Early onset sepsis is a serious condition with a substantial morbidity, and, thankfully, a relatively low mortality in recent years. Prompt recognition and early treatment are essential, but early clinical signs and risk factors tend to be non-specific. As a result many infants are evaluated for sepsis and treated empirically while waiting for culture results. The proportion is amazingly variable between hospitals, in an article from a couple of years ago the California Perinatal Quality Care Collaborative showed that the proportion of newborns receiving antibiotics ranged from 1.6% to 43% in various hospitals. Some of which was due to differences in hospital characteristics (and included preterm infants), but much was due to differences in practice patterns.
Among term infants and those very near to term (35 weeks and more) somewhere around 5 to 10 % currently have a sepsis evaluation, which almost always leads to temporary treatment with antibiotics. Despite the very high sensitivity of modern culture methods, it is difficult for some physicians to stop antibiotics when the cultures are negative, so “culture negative sepsis” is a frequent diagnosis. The actual incidence, or even the existence of such a phenomenon is uncertain, as very many studies have used the prolongation of antibiotics as a diagnostic criterion. Which leads to the following circular reasoning:
- 1. Infants with the following criteria were evaluated for sepsis and then had more than 2 days of antibiotics, therefore they had “culture negative sepsis”,
- 2. In the future we will use those criteria to define “culture negative sepsis” and treat the babies with a full course of antibiotics.
- 3. Lo and behold, the babies do well, so we must be doing the right thing,
- 4. We will continue to use the same criteria to diagnose “culture negative sepsis”
This is all compounded by the use of “inflammatory markers” such as CRP as part of the criteria. The criteria used in many centres are a combination of perinatal risk factors and higher concentrations of CRP, or perhaps procalcitonin, than the concentrations found in healthy normal babies without those risk factors.
Anyone reading this blog for a while will now my attitude to those markers. The new article that I will shortly discuss used a combination of “risk factors” and an elevated CRP >10 or procalcitonin (above “reference intervals for postnatal age”). About 10% of noninfected babies overall have a CRP over 10, I am not sure what the proportion is, however, of non-infected babies from births with “risk-factors” that have an elevated CRP. Especially as nowhere in this publication or in the attached protocol are the risk factors defined. One risk factor for sepsis, which is associated with a very high rate of empirical treatment, and a very low frequency of actual sepsis, is maternal chorioamnionitis. It is known that maternal fever and chorioamnionitis increase neonatal CRP even when the infant is not infected, as do fetal distress, prolonged rupture of membranes, prolonged labour, and meconium aspiration syndrome.
The trial I am discussing, (Keij FM, et al. Efficacy and safety of switching from intravenous to oral antibiotics (amoxicillin-clavulanic acid) versus a full course of intravenous antibiotics in neonates with probable bacterial infection (RAIN): a multicentre, randomised, open-label, non-inferiority trial. Lancet Child Adolesc Health. 2022;6(11):799-809.) with one of the longest titles of recent years, enrolled babies of at least 35 weeks gestation, who had an evaluation for early onset sepsis because of risk factors, or transient clinical signs, and who were well at 48 hours after the blood cultures were obtained, which were negative. As mentioned the risk factors are not defined in this publication, but the reference they give is to the NeoPINS study which defined the risk factors as any one of: maternal GBS positive; ruptured membranes >18h; GA <37 weeks; or maternal chorioamnionitis, defined as either fetal tachycardia or a maternal fever >38.5.
If this trial used those criteria, then 100% of the babies born at 35 and 36 weeks would have had a sepsis evaluation and antibiotics, which I think is crazy, to use the scientific terminology.
The combination of one of those risk factors with a negative culture, a baby who was well at 48 to 72 hours, and either a CRP >10 or a PCT over the postnatal age defined limits, made a baby eligible to be randomized to either continued IV antibiotics, or to switch to an oral suspension of amoxicillin and clavulanic acid.
The worst thing about this trial is labelling these healthy, probably uninfected babies, as having “probable bacterial infection”. They did not.
The primary outcome was the re-infection rate, defined as a clinical infection associated with either fever or hypothermia and an increase in inflammatory markers, prior to 28 days of age.
The primary outcome occurred in 1 of 252 IV and one of 252 po babies.
So if you don’t need antibiotics, it doesn’t matter whether you get them intravenously or orally.
The current AAP recommendations are to stop antibiotics in well-appearing infants after 48 hours if the cultures are negative and not to continue simply based on lab results (such as a raised CRP or abnormal white cell count). That is an evidence-based recommendation that I firmly agree with, and would have meant that the large majority of the infants in the RAIN trial, probably all of them, would have been sent home without antibiotics.
Many infants in the RAIN trial would not have had a sepsis work up at all if they had used a sepsis calculator instead of these simplistic perinatal risk factors (Achten NB, et al. Association of Use of the Neonatal Early-Onset Sepsis Calculator With Reduction in Antibiotic Therapy and Safety: A Systematic Review and Meta-analysis. JAMA Pediatr. 2019;173(11):1032-40), probably about half of them would never have been evaluated, and large numbers of the 35 and 36 week infants would have escaped antibiotics.
Systemic antibiotic therapy in the newborn, especially when prolonged, is not benign.
Messing up the neonatal microbiome, which has evolved along with us over many millions of years, should not be taken lightly. Oral antibiotics may even be worse than intravenous, depending on the IV antibiotics used, some have little intestinal excretion, whereas amoxicillin clavulanic acid is great at killing bifidobacteria.
The human intestinal microbiome is affected for months when you give that combination to adults. The long term impacts of giving systemic antibiotics, during the phase of initial development of the intestinal microbiome, are only now being appreciated. (Patangia DV, et al. Impact of antibiotics on the human microbiome and consequences for host health. Microbiologyopen. 2022;11(1):e1260).
Some of the derangement in the microbiome after neonatal antibiotic treatment has abated by 12 months of age (Reyman M, et al. Effects of early-life antibiotics on the developing infant gut microbiome and resistome: a randomized trial. Nat Commun. 2022;13(1):893), but the long term clinical impacts are numerous and important.
Neonatal antibiotic exposure has been linked with asthma, other forms of recurrent wheezing, colic, coeliac disease, abnormal development of recognition processes, reduced linear growth, and increased obesity, not to mention eczema, inflammatory bowel disease, and type 1 diabetes. (Duong QA, et al. Antibiotic exposure and adverse long-term health outcomes in children: A systematic review and meta-analysis. J Infect. 2022;85(3):213-300).
The RAIN trial showed that switching to oral antibiotics meant that the child could go home sooner, and had fewer iv attempts and therefore less pain, which are good things. But even better would be just stopping antibiotics when the baby doesn’t need them.
A mentioned above, all the RAIN trial really tells us is that if you don’t need antibiotics, then the risk of possible infection in the first month of life is the same if you give oral or IV antibiotics for a week.
I have corrected the title of the trial :
Efficacy and safety of switching from intravenous to oral antibiotics (amoxicillin-clavulanic acid) versus a full course of intravenous antibiotics in neonates who probably are not infected (RAIN), unfortunately without an untreated control group: a multicentre, randomised, open-label, non-inferiority trial.
Which we could rename as RAINING (Reduction of intravenous Antibiotics In Neonates, In a Non-infected Group).
Excellent article. Fully agree with your comments on misuse of CRP results as a justification for prolonging antibiotic duration.
Preach it brother!!!!