Background

Citrobacter species are ubiquitous in the environment, and have long been considered pathogens of low virulence, causing infections less frequently compared to other Enterobacterales [1, 2]. As such, they are not considered classic nosocomial pathogens [3, 4]. In recent years, however, nosocomial Citrobacter spp. infections and hospital outbreaks have been increasingly reported [5, 6]. For instance, a C. freundii outbreak in a neonatal intensive care unit attracted public attention in Korea, after four neonates died of bacteraemia following receipt of a contaminated intravenous (IV) infusion [7].

Together with the growing body of evidence on Citrobacter spp. infections in hospitals, reports on antibiotic resistance among Citrobacter isolates have been also evolving, including reports on carbapenemase-producing [8], and AmpC β-lactamase (Amp-C) carrying isolates [9]. Several carbapenemases, carried on plasmids, have been described in Citrobacter spp., that can easily spread to other Enterobacterales species [10]. Nonetheless, the magnitude of Citrobacter spp. involvement as a clinically significant pathogen in hospitalized patients is not well established, and antibiotic resistance patterns in Citrobacter spp. have not been yet reviewed. Understanding the epidemiological features of this emerging pathogen, is essential to uncover its role in healthcare and to develop effective control strategies.

We conducted a systematic review and meta-analysis to evaluate the epidemiology of infections due to Citrobacter spp., and their antibiotic resistance patterns among hospitalized patients. We also examined the occurrence of hospital outbreaks due to Citrobacter spp.

Methods

Eligibility criteria

The eligibility criteria for study selection were defined using the PICOS framework (Patient, Intervention/exposure, Comparison, Outcome, Study design) [11]. Eligible study populations were hospitalized patients of any age, diagnosed with Citrobacter spp. infections, as well as those identified with colonization and/or infection due to Citrobacter during hospital outbreaks. Antibiotic resistance mechanisms of interest were 3rd generation cephalosporin and/or carbapenem resistance. Outcomes included prevalence and incidence of Citrobacter infections among hospitalized patients, prevalence/incidence of nosocomial Citrobacter infections and resistance percentages to the above-mentioned antibiotics. Frequency of reported hospital outbreaks due to Citrobacter was also evaluated. Eligible study designs were observational studies (cohort, cross-sectional, case–control studies, and case series), clinical trials, outbreak reports, and surveillance studies (Additional file 1, Review definitions). For an outbreak report to be included, Citrobacter spp. had to be the main implicated pathogen, defined as the responsible pathogen for at least one third of the detected cases. Eligible surveillance studies needed to be of at least one year duration and include a minimum of 30 Citrobacter isolates to be included. Studies reporting aggregate data on multiple Enterobacterales, or on other Enterobacterales, and studies focusing only on community-acquired infections were excluded.

Information sources and search strategy

A detailed study protocol was published on 18 January 2023 on PROSPERO (CRD42023390084) [12]. Data sources were MEDLINE® (PubMed), Embase (Ovid), outbreak database [13], and grey literature including Global Index Medicus, US Centers for Disease Control and Prevention (CDC), and the European Centre for Disease Prevention and Control (ECDC) websites. The search included publications during the period Jan-2000 to Dec-2022 without language restriction. The Medline search strategy included a combination of MeSH terms and keywords, encompassing the following search concepts: Citrobacter, nosocomial (or healthcare- or hospital-acquired) infections, hospitalized patients, outbreak and surveillance. The search terms were modified as required for each of the other databases (Additional file 1, Search strategy). A systematic reference search was performed for all included Citrobacter outbreak studies.

Study selection

A summary list of all titles/abstracts was generated according to the search terms. Searches from different databases were combined and de-duplicated using Covidence (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia) [14]. Single screening of titles/abstracts was performed by one reviewer (PF), complemented by additional discussion with a second reviewer (NHK), as needed. Two reviewers (PF and NHK) performed double full-text screening; any uncertainties were resolved by consensus. Data extraction was completed by a single reviewer (PF), with double extraction of 50% of included publications by a second reviewer (NHK). Data was extracted into dedicated forms designed in Covidence.

Data extraction

The following data was extracted: bibliographic information, study design and setting, study characteristics (i.e., objectives, sample size, age groups). Type(s) of clinical infections, and unit of analysis (infected patient or cultured isolate). Microbiological analysis methods (i.e. phenotypic and genotypic resistance evaluation) were also recorded.

Data on prevalence and incidence of Citrobacter spp. infections among hospitalized patients and resistance percentages of Citrobacter spp. isolates was retrieved. We recorded the percentage of isolates that were resistant to third generation cephalosporins or carbapenems or that produced extended-spectrum beta-lactamases (ESBL), AmpC beta-lactamases or carbapenemases. For outbreak study reports, data on setting, timing, and duration of each outbreak, detected outbreak source(s) and interventions implemented to halt the outbreak were collected.

Methods of data synthesis

The characteristics of the included studies were described. Whenever available, prevalence and incidence rates were reported. Antimicrobial resistance percentages reported in observational and surveillance studies were meta-analysed to generate pooled estimates for both resistance mechanisms and resistance to specific antibiotic agents. Random effects models were used (when 3 or more studies reported specific resistance data). Freeman-Tukey double arcsine transformation was used to stabilize the variances [15], and statistical heterogeneity was assessed using the I2 statistic measure [16]. Studies focusing only on multidrug-resistant Citrobacter isolates and those reporting on < 10 Citrobacter isolates were excluded from the meta-analysis. Data on hospital outbreaks of Citrobacter spp. were summarized descriptively. Statistical analysis was done using ‘meta’ package, RStudio (Version 4.2.3).

Risk of bias assessment

Risk of bias was assessed using the Joanna Briggs Institute (JBI) study design tools [17]. A study was defined at low risk of bias when it scored was ≥ 75% of the applicable score. Quality of reporting in hospital outbreaks was evaluated by compliance with the ORION recommendations [18]. One reviewer (PF) assessed risk of bias and reporting quality; risk of bias in 50% of all included studies was also evaluated by a second reviewer (NHK), with no major inconsistencies.

Results

Study selection and characteristics

A total of 1609 de-duplicated publications were identified and reviewed by title/abstract. Of these, 148 full-text articles were reviewed. Finally, 41 studies fulfilled the inclusion criteria (Fig. 1): 15 observational studies (10 cohort studies, four cross-sectional studies and one case-series); 13 surveillance studies and 13 outbreak reports. The main reasons for exclusion were surveillance studies including less than 30 Citrobacter spp. isolates (n = 41) and incompatible study design (n = 20). Most included observational studies were single-center studies (87%) whereas surveillance reports often included data from multi-center networks or reference laboratories (10/13, 77%; Table 1). Intensive care units (ICUs) were the most frequently implicated hospital department (14/28, 50%), with six studies focusing only on ICU patients (Table 1). Most observational studies were conducted in Asia, with the highest number of studies from India (n = 5). Three studies reported international surveillance data. Germany, Spain and USA were the countries contributing most Citrobacter surveillance data (three studies each).

Fig. 1
figure 1

PRISMA* flowchart for the systematic review

Full size image
Table 1 Characteristics of the included studies in the systematic review (n = 41)
Full size table

Citrobacter infections among hospitalized patients

Out of 28 observational and surveillance studies, 15 studies (54%) focused on patients infected with Citrobacter, while the remaining included also other Enterobacterales infections. Across all studies, C. freundii was the most frequently species (reported in 22/28, 79%), followed by C. koseri/C. diversus (11/28, 39%) and C. braakii (5/28, 18%). Other species included C. amalonaticus, C. youngae, C. portucalensis and C. europaeus (Table 1). Most studies provided a clear definition of clinical infection; yet 13/28 studies (46%) only reported on Citrobacter spp. growth in clinical cultures without providing additional clinical information (Table 2). Citrobacter bloodstream infections (BSI) were the focus of four studies [2, 5, 22, 23].

Table 2 Summary of epidemiological and microbiological features in the included observational and surveillance studies
Full size table

In observational studies, a median of 65 patients with Citrobacter infections were included per study (interquartile range (IQR), 42–157), contributing to a total of 4617 Citrobacter patients. In surveillance studies, a median of 279 Citrobacter isolates were included per study (IQR, 52–834), contributing to a total of 6582 isolates. An increasing number of patients with Citrobacter infection/colonization were reported in observational studies after 2010 (Additional file 1, Figure S1).

Data scarcity prevented generating pooled incidence estimates; two studies provided denominator data quantifying the size of population at risk, yielding a cumulative incidence of 0.175 and 0.035 episodes per 1000 patients for Citrobacter BSI and invasive Citrobacter infections, respectively [23, 26].

Among hospitalized patients, UTI was the most frequently reported Citrobacter infection (17/28 studies, 61%) followed by BSI (15/28 studies, 54%), and respiratory-tract infection (RTI) in 8 studies (29%, Table 2). In most studies the exact date of infection-onset was not clearly defined. Yet, seven studies reported separately on patients with hospital-acquired Citrobacter infections; 85% (838/990) of hospitalized Citrobacter patients in these studies had a nosocomial infection. In three studies reporting patient mortality after nosocomial Citrobacter BSI, a case fatality ratio of 34% (36/106 patients) was found [2, 23, 25].

Citrobacter antibiotic resistance patterns among hospitalized patients

A total of 11,199 Citrobacter isolates were analyzed (4617 and 6582 from observational and surveillance studies, respectively). Urine and blood isolates were most common in observational studies, whereas the specimen type was often unspecified in surveillance studies (Additional file 1, Figure S2).

Phenotypic resistance to antibiotics was assessed in all included studies, and genotypic resistance in 11/28 studies (Table 2). Pooled resistance percentages from observational studies were higher than those from surveillance studies (Table 3). The pooled percentage of ESBL-producing Citrobacter was 22.2% (95% CI 3.5% – 50.3%, 8 studies), and for AmpC production, 33.3% (95% CI 13.2% – 53.4%, 4 studies, Table 3). Pooled resistance percentages for specific antibiotic agents in observational studies ranged between 26.4% for imipenem resistance (95%CI 0.0% – 54.6%, 6 studies) and 64.9% for ceftazidime resistance (95%CI 44.5%—82.9%, 6 studies), and in surveillance studies, between 0.1% for imipenem resistance (95%CI 0.0%-0.4%, 5 studies) and 21.7% for ceftazidime resistance (95%CI 5.0% – 45.4%, 6 studies). Of note, high resistance percentages were observed for other antibiotic agents, such as ciprofloxacin and gentamicin. Forest plots for resistance analyses are provided in Additional file 2.

Table 3 Pooled estimates of resistance percentage of included Citrobacter spp. isolates, per resistance mechanism and antibiotic agent stratified by study category
Full size table

Large heterogeneity was observed in the meta-analysis for all antibiotics. Significant subgroup differences between observational and surveillance studies were found for imipenem and ceftazidime (Additional file 2). In a subgroup analysis of observational studies focusing only on Citrobacter BSI, pooled resistance percentages to cefotaxime of 46.5% (95%CI 32.6–60.6, I2 = 71%, 4 studies) and negligible resistance to imipenem (95% CI 0 – 0.6, I2 = 0%, 3 studies) were found.

Nosocomial Citrobacter Outbreaks

Thirteen Citrobacter hospital outbreak reports were included, with a notable increase in reporting after 2016 (Table 4). Outbreaks frequently occurred in ICUs (n = 5), surgery and hematology units (3 each). C. freundi was the most often implicated species (10/13). Frequently detected carbapenemase and ESBL-production genes in outbreak isolates were OXA-48, KPC, CTX-M and AmpC cephalosporinase genes (Table 4). Two point-source outbreaks were identified, one tracked back to a staff member and the other to use of a contaminated solution for intravitreal injection [45, 56]. Other outbreaks were attributed to the hospital kitchen, or hospital toilets and sinks (Table 4). Non-point-source outbreaks lasted for a median duration of 212 days (IQR, 134–471), and a median of seven patients with Citrobacter infection and/or colonization were detected per outbreak (IQR, 5–16). The case fatality of Citrobacter infection was 7% (6/89 patients) based on three outbreak studies reporting mortality [48, 52, 55]. Citrobacter outbreaks were reported as definitively controlled following the implementation of various preventive measures in 7/13 reports (Additional file 1, Table S1).

Table 4 Main characteristics of included outbreak reports
Full size table

Risk of bias and quality of reporting assessment

High risk of bias was observed (6/10 cohort studies and 2/4 cross-sectional studies, additional file 1, Figures S4-S7). Domains of high risk of bias were confounder identification and adjustment, exposure classification and adequacy of follow-up. Conversely, a good quality of outbreak reporting was found as evaluated by the ORION statement.

Discussion

To the best of our knowledge, this is the first systematic review focusing on Citrobacter spp. infections among hospitalized patients. By including 41 studies across different study designs, we could portray a comprehensive picture of endemic and epidemic Citrobacter spp. infections in the hospital setting. C. freundii was found as an important, emerging multidrug-resistant pathogen, causing diverse nosocomial infections and outbreaks, increasingly reported since 2016. Interestingly, half of all included studies (21/41) were conducted in Asian countries, hinting at the importance of Citrobacter as a multidrug-resistant pathogen in that region.

Our findings confirm that Citrobacter spp. frequently harbour multiple resistance elements; several types of carbapenemase, beta-lactamase and AmpC-cephalosporinase resistance genes were found in the included studies. Overall, high antibiotic resistance percentages were identified in Citrobacter isolates, especially for 3rd generation cephalosporins, gentamicin and fluoroquinolones. This is an alarming finding, limiting the available treatment options for Citrobacter infections [4, 57]. Of note, pooled resistance percentages were lower among isolates collected for surveillance purposes compared to those in observational studies, a finding that can be explained by the different target populations in these types of studies [58].

We found substantial resistance to cefotaxime in Citrobacter blood isolates (46.5%), which is comparable to cefotaxime resistance in other Enterobacterales monitored in the Global Antimicrobial Resistance Surveillance System (GLASS) network, with 63% of Klebsiella pneumoniae and 38.5% of Escherichia coli found resistant to cefotaxime in blood isolates collected in 2020 [59]. In light of this finding, systematic monitoring of antimicrobial resistance in Citrobacter spp. should be considered.

Although resistance percentages are important for the clinician prescribing an empirical therapy, these are less informative for public-health purposes; they are often based on biased estimates, and do not reflect the magnitude of the problem as rate-based estimates [60]. Due to data scarcity, we were unable to generate pooled estimates of the incidence of multidrug-resistant Citrobacter infections.

Many Citrobacter hospital outbreaks identified in our review were related to the hospital environment (sinks, toilets, and kitchens); this finding aligns with the study by Hamerlinck et al., who showed that carbapenem-resistant Citrobacter can evolve in the hospital aquatic environment, and suggested long-term persistence of this pathogen in the hospital plumbing system [61]. Of note, Citrobacter was also responsible for two point-source outbreaks, emphasizing its ability to contaminate a common source. Transition from epidemic to endemic occurrence was observed in almost one third of included outbreaks, for which definitive outbreak control was not achieved according to the publication, despite multiple interventions. The diverse outbreak sources and transmission patterns of Citrobacter call for increased awareness of the risk of nosocomial Citrobacter clusters and reinforcement of infection control measures related to aseptic procedures, pharmaceutical preparations and environmental hygiene.

Citrobacter infections may cause life-threatening infections [24, 62]. In a historical cohort study from Taiwan, 45 patients with Citrobacter BSI had an overall case-fatality ratio of 33% [63], similar to the ratio of 34% found in our review. Moreover, we documented a case-fatality ratio of 7% among patients affected by Citrobacter outbreaks.

Large heterogeneity was observed in the pooled resistance estimates that could be related to true differences in epidemiologic or microbiological methods, or patient case-mix. We tried to control for heterogeneity due to study design/case-mix by analysing resistance percentages in observational and surveillance studies separately. A subgroup analysis of resistance percentages in blood isolates was also conducted. However, the number of studies identified did not allow for further subgroup analyses.

This systematic review has limitations. First, our findings might underestimate resistant Citrobacter involvement in surveillance studies and hospital outbreaks, as we excluded surveillance studies with less than 30 Citrobacter isolates and outbreaks in which Citrobacter spp. was not the main pathogen. Second, we aimed to assess the magnitude of hospital-acquired Citrobacter infections; however, only seven studies clearly distinguished between community vs. hospital-acquisition. Nonetheless, 85% of Citrobacter infections were hospital-acquired when reported. Third, there was large variability in microbiologic methods, which might have affected the results of the individual studies. Forth, multiple specimen types were included and stratified analysis was only possible for blood isolates. Last, publication bias might have affected our findings both for resistance percentages and involvement of Citrobacter spp. in hospital outbreaks.

Conclusions

In conclusion, based on the reviewed studies, Citrobacter represents an emerging multidrug-resistant pathogen in hospitalized patients. The increased resistance among Citrobacter isolates, its ability to harbor numerous resistance genes, and its active role in hospital outbreaks all make Citrobacter an important, global patient safety risk. Our findings call for inclusion of Citrobacter spp. in surveillance networks as a pathogen of epidemiological significance, as done for Enterobacter spp. In addition, future studies need to address the role of Citrobacter spp. in nosocomial infections and better elucidate its reservoirs and transmission routes in the hospital environment.