Endotracheal intubation is hazardous

Which is no surprise, I hope, to any of us.

Neil Finer has been a leader in the field of recognizing and quantifying the adverse physiologic effects of endotracheal intubation, and of finding ways to reduce those effects using premedication.

It was largely as a result of his work that professional societies now recommend using medications prior to intubation; medications that reduce pain, help to stabilize the babies physiology and facilitate the intubation.

We have recently, in our NICU, restricted the performance of endotracheal intubation in our most fragile patients to only those professionals who have already demonstrated their competence with larger, more stable babies. Infants under 29 weeks gestation are now only intubated by physicians, NNPs, or respiratory therapists who have shown that they can intubate larger infants. Babies with diaphragmatic hernia are also only intubated by a restricted list of people. I think that the sickest babies are not the place where a first-timer (or a second-timer) should be learning what is a difficult skill. Although junior residents who may in the future become the physician covering a delivery service in a peripheral hospital, and may well become the NRP team leader, do need to learn how to perform this skill well, our first priority has to be the babies, and making sure the most at-risk, tiniest, babies have the most skillful person performing procedures. It would be better, I think, to also have restrictions for the sickest larger babies with very stiff lungs who often desaturate severely and very quickly during intubation, but we have to find ways to ensure that residents leaving the training program have enough exposure to become competent.

We have submitted an abstract with our data to the next PAS meeting, and I will reveal our results at a later date, except to say that more experienced intubators are much more likely to intubate on the first attempt.

Does this matter? Well a new observational study from Hatch and co-workers in Vanderbilt studied 273 intubations in 162 patients. Adverse events occurred in 107 (39%) intubations with nonsevere and severe events in 96 (35%) and 24 (8.8%) intubations, respectively.

Nonsevere events included : Esophageal intubation with immediate recognition, Mainstem bronchial intubation (confirmed by chest radiograph), Oral/airway bleeding, Difficult bag-mask ventilation, Emesis, Chest wall rigidity

Severe complications included: Hypotension receiving treatment, An urgent or elective intubation becoming an emergency, Chest compressions, Code medications (presumably this means needing an epinephrine bolus), or Pneumothorax.

Adverse events were much more frequent for emergency, rather than elective or urgent intubations, and the odds of having an adverse event were doubled if there was more than one attempt at intubation. Infants who needed 3 or 4 attempts had about a 75% chance of an adverse event. Novice intubators had 22% success on the first try while for experienced intubators that was up to 57%.

Hypoxia and bradycardia were not counted as adverse events, as the authors wanted to be able to compare their data to studies looking at intubation for older children. But those two “secondary” outcomes were very common. 44% of the babies desaturated to below 60% saturation, and 24% had a bradycardia to less than 60 per minute.

We continue, in our NICU, to allow less experienced intubators to be the first to attempt intubation on larger babies as we think that the consequences, in the  long term, for the larger baby are probably less important, but are there long term consequences of all these adverse events in very immature babies?

A study from Stanford by Wallenstein and colleagues suggests that there are serious consequences. They compared the outcomes of babies under 1000g who needed intubation in the delivery room between those who were successfully intubated on their first attempt to those who required more than 1 attempt.

There were 88 babies over a 6 year period who were in their cohort.  40% were intubated on the first attempt and 60% required multiple attempts. Babies who were not intubated the first time were more likely to need chest compressions after the first attempt, were more likely to develop a grade 3 or 4 IVH, were more likely to develop NEC, a pneumothorax or PVL. Differences which remained after adjusting for other risk factors, and each of which individually was not “statistically significant”.

Death or neurodevelopmental impairment occurred in 29% of infants intubated on the first attempt, compared with 53% of infants that required multiple attempts, adjusted odds ratio 0.4 (95% confidence interval 0.1 to 1.0), P<0.05. Which was due to a more than doubling in the odds of dying (from 11 to 25%), and nearly a doubling in the odds of “neurodevelopmental impairment”.

What should we do about this?

Firstly, I think that our approach, of restricting intubation attempts by inexperienced personnel to only the more stable babies, is the way to go. More experienced personnel have a much higher rate of success on the first attempt, and fewer attempts means much fewer complications.

Secondly, we should all have strict protocols for premedication for elective intubations. The evidence of benefit, and the reduction of adverse events during premedicated intubations is very clear. One criticism of the data from Vanderbilt is that almost none of the intubations were preceded by muscle relaxation, and those that were pre-medicated used an opiate and a benzodiazepine. This despite the good evidence that muscle relaxation facilitates and shortens intubation, improves intubation conditions, and improves success on the first attempt; and the complete lack of evidence for a benefit of benzodiazepines.

Thirdly we need to find ways of training junior staff in this skill that do not include exposing critically ill babies to inexperienced intubators with a low success rate, including much more extensive use of simulation, simulations which are much more realistic, video-laryngoscopy, improved video-laryngoscopes, and so on.

Fourthly, we need to continue to investigate ways to make intubation less traumatic, less painful, less frequently unsuccessful, and faster.

Finally whenever we are about to intubate a baby, we should ask ourselves if we are in the best place to do it (can we wait till we get the baby to the NICU, place an IV and pre-medicate?) with the best personnel (do we have rules that there is always an experienced person present for every birth of an extremely preterm infant?) the best environment, the best equipment, and the best monitoring.

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

Another amazing publication (insert jokey emoticon here)

This one’s just me, and is a bit of a “how do I do it” article. I recount how I try and keep up with the emerging literature, and then some simple rules for determining if a new publication is worth the effort of trying to analyze the results. I hope people find it useful.

Barrington KJ. How to find and how to read articles in neonatology. Seminars in fetal & neonatal medicine. 2015.

The whole issue of the journal is dedicated to clinical research in perinatal medicine, from small single center trials to large multicenter RCTs and on to systematic reviews: their importance, their design, their interpretation. Peter Davis put together an interesting group of authors, I think, and has produced what I hope trainees, and others interested in neonatal clinical research, will find a good resource for several years to come.

BTW, one of the articles is by the amazing Annie Janvier, and our friend and parent collaborator and advocate Barbara Farlow.  Janvier A, Farlow B. The ethics of neonatal research: An ethicist’s and a parents’ perspective. Seminars in fetal & neonatal medicine. 2015. My only criticism of that piece is that the apostrophe is in the wrong place. I should be “a parent’s perspective”. Please fix that, Seminars!


Posted in Neonatal Research | Leave a comment

Late Surfactant may not be effective, probably.

A large multi-center trial (n=511) led by Roberta Ballard has just been published. (Ballard RA, et al. Randomized Trial of Late Surfactant Treatment in Ventilated Preterm Infants Receiving Inhaled Nitric Oxide. J Pediatr 2015.)

In this trial infants had similar enrollment characteristics to the NOCLD trial; babies were between 500 and 125o grams birthweight and less than 32 weeks gestation. They had to be receiving assisted ventilation. There were the following differences to the previous study I alluded to: in NOCLD infants were 7 to 21 days, in this trial 7 to 14 days; in NOCLD the smallest infants could be enrolled if they were on CPAP; not in this trial, everyone had to be intubated and ventilated.

The idea was, that infants with persistent respiratory distress after a week of age have evidence of surfactant dysfunction, so perhaps if we gave them more functional surfactant they would be able to overcome this, and then have reduced lung function abnormalities, would be able to breathe more efficiently and would end up with less lung injury. There are a couple of pilot studies showing short term improvements in pulmonary function and gas exchange in very preterm infants who were still getting respiratory support at a week of age, and who received surfactant. So the investigators thought that a big RCT to examine clinically relevant outcomes was warranted.

Which I think is fine. This was a reasonable question to ask, and a reasonable, clinically important, outcome to investigate (especially with a local treatment very unlikely to have systemic adverse effects). Given the previous data on inhaled NO in a very similar group of babies (in whom a secondary analysis suggested that the earlier part of the postnatal age group, i,e, 7 to 14 days, was more effective) you can’t fault the investigators for using iNO in all the babies. Even if the, as yet still unpublished, NewNO trial did not show a benefit.

All the babies were getting inhaled NO, according to the NOCLD protocol. The surfactant was randomly given to half of the babies.

But I can’t tell you how much surfactant was given, or with what frequency. A major problem with this report of the study is that I can’t figure out exactly what was the intervention. Which is a big problem. The investigators went to great (and probably unnecessary) lengths to mask the procedure, with a separate team, not otherwise involved in clinical care, who gave the surfactant (or didn’t) into the ETT behind screens. But they don’t actually say what dose was given.

Babies in the study got a dose of surfactant (or a sham procedure), as I said, the study report doesn’t even say how much they got (it was “standard clinical doses”) or how often they got it (it was every 24 to 72 hours if they remained intubated, starting at 48 hours after the first dose with a maximum of 5 doses; but 24 to 72 is a huge range…), they don’t say what were the criteria for retreatment, or for not retreating and extubating etc. There are several guidelines presented for steroids, for re-intubation etc, but not for surfactant/sham administration.

Table 3 of the results does show that about 80% in each group got 5 doses (of either surfactant or standing behind a curtain).

There was no benefit shown. Nothing, not even a whisper of a hint of a benefit. Which is disappointing, but at least seems at first look rather definitive. Or at least it would be definitive if we knew what the intervention group had received.

Even though surfactant dysfunction is a real problem in these babies, giving them additional Infasurf, according to this uncertain schedule, isn’t sufficiently effective to improve their outcomes.

This does, I think, help to improve care, (as there is no longer any stimulus to give surfactant to babies at this age) but it would have been much more useful, after what is probably several million dollars of investment, to know exactly what was done.

All we know is that, giving some dose of surfactant (Infasurf) and giving, mostly, 5 of those doses,  didn’t reduce BPD or death with a certain degree of confidence (see below).

To return to a comment I made above, why did I say that the sham procedure was unnecessary? Masking the intervention has become an essential feature of neonatal (and much clinical) research in order to get good funding; however, there is actually little empirical evidence that blinding/masking the intervention makes much difference to the size or direction of the effect of an intervention, particularly if objective outcomes are being studied. Diagnosis of BPD, if the ‘physiologic” definition is being used, is relatively objective, and is unlikely to be influenced by knowledge of an intervention performed several weeks earlier.

I think, if it is relatively easy, and relatively inexpensive, to mask an intervention (such as an orally administered drug, for example) then go for it, there is often no good reason to not do so. But having an on-call surfactant or sham administration team, who will go through the ritual of masking used in this study, will have enormously increased the cost. They could have studied twice as many babies (I guess) for the same cost, and have a much better estimate of the size of the effect, or of the confidence with which we can eliminate a benefit or risk. Many of the original surfactant trials for treatment of HMD were masked, in a similar fashion, but not all. There is no clear difference in the estimates of efficacy between those that were masked and those that were not.

Which brings me neatly to the final comment, the study was stopped by the DSMB, because “based on a determination that the study treatment is very unlikely to demonstrate efficacy” they didn’t think they should continue. They actually made this determination when they had the outcome data of 301 infants. There is a lot of debate about stopping trials early for futility, one paper in Critical Care (freely available on-line) is actually a real debate. But I am a bit mystified in this case, when the decision to stop the trial was taken they had actually randomized 511, of the planned 524 babies. One of the justifications for early stopping for futility is that it saves wasting money. That clearly isn’t an issue in this case. But even with all the data from 511 babies available there is still major uncertainty about whether this intervention is actually futile; the 95% CI for death or BPD include a 25% increase or decrease in that outcome. Which is huge, and clinically important. A 25% reduction (or increase) in death or BPD is something I would be interested in.

When the DSMB recommended stopping the trial they only had data from 300 babies, which means the confidence for saying there is no benefit (or harm) was extremely lacking in, er, confidence. Depending on how you calculate it, (assuming that the groups both had a 40% incidence of death or BPD at that point) when they stopped the trial showed that the likely real difference in that outcome was between about a 35% increase or decrease in risk of death or BPD. The sample size for the study was based on a hypothesized 13% change in the incidence of “death or BPD”, so why would the trial be stopped early when the confidence intervals included the hypothesized difference?

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

When should we treat hyperbilirubinemia in preterm infants?

I missed this when it was first published, but it came up on one of my regular searches at the weekend. Hulzebos CV, et al. The bilirubin albumin ratio in the management of hyperbilirubinemia in preterm infants to improve neurodevelopmental outcome: a randomized controlled trial–BARTrial. PLoS One. 2014;9(6):e99466. The multi-centered Dutch group of authors randomized 615 preterms of less than 32 weeks gestation. Babies had their bilirubin treated according to some arbitrary nomograms for total serum bilirubin. They had to be arbitrary of course, because there are few good data to inform an evidence-based nomogram. The thresholds for treatment seem reasonable, and are all available on the free access web page where the article is published. The group randomized to have the bilirubin-albumin ratio taken into account were in addition treated if their B/A ratio exceeded a line on another nomogram, also somewhat arbitrary. The 2 nomograms were different for different strata of birthweight, and differed over the first couple of days of life.

The idea being that if you start treatment earlier for those babies with a low albumin, and a bilirubin/albumin ratio which is over the threshold, then they might have less bilirubin induced brain injury (often evident as motor disturbance later in life, they thought).

What this meant in the end was that the group with the B/A ratio included had an average of only 6 more hours of phototherapy (which was not statistically significant) and 2 exchange transfusions vs none in the TSB alone group. There were fewer deaths in the B/A group, 5.2% vs 8,1%, p=0.2, i.e. a difference which could easily be due to chance. The primary outcome was the composite motor score at 18 to 24 months corrected age.

There was no difference between the groups in the primary outcome. Or any indices of development at 2 years of age.

The only intervention which occurred as a result of this change in screening policy was a minor increase in the duration of phototherapy; I don’t think you can postulate that this was the cause of a minor reduction in mortality. The study was not designed, or powered, to show a difference in mortality. I think even doing the subgroup analysis that they present is really questionable, a subgroup analysis of an unexpected non-significant difference in a secondary outcome is something you should think twice about, and then think another three times, before presenting. It could easily mislead people into thinking that there was a real difference to worry about.

So either the B/A ratio is not of any value for directing therapy and improving motor outcomes in preterm infants, or the thresholds they created were not right. I guess its possible that a much stricter threshold for intervention would have shown some differences in outcome, which if it just led to more phototherapy might be a benign intervention. But on the other hand phototherapy is not entirely benign, it interferes with mother infant interactions, and affects developmentally sensitive care. Exchange transfusions are certainly not benign.

I think it is hard to believe that there will be another investigation of this issue, certainly not of this high quality. The difference in interventions would have to be greater that this to be able to show an effect. Perhaps another trial of prophylactic early phototherapy is warranted, as the large NICHD trial did show a reduction in “neurodevelopmental impairment”.

For now I think the thresholds for treatment suggested by Jeff Maisels and his pals are about as good as we can get.

Posted in Neonatal Research | 3 Comments

A new book, with great chapters!

A new book about neonatal clinical ethics has just been released. Eduard Verhagen and Annie Janvier are the editors.

Here is the blurb:

Ethical Dilemmas for Critically Ill Babies

Editors: Verhagen, Eduard, Janvier, Annie.

Series: International Library of Ethics, Law, and the New Medicine, Vol. 65


Addresses important ethical questions about human life and medical interventions

Only volume which covers the important issues related to ethical, legal and medical aspects of neonatology intensive care

Contains contributions from the leading experts in the field of neonatology, bioethics and health law. Most neonates who now survive intensive care would have died 50 years ago, and “nature” would have decided the outcomes, making ethical discussions about initiating or withholding resuscitation irrelevant. Medical developments in neonatology have changed the way we respond to diseases of neonates, to their illness, and to their parents. Not only as physicians, but also as a society. Decisions on when to start, withhold, or withdraw life-saving interventions in critically ill neonates are among the most difficult decisions in pediatric practice. These decisions are fraught with ethical dilemmas, for example deciding whether withholding intensive care –leading to death- is superior to uncertain survival with a risk of disability and the additional burden of intensive care. This book covers important ethical questions that arise in neonatal intensive care units. Questions such as, whether to intervene medically and whether we are good at predicting the outcome of fragile neonates; whether a medical intervention should be withheld or withdrawn, and who should be primarily responsible for these decisions and how?


The book is derived from presentations that were given at a conference near Geneva that was sponsored by the Brocher Foundation.   There are 2 chapters written by yours truly:

Predicting Outcomes in the Very Preterm Infant.

Keith J Barrington (page 51).

A perceptive and disturbing evaluation of why we do so many imperfect tests in the NICU to try and predict the future life of newborn babies.

(that’s not the publisher’s blurb, I wrote that)


Neonates are Devalued Compared to Older Patients

Annie Janvier, Carlo Bellieni, and Keith Barrington (page 25).

A perceptive and disturbing evaluation of the way newborn infants are treated differently, and negatively compared to older children.

(Guess who wrote that).

You can buy it here, as an ebook it’s a steal at $99.

Posted in Neonatal Research | Leave a comment

A step forward in neonatal resuscitation. And Oh So Simple.

When you are resuscitating a baby, and you ask, how is the heart rate? What kind of answer do you get? “It’s good” “pulse is a bit slow” “I think its around 80”?

As Lou Halamek and his team recount (Yamada NK, et al. Impact of Standardized Communication Techniques on Errors during Simulated Neonatal Resuscitation. American journal of perinatology. 2015), such kinds of communication in an Air Traffic Control tower would see you booted out. In air traffic control, they realized that imprecise, or even just variable, ways of recounting what was going on, were leading to errors; so now, when there is a communication, it has to follow a strict format. Yamada et al have followed that lead and developed a standard lexicon of information transfer for neonatal resuscitation, including a closed-loop communication system for completing orders for medication or volume resuscitation.

In answer to the question above, the responses have to be one of the following “heart rate above 100″; ‘heart rate below 100”; “heart rate below 60”; or “we’re in deep shit’ (I made that last one up, it should be “heart rate zero”).

You can see how that would be better than “I can’t hear a heart rate” which might mean, “I have a middle ear infection”, or “my stethoscope just dropped on the floor”, or “the baby is asystolic”!

Similar structured phrases are presented for other issues during resuscitation.

For medications or volume, the order must include the name of medication, dose, concentration and route AND it must be repeated by the person taking the order and include all the same information.

They studied this in an RCT, with nurses trained in the lexicon. A cross-over design was used, so the trained nurses either used the lexicon or did not, and 13 people with some experience in NRP were enrolled as the study subjects.

This is a weakness in design of this study,  I think, the trained nurses were told for the control group to “follow the general pattern of imprecise communication that is typical of nonstandardized speech”. It is certainly possible that the nurse participants, who were probably invested in showing that this works, might have used even more imprecise phrases than usual. I don’t remember ever hearing “wow he’s crackly” (one of their examples) when asking about air entry, but I guess the point is that it could happen. I think it would have been better to use non-trained nurses as the controls, but I can see that would introduce other biases also.

The other problem is that although this seems like a major, and very obvious, improvement (although I didn’t think about it), the sample size was so small that most of the changes seen were not that significant. There were fewer errors of omission (failure to perform an intervention that was clinically indicated), cardiac compressions were started earlier, by about 8 seconds, PPV was started earlier by about 2 seconds. Communication techniques were used much more frequently; that comparison was statistically significant.

This is another place where I am not sure we need another trial; we should probably all start doing this, making sure communication is clear, by using standard phrases. I can’t see any down side, it wouldn’t cost anything, and the only thing a larger study is likely to find is that sometimes communication errors lead to screw ups.

Posted in Neonatal Research | Tagged | 1 Comment

Also, still no parents!

My previous post was long enough without addressing another serious deficiency in these guidelines. It is worth its own post.

The guidelines are written by doctors. And only doctors (actually only Obstetricians). There is no input mentioned from any other stakeholders.

Most importantly no parents.

Surely the time has come to stop doing this, to stop physicians gathering in closed rooms and deciding what is best for the rest of the population. This is not a new idea, it has been strongly recommended for many years. The Institute of Medicine published guidance 4 years ago (to my mind their guidance leaves the patient/parent out of the process until too late) about involving patients/parents in the guideline development process. There are very high profile editorials in leading journals. There are even blog posts about it!

So why are bodies like ACOG and the SMFM not doing this? Is it because they may not like what they would hear? Is it easier to meet around a table with other physicians and not be challenged by a patient? Is it even easier if you limit the physicians to just your Obstetrical colleagues?

One of the reasons identified for why clinical practice guidelines have little effect on practice is the lack of involvement of the people most affected by them.

Because that is the whole point. The Obstetrician will leave work at the end of the day to go home, the families are marked forever by these decisions. Whichever way they go.

Posted in Clinical Practice Guidelines | Tagged , , , | Leave a comment