Recently, Levy and colleagues [1] presented the results of a study evaluating the effects of implementing the Surviving Sepsis Campaign Guidelines over a 7.5-year period.

The authors concluded that performance metrics can drive change in clinical behavior, improve quality of care, and may decrease mortality in patients with severe sepsis and septic shock.

Many studies have shown that implementation of performance metrics drives change and improves outcome. Even when the target variable behaves similarly in the intervention and control groups, mortality may still be lower in the intervention group [2]. This could result from the so-called Hawthorne effect, which has been shown to be present in many interventions. In a recent study, Van Zanten et al. [3] showed that implementing the national sepsis program in the Netherlands decreased mortality in both sepsis and non-sepsis patients, whereas mortality in non-participating hospitals did not change over time. Considering that methods of screening were established locally and that no effort to supervise the quality or completeness of screening was attempted, the increase in physician awareness in detecting severe sepsis [3] and the definition of patients enrolled may also have changed over time.

The definition of high compliance versus low compliance hospitals in the current study is interesting. A high compliance hospital was defined as one having a compliance rate of >15 % of the resuscitation bundle and >20 % of the management bundle. Only meeting 1 or 2 items from the bundles would thus already characterize a high compliance hospital. On average, it took hospitals 3–4 years to increase overall compliance to 30 %. This is in line with other studies on compliance over time when implementing the Surviving Sepsis Campaign. Even very small increases in compliance were associated with improvements in outcome, whereas a subsequent decrease in compliance during long-term follow-up was not associated with worsening of the outcome [4].

Therefore, the claim by the authors that part of the observed outcome improvements was related to increasing compliance in the resuscitation and management bundles over time may be over-interpretation of the data. It seems that protocolized screening of patients and adding clinical tools for therapy of itself improves performance. An alternative explanation would be that a continuous quality improvement process with education, improvement in skills and behavior, creation of dedicated groups that review and adapt the protocols, performance of data monitoring and feedback to the healthcare workers all result in improved performance. In a recent program to prevent central line-associated bloodstream infections, even compliance as low as 38 % decreased infection rates due to the overall effect of the program [5].

Regarding clinical interventions, it is remarkable that an increased compliance resulted in improved outcome, as several items of the bundles have been either questioned or shown to be ineffective in improving outcome. Nevertheless, in the current study, almost all items of the resuscitation and management bundle were associated with improved hospital mortality. First, recent studies have shed doubt on the use of a lactate measurement or central venous oxygenation measurement in optimizing resuscitation [6]. Second, several studies have associated central venous pressures above 8 mmHg with increased risk of acute kidney injury and mortality [7, 8]. Third, activated protein C was shown not to improve mortality in the confirmatory trial in patients with septic shock [9]. Fourth, glucose management, although initially found to improve outcome in an efficacy study [10], showed no effect or even an increase in mortality in efficiency studies [11, 12]. And fifth, the use of steroids is subjected to continuous debate [13, 14], and in the current study their potential effects were less clear. There is one undisputed element in the bundles, which is the early administration of antibiotics. Many studies have underscored the importance of this, particularly in patients with septic shock [15, 16].

Should we then conclude that the Surviving Sepsis Campaign Guidelines [17] should be abandoned as they are ineffective and may even reverse the observed trends in decreased sepsis-associated mortality worldwide? We think not, as the Surviving Sepsis Campaign has had a tremendous impact on early recognition and treatment of patients with a syndrome that is difficult to recognize and evaluate. The remarkable change in baseline characteristics of recent studies [6], together with the landmark study [18], shows the impact of early recognition and treatment following the general adoption of the Surviving Sepsis Campaign Guidelines. Although improved screening with new techniques might add patients with a low risk of mortality to the pool of sepsis patients, this should strengthen the need for adjustments and profiling of the Guidelines rather then throwing them overboard.