When planning a research project with neonatal patients the first question should be, what am I investigating? The PICO outline : standing for Patients, Intervention, Controls (or comparison) and Outcome, is a standardized way of asking the simple question. If I do X to babies with certain characteristics, what happens compared to not doing X?

‘What happens’ is called the outcome or outcomes of interest. You would think that asking that question in a simple way is very straightforward. Measuring ‘what happens’ in ways that are standardized and reproducible and easily understandable to the potential users of your research should be the goal. How are we doing as a neonatal community?

Regular readers of this blog will know that it is not that rare for a published article to be quite unclear about what exactly the primary outcome was, or how it was defined or measured.

Surely large multicentre trials do better? Not always! Webbe JWH, et al. Inconsistent outcome reporting in large neonatal trials: a systematic review. Archives of disease in childhood Fetal and neonatal edition. 2019.

In this systematic review the authors studied published RCTs in NICUs that had at least 100 babies per group over a recent 5 year period, 2012-2017. They found 76 such trials with over 43,000 total participants.

Across 76 trials 216 distinct outcomes were reported. The most commonly reported outcome was survival; reported in 67 trials (88%). The next most commonly reported outcomes were necrotising enterocolitis (53 trials (70%); bronchopulmonary dysplasia (50 trials (66%); sepsis (48 trials (63%) and retinopathy of prematurity (43 trials (57%). In relation to neurodevelopmental outcomes, visual impairment or blindness were only reported in 21 trials (28%) and 42 trials (55%) did not report any developmental outcomes. Even among the 10 trials involving the largest numbers of infants, major neonatal conditions were not universally reported. Of the 216 outcomes reported, 92 were only reported in a single trial.

Where trials reported the same outcomes, for example, retinopathy of prematurity, these may not be comparable if different outcome measures are used; for example, bilateral
retinopathy of prematurity stage ≥3 and retinopathy of prematurity needing surgery. Sepsis was recorded using 43 different outcome measures; bronchopulmonary dysplasia 16 outcome measures and necrotising enterocolitis 13 outcome measures. In relation to the 216 outcomes, 889 different outcomes measures were reported; of these, 639 were only reported in a single trial.

I have written many times about composite outcomes, and the problems with their use, this review shows their prevalence:

Composite outcomes were used in 41 trials (54%); the most
commonly reported composite being a composite of death and
bronchopulmonary dysplasia (13 trials (17%). This composite
was reported using six different measures and at two time points. There was heterogeneity among composites: 33 different
composite outcomes were reported using 69 incomparable composite outcome measures, with 58 of these outcome measures only reported in a single study.

I have also published a paper about parent and family involvement in neonatal researach, this article clarifies just how far we have to go:

Among 76 included trials and after reviewing published papers and protocols where available, we found that no trial reported patient or parent involvement in outcome selection.

Large clinical neonatal trials are usually clear about what the primary outcome was, and, most often, how it was measured, but if they are not comparable between trials it becomes very difficult to summarize the available literature. Even though certain trials may indeed need to study outcomes that are not universally collected, such as feeding trials, which may need to report outcomes of time to oral feeding competence, as one possible example. It would be very helpful if all trials report a core outcome set, with universally applied definitions and timing. That way different interventions can be summarized and compared, for their impact on the same outcomes, which should be outcomes which are important to families, not just to the medical team.

This study came from the COIN group (Core Outcomes in Neonatology) who are working to do just that. Hopefully the items that we measure in the future include measures of impacts of our care that are standardized and which make a difference in the lives of the babies we care for.

Other groups are also working on the same issue (ICHOM, for example, the International Consortium for Health Outcomes Measurement). I hope they all get together so that we do not have more than one Core Outcomes Set!