Efficacy of Carbapenems Compared With Noncarbapenem Broad-Sp… : Critical Care Medicine


Question: Do carbapenems as initial antibiotic therapy improve mortality in patients with sepsis compared with the other broad-spectrum beta-lactam antibiotics?

Findings: In the current retrospective observational study adjusted by propensity score methods, carbapenems as an initial therapy for sepsis were not associated with significantly lower mortality. However, significant mortality reductions were observed in specific populations with septic shock or higher disease severity.

Meanings: To reduce indiscriminate use of carbapenems, initiating carbapenem therapy without appropriate selection of target patients should be avoided.

Sepsis is defined as life-threatening organ dysfunction caused by dysregulated host response to infection (1). Despite the progress made in medical management over the past few decades, sepsis remains an important global health problem causing millions of deaths around the world (2,3). One essential facet of sepsis management is the administration of appropriate antimicrobials. Multiple lines of evidence have shown that failure or delay in the administration of appropriate antimicrobials is associated with substantial increases in the mortality of sepsis (4–9). On the basis of the evidence, the Surviving Sepsis Campaign guidelines recommended empiric broad-spectrum therapy to cover all likely pathogens as soon as possible after sepsis is recognized (10,11).

Broad-spectrum beta-lactam antibiotics are commonly used as empiric therapy for sepsis (12). Among them, carbapenems offer an exceptionally broad spectrum of antimicrobial activity. The existing evidence suggests that carbapenems are associated with relatively low mortality in several infectious diseases, especially those caused by extended-spectrum beta-lactamase (ESBL)-producing pathogens (13,14). For these reasons, carbapenems are the preferably used antibiotics for the initial treatment of sepsis, and consumption of carbapenem has been reported to be increasing in different parts of the world (15–17). However, indiscriminate use of carbapenems against sepsis should be avoided because it has contributed to the emergence of multidrug-resistant pathogens (18,19). Therefore, updating knowledge on the efficacy of alternative empiric regimens for sepsis is required.

The fourth-generation cephalosporins, such as cefepime and cefpirome, and piperacillin/tazobactam are nearly equal to carbapenems in antimicrobial activity and thus are considered alternative therapies against sepsis. Previous studies have evaluated the efficacy of carbapenems compared with these antibiotics against several specific pathogens, such as Acinetobacter species and ESBL-producing Escherichia coli (20–23). However, there is little evidence of the efficacy of carbapenems against unidentified mixed pathogens. Causative pathogens are not usually identified at the time of initiating antibiotics in sepsis because of the strong recommendation to initiate immediate treatment, ideally within 1 hour after sepsis is first suspected (11). Thus, there is a significant need to comprehensively evaluate differences in mortality and morbidities between carbapenems versus noncarbapenem broad-spectrum antibiotics as initial treatment of sepsis, including various underlying infectious diseases and causative pathogens. This study aimed to evaluate mortality and morbidities in patients with community-acquired sepsis treated with carbapenems as initial antibiotic therapy compared with those treated with noncarbapenem antibiotics.


Design and Setting

This retrospective observational study used a part of large-scale database covering approximately 23% of acute-care hospitals in Japan that contained about 30 million patients as of October 2019 (provided by Medical Data Vision (MDV) Co., Tokyo, Japan; https://en.mdv.co.jp/about-mdv-database/) (24). The database consists of Japanese administrative claims data and includes data on age, sex, primary diagnoses, concomitant diagnoses, complication diagnoses, procedures, drug administrations, discharge status, and laboratory tests. The diagnoses are recorded in this database using the International Classification of Diseases, Tenth Revision (ICD-10) codes. From the original MDV database, we extracted the clinical data of 534,739 patients who required unplanned admissions to 42 hospitals that recorded the laboratory data from February 2006 to December 2019.

This study followed the principles of the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology guideline. The study protocol was approved by the Institutional Review Board for Clinical Research of Osaka General Medical Center (IRB No. S201916015, Study title: Evaluation of clinical issues in emergency and intensive care medicine using medical big data, approved on July 7, 2020). The board waived the requirement for informed consent because of the anonymous nature of the data.


Patients were eligible for this study if they met all of the following criteria: 1) required unplanned hospital admission, 2) were not transferred from other hospitals, 3) were older than 18 years at the time of admission, 4) were diagnosed as having sepsis, and 5) received IV broad-spectrum beta-lactam antibiotics as initial antibiotic therapy within 24 hours from hospital admission.

Sepsis was defined as having a proven/suspected infection and the development of organ dysfunction. In this study, the original ICD-9 codes for infection and organ dysfunction by Angus et al (25,26) were adapted to corresponding ICD-10 codes by previously published methods. The sensitivity and specificity of ICD-10 codes to detect sepsis were reported to be 35% and 98%, respectively (27). An infection was defined as having any of the infection-related ICD-10 codes in the primary diagnosis, and organ dysfunction was defined as having any of the organ-dysfunction-related ICD-10 codes in the primary, concomitant, or complication diagnoses (Supplementary Table S1, https://links.lww.com/CCM/H346). Broad-spectrum beta-lactam antibiotics in this study included carbapenems, fourth-generation cephalosporins, and tazobactam/piperacillin.

Exclusion criteria included patients who died within 24 hours after admission and concomitantly received carbapenems and other broad-spectrum beta-lactams. We also excluded data from the second and subsequent hospitalizations of patients who were hospitalized multiple times during the study period and used only the data from the first hospitalization.

We divided the study patients into two following groups: patients who received carbapenems as initial treatment (carbapenem group) and patients who received the noncarbapenem broad-spectrum agents (i.e., fourth-generation cephalosporins or tazobactam/piperacillin) as initial treatment (noncarbapenem group). To evaluate the effect of antibiotics as “initial treatment of sepsis,” patients whose antibiotics were switched from noncarbapenem beta-lactams to carbapenems were categorized into the noncarbapenem group, and patients whose antibiotics were switched from carbapenems to noncarbapenem beta-lactams were categorized into the carbapenem group. As in previous studies, we defined the “escalation” and “de-escalation” of antibiotics based on ranking according to the activity spectrum against Gram-negative bacteria (28,29). Namely, switching from noncarbapenems to carbapenems was regarded as escalation, and switching from carbapenems to noncarbapenems was regarded as de-escalation.

Data Collection

We collected the following data: age, sex, height, weight, Charlson Comorbidity Index (CCI) (30), site of infection, use of catecholamines, use of mechanical ventilation, and laboratory tests. We used predefined ICD-10 coding algorithms for evaluating CCI (31). The renal, hepatic, and coagulation subscores of the Sequential Organ Failure Assessment (SOFA) were calculated based on creatinine levels, total bilirubin levels, and platelet counts.


The primary outcome was all-cause in-hospital mortality. The following secondary outcomes were: 1) occurrence rate of acute kidney injury (AKI), 2) occurrence rate of enterocolitis due to Clostridium difficile, 3) prevalence of methicillin-resistant Staphylococcus aureus (MRSA) infection or methicillin-resistant coagulase-negative Staphylococci (MRCNS) infection, 4) prevalence of multidrug-resistant pathogens infection, 5) rate of patients who received antibiotics for more than of multidrug-resistant pathogens infection, 5) rate of patients who received antibiotics for more than 7 days, 6) rate of patients who received antibiotics for more than 14 days, 7) rate of patients who received broad-spectrum antibiotics for more than 7 days, and 8) rate of patients who received broad-spectrum antibiotics for more than 14 days.

AKI was diagnosed based on serum creatinine levels from the Kidney Disease Improving Global Outcomes (KDIGO) criteria, when serum creatinine increases by 0.3 mg/dL or more in 48 hours or rises to at least 1.5 times from baseline within 7 days, in patients without chronic kidney diseases requiring chronic maintenance dialysis (32). Enterocolitis caused by C. difficile was diagnosed when patients received ICD-10 code A04.7 and were administered oral vancomycin or metronidazole from 7 days after admission. The sensitivity and specificity of ICD-10 codes for C. difficile infection were reported to be 35.6% and 99.9%, respectively (33). Similarly, MRSA and MRCNS infections were diagnosed when patients received the ICD-10 codes A04.8, A41.0, A41.1, A49.0, A49.0, G00.3, I33.0, J15.2, J15.2, J85.2, J86.9, K65.0, M00.01, M00.02, M00.05, M00.06, M00.09, M86.99, N30.8, T81.4, or Z22.8, and they were administered IV anti-MRSA agents, including arbekacin, daptomycin, linezolid, teicoplanin, and vancomycin from 7 days after admission. In this study, multidrug-resistant pathogens included multidrug-resistant Pseudomonas species and multidrug-resistant Acinetobacter species, which were diagnosed in accordance with ICD-10 codes A49.8, A49.1, and A49.9 and the diagnosis name determined from 7 days after admission. Broad-spectrum antibiotics included carbapenems, fourth-generation cephalosporins, tazobactam/piperacillin, and third- and fourth-generation quinolones.

Statistical Analysis

Due to the retrospective nature of this study, imbalances in the covariates at baseline between the groups were assumed to be present. To account for this imbalance, we used a propensity score method for inverse probability treatment weighting (IPTW). The propensity score was determined using logistic regression with the 23 variables assumed to be associated with carbapenem use as covariates (Supplementary Table S2, https://links.lww.com/CCM/H346). The C-statistic was calculated to evaluate the goodness of fit. Additionally, the distribution of propensity scores in each group was graphically compared to assess the overlap assumption.

Descriptive statistics were calculated as group means with sds for continuous variables and as frequencies with percentages for categorical variables. Standardized mean difference (SMD) was used to examine the balance of patient characteristics between the two groups before and after IPTW adjustment, where greater than 0.1 was considered imbalance.

An IPTW-adjusted logistic regression model was leveraged to assess the association between treatment and mortality. In the sensitivity analysis, we applied two more logistic regression models: a doubly robust augmented IPTW-adjusted logistic model (34–37), and a multivariate logistic model adjusted by the 23 covariates used to calculate the propensity score. Furthermore, we evaluated the differences in survival benefit between 1) carbapenems and tazobactam/piperacillin, 2) carbapenems and fourth-generation cephalosporins, 3) meropenem and tazobactam/piperacillin, and 4) meropenem and fourth-generation cephalosporins separately. We also fitted logistic regression models adjusted by IPTW method to evaluate the survival benefit of carbapenem therapy in several subsets, including; 1) older (≥ 70 yr) versus younger (< 70 yr) patients, 2) serum C-reactive protein (CRP) level of greater than or equal to 15.0 mg/dL versus less than 15.0 mg/dL, 3) CCI of greater than or equal to 5 versus less than or equal to 4, 4) septic shock versus nonshock, 5) general wards versus ICUs, 6) with mechanical ventilation versus without mechanical ventilation, 7) anatomical sites of infection, 8) admission during the first half versus the second half of the study period, 9) without switching of antibiotics (i.e., from carbapenems to noncarbapenems or from noncarbapenems to carbapenems) within 3 days, 10) without switching of antibiotics within 7 days, and 11) without switching of antibiotics within 14 days after initiation of antibiotic therapies in the patients.

In this study, septic shock was diagnosed when patients were administered IV catecholamine on the day of hospital admission. The categories of subgroups were determined before extracting the study data, but the cutoff points were determined after extracting the data to allow well-balanced stratifications. To evaluate the effect modification of these factors on the survival effect associated with carbapenems, we included cross-product terms between these factors and carbapenem therapy as covariates of the regression analyses. In the subsets based on the site of infection, effect modifications were evaluated in relation to patients with respiratory infection as a reference subset. For sensitivity analysis in the subgroup analyses, we also estimated the survival association of carbapenem in the subsets by multivariable logistic regression models adjusted by the covariates. The impact of carbapenem therapy on the secondary outcomes was estimated by logistic regression analyses adjusted by IPTW. Missing values were not imputed in the regression models.

All statistical inferences were two-sided, and a p value of less than 0.05 indicated statistical significance. Because of the underpowered nature of the interaction analysis, we used a significance level of 0.20 with statistical inferences for the effect modifications. All statistical analyses were conducted using Stata Data Analysis and Statistical Software version 17.0 (StataCorp, College Station, TX).


Study Population

Among 534,739 patients in the study registry, 122,843 were adults, were admitted from outside the hospitals, and were diagnosed as having infectious diseases. Among them, 36,785 patients were diagnosed as having sepsis at the time of hospital admission, and we included 7,907 patients who received broad-spectrum beta-lactam antibiotics within 24 hours from admission. After excluding 515 patients who met the exclusion criteria, we included 7,392 patients in the final study cohort. The carbapenem group comprised 3,547 patients, and the noncarbapenem group comprised 3,845 patients (Fig. 1). Supplementary Figure S1 (https://links.lww.com/CCM/H346) shows the distributions of propensity scores in the two groups. The C-statistic for the fit of the propensity scores was 0.677. The frequency of administration of each antibiotic agent is summarized in Supplementary Table S3 (https://links.lww.com/CCM/H346). Meropenem and tazobactam/piperacillin were the most commonly used in each group (76.8% and 82.1%, respectively).

Figure 1.:

Patient flow diagram. ICD-10 = International Classification of Diseases Tenth Revision, ICD-9 = International Classification of Diseases, Nineth Revision.

Numbers and cumulative numbers of antibiotics-switched patients on each day from the initiation of primary antibiotics are shown in Supplementary Table S4 (https://links.lww.com/CCM/H346). Within 14 days after the initiation of antibiotic therapies, 3.7% (130 of 3,547 patients) in the carbapenem group had been switched to noncarbapenem beta-lactams, and 7.1% (272 of 3,845 patients) in the noncarbapenem group had been switched to carbapenems.

Supplementary Table S5 (https://links.lww.com/CCM/H346) shows the baseline characteristics of the two groups before and after IPTW adjustment. Patients in the carbapenem group were slightly younger (mean, 73.8 vs 72.5 yr old, SMD = 1.1414). There were no significant differences in sex ratio (males, 61.6% vs females, 59.6%; SMD = 0.0417). The rates of septic shock, mechanical ventilation use, ICU admission, and renal replacement therapy were significantly higher in the carbapenem group. Similarly, renal, coagulation, and hepatic SOFA subscores were all significantly higher in the carbapenem group. After IPTW adjustment, the SMDs of all parameters between the two groups were less than 0.1, indicating that patient baseline characteristics, use of antifungals, and use of concomitant antibiotics, including anti-MRSA agents and quinolones, were well balanced between the groups after IPTW adjustment. As there were no missing data in the ICD-10 code and receipt data in the study database, data on the preexisting comorbidities, site of infection, and concomitant therapies were not missing. We show the rates of missing data for the demographic and laboratory data in Supplementary Table S6 (https://links.lww.com/CCM/H346).

Effect of Carbapenem Therapy on Survival

The survival odds ratios (ORs) associated with carbapenem therapy adjusted by the three logistic regression models are shown in Figure 2. There were no significant differences between the groups in all three models. Similarly, there were no significant differences in mortality between carbapenems and tazobactam/piperacillin, between carbapenems and fourth-generation cephalosporins, between meropenem and tazobactam/piperacillin, and between meropenem and fourth-generation cephalosporins (Supplementary Fig. S2, https://links.lww.com/CCM/H346).

Figure 2.:

Differences in in-hospital mortality between the carbapenem and noncarbapenem groups as evaluated by three logistic regression models. AIPTW = augmented inverse probability treatment weighting, IPTW = inverse probability treatment weighting, OR = odds ratio.

A summary of survival benefits associated with carbapenems in the subgroups is shown in Figure 3. There were significant multiplicative effect modifications between septic shock and nonshock patients (p < 0.001), between patients admitted in general wards and ICUs (p = 0.014), and between patients with and without mechanical ventilation (p = 0.105). Significant associations between carbapenem therapy and lower mortality were observed in the subsets of septic shock (OR 0.49; 95% CI, 0.34–0.71), ICU admission (OR 0.57; 95% CI, 0.38–0.85), and mechanical ventilation (OR 0.65; 95% CI, 0.43–0.99), whereas risks of death were similar between the groups in the nonshock, general wards, and nonmechanical ventilation subsets. In contrast, there were no significant effect modifications by age, CRP, CCI, and admission date. We show the results of sensitivity analyses evaluating survival benefits associated with carbapenem therapy in the subgroups using logistic regression analyses adjusted by covariates in Supplementary Figure S3 (https://links.lww.com/CCM/H346). Survival benefits associated with carbapenem therapy in the subgroups were similar to those in the IPTW analyses. Similarly, there were no significant differences in survival associated with carbapenem therapy in the three subsets excluding those patients in whom antibiotics were switched.

Figure 3.:

Differences in in-hospital mortality between the carbapenem and noncarbapenem groups in several subsets according to patient characteristics. CRP = C-reactive protein, CCI = Charlson Comorbidity Index, OR = odds ratio.

Association Between Carbapenem Therapy and Secondary Outcomes

The secondary outcomes are summarized in Figure 4. The adjusted risk of MRSA/MRCNS infection was significantly higher in the carbapenem group (adjusted OR 1.45; p = 0.011). The adjusted risks of AKI, enterocolitis due to C. difficile, and MRSA/MRCNS infection were not different between the groups. The rates of patients who received antibiotics for more than 7 days and 14 days were significantly higher in the carbapenem groups (adjusted OR 1.19; p = 0.001 and adjusted OR 1.25; p = 0.001, respectively). Similarly, the rates of patients who received broad-spectrum antibiotics for more than 7 days and 14 days were significantly higher in the carbapenem groups (adjusted OR 1.17; p = 0.002 and adjusted OR 1.24; p = 0.007, respectively).

Figure 4.:

Differences in the risk of secondary endpoints between the carbapenem and noncarbapenem groups. MRSA = methicillin-resistant Staphylococcus aureus, MRCNS = methicillin-resistant coagulase-negative Staphylococci, OR = odds ratio.


Rapid administration of appropriate antimicrobials is essential for reducing mortality in sepsis (6). As causative pathogens are not usually confirmed when initiating antibiotics, updated knowledge is strongly required on the efficacy associated with the initial empiric use of broad-spectrum antibiotics. This nationwide multicenter study compared carbapenems with other broad-spectrum beta-lactams as initial therapy and revealed the following: 1) initial use of carbapenems was not associated with significantly lower mortality overall in patients with sepsis, 2) a survival benefit associated with carbapenem was observed in patients with septic shock, in ICUs, or with mechanical ventilation, and 3) carbapenem therapy was associated with a significantly higher risk of MRSA/MRCNS infection and longer use of antibiotics.

Differences Between Carbapenem and Noncarbapenems

The fourth-generation cephalosporins and piperacillin/tazobactam exhibit broad spectrums of antibacterial activity against most aerobic bacteria (21,38). However, several concerns might limit the clinical effectiveness of these agents compared with carbapenems. First, carbapenems possess an even wider antibacterial spectrum compared with the noncarbapenem broad-spectrum agents. For example, in common with other cephalosporins, fourth-generation cephalosporins have limited inhibitory activity against nonfermentative bacteria and minimal activity against Enterococci and anaerobic bacteria (39).

Second, the efficacy of the noncarbapenem agents is considered to be inferior to carbapenems against several drug-resistant pathogens such as AmpC beta-lactamase-producing and ESBL-producing Enterobacteriaceae. Especially, ESBL-producing Enterobacteriaceae is one of the main targets for antibiotic treatment because of the global increase in its community prevalence (40). Although noncarbapenem beta-lactams are frequently active in vitro against ESBL, several previous studies suggested that these agents were associated with suboptimal clinical outcomes to carbapenems for the treatment of ESBL infections (41–43). However, approximately 12% of E. coli and 6% of Klebsiella pneumoniae were reported to produce ESBL in Japan, which might not be high enough to cause the differences in survival benefit between carbapenem and the other beta lactams (44). Actually, carbapenems were not associated with significantly lower mortality overall in the patients in this study. However, mechanical ventilation and septic shock were reported to be associated with a higher prevalence of ESBL-producing pathogens (45), which might explain the difference in survival benefits associated with carbapenem therapy observed in several of the subsets.

In this context, carbapenem therapy is considered to be a safer choice especially when causative pathogens are not confirmed. Furthermore, our finding of an association between carbapenems and lower mortality in the subset of septic shock is consistent with that in a previous study, which reported that the association between early administration of antibiotics and mortality was stronger among patients with rather than without septic shock (46).

Optimal Target for Carbapenem Therapy

Although carbapenems have potentially higher efficacy as initial therapies in sepsis, their indiscriminate use should be avoided because the extremely higher selective pressure would increase the risk of drug resistance (18,19,47). Actually, carbapenem use has been globally reported to be associated with a significant increase in the risk of carbapenem-resistant Enterobacteriaceae (48). The global spread of drug-resistant pathogens associated with the liberal use of broad-spectrum antibiotics has become a serious public health problem (49). Approximately 4%–14% of Pseudomonas aeruginosa infections in the United States and 2.1%–4.1% in Italy were reported to be caused by multidrug-resistant strains (50). In nationwide multicenter surveillance conducted in Japan, the prevalence of multidrug-resistant P. aeruginosa was reported to be 2.4%, which was comparable to that in other countries (51). Some investigators warn that effective antibiotics may not be available in the near future because the pipeline of new antibiotics is drying up despite the increasing prevalence of resistant bacteria (52).

This study also showed that the use of carbapenems as initial therapy was associated with a significant increase in the rate of the prolonged use of antibiotics. Although patient backgrounds would influence the length of antibiotic therapy, a treatment duration of 7–10 days is generally adequate for sepsis. For example, recent guidelines for pneumonia recommend a 7-day course of antibiotic therapy (53). In the present study, 2190 of 3394 patients (64.5%) in the carbapenem group received antibiotic therapy for more than 7 days, and 725 of 3319 patients (21.8%) received antibiotic therapy for even more than 14 days, which were significantly higher compared with the durations in the noncarbapenem group. Although there is no clear explanation as to why carbapenems were associated with the longer duration of antibiotic therapy, such administration might cause a further increase in the risk of drug-resistant pathogens.

Therefore, the decision to use carbapenems should be carefully considered after selecting the optimal targets that could receive the maximum benefit from such therapy. The present study also showed multiplicative effect modifications of carbapenem therapy on mortality according to the presence of shock and the requirement for mechanical ventilation, suggesting that higher severity might be a key component in selecting those patients expected to receive the maximum survival benefit associated with initial carbapenem therapy. Further prospective controlled studies are needed to confirm or refute these findings.

Strengths of This Study

This study has several strengths. First, there is little evidence so far of the efficacy of carbapenems as an initial choice of antibiotic therapy against an overall sepsis population that includes various sites of infection and pathogens. Therefore, this study provides a novel insight into empiric antibiotic therapy at the time of initiating antibiotics when causative pathogens are not usually identified. Second, the large sample size allowed us to explore the efficacy of carbapenem by taking a large number of confounding factors into account, which enhanced the robustness of the methods and minimized biases.


This study has several limitations. First, the diagnoses recorded in administrative claims databases are generally less accurate than those in prospective studies. Second, selection bias may exist due to the data extraction and study inclusion criteria, which did not include patients whose laboratory data were not recorded in the database and patients who received concomitant beta lactams. Third, baseline characteristics and severity of illness were different between the two groups. We adjusted the imbalances using propensity scores in the regression analyses. However, it is hard to completely remove residual confounding due to unmeasurable confounding factors. Fourth, although the study results are based on nationwide data in Japan, it may be difficult to directly apply these results to other regions because of the regional variations in pathogens and patterns of drug resistance. Nevertheless, we consider the findings of this study to be applicable to many countries because community prevalence of ESBL-producing Enterobacteriaceae has been reported to be approximately 5%–10% in many countries and regions (38), which is comparable to the percentages reported from Japan (42). Fifth, although the database recorded the accurate date when the antibiotics were used, it did not include the detailed time of administration of the antibiotics. Thus, we did not interpret the study findings from the viewpoint of pharmacokinetics and pharmacodynamics. Sixth, we could not speculate on the reason for the switch of antibiotics from the viewpoints of cultures and susceptibility because the database does not contain such data. Finally, although several clinically important antimicrobial resistance pathogens, such as MRSA, were diagnosable according to the ICD-10, detailed information on other pathogens and drug resistance was not recorded. Although the sensitivity and specificity of ICD-10 codes for C. difficile infection have been reported so far, there are no sufficient data on the sensitivity and specificity of ICD-10 codes to detect infection by MRSA and multidrug-resistant pathogens. Furthermore, as our definition of multidrug-resistant pathogens was also based on the ICD-10 coding, the inability to evaluate the exact prevalence of multidrug-resistant pathogens based on antimicrobial susceptibility was a major limitation that potentially affected the generalizability of the study findings. Namely, the effect of carbapenem therapies as initial therapy for sepsis would vary according to the prevalence of multidrug-resistant pathogens in the regions. However, because the present study was conducted using a nationwide multicenter cohort, the prevalence of multidrug-resistant pathogens was assumed to approximate that reported in previous national surveillance (51). Therefore, the study findings could be generalizable to many other nations where a prevalence of multidrug-resistant pathogens similar to that of Japan has been reported.


Compared with the noncarbapenem broad-spectrum antibiotics, carbapenems as initial empiric therapy were not associated with significantly lower mortality overall in patients with sepsis. Scientific evidence on the benefits of carbapenem therapy as initial therapy in sepsis is still limited, and further prospective studies need to be performed.


The authors thank all of the nurses and physicians in the participating institutions and all of the patients who contributed to this study.


1. Singer M, Deutschman CS, Seymour CW, et al.: The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016; 315:801–810

2. Fleischmann-Struzek C, Mellhammar L, Rose N, et al.: Incidence and mortality of hospital- and ICU-treated sepsis: Results from an updated and expanded systematic review and meta-analysis. Intensive Care Med 2020; 46:1552–1562

3. Rudd KE, Johnson SC, Agesa KM, et al.: Global, regional, and national sepsis incidence and mortality, 1990-2017: Analysis for the global burden of disease study. Lancet 2020; 395:200–211

4. Ibrahim EH, Sherman G, Ward S, et al.: The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in the ICU setting. Chest 2020; 118:146–155

5. Barie PS, Hydo LJ, Shou J, et al.: Influence of antibiotic therapy on mortality of critical surgical illness caused or complicated by infection. Surg Infect (Larchmt) 2005; 6:41–54

6. Kumar A, Roberts D, Wood KE, et al.: Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006; 34:1589–1596

7. Kumar A, Ellis P, Arabi Y, et al.; Cooperative Antimicrobial Therapy of Septic Shock Database Research Group: Initiation of inappropriate antimicrobial therapy results in a five-fold reduction of survival in human septic shock. Chest 2009; 136:1237–1248

8. Paul M, Shani V, Muchtar E, et al.: Systematic review and meta-analysis of the efficacy of appropriate empiric antibiotic therapy for sepsis. Antimicrob Agents Chemother 2010; 54:4851–4863

9. Ferrer R, Martin-Loeches I, Phillips G, et al.: Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: Results from a guideline-based performance improvement program. Crit Care Med 2014; 42:1749–1755

10. Rhodes A, Evans LE, Alhazzani W, et al.: Surviving sepsis campaign: International guidelines for management of sepsis and septic shock: 2016. Intensive Care Med 2017; 43:304–377

11. Evans L, Rhodes A, Alhazzani W, et al.: Surviving sepsis campaign: International guidelines for management of sepsis and septic shock 2021. Intensive Care Med 2021; 47:1181–1247

12. Dulhunty JM, Paterson DL, Webb SAR, et al.: Antimicrobial utilization in 37 Australian and New Zealand intensive care units. Anaesth Intensive Care 2011; 39:231–237

13. Paterson DL, Ko WC, Von Gottberg A, et al.: Antibiotic therapy for Klebsiella pneumoniae bacteremia: Implications of production of extended-spectrum beta-lactamases. Clin Infect Dis 2004; 39:31–37

14. Perez F, Bonomo RA: Can we really use β-lactam/β-lactam inhibitor combinations for the treatment of infections caused by extended-spectrum β-lactamase-producing bacteria? Clin Infect Dis 2012; 54:175–177

15. Abe T, Ogura H, Shiraishi A, et al.; JAAM FORECAST group: Characteristics, management, and in-hospital mortality among patients with severe sepsis in intensive care units in Japan: The FORECAST study. Crit Care 2018; 22:322

16. Rhodes NJ, Wagner JL, Davis SL, et al.; MAD-ID Research Network: Trends in and predictors of carbapenem consumption across North American hospitals: Results from a multicenter survey by the MAD-ID research network. Antimicrob Agents Chemother 2019; 63:e00327–e00319

17. European Surveillance of Antimicrobial Consumption Network: Antimicrobial consumption in the EU/EEA (ESAC-Net)—Annual epidemiological report for 2021. Available at: https://www.ecdc.europa.eu/en/publications-data/surveillance-antimicrobial-consumption-europe-2021. Accessed December 21, 2022

18. Hussein K, Sprecher H, Mashiach T, et al.: Carbapenem resistance among Klebsiella pneumoniae isolates: Risk factors, molecular characteristics, and susceptibility patterns. Infect Control Hosp Epidemiol 2009; 30:666–671

19. Patel G, Huprikar S, Factor SH, et al.: Outcomes of carbapenem-resistant Klebsiella pneumoniae infection and the impact of antimicrobial and adjunctive therapies. Infect Control Hosp Epidemiol 2008; 29:1099–1106

20. Benanti GE, Brown ART, Shigle TL, et al.: Carbapenem versus cefepime or piperacillin-tazobactam for empiric treatment of bacteremia due to extended-spectrum-beta-lactamase-producing Escherichia coli in patients with hematologic malignancy. Antimicrob Agents Chemother 2019; 63:e01813–e01818

21. Chang YY, Yang YS, Wu SL, et al.; ACTION Study Group: Comparison of cefepime-cefpirome and carbapenem therapy for Acinetobacter bloodstream infection in a multicenter study. Antimicrob Agents Chemother 2020; 64:e02392–e02319

22. Palacios-Baena ZR, Gutiérrez-Gutiérrez B, Calbo E, et al.; Spanish Network for Research in Infectious Diseases (REIPI)/European Study Group of Bloodstream Infections and Sepsis (ESGBIS)/INCREMENT Group: Empiric therapy with carbapenem-sparing regimens for bloodstream infections due to extended-spectrum beta-lactamase-producing Enterobacteriaceae: Results from the INCREMENT cohort. Clin Infect Dis 2017; 65:1615–1623

23. Vardakas KZ, Tansarli GS, Rafailidis PI, et al.: Carbapenems versus alternative antibiotics for the treatment of bacteraemia due to Enterobacteriaceae producing extended-spectrum beta-lactamases: A systematic review and meta-analysis. J Antimicrob Chemother 2012; 67:2793–2803

24. Tashiro M, Obata Y, Takazono T, et al.: Association between fluid infusions and the recovery from acute kidney injury in patients administered liposomal amphotericin B: A nationwide observational study. Ren Fail 2022; 44:282–292

25. Angus DC, Linde-Zwirble WT, Lidicker J, et al.: Epidemiology of severe sepsis in the United States: Analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001; 29:1303–1310

26. Wilhelms SB, Huss FR, Granath G, et al.: Assessment of incidence of severe sepsis in Sweden using different ways of abstracting International Classification of Diseases codes: Difficulties with methods and interpretation of results. Crit Care Med 2010; 38:1442–1449

27. Liu B, Hadzi-Tosev M, Liu Y, et al.: Accuracy of International Classification of Diseases, 10th revision codes for identifying sepsis: A systematic review and meta-analysis. Crit Care Explor 2022; 4:e0788

28. Kollef MH, Morrow LE, Niederman MS, et al.: Clinical characteristics and treatment patterns among patients with ventilator-associated pneumonia. Chest 2006; 129:1210–1218

29. Leone M, Bechis C, Baumstarck K, et al.; AZUREA Network Investigators: De-escalation versus continuation of empirical antimicrobial treatment in severe sepsis: A multicenter non-blinded randomized noninferiority trial. Intensive Care Med 2014; 40:1399–1408

30. Charlson ME, Pompei P, Ales KL, et al.: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 1987; 40:373–383

31. Quan H, Sundararajan V, Halfon P, et al.: Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005; 43:1130–1139

32. Kidney Disease: Improving global outcomes (KDIGO) acute kidney injury work group: KDIGO clinical practice guideline for acute kidney injury. Kidney Int 2012; 2:1–138

33. Jones G, Taright N, Boelle PY, et al.: Accuracy of ICD-10 codes for surveillance of Clostridium difficile infections, France. Emerg Infect Dis 2012; 18:979–981

34. Gayat E, Pirracchio R, Resche-Rigon M, et al.: Propensity scores in intensive care and anaesthesiology literature: A systematic review. Intensive Care Med 2012; 36:1993–2003

35. D’Agostino RB Jr, D’Agostino RB Sr: Estimating treatment effects using observational data. JAMA 2007; 297:314–316

36. Feng M, McSparron JI, Kien DT, et al.: Transthoracic echocardiography and mortality in sepsis: Analysis of the MIMIC-III database. Intensive Care Med 2018; 44:884–892

37. Allan V, Ramagopalan SV, Mardekian J, et al.: Propensity score matching and inverse probability of treatment weighting to address confounding by indication in comparative effectiveness research of oral anticoagulants. J Comp Eff Res 2020; 9:603–614

38. Cheng L, Nelson BC, Mehta M, et al.: Piperacillin-tazobactam versus other antibacterial agents for treatment of bloodstream infections due to AmpC beta-lactamase-producing Enterobacteriaceae. Antimicrob Agents Chemother 2017; 61:e00276–e00217

39. Barradell LB, Bryson HM: Cefepime. A review of its antibacterial activity, pharmacokinetic properties and therapeutic use. Drugs 1994; 47:471–505

40. Woerther PL, Burdet C, Chachaty E, et al.: Trends in human fecal carriage of extended-spectrum beta-lactamases in the community: Toward the globalization of CTX-M. Clin Microbiol Rev 2013; 26:744–758

41. Vardakas KZ, Tansarli GS, Rafailidis PI, et al.: Carbapenems versus alternative antibiotics for the treatment of bacteraemia due to Enterobacteriaceae producing extended-spectrum beta-lactamases: A systematic review and meta-analysis. J Antimicrob Chemother 2012; 67:2793–2803

42. Tamma PD, Han JH, Rock C, et al.; Antibacterial Resistance Leadership Group: Carbapenem therapy is associated with improved survival compared with piperacillin-tazobactam for patients with extended-spectrum beta-lactamase bacteremia. Clin Infect Dis 2015; 60:1319–1325

43. Lee NY, Lee CC, Huang WH, et al.: Cefepime therapy for monomicrobial bacteremia caused by cefepime-susceptible extended-spectrum beta-lactamase-producing Enterobacteriaceae: MIC matters. Clin Infect Dis 2013; 56:488–495

44. Nakama R, Shingaki A, Miyazato H, et al.: Current status of extended spectrum β-lactamase-producing Escherichia coli, Klebsiella pneumoniae and Proteus mirabilis in Okinawa prefecture, Japan. J Infect Chemother 2016; 22:281–286

45. Capsoni N, Bellone P, Aliberti S, et al.: Prevalence, risk factors and outcomes of patients coming from the community with sepsis due to multidrug resistant bacteria. Multidiscip Respir Med 2019; 14:23

46. Seymour CW, Gesten F, Prescott HC, et al.: Time to treatment and mortality during mandated emergency care for sepsis. N Engl J Med 2017; 376:2235–2244

47. Teng C, Reveles KR, Obodozie-Ofoegbu OO, et al.: Clostridium difficile infection risk with important antibiotic classes: An analysis of the FDA adverse event reporting system. Int J Med Sci 2019; 16:630–635

48. van Loon K, Voor In ‘t Holt AF, Vos MC: A systematic review and meta-analyses of the clinical epidemiology of carbapenem-resistant Enterobacteriaceae. Antimicrob Agents Chemother 2017; 62:e01730–e01717

49. World Health Organization: Antimicrobial resistance: Global report on surveillance. World Health Organization, 2014. Available at: https://apps.who.int/iris/handle/10665/112642. Accessed April 31, 2022

50. Nathwani D, Raman G, Sulham K, et al.: Clinical and economic consequences of hospital-acquired resistant and multidrug-resistant Pseudomonas aeruginosa infections: A systematic review and meta-analysis. Antimicrob Resist Infect Contr 2014; 3:32

51. Kirikae T, Mizuguchi Y, Arakawa Y: Investigation of isolation rates of Pseudomonas aeruginosa with and without multidrug resistance in medical facilities and clinical laboratories in Japan. J Antimicrob Chemother 2008; 61:612–615

52. Bad Bugs, No Drugs, as Antibiotic Discovery Stagnates … A Public Health Crisis Brews: Available at: https://www.idsociety.org/globalassets/idsa/policy–advocacy/current_topics_and_issues/antimicrobial_resistance/10×20/statements/070104-as-antibiotic-discovery-stagnates-a-public-health-crisis-brews.pdf. Accessed April 18, 2022

53. Kalil AC, Metersky ML, Klompas M, et al.: Management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 Clinical Practice Guidelines by the Infectious Diseases Society of America and the American Thoracic Society. Clin Infect Dis 2016; 63:e61–e111

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *