Abstract
We would like to thank the reader for the commentary and questions regarding the GENESIS publication 1 . The question regarding potential influences on the primary outcome of mortality via study design or by confounding from baseline variables of illness severity is an important one.
As pointed out by the reader, the control group did have several statistically significant worse values (lactate [4.8 vs 4.4 mmol/L], bicarbonate [20.5 vs 21 meq/L], and Scv02 [66.9% vs 69.5%]) than the intervention group. However, when comparing universally accepted organ dysfunction scores as a metric for overall illness severity, the intervention group (group II) began with a higher organ dysfunction score than the control group (Acute Physiology And Chronic Health Evaluation [APACHE] 22.9 vs 20.6, P < .001). This translated into a higher predicted mortality (APACHE 46% vs 39%) in the intervention versus the control group, respectively. The baseline Sequential Organ Failure Assessment score that reflects early hemodynamic status was also higher in the intervention than in the control group (7.9 vs 7.0, P = .002). These findings not only indicate that the intervention group was of higher baseline illness severity, but also emphasizes the robust treatment effect that led to a substantial decrease in mortality despite starting at a greater baseline illness severity.
As far as the origin of presentation being a potential confounder, we agree with the comment that “the intervention group was more likely to be drawn from the emergency department (ED; 66% vs 50%)” and had “more inclusive enrollment of lower mortality cases” as evidenced by an ED intervention mortality rate of 23.8% which was less than that observed in the general practice unit (GPU; 33.5%) and intensive care unit (ICU; 31.3%). However, despite this higher origin of presentation in the ED, it is important to note that the relative risk of mortality among the origin of presentation for the control and interventions groups remained rather homogeneous Change so it reads: “(ED 0.66, GPU 0.66, ICU 0.69)” This preserved consistency in treatment effect would indicate that more of a global treatment effect was equally spanning the origins of presentation as opposed to one single origin unduly influencing the aggregate mortality.
Although randomized controlled trials are considered the research standard, their use in evaluating therapies among the critically ill have been called into question for multiple reasons 2 . Multiple investigators have shown that estimates of treatment effects in large case-controlled observational studies are qualitatively similar to those obtained in randomized, controlled trials. Thus, the results of studies such as GENESIS provide important and equally acceptable contributions to the outcome science of severe sepsis and septic shock. 3,4