Use of piperacillin/tazobactam and meropenem in a Danish intensive care unit

Severe infections, including sepsis, are life-threatening conditions that often require admission to the intensive care unit (ICU) and are associated with high mortality, morbidity and healthcare costs [1, 2]. Early empirical treatment with broad-spectrum antibiotics is recommended in these patients [3], and approximately three in every four ICU patients receive antimicrobial agents during their ICU stay [4]. However, excessive use of broad-spectrum antimicrobial agents contributes to the development of antibiotic resistance [5].

The World Health Organization (WHO) has designated antibiotic resistance as one of the greatest threats against global health [6]. In the past decade, the use of broad-spectrum antibiotics such as piperacillin/tazobactam and meropenem has increased globally [7-9], including in Denmark [10]. Piperacillin/tazobactam may be preferred as an empirical agent over meropenem as the use of carbapenems is associated with selection of carbapenemase-producing bacteria [11-13]. However, a recently published systematic review suggested that piperacillin/tazobactam may be associated with less favourable outcomes than carbapenems in patients with severe bacterial infections, although the certainty of the evidence was low to very low [14]. Some included trials focused exclusively on bacteremia with extended-spectrum ß-lactamase (ESBL) producing bacteria [12, 13]. Therefore, randomised clinical trials are warranted to assess the effects of piperacillin/tazobactam versus carbapenems on mortality, serious adverse events and the occurrence of resistant bacteria in general [12, 14].

As part of a research programme preparing for the Empirical Meropenem vs. Piperacillin/Tazobactam for Adult Patients with Sepsis (EMPRESS) trial, we described the use of piperacillin/tazobactam and meropenem in all ICU patients admitted to a Danish university hospital ICU during 12 consecutive weeks.

Methods

Study design

We prospectively screened all patients admitted to the Department of Intensive Care at Copenhagen University Hospital - Rigshospitalet, Denmark, from 1 November 2022 to 23 January 2023. This general 24-bed ICU provides Critical Care Level 3 (advanced respiratory support, monitoring and support for two or more organs) for surgical and medical patients, including haematology and oncology patients and children aged more than one year. The hospital has specialised ICUs for cardiothoracic/cardiac, neurosurgical/neurological and neonatal cases.

Because of its observational and descriptive design, the study was approved as a quality control study. The head of the ICU approved the protocol, and no consent was required.

Participants/patients

Patients were eligible for inclusion if they received piperacillin/tazobactam and/or meropenem during their ICU stay. They were only included once in the study. We did include patients who had not been included during their primary admission if they met the eligibility criteria during their readmission (within the 12 weeks).

Data collection

Data were collected using RedCap, including demographics, baseline characteristics, daily use of piperacillin/tazobactam and meropenem, life support use, length of ICU stay and mortality. The complete list of variables is presented in the Supplementary File. Trained research staff performed data entry, and one author (HJ) validated the data. If any questions emerged during data entry, the responsible intensivist or co-author MHM or AP was counselled.

Outcomes

The primary outcome was 90-day mortality. Secondary outcomes were discharging to a ward or another ICU, duration of ICU stay, duration of antibiotic treatment and occurrence of antibiotic-resistant bacteria during index hospital admission, including a positive culture with methicillin-resistant Staphylococcus aureus, carbapenemase-producing organisms (CPO), ESBL, vancomycin-resistant enterococci (VRE) or other bacteria with acquired resistance.

Statistical methods

We analysed the data using R version 4.1.0 (R Core Team, R Foundation for Statistical Computing, Vienna, Austria). We present data descriptively, i.e. as numbers (%) with 95% confidence intervals (CI) for categorical data and medians with interquartile ranges (IQR) for continuous data. We stratified the data using piperacillin/tazobactam versus meropenem versus both agents during the ICU stay.

Trial registration: not relevant.

Results

Demographic and clinical characteristics of patients

A total of 286 patients were admitted to the ICU during the study period. Among these, we included 184 patients (64.3%) who received piperacillin/tazobactam and/or meropenem; 112 (61.9%) were men (Figure 1). The population had a median age of 63.3 (IQR: 48.0-71.3) years. Acute kidney injury (28.8%) was the most common coexisting condition at baseline. Most patients, 161 (87.5%), received at least one life-supportive intervention. During their ICU stay, a total of 80 (43.5%) patients received piperacillin/tazobactam, 76 (41.3%) received meropenem and 28 (15.2%) received both within 90 days of their ICU stay (Table 1). A total of 29 (38.2%) of the patients receiving meropenem had suspected skin and/or soft tissue infection at baseline. The most common focus of infection in the other treatment groups was pulmonary (32.5% of patients treated with piperacillin/tazobactam and 39.3% among patients receiving both agents during admission). More patients treated with meropenem received invasive mechanical ventilation (75.0%), vasopressors (86.6%) and renal replacement therapy (25.0%) than patients treated with piperacillin/tazobactam (52.5%, 72.5% and 16.3%, respectively) (Table 1).

Patients who received both agents during their ICU stay received these sequentially, i.e. they received one of the two agents at inclusion, which was hereafter changed to the other (typically piperacillin/tazobactam first). The most common reason for switching from piperacillin/tazobactam to meropenem was a need for extended-spectrum coverage (46.4%) (Table 1). Patients who received both antibiotics during their ICU stay had higher age, Simplified Acute Physiology Score III (SAPS-3), were more often transferred from another ICU and had a longer ICU stay than patients only receiving piperacillin/tazobactam or meropenem during their ICU stay (Table 1).

Most patients (71.2%) received the antibiotics upon ICU admission; 69 (37.5%) received piperacillin/tazobactam, 60 (32.6%) meropenem and two (1.1%) both agents. The two patients who received both agents upon ICU admission were adjusted to monotherapy soon after their arrival at the ICU. None of the patients treated with piperacillin/tazobactam upon ICU admission had received meropenem within 24 hours before receiving piperacillin/tazobactam. Twenty-four (40.0%) of the patients administered meropenem upon ICU admission had received piperacillin/tazobactam within 24 hours before receiving meropenem (Table 1). The most prescribed dose of piperacillin/tazobactam, pre-ICU and in the ICU, was 4/0.5 g four times daily. The most prescribed dose of meropenem pre-ICU was 2 g three times daily and 1 g four times daily in the ICU (Table 2).

Primary outcome

A total of 182 patients were included in the primary outcome analysis; two patients in the meropenem group were lost to follow-up at day 90 due to foreign residency. The 90-day mortality appeared similar between the groups: 19 (25.7%) (95% CI: 16.2-37.2%) in patients treated with meropenem, 22 (27.5%) (95% CI: 18.1-38.6%) in patients treated with piperacillin/tazobactam and eight (28.6%) (95% CI: 13.2-48.7%) in patients who received both agents upon ICU admission died within 90-days (Table 3).

Secondary outcome treatment duration

The duration of antibiotic treatment in the ICU and in total for patients treated with meropenem was five (IQR: 3-11) days and 12 (IQR: 8-18) days, respectively. The corresponding results for those treated with piperacillin/tazobactam were three (IQR: 2-5) days and six (IQR: 4-10) days, respectively (Table 3).

Antibiotic resistance

At baseline, eight patients (4.3%) had a known colonisation with a bacterium with acquired antibiotic resistance; six (3.3%) of these were VRE, one (0.5%) was CPO and one (0.5%) was methicillin-resistant S. aureus. Nineteen cases of a new positive culture with a bacterium with acquired antibiotic resistance were identified in 17 patients (9.2% of our study population). Among these, VRE was the most common pathogen (Table 3); where identification was possible, most patients had these identified in hospital after ICU discharge (7 (3.8%)). There appeared to be no major differences in resistance patterns between the three antibiotic groups (Table 3). See Table 3 for a full report on secondary outcomes.

Discussion

In this single-centre cohort study with prospective data collection, we observed that more than half of the general ICU patients received piperacillin/tazobactam and/or meropenem during their ICU stay. The 90-day mortality and occurrence of antibiotic-resistant bacteria appeared similar between the groups.

In our report, piperacillin/tazobactam was the preferred antibiotic treatment before ICU admission. Patients administered meropenem upon ICU admission were more likely to have previously received piperacillin/tazobactam within 24 hours. In contrast, none of the patients treated with piperacillin/tazobactam had received meropenem within the same timeframe before receiving piperacillin/tazobactam. This may indicate that clinicians were more restrictive in prescribing meropenem than piperacillin/tazobactam. Restrictive use of carbapenems as a last resort antimicrobial drug is in line with the guidelines on antibiotic use issued by the Danish health authorities [15].

Slightly less than one in every ten patients were identified with a new positive culture with a bacterium with acquired antibiotic resistance during their hospital stay, most of which were VRE. Infections with Enterococcus are often caused by E. faecalis or E. faecium, the latter having limited susceptibility to most antibiotics, including piperacillin/tazobactam and meropenem. This gives E. faecium a selective advantage in healthcare settings [10]. Thus, piperacillin/tazobactam and meropenem do not a priori induce vancomycin resistance, but they select for E. faecium. Furthermore, E. faecium has a high prevalence of acquired vancomycin resistance due to use and overuse of vancomycin, contributing to VRE selection [16, 17]. The incidence of VRE has increased in Denmark since 2002 [18]. In 2022, vancomycin resistance was found in 12.0% of invasive human E. faecium isolates [10]. In total, 22 (12.0%) patients were colonised or infected with VRE in our cohort. Our findings are, therefore, expected and in line with the previously documented incidence [10].

The study period coincided with an increased incidence of invasive Group A streptococcus infections [19]. In severe cases, infection with this bacterium may progress to necrotising soft tissue infection (NSTI). Copenhagen University Hospital - Rigshospitalet is national referral for NSTI. During the study period, we experienced an increase in NSTIs. Although we did not register the primary reason for ICU admission, the outbreak is reflected in the high proportion of patients with skin and soft tissue infections. In Denmark, meropenem, in combination with clindamycin, is the recommended empirical treatment for NSTI. The standard empirical dose of meropenem for NSTI is 2 g three times daily [20], which was the most prescribed dose of meropenem before ICU admission. Thus, meropenem may have been used more frequently and in higher doses during the study period than in general.

Patients treated with meropenem were younger, more likely to require life-support at baseline, and antibiotic treatment duration and length of ICU stay was longer for these patients than for patients treated with piperacillin/tazobactam. Piperacillin/tazobactam was more often used for patients who had undergone elective surgery, and patients treated with piperacillin/tazobactam represented the highest proportion of patients who did not use any type of life-support during their first 24 hours in ICU. This may explain some of the observed differences in disease severity, duration of antibiotic treatment and length of stay. Despite these differences, mortality appeared similar between the groups. While our study was descriptive, it has previously been reported that piperacillin/tazobactam may be associated with less favourable outcomes than carbapenems in patients with severe bacterial infections. However, the certainty of evidence was low to very low [14]. While our findings cannot be used to infer causality, they support the need for relevant trials assessing this question.

Strengths and limitations

We included patients from a large, general ICU with mixed medical and surgical beds for 12 consecutive weeks with 90-day follow-up. Data were entered manually and validated in a dedicated database by trained research staff. ICU specialists (co-authors) were available for counselling to minimise risk of data entry errors. There are, however, also limitations. First, this was a single-centre observational study conducted within a limited period; therefore, the sample size was limited. Although our cohort represents a wide range of ICU patients, it may not be directly representative of ICUs with a different case mix. We did not register the reason(s) for termination of antibiotic treatment, which could have contributed to further knowledge about antibiotic escalation and de-escalation in the ICU. We were unable to control for confounders.

Our findings will inform the design of the upcoming EMPRESS trial, ClinicalTrial.gov identifier NCT06184659.

Conclusions

In this single-centre cohort study with prospective data collection, we observed that more than half of the general ICU patients received piperacillin/tazobactam and/or meropenem during their ICU stay, mainly as empirical treatment. Mortality and the occurrence of resistant bacteria with new acquired resistance, mainly VRE, appeared to the same extent between the groups.