In this post we’ll discuss how the CAPTIS™ PubMed Integration helps medical writers complete their literature reviews quicker.
Literature searches are a key part of clinical evaluations, usually involving countless hours of work. Medical writers and clinical experts compile scientific publications as a part of their literature review for various reasons ranging from gathering evidence for claims made on the device, documenting the state of the art and most importantly, to provide proof of the reliability, effectiveness, and clinical value of the device under evaluation.
Any gaps that clinical data cannot answer must also be covered in the literature review. In order to uncover relevant literature and evaluate it, a librarian or a medical writer with knowledge of choosing the right search keywords and experience using scientific research databases like PubMed or Embase must be involved.
For regulatory documents created to satisfy requirements of the EU MDR or IVDR, one of the key components that notified bodies are interested in is your literature search strategy.
A search strategy is a plan to find relevant literature, including details of the databases searched, and the keywords and search filters used. There must also be a systematic and unbiased way of analyzing said literature which essentially translates into the inclusion and exclusion criteria used in the title and abstract screening, and the appraisal method used in the full text evaluation.
While one can use focused searches, i.e., input keywords in a database and pick articles that seem relevant, this is not a systematic approach and may result in an observation from the Notified Body citing risk of bias. Cherry-picking evidence is a crime in the regulatory world and re-doing your report’s literature section within your notified body response time will take time and effort (and blood, sweat and tears).
To have a more focused and smaller literature dataset to review and to get to your relevant literature quicker, writers should use search strings instead. Databases like PubMed allow combinations of multiple keywords using Boolean operators such as AND, OR, and NOT which can greatly reduce the number of hits and give you more refined and productive results.
Apart from search strings, other parameters can come into play when building a search strategy, such as the chosen date range for the clinical or performance evaluation. A defined start and end date range helps you get to the latest research and limit the number of overall hits. You can also use filters for article types such as meta-analysis, systematic review, clinical trial, review etc. depending on the type of publications you’re interested in.
While there are no limitations on the filters used when searching PubMed, ensure that your search strategy is not limiting and returns a sufficient number of articles to review. Needless to say, a “sufficient number” is subjective and will depend upon various factors such as the device’s time in the market and it being a popular topic of research. However, do not forget that the goal of a literature evaluation is to provide an unbiased picture of the device’s safety and performance in the market, and not to have a comfortably small literature dataset to review.
Another key feature of a literature search strategy is its reproducibility. One should be able to arrive at the same literature dataset even if the same search parameters are run on a different date. Hence, the strategy must be documented accurately.