Systematic review of invasive Acinetobacter infections in children

Abstract

INTRODUCTION:

Clinicians are generally familiar with Acinetobacter as an etiological agent for serious nosocomial infections in intensive care units. However, there are no previous reviews of the full spectrum of invasive infections in children.

METHODS:

A systematic review of the literature was completed up to December 2008 for reports of invasive Acinetobacter infections in children.

RESULTS:

There were 101 studies that met the inclusion criteria including 18 possible outbreaks, 33 case series and 49 case reports. Suspected outbreaks were concentrated in neonatal intensive care units (16 of 18 outbreaks) and involved bacteremia or meningitis. Proof of isolate clonality or identification of the source of the outbreak was seldom established. Case series were primarily of children younger than five years of age presenting with bacteremia (sometimes multiresistant), meningitis, endocarditis or endophthalmitis, with many community-acquired infections being reported from India. Case reports consisted of unique presentations of disease or the use of novel therapies. Attributable mortality in the outbreaks and case series combined was 68 of 469 (14.5%).

DISCUSSION:

Invasive Acinetobacter infections in children usually manifest as bacteremia, meningitis or both, but can result in a wide variety of clinical presentations. Outbreaks are primarily a problem in newborns with underlying medical conditions. Most reports of community-acquired infections are from tropical countries. The study of the mechanism of colonization and infection of children in intensive care units and of neonates in tropical countries may provide some insight into prevention of invasive infections.

The Acinetobacter genus is a group of immotile, aerobic, non-fermenting Gram-negative coccobacilli found in soil and fresh water (1). Acinetobacter species have become an important culprit in nosocomial infection (1) and, in recent years, have displayed increasing resistance to a broad range of antimicrobials (2–4). The purpose of the present study was to perform a systematic review of the literature on Acinetobacter invasive infection in pediatric patients to describe the epidemiology and outcome of reported cases. This may aid clinicians in outlining the prognosis for children with invasive infections and guide them when they suspect an outbreak due to Acinetobacter species.

METHODS

Inclusion criteria

All studies that provided information on the clinical course of one or more cases of Acinetobacter infection isolated from a normally sterile site in children up to 18 years of age were included in the present review (studies till December 2008). There was no date or foreign language restrictions. Translation was only available for French language articles; otherwise data were mainly derived from an English abstract if provided. Reports that described colonization, pneumonia or soft tissue infections were excluded, unless the organism was also grown from blood cultures or from a normally sterile site. Studies that involved both children and adults were included if the clinical information on pediatric patients could be ascertained.

Classification of studies

Studies were classified into one of three categories: outbreaks, case series or case reports. Studies describing nosocomial Acinetobacter infections or colonization (with a minimum of one case of invasive disease) that were believed by the study authors to be epidemiologically linked were classified as outbreaks. Studies describing two or more cases (including at least one child) that were not believed to have a common source were classified as a case series. The remaining studies, generally individual reports of interesting manifestations of Acinetobacter infection, were classified as case reports.

Search strategy

The English and foreign-language literature on Acinetobacter species was searched using MEDLINE, EMBASE, PUBMED, and SCOPUS databases in December 2008. For MEDLINE, all subject headings were checked off when terms were mapped. The study was limited to “all children, 0–18 years”. For EMBASE, all subject headings were checked off when terms were mapped. The study was limited to “infant”, “child”, “preschool child”, “school child” and “adolescent”. For PUBMED, the study was limited to “all child: 0 to 18 years”. For SCOPUS, an advanced search was conducted with “Acinetobacter AND infant* or child* or newborn* or neonate* or neonatal or preschool* or adolescent* or teen*or toddlers*”.

For the above four database searches, “Acinetobacter” was used as the first search term. The nomenclature for Acinetobacter species has changed over the decades, and all other known terms for any species of the bacteria were also searched with the four databases – these were Mima polymorpha, Herellea vaginicola, B5W, Bacterium anitratum, Achromobacter haemolyticus var. glucidolytica, Achromobacter conjunctivae, Moraxella glucidolytica, Achromobacter anitratus, Neisseria winogradskyi, Micrococcus calcoaceticus, Diplococcus mucosus, Achromobacter citroalcaligenes, Achromobacter haemolyticus var.alcaligenes, Alcaligenes metalcaligenes, Achromobacter lwoffii, Moraxella lwoffii and Alcaligenes hemolysis.

Data extraction

Data extraction was performed by both authors. Relevant clinical information on the course of Acinetobacter infection was collected.

RESULTS

There were 101 studies that met the inclusion criteria including 18 outbreaks, 33 case series, and 49 case reports. There were 28 studies up to 1970, 13 from 1971 to 1990, and 70 from 1991 to 2008.

Outbreaks

Eighteen of the studies (5–22) described outbreaks, beginning in 1982 (Table 1). The frequency of outbreak reports has been increasing, with nine of the 18 outbreaks reported within the past 10 years. Outbreaks were reported from Europe (seven studies), Asia (three studies), South America (three studies), Africa (two studies), the Middle East (two studies) and the Carribbean (one study).

TABLE 1.

Characteristics of pediatric Acinetobacter outbreaks with at least one case of invasive infection

Ref	Year	Country	Cases of invasive infection, n	Setting	Bacteremia, n	Meningitis, n	Deaths from any cause, n	Deaths fromAcinetobacter, n	Species
5	2008	Turkey	5	NICU	5	NR	NR	2	A septicus
6	2007	Taiwan	1	NICU	1	NR	0	0	A baumannii
7	2006	Venezuela	16	NICU	16	2	NR	NR	A RUH 1139
8	2005	Brazil	11	NICU	11	NR	3	3	A baumannii
9	2003	India	36	NICU	36	9	13	NR	A baumannii
10	2003	Israel	9	NICU	9	2	3	NR	A baumannii
11	2002	Taiwan	9	NICU	9	NR	3	0	A baumannii
12	2001	South Africa	7	PICU/NICU	5	NR	NR	NR	A calcoaceticus
13	2000	Germany	3	Pediatric oncology ward	3	0	NR	NR	A junii
14	1999	Netherlands	6	NICU	6	NR	NR	0	A junii
15	1999	South Africa	4	NICU	4	NR	1	1	Not specified
16	1998	Bahamas	8	NICU	8	0	6	3	A baumannii
17	1993	Israel	9	NICU	9	0	4	4	Not specified
18	1991	Chile	21	NICU	21	NR	NR	NR	A calcoaceticus
19	1990	Germany	6	NICU	6	NR	0	0	A calcoaceticus
20	1989	United Kingdom	7	NICU	7	NR	0	0	A lwoffi
21	1983	United Kingdom	2	Burn unit	2	NR	2	NR	A calcoaceticus
22	1982	United Kingdom	4	NICU	NR	4	NR	0	A calcoaceticus

All species belong to the Acinetobacter genus. NICU Neonatal intensive care unit; NR Not reported; PICU Pediatric intensive care unit; Ref Reference

Fifteen of the 18 outbreaks occurred in neonatal intensive care units (NICUs) with another occurring in a combined NICU and pediatric intensive care unit (PICU). There was an outbreak reported from a burn unit (21) and from a pediatric oncology ward (13). The number of cases of invasive Acinetobacter infection varied from one to 36 per outbreak (median: seven cases), with colonized patients also being reported in many outbreaks.

A variety of Acinetobacter species were implicated in the outbreaks: Acinetobacter baumannii (six studies), Acinetobacter calcoaceticus (five studies), Acinetobacter junii (two studies), Acinetobacter lwoffii (one study) and unspecified species (two studies). Two studies described cases with novel strains of Acinetobacter, including Acinetobacter septicus (5) and Acinetobacter RUH 1139 strain (7). In most studies, the authors assumed an outbreak based solely on an increased number of children with Acinetobacter colonization or infection in a medical unit over a period of weeks to months. Typing of isolates was described in only one study (11), identifying six of nine strains in bacteremic neonates as identical.

The source of the outbreak was proven only in one study in which an Acinetobacter RUH 1139 strain caused bacteremia infections in 16 neonates and was isolated from the total parenteral nutrition solution (7). Many studies described isolation of Acinetobacter from environmental sources, air conditioners, aerators, dressings, suction catheters or the hands of health care workers; however, typing was never performed to confirm whether the patient and environmental strains were identical.

The vast majority of patients in the outbreaks had Acinetobacter bloodstream infections. Of the 18 studies, 17 described at least one bacteremic patient. The other study (22) did not report blood culture results in four preterm infants with Acinetobacter meningitis. Meningitis was the second most common infection after bacteremia, with four of the 18 studies describing patients with Acinetobacter meningitis with or without bacteremia. There were no other sites of proven invasive infection described in the outbreak studies.

Fourteen of the 18 studies provided information on patient outcome, with all these studies involving only neonates. For the 11 studies that reported deaths attributable to Acinetobacter, mortality varied from 0% to 44%; 13 of 70 neonates (19%) died.

Case series

Thirty-three of the studies (23–56) were case series dating back to 1953 (Table 2). As with the outbreaks, the number of case series reported also seems to be increasing, with 15 of 33 (45%) occurring within the past 10 years. Case series were concentrated in India (seven studies [21%]), Slovakia (four studies [12%]) and the United States (four studies [12%]).

TABLE 2.

Characteristics of pediatric Acinetobacter case series involving at least one child with invasive infection

Ref	Year	Country	Cases of invasive infection, n	Age range	Setting	Primary infection	Deaths fromAcinetobacter,n	Species
23	2008	Iran	40	<28 days	NICU	Bacteremia	NR	A baumannii
24	2007	Taiwan	52	3 years to 27 years, mean: 6 years	Wards, PICU, NICU	Bacteremia	NR	NR
25	2007	Slovak Republic	25	NR	Children’s hospital general wards, ICUs	Meningitis	5	A baumannii
26	2007	United States	92	IQR: 1.8 to 13.2	Children’s hospital general wards, ICUs, outpatients	Bacteremia	4	A baumannii, A lwoffii, A calcoaceticus, unidentified species
27	2006	Greece	26	Mean: 7.6 years	PICU	Bacteremia	NR	A baumannii
28	2006	India	23	NR	7 community acquired, 16 hospital acquired	Bacteremia	NR	A baumannii, A lwoffii
29	2006	India	42	34 were <12 h old	9 community acquired, 33 hospital acquired	Bacteremia and meningitis	NR	A baumannii, A lwoffii
30	2005	Greece	2	14 years, 16 years	NR	Meningitis	0	A baumannii
31	2004	India	78	NR	NR	Bacteremia	NR	A baumannii, A lwoffii
32	2003	Morocco	20	<28 days old	NICU	Bacteremia and meningitis	NR	NR
33	2003	India	4	7 years, 14 years, 14 years, 18 years	NR	Endophthalmitis	0	A calcoaceticus
34,35	2001	Slovak Republic	48	0 to 12 years	Multiple wards across 8 teaching hospitals	Bacteremia	6	A baumannii
36	2000	Slovak Republic	10	1 to 36 months	NR	Meningitis	4	A calcoaceticus
37	2000	Taiwan	3	3 years, 3 years and 6 years	NR	Bacteremia	2	A lwoffii
38	2000	Slovak Republic	14	1 to 36 months	3 pediatric hospitals	Meningitis	0	A baumannii
39	1999	Taiwan	1	17 years	Neurosurgery ward	Meningitis	0	Unidentified species
40	1998	India	79	NR	55 nursery, 24 community acquired	Bacteremia	NR	NR
41	1998	Bangladesh	138	82% were 5 years or younger	Community and hospital acquired	Bacteremia	22	A baumannii, A lwoffii, A hemolyticus
42	1995	Netherlands	6	1, 3, 8, 13, 20 and 29 days	NICU	Bacteremia	5	A baumannii, A lwoffii, A junii, genomospecies 3, genomospecies 14, unidentified species
43	1993	India	26	NR	NICU	Bacteremia	NR	NR
44	1993	Israel	3	2 months, 3 months, 1 year	Neurosurgery ward	Meningitis	0	A baumannii
45	1991	Nigeria	2	NR	NR	Endocarditis	NR	NR
46	1989	Japan	19	4 to 22 days	NICU	Bacteremia	NR	A calcoaceticus
47	1986	United States	29	NR	Oncology ward	Bacteremia	0	A lwoffii, A calcoaceticus, unidentified species
48	1973	Israel	2	2 years, 3 years	NR	Bacteremia	NR	A lwoffii
49	1969	India	2	2 years, 2 years	NR	Meningitis	NR	A lwoffii
50	1965	United States	2	23 months, 6 years	Community acquired	Meningitis	1	A lwoffii
51	1965	United States	2	4 years, 16 years	Community acquired	Meningitis	1	A lwoffii
52	1965	India	2	4 years, 18 years	Community acquired	Meningitis	0	A lwoffii
53	1965	Uganda	5	1 month, 1 year, 1 year, 6 years, 7 years	NR	Bacteremia and meningitis	0	A lwoffii, A calcoaceticus
54	1961	Puerto Rico	7	4 days to 2 years	NR	Bacteremia and meningitis	3	A lwoffii, A calcoaceticus
55	1957	Canada	2	4 days, 6 days	NR	Meningitis	1	A calcoaceticus
56	1953	Unknown	2	11 days, 21 days	NR	Meningitis	0	A calcoaceticus

All species belong to the Acinetobacter genus. ICU Intensive care unit; IQR Interquartile range; NICU Neonatal ICU; NR Not reported; PICU Pediatric ICU; Ref Reference

The number of children per case series ranged from one to 138 (median: nine cases). For 12 of the 33 studies (36%), all cases were from the same setting: five studies were from an NICU (15%), two were from neurosurgery wards (6%), one was from a PICU (3%), one was from a general pediatric ward (3%) and three (all from India) described only community-acquired infections (9%). A study (26) from Philadelphia (USA) investigating risk factors for bacteremia with Acinetobacter found that almost 50% of infections were acquired at home in children with central venous catheters, many of whom had malignancies.

In terms of types of infection, five of the 33 (15%) studies included cases of neonatal (younger than one month of age) meningitis, 10 (30%) included cases of pediatric (older than one month of age) meningitis and four (12%) included cases of post-surgical meningitis. Fifteen of the 33 (45%) studies reported cases of bacteremia with no meningitis, one reported cases of endophthalmitis and one reported cases of endocarditis (Table 2). Four of the case series were published primarily because they reported infection with strains of multidrug-resistant Acinetobacter infection, with all being published since 2004 (23,24,27,28).

There was no information regarding the ages of the patients in six of the 33 case series, but 24 of the remaining 27 case series reported on patients five years of age and younger. For the 21 studies in which sex was reported, there were 334 males (54.6%) and 278 females (45.4%). For the 20 studies that reported deaths directly attributable to Acinetobacter infection, mortality varied from 0% to 83%; 55 of 397 children (13.9%) died.

Twenty-eight of the 33 case series provided information on the strain(s) of Acinetobacter responsible for causing infection, with nine reporting multiple strains. The strains implicated included the following: A lwoffii (15 studies), A baumannii (13 studies), A calcoaceticus (nine studies), A hemolyticus (one study), A junii (one study), genomospecies 3 (one study), genomospecies 14 (one study) and unidentified species (four studies). It should be noted that in seven of the studies, A lwoffii species was reported as the former name M polymorpha and, in four of the studies, A calcoaceticus was reported as the former names D mucosus or B anitratus.

Case reports

There were 49 case reports of invasive pediatric Acinetobacter infection dating back to 1950. Unlike the outbreaks or the case series, a relatively smaller fraction of the case reports were reported recently, with only 12 from the past 10 years. The geographical distribution of the case reports also differs from the outbreaks and case series in that only 12 of the 44 that provided the location were from developing countries.

Case reports described unusual presentations including multiple nodules in the spleen and lung mimicking fungal infection in a child with leukemia (57), a mediastinal mass in a child with chronic granulomatous disease (58), skin abscesses in a neonate (59), osteomyelitis of a phalanx following a hamster bite (60), gangrene of the toes in an infant (61,62) and corneal perforation with iris prolapse in a child (63). Novel aspects of therapy from the case reports included responses to parenteral colistin (64–66), netilmicin (65), tigecycline (67) and intrathecal polymyxin (68).

DISCUSSION

A systematic literature review of pediatric Acinetobacter infection yielded 18 descriptions of possible outbreaks, 33 case series and 49 case reports from around the globe dating back to 1950. Early publications were case reports of primarily meningitis, followed by case series of infections that occurred in specific settings (such as intensive care units, burn units or oncology wards). The first possible outbreak from a United Kingdom NICU was reported in 1982 (22). None of the outbreaks occurred in North America, but a NICU outbreak was reported from New York (USA) in 2009 after the literature search was completed (69).

One of the striking features of the outbreaks was that the vast majority (89%) occurred in NICUs, suggesting that neonates with a complicated course are at particular risk of Acinetobacter infection. However, it is important to realize that outbreaks may be easier to recognize in a NICU because infection control surveillance is more likely in this setting, with patients often staying in a NICU until discharge; consequently, almost all nosocomial infections are detected. In settings with no surveillance, linked cases may not be noted if different physicians are caring for the patient. Evidence for an outbreak was simply based on clustering of cases in the majority of studies; therefore, it is very difficult to determine whether they were truly outbreaks. The source of infection was definitely proven in only one study (7); therefore, it was difficult to make specific recommendations to prevent such outbreaks in the future. In fact, one study (12) reported that although all isolates were A calcoaceticus, susceptibility patterns varied, which would make one question if this truly was an outbreak.

Among the 33 case series, it is interesting to note that a significant number were from India (eight studies), with four of these case series including community-acquired cases. It is possible that Acinetobacter is more predominant in the normal flora of this population or in warm humid climates and, therefore, causes more community-acquired infections.

Bacteremia and meningitis were the most common presentations of pediatric Acinetobacter infection, with dissemination to other sites being rare. There were five case reports (69–74) and one case series that described endocarditis (45). There was one case series with four patients with endophthalmitis (33). There were only three cases reports of well-documented pulmonary involvement with isolation of Acinetobacter from pleural fluid (75), pulmonary lymph node (58) and lung at autopsy (76). However, it is important to recognize that ventilator-associated pneumonia Acinetobacter may also be common in children, but such cases would have been excluded from the current study because Acinetobacter is typically isolated from pulmonary secretions in the absence of a tissue diagnosis. Acinetobacter grew from superficial abscesses in two cases (59,77) and from an intra-abdominal abscess in one case (78), but was otherwise not implicated in abscess formation. This distribution of infections suggests that Acinetobacter is more likely to be part of skin rather than respiratory or gastrointestinal flora.

Acinetobacter taxonomy has changed markedly over time. As of 2008, there were 31 genomic species of which 17 had valid species names (79). It remains controversial whether A calcoaceticus and A baumannii should be regarded as the same species because they are difficult to differentiate. A calcoaceticus appears to be more common from environmental sources and A baumannii from human sources (79). To clarify some of the terms used in older literature, A anitratus is actually a subtype of A calcoaceticus and is more accurately written as A calcoaceticus var anitratus. Mima polymorpha is an older name for A lwoffii, while B anitratum, D mucosus, M glucidolytica and H vaginicola are all A calcoaceticus. A wide variety of species have been described as etiological agents of invasive disease, with A baumannii traditionally being considered to be of greatest clinical significance, although A lwoffii and A calcoaceticus were commonly implicated in the published pediatric literature.

The outcome of infection with multiresistant or panresistant Acinetobacter in children remains unclear because it is probable that authors are more likely to publish cases of success than of failure. There are few published cases (64–66) in which the only reasonable therapeutic option was colistin. Fortunately, it appears that ill children develop less nephrotoxicity from colistin than adults (80).

The primary limitation of the present study is that infections reported in the literature may not mirror the relative incidence, spectrum and severity of infection, especially in developed countries or nonacademic centres where clinicians may not have the opportunity to report unusual infections or outbreaks. Pediatric cases that were reported as part of adult series may have been missed. Identification and speciation of Acinetobacter is complex, and may not always be accurately reported.

CONCLUSION

Reported invasive Acinetobacter infections in children generally manifest as bacteremia or meningitis. Ventilator-associated pneumonia may also be a common presentation, but this could not be ascertained from the current study. Infections that have been reported from developed countries are usually nosocomial, with community-acquired infections often being reported from India. Outbreaks appear to most likely occur in NICUs, with the source usually remaining unknown. A large multicentre, prospective study concentrating on NICUs and PICUs would be beneficial to further characterize risk factors for and current outcome of invasive infection with this organism. A study of colonization of normal newborns in tropical countries would also be of value.