Mortality and Risk Factors of Death in Patients with AmpC β-Lactamase Producing Enterobacterales Bloodstream Infection: A Cohort Study

Introduction

Bloodstream infections (BSI) are a growing public health concern, with Enterobacterales among the most common agents of community-acquired and healthcare-associated BSI.^1–3 Infections caused by multidrug-resistant (MDR) microorganisms are a growing threat to human health.⁴ Enterobacterales are opportunistic pathogens that are usually part of the human intestinal microbiota.⁵ They are often resistant to β-lactam antibiotics because of the production of AmpC β-lactamases encoded by chromosomal genes (Citrobacter spp., Enterobacter spp., Serratia spp., Providencia spp., and Morganella spp.–ESCPM group bacteria) and are frequent causes of BSI (BSI-ESCPM).^4,6

The production of AmpC β-lactamases by the ESCPM group bacteria is as follows: (i) inducibly expressed following exposure to β-lactam antibiotics, and (ii) constitutively expressed after spontaneous mutations with consequent hyperproduction of AmpC β-lactamases. The production of inducible AmpC causes resistance to penicillins, β-lactamase inhibitor-β-lactam combinations, oxyimino-beta-cephalosporins (1st-, 2nd-, and 3rd-generation cephalosporins), cephamycin, and aztreonam.^7–9 AmpC β-lactamases show little hydrolytic activity against cefepime, cefpirome, and carbapenems. Therefore, susceptibility to these antibiotics is usually unaffected. However, the hyperproduction of AmpC in mutants triggers additional resistance to cefepime and cefpirome. Both mechanisms of AmpC production can occur during the treatment of BSI-ESCPM. Consequently, strains that are initially susceptible to these β-lactam antibiotics may develop resistance. Enterobacter cloacae, Klebsiella aerogenes (formerly Enterobacter aerogenes)¹⁰ and Citrobacter freundii are considered as moderate to high risk of clinically significant AmpC production through induction and mutations.^11,12 Consequently, severe infections caused by β-lactams resistant ESCPM group bacteria are difficult to treat and usually associated with high mortality.^13,14 There is no conclusive evidence about the optimal treatment of BSI-ESCPM.^15,16 In this cohort, mortality, and possible causes of death in patients with BSI ESCPM group bacteria are studied.

Material and Methods

Study Design

A prospective cohort study of patients aged ≥ 18 years with BSI-ESCPM, hospitalized at a University Hospital (UH) in Niterói, Brazil, was conducted from January 1, 2013, to December 31, 2018. All BSI-ESCPM episodes detected in each patient were included in the analysis. Death within 30 days of BSI diagnosis, defined as the date of the 1st positive blood culture, was the primary outcome. The epidemiological, clinical, and microbiological characteristics of the study population were analyzed as potential determinants of death. This study was approved by the Board of Ethics in Research of the Federal Fluminense University (BER-FFU) (number 32570). The BER-FFU waived the requirement for informed consent for this study because it was observational, data were analyzed anonymously, and kept confidential. This study was conducted in accordance with the principles of the Declaration of Helsinki.

Study Setting

This study was conducted in a 287-bed public teaching hospital comprising 13 inpatient units (surgical and clinical wards, intensive care unit, and coronary unit), an emergency room, and an outpatient hemodialysis clinic.

The Detection of the BSI-ESCPM Episodes and Data Collection

Cases were identified through daily laboratory-based surveillance of positive blood culture results issued by the UH Microbiological Laboratory. Once a positive blood culture was detected, BSI was confirmed by patient examination or medical record review. Patients were excluded from the study: 1) if they were dead at the time a positive blood culture result was obtained; 2) and their medical record was not available for review. Each included patient was followed up for 30 days after the BSI-ESCPM diagnosis. The following data were collected by reviewing the medical records: i) Demographics - sex, date of birth, patient origin (community, ambulatory care clinics, UH hemodialysis unit, and transfers from other hospitals), ward of admission, clinical or surgical admission, and date of admission. ii) Clinical - type of admission (emergency and elective), cause of admission, main underlying illness, Charlson comorbidity index,¹⁷ presence of an invasive device up to 48 h before BSI diagnosis, use of antibiotics and surgical procedure within 30 days before BSI diagnosis, clinical outcome within 72 h (sepsis or septic shock), and 30 days (remaining hospitalized, discharge, or death) after BSI diagnosis and date. iii) Characteristics of the BSI episode included: date of positive blood culture, the microorganism(s) detected and its antibiotic susceptibility, type of BSI acquisition (hospital-acquired, healthcare-associated infection, and community-acquired); source of BSI episode that was classified as i) primary associated with vascular access, mucosal barrier damage, infective endocarditis, and undefined source; and ii) secondary to extravascular site of infection, urinary tract, respiratory tract, gastrointestinal and intra-abdominal tract, skin and soft tissues, and the central nervous system. Pitt bacteremia severity score from the day of BSI onset to 48 h before (upper value was used in the analysis and a value ≥4 was used to define severe BSI),^18–20 C-reactive protein (CRP; mg/dl) serum level from 1 d before to 1 d after BSI onset (the highest value was used in the analysis), and presence of neutropenia (absolute number of neutrophils <0.5 × 10⁹/L)²¹ from 1 d before to 3 days after BSI onset; and iv) Treatment of the BSI episode included antimicrobial(s) initiated with the date of beginning and ending, if it was empirically initiated or guided by microbiological results), and other approaches such as removal of invasive devices and drainage of abscess.

Definitions

The community-acquired BSI was defined as either an infection detected or incubated at the time of patient admission to the hospital that was not associated with health care. The healthcare-associated BSI were detected in outpatients receiving healthcare. The hospital-acquired BSI was defined as an infection that occurred on or after the 3rd day of hospitalization.²² The definitions of primary BSI/laboratory-confirmed BSI, central line-associated BSI, BSI associated with mucosal barrier injury and secondary BSI were in accordance with The Centers for Disease Control and Prevention (CDC) definitions.²³

The multidrug resistant ESCPM pathogens were considered according to the following criteria: non-susceptibility to at least one agent in three or more antimicrobial categories, except for those with intrinsic resistance.²⁴ Resistance to 1st, 2nd, and 3rd-generation cephalosporin, amoxicillin-clavulanate, and piperacillin-tazobactam was considered intrinsic owing to AmpC-chromosomal production by the ESCPM group.⁹

Appropriate antimicrobial therapy was defined as: (i) at least one antibiotic used in the therapy must have action against the microorganisms(s) isolated from blood, (ii) the antimicrobial must be initiated within 24 h after blood culture is obtained, and (iii) the antimicrobial must be used for at least 72 h.¹⁴ Monotherapy is defined as the use of antibiotics during the treatment period. Otherwise, polytherapy was defined as combined therapy when two or more antibiotics were started within a maximum interval of 24 h and maintained concurrently for 72 h or more, and sequential therapy for the use of two or more antibiotics during the treatment period.

The Pitt bacteremia score is a previously validated scoring system to evaluate the severity of acute illness based on mental status, vital signs, requirement for mechanical ventilation, and recent cardiac arrest, at the day of the positive blood culture and 48 h before. The highest point score during that time is recorded. It ranges from 0 to 14 points [mental status (disoriented: 1; stupor: 2; coma: 4 points); fever (> 37.6°C and < 40°C: 1; > 40°C: 2 points); hypotension (drop in systolic > 20 mm Hg or diastolic > 10 mm Hg or on intravenous pressor agents: 2 points); mechanical ventilation: 2 points; and cardiac arrest: 4 points]. Patients accumulating 4 or more points (Score ≥ 4) is defined as critical illness and have increased risk of death.^18–20

Microbiological Analysis

BSI episodes were investigated by obtaining at least one blood sample (20 mL) from two different venipuncture sites in each patient using an aseptic technique, as recommended by the Infection Control Division of UH. Each blood sample was inoculated into one aerobic or anaerobic blood culture bottle (10 mL each). Microorganism growth was detected using BacT/ALERT^® (BioMérieux). Both microbiological identification and antibiotic susceptibility tests were performed using the Vitek^® 2 automated system (BioMérieux). The antimicrobial susceptibility test results were interpreted according to The Clinical and Laboratory Standards Institute, applying the updated versions for the year in which the blood culture was performed.^25,26 Resistance to cefoxitin is considered a phenotypic marker of the presence of chromosomal AmpC.⁹

Statistical Analysis

Descriptive analyses of categorical and continuous variables were presented as proportions and median values, respectively. The χ2 or Fisher’s exact test was used to compare categorical variables, and Student’s t-test or Mann–Whitney test was used to compare continuous variables, as appropriate.

Death within 30 days of BSI-ESCPM diagnosis was the primary outcome of interest. Mortality was calculated by dividing the number of deaths within 30 days after BSI-ESCPM diagnosis by the total number of BSI-ESCPM cases in the same period. The demographics, clinical and microbiological characteristics of the patients, and BSI episodes were analyzed as potential risk factors for death within 30 days. Inadequate antimicrobial therapy was hypothesized to be the main factor independently associated with death within 30 days. Variables were included in the multivariable logistic regression models if they met the criteria for statistical significance (P < 0.05) in the univariate analyses. Additional clinically relevant variables, based on biological plausibility and prior knowledge, were included as a priori variables in the adjusted models. A backward selection approach was applied, using the 10% rule, to identify potential confounders in the final adjusted model. Analyses were performed using the Epi info 7.2.5.0 software (https://epi-info.software.informer.com/7.2/).

Results

Study Population and BSI-ESCPM Episodes

From January 2013 to December 2018, 1394 BSI episodes were detected in the UH. Of these, 11.1% (n: 155) were caused by Enterobacterales from the ESCPM group. These 155 BSI-ESCPM episodes occurred in 138 patients during 146 hospitalizations. Among 138 patients, 11% had two or more BSI-ESCPM episodes. Of 155 episodes of BSI-ESCPM, 18% were polymicrobial with two or more microorganisms identified, adding up 184 microorganisms. All the detected BSI-ESCPM episodes were included in the analysis, as shown in Figure 1.

Figure 1 Population and episodes of bloodstream infection due to the ESCPM bacteria group detected at a University Hospital from January 1st, 2013, to December 31st, 2018. Abbreviations: BSI, bloodstream infection; ESCPM, Enterobacter spp. Serratia spp. Citrobacter spp. Providencia spp. Morganella spp.

Clinical and Demographic Characteristics of the Study Population

Among the 138 patients, 63.0% (n: 87) were male. The median age of the patients was 60 years (range: 18–85) years. Charlson comorbidity index > 5 was observed in 21.0% (n: 29) of patients, and the most frequently detected underlying disease was kidney disease in 27.5% (n: 38) of patients. These patients had 146 admissions; 80.1% (n: 117) due to a medical emergency. They occurred in the emergency room mostly, 65.8% (n: 96). Infectious diseases (30.1%; n: 44), mainly skin, soft tissue, and osteoarticular infections (10.3%; n: 15), followed by cardiovascular diseases (14.4%; n: 21), were the main reasons for admission.

Clinical Characteristics of the BSI-ESCPM Episodes

The 155 BSI-ESCPM episodes were hospital-acquired (HAI) and occurred mainly in the intensive care unit. Over a third of the episodes involved invasive procedures and antibiotic use, 48 h and within 30 days before BSI diagnosis, respectively. BSI-ESCPM episodes were associated with short-term central venous catheter (37.4%) mainly, followed by those secondary to respiratory infections (19.4%), as details Table 1.

Table 1 Univariate Analysis of Epidemiological and Clinical Characteristics as Risk Factors for 30-Day Death in Bloodstream Infection Due to ESCPM Bacteria Group (Enterobacter spp., Serratia spp., Citrobacter spp., Providencia spp., Morganella spp.)

Microbiological Profile of the BSI-ESCPM Episodes

Enterobacter spp. and Serratia spp. were the most frequent agents detected, 49% and 29%, respectively. Resistance to cefepime, or cefepime and meropenem was observed in 18.7% and 8.4% of the isolates, respectively. Most MDR Enterobacter spp. and Serratia spp. were detected (Table 2).

Table 2 Univariate Analysis of Microbiological Characteristics as Risk Factors of 30-Day Death in Bloodstream Infection Due to ESCPM Bacteria Group (Enterobacter spp., Serratia spp., Citrobacter spp., Providencia spp., Morganella spp.)

Treatment and Outcome

The initial antibiotic therapy was classified as inadequate in 16.1% of the cases. Therapy was empirically initiated in 78.1% of the episodes. Mostly, combined therapy included meropenem, comprising 14.2% of the BSI-ESCPM episodes. Septic shock within 72 h after BSI-ESCPM diagnosis and Pitt score ≥ 4 were observed in 18.1% and 32.9% of the BSI episodes, respectively, as shown Table 3.

Table 3 Univariate Analysis of Severity of Infection and Type of Antibiotic Treatment as Risk Factor for 30-Day Death in Bloodstream Infection Due to ESCPM Bacteria Group (Enterobacter spp., Serratia spp., Citrobacter spp., Providencia spp., Morganella spp.)

Mortality and Risk Factors of Death

The overall mortality up to 30 days after BSI-ESCPM diagnosis was 24.5% (n: 38) and ranged from 11.8% to 28.0% during the study period (2013–2018). The median time to death was 7.5 (range: 0–29) days. The temporal distribution of the mortality is shown in Figure 2.

Figure 2 Frequency of deaths within 30 days of the 155 episodes of bloodstream infection caused by ESCPM bacterial group (Enterobacter spp., Serratia spp., Citrobacter spp., Providencia spp., Morganella spp.).

The following variables were included in the initial multivariable logistic regression model: i) inadequate empirical antibiotic therapy (primary analysis); ii) BSI secondary to respiratory infection, BSI secondary to urinary infection, BSI caused by Enterobacter spp., and septic shock within 72 h after BSI-ESCPM (P≤0.05) in the univariate analyses); iii) polymicrobial BSI-ESCPM; combined antibiotic therapy, Charlson comorbidity index, and ESCPM group bacteria considered at moderate to high risk of clinically significant AmpC production (including prior knowledge and biological plausibility). Tables 1–3 present details of the analysis.

The final model after backward selection included septic shock within 72 h after BSI-ESCPM diagnosis (OR: 70.26; 95% CI: 16.69–295.77), treatment with combined antibiotic therapy (OR: 0.23; 95% CI: 0.05–0.94), BSI due to Enterobacter spp. (OR: 0.16; 95% CI: 0.05–0.56), and BSI secondary to urinary infection (OR: 0.11; 95% CI: 0.01–0.99) remained independently associated with 30-day death. There was a tendency for association between inadequate empirical antibiotic therapy and 30-day death (OR: 2.19; 95% CI: 0.51–9.42), as shown in Table 4. The potential interactions between risk factor and protective factors were tested, however no statistically significant associations were detected.

Table 4 Multivariable Analysis of Risk Factors for 30-Day Death in 155 Episodes of Bloodstream Infection Due to ESCPM Bacteria Group (Enterobacter spp., Serratia spp., Citrobacter spp., Providencia spp., Morganella spp.)

Discussion

In this 6-year cohort of patients with BSI-ESCPM, 30-day mortality was relatively high (24.5%). Shock within 72 h after BSI diagnosis was independently associated with death while combined antibiotic therapy, BSI secondary to urinary infection, and Enterobacter spp. BSI were protective factors for death within 30 days of BSI diagnosis. There was a positive association between inadequate empirical antibiotic therapy and 30-day death, however it did not reach statistical significance.

BSI-ESCPM episodes were detected mostly in males and older patients, with a median age of 60 years and a Charlson comorbidity index of three. These findings suggest males, the elderly, and patients with comorbidities are more susceptible to BSI-ESCPM, as described in other studies.^27–29 The following characteristics of these patients can partially explain their higher susceptibility to BSI: (i) lower healthcare, hormonal, and microbiological characteristics of males compared to females could increase the susceptibility to acquiring infection and (ii) the immunosenescence of aging with the greater susceptibility of the elderly to acquire infection.^2,28–30 This may have been partly attributed to UH’s characteristics since (i) most patients treated there are males; (ii) UH predominantly cares for patients with chronic diseases; and (iii) there is no emergency room for the general population.

Patients with BSI-ESCPM episodes were admitted to emergency room mainly. This finding can be explained by the greater risk of infection acquisition in patients with medical emergencies^31,32 suggesting that such patients must be closely surveyed for BSI, including BSI-ESCPM. Infectious diseases are the main cause of hospitalization, as reported by others.²⁸ The hospital preference for caring for patients with chronic diseases who are more susceptible to infections might explain these results.

The proportion of administered antibiotics and invasive procedures was high before the BSI-ESCPM acquisition. These characteristics are associated with an increased risk of HAI.^28,33,34 Not surprisingly, most BSI-ESCPM episodes are HAI, mainly in the ICU, where patients are usually more exposed to antibiotic use and invasive procedures. The presence of a short-term central venous catheter was the main cause of BSI-ESCPM, as described in previous studies^27,29 indicating that a large proportion of BSI-ESCPM episodes could be avoided by adhering to strict protocols to prevent vascular catheter-related infections.

Generally, empirical antibiotic therapy is initiated. A high proportion of patients received inadequate antibiotic therapy. Patients who received inadequate therapy had higher mortality than patients treated with combined therapy, including at least one adequate antimicrobial, as described by others for Gram-negative BSI in general^29,35–37 suggesting that combined antibiotic therapy can increase the proportion of adequate therapy and reduce mortality of BSI-ESCPM.

Enterobacter spp. and Serratia spp. were the most common agents of BSI-ESCPM, as seen in other studies.^27–29,33 A high proportion of BSI caused by bacteria resistant to cefepime alone as well as to cefepime and carbapenem was detected, involving Enterobacter spp. and Serratia spp. mainly. Enterobacter spp. and Citrobacter spp. are considered to have moderate to high risk of broad-spectrum cephalosporins resistance due to AmpC production.^11,38 However, we found a higher frequency of MDR Serratia spp. than MDR Citrobacter spp. which could be explained by the low number of BSI due to Citrobacter spp. detected. This also suggests Serratia spp. has an elevated potential to produce clinically relevant AmpC. Surprisingly, an elevated proportion of carbapenem resistance has been detected in these species. Thus, we wondered whether these species would have an increased risk of acquiring additional mechanisms of antibiotic resistance such as carbapenemase production.

The 30-day mortality rate was high, reinforcing that BSI-ESCPM is a severe infection with an elevated burden.^16,27,33,36 Septic shock within 72 h after BSI-ESCPM diagnosis was identified as a significant risk factor for death, as described by others for BSI in general.^{34,35,39–41} These findings emphasize the severity and potential for fatal health outcomes in BSI-ESCPM. Thus, better strategies for managing septic shock are critical for reducing the risk of death in patients with BSI-ESCPM.

Combined antibiotic therapy was a protective factor against death in patients with BSI-ESCPM. Similarly, other studies have described that combined therapy reduces the mortality of Gram-negative bacteria BSI.^42,43 To the best of our knowledge, this finding is unprecedented for BSI-ESCPM. In fact, combined antimicrobial therapy could reduce the selection of mutant ESCPM group bacteria with AmpC hyperproduction and could result in a patient cure.¹² In this study, the combination therapy mainly included meropenem, followed by combined therapies with other drugs that are active against Gram-negative bacteria. However, combination therapy with meropenem did not have a protective effect against death, suggesting that additional combinations of antibiotics had an impact on these results.

A trend towards the association between inadequate empiric antimicrobial therapy and death was detected (OR: 2.19; 95% CI: 0.51–9.42), reinforcing inadequate initial antimicrobial therapy is associated with negative health outcomes in patients with Gram-negative bacteria BSI, as described by others.³⁵ In the present study, a larger sample size would have provided a greater statistical power to test this association.

BSI-ESCPM secondary to urinary infection was also protective against death within 30 days of diagnosis. Other studies have demonstrated that BSI originating from urinary tract infections is associated with lower mortality rates than BSI originating from other sites of infection. Therefore, these findings corroborate the results of previous studies.^2,44–49

Surprisingly, Enterobacter spp. BSI was a protective factor against 30-day death. This finding differs from those of other studies, in which Enterobacter spp. BSI was associated with an increased risk of death^34,49 and most BSI-ESCPM caused by Enterobacter spp. occurred in patients with less severe clinical conditions, which can explain the lower risk of death observed compared to other agents in the ESCPM group. In fact, the patients with Enterobacter spp. BSI were preferentially admitted outside the intensive care unit and had BSI secondary to sources of infection other than the respiratory tract, which is associated with high mortality.

This study has some limitations (i) it was a single-center study; therefore, external validation to increase the generalizability of the findings is limited; (ii) only 155 episodes of BSI were detected; this small sample size may have reduced the power of detecting variables associated with death (for instance, the impact of therapy with piperacillin/tazobactam and antibiotic combination therapy without meropenem were not analyzed). Nevertheless, by including all BSI-ESCPM episodes detected during the study, this strengthened our findings; and (iii) BSIs due to different bacterial species were analyzed together. However, their principal mechanism of antimicrobial resistance is similar, and species considered to be at a higher risk of clinically relevant AmpC production were analyzed separately. Despite these limitations, this study presents new information regarding the risk factors for death in patients with BSI-ESCPM.

Conclusion

Finally, BSI-ESCPM is a severe infection with poor outcomes such as sepsis associated with death. Combined antibiotic therapy was found to be a protective factor against death and could reduce the burden of BSI-ESCPM in settings with elevated incidence. A tendency of association between inadequate empirical antibiotic therapy and 30-day death was detected in these patients, and BSI-ESCPM episodes caused by species considered to be at moderate to high risk of clinically significant AmpC production were not associated with a higher risk of death.