Managing Post-Haemorrhagic Hydrocephalus

Posted on 2 March 2026 by Keith Barrington

PHH, as I will call it, is an extremely important determinant of outcomes in a small subgroup of preterm infants. Infants with severe IVH who don’t develop PHH have outcomes that are little affected. As our group reviewed, even grade 4 IVH, if unilateral, and affecting 1 or 2 of the Bassan zones, has little impact on motor or cognitive outcomes unless complicated by PHH.

Andrew Whitelaw died recently (a touching eulogy is on the site of the Newborn Brain Society), I knew him personally, having worked as his resident during a summer in Jersey (Channel Isles, UK), when he covered for my consultant during their vacation. I remember meeting him on the beach with a surf board under his arm, and a pager tucked in his wet suit! Andy was interested in PHH throughout his productive career, and led some important trials in its management, he was the PI of the Ventriculomegaly Trial Group for their pivotal trial, Ventriculomegaly Trial Group. Randomised trial of early tapping in neonatal posthaemorrhagic ventricular dilatation. Arch Dis Child. 1990;65:3–10. That trial did not show any real benefit of routine early LP (or ventricular tapping) compared to “conservative” management of PHH. Of interest, the indications for permanent shunting were typical of the time it was done, “failure to control head size“, which clearly implies rather late intervention. Andy was also critically involved in the development of DRIFT treatment of PHH (DRainage, Irrigation and Fibrinolytic Therapy) for infants with bilateral ventricular enlargement to more than 4mm over the 97%le This involved inserting a ventricular reservoir, then injecting tPA into the ventricles, then perfusing the lateral ventricles with a solution designed to resemble CSF.

Although this invasive therapy has not been widely adopted, it did show improved long term outcomes in a small RCT.

More recently the ELVIS trial (Early versus Late Ventricular Intervention Study, in which Andy was a collaborator) showed improved 2 year outcomes in infants in whom intervention was started early. The protocol for the early intervention group included a trial period of lumbar punctures (maximum of 3) followed by reservoir placement with repeated drainage until CSF accumulation stopped or a shunt was inserted. The primary outcome for ELVIS was survival or needing a permanent VP shunt, which was not different between the groups. But, to be honest I don’t think there was any real chance that survival would be affected by earlier intervention, and, re-reading the original publication, I realize that it is difficult to figure out what the primary outcome rate was in each group, as it is never stated in the text, but from the figure, it looks like it was 19/64 early intervention babies (7 deaths and 12 VP shunts) and 22/62 in the late group (8 deaths and 14 shunts).

The 2 year outcomes, (Cizmeci MN, et al. Randomized Controlled Early versus Late Ventricular Intervention Study in Posthemorrhagic Ventricular Dilatation: Outcome at 2 Years. J Pediatr. 2020;226:28–35 e3) however show definite benefit in the early treatment group.

The adjusted OR for adverse outcome (death or any grade of CP or Bayley (version 2 or 3) motor or cognitive score <-2SD) was 0.24; 95% CI, 0.07- 0.87.

Based on these trial results and other observational data, we developed an intervention protocol which follows best practice, and is similar to suggestions in a fairly recent review (El-Dib M, et al. Management of Post-hemorrhagic Ventricular Dilatation in the Infant Born Preterm. J Pediatr. 2020;226:16–27 e3)

We use criteria very similar to the Yellow zone criteria to start LP attempts, with neurosurgical intervention if dilatation is persistent or progressive according to clear criteria.

One thing I appreciate about the group where I work is our production, and following of, such clinical protocols. We have large numbers of such protocols in every area of neonatal care, and try to ensure that they are followed unless there is a good clinical reason in an individual case for deviating. For such a difficult decision as this, requiring an understanding of the literature, the prognosis, and the therapeutic options, I can’t imagine nowadays trying to make a decision for each infant without the guidance of such a protocol. Which is why I was surprised that so many centres do not have such a protocol.

Coletti K, et al. Post-hemorrhagic ventricular dilatation: Comparison of management pathways among North American level IV NICUs. J Perinatol 2026. This article was from the Children’s Hospitals Neonatal Consortium, of mostly US NICUs (plus Toronto Sick Kids). Among the 40 NICUs that replied, only 12 had a written guideline, another 17 said they had a consensus. I find it hard to imagine what you would do in a NICU without a guideline: “if it’s Monday we intervene at at anterior horn width of 8 mm, but tomorrow it will be 6 mm (unless its the other neurosurgeon)”? Come on guys! What do you say to the parents? “We don’t know what to do, so we toss a coin”?

Even among those few with a guideline, there was a lot of variation, and most seem to have quite late thresholds for surgical intervention:

Mahaney KB, et al. Wide variation in death rates and post-hemorrhagic hydrocephalus (PHH) treatment in preterm severe intraventricular hemorrhage (IVH). J Perinatol 2026. These data from California show wide variation in the approaches to PHH, with many babies not having temporary interventions with reservoirs or subgaleal shunts prior to definitive VP shunting, which strongly suggests late intervention in most such babies. There were about 2000 deaths among about 5,500 babies with severe IVH, another 587 had a permanent VP shunt, 268 of them following a temporizing neurosurgical intervention. Another 89 had a temporary procedure and never had a shunt, either because they died, (n=16) or because, I suppose, it resolved or stabilized.

The evidence supporting early intervention for PHH is limited, it is true. If you want to be critical, you could argue that the improvement in the long-term outcomes in ELVIS is only statistically significant after adjustment for other risk factors. But even if that is your argument, it is important to come to a local, clearly-defined consensus. You also should take into account the observational data, comparing early to late intervention, such as this study comparing the Toronto Sick Kids results (at the time, a late approach) to results from Holland with an early intervention protocol, which was similar to that suggested by El-Dib and colleagues (Leijser LM, et al. Posthemorrhagic ventricular dilatation in preterm infants: When best to intervene? Neurology. 2018;90(8):e698–e706). The early intervention babies in that study had excellent outcomes, similar to babies without IVH.

At each stage in an ideal protocol there are important decisions to be made, to be discussed with parents: a trial of lumbar punctures; how many and when to to stop LPs; when to proceed to temporizing intervention; whether to insert an Omaya reservoir or a Ventriculo-Subgaleal shunt (which may be as much an institutional decision as a personalized one); and when to decide on definitive shunting, which will usually be a Ventriculo-Peritoneal shunt.

I don’t know how a parent could make most of these decisions without clear guidance from a treatment pathway agreed by all involved parties, especially neonatology and neurosurgery.

For babies without contra-indications, and especially for those without a major parenchymal injury, a protocol following the broad outlines of the El-Dib guidance referenced above will minimize secondary injury caused by ventricular dilatation. Of importance, there is no requirement for clinical signs of intracranial hypertension before intervention being indicated. Such signs occur very late in preterm infants, in the “Red Zone”, and should be avoided whenever possible.

Posted in Neonatal Research | Tagged , , , | Leave a comment

Give your opinion: what should we call the most immature babies?

Posted on 27 February 2026 by Keith Barrington

There is an upcoming workshop, sponsored by the AAP, and other groups, that is investigating what shared language should be used for the babies of less than 25 weeks gestation.

At present there are not as many responses as they would like from families, or from former preterm infants. I would encourage any readers of the blog, and especially those from the 2 groups I just mentioned, a few of whom, I know, follow this blog, to participate. The following is the generic notice that was circulated:

” We invite you to take a brief survey to provide feedback for an upcoming workshop about shared language for infants born at <25 weeks’ gestation. The workshop is sponsored by the American Academy of Pediatrics Section on Neonatal-Perinatal Medicine with representatives from the American College of Obstetricians and Gynecologists, Society for Maternal-Fetal Medicine, National Association of Neonatal Nurses, Association of Women’s Health, Obstetric and Neonatal Nurses, Vermont Oxford Network, Tiny Baby Collaborative, and NICU Parent Network. 

Survey participation is voluntary and confidential. The results may be used in workshop proceedings and future presentations or publications. For questions, you may e-mail matthew.a.rysavy@uth.tmc.edu. “

Survey link: https://uthtmc.az1.qualtrics.com/jfe/form/SV_dg9qrNqXGy5Orzw?list_source=ep

Posted in Neonatal Research | Leave a comment

Automated oxygen control; is it worth it?

Posted on 18 February 2026 by Keith Barrington

The history of automated controls of inspired oxygen goes back many decades, to before the invention of pulse oximetry. The first studies I remember used transcutaneous PO2 as the target variable, which had major limitations, as well as the advantage of being better at detecting hyperoxia, at least when it was working well.

There are numerous recent publications about automated FiO2 control, at the end of this post I have put a list of a selection of publications from the last 8 years or so. Such systems have usually been shown to reduce the time an infant spends outside of the desired saturation range, and to reduce the need for nursing intervention.

The different systems use different algorithms, and thus their efficacy in improving “time in range” differs. One of the publications below (Salverda et al) compared outcomes of 2 epochs using different controllers. They had previously noted a major difference in efficacy of the systems in reducing hyperoxia, with the OxyGenie system on the SLE6000 ventilator being much better at reducing hyperoxia than the CliO2 system on the Avea ventilator; but with similar efficacy in terms of percentage time in hypoxia. The clinical outcomes of the babies were different between the 2 epochs, with the OxyGenie epoch having fewer babies developing severe RoP, shorter duration of invasive ventilation, and shorter hospitalisation.

One big concern I have is that increasing FiO2 when an infant is apnoeic is unlikely to improve their oxygenation! Intermittent hypoxic episodes are mostly due to apnoeic pauses, and increasing the FiO2 during such pauses will likely have no impact on the depth of the hypoxia, but may lead to post-apnoeic hyperoxia. It is, unfortunately, difficult to tell if an infant is actually breathing. Central apnoeas are easy to detect, but obstructive apnoeas, and the obstructive component of mixed apnoeas, cannot be easily detected.

The algorithms should be designed to avoid changes in FiO2 during respiratory pauses, but rather to adjust FiO2 when the infant is breathing and their requirements for oxygen change. Exactly how to do that I am not sure; to determine whether an infant is actually breathing, a unidirectional noise-cancelling microphone attached to the chest, with the appropriate software for detection of breath sounds, is an idea I had many years ago and wasn’t able to pursue. Anyone who wants to investigate that idea, feel free.

Randomized trials, without crossover to periods of manual control, have been very few. One small trial listed below was from King’s College in London (Kaltsogianni et al), with only 70 babies in total, using the OxyGenie system on the SLE6000. Babies of GA 22-34 weeks (mean about 27) were enrolled, and randomized to automated vs manual control. The automated group had less hypoxia and less hyperoxia, and had improved clinical outcomes, shorter mechanical ventilation and less BPD.

What was needed was a large multicentre RCT. (Franz AR, et al. Automatic versus manual control of oxygen and neonatal clinical outcomes in extremely preterm infants: a multicentre, parallel-group, randomised, controlled, superiority trial. Lancet Child Adolesc Health. 2026;10(3):179–88) In this trial performed in China, Germany (80% of the subjects), the Netherlands and the UK, infants of 23 to <28 weeks GA were randomized, within 96 hours of birth, to either automated FiO2 control, using whatever ventilator system was available, or manual control, if possible using the same ventilator. A variety of systems were used, most commonly the Stephan Sophie ventilator, at just over half of the sites, followed by the Leoni+ at another 8 sites. The primary endpoint was a composite of any of the following: death, necrotising enterocolitis, or bronchopulmonary dysplasia up to 36 weeks PMA, or severe retinopathy of prematurity by 44 weeks PMA. I have no idea why death was only of interest up to 36 weeks, while the infants were followed in any case until 44 weeks. I don’t know if there were any deaths after 36 weeks.

The study was designed to have 2340 infants enrolled, but, because of recruitment difficulties, the analysis and the sample size had to be adjusted, and 1080 infants were finally included.

There were no major differences in any of the outcome measures, or in the primary, composite, outcome. The primary outcome was 2.7% less frequent among intervention group babies (Absolute difference). In a subgroup analysis, the FiO2 control with ventilator “3” (which they are careful not to name but was used in over half the infants, and therefore must have been the Stephan Sophie) was associated with a larger reduction in the primary outcome, from 41% to 35%, while the other 2 ventilators had slightly worse outcomes in the FiO2-control groups, by 2 or 4%. One always has to be very careful about such subgroup analyses, and the interaction term was not ‘significant’, but it does suggest the possibility that the impact of the automated control differed by algorithm. The BPD part of the primary outcome was significantly different by subgroup of ventilator/algorithm type, again showing a greater reduction with ventilator 3 (17% vs 23%) in the FiO2 control group.

This study did not measure the impacts on nursing workload, or alarm fatigue, 2 things that may be positively impacted by automated FiO2 control.

Currently none of these systems are available in north America, largely because of a lack of trials such as this. From my fairly limited review of this large literature, it seems that the most promising algorithms are the OxyGenie on the SLE6000, which was not used in this recent RCT, and the Stephan Sophie with SPO2C, which was. This trial, with its lack of a clinical benefit, at least showed no safety concerns.

It is hard to argue with a large multicentre trial such as this, but it is also hard to imagine that spending less time hypoxic, and less time hyperoxic, will not eventually be proved to have benefits for our patients.

A Selection of Recent Publications

Dani C, et al. Cerebral and splanchnic oxygenation during automated control of inspired oxygen (FiO2 ) in preterm infants. Pediatr Pulmonol. 2021.
Dani C. Automated control of inspired oxygen (FiO2 ) in preterm infants: Literature review. Pediatr Pulmonol. 2019;54(3):358–63.
Poets CF, Franz AR. Automated FiO2 control: nice to have, or an essential addition to neonatal intensive care? Arch Dis Child Fetal Neonatal Ed. 2016.
Wilinska M, et al. Automated FiO2-SpO2 control system in neonates requiring respiratory support: a comparison of a standard to a narrow SpO2 control range. BMC Pediatr. 2014;14(1):130. Kaltsogianni O, et al. Closed-loop automated oxygen control in preterm ventilated infants: a randomised controlled trial. Arch Dis Child Fetal Neonatal Ed. 2025.
Brouwer F, et al. Comparison of two different oxygen saturation target ranges for automated oxygen control in preterm infants: a randomised cross-over trial. Arch Dis Child Fetal Neonatal Ed. 2024;109(5):527–34.
Langanky LO, et al. Pulse oximetry signal loss during hypoxic episodes in preterm infants receiving automated oxygen control. Eur J Pediatr. 2024;183(7):2865–9.
Salverda HH, et al. Clinical outcomes of preterm infants while using automated controllers during standard care: comparison of cohorts with different automated titration strategies. 2022:fetalneonatal–2021–323690.
Salverda HH, et al. The effect of automated oxygen control on clinical outcomes in preterm infants: a pre- and post-implementation cohort study. Eur J Pediatr. 2021;180(7):2107–13.
Ali SK, et al. Preliminary study of automated oxygen titration at birth for preterm infants. Arch Dis Child Fetal Neonatal Ed. 2022:fetalneonatal–2021–323486.
Dargaville PA, et al. Automated control of oxygen titration in preterm infants on non-invasive respiratory support. Arch Dis Child Fetal Neonatal Ed. 2022;107(1):39–44.
Sturrock S, et al. A randomised crossover trial of closed loop automated oxygen control in preterm, ventilated infants. Acta Paediatr. 2021;110(3):833–7.
Sturrock S, et al. Closed loop automated oxygen control in neonates – a review. Acta Paediatr. 2019.
Gajdos M, et al. Effects of a new device for automated closed loop control of inspired oxygen concentration on fluctuations of arterial and different regional organ tissue oxygen saturations in preterm infants. Arch Dis Child Fetal Neonatal Ed. 2018.

Posted in Neonatal Research | Tagged , , , , | 1 Comment

Prophylactic acetaminophen in the preterm

Posted on 17 February 2026 by Keith Barrington

In a newly published trial (Roze JC, et al. Prophylactic Treatment of Patent Ductus Arteriosus With Acetaminophen: A Randomized Clinical Trial. JAMA Pediatr. 2026) nearly 800 infants of 23 to <29 weeks GA were randomized within 12 hours of birth. There were few exclusions, basically only congenital anomalies and Twtin-Twin transfusion syndrome. The doses were different among the 27-28 weeks infants compared to those more immature, and continued for 5 days in both groups.

The primary outcome variable was survival without “severe morbidity” which was any one of : severe bronchopulmonary dysplasia, stage II or III NEC, grade III or IV IVH, or cystic leukomalacia. Severe BPD used the 2018 NIH consensus definition, i.e. invasive ventilation with more than 21% O2, or CPAP with more than 30% O2, or high flow at more than 3 lpm. Outcome was determined at 36 weeks or discharge home.

There was no outcome difference of importance between the groups, as you can see below.

I don’t know why they inverted the groups between part A and B of the figure, which misled me briefly, as you can see there is a minor difference in NEC in the most immature infants, and among females, but this may well be a random variation.

It is also revealing that Acetaminophen was not very effective at closing the PDA :

Among babies who had an echocardiogram at 7 days of age, the rate of PDA closure was 43% in the more immature controls, and 61% with active medication. In comparison, prophylactic ibuprofen seems probably a little more effective for PDA closure; the Cochrane review of prophylactic ibuprofen shows closure of the PDA on day 3 to 4 of life of 58% in the controls compared to 83% with ibuprofen, from meta-analysis of 9 studies with a total of about 1000 babies.

21% of controls and 14% of acetaminophen infants had medical PDA “backup” treatment, which started a median of 7 days after birth, with a 1st quartile of 3 days. So, many of those “back-up” treatments started during the initial randomization period. A total of around 3% in each group had, eventually, formal closure (surgical or catheter) of the PDA.

The more immature GA stratum had the poorest response to acetaminophen, but still had many more PDAs closed at 7 days of age with acetaminophen than placebo. Apart from numerically fewer cases of NEC in this subgroup, with 95% confidence intervals which include no difference, there is no apparent difference in outcomes, either in the components of the primary or any of the secondary outcomes. Among the secondary outcomes, in the more immature group, there were also numerically fewer pulmonary haemorrhages, 4.5% vs 8%, and fewer babies who had catecholamine support or postnatal steroids in the first 7 days with acetaminophen treatment. But again, for all of these findings, the confidence intervals include no difference.

This trial confirms what we have been seeing in many recent studies. There is no advantage to routine PDA closure. There are still those who seem certain that there are some PDAs that should be closed in small preterm infants. My response remains that it may be true, but we really need a way to define that subgroup of infants, and then find the most effective, least toxic way of closing their PDA.

Posted in Neonatal Research | Tagged , , , , | 2 Comments

Not just neonatology, trip to Rwanda

Posted on 17 February 2026 by Keith Barrington

I was very fortunate to be able to take a trip to Rwanda to participate in their neonatal training scheme, in what was termed a “respiratory bootcamp”.

Rwanda is a small country of 23000 km2, with a young population of over 14 million, and 400,000 births annually. Which I can’t help comparing to Quebec, with an area of 1.5 million km2, a population of 9 million and annual births of 80,000.

There has been remarkable progress in perinatal health care and outcomes over the last decade, with a marked improvement in the proportion of deliveries attended by a trained birth assistant, and taking place in an institution (now >95%), with a dramatic reduction in maternal mortality from over 1000 to about 300/100,000 births. This has been accompanied by a decrease in stillbirth rates, and in neonatal mortality. There are now 3 NICUs in Kigali, and another in development in Butare, with 4 neonatologists in the country, all in Kigali. Their young fellowship program will soon have the first locally trained neonatologists. I feel privileged to have been a small part of this program.

In addition to visiting the NICUs and spending time with fellows, nurses and staff, I was able to take a tour of the country, visiting the 3 of their 4 national parks where there is accommodation. The fourth is currently being developed. I started in Volcanoes national park, where I was able to pursue one of my interests, viewing wildlife and bird photography. You can click on any photograph to see it full-sized.

Regal Sunbird
Scarlet-breasted Sunbird
Northern Double-collared Sunbird

These 3 are all Sunbirds, a large family with very colourful males, and almost uniformly drab females, who eat nectar, filling a similar ecological niche to the Hummingbirds of the Americas.

In Volcanoes park, a small troupe of Golden Monkeys were eating and watching us carefully, but I decided against buying the permit to view the Mountain Gorillas, maybe next time. I also visited some wetlands in the region, and photographed the birds below, among many others.

Black Crake
Hadada Ibis
Pied Kingfisher

Then on to Nyungwe national park, which is in the South-west corner of the country

Augur Buzzard
Yellow-eyed Black Flycatcher
Black-headed Waxbill

And has troupes of Black-and-white Colobus and Blue Monkeys

Black-and-White Colobus
Blue Monkey

We visited another wetlands as we headed back eastwards

Fulvous Whistling Duck
Southern Red Bishop
Black-headed Gonolek
Swamp Flycatcher

Then on to Akagera national Park, the oldest and largest in Rwanda

Masai Giraffes and Plains Zebras in a Valley of the Akagera park.

Long-crested Eagle
Bare-faced Go-away Birds on termite nest
Cape Buffalo and Yellow-billed Oxpecker
Cardinal Quelea
Woolly-necked Storks
White Rhinoceros and Western Cattle Egret
White-browed Coucal
Rufous-breasted Heron
Blue-headed Coucal
Woodland Kingfishers
African Fish Eagle

I was able to return to Akagera for another day, later in my trip

Black Cuckooshrike
Pin-tailed Whydah
African Openbill
Lilac-breasted Roller

In Kigali, close to the airport, there is a preserved wetlands, which has been turned into an Eco-park.

White-collared Oliveback
Black-and-white Mannikin
Gabar Goshawk
Red-chested Sunbird
Village Weaver and nest
Little Bee-eater

If you are interested, there are many other photos on my other blog, keithbarrington.com

Posted in Not neonatology | 3 Comments

How much oxygen for the resuscitation of the preterm?

Posted on 29 December 2025 by Keith Barrington

I like a good acronym, so my initial response to the new TORPIDO trial was very positive! TORPIDO 30/60 was a large multicentre RCT comparing initial FiO2 concentrations for resuscitation of the preterm infant (Oei JL, et al. Targeted Oxygen for Initial Resuscitation of Preterm Infants: The TORPIDO 30/60 Randomized Clinical Trial. JAMA. United States2025). The first TORPIDO study (Targeted Oxygen in the Resuscitation of Preterm Infants and their Developmental Outcomes) was a comparison in 290 infants <32 weeks GA, of 21% starting O2 concentration vs 100% starting concentration (Oei JL, et al. Targeted Oxygen in the Resuscitation of Preterm Infants, a Randomized Clinical Trial. Pediatrics. 2017;139(1)) . It had to be stopped early because it became difficult to enroll patients as we became reluctant to use 100% oxygen. Although underpowered, there was a numerically greater mortality in the 21% oxygen group, especially among the infants <28 weeks.

This contributed to an apparent increase in mortality among lower FiO2 regimens when all the Individual Patient Data were included in a Network Meta-Analysis, NETMOTION. (Sotiropoulos JX, et al. Initial Oxygen Concentration for the Resuscitation of Infants Born at Less Than 32 Weeks’ Gestation: A Systematic Review and Individual Participant Data Network Meta-Analysis. JAMA Pediatr. 2024).

That NMA suggested that the higher starting oxygen groups (90% or higher), had a lower mortality than low starting concentration (30% or less), while intermediate, >30% to <90%, was similar to low starting concentration. Of note, the 1st TORPIDO study was the largest in that NMA.

Unfortunately, 10 of the 12 trials in that NMA were very small, 31 to 95 subjects, the 2 larger ones were modestly sized (287 and 193). For such an important outcome, and an easily applied, no-cost intervention, better, larger trials were needed.

The new TORPIDO 30/60 trial started in 2018, well before NETMOTION was performed. According to the choices made by the authors of NETMOTION, it would have been considered a comparison of intermediate (60%) to low (30%) oxygen. At the time of designing it, it was considered unlikely that very high starting oxygen concentrations would be a good idea, so the choice of a comparison between 30 and 60% was entirely reasonable.

The new trial results show no difference in the primary outcome, or in any indices of brain injury on ultrasound.

There were some differences in immediate responses in the delivery room, “Newborns in the FIO2 of 0.6 group, compared with the FIO2 of 0.3 group, were less likely to receive chest compressions (2% vs 5%) or epinephrine (1% vs 2%) and more likely to reach SpO2 > 80% by 5 minutes (58% v 44%) (Table 3) and had higher initial SpO2.” Which are interesting differences, but did not translate to any measurable difference in clinical outcomes.

This suggests to me that we have to push this further, and compare a starting FiO2 of 60% (which seems no worse than anything lower, and might have some short term advantages) to a starting FiO2 of 90 or 100%. It seems weird, after all this time, to talk of going back to 100% oxygen for initiating resuscitation. It seems only a few years ago (but I guess it is more than 2 decades now) that I had to approach the hospital administration at my previous post to have pressurized room air and air/oxygen mixers installed in the delivery room, so that we could start resuscitation with 21%, and titrate if needed. The DR, in a relatively new building, had been built with just piped oxygen for neonatal resuscitation. The administrators of the Royal Victoria Hospital in Montreal were very responsive, and immediately approved the budget when I presented the evidence.

Most of my readers will know that the evidence for room air resuscitation in full term infants was quite strong, and that routinely starting with 21% oxygen was associated with a 27% lower mortality than 100 % oxygen, as this ILCOR meta-analysis showed. https://publications.aap.org/pediatrics/article/143/1/e20181825/76868/Room-Air-for-Initiating-Term-Newborn-Resuscitation

But if the data for preterm infants is taking us back to much higher oxygen concentrations, I guess we will have to decide whether 34 week infants should be treated like the 35 weekers, or like the more immature infants. Just like my comments about cooling, I think it is unlikely that the risks change dramatically at midnight between 34 weeks 6 days, and 35 weeks. We will need more granular information in order to make decisions for infants in the late preterm age ranges. Hopefully. those doing the trials will provide Individual Patient Data, that will allow analyses according to gestational age, and other risk factors.

Posted in Neonatal Research | Tagged , , , , | 2 Comments

Universal Neonatal Genome sequencing?

Posted on 13 December 2025 by Keith Barrington

It is over 2 years since I posted about this issue, in the meantime there have been a number of new initiatives, and several review articles and opinion pieces. Including a JAMA, “research of the year” article”.

I was triggered to return to this issue by a recent example of such an editorial which claimed that universal neonatal genome sequencing (UNGS) “pilot programmes like the Generation Study by Genomics England, which showed that whole-genome sequencing could identify rare, treatable conditions in about one in every 200 babies.” (Of note there is no author given for this editorial, perhaps it was generated by AI, interestingly a letter from a genomics researcher in the UK, in response to a letter to the editor, actually quotes this editorial as if it were reliable data from the project!)

This sounded to me to be a bit higher than previous estimates I have seen. Sometimes the supposed advantages of a Whole Genome Sequence have been dramatically overstated, for example, counting “referral to a pulmonologist” in the accumulated sum of benefits of screening. For large numbers of WGS performed in the NICU in sick infants, when there is actually a diagnostic result, the consequence has been redirection of care to comfort/palliative care, and sometimes the delay, waiting for the result, leads to a delay in redirection of care, when it could have been performed sooner without the WGS. For UNGS on the other hand, we really want to detect conditions in a pre-symptomatic phase which are amenable to intervention, and for which intervention improves outcomes, either survival, illness, or quality of life.

We also want to avoid false positives, that is detection of genetic variants that are abnormal, and associated with a disease, but do not cause the disease in that individual. The case of Krabbe’s disease, screened for in New York state, comes to mind. In nearly 2 million screens, positive results occurred in 620, of whom 346 had the genetic abnormality. Only 5 of them actually turned out to have infantile Krabbe’s disease, a PPV of only 1.4% for the disease among those who had the abnormal gene.

I decided to do my usual, and verify the source of the “treatable diagnosis in 1/200”, statement, and it seems that the authors (or the AI) have severely misinterpreted the results from Genomics England. That reference noted that, in the proposed 100,000 neonatal genomes they are accumulating, they are searching for a list of 200 different diagnoses, which they have selected as being candidates based on 4 principles. The principles are (a variation on the Wilson and Jungner criteria, which were modified to apply specifically to genetic screening): that a known gene is responsible for a disease; that the gene variant usually causes an important condition; that the condition is treatable when found on screening; and that all persons have access to treatment equitably. Which are principles that I think we can all agree with.

Other publications from Genomics England state that “fewer than 10% of rare genetic conditions have a treatment that is effective in improving patient outcomes”. (That isn’t referring to the conditions in their screening list, but in general). Of course, for several of the more common treatable conditions, universal neonatal screening is already in place (PKU, cystic fibrosis). What is the additional yield of treatable diagnoses, compared to standard screening? Surely that is the most important question, but I have had a lot of difficulty finding an answer. That is partly because standard screening is not standard around the world, but extremely variable, and the list of genetic conditions included as the targets in each program is extremely variable.

The various pilot studies, and there are now many, are also extremely variable, and have a diagnostic yield which is very variable, depending on which conditions are included, and usually do not state how many of them are actually treatable. Indeed there are over 30 current research programs into UNGS, as well 8 commercial bodies offering NGS if you pay for it (shown with the + symbol in the figure).

Of course, there may be benefits of a diagnosis, without an actual “treatment” as such. In one pilot, from Belgium, the largest diagnostic group, by far, was G6PD deficiency, which accounted for almost 50% of the positive UNGS results. That was a modestly sized study (about 3,900 babies included) with a diagnostic result in 71 cases, 44 of which were G6PD deficiency. The remainder were a variety of diagnoses with a variety of actions that were required. Four cases of CF, for example, 2 of biotinidase deficiency requiring biotin supplementation, and 2 of haemophilia with mild reductions in Factor VIII levels requiring intervention if they need surgery. There were several diagnoses that required enhanced specific surveillance in the future (cardiomyopathy variants for example).

The diagnosis of G6PD deficiency may indeed be a real benefit to the patient, who can avoid triggering factors for life, so I would count that as a potential benefit of UNGS; many screening programs, including in Quebec where I live, do not currently include G6PD. Indeed there are published standards which include the regulations in New York State where “public health law was amended to include quantitative diagnostic G6PD deficiency testing for infants with hemolytic anemia, hemolytic jaundice, early-onset increasing neonatal jaundice persisting beyond the first week of life (bilirubin >40th percentile for age in hours), admitted to hospital for jaundice following discharge, or familial or population risk for G6PD deficiency. G6PD testing is performed as a hospital diagnostic test, not by dried blood spot testing”. Does the detection of G6PD deficiency, by UNGS, confer an additional, measurable benefit to the infant, over such an aggressive case finding policy as that in New York?

One of the “publications of the year” in the JAMA group of journals, was the description of the GUARDIAN pilot of 4000 UNGS results. That study found 120 true positive results in the infants, of which 110 would not have been detected by traditional neonatal screening, 92 of those were G6PD deficiency. The remaining 18 infants include 1 with a fatty acid disorder (MCAD) 1 with achondroplasia (surely, a good physical exam would have picked this up) 1 with hypochondroplasia, 1 with Wilson’s disease, and 2 with prolonged QT. This study used a list of 156 “actionable conditions”, and there was a subgroup of infants whose parents consented to expanded testing including a further 99 conditions which were developmental disorders with an increased seizure risk. Among the infants who had that expanded testing they found a variety of disorders, total n=6, including, for example, 2 with different genes linked with Rett syndrome.

The large number of studies in progress is an indication of the uncertainty of the benefits. I also am very concerned about the potential for abuse in the future, with every infant’s genome sequenced in a database somewhere. My provincial government has just imposed a law which allows government officials access to everyone’s medical record. Ostensibly this is to improve co-ordination and access, but it is a serious intrusion into the confidentiality of medical records, and is being strongly contested by medical organisations here. Once everyone’s genome is recorded in a database, the potential for abuse is enormous.

The largest array of genes being screened for appears to be in BabySeq, with over 4000 genes in their list. Before instituting such a program as the universal norm, we have to be certain that the benefits, for the small number of infants who are detected by UNGS over and above the regular neonatal screening programs, outweigh the possible risks. A first step would be to determine whether detecting G6PD deficiency in asymptomatic infants actually improves their outcomes in some measurable way. The second step would be to analyse the outcomes of the 0.5% of babies who have another diagnosis, some of whom require follow-up, and some require a preventative intervention.

Most of the large programs around the world seem to be taking such a cautious approach. I am unsure if the commercial programs, whose primary motivation is profit, are quite as reasonable.

Posted in Neonatal Research | Tagged , , | Leave a comment

Why doesn’t closing the PDA help preterm infants?

Posted on 12 December 2025 by Keith Barrington

I have been continuing to think about this issue, so will subject you all to some of my thoughts. The large RCTs of medication for closure of the PDA show no benefit, and some evidence of harms, but we all know that the PDA usually closes shortly after birth in the term infant, and there is a statistical association between a persistently patent ductus in the very preterm infant and several neonatal complications.

So why don’t babies benefit from closure of the PDA?

I think there are 2 possibilities. One is that the medications are the problem. They are not very effective, and we expose the babies to reduced urine output, and other side effects, for unreliable constriction of the PDA, and a relatively small proportion of complete closure of the duct. Especially among the most immature babies, the drugs don’t work very well, and they have more secondary effects. Perhaps there are benefits to closing even modestly sized PDAs, but they are overwhelmed by the adverse effects of the medications we are using.

It is interesting that in the causes of mortality, as far as they can ever be definitively determined, there does not appear to be a difference in NEC, according to the meta-analysis that I referenced before, despite concerns of some that there may be an impact on GI perfusion of the meds. That SR/MA does, however, show an increase in severe IVH, PVL, and culture positive sepsis. None of which are individually statistically significant, but all of which suggest potential impacts that we often don’t consider. The decrease in urine output could also lead to an increase in lung water, ventilatory requirements and lung injury. So maybe the medications are the problem.

The other possibility is that the PDA is not etiologically linked to the associated complications, but is a bystander. I think for PDAs with a small shunt, this is probably the case. Centres with a very restricted approach to the PDA do not, generally, have worse outcomes overall. However, I think that is likely that there are babies who benefit from closure, including, perhaps, early closure.

Are there any babies, in controlled trials, who seem to benefit from PDA closure, in the current literature? Afif El-Khuffash and colleagues performed a secondary analysis of their trial which only included infants with a large shunt. The original trial publication (El-Khuffash A, et al. A Pilot Randomized Controlled Trial of Early Targeted Patent Ductus Arteriosus Treatment Using a Risk Based Severity Score (The PDA RCT). J Pediatr. 2021;229:127–33) showed a null result, but it was intended as a pilot trial, and was therefore under-powered for clinically important outcomes. The secondary analysis, was, therefore, even more underpowered, (Bussmann N, et al. Patent ductus arteriosus shunt elimination results in a reduction in adverse outcomes: a post hoc analysis of the PDA RCT cohort. J Perinatol. 2021;41(5):1134–41) there were only 17 infants who had successful closure of the PDA in the group with large PDA shunts who were randomized to treatment with ibuprofen. Among those who actually managed to close their PDA, they were better off than the placebo group infants in terms of “death or BPD”, 5/18, compared to 18/30 control group babies, which isn’t “statistically significant” by a simple chi-square. All secondary analyses are suspect, but, it does suggest that there may be a subgroup of babies who benefit from PDA closure. Interestingly, the active treatment group infants who did not close their PDA, were actually worse off (11/13 had “death or BPD), and possibly had more deaths.

One way of eliminating the adverse medication effects would be to close the PDA without medication. For those who are not aware, there is a trial in progress of ductal closure by catheter, the PIVOTAL trial, which randomizes infants of 7 to 32 days of age, <28 weeks GA, and over 700 g birth weight, who have a PDA score of 6 or more. The PDA score in use for this trial has been adjusted from that PDA RCT. I asked my good friend, the PI of the study, Carl Backes which score they were using and he very helpfully sent me this table, which I am sure he won’t mind me sharing.

ECHO-Based Scoring System0 points1 point2 points
PDA size indexed to weight<1.51.5-3>3
Mitral inflow velocity (E) (cm/s)<4545-80>80
IVRT (ms)>5030-50<30
PV D-wave (cm/s)<0.30.3-0.5>0.5
LA:Ao<1.31.3-2.2>2.2
LVO:RVO ratio OR LV-VTI:RV-VTI*<1.51.5-2.0>2.5
Descending aorta AND/ORceliac AND/OR middle cerebral artery flowForward Reversed
PDA size indexed to weight= (minimum PDA diameter [mm]) / (weight [kg] at time of ECHO); IVRT=isovolumic relaxation time; PV=pulmonary vein; LA:Ao=left atrium to aortic ratio; LVO=left ventricular output; LV-VTI: LV output velocity time integral; RVO=right ventricular output; RV-VTI: RV output velocity time integral*obtain both LVO:RVO and LV-VTI: RV-VTI measurements; the highest score from either ECHO parameter is counted towards overall score*use the high suprasternal arch view to obtain the post-ductal diastolic flow measurement;  reversal present from an abdominal view is acceptable, but the “absence of reversed flow” from an abdominal view may be a false negative, thus the need for the suprasternal view.

I hope that the PIVOTAL trial is planning a subgroup analysis by PDA shunt size or by gradations of the score, I think it is unlikely that a score of 6 or more was exactly the right choice, as the threshold for enrolment, but I could be proved wrong. The planned sample size was 240, and they are apparently about 75% complete. The trial will at least eliminate the problem of medication toxicity, but instead introduces the risks of catheterisation!

Catheter closure has rapidly been growing in the USA, the latest data I can find are from 2021, and they show things like the following, from the NICHD NRN

Indomethacin is disappearing, ibuprofen is relatively stable, acetaminophen has become the commonest intervention for the PDA, and the progressive disappearance of surgery is being replaced with the very rapid growth of catheter closure, being nearly 5% of babies born <29 weeks.

Given that there is an absolute lack of evidence that catheter closure of the PDA improves any clinical outcomes, it is very depressing to see that 1 in 20 extremely preterm infants in the NRN are being subjected to this procedure, almost all of them outside of any clinical trial.

Equally depressing the “Society for Cardiovascular Angiography & Interventions” (SCAI, a new one for me) have just published a position statement with a pre- and post- procedure checklist. Carl Backes, and his co-author wrote an accompanying editorial and they were much kinder about this position statement than I would have been. My response would have been : WTF? How can you write a position statement for a procedure completely lacking in supporting evidence that it is effective or safe, in terms of clinically important outcomes? Surely the position statement should have stated, that this procedure should not be performed outside of a prospective study.

Hopefully the PIVOTAL trial will prove to be pivotal, and will have a clear result, of benefit or no benefit. The primary outcome is ventilator free days, a reasonable way of analyzing the impacts on ventilator dependence or death (as deaths will be counted as 0 VFDs). Although I would have wanted other outcomes, in particular some evidence that effects on lung injury, using measures that are important to families, are improved, without an increase in mortality.

Posted in Neonatal Research | Tagged , , , | 1 Comment

Should we ever close the PDA?

Posted on 10 December 2025 by Keith Barrington

The latest large multicentre RCT has just been published. Laughon MM, et al. Expectant Management vs Medication for Patent Ductus Arteriosus in Preterm Infants. JAMA. 2025. In this trial, infants of 22 to 28 weeks GA were randomized at between 48 hours and 21 days of life after an echocardiogram. They were classified into: 1. no or asymptomatic PDA; 2.symptomatic PDA; or 3. cardiopulmonary compromise. Only group 2 were randomized. The definitions are shown below, including, at the bottom of the figure, the definition of group 2. Of note, infants receiving hydrocortisone were ineligible.

481 infants were randomized, within 48 hours of being eligible, and at a median of 10 days (IQR7-14) of age to either medical treatment, with ibuprofen, indomethacin or acetaminophen, or to control, expectant management. Controls were not supposed to receive any of those drugs unless they progressed to group 3 (cardiopulmonary compromise) or reached 36 weeks. There were unfortunately a large number of protocol violations, 60, or 25% of control, expectant management, infants received medical intervention (or surgical/catheter closure) for their PDA, of which 44 did not meet the agreed treatment criteria, and were therefore protocol violations.

I find this a little hard to understand, why get involved in the study if you are not prepared to abide by the protocol? Nearly 1 in 5 expectant group babies had PDA closure attempted even though they were not in the category of having severe clinical criteria with a large PDA, and therefore did not satisfy the protocol indications for treatment.

The primary outcome criterion was survival without BPD. I know, don’t get me started, designing a trial with a dichotomous outcome, that equates death with being in oxygen at 36 weeks, would be ridiculous in this day and age, in 2025, when so much better ways of designing trials with potentially conflicting outcomes, and analysing them in ways that take into account the relative importance of the outcomes, exist, and are now being used in other fields. It also leads to other rather, er, questionable decisions, such as defining death as death up to 36 weeks PMA. Really? Who cares about death up to 36 weeks, so an infant who dies at 37 weeks wasn’t counted in this trial?

The trial was, strictly speaking, a null trial. The primary outcome was identical between groups.

The breakdown of the primary outcome shows a lower mortality with expectant management than with intervention. At least, that is, mortality up to 36 weeks.

Surely the most important single question about any trial in sick preterm infants is : “If the baby receives treatment for the PDA, compared to expectant management, are they more likely to go home alive, or not?”

It takes a search of the supplemental materials, supplemental document 3, eTable 9, to find a partial answer to that question. The answer is that, by discharge or transfer, or 120 days after randomization (there were 18 infants still in the NICU at 120 days, and the investigators terminated the data collection), there were 14 deaths of the 241 expectant treatment infants. In the active treatment group there were 26 deaths among the 235 who actually got active treatment, there were 5 babies in this group who were not treated within 48 hours, and we don’t know if they survived or not.

If I do an ad hoc ITT chi-square, removing the unreported infants who had data collection truncated at 120 days, (as we don’t know if they lived or died) then the mortality is 14/237 vs 26/234. Or 5.9% vs 11.1%, a risk difference of 5.2% (95% CI of -0.2% to 11%), in other words not “statistically significant”.

As regular readers will know, I don’t think the arbitrary cutoff of p<0.05 is a good way to define what is real or not, which is why I almost always put “statistically significant” in quotation marks. But still, surely it was important to know that there were 7 pre-discharge deaths, at least, of the 40 total deaths, that occurred after 36 weeks. And there may have been more deaths after 120 days.

This has been a recurrent problem in similar studies, in BabyOscar for example, (Gupta et al 2024 in the figures below) mortality was reported at 36 weeks, and I can’t find any data anywhere about survival to discharge.

This is vitally important, the most recent meta-analyses show that medical treatment of the PDA increases BPD. Infants with severe BPD, still ventilated at 36 weeks have a measurable late mortality. Surely that is an outcome that we should know about?

A recent SR/MA, published in May this year : Buvaneswarran S, et al. Active T “reatment vs Expectant Management of Patent Ductus Arteriosus in Preterm Infants: A Meta-Analysis. JAMA Pediatr. 179. United States2025. p. 877–85, included 10 trials of infants <33 weeks GA with “hemodynamically significant PDA (diagnosed by clinical or echocardiographic criteria)” who were randomized to active treatment compared to expectant management. There were a total of over 1000 babies per group and the primary outcome Forest plots are below

As you can see, this analysis suggests that there may well be an increase in “death at 36 weeks or at discharge (whichever occurred later)”, by which they actually mean whichever were the latest reported mortality figures.

They also include in the secondary outcomes, mortality before discharge, reported for less than half of the babies included in the various studies, which also shows an increase, but less marked, and with 95% CI including no difference.

To return to the new trial publication. I should make it clear that the trial was started in 2018, therefore probably designed in 2016 or so, and the authors are the brilliant Matt Laughon and the NICHD NRN centre representatives. The outcomes that they chose back then were the typical outcomes of PDA trials. It is easy for me, tapping away for my blog, to criticize, in retrospect, decisions that were taken a decade ago…

Nevertheless this trial, if you add it to the meta-analysis above, would surely confirm that closing a “haemodynamically significant” PDA does not appear to have any measurable benefits, and may well increase both mortality and oxygen requirements at 36 weeks, with most of the increase appearing to be in milder BPD, but probably a small increase in more severe disease also.

I am struggling to think of an evidence-based indication to close the PDA in a preterm infant. The data are, unfortunately rather muddy, there have been many protocol violations in the majority of the large trials. The exception was Hundscheid et al in the figure above, the BeNeDuctus trial, which only had 1 protocol violation in the expectant group. In BabyOscar, in contrast, 30% of the placebo group received open label medical or surgical PDA closure, about 1/3 of whom did not satisfy the protocol defined criteria. In the Rozé trial, 62% of the placebo babies had open label treatment, it is not clear how many satisfied their criteria for open-label treatment.

Which brings me to the question in the title, “Is there any indication for PDA closure?” The evidence-based answer to that question is that there are no criteria for defining a clinically important PDA, for which medical or surgical intervention has shown a survival benefit, or a reduction in lung injury.

I put it that way because there are some experts who continue to suggest that the big problem is with how we define a clinically important or “haemodynamically significant” PDA. (Bischoff AR, et al. Beyond diameter: redefining echocardiography criteria in trials of early PDA therapy. J Perinatol. 2025). And that all we have to do is better define the phenomenon. Which may be true, but requires that we prove it.

That recent opinion piece suggested that the criteria that should be used are those being tested in a pilot trial, the “Smart PDA” trial, which are not enormously different to the criteria used in the NICHD trial above. The authors of that piece remark that simply defining a significant PDA by diameter is insufficient; some PDAs with a large diameter have relatively modest impacts, and others of the same size may be associated with major shunts. The big difference between the Smart PDA trial, and the newly published NICHD RCT is that, in the newly published trial, being on CPAP, and having a PDA of over 1.5 mm would qualify for enrolment, without any other signs of a large shunt. In Smart PDA, you will also need at least one of the following signs of a L-R shunt

  1. Left atrium: aortic root ratio 1.5–2.0
  2. Transductal peak systolic velocity 1.5–2.0 m/s
  3. Left ventricular output (ml/kg/min) 200–400
  4. Diastolic flow pattern in the descending aorta: Absent/ retrograde

It seems to me that this has to be the next stage in the process, we should stop treating PDAs that do not have signs, such as those, of a substantial L-R shunt, doing so seems to have no benefit, and may well increase both oxygen requirements at 36 weeks, and perhaps even mortality. I think we have now reached the point where medical or surgical closure before 36 weeks PMA should only be attempted, in the context of an RCT, in infants with signs of a large shunt. There is currently no proven benefit to early PDA closure, and only harms.

In my recent practice I have seen some babies, usually “older” infants still ventilator dependent near term, with large shunts from a PDA, who have improved rapidly after ductal closure. In an NICU with 100 extremely preterm babies a year, there were maybe 3 or 4 in the last 4 years. It may be that those infants would have benefited from earlier closure, but in the absence of clearly defined criteria, which have been shown to predict benefit in preterm infants with a PDA, we frequently hesitate before performing a procedure with known risks. We have also had babies who have had little change in their clinical status after late PDA closure.

There do seem to be some babies who, anecdotally at least, seem to benefit from closure of the PDA. Our challenge as a community is to identify them, and hopefully to be able to identify them early enough that we can improve their outcomes.

Posted in Neonatal Research | Tagged , , , , , | 2 Comments

Is this article trustworthy?

Posted on 21 November 2025 by Keith Barrington

There are many ways that an article in the medical literature may not be worthy of trust, and whose results may be unreliable. I remember one of my own trials, LACUNA, which was a pilot RCT of lactoferrin supplementation to prevent late-onset sepsis (LOS). As a pilot. the primary outcome was feasibility of a larger trial, but the clinical outcomes that I reported included the incidence of culture-positive sepsis. I calculated that as planned, as the proportion of babies who had at least one episode, out of the total number included in each group.

When I looked at the data, I realized that if I presented the results as infections per 100 patient days, there was actually a statistically significant difference between groups! I must admit to being tempted to report that outcome, as if it had been the primary outcome of my trial. I think the trial would have had a greater impact than it actually did, and could potentially have led to widespread adoption of lactoferrin as prophylaxis against late-onset neonatal sepsis. I could have been famous!

I am glad I resisted the temptation, because I know that it is vitally important to follow the trial design, outcomes, and statistical analysis that are planned prior to the trial. Any new analysis, or redefinition of outcomes, which is decided after examining the data, is inherently, extremely unreliable.

If I had reported that outcome, probably few people would have realized that I had changed the definition after performing the trial. The trial was pre-registered, but few people look at the registration documents, or compare the reported outcomes to those which were pre-planned. That includes peer reviewers, and meta-analysers.

It tuns out, of course, that the lactoferrin preparation that we used almost certainly has no real impact on LOS, and if I had reported the other outcome, it would eventually have been proven to be a red herring.

I was stimulated to think about this issue, which reminded me of the LACUNA outcomes, by a post from the Health Nerd https://gidmk.substack.com/p/the-terrifying-problem-of-fraud in women’s health. It is a really interesting blog, in general, and this post is indeed terrifying. He refers to several Cochrane reviews, including 2 that have major neonatal implications. He starts by discussing an influential Cochrane review of vitamin D supplements in pregnancy, that, in previous versions, was very positive, and led to recommendations from various health groups to give such supplements. He notes that the new version of the review has removed 21 articles because of trustworthiness issues. Twenty-one! The review now shows no benefit, and, in fact, potential harms.

A review with major neonatal impacts is the Cochrane review of steroids prior to Caesarian Delivery at term. The previous version included 4 trials, with nearly 4000 babies, and showed some significant benefits. The new version has removed 3 of those trials, for being untrustworthy, and now only shows a possible reduction in NICU admission among the 900 babies included in the 1 remaining trial, but not in RDS, or respiratory support. That single included trial did not report hypoglycaemia, which other trials, in late preterm infants, have shown to be a major potential complication of steroids in that population. Based on this new Cochrane review, I think such use of steroids is highly questionable.

Another Cochrane review from the obstetric group with a neonatal impact is their review of sildenafil and other NO pathway medications for fetal growth restriction. The initial search identified 19 studies, but 11 had to be excluded because of trustworthiness issues. Even one of those included was not prospectively registered, so they performed a sensitivity analysis excluding those data. Only sildenafil had enough data to be informative, and the review shows no benefit.

It is now essential that systematic review authors evaluate the trustworthiness of the trials being included. Almost all journals state that they follow the ICJME guidelines, which preclude the publication of trials which are not pre-registered. On many occasions however, I have seen articles published that are reports of clinical trials, but do not mention registration, or were retrospectively registered. Systematic reviews often include data from those trials without commenting on the issue. There should always be a sensitivity analysis excluding unregistered or retrospectively registered trials in a SR. The SR of erythopoietin for prevention of NEC, that I discuss in the post linked below, went from showing a reduction in NEC with erythropoietin, to a null effect when the retrospectively registered trials were deleted.

Just like my example above, untrustworthy research is not necessarily fraudulent. It may be a minor twist to the definition of the outcome, or deciding post hoc, that a certain observation is an outlier, or changing the analysis from treating a value as a dichotomous to a continuous outcome. Sometimes, of course, there is overt fraud, with entirely imaginary numbers. The pressure on individuals in some countries to produce published research is so enormous, that there is no consideration of the downstream harms that can be caused by such fraudulent results. Or by results which are “tweaked” to show a benefit.

The problem of untrustworthy published research is not new, but I fear that it will continue to grow, with articles now being generated entirely by AI. The least we can do is to demand that all published RCTs are publicly registered, in a database including projected sample size, eligibility criteria, and primary outcomes.

Badly performed, badly presented, unethical published research; what to do?

Those of us performing systematic reviews must now perform trustworthiness assessments, using one of the published tools. It substantially increases the workload involved, but will also improve the reliability of the results.

Posted in Neonatal Research | Tagged , , , | Leave a comment