assess the methodological quality of publications that have been positively screened, and to confirm the eligibility for inclusion in the analysis.
  • Endpoints to be extracted from eligible publication.
  • Quantitative methods to analyze extracted endpoints and draw conclusions.

Several guidelines and recommendations have been published to standardized systematic review methods in order to produce quality reviews.

Execution plan

It is influenced by the number of questions to be addressed, the type of evidence expected to be found, the amount of evidence reports to screen and to summarize.
That stage is the moment to quantify workload and difficulty, to select associate reviewer with relevant skills, to request from authors permissions to have access to additional information.

Forrest plot

Analysis

Analysis is quantitative when possible. When data extracted from publications are summary statistics, these should be preferably analyzed using the techniques of meta-analysis. These need to include robustness checks. Several statistical software provide meta-analytic modules.

When the review enables to extract observation-level data, then other statistical techniques may be used, the choice of which depends on the methods.

Qualitative analysis may be the only possibility when results from the different sources are not in a format that lends itself to statistical compilation.

Reporting

  • A formal systematic literature report is written according to the protocol.
  • Manuscripts for submission to peer-reviewed journals are written according to current guidelines.