Citation

Sladek RM, Tieman J. Health Info. Libr. J. 2008; 25(4): 295-301.

Copyright

(Copyright © 2008, Health Libraries Group, Publisher John Wiley and Sons)

DOI

10.1111/j.1471-1842.2008.00778.x

PMID

unavailable

Abstract

Background/objectives: Methodological decisions made during the research process can influence generalizability of findings to real world practice. The aims of this study were to explore the impact of decisions made in the development of a palliative care search filter and to consider the implications for implementation.

Methods: Three elements of the original study methodology were explored: (i) choice of OVID medline field delimiters; (ii) use of the general medical literature to evaluate the filter's performance; and (iii) use of the OVID interface. Sensitivity, specificity, accuracy and precision rates of variant search strategies were compared to consider each issue.

Results: The delimiter .af. outperformed the alternatives of .tw. or .mp. in OVID medline, improving sensitivity from 45.4 to 46.2%. Applying the filter in the specialist palliative literature resulted in 87.5% (692/791) of articles being retrieved using either .tw. or .mp., increasing to 100% (791/791) with the .af. delimiter. Finally, a PubMed version of the filter was successfully validated.

Conclusions: Reviewing three methodological decisions that preserved validity in an original study led to the improved utility of a search filter in practice. Generating high‐quality evidence is only part of evidence‐based practice: consideration of generalizability issues can inform further research and effective evidence implementation.