The research team, led by investigators from Zhongda Hospital and the Jiangsu Provincial Key Laboratory of Critical Care Medicine at Southeast University in Nanjing, China, scoured four major databases—PubMed, EMBASE, Web of Science, and the Cochrane Library—for randomized controlled trials published between January 2000 and July 2025. After screening 2,516 records and reviewing 191 full-text articles, they identified 25 eligible randomized trials encompassing 9,915 critically ill adults. The trials compared three intravenous administration strategies: continuous infusion, defined as an uninterrupted 24-hour drip typically preceded by a loading dose; extended infusion, in which each dose is delivered over three to four hours, roughly 40 to 50 percent of the dosing interval; and intermittent infusion, the conventional approach in which each dose is pushed or infused over an hour or less. Strikingly, despite decades of clinical experience with both prolonged strategies, not a single trial directly pitted continuous infusion against extended infusion in a head-to-head comparison.

The pharmacological rationale behind prolonged infusion is elegant. Beta-lactams are time-dependent bactericidal agents: their killing power depends on the percentage of time that free, unbound drug concentrations remain above the pathogen’s MIC, a metric pharmacologists abbreviate as %fT greater than MIC. Conventional intermittent dosing produces high peak concentrations, but those levels plummet between doses, and in septic patients—whose fluid shifts, organ dysfunction, and augmented renal clearance dramatically alter drug distribution and elimination—concentrations can dip into subtherapeutic territory for dangerous stretches. Prolonging the infusion, whether continuously or over several hours per dose, keeps drug levels above the killing threshold for a greater fraction of the dosing interval, suppressing bacterial regrowth and improving the odds of infection control.

Yet the two prolonged strategies differ in ways that matter enormously at the bedside. Continuous infusion typically demands a dedicated intravenous line, because beta-lactams can be chemically incompatible with other continuously infused drugs, forcing clinicians to reserve a lumen solely for the antibiotic. Many beta-lactams—meropenem being the notorious example—are chemically unstable at room temperature over 24 hours, requiring frequent replacement of infusion bags and driving up nursing workload, catheter utilization, and costs. Extended infusion, by contrast, piggybacks on existing central or peripheral venous access, preserves conventional dosing intervals, and frees the line between doses for other medications. These practical considerations explain why extended infusion has been adopted more widely in intensive care units worldwide, even as pharmacodynamic theory favored the steadier exposure of continuous infusion.

The new analysis, registered prospectively in PROSPERO and conducted according to PRISMA guidelines for network meta-analysis, used Markov chain Monte Carlo methods in the R statistical environment to weave together direct and indirect evidence across the trial network. Two reviewers independently screened studies, extracted data, and assessed risk of bias using the Cochrane tool, with publication bias evaluated through funnel plots and Egger’s test. Certainty of evidence was graded with the GRADE framework adapted for network meta-analysis, and a trial sequential analysis was performed to gauge whether the accumulated evidence was sufficient to rule out random error.

The headline findings are nuanced. Against intermittent infusion, continuous infusion was associated with a statistically significant reduction in all-cause mortality, with a relative risk of 0.85 and a 95 percent confidence interval of 0.66 to 0.98, rated as moderate certainty evidence. In absolute terms, that translates to roughly 39 fewer deaths per 1,000 patients. Extended infusion also showed a mortality benefit relative to intermittent dosing—a relative risk of 0.80—but with a confidence interval spanning 0.60 to 1.10, the effect was statistically uncertain and carried only low certainty. Both prolonged strategies significantly improved clinical cure rates compared with intermittent infusion: continuous infusion achieved a relative risk of 1.10 (95 percent CI 1.03 to 1.20), while extended infusion posted a more pronounced relative risk of 1.28 (95 percent CI 1.07 to 1.55), both at moderate certainty. Neither strategy, however, demonstrated a clear effect on intensive care unit length of stay, with wide confidence intervals straddling the null for both comparisons.

The pivotal question—continuous versus extended infusion—could only be answered indirectly, since no direct trials exist. The network meta-analysis found no statistically significant differences between the two strategies in mortality (relative risk 1.06, 95 percent CI 0.73 to 1.44), clinical cure (relative risk 0.86, 95 percent CI 0.70 to 1.05), or ICU length of stay (mean difference 0.86 days, 95 percent CI -3.32 to 4.54). But the Bayesian posterior probability analysis added a provocative twist: there was a 93.2 percent probability that extended infusion achieves higher clinical cure rates than continuous infusion, and a 63.7 percent probability that extended infusion is superior for reducing mortality. The authors suggest the stronger cure signal may reflect more reliable drug exposure with extended infusion, particularly for carbapenems whose 24-hour stability is questionable—degradation during prolonged room-temperature infusion could silently undermine continuous regimens.

The analysis confronted several methodological challenges head-on. The massive BLING III trial, published in JAMA in 2024, contributed approximately 70 percent of the total study population, raising concerns that a single study could dominate the pooled estimates. A prespecified leave-one-out sensitivity analysis excluding BLING III nonetheless preserved the core findings: continuous infusion retained its mortality advantage over intermittent infusion (relative risk 0.78, 95 percent credible interval 0.60 to 0.96) and its clinical cure benefit (relative risk 1.09, 95 percent credible interval 1.01 to 1.23), while the indirect continuous-versus-extended comparisons remained nonsignificant. Trial sequential analysis confirmed that the cumulative evidence for the continuous-versus-intermittent mortality comparison crossed the required information size and statistical significance boundary, whereas the extended-versus-intermittent mortality estimate never reached conventional significance, underscoring the lingering uncertainty around that strategy’s survival effect.

The authors are careful to frame their conclusions with appropriate humility. Most included trials were open-label, exposing patients and clinicians to performance bias, which forced downgrades in certainty of evidence. Outcome definitions varied considerably—mortality was captured at timepoints ranging from ICU discharge to 90 days, and clinical cure was defined differently across studies. Pharmacokinetic heterogeneity among critically ill patients, driven by renal replacement therapy, albumin levels, and dynamic organ function, may produce interindividual variation in drug exposure that dwarfs any difference between the two prolonged strategies. Moreover, few trials employed systematic therapeutic drug monitoring, leaving actual drug concentrations unknown and potentially masking true differences between regimens.

Nevertheless, the practical implications are hard to ignore. If extended infusion delivers outcomes statistically indistinguishable from continuous infusion—while sparing patients dedicated lines, extra infusion pumps, and hospitals the associated costs and nursing burden—then the simpler strategy may be sufficient for routine critical care. The 93.2 percent posterior probability favoring extended infusion for clinical cure, though hypothesis-generating rather than definitive, gives intensive care clinicians a data-driven nudge toward a regimen that is easier to implement at scale. What the field now desperately needs is a large, adequately powered randomized trial directly comparing continuous infusion with extended infusion, ideally incorporating therapeutic drug monitoring to illuminate whether concentration control, rather than infusion duration per se, drives the outcomes that matter. Until such a trial arrives, this analysis suggests that the ICU’s antibiotic drip debate may be closer to resolution than anyone expected—and that the answer may favor the pragmatic option already sitting at most patients’ bedsides.

Subject of Research: Comparison of continuous, extended, and intermittent beta-lactam antibiotic infusion strategies in critically ill adults

Article Title: β-lactam infusion strategies in critical illness: is continuous superior to extended infusion? A systematic review and network meta-analysis

Article References: Peng, F., Tang, Z., Hu, Z., Zhou, X., Kang, L., Chang, W., Sun, Q., Xie, J., Qiu, H., & Yang, Y. (2026). β-lactam infusion strategies in critical illness: is continuous superior to extended infusion? A systematic review and network meta-analysis. Frontiers in Epidemiology, 6, Article 1883520. https://doi.org/10.3389/fepid.2026.1883520

Image Credits: AI Generated

DOI: 10.3389/fepid.2026.1883520

Keywords: beta-lactam antibiotics, sepsis, intensive care, continuous infusion, extended infusion, network meta-analysis, pharmacokinetics, clinical cure, mortality, critical care, antibiotic dosing, BLING III trial