• J González de Dios.
• Departamento de Pediatría, Hospital Universitario San Juan, Universidad Miguel Hernández, Prof. Manual Sala 6, 3.o A. 03003 Alicante, Spain. gonzalez_jav@gva.es
• An Pediatr (Barc). 2004 May 1; 60 (5): 417-27.

IntroductionThe Cochrane Neonatal Collaborative Review Group (CNRG) is one of the most important collaborative review groups registered in the Cochrane Collaboration (CC) and provides a growing and readily accessible source of information to help neonatologists base their care on detailed, critical, and up-to-date reviews of the best available evidence. The objective of this article was to perform a bibliometric analysis (quantitative and qualitative) of the systematic reviews (SR) published in the Neonatal Cochrane Collaboration (NCC) and to identify the subject areas studied by the CNRG in which decision-making in clinical practice is based on the best available evidence.MethodsWe performed a bibliometric analysis of the Cochrane Database of Systematic Reviews in Neonatology, Issue 2, 2003 (n = 147 SR). The following variables were registered in each SR: subject area of the study, authors (number and country), dates (last review and update), characteristics of the clinical trials included (number and type), characteristics of included newborns (number and gestational age), reviewers' conclusions and potential conflicts of interest.ResultsThe main subject areas of studies in the NCC were respiratory diseases (69 SR) and gastroenterology-nutrition (23 SR). Reviews were relatively scarce on some subjects that are very important in clinical practice in neonatology, such as cardiovascular diseases (10 SR), neurology (10 SR) and infectious diseases (6 SR). All the SR dealt with interventions for the treatment or prevention of diseases of the newborn infant (mainly preterm) and we found no reviews on diagnostic tests. Eighty-two per cent of the SR were signed by 2 or 3 authors, and we found four Bradford's zones of productivity. Ninety-two percent of the authors of SR were from four countries (Australia, United States, United Kingdom and Canada). Seventy-nine percent of all SR were published between 2000 and 2002, and 48 % have been updated. The median number of clinical trials per SR was 5 (range 0-30) and the median number of newborns included per SR was 559 (range 0-5448). Sixty percent of SR also included quasi-randomized clinical trials, and 67 % of SR included only preterm infants. In 54 % of the SR, the reviewers' conclusion was sufficient to infer probable effects (favorable or unfavorable) in clinical practice, and potential conflicts of interest were registered in 22 %.DiscussionAlthough CC reviews are based on specific guidelines, errors and biases occur, suggesting the need to "review the reviews". The bibliometric analysis of the NCC shows that the quality of published SR varies, and the main conclusion is that almost half of the reviewers' conclusions are insufficient to infer probable effects in clinical practice. There is a lack of scientific evidence on many therapies in neonatology and, at the moment, the SR published in the NCC enable evidence-based decision making mainly in the subject area of respiratory diseases and gastroenterology-nutrition. Challenges for the future include the need for bigger and better clinical trials in this field (with long-term outcomes among surviving children), the preparation of systematic reviews on topics not yet covered, and keeping an increasing number of reviews up-to-date. Every decision in neonatology should be based on a systematic appraisal of the best evidence available in the context of the prevailing values and resources at our disposal.

4 keywords...

hide…