Rebuttal From Dr Jones

Abstract

In their counterpoint, Dr Rivers and colleagues1 present the theoretical view that patients with septic shock present in very distinct “hemodynamic phases” and that Jones et al2 enrolled patients in a different phase of septic shock than did Rivers et al.3 According to their theory, decreased central venous oxygen saturation (Scvo2) always precedes the appearance of lactate—a concept not observed in my clinical practice. Clinicians who routinely care for the critically ill encounter patients with elevated lactate and normal Scvo2. Furthermore, as shown in Table 1,2-10 the hemodynamic patterns of the subjects enrolled by Rivers et al3 are markedly different from any other reported populations of patients with septic shock treated with quantitative resuscitation. The study by Rivers et al3 patients had much higher lactate, much lower Scvo2, and much higher mortality than described elsewhere. Possible explanations for this discrepancy may include that patients with septic shock in Detroit between 1997 and 2000 were markedly different than any other septic shock population reported in the world’s literature and/or that systematic selection bias was a significant problem in the their study. In such a scenario, their results have questionable external validity. Supporting either of these assertions is the fact that mortality in the control group of the Rivers et al3 study was 20% higher than any septic shock mortality reported in the recent literature, leaving one to question exactly what care they received.3 Little evidence supports the contention that Jones et al2 enrolled patients in a different phase of septic shock than did Rivers et al,3 because the population enrolled by Jones et al2 used inclusion criteria identical to the Rivers et al3 study. Furthermore, the population in the Jones et al3 study appears to be an accurate contemporaneous population from three US EDs; it has characteristics, including mortality rates, nearly identical to those described from other studies (Table 1). Table 1 —Summary of Initial Clinical Characteristics and Mortality Rates of Recent Studies of Quantitative Resuscitation for Septic Shock Rivers et al3 have raised the issue of potential methodologic limitations. They state that one must consider “weaknesses of the non inferiority study design” in the Jones et al2 study. However, our study did not show equivalence (not worse than and not better than) but rather noninferiority (not worse than and maybe better than). This study design is the only design that could test the hypothesis at hand and, furthermore, the methodologic quality of a noninferiority and superiority randomized controlled trial is the same if properly conducted. So it is unclear exactly what “weaknesses” in our study are being criticized. On the other hand, there appear to be real methodologic concerns in the Rivers et al3 trial. They failed to follow recommendations by not reporting whether any patients were randomized but not analyzed or performing an intention-to-treat analysis that could easily change the statistically significant results of the trial. Finally, the counterpoint1 states that 45% of patients with septic shock present with a normal lactate and suggest that because of this, lactate is an inconsistent guide of resuscitation.1 In fact, >65% of patients in contemporaneous studies present with an Scvo2 >70%, and using their argument, these data make Scvo2 an even more inconsistent resuscitative guide, particularly given that there is no therapy for markedly elevated and pathologic Scvo2 (ie, >90%).2,5 Because data from an experimental clinical trial are the only way to scientifically deduce the clinical efficacy of lactate clearance vs Scvo2 and because data from a large multicenter clinical trial demonstrated that lactate clearance is not inferior to Scvo2 as an end point of early sepsis resuscitation, as described herein, lactate clearance has principles that may make it the more appropriate end point to choose.