Background The ndings of randomised controlled trials (RCT), observational studies, and meta-analyses vary regarding the e ectiveness of prolonged β-lactam infusion. We aimed to identify the e ectiveness of prolonged versus short-term infusion of antipseudomonal β-lactams in patients with sepsis.
Methods We did a systematic review and meta-analysis to compare prolonged versus short-term intravenous infusion of antipseudomonal β-lactams in patients with sepsis. Two authors independently searched PubMed, Scopus, and the Cochrane Library of clinical trials until November, 2016, without date or language restrictions. Any RCT comparing mortality or clinical e cacy of prolonged (continuous or ≥3 h) versus short-term (≤60 min) infusion of antipseudomonal β-lactams for the treatment of patients with sepsis was eligible. Studies were excluded if they were not RCTs, the antibiotics in the two arms were not the same, neither mortality nor clinical e cacy was reported, only pharmacokinetic or pharmacodynamic outcomes were reported, or if ten or fewer patients were enrolled or randomised. Data were extracted in prespeci ed forms and we then did a meta-analysis using a Mantel-Haenszel random-e ects model. The primary outcome was all-cause mortality at any timepoint. This meta-analysis is registered with the PROSPERO database, number CRD42016051678, and is reported according to PRISMA guidelines.
Findings 2196 articles were identi ed and screened, and 22 studies (1876 patients) were included in the meta-analysis. According to the Grading of Recommendations Assessment, Development, and Evaluation tool, the quality of evidence for mortality was high. Carbapenems, penicillins, and cephalosporins were studied. Patients with variable age, Acute Physiology and Chronic Health Evaluation (APACHE) II score, severity of sepsis and renal function were enrolled. Prolonged infusion was associated with lower all-cause mortality than short-term infusion (risk ratio [RR] 0·70, 95% CI 0·56–0·87). Heterogeneity was not observed (p=0·93, I2=0%). The funnel plot and the Egger’s test (p=0·44) showed no evidence of publication bias.
Interpretation Prolonged infusion of antipseudomonal β-lactams for the treatment of patients with sepsis was associated with signi cantly lower mortality than short-term infusion. Further studies in speci c subgroups of patients according to age, sepsis severity, degree of renal dysfunction, and immunocompetence are warranted.
We did not do analyses regarding microbiologically proven infections (according to individual or groups of bacteria) because these data were absent in the literature. Notably, continuous infusion has been associated with lower mortality in culture-negative (13·4% vs 26%, p=0·001) but not culture-positive (33·3% vs 26·8%, p=0·3) infections.40 The clinical signi cance of this nding warrants further study. Data regarding the speci c site of infections also need to be generated. Additionally, duration of masking and in some studies the duration of treatment was relatively short (in ve of 13 RCTs the mean or median treatment duration was 5 days or less). Although for both community and nosocomial infections short-duration treatments have been associated with similar outcomes compared with longer ones,77–80 we are not aware of studies evaluating the e ect of such short treatment duration on the outcomes of patients with severe sepsis. Finally, safety assessment was di cult because of under-reporting of adverse events. Although the dose of prolonged infusion in several studies was lower than the dose for short-term infusion, the higher serum concentrations achieved for a longer period of time with prolonged infusion and the higher peak concentrations achieved with short-term infusion could have resulted in more adverse events in either group.
In conclusion, prolonged infusion of antipseudomonal β-lactams in patients with sepsis was associated with lower mortality than short-term infusion; a signi cant association was evident in several subgroup and sensitivity analyses. The overall quality of evidence was high. The dissociation between the signi cant reduction in mortality and the non-signi cant di erence in clinical cure requires further investigation. Although the majority of RCTs included only ICU patients, prolonged infusion might bene t all hospitalised patients with sepsis; further studies in speci c subgroups of patients according to age, sepsis severity, degree of renal dysfunction and immunocompetence are warranted. The contribution of therapeutic drug monitoring on the outcome of patients treated with prolonged infusion of antipseudomonal β-lactams merits further study.