Severe sepsis and septic shock are the leading causes of death in non-cardiac intensive care units (ICU).[1,2] Management of sepsis based on the Surviving Sepsis Campaign (SSC) guidelines[3] has been associated with survival benefits.[4–9] For the management of severe sepsis, the SSC recommended the implementation of an early goal-directed sepsis resuscitation bundle. Although the guidelines provide a road map for the care of severe sepsis patients, their success depends on the compliance of clinicians. In previous studies a national training program based on the SSC guidelines improved compliance with recommended processes of care for severe sepsis in ICUs. However, a long-term follow-up study showed that the compliance with the 6-hour early goal-directed resuscitation bundle returned to the baseline value, while the compliance with the 24-hour sepsis management bundle remained stable.[10] Considering the impact of the 6-hour early goal-directed resuscitation bundle and the 24-hour sepsis management bundles on clinical outcomes, use of the resuscitation bundle seems to be more strongly associated with clinical outcomes.[4] We designed this before-and-after study to test the hypothesis that a biweekly repetitive educational program for the emergency department (ED) and ICU residents, nurses, and faculties led by a multidisciplinary team could improve the compliance rate for the sepsis resuscitation bundle and further reduce the mortality of severe sepsis and septic shock patients.

This before-and-after study was approved by the Institutional Review Board (IRB) of Severance Hospital (IRB No 4-2009-0295), and the need for informed consent was waived by the IRB. Adult patients aged 18 years or more who were admitted to the ED and diagnosed with severe sepsis/septic shock during a 20-month period after initiation of educational sessions were included as the intervention group (between November 2007 and July 2009). For the control group, patients who were admitted to the ED and diagnosed with severe sepsis/septic shock under the same criteria between January 2006 and September 2007 were included. The clinical data were retrospectively collected by reviewing the electronic medical records (EMR) of the patients. Severe sepsis was defined as sepsis associated with acute organ dysfunction. The term organ dysfunction was defined by the SOFA scoring system.[11] Each subgroup scoring higher than 3 was defined as specific organ failure. Septic shock was defined as acute circulatory failure (systolic blood pressure less than 90 mmHg despite volume resuscitation).[12] We excluded 22 patients who were younger than 18 years, pregnant, with trauma, readmitted within 1 week due to previous diagnosis of sepsis, transferred from another institution, or had a do-not-resuscitate order or a previous history of heart failure. We compared clinicians’ adherence to the SSC guidelines for the implementation of the early goal-directed resuscitation bundle and clinical outcomes of the patients, including 28-day mortality, between the two groups.

The intervention group had 87 patients and the control group had 88 patients. Age, sex, and APACHE II score were comparable between the two groups (Table 1). At time-point zero, there were no significant differences in the origin of the infection between the groups. In the study population, the most common origin of sepsis was pneumonia and abdominal infection. The prevalence of organ dysfunction was comparable between the groups. The most common major organ dysfunctions were cardiovascular and respiratory dysfunction.

Considering each of the resuscitation bundle items, antibiotics were more frequently given within 3 hours in the intervention group (85% vs. 41%, p < 0.001). The rate of achieving hemodynamic goals was also higher in the intervention group (CVP 83% vs. 27%, p < 0.001; MAP 98% vs. 89%, p = 0.013; CVP and MAP, 94% vs. 25%, p < 0.001). ScVO₂ level was incommensurable because it was not measured in the control group. The rate of patients who achieved all hemodynamic goals within 6 hours after diagnosis (80% vs. 0%, p < 0.001) was definitely higher in the intervention group (Table 2).

The time from the diagnosis to the achievement of variables of the early goal-directed resuscitation bundle were compared between the two groups. After completing the training program, the times to administering antibiotics, insertion of central venous catheter, and achievement of CVP and MAP goals were significantly reduced (118 vs. 245 min, p < 0.001; 40 vs. 122 min, p = 0.0007; 135 vs. 197 min, p = 0.0163; 91 vs. 179 min, p < 0.001, respectively).

The patients in the intervention group showed a significantly lower mortality rate at 28 days after the diagnosis (16% vs. 32%, p = 0.04) than those in the control group. Kaplan-Meier plots of the probability of remaining alive are shown in Fig. 1. The Breslow test showed statistical significance in the difference in 28-day mortality between the groups (Fig. 1, p = 0.040). LOS in the ICU analyzed by Kaplan-Meier test was significantly longer in the control group (Fig. 2, p = 0.001), while LOS in the hospital was not significantly different between the two groups (Fig. 3, p = 0.078). Discontinuation of the training for 4 months reduced the goal achievement rate and increased the 28-day mortality rate, although they were restored after resuming the program (Fig. 4).

In the intervention group, 17 patients did not have their resuscitation goals completed, while the remaining 70 patients did. There was a statistically significant difference in the 28-day mortality rate in the patients who did or did not complete the bundle (11% vs. 41%, p = 0.004), while LOS in the ICU and LOS in the hospital were not significantly different according to the bundle completion rate. (p = 0.062)

Norepinephrine (53% vs. 34%, p = 0.019) was more frequently administered in the intervention group, compared with the control group. Fewer patients received dopamine (50% vs. 91%, p < 0.001) and dobutamine (4% vs. 19%, p = 0.006) in the intervention group, respectively. Steroids were more frequently administered in the intervention group, compared with the control group (34% vs. 19%, p = 0.036), and there was no significant difference in the use of vasopressin (14% vs. 7%, p = 0.138).

The present study showed that quality improvement action including repeated education provided by a multidisciplinary team could improve the adherence to SCC guidelines, thereby improving clinical outcomes in severe sepsis and septic shock patients. The adherence rate to the SCC guidelines before education or quality improvement actions was poor, a finding consistent with previous studies.[10,14] In our study, a specially organized multi-disciplinary team with good collaboration and a training program improved the management process of severe sepsis and septic shock patients. Trzeciak et al.[9] showed that effective emergency medicine and collaborative critical care improved the early achievement rate of goal-directed therapy. Our collaborative sepsis management team showed improved compliance and outcomes.

A nation-wide study demonstrated that a training program based on the SCC guidelines improved compliance with the recommended processes of care for severe sepsis in the ICU.[10,15] However, Ferre also showed a discrepancy in adherence to the protocol for the early goal-directed resuscitation bundle and the 24-hour management bundle. In this study, compliance with the protocol for the sepsis management bundle was maintained even after the 1-year follow-up was finished, while the compliance for the early goal-directed resuscitation bundle returned to the baseline level.

In the intervention group, we found a significant relationship between the achievement rate of the early goal-directed resuscitation bundle as recommended by the SCC guidelines and the mortality rate. Recent studies showed that non-compliance with the 6-hour sepsis resuscitation bundle increased mortality while non-compliance with the 24-hour sepsis management bundle did not.[4]

Studies have been conducted on the effect of education in clinical fields and reports have revealed the effect of education on the behavior of the medical staff.[16] Ordinary didactic approaches such as printed materials or practice guidelines for health care providers, including physicians, failed to change their behavior.[17] Only evidence-based educational approaches are known to have an impact on health care providers’ behavior. Such approaches include interactive educational activities, sequenced and multifaceted interventions, educational outreach or academic detailing, and audit and feedback to physicians.[18,19] Constant re-educational programs were essential for the successful implementation of the sepsis bundle.

Standardization and protocolization are important for improving clinical practice.[20] Many researchers tried to demonstrate that implementation of an evidence-based management protocol improved clinical outcomes of sepsis patients, but their results were not consistent.[5–7,9,14,21] Most of the trials used more rigorous fluid management earlier in the treatment protocol, more appropriate antibiotics therapy, and shorter LOS in the ICU or hospital, and the effects of the tested interventions on mortality varied. A few studies demonstrated mortality benefits,[6,22] while others did not.[21] These inconsistent results were thought to be related to clinicians’ adherence to the protocol. Nguyen et al. demonstrated that the completion of early goal-directed therapy was significantly associated with lower mortality[7] (odds ratio, 0.36; 95% confidence interval, 0.17–0.79; p = 0.01). Compared to previous studies, the completion rate after intervention in our study was relatively high. This high rate might result from the fact that we organized the clinical pathway with a periodic survey of the treatment process and continuous provision of training sessions.

The time to administration of broad-spectrum antibiotics was improved after intervention in this study, while the administration rate itself did not show significant changes. Similar improvement in earlier administration of antimicrobial agents was noticed in several recent studies.[21].

Among vasopressors, dopamine and norepinephrine are most frequently used, and are recommended by the SCC guidelines. In our study, there was a noticeable change in the use of vasopressors. More vasopressin and less dopamine were used in the intervention group. This change was thought to result from the physicians’ concerns about the arrhythmogenicity of dopamine and relative deficiency of vasopressin.[23,24] In addition, during the study periods, more physicians would tend to use steroids in patients with septic shock who, despite adequate fluid replacement, required vasopressor therapy to maintain adequate blood pressure. Therefore, steroid use showed a significant difference between the two groups in the course of time (34% vs. 19%, p = 0.036).

As demonstrated by Rivers,[22] time-dependant characteristics of sepsis management required involved clinicians to make standardized protocols, but its multi-disciplinary nature made implementation of protocols challenging. The present study confirmed that the sepsis resuscitation bundle, as well as the treatment bundle, could be implemented and maintained by continuously conducting surveys and education, and benefit sepsis patients with standardized care and early admission to the ICU.

Limitations of this study are as follows. Since creating a multi-disciplinary team, provision of repeated training, and formulating a resuscitation bundle implement pathway were simultaneously carried out, we could not differentiate the impacts of each factor. Moreover, since the control group data were retrospectively collected, randomization could not be performed. However, since all patients who were admitted to the ED and diagnosed as having severe sepsis/septic shock between January 2006 and September 2007 were allocated to the control group, selection bias would be minimal. Random allocation of patients to the intervention group would not be ethical, so the comparison between the intervention and control groups could be performed only observationally. In our institution, moreover, since patients’ data were recorded in the EMR system according to the institutional guidelines, we gathered all parameters relevant to the resuscitation bundle and patients’ outcome. Another important limitation was that we did not know how long the effect of training could be maintained and we could not confirm the ideal interval between training sessions. However, discontinuation of the training program for 4 months reduced the goal achievement rate and increased the 28-day mortality rate, although the goal achievement rate and mortality rate were restored after resuming the training (Fig. 4). We would like to suggest that repeated training with an interval between training sessions of at least 2 or 3 months is necessary to maintain improved compliance and clinical outcomes. The other limitation was that our hypothesis was based upon a premise that we could exclude the effect of the 24-hour sepsis management bundle on the clinical outcome. However, compliance with the 6-hour early goal-directed resuscitation bundle was in connection with the 24-hour sepsis management bundle.

In conclusion, we demonstrated that implementation of an evidence-based management protocol for severe sepsis or septic shock patients improved clinical outcomes. Repeated exposure to an educational program and interdisciplinary communication was helpful to maintain a high completion rate for the management bundles, which in turn decreased ICU mortality in sepsis patients.