Skip to main content
NCBI home page
Medicine logoLink to Medicine
. 2026 Jan 2;105(1):e47024. doi: 10.1097/MD.0000000000047024

Association between empiric combination therapy using antipseudomonal agents and in-hospital mortality among patients with sepsis in intensive care units in Japan: A propensity score-matched analysis

Hiroki Iriyama a, Toshikazu Abe b,c,a,*, Taro Imaeda d, Akira Komori e, Takehiko Oami d, Tuerxun Aizimu d, Nozomi Takahashi d, Yasuo Yamao d, Satoshi Nakagawa f, Hiroshi Ogura g, Yutaka Umemura g, Asako Matsushima h, Kiyohide Fushimi i, Nobuaki Shime j, Taka‑aki Nakada d
PMCID: PMC12778166  PMID: 41496138

Abstract

Sepsis mortality due to drug-resistant organisms remains high. However, the effectiveness of empiric combination therapy for sepsis caused by potential drug-resistant gram‑negative bacilli remains debated. This was a nested case-control study using Japanese nationwide data from a medical reimbursement system between 2010 and 2017. The eligibility criteria were adult patients who were admitted to the intensive care unit, with new-onset sepsis at or after admission. We enrolled patients with combination therapy and monotherapy by propensity score matching using their baseline characteristics. We developed a logistic regression model including combination therapy, severity of organ dysfunction, and interaction terms as the adjusting variables. Overall, 7084 patients were included in this study, 3542 and 3542 of whom were enrolled in the combination therapy and monotherapy groups, respectively. The logistic regression model revealed combination therapy and moderate and severe organ dysfunction were significantly associated with in-hospital mortality (odds ratio [OR] 95% confidence interval [CI]: 1.52 [1.20–1.91], OR [95% CI]: 1.78 [1.43–2.23], and OR [95% CI]: 4.22 [2.61–6.81], respectively). Interaction between combination therapy and severe organ dysfunction was statistically significant (OR [95% CI]: 0.41 [0.21–0.80], P for interaction = .009). However, adjusted in-hospital mortality based on the model was not significantly different between the combination therapy and monotherapy groups among patients with severe organ dysfunction (44% [95% CI: 38–51] vs 47% [95% CI: 41–54]). Our results did not reveal any specific effectiveness of combination therapy; however, the significant interaction between combination therapy and severe organ dysfunction may indicate the possible survival benefit of combination therapy for severe sepsis subset.

Keywords: combination therapy, drug resistance, gram‑negative bacilli, sepsis

1. Introduction

Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection.[1] It is one of the most substantial global health concerns, causing millions of deaths annually.[2] Although treatment for sepsis has improved over the past decade, sepsis mortality due to drug-resistant organisms remains high.[3,4] Combination therapy using multiple antimicrobial agents has conventionally been administered for sepsis caused by potential drug-resistant organisms.

A delay in administering at least one effective antimicrobial agent for sepsis caused by drug-resistant organisms is associated with increased mortality.[5,6] The current sepsis guideline suggests combination therapy using 2 antimicrobial agents with coverage of gram‑negative bacilli (GNB) for empiric therapy in patients at a high risk for drug-resistant organisms, especially multidrug-resistant organisms; however, the current limited data could not provide specific criteria for combination therapy administration.[7–11] Prompt evaluation of drug-resistant organism risk and determination of empiric combination therapy in patients with sepsis, while adhering to the current 1-hour bundle, remains a clinical challenge.[11]

The effectiveness of empiric combination therapy for sepsis caused by potential drug-resistant GNB is controversial. There may be various reasons for this controversy. First, there could be variation of local prevalence of drug-resistant organisms, proportion of participating patients at high risk for drug-resistant organisms, and severity of sepsis among clinical studies.[12–15] Second, relevant studies focus not on combination therapy and monotherapy but on conventional antimicrobial and newly developed antimicrobial agents.[16–20] Third, some previous studies have implemented combination therapy that covered only drug-resistant organisms of community-onset sepsis, without addressing those linked to nosocomial sepsis despite the overlapping risk between drug-resistant organisms and nosocomial infection organisms.[18,20–22]

Empiric combination therapy is intended to treat community-onset and nosocomial sepsis caused by drug-resistant GNB. This study aimed to investigate whether clinical outcomes could be improved in patients with severe sepsis at risk for drug-resistant GNB.

2. Materials and methods

2.1. Design and setting

We conducted a nested case-control study using Japanese nationwide data from a medical reimbursement system database for acute care named “Diagnosis Procedure Combination (DPC)” between 2010 and 2017. It included 1237 acute care hospitals, accounting for 71.5% of acute care hospitals in Japan in 2017.[23,24] The database comprises patient data, including age, sex, diagnostic codes, procedures, drug administrations, and discharge status. Diagnostic codes include primary diagnosis, comorbidities on admission, and post-hospitalization complications. Clinical data including vital signs, laboratory data, blood culture results, and x-ray or computed tomography image were not included in the database.

2.2. Participants

The eligible criteria were as follows: age ≥ 18 years at the time of admission, new-onset sepsis at or after hospital admission, intravenous antimicrobial agent administration during the empiric therapy phase of sepsis, and admission to the intensive care unit (ICU) more than 48 hours. The number of patients with monotherapy was much higher than that with combination therapy after inclusion criteria. We used PS matching to select patients treated with monotherapy to ensure that the monotherapy group had as high risk for drug-resistant organisms as the combination therapy group.

2.3. Data collection and definitions

In-hospital mortality was the primary outcome of interest. Sepsis was defined using criteria similar to those reported in a previous study, based on sepsis-3: presumed infection and at least 1 acute organ dysfunction (Fig. 1).[24] Presumed infection was defined as blood culture collection at or after admission and the administration of new antimicrobial. As the database does not include clinical data, acute organ dysfunction was defined using diagnostic codes and organ-specific treatment data.[24] Acute organ dysfunction encompasses cardiac dysfunction, respiratory dysfunction, renal dysfunction, hepatic dysfunction, thrombocytopenia, coagulation dysfunction, and acidosis. Severity of acute organ dysfunction was defined using the number of organ dysfunctions as follows: mild, 1 to 2; moderate, 3 to 4; and severe, 5 or more. Antimicrobial agents in this study were defined as intravenous GNB agents covering Pseudomonas aeruginosa because this definition excludes the possibility that monotherapy was less effective because of not covering nonresistant GNBs responsible for nosocomial infections.[25,26] P. aeruginosa is the most common GNB causing nosocomial infection. Typically, combination therapy has activity against P. aeruginosa because it consists of β-lactam agents plus quinolone or aminoglycoside. We focused on resistant GNB observed mainly in community-onset infection; however, studies have reported that resistant GNBs, such as extended spectrum β-lactamase (ESBL) producing Escherichia coli and Klebsiella pneumoniae, are less frequently observed in Japan than in the western countries.[27–30] In fact, third-generation cephalosporin, ceftriaxone, is a screening drug to detect ESBL producing pathogen that is empirically administered to patients with sepsis in many acute care hospitals and even in tertiary-level facilities in Japan.[31,32] Conversely, common nosocomial pathogen like P. aeruginosa is observed in Japan as frequently as in western countries among sepsis population.[25,32,33] Based on such epidemiological feature in Japan, we included only GNB agents with antipseudomonal coverage to focus on drug-resistant GNB. Antimicrobial agents included in this study are in Table S1, Supplemental Digital Content, https://links.lww.com/MD/R86. The antimicrobial administration was required to begin withi ±1 calendar days of blood culture. The duration of administration was required for ≥3 calendar days to reduce the risk for overinclusion and lower specificity for sepsis. Combination therapy was defined as typical concomitant administration of both β-lactam antimicrobials and non-β-lactam antimicrobial agents.[34–37] Patients were excluded if they received atypical combination therapy, including double β-lactam agents or double non-β-lactam agents. Monotherapy was defined as administration of single antimicrobial agents. As aminoglycoside agents and fosfomycin have inadequate clinical efficacy at most in the setting of monotherapy, patients were excluded if an aminoglycoside agent or fosfomycin was administered to them as a single agent.[36,38–40]

Figure 1.

Figure 1.

Open in a new tab

Definition of sepsis and acute organ dysfunction.

Diagnostic codes were defined using the international classification of diseases, 10th revision (ICD-10) in the diagnosis procedure combination database. The focus of infection was also determined by referring to the ICD-10 codes for presumed focus of infection. Comorbidities were also identified by referring to ICD-10 codes using predefined algorithms.[41]

As outcomes, we collected data for in-hospital mortality, length of antibiotic treatment, and length of hospital and ICU stays. Data regarding ICU admission were extracted by providing additional medical fee for intensive care.

2.4. Statistical analysis

We included patients with sepsis treated with either combination therapy or monotherapy from the database using propensity score (PS) matching to ensure that these 2 groups were equally balanced in terms of baseline characteristics associated with risk for drug-resistant organisms. Variables including age, sex, admission times, number and type of comorbidities, blood culture day, number and type of infection site, and concomitant administration of antimethicillin-resistant Staphylococcus aureus agents were used to calculate the PS. We selected these variables on the basis of clinical relevance and previous research. We employed nearest neighbor matching without replacement in a 1:1 manner and used a caliper of 0.001 standard deviation of the logit of the PS. We evaluated the standardized mean difference of the variables over 0.1 as a meaningful imbalance following PS matching. We calculated descriptive statistics comparing the combination therapy and monotherapy groups, using the Wilcoxon signed-rank and chi-square tests for continuous and categorical variables, respectively, having first evaluated the normality of all continuous variables using the Kolmogorov–Smirnov test. No continuous variables had normal distribution.

To evaluate the association between combination therapy and in-hospital mortality among participants with sepsis ranging from mild to severe acute organ dysfunction, we developed a conditional logistic regression model. Combination therapy, severity of acute organ dysfunction, and interaction terms were included as adjusting variables in the model because we hypothesized that interactions existed between combination therapy and severity of sepsis with in-hospital mortality as the dependent variable.[15] Consequently, we estimated the adjusted in-hospital mortality on the basis of the results of the logistic regression model using the marginal standardization method.[42]

All P values were 2-sided; P values < .05 were considered statistically significant. Statistical analyses were performed using R software (version 4.4.1).[43]

2.5. Sensitivity analysis

We conducted a sensitivity analysis by focusing on participants without pneumonia because patients with pneumonia may receive combination therapy to cover typical and atypical pneumonia. To evaluate the association between combination therapy and in-hospital mortality among participants without pneumonia, we developed a conventional logistic regression model. The adjusting variables were the same as the main analysis.

2.6. Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Chiba University Graduate School of Medicine (approval number: 3429). We confirmed that all methods were performed in accordance with the relevant guidelines. The need for informed consent was waived by the Institutional Review Board of Chiba University Graduate School of Medicine because the database was anonymized.

3. Results

A total of 78,313 patients who received at least a GNB antimicrobial agent in the database were eligible for this study after applying the eligible criteria. After exclusion criteria and PS matching, the combination therapy and monotherapy groups comprised 3542 patients, respectively (Fig. 2). No meaningful imbalance of the variables for the PS matching was detected between the 2 groups (Tables 1 and 2).

Figure 2.

Figure 2.

Open in a new tab

Patient selection. MRSA = methicillin-resistant Staphylococcus aureus.

Table 1.

Baseline characteristics of the study participants.

Overall Combination therapy Monotherapy SMD P value
7084 3542 3542
Age 72 (63–80) 72 (63–79) 72 (62–80) 0.009
Male 4881 (68.9) 2438 (68.8) 2443 (69.0) 0.003
Admission times
 1 5508 (77.8) 2742 (77.4) 2766 (78.1) 0.024
 2 or more 1576 (22.2) 800 (22.6) 776 (21.9)
Comorbidity
 Malignant tumor 1284 (18.1) 662 (18.7) 622 (17.6) 0.029
 Hypertension 1361 (19.2) 679 (19.2) 682 (19.3) 0.002
 Diabetes mellitus 1741 (24.6) 868 (24.5) 873 (24.6) 0.003
 Heart failure 2025 (28.6) 1021 (28.8) 1004 (28.3) 0.011
 Cerebrovascular disease 729 (10.3) 373 (10.5) 356 (10.1) 0.016
 Ischemic heart disease 817 (11.5) 428 (12.1) 389 (11.0) 0.034
 Chronic respiratory disease 821 (11.6) 416 (11.7) 405 (11.4) 0.01
 Chronic renal failure 257 (3.6) 122 (3.4) 135 (3.8) 0.02
Blood culture day
 1–2 5204 (73.5) 2541 (71.7) 2663 (75.2) 0.078
 3 or more 1880 (26.5) 1001 (28.3) 879 (24.8)
Infection site
 Respiratory 4315 (60.9) 2163 (61.1) 2152 (60.8) 0.006
 Abdominal 570 (8.0) 289 (8.2) 281 (7.9) 0.002
 Bone and soft tissue 296 (4.2) 142 (4.0) 154 (4.3) 0.017
 Urogenital 197 (2.8) 98 (2.8) 99 (2.8) 0.002
 Other 530 (7.5) 268 (7.6) 262 (7.4) 0.006
 Unknown 1397 (19.7) 704 (19.9) 693 (19.6) 0.008
Severity of acute organ dysfunction
 Mild 2645 (37.3) 1245 (35.1) 1400 (39.5) .001
 Moderate 3991 (56.3) 2072 (58.5) 1919 (54.2)
 Severe 448 (6.3) 225 (6.4) 223 (6.3)
Open in a new tab

Data are presented as counts (proportions) for categorical variables and medians (interquartile ranges) for continuous variables. The respiratory infection includes pneumonia and pyothorax, and other infections include central nervous system, heart, and blood infections, and sexually transmitted disease.

SMD = standardized mean difference.

Table 2.

Empiric antimicrobial agent of the study participants.

Overall Combination therapy Monotherapy SMD P value
7084 3542 3542
β-Lactam agents
 Piperacillin derivatives 2203 (31.1) 1008 (28.5) 1195 (33.7) <.001
 Sulbactam/cefoperazone 19 (0.3) 5 (0.1) 14 (0.4) .066
 Third-generation cephalosporins 67 (0.9) 28 (0.8) 39 (1.1) .22
 Fourth-generation cephalosporins 279 (3.9) 136 (3.8) 143 (4.0) .714
 Carbapenems 4394 (62.0) 2452 (69.2) 1942 (54.8) <.001
 Monobactam 14 (0.2) 6 (0.2) 8 (0.2) .789
Non-β-lactam agents
 Aminoglycosides 500 (7.1) 500 (14.1) 0 (0)
 Quinolones 3229 (45.6) 3028 (85.5) 201 (5.7) <.001
 Fosfomycin 21 (0.3) 21 (0.6) 0 (0)
 Another antipseudomonal agent 2 (0.0) 2 (0.1) 0 (0) .48
 Anti-MRSA agents 513 (7.2) 255 (7.2) 258 (7.3) 0.003
Open in a new tab

Data are presented as counts (proportions) for categorical variables.

Piperacillin derivatives include tazobactam/piperacillin and piperacillin.

Another antipseudomonal agent is colistin.

MRSA = methicillin-resistant Staphylococcus aureus, SMD = standardized mean difference.

The baseline characteristics of the study participants are shown in Table 1. The median age was 72 (interquartile range [IQR]: 63–80) years, and males accounted for 4881/7084 (68.9%) of the study participants. A blood culture day of 3 or later indicating nosocomial infection was conducted in 1880 (26.5%) participants. The most frequent infection site was respiratory (4315, 60.9%), followed by abdominal (570, 8.0%) and bone/soft tissue (296, 4.2%). The combination therapy group had a higher proportion of patients with moderate and severe acute organ dysfunction than the monotherapy group (58.5% vs 54.2%, 6.4% vs 6.3%, respectively).

The empiric antimicrobial agents in this study are presented in Table 2. Frequent β-lactam agents were carbapenems (62.0%) and piperacillin derivatives (31.1%) in the entire cohort. The combination therapy group more frequently used carbapenems than the monotherapy group (69.2% vs 54.8%). The combination therapy group less frequently used piperacillin derivatives than the monotherapy group (28.5% vs 33.7%). Frequent non-β-lactam agents as the second agent in the combination therapy group were quinolones (85.5%) and aminoglycosides (14.1%).

The study outcomes are shown in Table 3. The combination therapy group had significantly higher crude in-hospital mortality than the monotherapy group (31.9% vs 27.4%, P < .001). The combination therapy group also had significantly longer length of antibiotic treatment (23 [IQR: 16–33] vs 15 [IQR: 10–25] days, P < .001) and longer length of ICU stay than the monotherapy group (8 [IQR: 5–14] vs 7 [IQR: 5–13] days, P < .001). The conditional logistic regression model showed that combination therapy and moderate and severe acute organ dysfunction were independently associated with in-hospital mortality (odds ratio [OR] 95% confidence interval [CI]: 1.52 (1.20–1.91), OR (95% CI): 1.78 (1.43–2.23), and OR (95% CI): 4.22 (2.61–6.81), respectively; Table 4). Interaction between combination therapy and severe acute organ dysfunction was statistically significant (OR [95% CI]: 0.41 [0.21–0.80], P for interaction = .009). The adjusted in-hospital mortality between the 2 groups by severity of acute organ dysfunction is depicted in Figure 3. Among patients with mild acute organ dysfunction, the combination therapy group had significantly higher adjusted in-hospital mortality than the monotherapy group (27% vs 20%). Conversely, among patients with severe acute organ dysfunction, the combination therapy group had lower adjusted in-hospital mortality than the monotherapy group; however, it was not statistically significant (44% vs 47%).

Table 3.

Outcomes among the study participants.

Overall Combination therapy Monotherapy P value
7084 3542 3542
In-hospital mortality 2099 (29.6) 1129 (31.9) 970 (27.4) <.001
Length of antibiotic treatment 19 (12–30) 23 (16–33) 15 (10–25) <.001
Length of hospital stay 29 (17–50) 29 (17–49) 29 (16–50) .526
Length of ICU stay 8 (5–13) 8 (5–14) 7 (5–13) <.001
Open in a new tab

Data are presented as counts (proportions) for categorical variables and medians (interquartile ranges) for continuous variables. The length of antibiotic treatment not only includes empiric therapy but also the antibiotics administered during hospitalization.

ICU = intensive care unit.

Table 4.

Association of combination therapy and acute organ dysfunction severity to in-hospital mortality among the participants.

OR 95% CI P value
Main effect
 Monotherapy Reference
 Combination therapy 1.52 1.20–1.91 <.001
 Mild acute organ dysfunction Reference
 Moderate acute organ dysfunction 1.78 1.43–2.23 <.001
 Severe acute organ dysfunction 4.22 2.61–6.81 <.001
Interaction
 Combination therapy * moderate acute organ dysfunction 0.76 0.55–1.05 .10
 Combination therapy * severe acute organ dysfunction 0.41 0.21–0.80 .009
Open in a new tab

CI = confidence interval, OR = odds ratio.

Figure 3.

Figure 3.

Open in a new tab

Adjusted in-hospital mortality by severity of acute organ dysfunction among the study participants.

The sensitivity analysis included 2806 patients without pneumonia. The conventional logistic regression model revealed that combination therapy and moderate and severe acute organ dysfunction were independently associated with in-hospital mortality, as well as in the main analysis (Table S2, Supplemental Digital Content, https://links.lww.com/MD/R86). Furthermore, the sensitivity analysis model revealed similar results regarding the interaction between combination therapy and acute organ dysfunction.

4. Discussion

This nationwide nested case-control study aimed to investigate the association between empiric combination therapy and in-hospital mortality among patients with sepsis due to potential drug-resistant GNB. Our results did not show an association between empiric combination therapy and low in-hospital mortality. However, our results showed a significant interaction between empiric combination therapy and severe acute organ dysfunction, suggesting the possible effectiveness of combination therapy for the most severe subset of sepsis.

In clinical practice, compared with monotherapy, empiric combination therapy for sepsis tended to be administered to patients with more severe conditions. In our cohort matched by baseline characteristics, the combination therapy group had a higher proportion of patients with moderate and severe acute organ dysfunction, higher in-hospital mortality, and longer length of antibiotic treatment and ICU stay than the monotherapy group. The current sepsis guideline recommendation for empiric combination therapy does not include sepsis severity.[11] In clinical settings, physicians may consider sepsis severity as an essential factor for the decision of combination therapy, although our data did not include all known risk factors for drug-resistant organisms.

Empiric combination therapy covering drug-resistant GNB may be effective for severe subset of sepsis because of effect modification according to the severity of acute organ dysfunction. Our study showed significant interaction between combination therapy and severe acute organ dysfunction in the logistic regression model. In the same model, among patients with severe acute organ dysfunction, our study could not show significant different adjusted mortality between the 2 groups probably because the combination therapy group had higher crude mortality of patients with mild acute organ dysfunction than the monotherapy group. However, the combination therapy would increase in-hospital mortality is theoretically difficult to consider. Our study probably could not adjust residual confounding severity of disease by using only acute organ dysfunction. Several previous meta-analyses showed no clinical benefit of combination therapy; however, a previous meta-regression study showed clinical benefit of combination therapy alone among patients at a high risk of mortality.[15,34,44,45] In the meta-regression study, combination therapy was beneficial among patients at higher mortality risk, such as more than 25%. In our study, the mortality rate among patients with severe acute organ dysfunction was 45.8%, which is consistent with the meta-regression analysis. Recent randomized clinical trials that did not demonstrate the effectiveness of combination therapy had low mortality rates[46–48] or small sample sizes.[16,49] In our study, the reason why empiric combination therapy was possibly effective for severe subset of sepsis could be because of the high mortality event rate caused by inappropriate empiric therapy within the initial empiric phase. Prospective interventional study is needed to confirm the finding of the present study.

Our study had several limitations. First, there could be information bias regarding the definition of sepsis. Sepsis is a clinical diagnosis based on suspicion of infection and life-threatening organ dysfunction; however, our database did not include clinical data. A previous study using the same database employed a similar method and showed epidemiological characteristics consistent with those in the US and Europe.[24] The epidemiological data were similar to those reported for the US and European countries. Second, there could be survival bias because our study included only patients who survived for at least more than 3 calendar days because combination therapy administration was required for 3 or more calendar days. However, a study design that includes patients who died within 2 days of sepsis onset would risk overinclusion and lower specificity for sepsis because we used a medical reimbursement system database. Such study design could include patients with shock who were initially administered antimicrobial agents but subsequently revealed without sepsis. Third, our database did not explain the reason for combination therapy administration. Our study could include fluoroquinolones to cover atypical pneumonia or aminoglycosides aimed at synergistic effect for infective endocarditis. However, sensitivity analysis that excluded pneumonia revealed the similar results as the main analysis. The proportion of patients with infective endocarditis was lower than 5% in our cohort (infective endocarditis was included in other infection sites). Additionally, our study may have included second-line empiric therapy initiated in response to clinical deterioration. Fourth, there could be residual confounder such as baseline disease severity between the 2 groups because we could not use clinical data such as SOFA score during patient selection. The combination therapy group had significantly higher in-hospital mortality of patients with mild acute organ dysfunction than the monotherapy group. Previous studies have indicated the potential harm of combination regimens, such as nephrotoxicity, but not with high mortality.[21,34] We may have not removed confounding disease severity. Fifth, the higher prevalence of carbapenems administered in the combination therapy group than in the monotherapy group could contribute to the low mortality in the severe subset of the cohort. In the case of Enterobacteriaceae, such as Escherichia and Klebsiella, resistance mechanisms like ESBL production are common and carbapenems are typically first-line treatment. Instead of carbapenems, piperacillin/tazobactam was administered in the higher proportion of the monotherapy group; however, prevalence of resistance to piperacillin/tazobactam among ESBL producing Enterobacteriaceae is low in Japan.[50,51] Moreover, we developed an additional conditional logistic regression model adjusting for carbapenem use. The significant interaction between combination therapy and severe acute organ dysfunction remained (Table S3, Supplemental Digital Content, https://links.lww.com/MD/R86). Sixth, definitive therapy could be residual confounder in this study because we analyzed only antimicrobial administration as empiric therapy. Seventh, the exposure – combination therapy – may reflect clinician behavior rather than a clinical indication of drug-resistant organisms due to the retrospective study design. This study could not include several known risk factors for drug-resistant organisms in the PS matching, including the presence of invasive devices, broad-spectrum antimicrobial agent administration before sepsis, selective digestive decontamination, and travel history to highly endemic countries. Instead, we substituted several variables in the database, including admission times, blood culture day after admission, and concomitant administration of antimethicillin-resistant S. aureus agent. Eighth, as our database did not have culture results and susceptibility, we cannot perform sensitivity analyses based on resistance profiles to confirm the association between the effectiveness of combination therapy and the appropriateness of empiric antimicrobial therapy. Ninth, the findings may not be generalizable to patients with limited antibiotic options, such as those with drug allergies. Combination therapy was defined as the concomitant administration of both β-lactam and non-β-lactam antimicrobial agents, implying that patients with β-lactam allergies were not included. Finally, our study could have information bias about infection sites. The medical reimbursement system database in Japan needs at least 1 primary admission diagnosis for each admission. Multiple infection sites such as infective embolism from infective endocarditis could be underestimated. Subsequently, we used the number and site of infection only for PS matching.

The administration of empiric combination therapy for sepsis should be carefully considered when the likelihood of clinical benefit is high and the risk of adverse effects is low. In this study, empiric combination therapy may have modified the rate of in-hospital mortality among patients with acute organ dysfunction, particularly benefiting those with severe organ dysfunction. Therefore, considering the high rate of organ dysfunction in sepsis, along with known risk factors for drug-resistant organisms, may be important in identifying patients who are most likely to benefit from combination therapy. Further prospective randomized controlled trials are needed to confirm our findings.

5. Conclusions

The results of this study did not reveal an association between empiric combination therapy with multiple GNB agents and low in-hospital mortality among patients with sepsis. However, the results demonstrated an interaction between empiric combination therapy and severe acute organ dysfunction, indicating the possible survival benefit of empiric combination therapy for sepsis covering drug-resistant organisms for the most severe subset of sepsis.

Acknowledgments

The authors thank Enago (www.enago.jp) for the English language review.

Author contributions

Conceptualization: Hiroki Iriyama, Toshikazu Abe, Taro Imaeda, Akira Komori, Takehiko Oami, Nozomi Takahashi, Satoshi Nakagawa, Hiroshi Ogura, Yutaka Umemura, Asako Matsushima, Kiyohide Fushimi, Nobuaki Shime, Taka‑aki Nakada.

Data curation: Hiroki Iriyama, Taro Imaeda, Tuerxun Aizimu, Yasuo Yamao.

Formal analysis: Hiroki Iriyama.

Methodology: Hiroki Iriyama, Toshikazu Abe.

Project administration: Hiroki Iriyama.

Software: Hiroki Iriyama.

Validation: Hiroki Iriyama.

Visualization: Hiroki Iriyama.

Writing – original draft: Hiroki Iriyama, Toshikazu Abe, Akira Komori.

Writing – review & editing: Hiroki Iriyama.

Supplementary Material

medi-105-e47024-s001.docx (43.9KB, docx)

Abbreviations:

CI
confidence interval
ESBL
extended spectrum β-lactamase
ICD-10
international classification of diseases, 10th revision
ICU
intensive care unit
IQR
interquartile range
OR
odds ratio
PS
propensity score

The study protocol was approved by the Institutional Review Board of Chiba University Graduate School of Medicine (approval number: 3429). The requirement for obtaining informed consent from the patients was waived since all data used were anonymized.

The authors have no funding and conflicts of interest to disclose.

The datasets generated during and/or analyzed during the current study are not publicly available, but are available from the corresponding author on reasonable request.

Supplemental Digital Content is available for this article.

How to cite this article: Iriyama H, Abe T, Imaeda T, Komori A, Oami T, Aizimu T, Takahashi N, Yamao Y, Nakagawa S, Ogura H, Umemura Y, Matsushima A, Fushimi K, Shime N, Nakada T. Association between empiric combination therapy using antipseudomonal agents and in-hospital mortality among patients with sepsis in intensive care units in Japan: A propensity score-matched analysis. Medicine 2026;105:1(e47024).

Contributor Information

Hiroki Iriyama, Email: hiroki.iriyama@showa-u.ac.jp.

Taro Imaeda, Email: taroh_2215@hotmail.com.

Akira Komori, Email: akg_6412@yahoo.co.jp.

Takehiko Oami, Email: seveneleven711thanks39@msn.com.

Tuerxun Aizimu, Email: ezim@smart119.biz.

Nozomi Takahashi, Email: nozomi_t1991@hotmail.com.

Yasuo Yamao, Email: yamao@smart119.biz.

Satoshi Nakagawa, Email: nakagawa-s@ncchd.go.jp.

Hiroshi Ogura, Email: ogura@hp-emerg.med.osaka-u.ac.jp.

Yutaka Umemura, Email: plum0022@yahoo.co.jp.

Asako Matsushima, Email: asakouab@gmail.com.

Kiyohide Fushimi, Email: kfushimi.hci@tmd.ac.jp.

Nobuaki Shime, Email: shime@koto.kpu-m.ac.jp.

Taka‑aki Nakada, Email: takanakada0@gmail.com.

References

  • [1].Singer M, Deutschman CS, Seymour CW, et al. The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA. 2016;315:801–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].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;395:200–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Spanik S, Novotny J, Mateicka F, et al. Bacteremia due to multiresistant gram-negative bacilli in neutropenic cancer patients: a case-controlled study. J Infect Chemother. 1999;5:180–4. [DOI] [PubMed] [Google Scholar]
  • [4].Cattaneo C, Antoniazzi F, Casari S, et al. P. aeruginosa bloodstream infections among hematological patients: an old or new question? Ann Hematol. 2012;91:1299–304. [DOI] [PubMed] [Google Scholar]
  • [5].Rodriguez-Bano J, Navarro MD, Romero L, et al. Bacteremia due to extended-spectrum -lactamase-producing Escherichia coli in the CTX-M era: a new clinical challenge. Clin Infect Dis. 2006;43:1407–14. [DOI] [PubMed] [Google Scholar]
  • [6].Kang C-I, Kim S-H, Park WB, et al. Bloodstream infections caused by antibiotic-resistant gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome. Antimicrob Agents Chemother. 2005;49:760–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Aloush V, Navon-Venezia S, Seigman-Igra Y, Cabili S, Carmeli Y. Multidrug-resistant pseudomonas aeruginosa: risk factors and clinical impact. Antimicrob Agents Chemother. 2006;50:43–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [8].Nakamura A, Miyake K, Misawa S, et al. Meropenem as predictive risk factor for isolation of multidrug-resistant Pseudomonas aeruginosa. J Hosp Infect. 2013;83:153–5. [DOI] [PubMed] [Google Scholar]
  • [9].Tartof SY, Kuntz JL, Chen LH, et al. Development and assessment of risk scores for carbapenem and extensive β-lactam resistance among adult hospitalized patients with Pseudomonas aeruginosa infection. JAMA Netw Open. 2018;1:e183927. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Seki M, Machida N, Yamagishi Y, Yoshida H, Tomono K. Nosocomial outbreak of multidrug-resistant Pseudomonas aeruginosa caused by damaged transesophageal echocardiogram probe used in cardiovascular surgical operations. J Infect Chemother. 2013;19:677–81. [DOI] [PubMed] [Google Scholar]
  • [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;47:1181–247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [12].Sievert DM, Ricks P, Edwards JR, et al. ; National Healthcare Safety Network (NHSN) Team and Participating NHSN Facilities. Antimicrobial-resistant pathogens associated with healthcare-associated infections summary of data reported to the national healthcare safety network at the centers for disease control and prevention, 2009–2010. Infect Control Hosp Epidemiol. 2013;34:1–14. [DOI] [PubMed] [Google Scholar]
  • [13].Itokazu GS, Quinn JP, Bell-Dixon C, Kahan FM, Weinstein RA. Antimicrobial resistance rates among aerobic gram-negative bacilli recovered from patients in intensive care units: evaluation of a national postmarketing surveillance program. Clin Infect Dis. 1996;23:779–84. [DOI] [PubMed] [Google Scholar]
  • [14].NNIS System. National Nosocomial Infections Surveillance (NNIS) System Report, Data Summary from January 1990-May 1999, issued June 1999. A report from the NNIS System. Am J Infect Control. 1999;27:520–32. [DOI] [PubMed] [Google Scholar]
  • [15].Kumar A, Safdar N, Kethireddy S, Chateau D. A survival benefit of combination antibiotic therapy for serious infections associated with sepsis and septic shock is contingent only on the risk of death: a meta-analytic/meta-regression study. Crit Care Med. 2010;38:1651–64. [DOI] [PubMed] [Google Scholar]
  • [16].Lerma FA, Group OB of TSIS. Efficacy of meropenem as monotherapy in the treatment of ventilator-associated pneumonia. J Chemother. 2001;13:70–81. [DOI] [PubMed] [Google Scholar]
  • [17].Christen D, Buchmann P, Geroulanos S. Imipenem/cilastatin versus aminoglycoside plus amoxicillin plus clindamycin in the treatment of serious postoperative infections. Scand J Infect Dis Suppl. 1987;52:11–4. [PubMed] [Google Scholar]
  • [18].Mouton Y, Deboscker Y, Bazin C, et al. Prospective, randomized, controlled study of imipenem-cilastatin versus cefotaxime-amikacin in the treatment of lower respiratory tract infection and septicemia at intensive care units]. Presse Med. 1990;19:607–12. [PubMed] [Google Scholar]
  • [19].Geroulanos S, Stern A, Christen D, Buchmann P. Antimicrobial management of postoperative infections in abdominal surgery: single or combination regimen? Clin Ther. 1990;12(Suppl B):34–42. [PubMed] [Google Scholar]
  • [20].Jaspers CAJJ, Kieft H, Speelberg B, et al. Meropenem versus cefuroxime plus gentamicin for treatment of serious infections in elderly patients. Antimicrob Agents Chemother. 1998;42:1233–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Schentag JJ, Wels PB, Reitberg DP, Walczak P, Van Tyle JH, Lascola RJ. A randomized clinical trial of moxalactam alone versus tobramycin plus clindamycin in abdominal sepsis. Ann Surg. 1983;198:35–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Kumar A, Zarychanski R, Light B, et al. ; Cooperative Antimicrobial Therapy of Septic Shock (CATSS) Database Research Group. Early combination antibiotic therapy yields improved survival compared with monotherapy in septic shock: a propensity-matched analysis*. Crit Care Med. 2010;38:1773–85. [DOI] [PubMed] [Google Scholar]
  • [23].Yamana H, Moriwaki M, Horiguchi H, Kodan M, Fushimi K, Yasunaga H. Validity of diagnoses, procedures, and laboratory data in Japanese administrative data. J Epidemiol. 2017;27:476–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Imaeda T, Nakada TA, Takahashi N, et al. Trends in the incidence and outcome of sepsis using data from a Japanese nationwide medical claims database-the Japan Sepsis Alliance (JaSA) study group. Crit Care. 2021;25:338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [25].Luzzaro F, Viganò EF, Fossati D, et al. ; AMCLI Lombardia Hospital Infectious Study Group. Prevalence and drug susceptibility of pathogens causing bloodstream infections in northern Italy: a two-year study in 16 hospitals. Eur J Clin Microbiol Infect Dis. 2002;21:849–55. [DOI] [PubMed] [Google Scholar]
  • [26].Sligl W, Taylor G, Brindley PG. Five years of nosocomial Gram-negative bacteremia in a general intensive care unit: epidemiology, antimicrobial susceptibility patterns, and outcomes. Int J Infect Dis. 2006;10:320–5. [DOI] [PubMed] [Google Scholar]
  • [27].Mazzariol A, Bazaj A, Cornaglia G. Multi-drug-resistant Gram-negative bacteria causing urinary tract infections: a review. J Chemother. 2017;29:2–9. [DOI] [PubMed] [Google Scholar]
  • [28].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;22:281–6. [DOI] [PubMed] [Google Scholar]
  • [29].Matsumura Y, Johnson JR, Yamamoto M, et al. ; Kyoto–Shiga Clinical Microbiology Study Group. CTX-M-27- and CTX-M-14-producing, ciprofloxacin-resistant Escherichia coli of the H30 subclonal group within ST131 drive a Japanese regional ESBL epidemic. J Antimicrob Chemother. 2015;70:1639–49. [DOI] [PubMed] [Google Scholar]
  • [30].Castanheira M, Farrell SE, Krause KM, Jones RN, Sader HS. Contemporary diversity of β-lactamases among enterobacteriaceae in the nine U.S. census regions and ceftazidime-avibactam activity tested against isolates producing the most prevalent β-lactamase groups. Antimicrob Agents Chemother. 2014;58:833–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Abe T, Iriyama H, Imaeda T, et al. Epidemiology and patterns of empiric antimicrobial therapy practice in patients with community-onset sepsis using data from a Japanese nationwide medical claims database – the Japan Sepsis Alliance (JaSA) study group. IJID Reg. 2024;10:162–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].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. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Pfaller MA, Jones RN, Doern GV, Kugler K, Group TSP. Bacterial pathogens isolated from patients with bloodstream infection: frequencies of occurrence and antimicrobial susceptibility patterns from the SENTRY antimicrobial surveillance program (United States and Canada, 1997). Antimicrob Agents Chemother. 1998;42:1762–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Paul M, Benuri-Silbiger I, Soares-Weiser K, Leibovici L. β lactam monotherapy versus β lactam-aminoglycoside combination therapy for sepsis in immunocompetent patients: systematic review and meta-analysis of randomised trials. BMJ. 2004;328:668. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [35].Chatzinikolaou I, Abi-Said D, Bodey GP, Rolston KVI, Tarrand JJ, Samonis G. Recent experience with pseudomonas aeruginosa bacteremia in patients with cancer: retrospective analysis of 245 episodes. Arch Intern Med. 2000;160:501–9. [DOI] [PubMed] [Google Scholar]
  • [36].Kuikka A, Valtonen VV. Factors associated with improved outcome of pseudomonas aeruginosa bacteremia in a Finnish university hospital. Eur J Clin Microbiol Infect Dis. 1998;17:701–8. [DOI] [PubMed] [Google Scholar]
  • [37].Chamot E, Boffi El Amari E, Rohner P, Van Delden C. Effectiveness of combination antimicrobial therapy for pseudomonas aeruginosa bacteremia. Antimicrob Agents Chemother. 2003;47:2756–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [38].Phe K, Bowers DR, Babic JT, Tam VH. Outcomes of empiric aminoglycoside monotherapy for Pseudomonas aeruginosa bacteremia. Diagn Microbiol Infect Dis. 2019;93:346–8. [DOI] [PubMed] [Google Scholar]
  • [39].Anastasia A, Bonura S, Rubino R, et al. The use of intravenous fosfomycin in clinical practice: a 5-year retrospective study in a tertiary hospital in Italy. Antibiotics (Basel, Switzerland). 2023;12:971. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [40].Sojo-Dorado J, López-Hernández I, Rosso-Fernandez C, et al. ; REIPI-GEIRAS-FOREST group. Effectiveness of fosfomycin for the treatment of multidrug-resistant escherichia coli bacteremic urinary tract infections: a randomized clinical trial. JAMA Netw Open. 2022;5:e2137277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [41].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;43:1130–9. [DOI] [PubMed] [Google Scholar]
  • [42].Leeper TJ, Arnold J, Arel-Bundock V. Margins: marginal effects for model objects. R Package Version 0.3; 2018.
  • [43].Team RC. R: A Language and Environment for Statistical Computing. 2019. https://www.R-project.org. Accessed May 1, 2025.
  • [44].Safdar N, Handelsman J, Maki DG. Does combination antimicrobial therapy reduce mortality in Gram-negative bacteraemia? A meta-analysis. Lancet Infect Dis. 2004;4:519–27. [DOI] [PubMed] [Google Scholar]
  • [45].Sjövall F, Perner A, Hylander Møller M. Empirical mono- versus combination antibiotic therapy in adult intensive care patients with severe sepsis – a systematic review with meta-analysis and trial sequential analysis. J Infect. 2017;74:331–44. [DOI] [PubMed] [Google Scholar]
  • [46].Brunkhorst FM, Oppert M, Marx G, et al. ; German Study Group Competence Network Sepsis (SepNet). Effect of empirical treatment with moxifloxacin and meropenem vs meropenem on sepsis-related organ dysfunction in patients with severe sepsis. JAMA. 2012;307:2390–9. [DOI] [PubMed] [Google Scholar]
  • [47].Damas P, Garweg C, Monchi M, et al. Combination therapy versus monotherapy: a randomised pilot study on the evolution of inflammatory parameters after ventilator associated pneumonia [ISRCTN31976779]. Crit Care. 2006;10:R52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [48].Leroy O, Saux P, Bédos J-P, Caulin E. Comparison of levofloxacin and cefotaxime combined with ofloxacin for ICU patients with community-acquired pneumonia who do not require vasopressors. Chest. 2005;128:172–83. [DOI] [PubMed] [Google Scholar]
  • [49].Heyland DK, Dodek P, Muscedere J, Day A, Cook D; Canadian Critical Care Trials Group. Randomized trial of combination versus monotherapy for the empiric treatment of suspected ventilator-associated pneumonia*. Crit Care Med. 2008;36:737–44. [DOI] [PubMed] [Google Scholar]
  • [50].Matsumura Y, Yamamoto M, Nagao M, et al. ; Kyoto-Shiga Clinical Microbiology Study Group. Emergence and spread of B2-ST131-O25b, B2-ST131-O16 and D-ST405 clonal groups among extended-spectrum- -lactamase-producing Escherichia coli in Japan. J Antimicrob Chemother. 2012;67:2612–20. [DOI] [PubMed] [Google Scholar]
  • [51].Kakuta N, Nakano R, Nakano A, et al. Molecular characteristics of extended-spectrum β-lactamase-producing Klebsiella pneumoniae in Japan: predominance of CTX-M-15 and emergence of hypervirulent clones. Int J Infect Dis. 2020;98:281–6. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

medi-105-e47024-s001.docx (43.9KB, docx)

Articles from Medicine are provided here courtesy of Wolters Kluwer Health

RESOURCES