Finding the Evidence: Identifying nurse staffing research in PubMed

Tuesday, 14 July 2009

Michael Simon, PhD, RN
Susan F. Klaus, PhD, RN
Nancy Dunton, PhD
National Database of Nursing Quality Indicators, University of Kansas School of Nursing, Kansas City, KS

Learning Objective 1: The learner will be able to search for nurse staffing research in PubMed.

Learning Objective 2: The learner will be able to discuss sensitivity and precision as indicators of search strategy performance.

Purpose: Identification of nurse staffing research literature is a cumbersome task. Search filters focused on 1) sensitivity 2) precision or 3) balancing sensitivity and precision support the work of researchers conducting systematic reviews (SR).

Methods: Using three comprehensive SR of nurse staffing research a pool of articles was identified in PubMed. This pool was used to extract candidate free text- and MeSH-terms, which were employed to develop three search strategies aimed at highest sensitivity, highest precision and balanced between sensitivity and precision. The newly developed search strategies were validated against a) the pool of relevant articles derived from the reviews and b) a test set identified from a hand search of three relevant journals (Med Care, Health Serv Res, J Nurs Adm). The hand search was conducted by two nurses with doctoral degrees and based on pre-defined eligibility criteria. Additionally all new developed strategies were compared to PubMed’s health services research queries.
Results: 78 articles from systematic reviews (test set 1) and 17 articles from a hand search (test set 2) were identified in PubMed. The highly sensitive search strategy (HSSS) identified 99%, highly precise search strategy (HPSS) 51%, and the balanced strategy (BSS) 79% of test set 1. For test set 2, HSSS identified 100%, HPSS 47% and the BSS 77% of all articles. High consistency between both test sets confirms the validity of the approach.

Conclusion: Like other health service research, nurse staffing studies are difficult to identify in PubMed. Depending on the purpose of the search, researchers can chose between high sensitivity (and high load of literature) or high precision (overlooking a noticeable amount of papers). A more homogenous terminology (e.g. through the term “nurse staffing”) could improve precision of future searches.