Publication

• Title: Continuous vs Intermittent Meropenem Administration in Critically Ill Patients With Sepsis: The MERCY Randomized Clinical Trial
• Acronym: MERCY
• Year: 2023
• Journal published in: JAMA
• Citation: Monti G, Bradić N, Marzaroli M, et al; for the MERCY Investigators. Continuous vs intermittent meropenem administration in critically ill patients with sepsis: the MERCY randomized clinical trial. JAMA. 2023;330(2):141-151.

Context & Rationale

• Background

  • Meropenem is a time-dependent β-lactam; efficacy is linked to maintaining free concentrations above the pathogen MIC for a sufficient proportion of the dosing interval.
  • Critical illness produces marked pharmacokinetic variability (augmented renal clearance, altered volume of distribution, organ dysfunction, renal replacement therapy), increasing the risk of underexposure with standard intermittent dosing.
  • Sepsis guidelines suggest prolonged infusion of β-lactams (after a loading dose) as a pragmatic strategy to improve PK/PD target attainment, despite ongoing uncertainty about patient-important outcomes.¹
  • Prior meta-analyses of prolonged/continuous β-lactam infusion reported lower mortality versus intermittent infusion in randomised trials (e.g., RR ~0.70; 95% CI 0.56 to 0.87), but the evidence base was heterogeneous and often at risk of bias.²³
• Research Question/Hypothesis

  • In adult ICU patients with sepsis or septic shock for whom clinicians had decided to start meropenem, would continuous administration (following a standardised loading dose) reduce the composite of 28-day all-cause mortality or emergence of new PDR/XDR organisms, compared with intermittent administration delivering the same total daily dose?
• Why This Matters

  • Continuous infusion increases complexity (infusion pumps, nursing workload, drug stability constraints) and requires outcome-level evidence to justify routine adoption.
  • Meropenem is widely used empirically in ICU sepsis; even modest improvements in mortality or prevention of highly resistant organisms would have major downstream impact.
  • A high-quality meropenem-specific RCT could clarify whether any benefits seen in earlier heterogeneous β-lactam literature apply to contemporary ICU practice.

Design & Methods

• Research Question: Among critically ill adults with sepsis/septic shock receiving meropenem, does continuous administration (vs intermittent administration) reduce a composite of 28-day mortality or emergence of new PDR/XDR organisms?
• Study Type: Randomised, multicentre, investigator-initiated, international, double-blind, double-dummy, parallel-group superiority trial; 31 ICUs across 26 hospitals in 4 countries; enrolment June 2018 to August 2022; ICU setting; trial registration NCT03452839.
• Population:
  • Inclusion: adults (≥18 years) in ICU with sepsis or septic shock; suspected/proven infection plus ≥2 SIRS criteria and SOFA score ≥2; clinicians had prescribed a new course of meropenem expected to start within 24 hours after randomisation; anticipated ICU stay >48 hours.
  • Exclusion: refusal/withdrawal of consent; known allergy to carbapenems; meropenem within 48 hours before screening; Gram-negative pathogen known to be meropenem resistant at enrolment; SAPS II score ≥65; AIDS; immunosuppressive therapy or long-term corticosteroids.
• Intervention:
  • All participants: a loading dose of 1 g meropenem.
  • Continuous administration: meropenem infused continuously to deliver the protocol dose (typically 3 g per 24 hours when creatinine clearance >50 mL/min; dose reduction for creatinine clearance 10–50 mL/min; dose escalation permitted in selected cases); double-dummy intermittent placebo boluses given to maintain blinding.
• Comparison:
  • All participants: a loading dose of 1 g meropenem.
  • Intermittent administration: meropenem delivered as bolus infusions over 30–60 minutes with the same total daily dose as the continuous group (e.g., 1 g every 8 hours when creatinine clearance >50 mL/min; renal-adjusted schedules; first 24 hours used a more frequent schedule in those with preserved renal function); double-dummy continuous placebo infusion given to maintain blinding.
• Blinding: Double-blind, double-dummy (patients, treating clinicians, investigators, and outcome assessors blinded; site pharmacy and designated ICU research staff prepared masked study drug).
• Statistics: A total of 600 patients (300 per group) were planned to detect a 12% absolute reduction in the primary composite outcome (from 52% to 40%) with >80% power at the 5% significance level; primary analysis was intention-to-treat with effect estimates reported as risk ratios with 95% confidence intervals (secondary outcomes reported without multiplicity adjustment).
• Follow-Up Period: Primary endpoint at 28 days after first study-drug bolus; mortality follow-up to 90 days.

Key Results

This trial was not stopped early. It completed enrolment (n=607; 303 continuous vs 304 intermittent) with primary outcome ascertainment in the full intention-to-treat population.

• The point estimate for the primary composite favoured continuous administration, but the confidence interval included clinically important benefit and harm (RR 0.96; 95% CI 0.81 to 1.13).
• Mortality was numerically lower at day 28 with continuous administration (30% vs 33%) but identical at day 90 (42% vs 42%).
• Prespecified subgroup analyses did not identify a clear treatment-responsive phenotype; for example, acute kidney injury at randomisation: 36/86 vs 54/98 (RR 0.76; 95% CI 0.56 to 1.03; interaction P=0.071), and pathogens with high carbapenem MIC: 56/92 vs 49/82 (RR 1.02; 95% CI 0.80 to 1.30; interaction P=0.55).

Internal Validity

• Randomisation and Allocation
  • Central web-based randomisation with concealment until assignment; stratified by study site.
  • Double-dummy preparation by pharmacy/research personnel reduced allocation disclosure to treating teams.
• Drop out or exclusions
  • Intention-to-treat population included all randomised patients (n=607) for the primary endpoint.
  • Per-protocol analysis excluded 12 participants meeting exclusion criteria post-randomisation (7 continuous vs 5 intermittent) and showed similar results for the primary endpoint (RR 0.97; 95% CI 0.82 to 1.15).
• Performance/Detection Bias
  • Double-blind, double-dummy design minimised performance bias for co-interventions and escalation decisions.
  • Mortality is objective; resistant organism emergence is culture-based and prespecified.
• Protocol Adherence
  • Prompt delivery: median time randomisation-to-first meropenem dose 7 minutes (IQR 0–14) vs 7 minutes (IQR 0–15).
  • Crossover was rare (1 patient [0.3%] received intermittent instead of continuous administration).
• Baseline Characteristics
  • Severity and support were balanced: septic shock 62% vs 60%; SOFA median 9 vs 9; SAPS II median 44 vs 43; vasopressor use 62% vs 61%; invasive airway (tracheal tube/tracheostomy) 74% vs 78%.
  • Timing and prior exposure were similar: hospital-to-randomisation 9 days (IQR 4–19) vs 8 days (IQR 3–18); ICU-to-randomisation 5 days (IQR 2–12) vs 5 days (IQR 2–11); antibiotics within prior 3 months 67% vs 65%.
• Timing
  • Intervention separation began after a standardised 1 g loading dose, with protocolised continuous vs intermittent administration thereafter.
  • Randomisation occurred several days into ICU stay (median 5 days), which reflects typical ICU prescribing but may reduce sensitivity to detect early-phase benefits.
• Dose
  • Dose distribution was similar: standard 3 g/day 62% vs 64%; low 2 g/day 24% vs 28%; high 6 g/day 11% vs 6.9%.
• Separation of the Variable of Interest
  • Administration strategy clearly differed by design (continuous vs intermittent infusion) with matching total daily dose and mandatory loading dose.
  • Delivered exposure proxy measures were similar: median treatment duration 11 days (IQR 7–15) vs 11 days (IQR 7–15); median total dose 24 g (IQR 15–42) vs 21 g (IQR 15–36); concomitant antibiotics 73% vs 74%.
• Outcome Assessment
  • Blood cultures were obtained in 95% vs 94% at baseline, with positivity 23% vs 20%.
  • Resistance component required identification of new PDR/XDR organisms by day 28 and excluded those dying within 48 hours (15 vs 24) from the resistance denominator.
• Statistical Rigor
  • Planned sample size was achieved (n=607); primary analysis was intention-to-treat with risk ratios and 95% confidence intervals.
  • Secondary outcomes were reported without multiplicity adjustment.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given robust allocation concealment, double-blind double-dummy delivery, minimal crossover, and complete intention-to-treat ascertainment; the main internal vulnerability is that the resistance component depends on culture ascertainment and prespecified denominator handling for early deaths.

External Validity

• Population Representativeness
  • Represents a high-acuity ICU sepsis cohort: septic shock in ~61% and invasive airway in ~76% at randomisation.
  • Restricted by exclusions (SAPS II ≥65; AIDS; immunosuppression/long-term corticosteroids; known meropenem resistance at enrolment), limiting applicability to the sickest and immunocompromised populations.
• Applicability
  • Directly applicable to ICUs able to operationalise continuous infusion workflows (pharmacy preparation, infusion pump availability, and stability governance).
  • Comparator was intermittent infusion over 30–60 minutes with a mandated 1 g loading dose; applicability to centres using extended intermittent infusions (e.g., 3–4 hours) or different dosing protocols is uncertain.

Conclusion on External Validity: Generalisability is good for high-resource ICUs treating severe sepsis/septic shock with meropenem using standard dosing pathways, but is limited for immunosuppressed cohorts, extreme severity (excluded by SAPS II), and settings with materially different dosing/infusion infrastructure.

Strengths & Limitations

• Strengths:
  • Large, international ICU RCT (n=607) with double-blind, double-dummy design.
  • Protocolised dosing with a universal loading dose and matched total daily dose between groups.
  • Clinically meaningful outcomes including a stewardship-relevant resistance endpoint and 90-day mortality follow-up.
  • High treatment separation with minimal crossover and timely initiation after randomisation (median 7 minutes).
• Limitations:
  • Primary composite endpoint combines mortality with culture-defined emergence of PDR/XDR organisms, introducing heterogeneity in component meaning and ascertainment.
  • No pharmacokinetic sampling or therapeutic drug monitoring to confirm PK/PD separation (e.g., fT>MIC) between strategies.
  • Patients were randomised a median of 5 days into ICU admission and 9 days into hospitalisation, after substantial prior antibiotic exposure (≈66%).
  • Exclusion of the highest SAPS II scores and immunosuppressed patients constrains generalisability to those groups.

Interpretation & Why It Matters

• Clinical effectiveness

  Routine continuous infusion of meropenem (after a 1 g loading dose) did not improve the primary composite endpoint or mortality compared with intermittent infusion when the total daily dose was matched.
• Resistance/stewardship signal

  No reduction was observed in emergence of new PDR/XDR organisms by day 28 (24% vs 25%; RR 0.94; 95% CI 0.71 to 1.26), arguing against a broad resistance-prevention effect of continuous infusion in an unselected ICU sepsis cohort.
• Implementation implications

  Given the neutral outcome and the operational burden of continuous infusion workflows, MERCY does not support default continuous infusion of meropenem for all ICU sepsis patients; it does not answer whether PK-guided, phenotype-targeted strategies (e.g., measured underexposure) are beneficial.

Controversies & Subsequent Evidence

• Composite endpoint construction
  • The choice to combine mortality with emergence of PDR/XDR organisms was intended to address both patient outcomes and stewardship, but creates interpretive complexity when components have different causal pathways and sampling dependencies.⁴
• PK/PD separation and mechanism
  • Continuous infusion benefits are mechanistically predicated on improved target attainment; MERCY did not include drug concentration measurements, limiting inference about whether neutral clinical outcomes reflect lack of PK separation or lack of outcome responsiveness to PK optimisation in this setting.⁴
  • MERCY used a universal loading dose and a comparatively optimised intermittent regimen (30–60 minute infusions and protocolised early dosing), which plausibly reduces incremental benefit attributable to continuous infusion versus historical “short bolus” comparators.
• Relationship to prior and subsequent evidence
  • Earlier meta-analyses (including randomised trials with heterogeneous designs and variable blinding) reported mortality reductions with prolonged/continuous infusion, contrasting with the neutral meropenem-specific MERCY result.²³
  • BLING III (JAMA 2024) evaluated continuous versus intermittent infusion across multiple β-lactam agents in ICU sepsis, addressing a potential class effect and complementing meropenem-specific data from MERCY.⁵
  • An updated systematic review and Bayesian meta-analysis in JAMA (2024) synthesised prolonged versus intermittent β-lactam infusion trials in sepsis/septic shock and provides the most contemporary pooled context in which to interpret MERCY’s neutral effect estimate.⁶
  • Guideline suggestions to consider prolonged infusion of β-lactams (after a loading dose) predate MERCY and were based on mixed-quality evidence; MERCY adds high-quality meropenem-specific evidence to inform future guideline updates.¹

Summary

• MERCY randomised 607 ICU patients with sepsis/septic shock prescribed meropenem to continuous vs intermittent administration in a double-blind, double-dummy design with a mandated 1 g loading dose.
• The primary composite (28-day death or new PDR/XDR emergence) did not differ: 47% vs 49% (RR 0.96; 95% CI 0.81 to 1.13; P=0.60).
• Mortality was similar at both 28 days (30% vs 33%) and 90 days (42% vs 42%).
• Emergence of new PDR/XDR organisms by day 28 was similar (24% vs 25%; RR 0.94; 95% CI 0.71 to 1.26).
• No study drug–related seizures or allergic reactions were reported.

Overall Takeaway

MERCY is a large, double-blind, double-dummy RCT directly testing whether continuous meropenem administration improves patient-centred and stewardship-relevant outcomes in ICU sepsis. Despite a strong mechanistic rationale, continuous infusion did not reduce the composite of 28-day mortality or new PDR/XDR emergence, nor did it improve 90-day mortality, when compared with an optimised intermittent regimen with a mandated loading dose.

Bibliography

• 1Evans 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(11):1181-1247.
• 2Vardakas KZ, Voulgaris GL, Maliaros A, et al. Prolonged versus short-term intravenous infusion of antipseudomonal β-lactams for the treatment of patients with sepsis: a systematic review and meta-analysis of randomised trials. Lancet Infect Dis. 2018;18(1):108-120.
• 3Roberts JA, Abdul-Aziz MH, Davis JS, et al. Continuous versus intermittent β-lactam infusion in severe sepsis: a meta-analysis of individual patient data from randomised trials. Am J Respir Crit Care Med. 2016;194(6):681-691.
• 4Shappell E, Klompas M, Rhee C. Do prolonged infusions of β-lactam antibiotics improve outcomes in critically ill patients with sepsis? JAMA. 2023;330(2):126-128.
• 5Dulhunty JM, et al; for the BLING III Investigators. Continuous vs intermittent β-lactam antibiotic infusions in critically ill patients with sepsis: a randomized clinical trial. JAMA. 2024;332(8):629-637.
• 6Abdul-Aziz MH, et al. Prolonged vs intermittent infusions of β-lactam antibiotics in adults with sepsis or septic shock: a systematic review and Bayesian meta-analysis. JAMA. 2024;332(8):638-648.