The efficiency of searches for randomized controlled trials in the International Journal of Eating Disorders: a comparison of handsearching, EMBASE and PsycLIT

Abstract

Our aims were: (i) to identify all reports of randomized controlled trials (RCTs) in the International Journal of Eating Disorders, (ii) to assess the number of these that were available in EMBASE and PsycLIT, and (iii) to investigate the precision of search strategies, designed for 100% sensitivity, for each of the above databases. The International Journal of Eating Disorders was handsearched for reports of RCTs. The 40 reports identified were then sought in EMBASE and PsycLIT. These records were assessed to identify methodological words/phrases which could be used to develop search strategies capable of identifying all reports of RCTs in the Journal. The precision of these searches was calculated. EMBASE contained 35 (88%) trials, and PsycLIT only 27 (68%). The 100% sensitive EMBASE strategy, designed to find all of the 35 trials, using methodological indexing text terms, identified 391 reports, and the 100% sensitive PsycLIT strategy identified 289 reports. The precision, therefore, of both searches was poor. Search strategies designed for high sensitivity resulted in very poor precision. A significant proportion of reports of RCTs were not indexed in PsycLIT.