Multi-Drug Resistant Gram-Negative Infections Among Critically Ill Patients: Analysis of Baseline Characteristics and Factors Associated with Mortality


 BackgroundBacterial infections are a frequent cause of hospitalization and a leading cause of death, particularly with the emergence of antibiotics resistance. The emergence of Carbapenem resistance among gram-negative bacteria (GNB) is one of the evolving alerts as its use is considered the last resort of treatment [1]. Therefore, this urged studying the risk factors for the development of multi-drug resistant [2] GNB, identify the clinical outcomes and factors associated with mortality, especially among critically ill patients who are expected to have the worst outcomes.Materials/methodsThis is a retrospective observational study of critically ill patients who had an infection with Carbapenem-resistant Enterobacteriaceae (CRE), or MDR Pseudomonas aeruginosa, or MDR Acinetobacter spp. between May 2016- Nov 2018. Baseline demographics, co-morbidities, and clinical outcomes were collected and were evaluated for association with 28 days mortality. ResultsA total of 255 patients with MDR Gram-negative cultures were screened, 77 patients met the inclusion criteria. Pseudomonas aeruginosa was the most common index organism (53% of patients), followed by Acinetobacter spp. and CRE, respectively. The mortality rate at 28 days was (59.7%). Non-survivors were significantly older (mean age 64 vs. 44 years, P= 0.0001), had significantly worse disease severity scores on ICU admission, higher incidence of chronic kidney disease (CKD) (43% vs. 16%, P= 0.010), required more continuous renal replacement therapy (CRRT) (54% vs. 13% P= 0.0001), had longer hospital length of stay prior to infection (median 34 vs. 13 days, P= 0.018), and required longer inotropic and vasopressors support (median 19 vs. 8 days, P = 0.0001). In multivariate logistic regression the following factors were significantly associated with mortality; requirement of inotropic support [OR 10.01 (95% CI 1.55-64.77); P= 0.015], age [OR 1.05 (95% CI 1.0-1.1); P=0.01], APACHE IV score on ICU admission [OR 1.03 (95% CI 1.0- 1.06); P= 0.04], and ICU length of stay [OR 1.03 (95% CI 1.0- 1.06); P= 0.035].ConclusionMDR Gram-negative infection is associated with significant in-hospital mortality among critically ill patients. Old age, high APACHE IV score, higher ICU length of stay, and higher hemodynamic support are associated with higher mortality.Trial registrationretrospectively registered.

Bacterial infections are a frequent cause of hospitalization and a leading cause of death, particularly with the devastatingly increasing antimicrobial resistance panel [3]. The lack of an antimicrobial stewardship program in many health institutions has contributed to the emergence of drug resistance partly through the malpractice of antimicrobial usage [4]. Antimicrobial resistance expanded through the different kinds of bacteria, primarily Gram-negative Enterobacteriaceae family, particularly Pseudomonas and Acinetobacter species. Gram-negative bacteria (GNB) has become more worrisome due to the emergence of strains resistant to several or even all the available antibiotics, with the lack of upcoming promising antibiotics matching the rate of resistance growth [5]. Multi-drug resistant (MDR) organisms are mainly prevalent in the healthcare settings, including outpatient facilities and general wards, but the highest prevalence, cost, morbidity, and mortality is in critical care units. It has contributed signi cantly to nosocomial infections leading to high nancial and health care burden [3,5].
The emergence of Carbapenem-resistant organisms is one of the evolving alerts as the use of carbapenems have been considered the last resort to treat such organisms for the last two decades [6].
Previous research has shown that infection with such organisms in the critical care settings is associated with high mortality, longer ICU and hospital length of stay and worse other morbidity indicators as the duration of mechanical ventilation and organ failure [5][6][7].
Several studies have evaluated risk factors for mortality in patients with MDR Gram-negative infections. A meta-analysis of patients with MDR Gram-negative infections found that mechanical ventilation, ICU length of stay, presence of septic shock, delay in initiating appropriate antimicrobial treatment, high disease severity scores as APACHE II, and advanced age were associated with higher mortality [5].
Speci c data is still de cient and con icting in critically ill patients regarding associated factors with mortality [5,[8][9][. Therefore, this urged studying the risk factors for the development of MDR GNB, identify the clinical outcomes and factors associated with mortality, especially among critically ill patients who are expected to have the worst outcomes [2,[8][9][10].

Study Population
This is a retrospective observational study of critically ill patients who were admitted to surgical/ or medical intensive care unit in a tertiary care academic hospital between May 2015-Nov 2017 who had an infection with Carbapenem-resistant Enterobacteriaceae (CRE), or Multidrug-resistant (MDR) Pseudomonas aeruginosa, or MDR Acinetobacter Baumani spp. Only patients determined to have active infection were included in the study, and all culture sources were incorporated to determine the presence of the bacteria of interest. Baseline demographics, co-morbidities, and clinical outcomes were collected and were evaluated for association with 28-days mortality. Approval of the research was obtained from the King Saud University research center Institutional Review Board (IRB) before the commencement of the study.

Variables collected
Demographic variables (including age, gender), co-morbidities, hospital admission during last 90 days, microbiologic data (including organism type, antibiotics resistance panel and source of infection), use of inotropes and vasopressors, use of steroids, mechanical ventilation and its duration, need for Continuous Renal Replacement Therapy (CRRT), disease severity scores as Acute Physiology and Chronic Health Evaluation (APACHE II and IV), hospital and ICU length of stay, and ICU and hospital mortality.

De nitions
Index culture: The rst culture during ICU admission that grew CRE, MDR Pseudomonas aeruginosa, or MDR Acinetobacter Baumani spp and associated with active infection.
Index organism: The organism that was grown on the index culture. These de nitions are adapted from the joint initiative by the CDC and the European Center for Disease Prevention and Control (ECDC) [3].
Empiric antibiotic therapy was de ned as the antibiotic therapy that the patient was receiving by the end of post-index culture day 1. Appropriate empirical antibiotics choice was de ned as empiric therapy that included an agent to which the index organism was susceptible. De nitive therapy was de ned as antibiotics that the patient received after the index culture results were available.
An expert review panel consisting of one clinical pharmacy specialists in critical care, one infectious disease physician, and one critical care physician, determined the presence of active infection versus colonization by reviewing the electronic medical record [2] and CDC de nitions in that regard. The expert review panel assessed the adequacy of empiric therapy according to documentation in the [2].

Statistical Analysis
Descriptive analysis was used for all data. Normally distributed data were analyzed using Mean ± Standard deviation (SD) and t-student test for differences. Whereas, in non-normally distributed data, the median was used for descriptive data and Mann-Whitney for any signi cant difference. Chi² was conducted for the association when required. The signi cance level was set at 0.05 with 2SD and 95% con dence intervals [7].

Patient Characteristics and Culture Results
Two hundred fty-ve cases were screened for inclusion criteria; 77 cases t the criteria corresponding to an incidence of 2% of all critical care areas admissions during the study period ( Figure 1). Study participants were predominantly male (60%) with a mean age of 55 years. Clinical characteristics and comorbidities of included patients are summarized in Table 1.
The median hospital length of stay was 115 days, and median ICU length of stay was 66 days, 93% of the patient's required mechanical ventilation, pneumonia was the most common infection (51%), followed by bloodstream infections (26%) as shown in Table 2.
Survivors and non-survivors had no signi cant difference regarding the hospital, ICU length of stay, duration of mechanical ventilation, and the use of steroids at any time during admission, as summarized in Table 4.
In multivariate logistic regression for independent variables associated with 28 days mortality, the following factors contributed signi cantly to mortality; use of inotropes [

Discussion
This study includes one of the largest cohorts of surgical and medical critically ill patients infected with MDR GNB in comparison to previous studies [1,[10][11], and represents mostly a sick group of critically ill patients based on the disease severity scores and predicted mortality as shown in Table 1.
The studied MDR organisms' colonization prevalence rate in our study was 2.9%, and the infection prevalence rate was 2% among the studied ICU population, while the incidence of infection among colonized patients was 70%, highlighting the high virulence of such organism in critically ill population, previous studies showed the variable range of colonization and infection rates among colonized population highlighting the effect of population variabilities, intensivist antimicrobial usage practice and antibiotic stewardship application affecting the prevalence of MDR organisms across different critical care units [12][13].
In this study, the 28-days mortality was 60% and occurred mostly within the ICU admission; previous studies have reported variable mortality rates with MDR GNB in critical care settings ranging between 15-40% [5,[14][15][16][17]. Hospital-acquired pneumonia was the most common infection (51%), even though 93% of our population required mechanical ventilation with long duration (mean= 54 days), this high rate of mechanical ventilation requirement points to the high morbidity and organ failure associated with those organisms like high requirement for renal replacement therapy and high incidence of cardiocirculatory failure evidenced by the high vasopressors/inotropic requirements ( Table 2). Our ndings regarding the high mortality and morbidity associated with those organisms highlight the importance of studying such organisms and their associated infection in critically ill populations more extensively due to the cost impact they exert on the healthcare system.
Non-survivors had a higher incidence of baseline CKD and were signi cantly older and sicker (high morbidity scores) on admission to ICU. The last two factors have been a consistent nding in many previous studies [2,[18][19], which can be utilized in the future as a predictor of outcome for patients infected with such organisms on admission to ICU and to decide level care according to them or plan further additional advanced supportive care accordingly, non-survivors had a signi cantly longer hospital stay before ICU admission which is a variable that was not investigated thoroughly in previous studies that might play a part in dictating the outcome of those patients, this can be explained by either longer duration of colonization with those organisms giving a chance to invade the body barriers and cause infection or those patients were sicker and had depleted physiological reserves, therefore, had a more extended hospital stay before acquiring infection and deterioration till ICU admission.
Both survivors and non-survivors did not have signi cant difference regarding de nite antibiotic therapy duration. Additionally the duration of empirical antibiotic therapy prior to de nite therapy (sensitivity based therapy) was signi cantly shorter in non-survivors, which signi es that those patients even had targeted therapy earlier than survivors, that variable was rarely studied in literature even though it's an essential factor potentially affecting outcome in relation to pharmacological source control, therefore being a non-contributing factor in mortality in our study hints to potentially the virulence of such organisms, the need for a high index of suspicion of infection in colonized patient and starting empirical therapy as early as possible before systemic microbial invasiveness and orid disease manifestation appear. Both groups did not differ regarding the appropriateness of empirical antibiotic choice or the number of empirical antibiotics used for the nal sensitivity of the culprit organism compared to previous studies [20]. However, an important observation in our study is that rate of the appropriate use of empiric antibiotic therapy was high in both groups reaching at least 75%, which is one of the major survival driving factors. However, mortality was still high, corresponding to other factors contributed to survival like age and high morbidity score at baseline on ICU admission [4,15,21].
Despite the fact that a large percentage of our cohort (43%) received systemic steroids during admission, there was no signi cant difference between both groups in terms of mortality even compared to previous studies indicating that it was not a contributing factor for mortality.
After controlling for other factors associated with mortality, high APACHE IV score on admission to ICU, need for inotropic support, longer ICU length of stay, and age were the only signi cant factors contributing to mortality.
Source control is an important factor in clinical success in treating bacterial infections, particularly in intra-abdominal infections commonly encountered in the SICU. Our population is a mix of surgical and medical patients with a small number of surgical patients, and we had limited access to surgical source control data; therefore we could not comment on its contribution to mortality as other studies that showed this factor to decrease mortality [13].

Study limitations:
This study has several limitations. It is observational; thus, a causal relationship between variables and mortality cannot be established. It may be underpowered to detect other important differences between survivors and non-survivors as we explored many variables, and even though our sample size was not small, but it was not large enough to study many variables regarding association with mortality. Our study was a single-center, which limits its generalization. We could not give a conclusion regarding surgical source control about the outcome as our surgical population is limited. Furthermore, exploring these ndings post the COVID-19 pandemic is warranted.

Conclusions
This study demonstrates that MDR Gram-negative infection is associated with signi cant in-hospital mortality among critically ill patients. Old age, high APACHE IV score, and higher hemodynamic support long ICU length of stay are associated with higher mortality. Further work in larger sample size is needed to distinguish between simple associations versus cause-effect relationships, but this study provides support for using these factors in further clinical research projects.   Flowchart of patients' selection.