Identifying risk factors for carbapenem-resistant Acinetobacter baumannii carriage upon admission: a case-case control study

Abstract

Background

Active screening programs and early detection of asymptomatic carriers are effective in preventing carbapenem-resistant Acinetobacter baumannii (CRAB) dissemination in healthcare facilities. This study aims to identify risk factors associated with CRAB carriage among patients upon admission to an acute care hospital.

Methods

A case-case-control study was conducted at an acute care hospital. Starting in June 2020, new admissions to medical wards underwent rectal and buccal screening. Patients with CRAB or carbapenem-susceptible A. baumannii (CSAB) carriage were compared to controls, randomly selected from patients with negative cultures, at a one-to-one ratio. Multinomial logistic regression using a backward stepwise method was employed to identify factors associated with CRAB and CSAB carriage. A Chi-square Automatic Interaction Detector analysis was also conducted to further elucidate risk factors.

Results

The study included 115 CRAB carriers, 117 CSAB carriers and 121 controls. Increasing age was associated with a reduced risk of CSAB (OR: 0.96, p < 0.001) and CRAB carriage (OR: 0.97, p = 0.02), while higher Charlson Comorbidity Index scores increased the risk for both. CRAB carriage was significantly associated with admission from long-term acute care hospitals (OR: 7.68, p < 0.001) and presence of pressure ulcers (OR: 89.98, p < 0.001). Decision tree analysis identified pressure ulcers, prior location, and Charlson score as key predictors, with CRAB carriage reaching 77.3% in patients admitted from long-term acute care hospitals with pressure ulcers.

Conclusion

Pressure ulcers were strongly associated with the carriage of both susceptible and resistant strains of A. baumannii. CRAB carriage was predominantly observed in patients transferred from long-term acute care hospitals, highlighting the need for targeted screening in this high-risk population.

Introduction

The rapid and widespread emergence of carbapenem-resistant Acinetobacter baumannii (CRAB) within healthcare facilities constitutes a significant threat to patient safety and the efficacy of healthcare delivery [1, 2]. The prevalence of CRAB significantly varies by geography and healthcare facility [3], necessitating the development of tailored strategies for its prevention and control. This involves ongoing monitoring of local CRAB prevalence, implementing antimicrobial stewardship programs, and reinforcing infection control practices [4].

Asymptomatic carriers of carbapenem-resistant Gram-negative bacteria serve as ongoing reservoirs, introducing these pathogens into healthcare facilities and contributing significantly to intrahospital and interhospital dissemination [5, 6]. Implementation of active screening programs targeting high-risk groups is a fundamental element in multifaceted strategies of CRAB prevention [7]. Hence, a comprehensive understanding of the epidemiological characteristics of asymptomatic CRAB carriers may contribute to the development of strategies aimed at preventing the transmission of these pathogens within healthcare facilities.

Acinetobacter spp. are commonly present in the commensal skin flora of humans, occurring in approximately 25–70% of individuals [8]. The predominant species found in skin flora include A. johnsonii and A. lwoffii. Yet, the majority of infections and hospital outbreaks are attributed to the A. baumannii complex [9]. Early studies over a decade ago reported low A. baumannii carriage rates in the community, with European and U.S. surveys showing less than 1% carriage among non-critically ill patients, and no CRAB colonization [10, 11]. This led to the assumption that routine screening of non-critically ill patients upon admission is expected to yield minimal benefit [11]. However, over the past decade, A. baumannii has become more prevalent in non-ICU wards, with a surge in early healthcare-associated infections [12]. Furthermore, the ongoing spread of CRAB, predominantly following the emergence of COVID-19 [13], may have altered the epidemiology of CRAB carriage among individuals admitted to healthcare facilities.

The incidence of CRAB infections in Israeli acute care hospitals has risen since the early 2000s [14]. In response to increasing CRAB rates, our facility implemented a comprehensive control program in June 2020 [15]. This included the screening of high-risk groups upon admission, resulting in a significant rise in the detection rate of imported CRAB cases from 24.3% of total cases identified in 2020 to 61.0% in 2022. Concurrently, there was a notable decrease in nosocomial CRAB infections. The active screening policy, as part of our multifaceted intervention to prevent CRAB infections, was developed based on general MDRO risk factors identified from the literature. However, risk factors for A. baumannii carriage in patients admitted to acute care hospitals in endemic regions remain poorly understood. The primary objective of this study was to refine and identify the risk factors associated with the carriage of CRAB among patients admitted to medical wards in an endemic region. A secondary objective was to evaluate the impact of such carriage on the subsequent clinical infections and in-hospital mortality.

Methods

Study design: A non-matched, hospital-based, case-case-control study was conducted to assess factors associated with the isolation of resistant strains upon hospital admission. Study participants included newly admitted hospital patients identified with either CRAB (case 1) or carbapenem-sensitive A. baumannii (CSAB, case 2) through screening cultures conducted within the initial 48 h of hospitalization. Patients with CRAB or carbapenem-susceptible A. baumannii (CSAB) carriage were compared to controls, selected randomly from patients with negative cultures, at a one-to-one ratio. The study was carried out from June 2020 to December 2022 in Wolfson medical center, a 670-bed teaching hospital in central Israel. The facility houses six medical wards with a median patient age of 77 years. Notably, approximately 50% of patients hospitalized in these medical wards have a recent history of hospitalization within the last six months, and an additional 20% were transferred from long-term care facilities.

Active screening policy

Rectal and buccal mucosa specimens were collected from patients upon admission to the medical wards. The screening was specifically targeted towards patient populations at higher risk, including those transferred from long-term care facilities, patients with a history of hospitalization within the prior six months and functionally disabled patients [16].

Sample collection and microbiologic methods

Buccal mucosa and rectum were sampled using TransystemTM swabs (Copan Diagnostics Inc.) and transferred to the laboratory. The swabs were inoculated onto CHROMagar™ mSuperCARBA™ (Hilab, Rehovot, Israel). Plates were then cultured at 37 °C for 18–24 h. Suspicious colonies (white colonies) were identified to the species level using VITEK-MS (bioMérieux, Marcy l’Etoile, France). Carbapenem resistance was determined using meropenem (10 µg) disks. Clinical isolates were also identified with the VITEK-MS system, and susceptibility testing was performed using the VITEK 2 AST-N308 card (bioMérieux, Carponne, France), with resistance determined according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [17].

Data collection

The following information was collected from the medical records of patients from all 3 patient groups including demographics; admission source (home, nursing home and long-term acute care (LTACH)); presence of comorbidities, Charlson Comorbidity Index, invasive devices on admission, and exposure to antimicrobial agents within the preceding month. Data regarding the use of invasive devices were tracked during the current hospital stay. Outcomes measured were the occurrence of positive clinical cultures and in-hospital mortality. Subsequent A. baumannii infection was referred as the growth from clinical specimens in the presence of clinical manifestations of infection.

Statistical analysis

We conducted a comparative analysis of patient characteristics among three distinct groups: CRAB carriers, CSAB carriers and controls. Categorical variables were described as frequency and percentage. Continuous variables were evaluated for normal distribution using histogram and reported as median and interquartile range. Chi-square test and Fisher’s exact test were used to compare categorical variables between groups while Kruskal-Wallis Test were used to compare continuous variables.

We used multinomial logistic regression to assess the association between independent variables and the outcomes: CRAB, CSAB, and controls. The model estimated the odds of CRAB and CSAB carriage relative to controls, adjusting for covariates. A backward selection method with a P-value threshold of > 0.1 was used to remove non-significant variables. In addition, a CHAID (Chi-squared Automatic Interaction Detector) classification tree model was applied to explore associations between the outcome and multiple predictors [18]. In the initial step, the most significant predictor (with the highest χ² value) was selected to split the root node, which contained all cases, into two subgroups (child nodes). This procedure was repeated for each subsequent node, with each split based on the next most significant predictor. The process continued until no further statistically significant predictors remained, resulting in terminal nodes.

In-hospital mortality was assessed using bivariate and multivariate logistic regressions. Variables with an unadjusted P value of less than 0.1 were included in the multivariate logistic regression to obtain adjusted P values. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated.

All statistical tests were two sided and p < 0.05 was considered as statistically significant. Statistical analysis was performed using SPSS statistical software (IBM SPSS Statistics for Windows, version 28, IBM Corp., Armonk, NY, USA, 2021).

Results

Patient characteristics

From June 2020 to December 2022, 2,832 patients were screened upon hospital admission, identifying 117 with CSAB and 115 with CRAB. A control group consisting of 121 patients were randomly selected from patients with negative screening cultures. The median age of the patients was 80 years, with an interquartile range (IQR) of 70 to 88 years, and 53.0% were male. High levels of functional dependence were noted in 77.3% of the patients, and 36.8% were transferred from LTACH. Predominant comorbidities included dementia (54.1%), and diabetes mellitus (50.0%), alongside with a notable incidence of pressure ulcers (45.0%). The median Charlson Comorbidity Index was 2 (IQR 2–5).

Risk factors for carriage

The bivariate analysis revealed significant demographic and clinical differences between the three groups (Table 1). CRAB and CSAB carriers were younger, with median ages of 80 (CSAB) and 76 (CRAB) compared to 84 in controls (p = 0.004). Pressure ulcers were more common among CSAB (53.8%) and CRAB carriers (80.9%) compared to controls (2.5%) (p < 0.001). CRAB carriers had a higher prevalence of heart failure (36.8% vs. 21.5% in controls, p = 0.019), dementia (69.6% vs. 44.6%, p < 0.001), and greater functional dependency (88.7% vs. 69.4%, p = 0.02). CRAB carriers were more often transferred from LTACHs (74.8% vs. 13.2% in controls, p < 0.001), had more frequent hospitalizations in the last six months (71.3% vs. 45.5%, p < 0.001), and had greater antibiotic exposure in the prior month (36.5% vs. 19.8%, p = 0.017). Invasive device use was also significantly higher in CRAB carriers, both upon admission (indwelling urinary catheter: 36.5% vs. 8.3%, p < 0.001; mechanical ventilation: 28.7% vs. 2.5%, p < 0.001) and during hospitalization (mechanical ventilation: 47.0% vs. 19.8%, p = 0.001; central venous catheter: 40.9% vs. 13.2%, p < 0.001).

Table 1 Demographic and clinical characteristics of the groups based on results of surveillance cultures

Table 2 presents the results of the multinomial logistic regression. Increasing age was associated with a lower risk for CSAB (OR: 0.96, p < 0.001) and CRAB (OR: 0.97, p = 0.02), while a higher Charlson Comorbidity Index increased the risk for both (CSAB: OR: 1.24, p = 0.03; CRAB: OR: 1.54, p < 0.001). CRAB carriage was strongly linked to LTACH residence (OR: 7.68, p < 0.001) and pressure ulcers (OR: 89.98, p < 0.001). Pressure ulcers were also a significant predictor of CSAB carriage (OR: 56.56, p < 0.001).

Table 2 Multinomial logistic regression analysis of risk factors associated with carriage of carbapenem-resistant and carbapenem-susceptible A. baumannii strains

Chi-squared automatic interaction detection decision tree analysis of A. Baumannii carriage

The predictive model included four variables: pressure ulcer, prior location before admission, dementia, and Charlson comorbidity index (Fig. 1). The risk of A. baumannii carriage was significantly higher among patients with pressure ulcers (98.1%) compared with those without (39.1%, P < 0.001). Among patients with pressure ulcer, transfer from LTACH demonstrated a 77.3% risk of CRAB carriage, significantly higher than the 29.0% observed in those from home or nursing settings (P < 0.001). Patients from domestic or nursing home settings without pressure ulcers and a low Charlson Comorbidity Index had a minimal CRAB carriage rate of 2.5%.

Fig. 1

Chi-square Automatic Interaction Detector to identify CRAB and CSAB carriage risk factors. Note: CRAB, carbapenem resistant A. baumannii; CSAB, carbapenem sensitive A. baumannii; LTACH, long-term acute care hospital

Outcomes

During hospitalization, the use of mechanical ventilation, central venous catheterization, and urinary catheterization was significantly more frequent among CRAB carriers (Table 1). Of the 115 CRAB carriers, 32.2% had at least one subsequent clinical culture positive for A. baumannii, predominantly with carbapenem-resistant strains (97%). Positive samples were primarily from sputum [19], followed by wound [12] blood [7], and urine [3]. Among CSAB carriers, 6.8% (8 out of 117) had positive cultures, with a single resistant isolate. No positive clinical cultures were identified among the control patients.

CRAB carries had significantly higher in-hospital mortality (47.0%) compared to those with CSAB (24.8%) or no colonization (25.6%; P = 0.01). Multivariate analysis identified several factors associated with mortality, including advanced age, renal disease, malignancy, admission infectious disease diagnosis, mechanical ventilation and indwelling urinary catheters, as shown in Table 3. Notably, the status of A. baumannii carriage and the presence of invasive CRAB infection was not associated with mortality.

Table 3 Bivariate and multivariate analysis of risk factors associated with in-hospital mortality

Discussion

Early identification of asymptomatic carriers, along with the enforcement of contact precautions, has a pivotal role in the control of CRAB [19, 20]. Our study aimed to identify risk factors for CRAB carriage upon admission to an acute care hospital in an endemic region. CRAB carriage was primarily detected in patients transferred from LTACH. These findings underscore the importance of targeted active screening for this at-risk population.

A case-case-control methodology was previously suggested to overcome limitations of traditional case-control studies [21]. Prior studies have predominantly focused on the carriage of resistant Acinetobacter species, with less emphasis on carbapenem-susceptible strains. Our study is among the few to report similar numbers of CSAB and CRAB in the same population, offering a broader understanding of A. baumannii carriage risk factors. By comparing resistant and susceptible strains to controls, we identified shared and unique risk factors, enhancing our understanding of resistance determinants. Carriage of both susceptible and resistant strains were linked to pressure ulcers, while resistant strains were associated with admission from LTACH. In addition, we employed the CHAID classification tree model to examine variable relationships. This model has been effective in identifying and predicting risk factors, offering a robust alternative to traditional logistic regression [18]. Notably, CHAID’s performance is uncompromised by collinearity and outliers, and can effectively manage interactions between multiple factors [22]. In the current study, the four predictors of A. baumannii carriage were presence of pressure ulcer, transfer from LTAC, dementia and increased Charlson Comorbidity Index. Notably, the presence of a pressure ulcer among patients transferred from LTACH were indicative of a high likelihood of CRAB carriage, with more than three-quarters of such patients being carriers. Therefore, the CHAID-generated model elucidates the complex interplay among risk factors, pinpointing patients at elevated risk for CRAB carriage.

Previous studies primarily focused on identifying risk factors for CRAB infections or carriage in intensive care units [23,24,25]. These studies found that older age, antibiotic exposure, and prior hospitalization within the previous six months were associated with an increased risk of CRAB infection or carriage. Nevertheless, identifying asymptomatic carriers upon hospital admission to non-ICU wards could serve as a crucial component in preventing the intrahospital dissemination of CRAB. This study is the first to explore risk factors for asymptomatic carriage among patients admitted to medical wards. We found a strong association between CRAB carriage and transfers from long-term care facilities. Notably, we observed different risk patterns when comparing patients admitted from LTACH to those from other long-term care facilities. LTACH admit patients with complex medical conditions, requiring specialized treatments including mechanical ventilation, prolonged IV antibiotics and complex wound care. LTACHs have been identified as significant contributors to regional dissemination of multidrug-resistant organisms, particularly carbapenemase-producing Enterobacteriaceae [26, 27]. The transfer of patients colonized with resistant pathogens from acute care facilities to LTACH serves as a pathway for introducing resistant organisms into these settings. In LTACHs, resistance persists due to a combination of host-related characteristics and institutional factors, contributing to ongoing transmission risks [28] and leading to the reintroduction of these organisms to acute care facilities at subsequent admissions. Although data on CRAB prevalence in LTACHs is limited, recent exposure to LTACH was associated with CRAB carriage upon admission to acute care hospitals [29]. These findings hold practical relevance, especially in regions with a high incidence of CRAB infections. They emphasize the necessity of instituting active screening programs that specifically focus on patients with recent exposure to LTACH.

Notably, we have observed that the presence of pressure ulcers predicts both carbapenem-sensitive and resistant A. baumannii carriage. Chronic wounds, including pressure wound, have been consistently linked to multidrug-resistant organisms carriage [30], and were also associated with MDR Acinetobacter infections among patients with spinal cord injuries [31]. These findings highlight the potential for cross-transmission via nursing staff, driven by the heightened need for more intensive nursing care.

Our results are in line with prior studies linking CRAB carriage with increased risk of clinical infections [25, 32]. In contrast, carriers of susceptible strains demonstrated lower infection rates, which may be attributed to the lower incidence of comorbid conditions and invasive devices utilization, While higher mortality rates were observed in CRAB carriers, adjusting for comorbidities and invasive device use diminished the apparent association between CRAB colonization and mortality.

The study has some limitations. First, the study was conducted within a single healthcare facility, which restricts the external validity of its findings to other healthcare settings. The specific temporal and geographical context in which the study was carried out may limit the generalizability of the results to other regions or timeframes. Nonetheless, increased risk of CRAB carriage among patients transferred from LTACH has been previously reported from other regions [29]. Furthermore, the median age of patients in our hospital’s medical wards exceeds the national average by five years [33], possibly reflecting local demographics and referral practices from nearby long-term care facilities. Secondly, the information was collected retrospectively from medical records, which may be subject to inaccuracies or incomplete data due to the inability to verify all details. Thirdly, our screening approach did not include skin sampling, a method previously demonstrated to increase the detection of asymptomatic CRAB carriers [34] This could introduce potential misclassification bias, as some patients not identified by rectal and buccal mucosa swabs might be inaccurately categorized as controls. Nevertheless, employing both buccal and rectal swabs was shown to achieve a sensitivity exceeding 90% [35]. To address this limitation, we selected as controls only patients with negative swabs from both sampling sites.

Conclusion

The presence of pressure ulcers was associated with both susceptible and resistant A. baumannii carriage. Notably, carbapenem-resistant strains were predominantly found in patients transferred from LTACH. These findings underscore the importance of targeted active screening for this at-risk population. The observed high prevalence of CRAB among patients from LTACH emphasizes the necessity for comprehensive regional intervention programs encompassing both acute and long-term care facilities to effectively manage and control the spread of CRAB.