Purpose
Very low birth weight (VLBW) infants are at increased risk of spontaneous focal intestinal perforation (FIP). FIP is characterized by a localized intestinal perforation with no or minimal tissue inflammation. Unlike necrotizing enterocolitis (NEC), which typically occurs later in VLWB infants [
1‐
3], FIP is commonly diagnosed within the first 10 days of life. Another differentiating aspect is the characteristic excessive inflammation occurring in infants with NEC [
3,
4]. Both, FIP and NEC are presumed to have multifactorial causes [
5‐
8].
Along with extreme prematurity, the use of non-steroidal anti-inflammatory drugs and steroids are key factors for FIP pathophysiology [
9,
10]. Standard of care treatment includes cessation of enteral food or medication, supportive care, intravenous antibiotics, peritoneal drainage and surgery [
11]. Clinical practice guidelines published by the Infectious Disease Society of America suggest that broad-spectrum antibiotics may be useful in patients with FIP [
12]. However, the evidence on FIP-causative pathogens and targeted antibiotic treatment strategies in VLBW infants with FIP is scarce. In Germany, two guidelines address antimicrobial therapy in neonates [
13,
14] but clear recommendations for VLBW infants with FIP are not given.
In this study we aimed to evaluate the antimicrobial use in VLBW infants with FIP in multiple tertiary care neonatal intensive care units (NICU) participating in the German Neonatal Network (GNN). The primary objective of this study was to evaluate empiric antibiotic therapy in VLBW infants diagnosed with FIP. The secondary objective was to characterize microbiological findings in a detailed subset analysis of patients treated at three GNN centers. We hypothesized that, based on clinical findings, laboratory results and pathogenic spectra the use of carbapenems may not be necessary in VLBW infants with FIP. This provides an opportunity to improve treatment concepts within the context of antimicrobial stewardship.
Methods
Study design and patients
Data on VLBW infants with FIP were prospectively collected within the GNN, a collaboration of 68 tertiary care NICUs in Germany. The GNN is an observational data collection initiative focusing on long-term development of VLBW infants. This collaboration started patient recruitment on January 1, 2009, and has been ongoing continuously. Patient characteristics and postnatal treatment was recorded on data sheets at the participating centers by study nurses or supervising physicians. Subsequently, data sheets were sent to the study center post-discharge. The GNN study was approved by the institutional review board at the University of Lübeck and at all participating study sites. Written informed consent was obtained by the infants’ guardians.
This study employed a retrospective design. Due to data retrieval from a predesigned database, available information was predetermined. With regards to our primary objective, data extraction from the GNN database was performed by applying the following inclusion criteria: (a) infants with birth weight less than 1500 g and (b) documented FIP requiring surgery. The last patient was included on December 31, 2021. Clinical features that were drawn from the database included birthday, gestational age, birth weight, sex, multiple births, antenatal steroids, 5-min Apgar score, inotrope support within 24 h of life, GNN site, occurrence of clinical sepsis or culture-proven sepsis, FIP and antibiotic prescriptions. Standard protocols dictate blood culture collection prior to any antibiotic exposure. However, a review of actual practice has not been conducted.
To investigate our secondary objective, we conducted additional on-site audits in a subset of VLBW infants with FIP from three participating GNN sites (Cologne, Lübeck, Würzburg). For this purpose, patient records were reviewed retrospectively from March 2016 until March 2023. Included were (a) infants with a birth weight < 1500 g and (b) documented FIP irrespective of whether they underwent surgical intervention. Excluded were infants with either only gastric or rectal perforation, malrotation or ileal atresia. Depending in center-specific policies, electronic and/or paper-based information was used in each patient. Various parameters were collected, including birthday, gestational age, birth weight, sex, laboratory results, histopathology, intraabdominal cultural results, data of multiplex-bacterial and -fungal PCR and antimicrobial treatment in timely association with FIP. Intraabdominal cultural collection comprised cultures obtained during intraabdominal catheter placement and during surgery. Standard protocols specify microbiological culture collection prior to antimicrobial treatment, if feasible. Microbiological data on subsequent surgeries were not included into the secondary objective analysis.
Definitions
Sepsis was defined according to the criteria of the national infection surveillance system (NEO-KISS) as either culture-negative clinical sepsis (clinical and laboratory signs of sepsis and clinician’s decision to treat with antibiotics for at least 5 days) or culture-confirmed sepsis [
15]. Furthermore, we differentiated inflammatory from non-inflammatory FIP. Inflammatory FIP was recorded as soon as one of the following criteria were met (1) greyish/mottled skin appearance and capillary refill time > 2 s, (2) painful palpation of the abdomen and (3) CRP ≥ 10 mg/L and/or IL6 ≥ 40 ng/L. FIP without inflammation was documented when the infant presented with (1) normal skin appearance and regular capillary refill time plus (2) soft abdomen after drainage plus (2) CRP < 10 mg/L and/or IL6 < 40 ng/L. This classification was not predefined and is not yet validated within any other cohort. We established this differentiation in anticipation of future antimicrobial stewardship efforts, with the aim of potentially identifying risk groups and implementing risk-adapted treatment.
Statistical analysis
Statistical analysis was performed with IBM® SPSS Statistics (version 28). For the primary and secondary objectives, normal distribution was determined by the Kolmogorov–Smirnov-test. Furthermore, proportions of nominal variables were compared using Pearson’s χ
2 test and Fisher’s exact test for both objectives. Median values were compared non-parametrically by two-sided Mann–Whitney
U-tests. Investigating the primary objective, multivariate analyses were performed by applying logistic regression models including known confounders which might have an impact on the clinical decision to treat with a carbapenem or vancomycin, i.e. gestational age, small for gestational age, sex, multiple births, antenatal steroids, 5-min Apgar score, inotrope support within 24 h of life, GNN site, year of birth, clinical sepsis or culture-proven sepsis and FIP [
16].
p-Values < 0.05 were considered significant.
Discussion
Carbapenems, vancomycin and aminoglycosides are the most common used antibiotic agents for VLBW infants with FIP in German NICUs, regardless of concomitant diagnosis of culture-confirmed or clinical sepsis. Microbiological findings of the audited subset of 124 VLBW infants with FIP showed a predominance of CoNS within the abdominal cavity, only a minor proportion presented with gram-negative bacteria. Among the identified Enterobacteriaceae, only three were susceptible to carbapenems exclusively. Consequently, broad-spectrum antibiotics such as carbapenems may not be justified for these patients.
FIP is a serious complication in VLBW infants with varying incidences depending on the population studied. Our cohort presented a rate of 2.8%, comparable to other previously published VLBW infants groups [
17]. The onset of FIP in our subset of 124 infants was at 5 days of life (median) with abdominal distension as the key clinical hallmark present in all cases. As discrimination between FIP and NEC may not always be easy [
1‐
3], a proportion of infants within our cohort presented with macroscopic and/or microscopic signs of inflammation or necrotizing tissue. Infants with macroscopic signs of inflammation were diagnosed later than infants without inflammation. Yet, necrotizing tissue was found very early in a smaller subset of patients. Furthermore, histopathological signs of inflammation were seen in 44 VLBW infants. In four cases, clear discrimination between NEC and FIP was not possible, as only one was diagnosed later than 14 days of life, while necrotic tissue was described histopathologically in three cases. With regards to the primary and secondary objective of our study, knowledge on the developing gut microbiota is crucial for future targeted treatment approaches of infants with FIP. The neonatal gut microbiota is highly dynamic and evolves rapidly within the first few weeks of life [
18,
19]. While
Staphylococcus spp. and
Streptococcus spp. dominate within the first week of life, the abundance of
Enterococci spp., gram-negative bacteria, and strict anaerobes like
Bifidobacterium spp. increases afterwards [
20‐
24]. Presence of fungi is also known [
8,
22,
25]. Since there is a variability in the occurrence of FIP, the timing of sample collection varied in our cohort. Still, 75% of all intraabdominal cultures were gathered within the first week of life. Literature on microbiological findings within the group of VLBW infants and FIP is scarce. A study from 2020 analyzed the incidence of bacteremia/fungemia in infants with FIP or NEC but found only 3/32 positive cases, i.e. two with CoNS, one with
E. coli [
26]. Abdominal culture results were not analyzed. The decreased abundances of Firmicutes and Bacteroidetes and the increased relative abundances of Gamma-Proteobacteria appear to be consistent with NEC-associated dysbiosis, while no dysbiotic feature has been linked to FIP so far [
4,
21,
27]. Our abdominal culture-based microbiological results obtained either from abdominal drainage catheter placement or during surgery mainly revealed gram-positive bacteria, especially CoNS. A minority presented with gram-negative organisms. Exclusive carbapenem sensitivity was evident in three infants with intraabdominal proof of
Enterobacteriaceae. Notably, the proportion of identifiable fungal species within the abdominal cavity is higher than what is typically found in blood cultures, likely due to the site of culture collection.
Evidence of intraabdominal pathogens alone do not provide conclusive evidence of invasive infection, however, they might help guiding antimicrobial treatment.
Antimicrobials are frequently prescribed in VLBW infants due to the difficulty of distinguishing infection from non-infectious conditions as symptoms in neonates are not specific [
28]. Furthermore, a significant number of infants in our cohort received broad-spectrum antibiotics independent of sepsis diagnosis. This special group of patients is highly vulnerable and at increased risk for severe infections, resulting in significant concerns regarding the potential severe outcomes if left undertreated. As a result, broad-spectrum agents are often preferred in VLBW infants with possible bacterial infection. ASPs aim to optimize the appropriate and judicious use of antimicrobial agents. The importance of ASPs in the NICU becomes apparent considering previously published data, showing not only gut microbiota disruption but also increased mortality and morbidities following non-rational use of antimicrobials [
29‐
35]. Feasibility and safety of such programs in the NICU were shown before [
36,
37]. In Germany, pediatric infectious diseases specialization is still evolving, and corresponding expertise is not comprehensively available nationwide. Consequently, ASP initiatives are currently implemented in only a limited number of NICUs. Our data emphasize the urgent need for expanding expertise in this area.
While empiric antibiotic treatment for early-onset sepsis typically involves a beta-lactam agent combined with an aminoglycoside, the management of late-onset sepsis varies from center to center in Germany. Common antibiotic regimens include a combination of a glycopeptide with an aminoglycoside, or a third-generation cephalosporin with a glycopeptide, or a beta-lactam beta-lactamase inhibitor. Treatment decisions also depend on the focus of infection. Moreover, in Germany, weekly screenings of VLBW infants are conducted to assess colonization status, which may influence the choice of empiric treatment [
38]. Literature on the use of antimicrobial substances within the setting of FIP is extremely limited [
39,
40], underlining the need for more studies to allow for improved future antibiotic treatment concepts. More data is available on infants with NEC. Still, a recently published review concludes, current evidence does not sufficiently favor any specific antibiotic regimen for the treatment of NEC [
39,
41,
42]. In our total cohort of 613 VLWB infants as well as in the smaller subset of 124 infants with FIP, antimicrobial therapy with meropenem and vancomycin were the most common prescribed agents. Considering the aforementioned microbiological findings within our cohort and the conclusions drawn from knowledge of the development of the neonatal gut microbiota, we suggest that broad-spectrum antimicrobials like carbapenems are not indicated in the majority of these patients. Instead, piperacillin/tazobactam might be a suitable option. Interpretation of the increased cultural identification of CoNS in our cohort remains challenging though. While sample contamination frequently leads to false-positive identification of CoNS, conversely, they form a significant part of the developing gut microbiota in VLBW infants at that age, potentially elevating the risk of infection and indeed being relevant in this group of patients.
Limitations of this analysis comprises the trial design, data collection within the prospective GNN trial allows associations only. Furthermore, documented antimicrobial treatment includes a wide period of time from birth until discharge. We lack reliable data on the prescribed antimicrobial therapy specifically related to FIP for the total cohort. Another point of discussion is the differentiation between NEC and FIP. Achieving a definitive differentiation between NEC and FIP in a clinical setting is not entirely feasible. Inflammatory FIP and NEC are particularly hard to discriminate, also the here applied definitions inflammatory vs. non-inflammatory FIP lack established standards. However, for our assessment, the timing of suspected infection appears to be more relevant.