LAUSR

To the Editor: The meta-analysis by Rhodes et al (1), published in a recent issue of Critical Care Medicine, confirms that a prolonged infusion of piperacillin-tazobactam over 3–24 hours, compared with bolus administration, reduces mortality in patients with sepsis and improves the cure rate. Since the Summary of Product Characteristics of this agent only includes a bolus administration, the findings by Rhodes et al (1) raise the question of whether this compound, when used as a prolonged infusion, is stable in vitro over more than 3 hours up to 24 hours. Data from the literature (2, 3) are not conclusive on this point because the stability of the drug depends on the presence of EDTA and the coadministration of other agents. Anyhow, an adequate stability is more likely at 3–4 hours (extended infusion, E inf ) than at 24 hours (continuous infusion, C inf ). In evaluating mortality, Rhodes et al (1) have included eight studies employing C inf and 10 studies employing E inf . If one considers C inf versus E inf as a covariate, a metaregression (based on the same data and the same random-effect model; computations made by the Open Meta-Analyst software, http://www.cebm. brown.edu/openmeta) shows that the odds ratio (OR) for mortality is numerically in favor of C inf but is far from statistical significance (OR for C inf vs E inf = 0.97; 95% CI, 0.65–1.45; p = 0.89). If one examines the absolute risk difference of C inf or E inf , each compared with the controls, its value is –5.1% for C inf (95% CI, –9.3% to –0.9%) and –3.3% for E inf (95% CI, –6.4% to –0.1%). These findings show that the absolute difference in mortality for E inf versus C inf can be estimated at +1.8% (with 95% CI from –3.8% to +7.4% determined as previously described [4]). Therefore, one can conclude not only that “no proof of difference” (5) exists favoring C inf versus E inf , but, more interestingly, C inf and E inf have an equivalent mortality with an equivalence margin set at ±7.5% (“proof of no difference” [5]). Based on this information, the administration of piperacillin/tazobactam by E inf in patients with sepsis should be preferred over the administration by C inf . Finally, it should be recalled that both dosage regimens are off-label. All authors disclosed off-label product use of piperacillin/ tazobactan in extended or continuous infusion.