There are no existing search filters to identify natural history studies or data. We designed one for a rare disease topic.

OBJECTIVES: i) to report the design and structure of a search filter to identify natural history studies/data; and ii) to reflect on the development of the filter

METHODS: We designed a multi-stranded search filter based on scoping searches and an informal review of studies and systematic reviews of natural history data (CRD42024536369). The filter took the following form:

Strand 1: terms for natural history or longitudinal data; OR

Strand 2: terms for study designs that include or would report data captured over time; OR

Strand 3: terms for data collection over time.

We combined the strands with terms for the condition and a date limit of 2014–present. Filter performance was compared to known eligible marker papers and drawn from the authors’ experience undertaking this type of review.

RESULTS: We identified 1,447 studies after deduplication based on a search of MEDLINE and Embase. We included 107 studies (7.4%) during the title and abstract screen. Additionally, eight reports were identified using backwards snowballing. No other eligible studies were identified from the suite of reviews we are undertaking in parallel (e.g. quality of life or economic evaluations) or following expert feedback. Irrelevant reports retrieved by the filter included animal and case studies.

CONCLUSIONS: The filter successfully retrieved relevant natural history reports and will be reported for future users via reporting using a Search Narrative. Going forward, further development can be undertaken to enhance the filter’s retrieval, for instance using a NOT cluster to exclude animal studies, study types unlikely to report data of interest, and certain types of syntheses.