Probiotics: are they cost effective?

Like almost any intervention in the NICU (even maternal breast milk requires pumps and equipment, and kangaroo care requires chairs) there is some cost associated with routine probiotic administration. A new publication attempts to calculate the cost-effectiveness relationship. (Craighead AF, et al. Cost-effectiveness of probiotics for necrotizing enterocolitis prevention in very low birth weight infants. J Perinatol. 2020).

They used baseline estimates of NEC incidence of 5.8%, reduced to 2.4% by probiotics, costs of NEC hospitalisations and rates of serious long term outcomes and their costs from the literature. The baseline cost of the probiotics they included was $2,200 per baby.

Using these estimates, routine use of probiotics in at-risk babies (with that baseline NEC risk) cost $1,868 per QALY (quality-adjusted life-year) saved. Which is probably the most cost-effective intervention ever seen in an ICU!

When you perform a study like this you start with baseline assumptions and then see how variations of those assumptions affect the calculations.

If your region has more NEC than 5.8% then the costs per QALY are even lower. In fact, they showed that at a NEC incidence above 6.5% probiotics don’t cost anything, they start to have a negative cost, the the system as a whole saves money by instituting routine probiotics.

Even at a NEC incidence of 2% probiotics only cost about $18,000 per QALY, which is dramatically below the threshold used to fund new interventions in the UK (about $100,000).

If your probiotics cost less than $2200 per baby then again the calculations are changed. Our probiotics currently cost 50 cents a day per baby (single-use sachets of Florababy). I haven’t calculated the average cost per baby, but a 24-week infant receives them for 10 weeks, or $35. If you don’t change any other assumptions then the total cost per QALY is -$5000! At that low a cost, probiotics are cost-effective at a NEC incidence of 0.1% (a figure I just made up, the real threshold may be much lower than that).

The results of this analysis are hardly a surprise, probiotics are currently relatively inexpensive to use in the NICU and prevent a serious condition, here’s hoping that they stay that way.

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

Probiotics, can’t get enough…

All probiotics are not equal. I think it is clear, for example, that the probiotic fungi (Saccharomyces) studied in 2 small trials had no impact on NEC. In addition one of the largest and best trials (PiPS) from the UK used B breve, and no clear benefit was shown. The lack of impact in that trial could be a random effect, with the confidence intervals suggesting that a reduction in NEC of up to 32% was consistent with the trial (or an increase in NEC of up to 27%). It could also be that the B breve strain used (BBG-001) is not effective for NEC, and/or that it is not very good at colonizing the neonatal GI tract. The intestinal microbiome analysis of some PiPS babies showed very little impact of B breve on the microbiome, even among those who were culture positive for the bug. That study looked at a single stool sample among babies who received B breve and in controls within a week of completing the PiPs trial.

In the network meta-analysis that I posted about recently, the majority of studies in the group with the highest impact on mortality and on NEC (at least one lactobacillus and at least one bifidobacterium) included a B longum subsp infantis, also sometimes called B infantis many others included a B longum subsp longum.

We have to be careful when lumping or splitting the trials, even two trials that used a Solgar mixture (Bin-Nun et al and ProPrems) had different components, B infantis, B bifidus, Streptococcus thermophilus in the first, B infantis, B animalis subsp lactis, Streptococcus thermophilus in the second. That second product was ABCDophilus produced by Solgar in New Jersey for the Australian trial, but as many of you will know, is no longer available as there was a case of intestinal mucormycosis in an infant who received it.

Infloran has been used in several trials, but the composition has not been identical in all. It currently seems to contain a Lactobacillus (L. acidophilus) and a Bifidobacterium (B bifidum, sometimes called B bifidus), but in the first trial by Lin et al it is reported as containing L acidophilus and B infantis, in the second trial by the same investigators Infloran was reported as containing L acidophilus and B bifidum.

One cannot, of course, sterilize probiotic preparations, as the whole idea is that they are live bacteria. In addition, as a general rule, it is very difficult to sterilize powders, so even, for example, powdered cellulose added to make a measurable volume has a risk of contamination. The exact identification of species, subspecies and strains, and what they are called is very confusing. The product that we use, and have published the results from (Florababy), has 4 Bifidobacteria (breve, bifidum, longum subsp infantis, and longum susbsp longum) and a Lactobacillus, (rhamnosus). As this post has already mentioned the same organism, or extremely similar organisms, are called different things by different groups. There are official naming conventions, but these are frequently not followed. The Abbott Tri-blend probiotic mix, which is now available in the USA is advertised as containing B lactis, B infantis, and S thermophilus. B lactis is synonymous with B animalis subsp lactis, which appears to be the officially preferred nomenclature. B infantis is, as previously mentioned, B. longum subsp infantis, S thermophilus is also called Streptococcus salivaris subsp thermophilus. The Abbott preparation seems to include the exact same strain of B infantis BB-02 that was in the ABCDophilus (ProPrems) product, as well as the exact same strains of B lactis and S Thermophilus.

I looked all this stuff up so that you don’t have to.

How you divide trials like these into groups for the network meta-analysis is also a question. It may have been just as valid to have one group being all the trials that studied B infantis, either alone or as a component of a multiple strain product, or perhaps B longum, and include both subsp longum and subsp infantis. Or perhaps S thermophilus along with at least one other Bifidobacterium; there are a number of potential choices, for how you could lump the trials.

As there have been almost no trials comparing different species/strains it is very difficult or impossible to be definite about which are the most useful, but we can identify, I think, the most likely to be useful. I believe from perusing the supplemental data of the Network Meta-Analysis (NMA) that all of the babies in the group of trials that the NMA identified as being most effective received a Lactobacillus and either B bifidum or a B longum, most of them received B longum subsp infantis, and some B longum subsp longum. I don’t think that any of the trials studied B infantis alone. Also, the 2 trials that were in the second most effective group, Bifidobacterium spp + S thermophilus both included a B Infantis.

This seems to me to confirm that a combination product which includes a B longum, either subsp infantis or subsp longum, or a combination product with a B bifidum would fit the profile of the most likely to be effective groups. As there are no trials I can see that studied a B longum subsp infantis alone in preterm babies for NEC prevention, and one subgroup with only 48 babies that studied a B longum subsp longum alone, and another single trial that studied B bifidum alone, then single strain Bifidobacterial preparations must remain uncertain in efficacy.

I am not enough of a bacteriologist (not one at all!) to know how closely related these different subspecies and species that seem likely to be effective are, or why combinations might be more effective than single strains. How the organisms metabolize human milk oligosaccharides (HMOs), how they interact with human enterocytes, how they modulate intestinal immunity, and their impacts on Toll-like receptors, may all be important.

The group in the UK that previously published their experience with introducing Infloran (in this iteration containing Lactobacillus acidophilus and B bifidum) have just published data on intestinal microbiome analysis on repeated sampling among their babies. They have a comparison of contemporary infants from other NICUs that did not routinely supplement, so an observational study rather than an RCT, but very interesting all the same. Alcon-Giner C, et al. Microbiota Supplementation with Bifidobacterium and Lactobacillus Modifies the Preterm Infant Gut Microbiota and Metabolome: An Observational Study. Cell Reports Medicine. 2020;1(5). Here is the visual abstract.

The genomic analysis of the strains involved showed the following:

the ability of Bifidobacterium to digest HMOs varies between species and strains of this genus. Thus, we analyzed B. bifidum genomes (our 5 isolates and Infloran strain) for the presence of genes involved in HMO utilization; all B. bifidum isolates contained specific genes involved in HMO utilization, and mucin degradation genes that may aid gut persistence. Notably, growth curves in whole BM confirmed that the B. bifidum Infloran strain utilized whole BM. Further phenotypic analysis indicated this strain was able to metabolize specific HMOs; 2-fucosyllactose (2′-FL) and Lacto-N-Neotetraose (LnNT)

Some fascinating findings from this study include the high relative abundance, in the Infloran group, of multiple strains of Bifidobacteria, not just those actually in the Infloran. In fact, there was much more B breve in the supplemented group stools, B breve was not in the supplement! Bifidobacteria seem to like the company of their own type of bug.

They showed evidence from the metabolomic analysis that the supplemented kids had gut flora that were actually doing their job and metabolizing HMOs to acetate and lactate, leading to a lower stool pH. They also confirmed the adverse effects of antibiotics on intestinal microbiome composition, and that the more immature infants had more difficulty developing a Bifidobacteria preponderant microbiome.

Previous studies of B breve have shown that they are extremely variable in their ability to metabolize HMOs, even between strains of what are labelled as the same species. As far as I know, there is less variability in this aspect of B longum subsp infantis activity, B bifidum, on the other hand, has a lower ability to metabolize HMOs but can absorb some of them, and even “altruistically” leaves some behind and feeds them to their cousins!

In this study, we observed altruistic behaviour by B. bifidum when incubated in HMOs-containing faecal cultures. Four B. bifidum strains, all of which contained complete sets of HMO-degrading genes, commonly left HMOs degradants unconsumed during in vitro growth. These strains stimulated the growth of other Bifidobacterium species when added to faecal cultures supplemented with HMOs, thereby increasing the prevalence of bifidobacteria in faecal communities

This truly weird behaviour of certain of these bugs might explain why combination products are more effective. Of course, we should be careful in our interpretation of the clinical trial data. Any comparison between multiple strain products and single strain products, each group of which contains many different species, is not particularly robust. In addition, the use of prebiotics, such as HMOs in particular, might make a difference to efficacy of different preparations.

It is surprising what you can discover from looking at babies poop.

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

Probiotics save the lives of preterm infants. Find a reliable source.

The American Gastroenterological Association has recently published its guidance about probiotics for various conditions. They found little evidence of benefit for most indications with the probable exception of preventing antibiotic-associated diarrhoea. The 8th indication that they reviewed was the prevention of NEC. Their conclusion after robust evidence review was

In this vulnerable population of premature infants … these data strongly suggest that probiotics may protect from mortality and do not increase rates of sepsis.

The full systematic review and network meta-analysis has been published. (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). They found 63 trials which included a total of over 15,000 infants.

The published article does not have one of those weird network plots, and I am sure you would miss it, so here is a plot from the supplemental data

B. adolescentis: Bifidobacterium adolescentis; B. & B.: Bifidobacterium animalis subsp. lactis & Bifidobacterium longum subsp. longum; Bac. coagulans: Bacillus coagulans; Bac & En.: Bacillus spp. & Enterococcus spp.; B. & Strp.: Bifidobacterium spp. & Streptococcus salivarius subsp. thermophilus; B. bifidum: Bifidobacterium bifidum; B. lactis: Bifidobacterium animalis subsp. lactis; B. longum: Bifidobacterium longum subsp. longum; B. breve: Bifidobacterium breve; B. clausii: Bacillus clausii; L. acidophilus: Lactobacillus acidophilus; L. & B.: Lactobacillus spp. & Bifidobacterium spp.; L. & B. & En.: Lactobacillus spp. & Bifidobacterium spp. & Enterococcus spp.; L. reuteri: Lactobacillus reuteri; L. & B. & Sac.: Lactobacillus spp. & Bifidobacterium spp. & Saccharomyces boulardii; L. & B. & Strp.: Lactobacillus spp. & Bifidobacterium spp. & S.
salivarius subsp. thermophilus; L. rhamnosus: Lactobacillus rhamnosus; S. boulardii: Saccharomyces boulardi.

The results are summarized in this table.

The meaning of the colours in the table is as follows

Compared with placebo, a combination of 1 or more Lactobacillus species and 1 or more Bifidobacterium species was the only intervention with moderate- or high-quality evidence of reduced mortality (odds ratio [OR], 0.56; 95% confidence interval [CI], 0.39–0.80).

Among interventions with moderate- or high-quality evidence for efficacy compared with placebo, combinations of 1 or more Lactobacillus spp and 1 or more Bifidobacterium spp, significantly reduced severe NEC (OR, 0.35 [95% CI, 0.20–0.59].

Unlike some other SRs with meta-analysis, they did not show a reduction in late-onset sepsis. They also did not show any significant incidence of invasive infection with the probiotic organisms.

One thing I have difficulty understanding is the lack of clear guidance from either the American Academy of Pediatrics or the Canadian Paediatric Society regarding probiotic use in the preterm. Anything else that had been studied in 63 RCTs with over 15000 babies randomized showing a reduction in mortality, and a large reduction in serious NEC would surely by now have been the subject of a neonatal-specific position statement. The AAP does have a statement from the Committee on Nutrition; Section on Gastroenterology, Hepatology, and Nutrition, which is 10 years old, it notes the evidence that was available then and promotes caution. The CPS also has an 8-year-old statement which covers multiple conditions, and wrongly states that there were no data for infants under 1kg.

I think that this new systematic review and network meta-analysis is clear enough. All at-risk babies should be receiving probiotics, and they should be receiving a mixture which includes 1 or more Lactobacillus species and 1 or more Bifidobacterium species. I think that clinicians need guidance on how to find a reliable safe source.

In Canada, we are fortunate that there is some regulation of such products by Health Canada, who have awarded an NPN to some probiotic products. A Natural Product Number is evidence of quality control and Good Manufacturing Practice standards, and that there is reason to believe that what is in the bottle is actually what it says on the label.

I have heard that EVIVO is being marketed very aggressively to hospitals in the USA, as a potential prophylactic agent against NEC. It is from an enterprise started by researchers from UCDavis, I believe, and as far as I can see they have very high manufacturing standards. However, single strain B. infantis, which is what EVIVO is, has not been studied in any RCT in the preterm to my knowledge. Many of the combination products, shown to be the most effective in this new network meta-analysis, did contain a B. infantis; but not all of them. I think it would be great if EVIVO performed an RCT against one of the other combination products shown to be effective. Indeed the network meta-analysis that I quote includes that as its final recommendation; “further research is needed”; how innovative a recommendation!


Posted in Neonatal Research | Tagged , , | 3 Comments

Lactoferrin: Does it decrease late-onset sepsis?

The first multicentre trial of bovine lactoferrin supplementation in preterm infants showed a dramatic reduction in late-onset sepsis of about 70%. I was very excited when I first saw Paolo Manzoni’s trial and rushed to set up a pilot in my NICU so that we could apply for funds to perform a confirmatory trial. The Lacuna trial which we published was not powered for sepsis outcomes, but for the feasibility of a subsequent masked multicentre RCT using our preparation, and ensuring that we could mask it easily. The proportion of babies with at least one infection was similar between our groups, but an exploratory analysis that I performed (and didn’t publish as it was unplanned and is therefore not scientifically reliable) suggested a reduction in infections per 100 patient days. That confirmed my desire to proceed with further trials.

The subsequent trials have generally shown only small differences in sepsis, with one or two exceptions from small trials. All the trials do show less sepsis with lactoferrin than control, but, in many of the trials, the difference was very small and may have been due to chance.

The most recent trial, LIFT Australia, is now published (Tarnow-Mordi WO, et al. The effect of lactoferrin supplementation on death or major morbidity in very low birthweight infants (LIFT): a multicentre, double-blind, randomised controlled trial. The Lancet Child & Adolescent Health. 2020;4(6):444-54). This was a multi-centre masked trial in 1500 infants of less than 1500g birthweight randomized during the first week of life. Unlike the original Manzoni trial, and my pilot, the dose was by body weight at 200 mg/kg/day. I gave everyone 100 mg/day, so except for a baby under 500g, this new trial gave more. The primary outcome of the trial was a composite of death, NEC (stage 2 or 3) Late-Onset Sepsis (LOS), retinopathy needing treatment, and brain injury. There was no difference in this outcome 21% vs 22%. The difference in LOS alone was very small 12% with lactoferrin and 14% in controls, it only included culture-positive sepsis.

The LIFT dose was higher than the previous, even larger, masked multi-centre trial from the UK in 2200 babies <32 weeks and ❤ days of age. (Griffiths J, et al. Enteral lactoferrin supplementation for very preterm infants: a randomised placebo-controlled trial. The Lancet. 2019;393(10170):423-33). This trial, known as ELFIN, gave 150 mg/kg. The primary outcome was LOS or “clinically suspected infection”, which was 29% in the lactoferrin group and 31% in controls. As frequent readers will know I am very sceptical about the occurrence of “culture-negative sepsis”, and their definition, which is given in detail in this publication, doesn’t make me any less sceptical. If we focus on confirmed LOS, the incidence was more similar to LIFT, 17% vs 17%, identical in the 2 groups.

One of the strengths of the LIFT trial is that in the publication they performed a PRISMA compliant systematic review of all the available evidence. That review suggested a decrease in Late-Onset Sepsis (LOS) with lactoferrin compared to control; as I mentioned before, all of the trials show fewer babies with LOS if they received Lactoferrin, but many of the studies had very small differences, including ELFIN, RR=0.94, and LIFT RR=0.83. However, as mentioned, this includes culture-negative sepsis in the ELFIN trial, but not in LIFT, or in Manzoni et al, or in Barrington et al (I didn’t check all the others, 2 are in Chinese and I can’t get hold of them anyway.)

The statistical tests of heterogeneity showed that the trials differed in this outcome. The I² from the random effects version of the meta-analysis is 58%, and the difference between trials is unlikely to be due to chance alone according to a chi-square analysis.

Why might this be? Of course, the difference between trials might just be a chance occurrence, but the statistical testing suggests it may be more than that. It is possible that the patients are different. One difference could be the breast milk intake of the babies. Maternal breast milk, unpasteurized, has higher concentrations of lactoferrin than pasteurized donor milk (human lactoferrin obviously), and there is almost no bovine lactoferrin in commercial formula. In the Lacuna trial, 90% of the babies received some maternal milk, and 60% had no formula at all, there was no donor milk at that time in our NICU. In ELFIN most of the babies received a mixture of formula and breast milk, secondary analysis suggests that there might be a differential impact of lactoferrin according to feed type, but it is not a clear difference between the groups., the statistical test for the interaction was not convincing. In LIFT there was a very high frequency of maternal breast milk use.

This may be one reason that we saw very little impact of lactoferrin supplementation on the intestinal microbiome of our babies, Grzywacz K, et al. Bovine Lactoferrin Supplementation Does Not Disrupt Microbiota Development in Preterm Infants Receiving Probiotics. J Pediatr Gastroenterol Nutr. 2020;71(2):216-22; in a side-study that we have just published among 70 of the 80 babies in our pilot, who were all receiving probiotics and mostly receiving breast milk, differences in microbiome development were minor between the probiotic group (P) and the probiotic with lactoferrin group (PL).
Breastfeeding rates were somewhat lower in the original Manzoni trial, and he has recently published a secondary analysis suggesting that lactoferrin was much more effective in babies receiving formula rather than breast milk. (Manzoni P, et al. Is Lactoferrin More Effective in Reducing Late-Onset Sepsis in Preterm Neonates Fed Formula Than in Those Receiving Mother’s Own Milk? Secondary Analyses of Two Multicenter Randomized Controlled Trials. Am J Perinatol. 2019;36(S 02):S120-S5) although after multivariate correction the data are less convincing.

I also wonder if we are all testing the same thing. Is all lactoferrin identical? A new publication from UC Davis shows that different sources of bovine lactoferrin have somewhat different properties, (Lönnerdal B, et al. Biological activities of commercial bovine lactoferrin sources. Biochem Cell Biol. 2020.) here, in what I believe is a first for the blog, is a Western blot!

bLf is the lactoferrin as processed by the investigators themselves, showing a single band at 75 kD, similar to Dicofarm which was the Manzoni product, Tatua was the ELFIN product, showing in addition 2 other bands. LIFT used “Australia’s own” lactoferrin, which may not be identical to any of these, and LACUNA and LIFT Canada (currently underway) use a product supplied by AOR, which seems to have its origin in Australia.

hLf is human lactoferrin, which has a similar structure to bLf, but not identical.

The authors of this study also looked at bioactivity, using various assays, and did proteomic analysis showing some contamination with small concentrations of other bovine milk proteins.

In summary, it is apparent that the commercial Lfs we investigated have different degrees of purity and that the contaminating proteins also vary. There are considerable variations in the bioactivities among the Lf samples, but there is no discernible consistency such that one sample is high for all activities and another has overall low activities. The mechanisms behind NEC and sepsis are not known yet but likely involve mucosal integrity (cell differentiation), infection (EPEC) and immune response (cytokines), but since we do not know the relative involvement of each of these processes it is difficult to assess the reason(s) why some clinical trials have shown positive outcomes while others have not. Finally, there seems to be limited effects of pasteurization of Lf on its bioactivities.

I think this means that there is still potential that some preparation of bovine lactoferrin will have a significant impact on LOS in preterm infants. Also, lactoferrin has never been associated with adverse impacts. Because of the very important effect shown in the Manzoni trial it would be worthwhile to do other studies, to complete LIFT Canada, and any other trials underway, and then perhaps to do another trial with the Dicofarm product. I think we should focus on culture-positive sepsis and NEC, and hope that the original promise of lactoferrin will be found to lead to an effective way to reduce at least partially the high frequency of LOS in very preterm babies.

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

When should we start Parenteral Nutrition?

Following on from the previous post:

This all brings me to a larger and very thorny issue, which is whether we should even be routinely starting parenteral nutrition (PN) immediately after birth at all! I hate to discuss this, as I will probably sound like a Luddite wanting to destroy the new-fangled mechanical looms that everyone knows are the future.

As mortality rates were progressively decreasing during the early development of neonatal care we assumed that everything we were doing was good (and most of it was, apart from a series of serious errors, hyperoxia, hexachlorophene, chloramphenicol, benzoic acid…). However, many innovations that were introduced were not subject to adequate scientific investigation. There were some innovations, such as the introduction of CPAP and PEEP, that led to immediate survival of babies who would previously have died, and did not require further testing. Some innovations such as early PN, which was initially accompanied by periods of prolonged fasting of very preterm babies, have never been well enough studied.

Immediate PN in the very preterm baby has been promoted (by me, among others) as a way of minimizing the otherwise dramatic nutritional deficits which arise in the early life of the very preterm infant. Very preterm babies have such low fat and protein content, and high metabolic demand, that they become catabolic quickly after birth. This catabolic phase can be eliminated with early parenteral nutrition (PN). Surely that is a good thing! Most of the trials of early PN have focussed on different components, different doses of protein etc. and they have mostly examined short term metabolic impacts, such as changes in protein balance. Whether we should really be doing this at all has not been well studied.

One study that made me pause and think about this was the PEPaNIC trial (Fivez T, et al. Early versus Late Parenteral Nutrition in Critically Ill Children. NEJM 2016;374:1111-22) this was a study from Europe and Canada that randomized children in the ICU to either immediate PN, or to wait 7 days before starting PN. Infants in the early PN group were much more likely to develop sepsis. It would be easy to say that this was older children of many different pathologies, and has limited relevance to the NICU, but there was a pre-planned subgroup analysis of the full-term babies involved.  The initial publication of the trial of 1440 children in PICUs showed the following :

new infection was 10.7% in the group receiving late parenteral nutrition, as compared with 18.5% in the group receiving early parenteral nutrition (adjusted odds ratio, 0.48; 95% confidence interval [CI], 0.35 to 0.66). The mean (±SE) duration of ICU stay was 6.5±0.4 days in the group receiving late parenteral nutrition, as compared with 9.2±0.8 days in the group receiving early parenteral nutrition; there was also a higher likelihood of an earlier live discharge from the ICU at any time in the late-parenteral-nutrition group (adjusted hazard ratio, 1.23; 95% CI, 1.11 to 1.37). Late parenteral nutrition was associated with a shorter duration of mechanical ventilatory support than was early parenteral nutrition (P=0.001), as well as a smaller proportion of patients receiving renal-replacement therapy (P=0.04) and a shorter duration of hospital stay (P=0.001).

More of the late PN group had a hypoglycemic event (9% vs 5% had blood sugar less than 2.2 mM/L (=40 mg/dL)), but apart from that all of the differences between groups were in favour of the late PN group.

The newborn group from that trial was analyzed separately (van Puffelen E, et al. Early versus late parenteral nutrition in critically ill, term neonates: a preplanned secondary subgroup analysis of the PEPaNIC multicentre, randomised controlled trial. Lancet Child Adolesc Health. 2018;2(7):505-15). I must say I am really pissed off that even 2 years after this was published it is still not open access. Even with my good fortune as a professor of paediatrics at the University of Montreal and our inter-library loan system, I still cannot get immediate access to the full text of this study. The Lancet group of journals, which includes the journal which published this article, asks for at least $22 (Canadian) to have short term access to the full text. The research was performed using public money (European money, but still), and now to see the results I have to spend my own cash! Thankfully the authors have been very helpful and sent me a copy of the pdf immediately when I asked.

There were 209 newborns in the PEPaNIC trial, 98 in the early PN group and 110 in the late group. Of those 145 were 1 week or less of age, and 45 were on day 1 of life. They were in the PICU rather than an NICU mostly because of a surgical diagnosis, including cardiac surgery; that applies particularly to the babies under 1 week of whom over 80% were surgical. The main comparison in the trial is described here:

For patients assigned to early parenteral nutrition, this was initiated within 24 h after admission to the paediatric ICU as supplementation if enteral nutrition provided less than 80% of the target to reach the local age-specific and weight-specific caloric targets. In patients assigned to late parenteral nutrition, parenteral nutrition was withheld during the first week in the ICU. To match the fluid administration of the early parenteral nutrition group, taking into account the volume of enteral nutrition delivered, a mixture of dextrose 5% and saline was provided. To prevent refeeding syndrome, patients from both groups received intravenous micronutrients (trace elements, minerals, and vitamins) early in similar amounts, until enteral nutrition reached 80% of caloric targets. For patients from both groups who were still in the ICU after a week and who were not yet receiving 80% of the caloric target enterally, parenteral nutrition was administered to reach the targets.

The two primary end points were new infection acquired during the ICU stay and the duration of ICU dependency, which was adjusted for five prespecified baseline risk factors (diagnostic group, age group, severity of illness, risk of malnutrition, and treatment center)

Among the newborn infants, just as for the overall study group, there were many more hospital-acquired infections when the babies received early PN, compared to PN not started until 1 week. 31% vs 16% for the whole newborn group, the younger the babies were at admission, the more striking the difference between groups. Under 1 week it was 36% vs 14%, and in the day 1 babies, the difference was 57% vs 12%. The early PN babies also had longer ventilator dependence, longer PICU length of stay, which continued to be different after adjustment for potential confounders.

The newborn subgroup was more likely to have at least one blood sugar <2.2 mM/L in the late PN group. As you can see from the supplementary tables (well you can’t actually because you have to pay for them also so I’ll copy one example here) the late PN group had much lower intakes of sugar, protein and fat during the 1st week, the 1-day olds were almost all nil by mouth, it seems, as their enteral nutritional intakes averaged 0.

You can see from this table showing the nutritional intakes of the babies admitted on day 1 that the parenteral glucose intakes were very low in the late PN group; probably using 10% dextrose, instead of 5%, and ensuring adequate glucose intake to prevent hypoglycaemia would be preferable.

A further secondary analysis of the results in this subgroup (and it is now starting to get risky, secondary analysis of one facet of the intervention in a subgroup, addressing the 2 primary outcomes and one secondary outcome) suggested that the main culprit was the amino acids

In neonates aged up to and including 4 weeks, higher average doses of aminoacids were associated with a lower likelihood of earlier live discharge from the paediatric ICU from day 2 to day 5 (HR 0·56–0·71, p≤0·04). In neonates aged up to and including 1 week, a similar association reached significance for doses up to days 3, 4, and 7 (HR 0·42–0·70, p≤0·04; figure). Higher average doses of aminoacids were also associated with a lower likelihood of earlier live weaning from mechanical ventilation from day 3 onwards in neonates aged up to and including 4 weeks (HR 0·44–0·66, p≤0·01; figure). A similar association with aminoacids occurred in neonates aged up to and including 1 week, with significance reached for average doses administered up to days 4, 6, and 7 (HR 0·32–0·58, p≤0·03; figure). No association occurred between the average doses of glucose and any of the efficacy endpoints. Higher average doses of lipids were associated with a higher likelihood of an earlier live discharge at day 7 in both neonates aged up to and including 4 weeks (HR 1·66, 95% CI 1·11–2·52; p=0·015) and aged up to and including 1 week (2·00, 1·15–3·45; p=0·013). Higher average doses of lipids were also associated with a higher likelihood of an earlier live weaning from mechanical ventilation in neonates aged up to and including 4 weeks (up to days 6 and 7, HR 1·46–1·73; p<0·05) and those aged up to and including 1 week (up to day 7, HR 2·10, 95% CI 1·17–3·93, p=0·016; figure).

The “figure” referred to is the following, which I found somewhat confusing. It suggests that none of the individual components of the PN were associated with nosocomial infections, which is weird.

Figure: Association of average total macronutrient doses in each of the first 7 days in the paediatric intensive-care unit with clinical outcome Data are hazard ratios (HR; 95% CIs) per g macronutrient/kg added. The figure shows associations of average daily doses of the individual macronutrients up to each of the 7 days with the likelihood of (A) acquiring a new infection in the paediatric intensive care unit, (B) earlier live discharge from the paediatric intensive care unit, and (C) earlier live weaning from mechanical ventilation. Results were obtained after adjustment for centre, paediatric logistic organ dysfunction score, paediatric index of mortality 2 score, diagnosis group, and weight-for-age Z scores on admission. HR >1 indicates a higher likelihood of acquiring a new infection (indicating harm), but a higher likelihood of earlier live weaning from mechanical ventilation and earlier live discharge from the paediatric intensive care unit (indicating benefit) and vice versa for HR <1. N represents the number of patients still in the paediatric intensive care unit on the day of analysis. The dotted lines represent a neutral relationship in form of hazard ratios being equal to 1 (border between harm and benefit).

I think it would be a mistake to assume that this means that lipids and glucose were beneficial. The babies on early PN almost all received both amino acids and lipids, and it is only by this data dredging, which might be suggestive but is certainly not conclusive, that the potential relatively different impacts of different components of the PN can be seen.

One strange thing about this trial is the infection outcome data, a 57% incidence of PICU acquired infection in full term babies seems enormously high, (16/28 1 day old babies in the immediate PN group). If you examine the tables the numbers of individual types of infection don’t add up to the total number of infections among the under 4 week and under 1 wk tables, even though there is a group called “others”.

Until the last few years I was starting PN immediately in the preterm infant, but I was willing to wait a few days in term babies, usually only starting the PN if they were likely to be without feeds for 3 or 4 days or more. Like many people I guess, as a result of indication drift, we now often start PN on day 1 when we admit term babies to the NICU.

We should rethink that approach. There may be no benefit to early PN in term babies apart from reducing hypoglycemia, which you can avoid with more glucose. An RCT of lipid and glucose vs glucose alone vs PN in babies expected to need less than 1 week of nil by mouth would be needed to really answer the questions in term babies. I think we really need to study this in NICUs where hospital-acquired infections among full-term babies are very much rarer than the figures in either group in PEPaNIC. In the latest CNN report the rate of late-onset sepsis among full-term babies admitted to the NICU was 1%. Of course, this is in no way comparable to the population studied in PEPaNIC, but it does make me wonder why so many of their babies developed infections, in both the early and the late PN group.

In the supplementary data, there are some tables examining risk factors for infections. Including this one for example which shows that the Odds Ratio for developing an infection were 0.0 in Edmonton, which I presume means that they didn’t have any. I don’t understand how the p-value for that can be 1.00, maybe there is an error there, or I am just not understanding. It is also possible that the definition of “airway infection” is the cause of the high frequency, and perhaps the cardiac surgery babies are at enormously high risk of satisfying that particular definition.

And what about the preterm? The nutritional fragility of the preterm, especially the very preterm, and preventing or reducing the usual protein catabolism after birth, has been the justification for the immediate introduction of PN. What if that is the wrong approach? The trials of earlier aggressive PN have I think not been adequately powered to answer questions of clinically important outcomes. The NEON factorial trial from the UK is one of the better ones, but even that only included outcomes of 133 babies of <31 weeks who either received high immediate amino acid intakes (3.6 g/kg/d) or a gradual increment, 1.7, 2.1 then 2.7 g/kg/d, which also started on day 1. The other arm of the factorial was comparing 2 different lipid emulsions, both arms of which received 2 g/kg/d on day 1 and 3 g/kg/d on day 2. The primary outcome for the amino acid arm was lean body mass at term equivalent which was not different between groups. Among secondary outcomes, the early higher dose AA group had smaller head circumference at term by 0.8 cm, but had slightly higher brain volumes on MRI.

I think it is important that we gain better information about the clinical outcomes of babies who receive early PN.

I suggest we obtain some observational data to start with (already underway in the UK) and then perform RCTs in full-term babies in the NICU, as suggested above we could have a group that just received extra calories as lipid. Then mildly preterm babies, and depending on those results we could consider studying moderately or very preterm infants. Early PN reduces or prevent catabolism in preterm babies, we should figure out whether that leads to better short, medium, or long term outcomes.

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

How much Intravenous lipid should we start with?

Since I started neonatology (just a little over 5 years ago…) we have given intravenous lipid emulsions starting at low doses and progressively increasing. As a reminder to everyone, when intralipid (as I will call it to refer to all soybean-oil based emulsions) was introduced there were no large multicenter RCTs to confirm safety and efficacy; as part of the early development of intravenous neonatal nutrition, it was really trial and error. Babies usually seemed to tolerate it; some developed milky serum with extremely high triglyceride concentrations, but most were without apparent complications and were getting extra calculated calories, so it was assumed that they were metabolizing the stuff and using it to grow.

I think it was the concern about occasional intolerant babies that led us to start at lower doses (often 0.5 or 1 g/kg/day) and progressively increase to 3 or 4 g/kg/day maximum. I don’t think there was ever good evidence that tolerance improves with postnatal age, we just didn’t want to start at 3 g and be surprised to find a baby with enormously high triglyceride or free fatty acid levels.

This gradual approach limits the potential calories that can be administered, as glucose tolerance tends to be limited in the first few days for the very preterm baby, and enteral tolerance often takes a few days before full feeds can be tolerated without regurgitation. So we could ask whether it would be better to start with higher doses right from intralipid initiation. This new trial studied exactly that (Alburaki W, et al. High Early Parenteral Lipid in Very Preterm Infants: A Randomized-Controlled Trial (HELP trial). The Journal of Pediatrics. 2020). The investigators randomized 83 infants of <1500g birthweight who were appropriate for gestational age (i.e. no babies under the 10th percentile) to either start intralipid at 0.5 g/kg/d starting at 12 to 24 hours of age if under 1kg, or at 1 g/kg/d if over 1 kg, and then increasing by 0.5 g/k/d per day up to 3 g/kg/d; the comparison group was to start at 2 g/g/d as soon as possible on day 1 and increase to 3 g/kg/d on day 2. The control group, lower dose, schedule is similar to the protocols of many tertiary centres. The primary outcome of the trial was the maximum percentage postnatal weight loss.

The main finding was of a reduction in maximum postnatal weight loss from -12.7% in the controls to -10.4% with early higher dose lipid (SD in each group approx 4), which was unlikely to be a chance occurrence. The investigators also examined the proportion of babies with extra-uterine growth restriction, which was defined as weight less than the 10th percentile at 36 weeks post-menstrual age (Fenton percentiles). 68% of their controls had EUGR, and 39% of the intervention group. It is very hard to ascribe the EUGR impact to the early postnatal nutrition, the babies in the intervention group overall only received an additional 2.6 g/kg total of lipid, but had 8 g/kg LESS carbohydrate, and 3 g/kg less protein. The energy intake over the first week was therefore almost identical between groups (the early higher dose lipid babies received only an extra 1 kcal/kg/d over the 1st week). It is very hard to believe that changing the source of calories during the first 5 or 6 days of life, without changing the overall calorie intake (total about 30 more kcal/kg from fat, but 25 fewer from carbohydrate and 12 fewer from protein), will have an effect on growth during the whole hospital stay.

This was not a masked study, which looks like it might have led to some compensatory changes in the nutrition, there is no discussion that I can see at all about the enteral nutrition of these babies, most of whom would have had enteral feeds started during the study period, and even whether the enteral nutrition was included in their intake calculations is not specified.

The frequency of EUGR they report is enormously high; with an appropriate, and well-tolerated nutrition protocol we have practically eliminated extra-uterine growth restriction (Lapointe M, et al. Preventing postnatal growth restriction in infants with birthweight less than 1300 g. Acta Paediatr. 2016;105(2):e54-9). I think there is no excuse for having 68%, or even 39%, of VLBW AGA babies arriving at 36 weeks and being <10 percentile.

The study is too small to say much as about the safety of this approach, we can say that many of the babies showed biochemical tolerance of early higher intralipid doses. But mean triglyceride concentrations were higher, and many more babies exceeded 2.8 mmol/L (33% vs 18%), their definition of hypertriglyceridemia.

I think that you can say based on this small unmasked trial, that most infants will tolerate earlier, higher intralipid infusion rates, but hypertriglyceridemia is more common, and the clinical benefits are questionable, perhaps a reduction in maximum postnatal weight loss by 2.3 percentage points. An impact on late-onset sepsis, or other complications of prematurity, would not likely be detected by such a small study. The safety of early parenteral nutrition is a subject for another blog post.

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

How much should we feed babies?

Surely after 60 years of neonatal intensive care, we have figured out what volume of milk to give to small preterm babies? Our local feeding protocol uses a standard of 160 mL/kg/d, which we will increase to 170 mL/kg/d if weight gain is sub-optimal and the 160 is tolerated. With our current protocol, we often have to increase the calorie density of the milk, by adding extra fortification (and go beyond what has been well studied), but maybe we should just increase the volume further. Will babies routinely tolerate 180 to 200 mL/kg/d, and is it safe?

In this new study from the group in Birmingham Alabama, (Travers CP, et al. Higher or Usual Volume Feedings in Very Preterm Infants: A Randomized Clinical Trial. J Pediatr. 2020) 224 babies of between 1000 and 2500 g birth weight were enrolled when they had reached “full-feeds” defined as 120 mL/kg/d. They then advanced their feeds at 20 to 30 mL/kg/d until they reached either 140-160 mL/kg/d or 180-200 mL/kg/d. In either group, if weight gain was considered insufficient the calorie density could be increased. It seems as though donor human milk was not used, supplemental feeds were with preterm formula.

Fluid and calorie intake were higher in the high-volume group, and weight gain (the primary outcome variable) was consequently also greater. This is the comparison of calorie intake between groups.

The high volume group gained weight at 20.5 g/kg/d compared to 17.9 g/kg/d in the standard volume group. As a result, they were heavier at 36 weeks by 165 g on average, had larger heads by 0.5 cm, were longer, by 0.5 cm, and were less likely to have a weight <10th %le, 12% vs 21%.

The average calorie density at study termination was 23 kcal/oz in each group (ranging from 20 to 30 kcal/oz), so they could possibly have achieved as good growth in the lower volume group by increasing calorie density. The feed tolerance was excellent in both groups and was not more of a problem in the higher volume group. There were no cases of Necrotizing Enterocolitis.

The implications for my practice are that we can safely increase feed volumes above our usual range, in babies over 1kg at least, rather than increasing calorie density with more fortifier. This will give our babies more breast milk (maternal or donor, in our unit), which always sounds like a good thing to do!

It is amazing that despite years of academic neonatology there were only 2 previous trials of similar comparisons, one from Sydney compared 150 to 200 mL/kg/d of fortified breast milk or formula in babies <30 weeks gestational age (n=57), and another from Vellore in India compared 200 to 300 mL/kg/d of unfortified breast milk in 64 babies <1500 g. There are so many important simple questions that we have not asked, much of current practice depends on approaches that gradually developed over the years and have never been adequately tested. It is time we changed that.

Posted in Neonatal Research | Tagged , , | 4 Comments

Not neonatology: Julian Bream 1933-2020 RIP

Most of my readers probably don’t know that for a few years I have been struggling to learn Classical Guitar.

One of my inspirations is Julian Bream, who was fundamental in the UK in the recognition of the guitar as a genuine Classical Music Instrument. After I watched one of his performances on youtube 2 days ago I wondered if he was still alive, and discovered that he had died less than 24 hours previously. There is an obituary in the Guardian which describes his contributions to music.

Bream was a genius in his interpretations of early guitar music, and in particular his renditions of lute music that had been re-transcribed for the guitar. He then learned lute, and for a long period was recognized as the best lute player in the world.

I think he was at his best with the modern Spanish composers and also, in particular, with the Brazilian Heitor Villa-Lobos. Here he is with the Villa-Lobos preludes, out of order, of course, the prelude no1, the best known and my favourite, starts at about 2:48. The second starts at 6:35 minutes, and I think he plays it too fast, but that is probably jealousy!

Villa-Lobos was himself a guitar player and wrote some of what I consider to be the best modern guitar pieces. Only one of which I can almost play all the way through!

A part of Bream’s legacy is the many pieces for classical guitar that he commissioned from many leading composers, including William Walton, Michael Tippett, Malcolm Arnold, and Benjamin Britten. The Bagatelles for guitar by Walton are among the masterpieces of 20th-century guitar music; according to the story, Walton didn’t really understand the guitar and asked Bream to send him a chart to explain what the guitarist could do, and what stretches were impossible. The piece by Benjamin Britten even Bream acknowledged as being fiendishly difficult to play, almost too difficult even for him!

There is a great documentary about his life on Youtube which among other great moments shows him playing the lute for a Dowland song with the tenor Peter Pears (Benjamin Britten’s long time lover), improvising jazz guitar with a group in his apartment in London, playing and improvising in a group in India with Sitar and Tabla, playing the Britten Nocturne written for him, and playing with his friend and the other 20th century English Classical Guitar genius John Williams.

A boy from a poor “broken home” in London who affected the lives of millions.

Exactly the kind of thing that the bunch of seriously incompetent fools who are currently in power in the UK want to prevent. The most recent catastrophe of re-grading school leavers exam results (under cover of COVID) to give extra credit to the most privileged, and downgrade those who have struggled all their lives to overcome their disadvantages, will hopefully make voters in the UK sit up and take notice. The profoundly perverse tabloids in the UK, however, will, I am sure, try to pretend that this is all a communist plot!

Goodbye Julian Bream, bon voyage.

Posted in Not neonatology | 1 Comment

ELVIS is alive!

I was searching through the lyrics of ELVIS songs for a quote that might be more apt for this new blog post than “It’s now or ever”, which I have used previously, when I realized that I am a loser, and I should spend more time with my kids!

The 2 year neurological and developmental outcomes of the ELVIS trial (Early vs Late Ventricular Intervention Study) have been published. Cizmeci MN, et al. Randomized Controlled Early versus Late Ventricular Intervention Study (ELVIS) in Posthemorrhagic Ventricular Dilatation: Outcome at 2 Years. The Journal of Pediatrics. 2020. Hopefully, you will remember (surely as a result of my previous ELVIS blog post) that this was a much-needed RCT comparing 2 strategies for intervening for post-haemorrhagic ventricular dilatation in preterm infants. You can imagine the importance of this study when you note that Linda de Vries, Floris Groenendaal, and Andy Whitelaw were all three involved in organizing and performing this trial. Post-haemorrhagic hydrocephalus (PHH, as I will call it from now on) has become less frequent over the years, as severe IVH has become less common, but it remains a major problem for a small number of babies.

In ELVIS, preterm infants under 35 weeks gestation who had an acute IVH with ventricular dilatation followed by progressive ventricular dilatation, were randomized to either of 2 approaches; in the early group:

treatment was started after the Ventricular Index had crossed p97 (97th percentile) but before crossing the p97+4 mm line, the Anterior Horn Width was >6 mm but <10 mm and/or the Thalamo-Occipital Distance was >25 mm. Intervention started with LPs (max 3), and if necessary, followed by insertion and taps from a Ventricular Reservoir, aiming for VI<p97 over the next 7–10 days.

In the late treatment group, the same sorts of interventions were performed, but they started when the ventricular index was greater than the 97th percentile + 4 mm, and the anterior horn width was over 10mm. This later threshold was based on old observational data, from Malcolm Levine I believe, that showed that once you passed this threshold progressive hydrocephalus was frequent. It had become a sort of default standard of care without any good data that it was the best indication for intervention.

In both groups, if a reservoir was inserted, it was tapped regularly

Ten mL/kg were removed once or twice a day, the volume adjusted according to cranial US. When taps from a reservoir were still needed 28 days after insertion to keep the ventricular index well below the p97+4 mm, one or two ‘challenges’ were performed with discontinuation of taps. Reservoir taps were resumed in case of expanding ventricles, clinical symptoms and/or excessive head growth. Taps were continued until infant’s weight reached 2000–2500 g and according to unit protocol, the protein had decreased to <1.5 g/L and erythrocytes <100/mm3, at which stage the infant became eligible for a VP shunt.

The primary outcome for the early phase of the study was whether or not you survived without needing a shunt.

As you might imagine, the early treatment group had many more LPs; and needed many more reservoirs, 40/64 compared to 27/62. They did not, however, need more VP shunts, in fact, slightly fewer shunts were needed with early treatment, 12 vs 14.

Of course what we really want to know is the long term outcomes, if they had been similar, then less intervention could be considered to always be a good thing, but if there was a difference between groups, then the increased need for intervention in the early group should be weighed against the outcomes. The 2 year follow up study analyzed outcomes from about 90% of survivors, so is a reliable indication of what outcomes to expect with these approaches.

Bayley scores were better with earlier intervention, and there was less cerebral palsy; there was a relatively small sample size, and therefore the differences shown may have been random effects. This figure suggests that, among those needing shunts, the differences were greater than could be expected by random variation.

Adjusted analyses showed that early intervention seemed to be associated with a decreased risk of an adverse outcome after correcting for gestational age, the severity of intraventricular haemorrhage and cerebellar haemorrhage (adjusted odds ratio: 0.24, 95% confidence interval [CI], 0.07 to 0.87. In this situation, I think adverse outcome means death, or cerebral palsy, or a Bayley version 3 motor or cognitive composite score more than 2 SD below the mean.

It is interesting that in the early treatment group 94% of the cognitive scores were above -2SD with early treatment, and 87% with later treatment. The large majority of preterm infants with PHH who were enrolled in this trial, therefore, did NOT have cognitive delay! There is obviously a great deal of selection that goes on before an infant would be randomized in a trial such as this, but good outcomes of PHH are not rare! Motor scores were more severely affected than cognitive scores, but were also somewhat better in the early treatment group.

The clinical implications of this trial to me are that PHH should be treated according to the early intervention thresholds in ELVIS. The differences in outcomes are not enormous, but all favour earlier intervention. The only advantage of waiting until the later thresholds are crossed is that there are fewer lumbar punctures and fewer reservoirs, but no reduction in permanent shunts. I also think the study confirms that the Levine thresholds from many years ago were reasonable, and, I think should now be considered to be late treatment. Once the late treatment thresholds are reached intervention becomes urgent and should not be further delayed, unless there is an understanding with the parents that interventions will be limited. If we are aiming for the best possible long term outcomes, then the early treatment thresholds of ELVIS should be followed.

Otherwise, we may be accompanying ELVIS singing “Maybe its too late, I sometimes even hate myself…”

Posted in Neonatal Research | Tagged , , , | 6 Comments

Does erythropoietin prophylaxis prevent NEC? Unreliable data.

Routine erythropoietin administration from early life in very preterm babies has been postulated to be neuro-protective. Unfortunately, despite some early promising results, 2 large well-performed RCTs showed no benefit, including the largest with 750 babies under 28 weeks, which showed very small differences between groups and nothing that appeared to be a real effect of the erythropoietin.

Another trial of about the same size, but with less immature babies (800 of less than 32 weeks gestation) used a dose about half of those other 2 trials, it was published in Annals of Neurology for some reason. Of note, it was registered on after it was completed, and the registered primary outcome variable is MDI<70. Even though they registered the trial after they performed it, the published primary outcome variable is not the same, in fact, the published manuscript methods section does not mention a primary outcome variable, the results show in table 2 as “Primary Outcomes” 3 outcomes, Death, Disability, and Death + Disability. The sample size reported in the publication is 743, the registration document (submitted after the trial was finished) states the sample size is 490.  For several years the Annals of Neurology has required that RCTs are registered, that the registered primary outcome is the same as that reported by the article, and that the publication complies with CONSORT guidelines. Fail, fail, fail.

You might ask why I am spending so much time criticising a journal and an article when it was published in 2016? That article purported to show a substantial reduction in major intracranial haemorrhage with erythropoietin administration (EPO) from 16% to 7%, an enormous reduction in PVL from 20% to 11%, a reduction in sepsis from 26% to 20% and a lowering of “Death + Disability” from 27% to 13%. The Cochrane review of outcomes of early erythropoietin in preterm infants (which you can reach full text via the Vermont Oxford website is heavily dependent on that trial “Song 2016”. For serious IVH, for example, that trial contributes 73% of the weight in the analysis, for PVL it is 90%. If the data are unreliable, which I contend they are, then the systematic review is unreliable also. The Cochrane Systematic Review (SR) does not include the results of the most recent large trial Juul et al, as it was published after the SR was completed, and after the first author of the SR Arne Ohlsson sadly died, last year. That trial from the PENUT consortium showed NONE of those effects, no difference in IVH, PVL or sepsis or in developmental delay.

The Song 2016 trial is also the only trial in the SR that shows a reduction in NEC. As a result of only that trial the overall impact of early EPO appears to show a reduction in NEC. The same authors from Zhengzhou appear to have carried on with their same poor research practices, and immediately performed a subsequent trial (Wang Y, et al. Erythropoietin prevents necrotizing enterocolitis in very preterm infants: a randomized controlled trial. J Transl Med. 2020;18(1):308). Song 2016 enrolled patients until June 2013, (or perhaps January 2013 according to the registration document) the new trial using the same dosage regime (500 units/kg every 48 hours for 2 weeks) started recruiting in January 2014. Enrolment continued until June 2017 and the trial was registered 2 years later in April 2019. 1285 babies of less than 33 weeks gestational age were randomized. The primary outcome in the registration document is “Incidence of NEC” not further specified. In the publication which, strangely, has appeared in BMC Journal of Translational Medicine, the primary outcome is the same, but whether all grades of NEC or stage 2 and higher was the primary outcome is never specified.

If we concentrate on “confirmed NEC” as they refer to it, which is Bell grade 2 or higher, it is very frequent in this trial, 17% of the controls under 28 weeks, but the numbers of babies involved under 28 weeks is just 110 (53 controls and 57 erythropoietin). The incidence of confirmed NEC in more mature babies is also very high, at 30 to 36 weeks it is over 3%, compared to, for example, recent data from the Canadian Neonatal Network which shows an incidence of 1% at that gestational age.

Strangely, in both the new trial, and in Song 2016, the controls received a saline placebo, but the trials are reported as not being masked. I am not sure why you would go to the trouble of preparing and administering a placebo saline injection if the caregivers knew it was a placebo! Interestingly, the CONSORT flow chart shows that there were 1327 potentially eligible infants screened and only 20 refused consent, which is the highest consent rate I can remember seeing for a long time.

Mortality was 50/644 controls, and 34/641 EPO babies, which looks like a difference which is likely to be due to chance, but amazingly enough there is no statistical test of mortality in the manuscript, and compatibility intervals for mortality differences are also never reported.

What also surprises me about this article is the failure to report any other clinical outcomes. NEC and mortality are reported, but the outcomes that this group has reported previously as being reduced by EPO (IVH, severe IVH, PVL, sepsis, retinopathy) are nowhere to be seen.

The Journal of Translational Medicine is supposed to adhere to CONSORT standards, but they are woefully lacking in this publication. One might wonder why a multicentre clinical trial in preterm infants with major potential clinical importance was submitted to this journal; perhaps it was not the first choice, or perhaps the editorial standards suited the authors, the article was submitted to that journal May 20th this year, and accepted July 26th. Many open access journals publish the names of reviewers, and make available all versions of the article from initially submitted to final revisions, but BMC Journal of Translational Medicine does not.

The data in this publication are not consistent with the results of Juul, who showed a very minor difference in NEC incidence among infants who were all less than 28 weeks gestation; 6% with EPO and 8% in controls for stage 2 or 3 NEC. I discussed that high-quality trial previously

This new trial creates a big problem for all of us interested in evidence-based practice. I think that the data are unreliable, but the sample size is so large that the results will overwhelm any attempt at systematic review of EPO impacts on NEC. Fortunately, there are few babies under 28 weeks, but even an analysis of EPO for prophylaxis of NEC among babies under 28 weeks will be heavily impacted because of the large number of events. I guess we can be grateful that all the other usual clinical three-letter outcomes were NOT reported! At least SR of early EPO impacts on IVH, PVL, BPD, RoP and sepsis won’t be impacted by the results of this trial.

Unfortunately, the same group appears to be continuing to perform their studies. There is another registration document on for a trial by this group (registered April 2016, enrolment was expected to start in May 2016 and be complete in December 2019) which is apparently underway. This, of course, overlaps with the trial just reported which finished enrolment in June 2017. In this most recent publication, they refer to the next trial as being among infants of under 28 weeks gestation, and examining retinopathy. In the registration document retinopathy of prematurity (not further specified) is listed as one of several short term secondary outcomes. The primary outcomes are listed as being two separate outcomes Mortality and the Incidence of neurological disability at 2 years old.

I for one won’t be giving routine EPO to my preterm babies to prevent NEC based on this data. For now, all that I think we can be sure about is that early routine EPO stimulates red cell production, which, depending on transfusion thresholds, is very likely to reduce blood transfusions. Any other potential benefits are unproven.

Prospective registration of RCTs has numerous benefits, including ensuring the completeness of the scientific record, ensuring that the analyzed primary outcome variables are those initially planned, and avoiding the addition of post hoc outcomes, unless they are clearly reported as being post hoc outcomes. Allowing publication of RCTs which are registered after they have been completed negates all those benefits, even when the authors still can’t get the sample size or the primary outcome correct!

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