How Often Are Patient-Important Outcomes Represented in Neonatal Randomized Controlled Trials? An Analysis of Cochrane Neonatal Reviews

Evaluation of the Quality of Perinatal Trials: Making the GRADE

BACKGROUND

Assessing the quality of clinical research is a key evidence-based practice skill. Clinicians, guideline producers, policy makers, service commissioners, and families need to have a sense of the validity, applicability, and certainty of research evidence when determining how it should inform their decision-making and practice.

METHODS

We consider the various methodological and study design factors that contribute to the validity and applicability of clinical research findings. We describe the "Grading of Recommendations Assessment, Development and Evaluation" (GRADE) methodology and discuss how this approach is used to assess and report certainty of evidence and strength of recommendations.

RESULTS

The randomized controlled trial (RCT) is the gold standard method for assessing interventions because randomization balances prognostic characteristics between comparison groups. The GRADE approach considers evidence from RCTs as high quality, but acknowledges that the quality and level of certainty of trial evidence may be "downgraded" based on consideration of threats across 5 domains: risk of bias in included trials, inconsistency between trials in outcome estimates, indirectness of the evidence, imprecision of estimates, and likelihood of publication bias.

CONCLUSIONS

Structured critical appraisal using GRADE methods to assess risk of bias and other threats to the internal and external validity of RCTs and systematic reviews and meta-analyses of their data facilitates transparency and consistency in using evidence to inform policy and practice.

The Choice of Population and Outcomes in Neonatal Trials on Hyperbilirubinemia: Are They Relevant? An Analysis of Cochrane Neonatal Reviews

BACKGROUND

Neonates with jaundice are usually managed according to their serum bilirubin despite an unclear overall correlation between bilirubin levels and patient-important outcomes (PIOs) such as kernicterus spectrum disorder (KSD).

OBJECTIVES

We examined data from Cochrane Neonatal reviews to assess whether conditions that constituted KSD were included as key outcomes and how commonly they occurred in the population studied.

METHODS

We identified Cochrane reviews, published till November 2017 that evaluated interventions for neonatal jaundice (NNJ). We extracted the following information at the review and study levels: included population, outcomes assessed (in particular, whether PIOs such as KSD were listed as the primary outcomes), as well as their cumulative incidence in the reviews.

RESULTS

Out of 311 reviews, 11 evaluated interventions for NNJ with 78 randomized controlled trials (RCTs) included. Among the reviews, a total number of 148 outcomes were predefined and 30 (20.3%) were PIOs related to KSD, with 11 (36.7%) listed as primary outcomes. Among the 78 included RCTs (total participants = 8,232), 38 (48.7%) enrolled predominantly high-risk and 40 (51.3%) enrolled predominantly low-risk population. A total number of 431 outcomes were reported, and 40 (9.2%) were PIOs related to KSD (of which 37 were from studies with high-risk infants), with 13 (32.5%) listed as primary outcome. Cumulatively, no infant developed KSD across all studies.

CONCLUSIONS

There is suboptimal representation of PIOs such as KSD in neonatal trials and Cochrane reviews on NNJ. Over half of the trials included populations with very low risk of KSD, which does not represent judicious use of resources. Amidst our continued search for a more reliable surrogate marker for NNJ, studies should evaluate the whole spectrum KSD alongside serum bilirubin in high-risk populations with sufficiently significant event rates, as this will make the trial methodologically feasible, with findings that will impact the population concerned.