key: cord-0004972-tzia719d
authors: Boccia, Delia; Stolfi, Ilaria; Lana, Susanna; Moro, Maria Luisa
title: Nosocomial necrotising enterocolitis outbreaks: epidemiology and control measures
date: 2001
journal: Eur J Pediatr
DOI: 10.1007/s004310100749
sha: 2b07dbfdbf340c4ea51587a7e2e2ebc1a4cc320d
doc_id: 4972
cord_uid: tzia719d

Necrotising enterocolitis (NEC) is one of the most serious gastrointestinal diseases among newborns and it mainly affects those in intensive care units. The aetiology of the disease has been reported to be multifactorial and both sporadic cases and nosocomial outbreaks have occurred. In this report, we review 17 epidemics of NEC reported in the literature between 1973 and 1999. The number of confirmed cases ranged from 1 to 32 with an average of 10.5 confirmed cases. On average, 16.15% of cases required surgery (range 0–66.6%). The average mortality rate was 6.25% (range 0–87.5%). The mean age at disease onset was 9.5 days (range 6.6–29 days). Most of the infants had low birth weight (median weight 1,395 g; range 1,112–2,788 g, calculated on the reported mean weights). The main risk factors associated with NEC were: low birth weight, low gestational age, low Apgar score, perinatal complications, hyaline membrane disease, and umbilical catheterisation. The bacteria involved often included Enterobacteriaceae, particularly Escherichia coli, Klebsiella pneumoniae and Enterobacter cloacae type 3305573. The causative role of Clostridia in NEC is controversial. With regard to viral agents, coronarovirus, rotavirus and enterovirus, such as echovirus type 22, were isolated during some of the epidemics. The recommended control measures for NEC epidemics are those used for epidemics of other orofaecally transmitted infections. Conclusion Understanding the epidemiology of necrotising enterocolitis is fundamental if adequate preventive control measures are to be developed and applied.

Necrotising enterocolitis (NEC) is one of the most serious gastrointestinal diseases among newborns. It mainly aects those with low birth weight and generally occurs as sporadic cases, though epidemics have also been reported. The incidence of NEC is relatively high in the neonatal period, ranging from 1±5% among newborns on intensive care units and from 0.3±2.4 per 1,000 livebirths (median 1.3 per 1,000) [28] . Most of the newborns aected by this disease are premature (62±94%), although sporadic cases of NEC have been reported among full-term newborns [32] and there is a linear correlation between incidence, low birth weight and low gestational age. NEC requires surgery in 23±70% of cases depending on the gestational age; although the mortality rate has signi®cantly decreased in the last 30 years, it continues to be high (9±28% in the 1990s) [28] .

The universally accepted case de®nition is that proposed in 1978 by Bell and collaborators [2] , according to whom the disease is characterised by the following: systemic symptoms, ranging from apnoea, bradycardia and temperature instability to diuse intravascular coagulation and septic shock, intestinal symptoms (e.g. abdominal distension and bloody stools) and radiological ®ndings such as pneumatosis intestinalis and gas in the portal vein. Based on the severity of symptoms, dierent grades of the disease, from suspected to advanced, have been de®ned ( Table 1) .

The aetiology of NEC is multifactorial, attributable to both infective and non-infective factors (e.g. ischaemic lesions of the gastrointestinal tract in premature newborns or in those with speci®c predisposing factors, formula milk feeding, and the presence of a pathogen or of imbalances in the intestinal microbial¯ora) [28] .

Within the last few decades, various NEC epidemics have been reported in the literature. Given the potential severe clinical impact and the preventability of these events, we conducted a systematic review of the scienti®c literature to evaluate the epidemiological characteristics of the epidemics reported, the methods adopted for their investigation and the control measures used, with the aim of providing indications for the proper and rapid identi®cation and management of NEC epidemics.

NEC epidemics were identi®ed by a literature search of PubMed, the online site of the National Library of Medicine. Since NEC epidemics are not very common, we did not restrict the search to period of time or language. The following terms were used:``necrotising enterocolitis'' (all ®elds: 1,729 records were identi®ed),`n ewborn'',``neonate'' or``neonatal'' (all ®elds: 382,420 records), and``disease outbreaks'' (MeSH term: 27,037 records). The search identi®ed 38 articles that included all three terms: three of these articles could not be traced [3, 15, 20] and 20 were excluded from the analysis because they were not strictly related to epidemics. We reviewed a total of 15 articles, published in English language journals and reporting on 17 epidemics. Table 2 describes the principal characteristics of the 17 epidemics (i.e. number of cases, mean age and weight at onset and clinical impact of the disease). For all the epidemic investigations, the case de®nition of con®rmed NEC developed by Bell et al. [2] was adopted; however, it is not clear whether the investigations also included suspected cases of NEC and/or simple cases of gastroenteritis.

Of the 17 epidemics analysed, 11 lasted from 8 to 10 weeks [4, 6, 9, 11, 13, 16, 17, 21, 27] ; one epidemic lasted for 20 weeks [13] and four lasted for less than 8 weeks [1, 8, 10, 23] (in one case the duration of the epidemic was not mentioned [24] ); no seasonal trend was evident. The number of con®rmed cases ranged from 1 [4] to 32 [6] , with an average of 10.5 con®rmed cases. In ®ve of the epidemics, additional cases of NEC were suspected but not con®rmed [4, 8, 9, 16, 17] . In eight of the epidemics, additional cases of gastroenteritis due to the same pathogen were diagnosed [4, 6, 8, 9, 10, 11, 23, 24] .

Surgery was required in 0% [4, 10] to 66.6% [12] of cases (median 16.15%). The mortality rate ranged from 0% [4, 11, 13, 16] to 87.5% [27] (median 6.25%). The highest mortality rates were reported in the 1970s, whereas the mortality rate of the epidemics described in recent years was less than 10% [1, 6, 8, 10, 17, 24] .

The mean age at disease onset ranged from 6.6 [13] to 29 days [1] (median 9.5 days). Most of the aected newborns had low birth weight (median 1,395 g, range 1,112±2,788 g calculated on the reported mean weights), although some epidemics involved newborns with a birth weight over 2,000 g and Apgar scores of more than 7 [13, 16] . Many of the epidemic investigations reported risk factors associated with NEC, the most common of which were: low birth weight [4, 9, 17, 24, 27] , low gestational age [1, 4, 9, 27] , low Apgar score [26] , perinatal complications [21, 27] , hyaline membrane disease [21] , and umbilical catheterisation [21] . However, many of the investigations reported no risk factors in that there were no clinical dierences between cases and healthy controls [6, 13, 16, 23] . Table 3 shows the pathogens isolated during the epidemic investigations, the epidemiological and laboratory methods used, the risk factors identi®ed and the control measures adopted. Only seven of the epidemics were investigated with a proper case-control study [1, 9, 13, 17, 24] , one of which also included the microbiological monitoring of cases, hospitalised newborns and/ or hospital sta [24] . Seven epidemics were investigated only by microbiological monitoring [6, 8, 10, 11, 16, 23, 24, 27] , and three included only an analysis of the epidemiological characteristics of the NEC cases [4, 18, 27] . With regard to the causal agent (Table 3) , the bacteria involved often included Enterobacteriaceae, particularly Escherichia coli (non-typable [27] , a heat labile toxin producer [8] , and serogroup 0142 H6 [10] ), Klebsiella pneumoniae (either alone [13] or with viral agents [23] or other types of bacteria [12] ) and Enterobacter cloacae type 3305573 [21] . Dierent species of Clostridia were isolated during some of the epidemics (i.e. Clostridium butyricum [16] ; Clostridium dicile was reported in other epidemics not included in this review [14, 20] ). The viral agents involved were coronarovirus [6] , rotavirus [23, 24] and enterovirus such as echovirus type 22 [4] . In six of the epidemics, the causal agent was not identi®ed [1, 9, 11, 13, 17] ; in some cases, the laboratory investigations led to the isolation of one or more microorganisms but their causal role was not demonstrated as they were also isolated in healthy controls or from sites where they are commonly commensal.

All but one of the investigations [12] performed stool cultures (Table 3) . Two investigations included rectal swabs [9, 24] to search for intestinal pathogens, such as Salmonella, Shigella, enterotoxic Staphylococcus aureus, Campylobacter and Yersinia [9, 24] . In most cases, blood cultures were taken for aerobes and anaerobes; cultures were carried out more rarely for urine [1, 10] , CSF [1] , peritoneal liquid [17] , and nasal and pharyngeal swabs [10] . In one epidemic [12] , all isolates were subjected to molecular ®ngerprinting. For four epidemics, microbiological investigations were carried out on environmental samples or on milk [1, 16, 24, 27] . Only some of the investigations included the search for Clostridia and their toxins using anaerobic cultures [1, 16] , gas-liquid chromatography [16] or immunoenzymatic techniques (e.g. ELISA) [8] . With regard to Clostridium butyricum, isolation required a very long period of time: the ®rst of the six samples that were positive on gas-liquid [16] . The techniques used to identify the virus were ELISA [8, 11, 23] and EIA [24] ; some studies used electron microscopy to detect viral particles in the faeces [6, 11, 17, 24] . In many epidemics, microbiological tests (either serological or culture) were also carried out on the hospital sta [4, 6, 8, 10, 11, 16, 23, 24] who were often shown to be positive, mostly to the serological tests [4, 6, 23, 24] but also to culture [10] . The measures used for controlling the epidemics were those for epidemics of other orofaecally transmitted infections, speci®cally: strict adherence to basic rules of hygiene such as hand washing and the use of gloves and masks [1, 4, 6, 11, 12, 13, 17, 27] , isolation of infected newborns [1, 4, 6, 13, 17, 23, 27] , cohorting cases (i.e., infected newborns always cared for by the same sta members) [1, 4, 6, 17, 27] , and, in some cases, closing of the ward involved [1, 11, 17 ].

This review shows that the reported epidemics greatly diered in terms of the number of cases, the spectrum of clinical presentations reported and the proportion of newborns who underwent surgery or who died. These dierences, in part, can be explained by dierences in the application of the case de®nition (i.e. whether or not only con®rmed cases or also suspected cases of NEC and/or simple gastroenteritis were included) and by dierences in diagnostic procedures. The reported causal agents also varied greatly. The causative role of Clostridia is controversial: some authors suggest that Clostridia may be part of the normal intestinal¯ora and that it is thus dicult to establish whether or not they act as pathogens [21, 30] . In one of the epidemics [16] , the causal role of Clostridia was corroborated by their isolation from the blood of nearly all of the cases; however, as discussed in the epidemic report, the seriousness of the clinical syndrome could probably be attributed to the particularly invasive nature of the speci®c strain which would account for healthy newborns not normally at risk for NEC also being aected. Furthermore, the authors suggested that the failure to isolate Clostridia in other NEC epidemics may depend on the fact that anaerobic cultures are not always performed or, when performed, the samples are not always incubated for a suciently long period and more sophisticated diagnostic techniques are not used (i.e. gas-liquid chromatography) [16] . None of the investigations reported Staphylococcus epidermidis as the causative agent; however, Ng et al. [19] described two cases of NEC known to have been caused by Staphylococcus epidermidis since this microorganism was isolated from blood, peritoneal liquid and breast milk, which was the likely source of infection. Moreover, other studies [18, 25] have suggested that coagulase-negative staphylococci are commonly involved in NEC and are associated with high rates of mortality and morbidity. [ Since viral agents are often the cause of epidemics of hospital-acquired gastroenteritis and are not generally associated with serious clinical syndromes such as NEC, it has been suggested that NEC may be caused by the concomitant presence of viral agents and predisposing factors in newborns or to a possible synergism among viral agents and other microorganisms. In one of the epidemic reports [23] , the authors suggested that a synergism exists between rotavirus and Klebsiella resulting in an increased pathogenicity. In four of the eight cases reported, both microorganisms were present, and in one case, although there was no rotavirus in the faeces, a high IgM antirotavirus titre was detected. In 33% (6/17) of the outbreaks, the causative agent was not identi®ed, probably as a result of incorrect or incomplete diagnostic testing. In contrast to sporadic cases of NEC which may be caused by various factors and that do not necessarily include a speci®c pathogen, in epidemics, the causative factor is the diusion of a particular pathogen in a speci®c ward and in a certain period of time. Thus the involvement of a speci®c pathogen must be taken into account when conducting epidemiological investigations and diagnoses and eective control measures must be rapidly adopted to stop the infection from spreading. Of note is the ®nding that hospital sta appeared to be frequently involved in these epidemics, probably because they are exposed to patients, although it cannot be excluded that the sta may play a direct role in transmitting the infection.

The reason for which a relatively high proportion of newborns with gastrointestinal symptoms developed NEC is still unclear, though two hypotheses can be proposed:

1. Epidemics of NEC occur when the pathogen is transmitted orofaecally among newborns considered to be particularly susceptible (i.e. premature or with predisposing factors in the intestinal ischaemia). For healthy newborns, simply being exposed to the infective agent would only result in colonisation or, depending on the faecal¯ora concentration, a simple case of gastroenteritis. For premature newborns or those with predisposing clinical conditions, exposure to the infective agent would lead to intestinal necrosis and to NEC. 2. The onset of NEC depends on the level of virulence of the microorganism concerned (e.g. Clostridium butyricum) or on the pathogenic eect consequent to the synergic action of two microorganisms (e.g. rotavirus and Klebsiella pneumoniae).

Given the severity of nosocomial outbreaks of NEC, it is fundamental that healthcare workers in neonatology and neonatal intensive care wards be made aware of the potential for epidemics to occur and that all necessary preventive measures be taken. Outbreaks must be iden-ti®ed promptly and measures must be taken to identify and interrupt the chain of transmission. This means being prepared to cope with the epidemic before it happens. The following is a brief summary of activities to be conducted in the case of an epidemic of NEC.

1. Identi®cation of the causal agent. In addition to routine investigations, it is necessary to perform the following: stool cultures for Clostridia, their toxins, and viral agents such as enterovirus and rotavirus; blood cultures for both aerobes and anaerobes; and appropriate serological investigations of cases and contacts when a direct search for the pathogen fails. 2. Exhaustive investigation of cases. Attempts must be made to identify all newborns who have been infected by the causal agent and any other newborns who have developed any form of NEC. Newborns discharged from nurseries could be monitored to verify any additional cases. A thorough identi®cation of the cases is important not only for epidemiological reasons but also for adopting adequate cohort programmes (i.e. having a separate nursing sta care for the infected or colonised newborns). 3. Adoption of eective measures for stopping the spread of the infection. These include: (a) isolating or adopting cohort programmes for both infected and colonised newborns; if colonised newborns cannot be identi®ed, then the isolation or cohort programmes can be organised for both exposed and non-exposed (i.e. newly admitted) asymptomatic newborns; (b) enforcing routine procedures for antisepsis (e.g. hand washing, barrier measures, adequate decontamination of equipment and protocols for the treatment of newborns exposed to an invasive procedure); (c) considering the option of closing the ward as a last resort if the epidemic cannot be contained and (d) surveillance of sta and removal of those persons with symptoms of gastroenteritis [5] . In three epidemics [5, 10, 26] , the adoption of these control measures stopped the epidemic. In one of the hospitals [5] , these measures continued to be strictly followed after the epidemic and the incidence of NEC decreased from 3.6% to 0.7% over a 3-year period. 4. Identi®cation of the possible means and mechanisms of transmission. This can only be achieved by conducting a descriptive epidemiological investigation, followed by, if possible, an analytical investigation (e.g. cohort and case-control studies). 5. Search for the causal agent in the environment. This only makes sense if a causal agent has been identi®ed and if the epidemiological investigation has led to a hypothesis on the possible source of infection.

In conclusion, epidemics of NEC are serious events that may, however, be prevented by speci®c measures for preventing transmission in hospitals. It is also possible to contain an epidemic provided that measures have been carefully planned in advance.

A widespread epidemic of mild necrotising enterocolitis of unknown cause

Neonatal necrotizing enterocolitis: therapeutic decisions based upon clinical staging

An outbreak of necrotizing enterocolitis in a special care newborn nursery

Echovirus type 22 outbreak associated with gastro-intestinal disease in a neonatal intensive care unit

Clustering of necrotizing enterocolitis: interruption by infection-control measures

Association of coronavirus infection with neonatal necrotizing enterocolitis

Neonatal necrotizing enterocolitis

Necrotizing enterocolitis with Escherichia coli heat labile enterotoxin

The role of host factors in an outbreak of necrotizing enterocolitis

An outbreak of gastroenteritis due to Escherichia coli O142-H6 in a neonatal department

Increased risk of illness among nursery sta caring for neonates with necrotizing enterocolitis

Klebsiella pneumoniae with extended spectrum beta-lactamase activity associated with a necrotizing enterocolitis outbreak

Epidemic occurrence of neonatal necrotizing enterocolitis

Clostridium dicile colonization in neonates admitted to a neonatal intensive care unit

Nosocomial colonisation with Klebsiella type 26 in a neonatal intensive care unit associated with an outbreak of sepsis, meningitis and necrotizing enterocolitis

Outbreak of necrotizing enterocolitis caused by Clostridium butyricum

An outbreak of necrotizing enterocolitis. An association with transfusion of packed red blood cells

The role of coagulase negative Staphylococcus in neonatal necrotising enterocolitis

Bacterial contaminated breast milk and necrotising enterocolitis in preterm twins

An epidemic outbreak of necrotizing enterocolitis due to Clostridium dicile in term newborn infants

Necrotizing enterocolitis. Epidemic following an outbreak of Enterobacter chloacae type 3305573 in a neonatal intensive care

Clostidium dicile and its toxin in healthy neonates

An outbreak of rotavirus-associated neonatal necrotizing enterocolitis

Neonatal rotavirus-associated necrotising enterocolitis: case control study and prospective surveillance during an outbreak

Delta like toxin produced by coagulase negative staphylococci is associated with neonatal enterocolitis

Clinical and bacteriological ®ndings in necrotising enterocolitis: a controlled study

Fulminating neonatal sepsis and necrotizing enterocolitis associated with``nonenteropathogenic'' strain of Escherichia coli

Epidemiology of necrotising enterocolitis

Necrotizing enterocolitis: treatment based on staging criteria

Quantitative study of the aerobic and the anaerobic faecal¯ora in neonatal necrotising enterocolitis

Epidemiologic characteristics of necrotizing enterocolitis: a population based study

Necrotizing enterocolitis in full term infants. A case control study