This is the authors' abstract. We don't add key points for this paper.
W e read with interest the systematic review and meta-analysis by Li et al. examining prolonged (continuous and extended) versus intermittent infusions of beta-lac tam antibiotics in ICU patients with pneumonia (1).Based on frequentist analysis, the authors found no statistically significant improvements in two critical endpoints for severe infection: mortality (RR 0.79, 95% CI 0.59-1.07)and clinical cure rate (RR 1.09, 95% CI 0.99-1.21).We identified several methodological considerations.First, many included RCTs contained sparse events or zero events in both arms, making risk difference (RD) within a Bayesian framework more appropriate than risk ratios (2, 3).RD also facilitates clearer communication of treatment effects to patients and families (4).Second, given inherent heterogeneity in patient severity and beta-lactam types, random-effects models should be preferred over the fixed-effects models used by the authors.Third, continuous and extended infusions might have sufficiently different pharmacokinetic profiles to warrant separate analysis.Finally, Bayesian approaches may be more suitable than frequent ist p-values for critical care research, providing complementary insights for clinical decision-making (5).Therefore, we conducted a Bayesian network meta-analysis using multinma in R with random-effects models and non-informative priors.Models demonstrated convergence and good fit.The posterior distribution half-eye plot is shown in Fig. 1A.Compared with intermittent infusion, mortality RD was -0.0550 (95% credible interval [CrI] -0.155 to 0.227, 72.4% probability of benefit) for continuous infusion and -0.0573 (95% CrI -0.134 to 0.0514, 90.1% probability of benefit) for extended infusion.The surface under the cumulative ranking curve (SUCRA) analysis ranked extended infusion first (0.708), continuous second (0.605), and intermittent third (0.187).Predicted mortality (Fig. 1B) was lowest with extended infusion (11.8%, 95% CI 4.2%-23.4%),similar with continuous (11.9%, 95% CI 2.4%-40.4%),and highest with intermittent (17.6%, 95% CI 14.2%-21.3%).Notably, continuous infusion showed greater uncertainty due to limited RCTs with rare events.Intermittent infusion could not achieve mortality below 5% (0% probability) and had a 93.1% probability of mortality exceeding 15%.Clinical cure rates improved with both prolonged infusion strategies compared to intermittent infusion.Continuous infusion showed RD 0.0885 (95% CrI -0.221 to 0.270, 76.2% probability of benefit) and extended infusion showed RD 0.104 (95% CrI -0.106 to 0.274, 89.0% probability of benefit) (Fig. 2A).The consistent SUCRA ranking was exten ded (0.723), continuous (0.603), and intermittent (0.174).Predicted absolute clinical cure rates were 71.7% (95% CI 50.2%-88.9%)for extended, 70.3% (95% CI 38.5%-88.4%)for continuous, and 61.1% (95% CI 56.6%-65.7%)for intermittent infusion (Fig. 2B).
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Lai et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: