Background: In 1996, members of the Child Health Field recognized the need for a comprehensive database of published, child-relevant clinical trials that could be used by paediatricians to inform their clinical practice and by researchers to identify knowledge gaps. An initial register was developed in 1998; the register was updated in 2002 and then again in 2006. Objectives: To describe the development of the Child Health trials register, to highlight the benefits as well as the methodological and logistical challenges, and to provide solutions to address these challenges. Methods: We searched CENTRAL Issue 3, 2002 and Issue 1, 2006 using a paediatric filter developed and validated from 1996 to 1998. Screeners classified the records as ‘include’, ‘exclude’ or ‘unsure’ based on titles and abstracts. Records were eligible for inclusion if: 1) the study was a randomized or controlled clinical trial as defined by The Cochrane Collaboration; and 2) outcomes were reported for children 0-18 years of age. Double screening was applied to a sample of the records to calculate interrater agreement. Decision rules were developed and applied to records classified as ‘unsure’. We retrieved full text for studies that could not be classified based on these decision rules. Results: There were 26,638 records in the original register. Our searches of CENTRAL retrieved an additional 64,987 records. These were classified as: include (n = 44,484); exclude (n = 19,678); and unsure (n = 27,463) (numbers include duplicates). Once decision rules were applied, the numbers in the unsure category were reduced to 2805 records for which we retrieved full text. There are currently 30,151 unique records in the register. These have been tagged ‘SR-CHILD’ in CENTRAL. They are also available online at: http://www.chtr.med.ualberta.ca. Conclusions: The many advantages of developing a trials register must be weighed with the extensive resources required. Further work is planned to update the register, regularly incorporate handsearch results, revise and validate the search filter, and test the register for comprehensiveness. Our experience will inform future
updates as well as the development of other specialized registers.