Erythropoietin does not protect the brain, what next?

Posted on 27 July 2022 by Keith Barrington

The latest trial to fail to find a benefit of erythropoietin (Epo) for neonatal brain protection has just been published (Wu YW, et al. Trial of Erythropoietin for Hypoxic-Ischemic Encephalopathy in Newborns. N Engl J Med. 2022;387(2):148-59). The HEAL trial, a multicentre randomized controlled trial in 500 asphyxiated term infants undergoing cooling found a similar incidence of death, of neurologic impairment, and of developmental delay at 2 to 3 years with Epo or placebo. There was also no difference in MRI injury, either as a percentage or when looking at the patterns of injury, or in discharge neurological examination.

Epo was started before 26 hours of age, in fact at a mean of just less than 18 hours, the dose was 1000 u/kg for a total of 5 doses, about 1/4 had severe (Sarnat 3) encephalopathy, the reminder being moderate.

Erythropoietin is effective in many animal models of HIE, so why did it not work in this trial? At first I thought that maybe it was given too late, but the authors suggest that maybe it was given too early! They put it this way “Other possible explanations for our negative findings include toxic effects of erythropoietin administration early in the injury cascade when combined with therapeutic hypothermia; suboptimal dosage or timing of administration, because later doses may be most effective; and differences in injury mechanisms between preclinical models of hypoxic–ischemic encephalopathy and human hypoxic–ischemic encephalopathy” In the first of the 2 animal studies they refer to in support of their statement about timing, in rats with cerebral artery occlusion the Epo was given starting a week after the injury, and improved outcomes, in the other, in mice with an MCA occlusion, Epo was given starting 3 days after injury and improved outcomes. But I haven’t seen anything to suggest that Epo only works if given late, but not if given early (I am by no means an expert in this literature). Indeed the lamb studies of Alistair Gunn show efficacy of Epo given starting 30 minutes after the injury. His studies, however, suggest that there is no additional benefit if given in combination with therapeutic hypothermia (Wassink G, et al. Recombinant erythropoietin does not augment hypothermic white matter protection after global cerebral ischaemia in near-term fetal sheep. Brain Commun. 2021;3(3):fcab172). Indeed there may be adverse effects of the combination (Dhillon SK, et al. Adverse neural effects of delayed, intermittent treatment with rEPO after asphyxia in preterm fetal sheep. J Physiol. 2021;599(14):3593-609).

It remains possible that if you don’t have access to therapeutic hypothermia, that Epo may have a role, and a much smaller RCT from northern India, (n=100) did suggest a benefit among non-cooled infants. (Malla RR, et al. Erythropoietin monotherapy in perinatal asphyxia with moderate to severe encephalopathy: a randomized placebo-controlled trial. J Perinatol. 2017).

The authors report that the overall incidence of serious adverse events was higher in the Epo group than the controls, but the definition of what is an SAE compared to an adverse event is rather subjective, so hypertension, is considered an SAE, whereas convulsions (which were a bit less common with Epo) are considered a non-serious AE, if you exchange those definitions then there is no “significant” increase in SAE. I know you should not redefine things after seeing the results, but lumping all the very different adverse events which they consider serious together seems to be already quite questionable. Individually, there is nothing which looks different to random variation. Which is to say that Epo is probably safe, but also very likely ineffective when used like this for this indication.

There are other trials, or at least 1 other trial in Australia (PAEAN), which seems to have completed enrolment but has not yet reported outcomes, so it is possible that this will change, but it seems unlikely as there isn’t a hint of an advantage of Epo in the HEAL trial. The PAEAN trial also is in babies under hypothermia, and also plans to randomize prior to 24 hours of age.

The outcomes of babies with HIE after hypothermia remain problematic, there are many infants with long term motor or intellectual difficulties, and further research will be essential. What should be the next trials? There are of course a few already being performed, one promising therapy being studied in a multicentre European trial is allopurinol, in the ALBINO trial, with a sample size of over 800 it will be the largest trial yet of HIE; large trials are needed for this condition, as the chance of seeing an impact as great as the impact of hypothermia are small, but moderate improvements in outcomes, needing large sample sizes to show them with certainty, could be very important. There are a couple of small trials of melatonin, which suggest a possible impact, and a current multicentre trial in Italy has a sample size of only 100, which is underpowered for all except dramatic effects.

This figure from a 2015 review article suggested some other possibilities

The most easily investigated of those possibilities are the same ones that I wrote about 10 years ago! I still think that prophylactic high-dose phenobarbitone is worth investigating, with the same rationale, a small RCT pre the cooling era showed benefit, there are theoretical and some pre-clinical data suggesting a role also.

But for now, there is no proven therapy that improves outcomes beyond the benefits, important and real but limited, of therapeutic hypothermia.

Posted in Neonatal Research | Tagged , , , | 4 Comments

What is happening for the most preterm deliveries.The Completed Week Heuristic.

Posted on 17 June 2022 by Keith Barrington

There are several recent publications about deliveries at 22 to 25 weeks gestation, evaluating the frequency of intensive care support of the babies, factors associated with support, and survival.

The California Perinatal Quality Care Collaborative reports the proportion of “active treatment of infants born at 22-25 weeks of gestation in California, 2011-2018” which also happens to be the title. (Chen X, et al. The Journal of Pediatrics. 2022) it shows that during those years, the frequency of active treatment of babies thought to be 23 completed weeks of gestation increased, a lot.

I was a bit surprised by the lack of change at 22 weeks, but there are only about 120 per year, which is why the intervals are so wide, and if you ignore 2012, it looks like there is probably an upward trend; perhaps more recent years will show a change. The CPQCC data collection hasn’t changed significantly during those years, with over 90% of California NICUs contributing, and over 97% of very preterm births included, so the data are very reliable for regional trends.

Another intriguing graph is this one

It is interesting, and understandable, that the rate of active treatment increases as the week goes on, from 22 weeks 0 days to 22 and 6, and the same during week 23. What makes no sense, however, scientifically, is the fact that this is not a smooth curve, the difference between 22+5 and 22+6, is so much smaller than the jump from 22+6 to 23+0.

With all the inaccuracies in gestational age dating, the big increases in active care at midnight on day 22+6, and again at midnight on day 23+6 make no sense, and are clear evidence of our preconceptions, I think I will call this the “completed week heuristic”: 22 weeks and 6 days sounds much less mature than 23 weeks, and again 23 weeks and 6 days sounds a lot less than 24 weeks! I hypothesized in a previous post that if human beings had decided a few millennia ago to divide our days into blocks of 10 days rather than 7, our resuscitation patterns would be very different.

Also interesting in this study is that the proportions of babies receiving active intensive care was higher in level 2 NICUs than level 3, which were again higher than level 4 NICUs. The data were analyzed by the hospital of birth, and many level 3 NICUs are associated with a perinatal service, and provide all the care that a level 4 provides except for surgery. It is hard for me to explain why the odds of having active treatment are 34% higher at a hospital with a level 3 NICU at 22 and 23 weeks, and are still 43% higher at 24 and 25 weeks, than a hospital with a level 4 NICU.

The even higher odds of receiving active treatment in a hospital with a level 2 NICU (about twice the odds) are discussed by the authors in the publication, they think that many such babies are delivered rapidly after arrival of the mother, when they do not have time to safely transfer her. As such, there is also less time to counsel the parents, so they may default to initiating intensive care and then transferring.

And what about survival? Well a new publication could have been helpful, but unfortunately ends up being very unhelpful. (Doshi H, et al. National Trends in Survival and Short-Term Outcomes of Periviable Births </=24 Weeks Gestation in the United States, 2009 to 2018. Am J Perinatol. 2022). It analyses data from an administrative database, and reports that there were 71,854 live births in the USA during the years noted in the title, but then ignores all the babies who die at less than 24 hours of life (34,251); I calculated, from the data they present, that the deaths at less than 24 hours were 31,850 babies of 22 or 23 weeks gestation, and about 2,400 babies of 24 weeks. The graph below, like all the others, only includes babies who survived the first 24 hours.

As they give no information about the trends of <24 hour deaths, we don’t know the survival trends among live born babies. Even more annoying, all the data about complications are given as “death or…”, the final figure shows the trends in “death or tracheostomy”, “death or VP shunt” etc. Most of the graphs look just about identical, because death is so much more common than the associated complication, and, as there is no graph for death in that figure, it is not at all clear what the results are for “VP shunt among survivors”, or “tracheostomy among survivors”, which would have been useful and helpful. “Death or BPD” is the one graph which is different, with a much higher percentage of adverse outcomes than the other outcomes, but still, without knowing what the overall mortality is by year, the reader cannot tell if the “BPD among survivors” is increasing or decreasing.

The last of these database analyses (Perry MF, et al. Factors Associated with Receiving No Maternal or Neonatal Interventions among Periviable Deliveries. Am J Perinatol. 2022), looked at the use of antenatal steroids, cesarean delivery and transfer to a tertiary hospital, among mothers who delivered at 22 or 23 completed weeks gestation in the USA. They also examined active interventions for the babies (neonatal intensive care unit admission, surfactant administration, antibiotic administration, or assisted ventilation). The results are somewhat concerning, and reflect, I think, some of the problems in the US medical system, but I also think reflect some cultural variations. Non-hispanic black mothers were significantly less likely to have interventions, which is presented, strangely, as the Odds ratio of having no interventions, which was 1.16 at 22 weeks and 1.21 at 23 weeks. Once born, black babies were less likely to receive no interventions, OR 0.75 at 22 and 23 weeks. SGA babies were more likely to have no interventions and families with private health insurance were also more likely to have no interventions.

When I think of our local practice, I think we also see cultural variations in choices made by and with families from different backgrounds. Parents with some cultural histories are more likely to choose active intervention in the “grey zone” than others, and my guess is, that if we did an objective analysis, we would also see a trend that families with more resources are less likely to choose active intervention. We obviously don’t have families with and without health insurance, this is Canada, we aren’t savages! The discrepancy between maternal and neonatal interventions in this new publication does raise major questions. As an analysis of an enormous administrative database, there are many of those questions that cannot be answered from this data, but they do warrant consideration.

The best outcomes for these babies come when obstetrics works closely with neonatology to co-ordinate active interventions, and ensure that babies who will receive active NICU care are born in the best possible condition. The potential for good survival and excellent long term quality of life is now such that I think I will no longer use the term “periviable”, I am not sure any more what that means, or how to define it.

Posted in Neonatal Research | Tagged , | 1 Comment

The list of shame, and the continued shame of the Journal “Pain”

Posted on 9 June 2022 by Keith Barrington

Randomized controlled trials of pain control measures prior to planned painful procedures that have an untreated control group are unethical.

I would have thought in 2022 that statement was about as uncontroversial as a statement about ethics could be, nevertheless such studies continue to be performed and published.

In one of my recent posts I wrote about an article in the journal “Pain” where a group of babies were randomized to experience pain. I wrote to the editor about my concerns and was asked to write a “letter to the editor”, which I did.

The text of the letter is here:

The article, by Gao et al [3], Effect of combined pharmacological, behavioral, and physical interventions for procedural pain on salivary cortisol and neurobehavioral development in preterm infants: a randomized controlled trial,” published in your journal describes a research project which is clearly unethical[1] and should not have been published.

In this study, newborn babies were assigned to a group designed to experience repeated painful interventions. The express purpose of the design was to create a group of newborn infants who had repeated, avoidable pain. The control group had a mean of 35 painful procedures per baby (SD 15), with mean PIPP scores indicating moderate to severe pain. Pain which could easily have been dramatically reduced or completely avoided.

This research is clearly in conflict with the declaration of Helsinki[6], which notes that untreated controls should only be included in a trial “in the absence of existing proven therapy”, whereas there is an abundance of existing proven therapies for the procedures studied in this trial, from randomized controlled trials involving many thousands of newborn infants. Effective methods to prevent pain caused by skin-breaking procedures are well known, easily available, and cheap or free. Those methods include kangaroo care/skin-to-skin contact, oral sucrose or glucose solutions, especially when combined with non-nutritive sucking, and breastfeeding. There is no valid reason for denying such pain reduction methods to research subjects. Publication in a high-quality journal such as Pain gives credibility to the research and suggests that it is acceptable to inflict pain on babies in order to complete a research project.

Research which compares an analgesic intervention for a painful procedure in newborn infants to an untreated control group is useless in improving care. As effective pain measures are already well known, the only research which could possibly improve care is that which compares different analgesic interventions, or examines the addition of measures to those already known to be effective. It is already clear that repeated painful procedures, as studied in this trial, have adverse effects on stress, neurological and developmental outcomes[5]. It is further already known that routine analgesic interventions mitigate these adverse effects; as a result, national[2] and international[4] guidelines require that analgesia is used for all painful procedures in newborn infants.

The most effective way your journal could improve pain control in newborn infants would be to cease publishing research which unethically randomizes babies to have avoidable pain. All future trials in newborn infants undergoing planned painful procedures that the journal publishes should ensure that all the babies receive proven effective methods of pain control.

I urge you to consider withdrawing this article, and to establish editorial guidelines which prevent the publication of research which imposes avoidable pain on the participants. Specifically, studies comparing pain responses in newborn infants which include an untreated control group should never be published.

The letter was sent out to referees, I guess the same ones who refereed the initial publication, I don’t know as it was all anonymous, referee numbers 1 and 2 had similar points in responses to my letter: They both made the point that the procedures were required for clinical care, that was not in dispute, and has no relevance to my arguments. Referee 2 made the following point:

The article by Gao and colleagues state that it is routine in the unit where the study took place for pain-relieving interventions to be administered once a baby starts crying rather than before. If this RCT had not been conducted, then all the babies in this study would have received this form of care.

Similar comments were made by referee 1, in other words, they are saying that the fact that pain is so badly handled, in this NICU and in many others, is sufficient justification for not providing effective analgesia in the trial. Waiting for the baby to demonstrate that they are in pain, and then trying to soothe them afterwards, is, according to these reviewers, acceptable as a method of pain control. There is, of course, a huge literature showing that pre-emptive pain control, prior to painful procedures, is much more effective than after the fact; waiting for the baby to be in pain, then trying to calm them down afterwards, is completely inadequate as an approach to painful interventions. Just because the NICU does not currently follow adequate pain control standards is not a good reason for publishing a trial comparing their inadequate usual approach to something else.

One of the referees of my letter also notes that it has been stated that neonatal providers in China need a high-quality, evidence-based guideline for the treatment and management of neonatal pain. Which may well be true, but that does not justify performing research on Chinese babies which inflicts avoidable pain. There is already enough published data about pain responses without hurting Chinese babies too.

As you might guess from this post, the editor rejected my letter, which I think shows a certain cowardice, they should at least let the readers consider the arguments for themselves.

One of the referees of the letter actually gives references to 2 articles and a registered protocol which come from a single group. In all of those references there are babies receiving planned painful skin breaking procedures without analgesia, which leads me to guess that there is some auto-justification going on here, and that perhaps this referee has been performing similar unethical studies. When I looked at one of those studies, published as correspondence in a non-paediatric journal (Gursul D, et al. Stroking modulates noxious-evoked brain activity in human infants. Curr Biol. 2018 Dec 17;28(24):R1380-R1381), I was appalled to find that the first of 2 studies, briefly reported together, inflicted pain for no valid reason at all. They took 30 full term infants and gave them pain using a pinprick stimulator and measured the EEG responses, showing noxious related EEG changes in the control babies who received no analgesic intervention. The principal investigator of that study has published articles suggesting that behavioural pain responses are not very important, and that we should use EEG noxious changes as the method to detect pain in newborns. I disagree with that contention, but even if you accept that argument, how does one justify creating pain in otherwise healthy normal babies by sticking needles into them, and “proving” that they have pain using an EEG, without any clinical justification for the pain? This is one of the least ethical studies I have seen in recent years in newborn infants. This part of the study was a comparison between pain from pinpricks with either no analgesic intervention or stroking the leg with a brush, in a specific fashion thought to stimulate C-fibres. The heels of the babies were pricked 27 times in 3 blocks, which appears to be in order to average out the EEG background changes so the the noxious related waves can be seen. I have to admit that this seems to be a relatively mild painful stimulus, the babies usually pulled their legs away, but in similar previous studies this group has shown that the pain they administer with this system is not enough to increase PIPP scores, even though they see EEG changes and the baby withdraws their leg.

In the second part of that letter (Gursul et al) they report a study which appears to be not randomized (it is so poorly reported that it is not clear how the babies were divided into intervention and control groups, at one point they say the controls were “age-matched” but the intervention group was 5 days old and the controls were 2.5 days), the babies had a heel poke for some clinically required blood work, and the (either14 or 16) controls had no analgesic intervention, the 20 in the intervention group had the brushing intervention thought to stimulate C-fibres, which, in 16 of them, decreased the EEG changes and shortened the duration of facial grimacing (they did not do formal assessment on a pain scale). The controls were therefore assigned to a group designed to have pain, with no pre-emptive analgesia, and without even any evidence that they had calming procedures after the heel poke when they were crying.

Other new publications to add to the list of shame include Talebi M, et al. The effect of concurrent use of swaddle and sucrose on the intensity of pain during venous blood sampling in neonate: a clinical trial study. BMC Pediatr. 2022;22(1):263, which studied 4 groups, swaddling, sucrose, combined sucrose and swaddling and nothing at all. Not surprisingly, nothing at all was worse than any of the other interventions, and that control group was completely unnecessary and subjected to avoidable pain.

Another study shows that you can do such research without an untreated control group, and have useful results. They compared breastfeeding to 20% glucose to non-nutritive sucking prior to a clinically required heel-poke, and did not have an untreated group. (Napiórkowska-Orkisz M, et al. Evaluation of Methods to Minimize Pain in Newborns during Capillary Blood Sampling for Screening: A Randomized Clinical Trial. Int J Environ Res Public Health. 2022;19(2):870), I would argue that NNS without glucose is already known to have a somewhat limited efficacy, but this smallish study (n=30 per group) did not show much difference between groups; as a very positive note, they mention in the introduction that they did not have an untreated control group for ethical reasons.

A new systematic review has examined the role of NNS in combination with sucrose, and shows that there is a consistently better control of pain with the combination than with either alone. (Li Q, et al. Efficacy and safety of combined oral sucrose and nonnutritive sucking in pain management for infants: A systematic review and meta-analysis. PLoS One. 2022;17(5):e0268033)

Most of these studies used the PIPP score, if you take out the Mandee study, and O’Sullivan, which used different scales, and should not be analyzed together, the results are similar, with better pain control during heel sticks, and slightly better pain scores during RoP screening, but as you can see from the mean scores, RoP screening still leads to high PIPP scores (means of 12 to 17) even with the combined intervention.

This review also compared combined NNS and sucrose to breastmilk or breastfeeding, but found very little data, 1 study of heelstick comparing NNS and sucrose to breast feeding and 1 of venepuncture comparing the combination to expressed breast milk, both showing similar pain control.

This review supports the idea that the optimal analgesia, prior to heel pokes, if the mother is not present, is a combination of sucrose and a soother; breast milk might be as effective, but there is less data. If mum is there, then breastfeeding for babies when that is appropriate, or skin to skin care, may be about as effective, and if dad is present then skin to skin care by him might also work, but I am having difficulty finding data about that.

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

Bronchopulmonary Dysplasia; does it matter?

Posted on 7 June 2022 by Keith Barrington

I just published a “different view” article addressing the question in the title, with a group of expert collaborators, Mai Luu who is our follow up doc in Montreal, and helps to run the Canadian Neonatal Follow-Up Network (CNFUN), Paige Church, who organizes the neonatal follow-up in Toronto, and Peter Davis, who needs no introduction.

Barrington KJ, et al. Respiratory outcomes in preterm babies, is bronchopulmonary dysplasia important? Acta Paediatr. 2022.

I was stimulated to write the article when reviewing publications about postnatal steroid use, we were trying to develop a common approach for our NICU, and be as evidence-based as possible. During the process I realised, what I guess I already knew, that there is no evidence from any of the trials that long term pulmonary health is improved by steroid use. Even though there are many trials showing a reduction in BPD. Regardless of how it is defined; all the various definitions of BPD are based on abnormal oxygenation, either at 28 days of life or at 36 weeks PMA, sometimes requiring x-ray changes. Our contention in the article is that such short term changes in gas exchange do not correlate very strongly with long term respiratory morbidity. In addition, it probably isn’t very important to families if a baby is on oxygen at 36 weeks PMA, if they go home with little support and few respiratory signs, and have minor respiratory problems in the first year or so of life.

You may remember that BPD was first described, and named, by Bill Northway (a radiologist) and colleagues in a seminal paper in the NEJM (Northway WH, Jr., et al. Pulmonary disease following respirator therapy of hyaline-membrane disease. Bronchopulmonary dysplasia. N Engl J Med. 1967;276(7):357-68.) which you can find easily on-line, along with the description of stages (which we rarely use now) and pathology specimens. This was entirely a clinical description of a new phenomenon among babies who had been supported with the new innovation, assisted mechanical ventilation! PEEP was not yet in use, so babies were being ventilated as a last resort to prevent death, using high peak pressures and zero PEEP, and oxygen concentrations were kept extremely high for long periods, as it wasn’t possible to measure oxygenation continuously, or even very frequently, because arterial blood gas results took a while.

The various newer definitions of BPD, have usually been developed because it was realized that many babies who satisfy the contemporary definition (initially, persistent oxygen needs up to a completely arbitrary 28 days of life) had little long term respiratory difficulty. Andy Shennan and colleagues in Toronto (including Michael Dunn, who must have been very young at the time) compared long term outcomes to the neonatal evolution of lung disease, and realized that adverse long term outcomes were much better predicted by oxygen need at 36 weeks Post-Menstrual Age. In other words they were acknowledging from the start that it is the long term respiratory morbidity that is important, and that the definition of BPD is important as an interim, surrogate, outcome. What they used as their long term outcome of interest is any one the following up to 2 years of age: death not due to congenital anomaly; O2 at 40 weeks PMA; respiratory surgery; 2 or more respiratory re-hospitalisations; recurrent wheezing requiring medication; or, X-ray changes or persistent wheezing/retractions/tachypnoea with growth failure, hypotonia, neurological impairment or developmental delay. Which, interestingly, includes neurodevelopmental outcomes in the respiratory morbidity definition. Among the group of babies they studied, the prevalence of that outcome was just under 20%, and the PPV for such an outcome, of needing oxygen at 36 weeks, was 63% with a NPV of 90%.

Other proposed definitions of BPD have also used different aspects of long term respiratory morbidity as their standard for evaluating their definition. The NICHD definition, which introduced grades of BPD, looked at re-hospitalisations in one analysis, and chronic respiratory medication use in another. They showed that having moderate or severe BPD was, statistically speaking, a predictor of both re-hospitalisation and chronic respiratory medication use, but if you look at the data, the PPV for either of those outcomes was only about 40%, and the NPV about 70%. In other words, many babies without BPD have chronic respiratory morbidity, and many babies with BPD, even those with severe BPD, do not, at least using these very broad simplistic descriptions of respiratory morbidity.

I think that our reliance on “BPD” as an outcome could well be having serious adverse impacts on our ability to improve long-term respiratory function among preterm infants.

If we take one example, the SUPPORT trial, the primary outcome of “death or BPD” was just about identical between groups (48% vs 51%) and BPD alone was 39% vs 41%. In marked contrast, all the long term respiratory outcomes, mostly things which probably matter to infants and their families, were better in the CPAP group. The CPAP babies had “fewer episodes of wheezing without a cold (28.9% vs 36.5%; P < .05), respiratory illnesses diagnosed by a doctor (47.7% vs 55.2%; P < .05), and physician or emergency room visits for breathing problems (68.0% vs 72.9%; P < .05) by 18-22 months CA”. Respiratory re-hospitalisations were also less frequent, 43% vs 50%.

According to these outcomes, CPAP is clearly preferable to immediate intubation for surfactant, even though BPD is not improved. In contrast, many studies of postnatal steroids (Yeh et al, Kothadia et al, UK dexamethasone trial….) have shown less BPD, but all those that have published long-term respiratory morbidity data show no benefit of postnatal steroids. My explanation of that is that steroids reduce inflammation and, as a result, improve gas exchange for a period, but they also have adverse effects on pulmonary growth and development. Of course, many of the steroid trials are relatively unreliable for long term outcomes because of high rates of treatment of controls with steroids, but that does hold true for the trials with low rates of control steroid administration (Such as Yeh et al and Kothadia et al). If we focus on improving long-term respiratory morbidity, then there is little indication for postnatal steroids, infants at risk of dying from their lung disease may be saved by postnatal steroids, but giving steroids to reduce FiO2 or to assist in extubation might be successful for those outcomes, but probably won’t improve pulmonary outcomes in the long term.

We should think of BPD as a short term, interim, surrogate outcome, a way of classifying infants and evaluating lung injury in the short term, but with poor predictive value for respiratory morbidity in the longer term. Much like brain injury, and neurological and developmental outcomes compared to head ultrasound findings; we know that head ultrasound abnormalities are statistically related to longer term neurodevelopmental outcomes, but in order to decide which therapies are likely to improve outcomes we do not rely on ultrasound findings, or even MRI, objective evaluation of longer term function is far more important. We cannot, of course, decide after the fact for each individual trial what long term outcomes are valuable, that would lead to a major risk of bias, and post hoc selection of features of respiratory morbidity which are consistent with one’s prejudices. What long term respiratory morbidities should we evaluate?

When I looked at what features of long term respiratory morbidity have been reported in various trials, it is extremely variable, in addition to the factors already discussed, as outcomes in the Shennan et al paper, and in the NICHD definition of BPD paper, and in the respiratory outcomes of SUPPORT, other articles have presented chronic use of inhaled respiratory medications, post-discharge use of systemic or inhaled respiratory medications, emergency room visits, more than 2 re-hospitalisations, cough without other signs of Upper Respiratory Infection, still requiring oxygen at 2 years after discharge, tracheostomy, and so on. Feeding difficulties, and many other outcomes which are important to families according to recent publications, were never reported.

In the article we plead for the development of long-term respiratory morbidity measures which reflect morbidity of importance to babies and families, and the use of such outcomes as the primary measure of efficacy of neonatal respiratory interventions. There have been one or two attempts to create a definition; one definition from the prematurity and respiratory outcomes study created the following definition “one of the following, hospitalization for respiratory indication, home respiratory support, respiratory medication administration, or respiratory symptoms, being present on at least two of the four 3-monthly questionnaires completed by families during the first year of life”. As far as I can tell the definition was arbitrarily created by the investigators; it was used as a dichotomous outcome. PROP showed a statistically significant association between a diagnosis of BPD and the presence of their outcome, that they called Post-Prematurity Respiratory Disease, PRD. But the predictive ability of a diagnosis of BPD or of severe BPD was poor. PPV of BPD for PRD was only 50%, and the NPV was only 30%. The PPV of severe BPD for PRD was only 37%, and the NPV was 83%.

Again, in other words, many babies with a diagnosis of BPD did not have PRD, and many babies without a diagnosis of BPD, nevertheless had this degree of post-discharge respiratory morbidity.

A more complete definition, which gives a score on an ordinal scale has been developed in the Netherlands for children aged 4 to 8 years, (Meijer-Schaap L, et al. Development and construct validation of a parent-proxy quality of life instrument in children with bronchopulmonary dysplasia aged 4-8 years old. Qual Life Res. 2019;28(2):523-33) it includes 33 items, in 5 domains, respiratory symptoms, school functioning, growth and nutrition, exercise and locomotion, and emotional functioning and health care concerns. It has the great advantage of being developed from items proposed by a multidisciplinary group including parents.

There is a real need for a respiratory outcomes scale developed for use in the first couple of years of life, it should be an ordinal scale, reflecting the range of respiratory morbidities among former very preterm infants, it should be simple to administer, and, most importantly, include items that parents report having an impact on the family.

Posted in Neonatal Research | Tagged , | 1 Comment

Preventing desaturation during intubation. Shine on, you crazy….

Posted on 26 May 2022 by Keith Barrington

Recommendations for older children and adults during endotracheal intubation frequently include the use of free flow oxygen, indeed when I was a fellow with Neil Finer we routinely provided additional free flow oxygen from a catheter placed near the nose, a practice based on data from older subjects. As practices changed with the introduction of routine premedication I didn’t continue the practice, reasoning that intubation with premeds was a lot faster, and that the babies were not breathing anyway. This was probably a mistake. Apnoeic oxygenation is something that is well described in various situations, and I have used it to prevent desaturation during apnoea tests for brain death in the PICU. At the same time the development of readily available high-flow nasal cannulae makes provision of a free gas flow during intubation easier, and probably more effective than just holding an oxygen tube near the baby’s nose.

Studies in adults and older kids have shown that having high-flow cannulae in place during intubation delays desaturation, which is a very common reason for terminating an intubation attempt, and leads to reduction in the number of attempts prior to successful intubation.

Not surprisingly it is the Melbourne group who have taken this to a clinical trial in newborn infants. (Hodgson KA, et al. Nasal High-Flow Therapy during Neonatal Endotracheal Intubation. N Engl J Med. 2022;386(17):1627-37), the Stabilisation with Highflow for Intubation of NEonates, SHINE trial. They randomized 251 intubations (in 202 infants) to either have high-flow nasal cannulae (HFNC) placed prior to intubation, or not. Babies of all gestational ages were eligible, and intubations in the delivery room or the NICU were included, if the intubation was not an emergency.

HFNC were used at a flow rate of 8 litres for all the infants, and used the same oxygen concentration that the baby was already receiving, with the opportunity to increase to 100% if the SpO2 fell below 90%. The primary outcome variable was success on the first attempt, without instability, which was defined as hypoxia (>20% decrease in SpO2) or bradycardia (<100 bpm). Three quarters of the intubations were in the NICU, and most were pre-medicated, with the same protocol that we use in my centre, and which is recommended in several guidelines, atropine, succinylcholine and fentanyl.

As you can see from the primary outcomes, there was a much higher rate of success on first attempt without desaturation or bradycardia, which appears to be mostly because of less desaturation, and there was not any significant occurrence of hyperoxia. You can also see that the biggest gain was among those with less experience, with a dramatic improvement in success, but there is probably an improvement among those with more experience also.

If you have access to the NEJM you can watch a video of a real-life intubation with HFNC, but it really is quite simple, as the image below shows.

There doesn’t to me seem to be any good reason to not implement this widely. There were unfortunately few intubations with a video laryngoscope (only about 8%), which other data show improve success, especially with inexperienced operators. But the two approaches are not in any way exclusive, using HFNC and a video laryngoscope might be the best possible approach, especially for inexperienced intubators, which is now the majority of paediatric trainees! Now that we intubate very few meconium stained babies, we try to maintain as many babies as possible on non-invasive ventilation, and we have a tiny baby intubation team, there are few opportunities for trainees to intubate real babies. Simulation training is vital in providing initial training, but making intubation safer for real babies when it becomes necessary, and ensuring that paediatricians in training develop the expertise needed are all important.

This new publication, and the technique that it shows is safe, effective, and simple, is a big advance for neonatal intensive care.

Posted in Neonatal Research | Tagged , , | 11 Comments

World NEC awareness day, part 2

Posted on 17 May 2022 by Keith Barrington

As well as all the high tech mechanistic studies in NEC, such as those that I just posted about, sometimes we need to take a step back and ask some very basic questions. What is the current natural history of NEC? How good are we at diagnosing NEC?

So much has changed in neonatology since NEC was first described, that this new study from the Pediatrix database is timely, (Mara K, et al. Necrotizing Enterocolitis in Very Low Birth Weight Neonates: A Natural History Study. Am J Perinatol. 2022). It includes data from babies born 23 to 29 weeks GA who had a birthweight less than 1500g. The babies were divided into those with suspected, medical or surgical NEC. It shows a peak in confirmed NEC incidence around 2007, with a fairly marked decrease since then, being relatively stable since 2015. There has also been a switch in antibiotics used, with much more Piperacillin/Tazobactam, much less 3rd generation cephalosporins, and a drop in vancomycin use over the last few years. There has also been a switch from clindamycin to metronidazole for anaerobic coverage.

There is a huge amount of data in this publication, so I will just mention a few highlights. Risk factors for development of NEC are confirmed as being lower GA, male sex, small for gestational age and being outborn. Mortality is elevated, and is extremely high for surgical NEC: 6% for suspected NEC, 16% for medical NEC and 43% for surgical NEC. Positive blood cultures within 7 days of diagnosis were positive in 20%, and among surgical NEC were positive in 40%. Postnatal growth restriction was common among these babies, especially the surviving surgical NEC infants.

In addition to the obvious human costs of NEC the calculated additional financial costs are enormous. Length of stay was 26 days longer for confirmed NEC babies, and just on that basis average increased hospital charges are about $200,000 per case. Which can be calculated for the whole of the USA to be nearly half a billion dollars a year. That estimate does not include post-discharge costs, which are also enormously increased for NEC babies. Any intervention which has a reasonable cost and which decreases NEC will save money, as well as lives. Another good reason for funding NEC research!

The study has the advantages and limitations of a large database, meaning that individual diagnoses were not verified (of course) and we have to assume that these were all cases of NEC. That assumption is challenged somewhat by this meticulous individual examination of cases; BErrington J, Embleton ND. Discriminating necrotising enterocolitis and focal intestinal perforation. Arch Dis Child Fetal Neonatal Ed. 2022;107(3):336-9. By examining in detail the case records of babies in their local database, it was clear that differentiating the 2 phenomena is difficult, there is overlap in age of presentation, clinical signs at presentation, complications and outcomes. There is even confusion when the histology is examined. Only by really carefully examining each individual case were they able to come up with a final decision, and I am sure they would agree that other people examining the same cases might sometimes have different conclusions! Spontaneous intestinal perforation, with the typical anti-mesenteric perforation in the absence of skip necrotic lesions, has some aetiologic characteristics in common with NEC, but I think it is unlikely that interventions to prevent NEC will also prevent SIP.

The long term outcome impacts of NEC are well known, the data about SIP outcomes are more inconsistent, but there do appear to be substantial adverse impacts on neurological outcomes and on developmental progress. There are also major impacts of NEC on feeding and nutritional outcomes, and associations with lung injury and retinopathy and behavioural problems.

On this day we need more than just awareness, a co-ordinated effort to reduce the incidence and the impact of NEC, and SIP, will improve the future lives of thousands of babies and their families.

Posted in Neonatal Research | Tagged , | 1 Comment

World NEC awareness day.

Posted on 17 May 2022 by Keith Barrington

As we approach world NEC awareness day (May 17th) I thought I’d do a quick PubMed search to see if I’ve missed anything recently, so I started typing “necrotizing” in the search bar, which immediately suggested “necrotizing enterocolitis” as one of the possible completions, so I clicked on that and found there were 9,700 potential hits, ranging from the first reports in premature infants in the 1950’s, the first US series in 1964, and a progressive increase in annual publications to 766 last year. Apart from the odd report in animals or older patients, almost all of these publications are about neonatal human NEC, or animal models of neonatal NEC.

Recent work that caught my eye included the following, a study of the intestinal “virome” in very preterm babies. This was an “-ome” that I hadn’t come across before, but it was fairly obvious what it meant. (Kaelin EA, et al. Longitudinal gut virome analysis identifies specific viral signatures that precede necrotizing enterocolitis onset in preterm infants. Nat Microbiol. 2022;7(5):653-62) The authors identified a number of bacteriophages and what they call “eukaryotic viruses” which stumped me for a minute, as I was stunned that a virus could be a eukaryote; but of course a virus cannot be a eukaryote, the term refers to viruses which infect eukaryotic cells. Such as COVID, or all of the other viruses causing disease in humans. Not too surprisingly, the large majority of viral signatures that they found in the stools of preterm infants were unclassifiable. To be really honest, I don’t understand a lot of the analyses that they performed comparing the sequential viral and bacterial microbiomes of 9 preterm babies who developed NEC and 14, matched for GA and birth weight who did not. So, unusually for me, I will have to take them at their word that they found that “the viromes of infants who developed NEC converged towards a reduced level of beta diversity before NEC ensued and this convergence was characterized by specific viral signatures”. Of course this is a small series form a single hospital, and will need to be confirmed, but it suggests that viruses may have a role in the development of NEC, and specifically some bacteriophages may be implicated.

Other evidence that examines how we mess up the microbiome having a significant impact on NEC comes from studies like this one in neonatal mice (Chaaban H, et al. Early Antibiotic Exposure Alters Intestinal Development and Increases Susceptibility to Necrotizing Enterocolitis: A Mechanistic Study. Microorganisms. 2022;10(3)) half of whom were randomized to receive antibiotics (amp and gent) for 10 days. 4 days later they all received oral Klebsiella pneumoniae or an oral vehicle. Just getting antibiotics for 10 days severely impacted on gut development with fairly dramatic effects on villi length and crypt depth.

Goblet cells and Paneth cells were also impacted, as were intestinal permeability which was dramatically increased, and the antibiotics, not surprisingly, had major effects on the intestinal microbiome. When they got both antibiotics and then later Klebsiella, intestinal permeability was even further increased, TNF alpha and IL-1 beta also shot up, and half of the animals developed an intestinal injury which looks a lot like NEC.

Another fascinating study looks at whether using AI to interpret intestinal microbiome composition could possibly predict NEC (Lin YC, et al. Interpretable prediction of necrotizing enterocolitis from machine learning analysis of premature infant stool microbiota. BMC Bioinformatics. 2022;23(1):104), again I can’t pretend to understand a lot of what they did, but by re-analysing data from 2 prior cohorts of sequential microbiome analysis in preterm babies, among whom a number developed NEC, they were able to train an AI system to predict NEC. The system predicted NEC in the majority of preterm babies at birth, which isn’t very useful, but you could just ignore the first couple of days, and then as the days progressed there was a progressive divergence leading eventually to a reasonably good sensitivity of 86% and specificity of 90% for predicting NEC, which was possible 8 days prior to clinical presentation. This hold out hope that sequential microbiome analysis could eventually prospectively be fed into the machine learning algorithm and give you a weeks notice that a baby is going to get NEC. What you would do about it at that point I have no idea, but maybe you should block the toll-like receptor-4.

There are couple of recent reviews which discuss the pathogenesis of NEC, this one (Hackam DJ, Sodhi CP. Bench to bedside – new insights into the pathogenesis of necrotizing enterocolitis. Nat Rev Gastroenterol Hepatol. 2022. ) concentrates on the role of the TLR4, in the text it is surprisingly dismissive of the importance of microbiome disturbance, which I find difficult to understand. I think it is clear that B infantis interacts with TLR4 and exerts anti-inflammatory effects as a result.

Improving gut colonisation with B infantis, and/or other organisms that reduce inflammation by interacting with TLR4, as well as potentially other mechanisms will probably reduce NEC incidence.

Speaking of which there is a new Cochrane review of synbiotics for the prevention of NEC which found 6 trials. Unfortunately 3 of them used inulin as the prebiotic part of the “syn-“, which is quite probably not the best prebiotic to consider, other oligosaccharides and in particular lacto-N-tetraose, are much more effective at promoting the growth of B infantis. The 3 small trials which used other oligosaccharides were of variable quality, and give no real confidence that the reduction in NEC that they showed is a reliable effect. Ongoing work confirms the probable importance of lacto-N-tetraose, such as this one (Wu RY, et al. Structure-Function Relationships of Human Milk Oligosaccharides on the Intestinal Epithelial Transcriptome in Caco-2 Cells and a Murine Model of Necrotizing Enterocolitis. Mol Nutr Food Res. 2022;66(4):e2100893.) which showed that mice which received one of 3 different human milk oligosaccharides were protected against NEC which occurred in the controls, induced by hypoxia, hyperosmolar feeds and lipopolysaccharide; but that the mechanisms of protection were different between the oligosaccharides. Lacto-N-tetraose appears to have been one of the most effective at reducing inflammation.

There is still so much to learn about this disease, and despite the many advances in neonatology over the years, progress in NEC has been slow; although probiotics reduce the incidence of NEC it still occurs, and still devastates some infants. Determining which probiotic is most effective, how to enhance efficacy with the most effective prebiotic/HMO, and what to do when early signs of NEC or of intestinal dysbiosis occur, are topics that require further investigation.

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

Maternal breast milk is risky too

Posted on 12 May 2022 by Keith Barrington

Human breast milk, when freshly expressed, contains all sorts of goodies, to use the technical scientific term. Many of which are adversely affected by standard pasteurisation (called Holder pasteurisation, which is very similar to what Louis Pasteur himself invented in the 1860’s). Holder pasteurisation entails heating the milk to about 63o Celsius and holding it there for 30 minutes, as far as I can tell, it is called the Holder method because of the process of “holding” the temperature stable for a period, but if that is wrong and someone called Holder should be credited I would like to know. I believe it is the only method used in human milk banks that follow HMBANA standards (or NICE standards or others internationally), even though it is no longer used much in the dairy industry, who use short term high temperature (72o for 15 seconds, or even for 5 seconds) processing instead. Other methods, including Ultra-High Temperature, high pressure processing, ultraviolet, and radiation treatments, can inactivate bacteria and other micro-organisms.

Holder pasteurisation has major effects on large molecules and less effect on small molecules (to over-simplify). So not much impact on oligosaccharides in general, but major decreases in immunoglobulins, various enzymes (e.g. lactoferrin), and larger hormones (adiponectin). Even some smaller molecules, such as insulin, are heat-labile and severely reduced by pasteurisation.

Occasional neonatal sepsis from bacterial organisms found in non-pasteurised maternal breast milk can occur, and even pasteurisation doesn’t kill everything, in particular spore-bearing organisms are resistant, which is why pasteurised milk is not the same as sterilised milk. Indeed contamination of pasteurised donor milk with Bacillus Cereus is a major reason for disposing of donated milk after pasteurisation, as it cannot safely be given to extremely preterm babies. In one recent study, the only culture positive organism after pasteurisation was B. cereus.

All forms of pasteurization, however, inactivate cytomegalovirus, and CMV from unpasteurized maternal milk is the major source of postnatal CMV infection in our patients.

There have been several publications recently about the importance of postnatal CMV acquisition in very and extremely preterm infants, the results of which are somewhat variable. I thought I would try to summarize what these recent publications tell us.

The most recent article is from Melbourne (Bimboese P, et al. Postnatal Cytomegalovirus Infection of Preterm and Very-low-birth-weight Infants Through Maternal Breast Milk: Does It Matter? Pediatr Infect Dis J. 2022;41(4):343-51) they followed very preterm (<32 wk) and/or very low birth weight (<1250g) infants of mothers who were CMV IgG positive. Just under half of the eligible mothers were antibody positive, and they were followed with breast milk and urine PCR for CMV. They eventually pared down the cohort to 58 babies, of 49 mothers, who were exposed to CMV PCR positive breast milk, and then compared the clinical course between those who became urine PCR positive, and those who stayed negative. The data are from around 20 years ago, for some reason, and at that time there was no human milk bank, so the babies received either fresh, refrigerated or frozen maternal breast milk, or formula, all of the babies received at least some fresh maternal BM.

“In total, 30% (8/27) of the CMV-positive infants were asymptomatic, 48% (13/27) mildly symptomatic, and 22% (6/27) severely symptomatic at the time of the first CMV-positive urine sample. Neutropenia was one of the most common presentations at the time of first CMV-positive urine (44%; 12/27), followed by respiratory deterioration (33%, 9/27; apnea (3), new CPAP requirement (1), new intubation (2), increasing oxygen requirement (7)). Nine of 27 infants had a partial or full septic workup”.

A much larger cohort of infants <1500 g birth weight was published in 2020, (Weimer KED, et al. Association of Adverse Hearing, Growth, and Discharge Age Outcomes With Postnatal Cytomegalovirus Infection in Infants With Very Low Birth Weight. JAMA Pediatr. 2020;174(2):133-40) from the Pediatrix group; because it is from a huge database the case definition and the comparison group are not quite as clear-cut, cases were defined by those who had a positive CMV culture or PCR from blood, urine, CSF, or respiratory secretions after 21 days. If they also had a positive test prior to 21 days they were classed as congenital, but if they never had a test before 21 days they were considered to have postnatally acquired CMV (and then analysed as a separate subgroup to see if that changed the conclusions). If a baby never had a CMV PCR or culture they were considered negative. The major outcomes of interest were: a failed hearing screen, length of hospitalisation and growth, with BPD and NEC being important secondary outcomes. Each of the 273 postnatally acquired CMV babies was propensity score matched with a control. Babies with postnatal CMV were much more likely to fail their hearing test, much more likely to develop BPD and there was very little NEC.

Length of stay was longer in the CMV babies, and there were some relatively small effects on weight growth, but not length or head circumference.

Remember that in that study there would have been many babies who never had a CMV pcr, so the overall rate of breast milk CMV transmission cannot be calculated.

A systematic review from 2021 (Park HW, et al. Incidence of Postnatal CMV Infection among Breastfed Preterm Infants: a Systematic Review and Meta-analysis. J Korean Med Sci. 2021;36(12):e84) suggests that among preterm babies of seropositive mothers, who receive breast milk known to be contaminated with the virus, somewhere between 2% and 87% will be infected. That is such a huge range that the data seem to be of little value, but the review includes some very tiny studies with only one or two cases. If you pool all of the studies then approximately 20% of preterm babies who receive CMV positive milk will become infected. It looks from that review, that the groups with more immature babies might have a higher risk, but transmission rates are so variable, that is not certain. However, a small study limited to babies of 22 to 24 weeks gestation (Mehler K, et al. High Rate of Symptomatic Cytomegalovirus Infection in Extremely Low Gestational Age Preterm Infants of 22-24 Weeks’ Gestation after Transmission via Breast Milk. Neonatology. 2013;105(1):27-32) showed a 65% rate of acquisition of CMV in the most immature infants, more than half of whom failed their hearing screen.

It also looks like freezing breast milk does not routinely lead to a lower rate of transmission. I mention freezing as it has been shown to partially inactive the virus, and is used as a routine in some centres for that reason. It has long been known that the impact on CMV inactivation is only partial. One RCT, despite limited power, showed that freezing was probably ineffective to prevent CMV transmission (Omarsdottir S, et al. Cytomegalovirus infection and neonatal outcome in extremely preterm infants after freezing of maternal milk. Pediatr Infect Dis J. 2015;34(5):482-9).

In contrast, pasteurisation of breast milk reliably inactivates CMV, and it doesn’t seem to make any difference how you pasteurise. Even a very short time, 5 seconds at 62O, is partially effective, (Bapistella S, et al. Short-term Pasteurization of Breast Milk to Prevent Postnatal Cytomegalovirus Transmission in Very Preterm Infants. Clin Infect Dis. 2019;69(3):438-44) and in that study they were able to pasteurise the milk daily, starting on day 4 postnatally using a system in the NICU milk kitchen, or even at the bedside. In this study they included infants of mothers who were CMV IgG positive, and then tested breast milk weekly. Breast milk was pasteurised routinely before being given to the babies unless CMV cultures of the milk proved negative. They ended up with 87 VLBW or very preterm <32 wk babies at risk of catching CMV from the breast milk, only 2 actually became CMV positive among the babies who received the short term pasteurised milk, whereas they had a historical control group of 83 babies who received raw milk, of whom 17 became infected. Although unfortunately not a randomized trial, this is certainly suggestive that this simple bedside method of pasteurisation can reduce, but not eliminate breast milk CMV transmission. This method has less effect on some large molecules, such as growth factors, but it probably has a major impact on Bifidobacteria, which are universally present in breast milk, are important for neonatal gut colonization and metabolism of oligosaccharides, and we should try and preserve if possible. I don’t know if there is any way to inactivate CMV without killing Bifidobacteria, but, if not, it warrants an RCT to determine if, in very preterm babies receiving CMV positive breast milk, there is an overall advantage of pasteurization or not.

In more general terms there is already one RCT of routine pasteurization of breast milk, (Cossey V, et al. Pasteurization of Mother’s Own Milk for Preterm Infants Does Not Reduce the Incidence of Late-Onset Sepsis. Neonatology. 2012;103(3):170-6) which I have mentioned a couple of times, and showed a 50% higher incidence of sepsis in infants fed pasteurized milk compared to raw maternal breast milk (possibly a random effect). A similar study, but confined to CMV positive breast milk, and with a sample size powered to determine sepsis, as well as growth outcomes, would be valuable.

In summary, somewhere around a quarter of very preterm babies exposed to CMV positive breast milk become infected, and the most extremely preterm babies are at much higher risk. The acute infection often presents with thrombocytopaenia, and sometimes neutropaenia, a sepsis like syndrome, respiratory deterioration, and apnoea. Postnatal, breast-milk acquired CMV, also appears to substantially increase the risk of failing hearing screening, likely to mean a real increase in hearing impairment. Breast milk acquired CMV also probably increases the risk of BPD, and, with, much less convincing data, perhaps increases retinopathy and NEC. Whether routine screening of breast milk for CMV, and some sort of pasteurization in selected cases is warranted, effective and safe is unknown.

One unanswered question I have, which is one of the reasons that I was reviewing this literature, is what about steroids? I have seen babies have very serious systemic CMV after steroids were given , but I struggle to find data about the impacts of steroid use on CMV infections in the preterm. Are babies who receive steroids and CMV positive breast milk at increased risk of serious CMV infection? Should we modify our decision-making in babies from CMV IgG positive mothers? If anyone knows any data about the issue, please let me know in the comments. A pubmed search using what I thought were the right terms came up with nothing relevant, and when I took “newborn” out of the search I had over 2000 hits!

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

AAP guidance on probiotics. As wrong-headed as a head can be wrong.

Posted on 9 May 2022 by Keith Barrington

The AAP issued guidance on probiotic use in the preterm infant last year in the form of what they call a “clinical report”, which I didn’t comment on at the time, I thought it might be a bit redundant as I have made my views pretty clear over the years. Unfortunately the AAP guidance was, in my view, based on a very limited and biased review of the available literature, comes up with recommendations which are really questionable, and continues to be challenged as more data accumulates.

Nothing in neonatology has been as extensively researched as probiotics. Somewhere over 12,000 preterm infants at risk of NEC have been randomized in trials so far. While it is true that trials are of extremely variable quality, from the poor to the exemplary, the overall quality of the evidence can be described as moderate to good. Of importance is that not a single trial has reported adverse outcomes as a result of probiotic use. Some have been null, some have been positive, but not one has been negative.

The variety of probiotic preparations used is another major problem. Which is why network meta-analyses have been performed (there are at least 4 published), examining the different probiotic formulations. The choices made regarding how the different formulations were grouped can be questioned, but the network meta-analysis which I think is the best quality (Morgan RL, et al. Probiotics Reduce Mortality and Morbidity in Preterm, Low-Birth-Weight Infants: A Systematic Review and Network Meta-analysis of Randomized Trials. Gastroenterology. 2020;159(2):467-80), suggested that a combination of bifidobacterial species and a lactobacillus is optimal, and reduces the risk of NEC by about 50%. The other network meta-analyses have reached similar conclusions. As I previously mentioned, comparing studies that use a B infantis (otherwise known as B longum subsp. infantis) to other organisms would probably also conclude that B. infantis is the most important. Most of the studies using multi-organism combinations have included a B. infantis in the combined preparation, a network meta-analysis comparing treatment with B. infantis, either alone or in combination with other probiotics would therefore, I think, probably conclude that it was the most important of the organisms.

It seems to me that individuals and centres that have made the decision to not use probiotics seek to justify their decisions by invoking the “poor quality” of the evidence, whereas centres that have nevertheless introduced probiotics routinely see a fall in NEC incidence.

Even worse than the AAP statement is the press release that accompanied it, including a statement not found in the guidance document ” The most recent trials have not shown a reduction in NEC in those at highest risk”, which is just not true.

One of the problems with this literature which allows this type of interpretation is that many trials have not presented their data in a stratified fashion. In other words, the data have not been presented for babies above and below 1kg. Even though the mean birth weight of participants is about 1 kg; this allows the AAP to suggest that few babies under 1 kg have been studied, and efficacy in such babies is uncertain.

In reality, about half of the 12000 infants in the studies were <1kg, and there is nowhere any evidence of a differential effect based on body weight, indeed that would be weird. This would be the only intervention that I am aware of that works less well in higher risk babies than in lower risk infants.

In addition, the AAP statement suggests that centres that choose to use probiotics “should discuss the potential risks and benefits of this therapy with parents and should strongly consider a formalized informed consent process”, I will never be someone to suggest that we should not discuss the potential risks and benefits of a therapy with parents, but what risks does the AAP want us to discuss? Among the huge numbers of babies in RCTs, there were no reported risks, indeed an evidence-based discussion of the risks would go something like: “there are a few reported cases of sepsis with probiotic organisms outside of the trials, and there is no sign in any of the trials that the risk of receiving probiotics is more than the risk of not receiving them, all the evidence shows that the risk benefit of probiotic administration is heavily on the side of benefit”.

More importantly, why don’t the AAP suggest an informed consent process for those centres who don’t give probiotics? Surely, the evidence, which shows that giving probiotics containing B. infantis, or a mixture of Bifidobacteria and Lactobacilli, has benefits and minimal risks, should be discussed with every parent in every NICU, not just in those who give them, but even more importantly, in those who don’t. Shouldn’t every parent have the opportunity to ask for a treatment which has only shown benefit in RCTs?

I’m trying to think of a parallel, but I think that all other low risk procedures and moderate risk procedures are universally available, I don’t think there are NICUs that have decided to never give surfactant, or antibiotics, or CPAP. So the only partial parallel I can think of is for extremely high risk procedures, such as ECMO. If you are looking after a sick full term baby with meconium aspiration who reaches ECMO criteria and you are in a centre that doesn’t offer ECMO, would the AAP consider it OK to not discuss transfer to an ECMO centre? That only centres offering ECMO discuss the risks and benefits, and the others should just pretend that ECMO doesn’t exist?

An decision to cannulate for ECMO is of course extremely high risk, and warrants extensive discussion with parents, and if you are in a non-ECMO centre, then the discussion should also involve the risks of the transport, but if it is a reasonable alternative for the individual baby, then that discussion should always take place.

I am not sure if many centres really use a “formalized consent process” for giving antibiotics, or surfactant? Interventions for which the risk/benefit is often much more questionable than for probiotics.

To put it bluntly, the current AAP advice is that parents in centres that do not currently offer probiotic prophylaxis for NEC should have the evidence hidden from them, and there is no obligation for them to be informed. In 2022 that is unlikely to be a successful policy, I am sure that many parents in the USA search the interwebs when their baby is in the NICU, and especially those whose baby develops NEC are likely to find the NEC society web site, whose educational materials clearly state :

“There is also good evidence that giving premature babies probiotics reduces their risk of NEC and increases their chance of survival. Neither human milk nor probiotics can eliminate the risks of NEC”.
And

“Are there any risks of getting probiotics?
There are risks and benefits to every treatment. The benefits of probiotics include maintenance of healthy bacteria in the intestine. This is believed to help prevent NEC. In rare situations, probiotic bacteria can get into the blood and cause infections. If babies develop an infection in the blood with the probiotic bacteria, they are given an antibiotic to kill the probiotic bacteria. When this has happened, the infections have been responsive to treatment. Based on the literature, it appears that the benefits of probiotic administration outweigh the potential risks.”

A new publication of a before and after study in an NICU in Portland Oregon (Tobias J, et al. Bifidobacteriumlongum subsp. infantis EVC001 Administration Is Associated with a Significant Reduction in the Incidence of Necrotizing Enterocolitis in Very Low Birth Weight Infants. J Pediatr. 2022) showed a substantial decrease in NEC after introduction of routine B Infantis supplementation, and an elimination of NEC deaths. As I have previously mentioned, I think it is scandalous that Evivo is being aggressively marketed without the kind of RCT evidence that would be needed if it were a medication, there really is no scientific justification for this. It would not be difficult to perform the sort of trial that is needed to prove that Evivo truly is effective, and also is more effective than alternatives. I believe that it has the manufacturing standards that are required, we just need proof of efficacy from an RCT.

In an excellent editorial accompanying that publication Mark Underwood notes that the AAP statement does not suggest a “formalized consent process” for other interventions that reduce NEC, unpasteurized maternal milk administration and banked human milk use, even though the small risks from those interventions are probably greater than the risks of probiotics. Of course, the risk/benefit ratio of fresh maternal milk, and supplementation with banked donor milk, are clearly in favour of their use, but that is also the case for probiotics.

I think the AAP missed an opportunity to advocate for the development of probiotic preparations that are of sufficiently high quality, with stringent quality control and safety data, they missed the opportunity to advocate for comparative trials, with individual or cluster randomization. There are still many unknowns with probiotics, and trials comparing different strains, and different combinations could have a huge impact on preventing this atrocious disease. I think that trials comparing a B, Infantis alone, to a B. Infantis in combination with a Lactobacillus, and perhaps to a B. Infantis and oligosaccharide combination (preferably with DSLNT), would have a chance of improving care of very preterm infants and reducing the terrible consequences of NEC.

May is NEC awareness month, and May 17th is NEC awareness day. With those dates in mind, consider supporting the NEC Society, an organisation which involves parents and professionals, with the overall goal to create a world without NEC. A worthy goal indeed.

Posted in Neonatal Research | Tagged , , | 3 Comments

Automated oxygen control, what’s taking so long?

Posted on 29 April 2022 by Keith Barrington

A long long time ago, in a galaxy… actually quite near here, the idea of servo control of inspired oxygen was already in the air. At the time I first heard about it, the idea was to control FiO2 based on PO2 derived from electrodes on the tip of an umbilical catheter (paper from the mists of time… 1979), so you can see how old an idea it is! Other systems were developed which used transcutaneous PO2, and although the systems worked, there were the obvious limitations of using TcPO2. Now multiple systems have been tested to control FiO2 based on the pulse oximeter, different algorithms compared, systems that work with high-flow nasal cannulae developed, and even use in the delivery room during resuscitation had been trialled.

What will it take for this to become standard of care?

Most babies receiving oxygen have adjustments to their oxygen supply several times a day, this is usually triggered by alarms from the pulse oximeter, which have an enormously high frequency, one recent study, for example, noted up to more than 300 alarms per day, which leads to alarm fatigue and even important alarms being ignored or with very delayed responses. Babies as a result may spend substantial amounts of time below and above target ranges. In our unit, as in many others, we have a target range of saturation for most babies receiving oxygen between 90 and 94%, but we set the alarms 2% wider than that in order to avoid too many alarms. In addition, our oximeter monitors have an adaptive alarm system, so if they are just 1% outside the alarm limit they don’t ring for 60 seconds, at 2% they ring after 30 seconds, and so on, until they ring immediately if more than 5% outside the limits. Even with these approaches, the alarms are very annoying to parents, nurses, and I dare say babies (and to me!).

What we really need are systems which reduces the number of episodes of hypoxia and hyperoxia, and reduces the duration and severity of over and undersaturation when they occur. I am not sure if the most important thing is to reduce the total duration of hypoxia and hyperoxia, or to reduce the number of episodes of hypoxia and hyperoxia. It may be that saturations rapidly changing up and down are worse than persistently low (or high) saturations, even if the total duration is the same. Re-saturation after hypoxia is a potent source of oxygen free radicals, so my guess is, that for the same total duration of hypoxia, it would be less harmful to have one long episode of mild hypoxia, rather than 100 brief episodes with intermittent re-saturation.

Hyperoxia is bad for your retina and your lungs, and probably your brain also, but is a persistent slightly high saturation worse than a saturation that goes up and down to 99% then back to normal? There might be animal data out there that address the issue, but I don’t know them.

There are a number of complicating factors, in designing automated control systems. Much hypoxia is caused by apnoea, which won’t be affected by giving more oxygen! Post apnoeic hyperoxia is already a real problem, which probably often occurs when caregivers increase oxygen during apnoeic spells (partly because mixed and obstructive apnoeas don’t look like apnoea, so people try increasing the oxygen anyway). An automated system which routinely increased oxygen administration during apnoeas could be problematic. But, with the appropriate safeguards built into the algorithms it should be possible to safely reduce overall oxygen instability, and reduce alarms.

Another fairly trivial problem for an automated system, would be to turn off the high saturation alarm when the baby is in 21% oxygen, and turn it back on again when the oxygen concentration is increased; this is by no means a trivial problem in the NICU currently, babies in variable very low concentrations of oxygen who have a period when they saturate above 96% in 21% oxygen, will currently, appropriately, have the high alarm switched off by caregivers, and then when the baby needs an increase in FiO2 for a temporary desaturation, they often end up in low supplemental oxygen concentrations, with saturations above the desired limits and the high alarms turned off.

Two years ago Carlo Dani reviewed the then available literature, finding 16 trials which compared oxygenation outcomes between standard and automated oxygen control. He showed that overall the proportion of time spent in normoxia was increased, with less time spent hyperoxic, and a smaller impact on hypoxia, with automated control. Since then there have been several other trials published, including one by his own group, but we still don’t know if any clinical outcomes might be improved.

One of the outcomes that always seems improved in the trials, when reported, is the number of manual oxygen adjustments that were required, which is usually dramatically reduced. Alarm frequency is not something that I have seen reported much, but is presumably substantially reduced. One interesting study (Warakomska M, et al. Evaluation of two SpO2 alarm strategies during automated FiO2 control in the NICU: a randomized crossover study. BMC Pediatr. 2019;19(1):142) randomly compared 2 alarm strategies in a cross-over study, among babies who are all on automated oxygen control, they crossed-over babies who had normally a target range of 88-95% saturation, and had either alarm limits set to 87 and 96%, with a 30 second delay, or looser alarms, set 2% wider with a 90 second delay. They only studied 21 babies and started on day 1 of life for up to 6 days, and the alarm strategy crossed over every 24 hours, they used the Avea ventilator for both invasive and non-invasive support and stopped the study when they were off CPAP. Although the alarm limits were wider and slower, and therefore rang much less frequently in the “loose” group, oxygen saturation profiles were identical, being 95% of the time normoxic (88-96% unless in room air, when >96% still considered normoxic). The number of actual alarms is shown below.

.

The holy grail, of course, to use an overused metaphor, would be to show that automated servo controlled FiO2 decreased retinopathy and shortened resolution of lung disease. I guess equipment manufacturers would then charge enormous amounts of money for their systems, so maybe we don’t want that! Just making life easier for families in the NICU by reducing alarms, and making life easier for caregivers (mostly nurses) by reducing alarm fatigue, that should be enough to make automated control standard of care, but it will be hard to find the money to buy the systems unless we can show there are fewer complications.

One of the reasons for writing about this now is this new publication from the Pediatrix group (Srivatsa B, et al. Oxygenation factors associated with retinopathy of prematurity in extremely low birth weight infants. J Pediatr. 2022), in a retrospective analysis of oxygenation changes during ventilatory assistance, which was respiratory support of HFNC of 2 litres or anything more than that (CPAP etc.) whether or not the infants were getting supplemental oxygen. They included 101 ELBW babies who had at least 45 days of data during the first 2 months of life and survived to have an eye exam, from a single NICU, of whom 33 developed mild and 15 had severe RoP (6 needed treatment). They found the babies with severe RoP were exposed to higher FiO2 and had more frequent FiO2 adjustments compared with the group without RoP. Room air hyperoxia (>95%) occurred more frequently in the group without RoP whereas iatrogenic hyperoxia occurred more frequently in the group with severe RoP. Babies with severe ROP had more hypoxic episodes and a longer time spent in severe hypoxia (<80%) than those without ROP. You would hope that reducing those fluctuations would reduce RoP, but of course, you can’t be sure of that from an uncontrolled observational study like this.

A recent pre-post study (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) was unable, however, to show a benefit of automated oxygen adjustments, they analyzed cohorts of 24 to 29 week gestation babies over about 3 years before and 3 years after routine use of automated oxygen adjustment. The babies in the cohorts look quite similar in terms of risk factors, during the after period, duration of intubation and CPAP was less, and HFNC was more, but that might of course have nothing to do with the O2 system. Rates of retinopathy were low and stable, and there was a bit more severe BPD and a bit less moderate BPD after implementation, but that could easily just be due to practice patterns changing and the use of HFNC, which tends to inflate rates of diagnosis of severe BPD (more about that another time!)

Showing a reduction in RoP, for example, when only 10% of the eligible babies develop the condition, and only half of them need treatment (as in these cohorts in Salverda et al) will need a very large trial. Which is why I am grateful that an international European collaborative will enrol over 2300 babies <28 weeks.

But, as I already mentioned, even if there is no improvement in clinical outcomes, as long as they are not worse, I think there are major benefits to families of having fewer alarms, and, if I had access to devices to fit to my ventilators/CPAP/HFNC devices right now, I would do so.

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