OBJECTIVES: Different approaches can be taken to conduct a TLR. This research investigated the effectiveness and value of different targeted search methods.

METHODS: We took a case study of identifying the global epidemiology of acromegaly based on observational studies published in English from inception till September 2022 in Embase^® and PubMed^®. The structured TLR searches used pre-defined keywords, exploded sub-headings, boolean/proximity operators, etc., while the focused TLR developed pragmatic searches utilizing the same keywords but with methodological shortcuts, e.g., search restricted to titles-abstracts. The pearl growing method included the identification of primary pearl/citation, then backward and forward citation mining in PubMed. The ‘similar articles’ section in PubMed was also explored.

RESULTS: The structured and focused search identified 1,293 and 681 citations, respectively. In comparison, 574 citations were identified from the primary pearl (512: similar articles; 39: cited by; 23: bibliography). Post analysis, the structured search identified 40 relevant studies, while 30 were identified from the pearl growing approach and 21 from focused TLR. The ten additional studies retrieved from structured TLR in comparison to pearl growing were potentially due to the word-weighted algorithm PubMed utilizes to identify similar articles (e.g., geography: US of primary pearl vs. others; disease indexing as acromegaly vs. pituitary adenoma) and the same publication period as that of the primary pearl hindering their linking to cited by and bibliography sections. Also, these additional studies assisted in identifying relevant data from five new geographies. Lastly, the time taken to conduct structured TLR was higher (9 days of screening at a rate of 400 citations/day) compared with focused TLR (5 days) and the pearl growing method (3 days).

CONCLUSIONS: This case study found the structured TLR approach most comprehensive. However, we recommend the pearl growing method for early-stage or rapid evidence synthesis as it is the most efficient.