Resources & Evidence Contact us

PNG Conference: The pro and cons of probiotics for preterm infants: Why I am still uncertain!

Results and outcomes described are anecdotal and specific to the institution and its protocol. For a list of published studies of EHMD, click here.

Learning Objective:

  • To develop an understanding of the strengths and weaknesses of evidence provided by systematic reviews and meta-analyses to inform our practice regarding the use of probiotics in preterm infants


Roger F. Soll, MD
Vice President of the Vermont Oxford Network
Coordinating Editor of Cochran Neonatal


Hi, I'm Roger Saul, and I have the privilege of discussing the pros and cons of probiotics for preterm infants, not so much to debate debate the point, but to discuss a little bit about why I'm still uncertain about exactly what we should be doing. As a matter of disclosure, I am the Vice President of the Vermont Oxford network, and the coordinating editor of Cochrane neonatal, and much of the data that I presented on trials will be coming from work that has been done with Cochrane neonatal, I have no other relevant financial issues to disclose. In this session, I hope we can develop an understanding of the strengths and the weaknesses of the evidence provided by the systematic reviews and meta analyses that inform our practice regarding the use of probiotics in preterm infants. First, we need to agree, there is a real problem that needs to be addressed by some intervention, probiotics or other, and that is necrotizing enterocolitis. Although we've seen some small decreases over the past two decades, we still see necrotizing enterocolitis in any place from five or 6% of our very low birth weight infants. We think we know that there are many factors that contribute to this disease, we know that there's the underlying compromise of the newborns gastrointestinal tract, immaturity, inflammation, hypoxia, we know there's a role that bacteria play in the role of NEC. And we think that there are issues regarding feeding from volume to advancement, to timing to what the substrate is that we feed babies. I have a pet theorem. And that is that the number of the arrows on the slide is inversely proportional to our understanding of the process. And I have to tell you, there are a lot of arrows still on the NDC trial, which tells me, there's a lot of research still to be done. Let me discuss some of the basics around NEC so that we can agree on the scope of the problem we need to address. We do know a few things about necrotizing enterocolitis. We know that breastfed babies have a lower incidence of NDC than formula fed babies. But it's unclear regarding the role of various feeding regimens in the etiology of NDC. Although conventional wisdom recommends slow initiation and advancement of enteral feeds for preterm infants, the randomised trials don't necessarily show that we can decrease the risk of NDC. in infants we advanced feeding more slowly and are more cautious with. So then the question of today's webinar, and that is what is the role of probiotic supplementation. There is, as Dr. Underwood undoubtedly has talked about a very strong biological rationale for why probiotics might be useful. Here's a quote from Michael specter, which makes it very clear, we are inhabited by as many as 10,000 bacterial species. And these cells outnumber those of which we consider to be our own by 10 to one. So this microbiome is obviously critical to human life and human health. And something that we do not understand well. The role of these bacteria is complex. They're responsible for synthesis of vitamins, digestion of harmful or non digestible compounds, energy production, and they may be involved in the protection and stimulation of our immune system, as well as providing a physical barrier against pathogenic bacteria.

Now, it's more complex than this. But at birth, the infant's GI tract is essentially sterile, we do know that there is migration of microbes into the gut, even as a fetus, but basically, there's nothing growing there, compared to what will grow as colonization begins, which is immediately after birth. But the initiation of enteral feeds and is well established within the first few days. intestinal flora varies widely from person to person, and is a huge number of bacterial cells, comprising more than 400 different species. Here's a nice slide from work that looks at the microbiome in infants who delivered vaginally versus those by this is arion section, and it points out the important role of colonization through vaginal birth and sort of the appropriate colonization through mothers that is disturbed by even something as simple as zerion birth and not passing through the birth canal. In formula fed infants, we know that the microbiome is clearly altered coliforms enter coxy and bacteroides predominance And agents like by fit of bacterium and lactobacillus are only occasionally present. That's very different in breastfed infants, by fit of bacteria and lactobacillus predominate in breastfed infants with other enteric organisms being present less frequently. This pattern is even further different in preterm infants, particularly those experiencing neonatal intensive care, whether that's from antibiotic use, delayed initiation of enteral feeds, they may all influence the type and amount of microorganisms that colonize the GI tract. Let's take a look at just how abnormal This is. In the extremely low birth rate infant and LBW infants, they're colonized by fewer than three bacterial species by the 10th day of life. And species of the probiotic bacteria that we'll be talking about. The fit of bacterium and lactobacillus are found in the stool of less than one in 20 of the infants studied within the first month of life. By a month of life, the predominant organisms are Enterobacter, coagulase, negative staph, which if you think about it are actually the organisms most responsible for nosocomial infection in neonatal intensive care. So what's the solution we're talking about today, with the possibility of using probiotic organisms. Probiotics are live non pathogenic bacteria that normally reside in our GI tract. And it has been postulated that introducing probiotics to preterm infants might be beneficial. In order to avoid the overgrowth of pathogenic organisms. They've been proposed to enhance our tolerance of feeds prevent necrotizing enterocolitis, prevent nosocomial infection, and even impact on mortality in preterm infants. These agents are widely available without prescription. And a variety of these agents are available for study. Not that we're talking about the issue in term babies, but this is how far it's gone. There are marketed infant formulas that include probiotics.

So let's talk a little bit about what the trials tell us about probiotics. What are the benefits and harms that have been shown in trials? The desirable effects, potentially undesirable effects? And importantly, how certain are we that the evidence is an accurate reflection of the effects of probiotics? So let's start with a simple question to probiotic agents improve growth and feeding tolerance? Well, I'm not going to go through a meta analysis of dozens and dozens of studies but give one good example to show that yes, in fact, we see better growth in infants who receive probiotic agents. Here's a single study from Kitajima in 1997, which illustrates this point. Well, it's a study of 91 infants, randomized either a bifidobacterium or control. And what you see here, in the black dots are the non colonized infants whose feeding volume toleration is measurably less than the infants who are receiving and colonized with probiotics. Similarly, you see, looking here at growth, body weight over time in weeks, we see improved growth in the infants who received and are colonized with probiotic agents. Well, that's great, but I think we're all mostly concerned about the more major clinical impacts on outcomes such as infection necrotizing, enterocolitis, and mortality. There is we have through Cochrane neonatal recently published an updated review of probiotics in the prevention of necrotizing enterocolitis, in very preterm infants, and this work from Sharif and colleagues is available to you. If you go on to the Cochrane Library, you can access that through Cochrane advance or through the Wiley Cochrane Library. One thing that's different about this meta analysis because if you search the literature, you'll find many is that the meta analysis is restricted to the very low birth weight infant. And so this isn't just a general population feeding of newborn infants with probiotics, but it's focused on the group that we have the most concern, specifically regarding issues like necrotizing enterocolitis. This is a very large review in involves 53 trials and over 10,000 very preterm infants. The risk of bias does vary, particularly regarding methods of allocation concealment, and lack of blinding is mostly though reasonable, but there is a great deal of the variation of the timing of when probiotics is given. The dose, the length of treatment, the formulation of the probiotics, and the feeding regimens that the infants were undergoing In fact, there is literally a tsunami of evidence

to look at. Here's an example of just one of the forest plots, looking at the outcome of necrotizing enterocolitis. And you can see that in this plot, and you'll have to look carefully and put your glasses on, that is broken down by organism. So we see here, that the fit of bacteriums studies, here, the lactobacillus studies, and the other agents and we also see an overall estimate here, which is very profound. But let's take the long view and look at the forest for the trees. Here are the results. The summary of the meta analyses on the big three outcomes were concerned about infection, NEC and mortality, we can see for infection, there are 45 studies that were conducted involving 9532 infants, there is a marginal decrease in the risk of infection, expressed here is relative risk with the loss of identity being one. But I've also calculated the risk difference, the absolute risk difference, saying that there will be an absolute difference of 2% risk of infection, which may be as much as a 4% risk of infection, or 1% risk of infection. Similarly, a large number of these studies 52 studies involving 10,306 babies reported on necrotizing enterocolitis. Here we see a very profound decrease, and almost 45% decrease looking at the point estimate, and a very narrow confidence interval, suggesting a very specific and precise effect. For numbers needed to treat the point estimate is minus 0.03, ranging from 4% less to 2%. Less. Similarly, a large number of studies report on mortality, and there is a similar reduction that is reported in mortality, approximately a 30% reduction in the risk with a point estimate of minus 0.02. These results, when you look at meta analyses would on the face of it appear to be very strong and precise results. But I also point out to you this issue of risk difference, even looking at the point estimates here, it says that to gain these benefits based on these studies, and these estimates, you'd have to treat someplace between 33 and 50 kids. And if you look at the confidence interval, it really could be as many as 100 kids that need to be treated before we see these benefits. I bring that up because anything you need to treat so many infants with to gain that effect, you also have to worry about safety. In the clinical trials, there were no probiotic related sepsis cases reported or other specific adverse events. So much so that people who I respect deeply like Keith Barrington, on his blog, say that he has difficulty understanding the lack of clear guidance from any of the major societies regarding the use of probiotics in preterm infants. So, you might think I'm busy making Dr. Underwood's case here. And it's important that we agree on what the possible benefits are. But let me spend the rest of this time talking about the possible risks and why I remain uncertain about whether or not this should be part of our routine practice. Because there's more to discuss. First, the question is, do we need more trials? There are few things you'll look up in the neonatal literature that are better studied in terms of numbers of trials and numbers of infants enrolled? Well, there are some specific questions to answer before you consider using probiotics. There are multiple agents that are available, there are multiple dosing strategies. Until recently, there were a few extremely low birth weight infants enrolled in trials. And in many studies, few exclusively breastfed infants. And importantly, and we'll get to this towards the end, there is no product that has cleared the regulatory hurdles. So let's try to take a look at some of these points and see if we can dissect them and understand them further. So let's start with this whole idea that there are many, many products that have been tested. One thing of some interest is that I'm not so sure that bugs are bugs, I think that it may matter what probiotic probiotic agent you choose. studies that have solely used bifidobacteria have shown distinctly less impact on the important outcomes of NEC and mortality. The same can be true of studies restricted to lactobacillus In fact, studies only using the fit of bacteria and studies only using lactobacillus don't show us any significant impact on mortality. Only studies with mixed combinations of probiotic agents begin to show us the types of effects that would motivate To think that this is an important new therapy.

What about the certainty of the evidence, this is a word that you've probably heard me use in other talks that comes from the whole idea of grade and the various domains of grade. When we think about the certainty of the evidence, we think of the factors that can lower the quality of the evidence, we think about the limitations of design and execution, we think about inconsistency or heterogeneity. We think about indirectness Is this the patient population I'm interested in is it applicable to my population, if I work in a lower middle income country, if I look into high income, country, etc, we look at imprecision, which is mostly a statistical issue based on sample size. And lastly, we'll get publication bias, which we'll inspect more closely for the probiotic studies, namely, which studies get into publication and which don't. So let's start with just a general overview. First of all, despite the fact that there are these 53 studies, the median sample size is 149 infants that's relatively small. And in fact, there are only two large trials, enrolling over 1000 participants. And we'll look at them more closely to see if they give us a somewhat different snapshot of probiotics. Some are unblinded, which is problematic. But in general, they have a lower risk of bias in terms of issues like blinding. Here's one of the issues that came up from Sharif and McGuire's meta analysis. And that is the so called funnel plot, a scatter plot using meta analyses to detect the presence of publication bias, you can see it's plotting the study precision the standard error against the log odds ratio. And in normal circumstances, what we would expect if there's no publication is a symmetrical distribution. So an ideal funnel plot is one where the included studies have scattered on either side of the overall effect line in a symmetrical manner. Severe asymmetry to either side is an indication that publication bias may be present. And let's take a look at the outcome that we're most interested in, namely, the reduction of any see. And you can see, there's actually a highly asymmetrical funnel plot for the studies of probiotics and their impact on any seat. How can you postulate that study bias would occur? Well, this is not a regulated product, people can go and get it off the shelf and use it off, you know, clinical trials, and then only choose to either present their data if the study is positive. And frankly, journals are more interested in publishing the data if it's positive. So it's an interesting skew in the data that suggests that some bias may have crept in. Let's take a look at those results from the two large trials that make up from around 35 to 40% of all patients in the trials, we can see that first of all, despite the fifth of their large trials, there are broad confidence intervals. We also see that neither the large Jacob style the pro pro premier trial done with ABC dockless, which is a mixture of probiotic agents, or Kate Costello's trial, the pyp study, which was used use bifidobacterium, they don't actually individually show an impact on mortality or sepsis. And really only Jacob study shows an impact on necrotizing enterocolitis. And again, because the risk of NEC is real, for infants in their families, even a large reduction of 50% reduction, maybe going from only 4% to 2%, and may involve treating as many as 50 babies to see these results. So the other things we need to think about is the acceptability of the intervention. And here, I hear a lot of positive feelings about probiotics. Here's a study that was done by fishermen colleagues, asking parents who are presented with some of the summaries of the data that we've just reviewed, how they think about probiotics, and actually, they're perhaps more positive than I am sounding. They thought at least an information sheet about probiotics, their availability, and what they've been proven to do would be very useful. About one in 10. We're worried about the use of live bacteria in their babies and a similar number. We're worried about the unknown risks, but invariably, 96% felt they wanted to be informed about this and two thirds wanted to have options around treatment. So strike one for probiotics in terms of parent advocacy.

So what am I still worried about though, in this unregulated environment? Well, I'd like to briefly share two cautionary tales of some humor, but also some seriousness. One is with a macro agent, the cane toad, a native to South and Central America. It is a voracious predator of insects and other small prey. And it was thought that maybe bringing this Toad into Australia would help reduce their problem with pest beetles in the sugarcane industry. Well, the problem is that there was no natural predator to cane toads. On top of it, cane toads didn't really want to eat the sugarcane pests in Australia. And you can see in the map at the bottom, that from their initial spread, they have spread to the entirety of Northern Australia, and consider to spread at a rate of 40 to 60 kilometers per year. If you're in a strange mood, you can google and look at YouTube videos of what happens when they hatch, and the roads are literally littered with cane toads. So not such a great idea, when we unleash a biologic agent, to take care of our problems with our limited understanding of all the downstream consequences. But here's something a little bit more specific to introducing bacteria as the story of serratia marcescens. So ratio was used early on in the 1970s, sort of as a biomarker because of the glow in red pigmentation. It can be used to take a look at issues around spread of toxins in water supply, or in the air, etc. And so in the 1950s, into the 60s, the army experimented with trying to understand these phenomenon by spraying serratia marcescens into the air in San Francisco, into the water systems in New York, in the subway system in New York. And what happened is that they were growing reports of serratia marcescens. Starting to infiltrate, the hospitals grow in all of the wet areas in hospitals and develop severe antibiotic resistance. Probiotics are not immune from this sort of phenomenon. They are not exempt from the natural processes that govern antibiotic resistance. And they could potentially be a double edged sword. As they enter into our environment in the nick you. We already know that there have been problems with unregulated probiotic agents. We know the report of fatal gastrointestinal fungal disease and infant who received Low and behold the same product that was used in the pro prims trial of product that we would normally have thought, Wow, that's great. This has been tested to some extent. FDA confirmed the presence of this and alerted us healthcare providers and certainly put some constraint on our willingness to consider widely using probiotics prop products, and probably is why we're still discussing this today. So what should we actually be doing? Well, there are some suggestions in terms of practice, dish, Bondi and colleagues have published this in 2011. And it's evidence based guidelines for considering the use of probiotics in preterm neonates. And I think there's some points here that are worth discussing. But I think you'll see it's a pretty rigorous and demanding approach. If you're going to consider probiotic agents. They suggest be reasonable to use probiotic products that have previously been shown to be effective in randomised controlled trials. Well, that's all well and good, but as I just mentioned, one of the ones that was proven most effective in the pro prems trial is in fact the one that had concerns about a fungal contaminant. They suggest that you should use your on site expert, microbiological support, to have independent culturing, confirmation of what bugs are actually in the probiotics and exclude that there are contaminants in the product you're using. So they suggest having a partnership with your laboratory. If you're going to use this and in fact, batch testing the probiotics, which I doubt most people are considering or doing. What are most people doing? Well, it's interesting, the use of probiotics in very low birth weight infants has steadily increased over the past nine years. We have original rates of around 8% and they are now hovering above 17%.

I like assuming That it's not that we pick and choose. And since the some of that goes on, but I'm assuming that someplace around 15% of centers are now routinely using probiotics, and whatever at risk population they have identified. I'll go back to the parent input here. And the search that was done by C sharp parents, I believe want information about whatever choice we are making in our units. These are available treatments. And I think it's very fair fair to consider whether or not parents should have a voice about starting these treatments in your unit. And it's something to consider with whatever parent advisory groups you have, or other groups like the parent group on NEC, but information they use for parents making this decision. I say this, because despite the slight lowering and rates remains a serious problem in our Nick use. And I look forward to our further research on this issue issue. Specifically, which products should we be using? How do we obtain regulatory approval for these products? as I alluded to, what are the roles of families and parents in making this decision, either in approving the use on their baby in your unit, or, and I think probably more importantly, helping us further our research agenda. There are many issues and many alternatives to consider. lactoferrin has been tested and is proven a disappointment, but other prebiotics compounds that induce the growth of bacteria of the beneficial microorganisms synbiotics combinations of probiotics and prebiotics, and perhaps most importantly, learning how to better supply mother's own milk to babies how to better harvest and give milk to babies and improve our breastfeeding rates and our delivery of breast milk may well be as important in our research agenda, as some of the problems I've alluded to with probiotics themselves. And so I've shared my uncertainty with you. Not a negative view, I hope, but one that explains why I have personal caution, and perhaps explains why it's only 17% and not 80% of VLBW w babies that are currently getting probiotics. So I'll stop here and I look forward to our discussion section. Thank you