Risk Factors for rectal colonization with carbapenem-resistant Acinetobacter baumannii in hospital: a matched case-control study

A retrospective matched case-control study (ratio 1:2) with a prospective inclusion of cases and concurrent selection of controls was conducted from January 2017 to December 2018 in a tertiary-care hospital. Universal active surveillance for CRAB was implemented. Univariate and multivariate logistic regression was carried out using a stepwise selection method to compare prognostic factors between cases and controls. A sub-analysis was carried out according to the type of department.


Abstract Background
During the last decade carbapenem-resistant Acinetobacter baumannii (CRAB) became hyper-endemic in hospitals due to di cult to control spreading. Our aim is to identify risk factors for nosocomial rectal CRAB colonization in an endemic hospital.

Methods
A retrospective matched case-control study (ratio 1:2) with a prospective inclusion of cases and concurrent selection of controls was conducted from January 2017 to December 2018 in a tertiary-care hospital.
Universal active surveillance for CRAB was implemented. Univariate and multivariate logistic regression was carried out using a stepwise selection method to compare prognostic factors between cases and controls. A sub-analysis was carried out according to the type of department.

Results
Forty-ve cases with nosocomial rectal CRAB colonization and 90 controls were included. One hundred and two (75%) patients were hospitalized in medical departments. At multivariable analysis signi cant risk factors associated with CRAB colonization were: use of permanent devices (OR: 10

Conclusions
Our data suggest that active surveillance for rectal CRAB colonization should be addressed to patients with an unfavourable prognosis, longer hospitalizations and carriers of multiple devices. To counter CRAB spreading in endemic settings, clinicians must limit the use of carbapenems, and reinforce interventions aimed at proper use of devices.

Background
Carbapenem-resistant Acinetobacter baumannii (CRAB) is emerging worldwide as a major cause of healthcare-associated infections (HAIs), especially in intensive care units (ICUs) [1][2][3][4]. Worryingly, during the last decade, the Italian antimicrobial resistance rates were among the highest in Europe and CRAB became hyper-endemic [5]. Its growing importance in the hospital setting is due both to the ability of these bacteria to accumulate mechanisms of resistance to antibiotics, and the ability to survive in unfavourable conditions for long periods of time [6]. CRAB infections frequently occur in patients with severe underlying diseases, the critically ill, the elderly, immunocompromised, severely debilitated or with life-threatening conditions and also in patients transferred from long-term health care facilities (LTHCFs) [7][8]. Risk factors mainly associated with CRAB infection and colonization are invasive procedures, indwelling devices and previous antibiotic treatment [9]. Moreover, cross-transmission of CRAB among hospitalized patients is promoted by poor adherence to hand hygiene practices and by repeated contact with the contaminated environment [4,[10][11][12][13]. Therefore, the rapid identi cation of CRAB asymptomatic carrier could allow an earlier introduction of infection contact precautions to prevent transmission to other patients and to the hospital environment [14]. An early recognition of CRAB carrier can also assist to identifying patients at risk of subsequent CRAB infection. Indeed, Latibeaudiere et al. demonstrated that previous CRAB colonization increased by 8 times the risk to develop a CRAB infection [4,15]. Several studies have been carried out worldwide with the aim of identifying risk factors for colonization and infection with A. baumannii, in particular focusing on CRAB [16][17][18][19]. Unfortunately, most of these studies were retrospective and conducted during CRAB outbreaks.
Furthermore, many studies were limited to ICUs where colonization and infections with Acinetobacter spp. are more frequent [1,16,[20][21], while only a few studies investigated non-ICUs settings [3,19]. Only two studies involved LTHCFs [17,22]. The most common outcome was risk factors for CRAB infection and subsequent mortality. These heterogeneous studies conducted in different epidemiological situations had methodological limitations mainly due to different selection criteria between cases and controls, and did not allow conclusive ndings [1,20,[23][24]. This study aims to identify the main risk factors associated with rectal CRAB colonization in a CRAB-endemic acute care facility.

Setting and de nitions
The study was conducted from January 2017 to December 2018 at the Modena University Hospital, a tertiary-care hospital with 1200 beds. The region is endemic for CRAB, thus, according to hospital infection control policy, a universal active surveillance was implemented during the study period with a rectal swab performed at hospital admission and repeated weekly in the whole hospital. A matched case-control study with a prospective inclusion of cases and concurrent selection of controls (ratio 1:2) was designed. A patient was de ned as a case meeting all of the following criteria: nosocomial isolation of a CRAB strain from rectal swab screening (isolated ≥ 72 hours from admission time), a negative rectal swab at hospital admission and no isolation of CRAB from any biological sample in the previous 6 months. Two controls were individually matched to each case by age, date of screening and department at the time of screening. Controls were selected among patients with a negative rectal swab for CRAB on admission and on the same day as the matched case. Similar to cases, controls had a hospital stay at time of screening longer than 4 days and no isolation of CRAB from any biological sample in the previous 6 months.
Several risk factors, based on previous literature studies were investigated and included both patient-related risk factors, such as age or the presence of comorbidities and extrinsic risk factors related to patient hospitalization. The following data were collected: age, sex, length of hospital stay (LOS) at time of positive screening, previous ICU stay, provenance of patients at admission, overall mortality at 90 days, previous hospitalizations within 6 months, comorbidities, assessed by the Charlson Comorbidity Index (CCI) and speci c covariates composing the index [25], McCabe prognostic classi cation score [26], disabilities de ned as independence in activities of daily living, permanent devices (including indwelling urinary catheters), exposure to invasive devices and medical procedures during hospitalization, immunosuppressive therapies (chemotherapy or minimum dose of 0.3 mg/kg/day of prednisone equivalent for > 3 weeks) and antibiotic administration prior CRAB colonization distinguishing between before and during the hospitalization (≤ 30 days before the admission). Antibiotic polytherapy was de ned as two or more classes of antibiotics simultaneously prescribed. The size of the study sample was calculated to detect associations with an odds ratio (OR) = 3.5, considering an error of 5%, power of 80%, and proportion of exposed controls of 25%, and allowing for the matched study design. A speci c study form has been used to collect data from hospital medical charts. The microbiology laboratory provided the list of potential matched controls for each identi ed case, ordered by the laboratory request code of the screening test. The list included all patients meeting the criteria for controls described above. All eligible cases and 2 matched controls were selected consecutively. This study was approved by the Modena University Hospital Institutional Ethics Committee with the following approval number: AOU 0025972/19 on the 25/09/2019. No written informed consent was provided to patients as all data were analysed anonymously after a de-identi cation process.

Microbiological Methods
All isolates were identi ed by MALDI-TOF MS using the VITEK MS (bioMérieux, Marcy l´Etoile, France) following the manufacturer's instructions. Antimicrobial susceptibility testing was performed by microdilution method using the antimicrobial susceptibility testing ITGNEGF panel (MICRONAUT, Merlin, Germany).

Statistical analysis
Descriptive statistics were performed for baseline demographic clinical characteristics of the entire group, as well as the groups of patients with and without colonization with CRAB. Continuous variables were presented as the number of patients (N), mean, standard deviation (SD). Unpaired Student's t test was used to compare groups. Categorical variables were presented as frequency (N, percentage [%]) and compared using Pearson's chi-squared test (Fisher's exact test was used for those variables with less than 5 events). A multivariate logistic regression model was carried out using a stepwise selection method to identify the prognostic factors between cases and controls. In the rst step, the intercept-only model was tted and individual score statistics for the potential variables were evaluated. A signi cance level of P < 0.05 was used to allow a variable into the model. In stepwise selection, an attempt was made to remove any insigni cant variables from the model before adding a signi cant variable to the model. Hosmer and Lemeshow tests were used to evaluate "goodness of t" in the selection model. Data from the univariate and multivariate logistic regression analyses were expressed as odds ratio (OR) and 95% con dence interval (CI). A P < 0.05 was considered statistically signi cant. Statistical analysis was performed using STATA® software version 14

Results
In this study, 45 (33.4%) patients colonized with CRAB were considered as cases; for each patient two controls were selected up to a total of 90 (66.6%) controls. One hundred and two (75.5%) patients were hospitalized in medical departments, including geriatric and internal medicine, 21 (15.5%) patients in ICUs and 12 (9%) in surgical departments. The baseline characteristics of the two matching groups and the comparisons are described in Table 1.  Concerning therapy, the use of antibiotics and antibiotic polytherapy during hospitalization was found to be very strong risk factors in the univariate analysis. With regard to the different antibiotic classes, carbapenems and glycopeptides statistically increased the rate of rectal colonization with CRAB, but not the use of 3rd generation Cephalosporins (3GC). Moreover, corticosteroid therapy was found to be a signi cant risk factor for CRAB. In the multivariate logistic regression analysis shown in Table 2, a signi cant independent risk factor for rectal CRAB colonization was the use of permanent devices. The use of UC and MV increased the risk by 5 and 40 times respectively. McCabe score elevated the risk for CRAB rectal colonization by 5.45 times. Another signi cant risk factor was LOS. Among antibiotic exposure, only carbapenems were signi cantly associated with rectal CRAB. Finally, there was no association with mortality.

Subgroup Analysis By Department
A subgroup analysis to identify speci c risk factors for rectal CRAB colonization in the different departments was performed (Table 3). We categorized the data according to the department in which the patient was hospitalized at the time of screening positive for CRAB.

Internal medicine department
Concerning 51 patients from the internal medicine department, among CRAB cases there were a signi cantly higher percentage of patients with partial disabilities or bedridden status compared to controls (data not shown). There were also signi cant differences between cases and controls regarding the use of devices, of MV, and antibiotics during hospitalization, antibiotic polytherapy (data not shown), and the average number of antibiotics administered during hospitalization (cases 3.11 ± 2.29 and controls 1.03 ± 1.24, P < 0.05). The univariate analysis showed that several variables had a statistically signi cant OR (Table 3). Prolonged hospitalization, previous admission to the ICU, MV, the use of permanent devices, and catheters during hospitalization represented important risk factors for rectal CRAB colonization for this speci c population.
Among these, only the use of permanent devices was con rmed as a signi cant risk factor at the multivariate analysis. Moreover, current antibiotic therapy and antibiotic polytherapy represented a signi cant risk factor for CRAB at the univariate analysis, also con rmed in multivariate analysis. In particular, the exposure to carbapenems and piperacillin/tazobactam signi cantly increased the risk of CRAB colonization by 9 and 5 times, respectively, as well as the higher number of antibiotics use (Table 3).

ICU
With regard to 7 patients acquiring rectal CRAB colonization in ICU vs 14 controls, some statistically signi cant differences were found ( Table 3). The univariate analysis showed that the McCabe Score represented a signi cant risk factor for nosocomial CRAB colonization, a fatal or rapidly fatal disease increased the risk by 8 times for acquiring CRAB in ICU. Concerning antibiotic exposure, the use of 3GC and carbapenems increased the risk of colonisation by 33 times and 15 times respectively. Multivariate analysis was not carried out due to limited simple size.

General surgery department
Concerning the population admitted to surgical departments, we could not identify any statistically signi cant risk factor for this population due to the limited sample (n = 12) also affecting the statistical analysis.

Discussion
Antibiotic therapy, disabilities and medical devices play a crucial role in rectal CRAB colonization. Our ndings support active screening strategies targeted to early identi cation of CRAB asymptomatic carriers in an endemic hospital setting, including non-ICU departments, a previously underestimated setting. The early identi cation of speci c risk factors for colonization with CRAB could become the cornerstone of a costsaving prevention strategy. In our population, diagnosis of a fatal condition and longer LOS had the greatest impact on CRAB nosocomial acquisition, especially in elderly patients. Cases had a twofold longer LOS compared to control patients, suggesting that longer hospitalization could facilitate CRAB acquisition. The risk of CRAB colonization was increased by several factors: carbapenem exposure increased the risk by 5 fold; permanent devices and MV increased the risk by 10 and 40 times respectively. Finally, among comorbidities and intrinsic risk factors [3], only peripheral vascular disease and dementia were found to be associated to CRAB colonization in our hospital setting. Several studies have shown that CRAB colonization represents the main signi cant risk factor for the development of CRAB-related infections, with very di cult therapeutic management due to limited antimicrobial treatment options, present and future [3,15]. In this scenario, active surveillance cultures with rectal screening represent one of the most important strategies of a multimodal approach in order to counter CRAB hospital spreading in highly endemic settings [4]. To our knowledge, few studies have investigated the risk factors for colonization and most of them are focused on ICUs or aimed at highlighting the variables involved in the development of infection rather than colonization [19,21,23]. These limitations could be explained by the fact that rectal screening on admission is not universally performed outside the ICU, so it is very di cult to distinguish between community-associated and nosocomial colonization. Until now the identi ed risk factors for CRAB nosocomial acquisition are: disease severity measured by scores (e.g. APACHE II, McCabe), antibiotic use, invasive procedures such as catheterisations, enteral feeding or MV, ICU stay, and LOS [3,16,19,21,27]. Therefore, multiple and permanent devices should always set up a wake-up call to screen patients admitted in non-ICU departments in endemic hospitals. Although polytrauma, blood transfusions and major surgery are signi cant known risk factors for CRAB acquisition [16] [21], while Tacconelli et al. [27], showed that methicillinresistant Staphylococcus aureus isolation and beta-lactam use were among the independent risk factors for CRAB colonization and infection. Our study has several strengths. First, it focused on CRAB colonization and not systemic infection. This is extremely important in order to better identify risk factors for colonization by eliminating possible confounders associated with CRAB infection. Second, our study is the rst to conduct a risk factor analysis according to the type of department, with a majority of patients from outside the ICU. The sub-analysis by department showed that patients admitted to different departments have different risk factors and that, as far as our centre is concerned, most of the colonization occurred in geriatric and internal medicine settings. We found that antibiotic exposure had a different effect depending on the department. In ICU, antibiotic use was the main risk factor for nosocomial CRAB colonization, and in particular the use of carbapenems and 3GC signi cantly increased the risk by 33 times and 14.7 times, respectively. In internal medicine department, the use of antibiotic polytherapy, carbapenems and penicillins (especially piperacillin/tazobactam) were among the most important risk factors for rectal CRAB colonization. The use of piperacillin/tazobactam has been poorly investigated in previous studies probably because most focused on ICU-CRAB carriers, while the highest consumption of this antibiotic occurs in medical department [3,26].
Conversely, in the geriatric setting the most important risk factors to become a CRAB carrier were the presence of invasive devices, both permanent and positioned during the hospital stay, rather than antibiotic exposure. A third strong point of our study is its design. The matching by department and date of screening allowed us to control for the variation of CRAB colonization pressure and the cross-transmission in the department, which are potential confounders of the main associations considered in this study. In fact, the likelihood of a patient to receive an antibiotic prescription in a speci c setting and the timing may be in uenced by the perceived risk of a multidrug resistant organism (MDRO) infection, as physicians might be more prone to prescribe a broad spectrum antibiotic when other patients in the department are colonized or infected by MDROs. Furthermore, concurrent selection of controls provided an accurate estimate of relative risk even in case-control studies assessing non-rare outcomes. We excluded community-colonized patients through an active surveillance on admission. This study has some limitations. It is a single-centre retrospective study carried out in a CRAB endemic area. For this reason, our model may not be applicable to different settings, and the results must be generalized with caution. Furthermore, the design of the study did not allow us to consider some important parameters such as risk factors related to infection control strategies. Since the controls for each CRAB case were selected through a matching by department and date of screening, it was not possible to analyse the role that the roommate had in the acquisition of CRAB colonization. Another limitation of our study is the selection bias due to the case-control approach. We have not analysed all the patients colonized in the time frame considered, but 45 randomly selected patients who tested positive for rectal screening after at least 4 days of hospitalization. Unfortunately, no speci c risk factors were identi ed for the surgical department, because the surgical sample was too small. Further investigations are needed in this setting.

Conclusions
As CRAB colonization is one of the most important risk factors for CRAB infection, targeted infection prevention and control (IPC) practices are crucial to limit CRAB spreading in hospitals, not only in ICUs, but also general departments where this pathogen is widely found. Rectal screening could represent a cornerstone to limit CRAB dissemination in endemic settings. By identifying the main risk factors associated to in-hospital rectal CRAB-acquisition in different wards, this study provides new real-life elements to limit CRAB cross-transmission and to target CRAB-IPC interventions. In the era of MDROs clinicians should strictly adhere to and apply bundles for permanent devices and invasive procedures management in order to target their use. Finally, our study supports the fundamental principles of antimicrobial stewardship, i.e. limiting the use of inappropriate antibiotic therapy, avoiding polytherapy and reducing carbapenem use in CRAB endemic hospitals.

Declarations
Ethics approval and consent to participate This study was approved by the Modena University Hospital Institutional Ethics Committee with the following approval number: AOU 0025972/19 on the 25/09/2019. No written informed consent was provided to patients as all data were analysed anonymously after a de-identi cation process.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.