Age-Specific Search Strategies for Medline

BACKGROUND: Many clinicians and researchers are interested in patients of a specific age (childhood, geriatrics, and so on). Searching for age-specific publications in large bibliographic databases such as Medline is problematic because of inconsistencies in indexing, overlapping age categories, and the spread of the relevant literature over many journals. To our knowledge, no empirically tested age-specific search strategies exist for Medline. OBJECTIVE: We sought to determine the retrieval characteristics of age-specific terms in Medline for identifying studies relevant for five clinical specialties: adult medicine, geriatric medicine, pediatric medicine, neonatal medicine, and obstetrics. METHODS: We compared age-specific search terms and phrases for the retrieval of citations in Medline with a manual hand search of the literature for 161 core health care journals. Six experienced research assistants who were trained and intensively calibrated read all issues of 161 journals for the publishing year 2000. In addition to classifying all articles for purpose and quality, study participants' ages were also recorded. Outcome measures were sensitivity, specificity, precision, and accuracy of single and combination search terms. RESULTS: When maximizing sensitivity, the best sensitivity and specificity achieved with combination terms were 98% and 81.2%, respectively, for pediatric medicine, 96.4% and 55.9% for geriatric medicine, 95.3% and 83.6% for neonatal medicine, 94.9% and 64.5% for adult medicine, and 82% and 97.1% for obstetrics. When specificity was maximized, all disciplines had an expected decrease in sensitivity and an increase in precision. Highest values for optimizing best sensitivity and specificity were achieved in neonatal medicine, 92.5% and 92.6%, respectively. CONCLUSION: Selected single terms and combinations of MeSH terms and textwords can reliably retrieve age-specific studies cited in Medline.