How conclusive are Cochrane neonatal reviews?

Article type
Year
Authors
Lai NM1, Ong MJ2, Chaiyakunapruk N3, Ovelman C4
1Taylor's University School of Medicine and Cochrane Malaysia, Malaysia
2Cochrane Malaysia, Malaysia
3Monash University School of Pharmacy, Malaysia
4Cochrane Neonatal Review Group, USA
Abstract
Background: The conclusiveness of a systematic review influences the strength of practice recommendations. A study in 2006 showed that two-thirds of Cochrane neonatal reviews were conclusive. Ten years later, with many new and updated reviews, it is unclear whether the finding has changed.

Objectives: We determined the proportion of Cochrane neonatal reviews that were conclusive, and assessed whether the size of the evidence gathered affected the conclusiveness of the review.

Methods: We analysed published Cochrane neonatal reviews up to issue 12, 2015, focusing on the main intervention-outcome combination. Two authors independently evaluated the abstract conclusions and selected one out of five possible options: clearly beneficial, clearly non-beneficial (conclusive), perhaps beneficial, perhaps non-beneficial and unclear either way (inconclusive).
We performed logistic regression to examine the association between the number of included studies and cumulative sample size and the likelihood of a review being conclusive (SPSS 22, Chicago, IL, USA).

Results: Overall, 319 reviews were assessed. Excluding 43 empty reviews, 103 reviews (37.3%) were conclusive (beneficial: 63 (22.8%), non-beneficial (40 (14.5%)) and 173 (62.7%) were inconclusive (perhaps beneficial: 80 (29.0%), perhaps non-beneficial: 5 (1.8%), unclear either way: 88 (31.9%)). Reviews with more studies were more likely to be conclusive (OR 1.07, 95% CI 1.03 to 1.12 for each added study, P 0.001), but no independent association was observed between cumulative sample size and conclusiveness of a review (P 0.30). Among reviews published pre-2000, 70% were conclusive, compared to 34.8% post-2000.

Conclusions: The majority of Cochrane neonatal reviews are inconclusive. The major decline in the proportion of conclusive reviews post-2000 was not clearly attributed to the size of evidence. Our findings suggest that in Neonatology, incremental benefits of newer interventions over older ones may no longer be clear-cut. Clinical decisions may increasingly depend on the clinical context, individual perception of benefit and harm, and values and preferences of those involved in the care of neonates.