The Surviving Sepsis Campaign bundles and outcome: results from the International Multicentre Prevalence Study on Sepsis (the IMPreSS study)

Despite evidence demonstrating the value of performance initiatives, marked differences remain between hospitals in the delivery of care for patients with sepsis. The aims of this study were to improve our understanding of how compliance with the 3-h and 6-h Surviving Sepsis Campaign (SSC) bundles are used in different geographic areas, and how this relates to outcome. This was a global, prospective, observational, quality improvement study of compliance with the SSC bundles in patients with either severe sepsis or septic shock. A total of 1794 patients from 62 countries were enrolled in the study with either severe sepsis or septic shock. Overall compliance with all the 3-h bundle metrics was 19 %. This was associated with lower hospital mortality than non-compliance (20 vs. 31 %, p < 0.001). Overall compliance with all the 6-h bundle metrics was 36 %. This was associated with lower hospital mortality than non-compliance (22 vs. 32 %, p < 0.001). After adjusting the crude mortality differences for ICU admission, sepsis status (severe sepsis or septic shock), location of diagnosis, APACHE II score and country, compliance remained independently associated with improvements in hospital mortality for both the 3-h bundle (OR = 0.64 (95 % CI 0.47−0.87), p = 0.004)) and 6-h bundle (OR = 0.71 (95 % CI 0.56−0.90), p = 0.005)). Compliance with all of the evidence-based bundle metrics was not high. Patients whose care included compliance with all of these metrics had a 40 % reduction in the odds of dying in hospital with the 3-h bundle and 36 % for the 6-h bundle.


Mitchell Levy,
Alpert Medical School at Brown University, Rhode Island Hospital, Providence, Rhode Island, USA Funding/Support: The SSC, the ESICM and the SCCM provided logistic support for mailings, the Web site, and meetings of the steering committee.Centres did not receive any payment for recruiting patients.

Introduction
Despite many advances in our understanding of sepsis [1] and recent reports of improved outcomes from the condition [2], the disorder remains of epidemic incidence with an unacceptably high death rate and devastating long-term effects.Quality improvement efforts through the application of sepsis care bundles have reduced mortality, but the number of hospitals participating in such initiatives remains low [3,4].
The Surviving Sepsis Campaign (SSC) was developed to reduce the mortality from severe sepsis and septic shock.SSC activities directed towards this goal included: the development of evidence-based guidelines [5][6][7], educational packages to improve the awareness and understanding of the condition and a quality improvement initiative to help healthcare professionals adopt the identified best practice [4,8,9].A recent analysis covering a 7.5-year period demonstrated that active participation in the campaign was associated with increased guideline adherence, as evidenced by improved compliance with established performance metrics.Additionally, these improvements were in themselves associated with reductions in sepsis-related mortality [4].Finally, the longer hospitals participated in the campaign and the more they improved their performance, the greater were the observed outcome improvements.
Despite evidence demonstrating the value of such performance initiatives, marked differences remain between hospitals in the delivery of care for patients with sepsis.Reviewing the inconsistent application of measures identifies an important opportunity to reduce sepsis-induced mortality further.It is recognized that the penetration of the Campaign to hospitals around the world is limited.To inform current and future quality improvement efforts in sepsis, there is a need to better understand how widely and well the evidence-based Surviving Sepsis Campaign bundles are used in different geographic areas, and how these relate to outcome.In particular, it is necessary to assess the compliance with the 2012 guidelines and associated bundles as all previous data assessed compliance with the previous iterations.A critical step in quality improvement efforts is a thorough assessment of current practice in order to identify on going gaps in clinical processes.This study was designed to address this need.

Study design and participants
This was a global, prospective, observational, quality improvement study of the prevalence of patients with either severe sepsis or septic shock, with evidence-based practices.On November 7 th 2013, (00 00 to 24 00 ), consecutive patients presenting to either the emergency department (ED) or being cared for in an ICU (either intermediate care or intensive care) with severe sepsis or septic shock were enrolled.To be eligible patients had to have a high clinical suspicion of an infection, together with a systemic inflammatory response and evidence of acute organ dysfunction and / or shock [10].Patients were excluded if they were less than 18 years of age.Participating hospitals were identified through membership of the European Society of Intensive Care Medicine (ESICM), the Society of Critical Care Medicine (SCCM) and the Surviving Sepsis Campaign (SSC) and through the networks of national and local coordinators.The project was approved as a quality improvement initiative in each participating country, thus precluding the necessity for written informed consent from participants.All demographic and clinical information were de-identified as part of data collection processes so that patient anonymity was strictly maintained throughout the study.

Procedures
Local investigators were identified and were supported by a network of national coordinators.Key study information was provided through a website (http://impress-ssc.com/) which included the protocol, answers to key questions and access to the electronic-case report form (eCRF). Upon entry into the eCRF, each patient was assigned a unique study identifier.No patient identifiable data was submitted to the online database housed on a secure server in Germany.
A multi-continental panel of critical care experts iteratively developed a "realistic data set."These data elements included all key and relevant clinical and demographic data points whilst not discouraging centres from participating because of an excessive burden of data collection [See ESM].
The data collected were all part of routine clinical care.Patients were followed up until 30 days after study enrolment or hospital discharge, whichever occurred first.Data were collected for every patient, on whether their management fulfilled the requirements of the SSC bundles [7].The 3-hour bundle for patients with severe sepsis / septic shock (i.e., elements completed within 3 hours) includes: a lactate level measurement; blood cultures obtained prior to the administration of antibiotics; the administration of broad spectrum intravenous antibiotics; and, administration of 30mL/Kg of intravenous crystalloid if hypotension was present or the lactate level was ≥ 4 mmol/L (36 mg/dL).The 6-hour bundle for patients with severe sepsis / septic shock (i.e., elements completed within 6 hours) includes: a re-measurement of lactate if it was initially raised; the application of vasopressors when hypotension (mean arterial pressure [MAP] ≤ 65 mm Hg) is persistent despite initial fluid resuscitation; and, measurement of central venous pressure (CVP) and central venous oxygen saturation (ScvO2) when there is persistent arterial hypotension despite volume resuscitation or the initial lactate concentration is ≥ 4 mmol/L.The 6 hour bundle was reported for all patients in the study and also just for those who remain with persistent hypotension and / or hyperlactataemia following volume resuscitation within the 6 hour period.

Statistical analysis
Categorical variables are described using frequencies and proportions and are compared using Fisher's exact test.Comparisons between geographic regions have been made excluding the data from Oceania, due to the low numbers of patients enrolled from this region making the estimates less reliable.Continuous variables are described as mean and standard deviation if normally distributed or median and inter-quartile range if not.A generalized estimating equation (GEE) population-averaged logistic regression was used to assess the association between prognostic factors and mortality where country was the clustering or panel variable with an exchangeable correlation structure.Both unadjusted and adjusted odds ratios are presented along with their associated 95% confidence intervals.The following adjustment variables of age, ICU admission (yes vs. no), sepsis status (severe vs. shock), location (ED, ward, ICU, OR, unknown), sepsis origin (community, health care, hospital, or ICU acquired), and APACHE II were determined a priori.All analyses were run using Stata 13.1, StataCorp, College Station, TX.

Results
We collected data describing patients presenting with severe sepsis and / or septic shock in 618 hospitals from 62 countries.Data were returned on 1927 patient records of which 133 were removed having been identified as duplicates or having missing hospital outcome data, leaving 1794 for analysis [ESM Figure 1].A median number of 9   Oceania had only 14 observations thus their results have very wide confidence intervals.
Tables 1 and ESM Table 2 and 3 show the baseline data and outcomes.Overall, 47% of the patients were over 65 years old and 59% presented with at least one co-morbid illness [ESM Table 1 and 2].
The majority of patients were diagnosed in the emergency department (54%) and the most frequent presentations were of community acquired sepsis (59.9%) and pneumonia (40%).The most common organ dysfunctions at presentation were hypotension (66%), acute respiratory distress syndrome [11] (57%) and acute kidney injury (46%).In 39% of the patients the sepsis progressed to septic shock [ESM Table 2].1545 (86%) of the patients were admitted to an intensive care unit and the overall hospital mortality was 28% with a median (IQR) length of hospital stay of 13.7(6.5 -24.6) days.
Demographic and clinical details of patients presenting by region are described in Table 1.Patients were more likely to be older in Western Europe and present with chronic illnesses in North America.
The diagnosis of severe sepsis/septic shock was most likely to be made in the emergency department in North America (64%), the ward in Asia (24%) and the intensive care unit in Eastern Europe (44%).Unadjusted hospital mortality was highest in Eastern Europe (44%) and lowest in Oceania (14%).When the crude mortality rates for each region were compared against North America and adjusted for ICU admission, sepsis status, location of diagnosis, origin of sepsis, APACHE II score and country, East Europe and Central/South America remained with higher odds of dying (OR = 2.46 (95% CI: 1.27-4.77),p = 0.008 and OR = 2.17 (95% CI: 1.16-4.03),p = 0.015), respectively) [Figure 1].There were no statistical differences found in adjusted mortality rates between North America and Asia, Oceania, West Europe and Africa/Middle East.
Overall compliance with all the 3-hour bundle metrics was 19%.This was associated with lower hospital mortality than non-compliance (20 vs. 31%, p < 0.001).Overall compliance with all the 6hour bundle metrics was 36%.This was associated with lower hospital mortality than noncompliance (22 vs. 32%, p < 0.001) [Table 2].For patients who had persistent hypotension and / or hyperlactatemia full compliance with the 6-hour bundle was reported in 90 (11%) of patients.The compliance with the 3-hour bundle was highest in North America (29%) and lowest in Central / South America (9.5%), whereas the compliance with the 6-hour bundle was highest in West Europe (41%) and lowest in Africa / Middle East (26%) [Table 3].After adjusting the crude mortality differences for ICU admission, sepsis status (severe sepsis or septic shock), location of diagnosis, APACHE II score and country, compliance remained independently associated with improvements in hospital mortality for both the 3-hour bundle (OR = 0.64 (95% CI: 0.47-0.87),p = 0.004)) and 6-hour bundle (OR = 0.71 (95% CI: 0.56-0.90),p = 0.005)) [Table 4].

Discussion
In this multi-national study of severe sepsis and septic shock the hospital mortality rate was 28.4% and this varied significantly between different geographic regions of the world.Compliance with all of the evidence-based bundle metrics for the treatment of this condition was not high: 19% for the 3-hour bundle and 35.5% for the 6-hour bundle.Patients whose care included compliance with all of these metrics had a 40% reduction in the odds of dying in hospital with the 3-hour bundle and 36% for the 6-hour bundle.
Despite recent reports of reducing mortality rates from septic shock [2,12] and data from recent randomized controlled trials suggesting the mortality is now quite low [13,14], we found a hospital mortality rate of 28.4%.This is consistent with reports from other observational studies [4] [15] that suggest the mortality rate may still be higher than reported from interventional studies [16] that often exclude the highest risk groups of patients, and also more formally structure the delivery of care.We have also found large differences in mortality between different geographic regions.We have previously reported similar findings when comparing Europe to North America where the crude mortality rate was lower in North America, but the difference did not remain after adjusting for baseline confounding influences [17].In this current study, the differences between West Europe and North America were non significant after adjustments, however we have been able to document significant differences between North America and Central / South America and Eastern Europe.
Other authors have described differences in the provision of intensive care facilitates and treatments between and within continents [18][19][20][21][22], but this study adds to this by extending the findings to a global scale.
The strengths of this study include the defined dataset, a web-based data entry portal, a website containing all relevant documents and training manuals and the participation from over 60 countries representing all parts of the globe.We describe a cohort of patients with severe sepsis and septic shock that could be identified in emergency departments and intensive care units in each participating country.We have previously published [7] and extensively marketed [23][24][25] the evidence based bundle metrics so they were familiar to all participating sites.We were then able to collect data describing compliance with these metrics and also data describing presentation patterns and severity of these patients enabling us to correct bundle compliance and outcome metrics for such differences.
Our study has some limitations.Our dataset was a compromise between being an exhaustive list describing all facets of a patient with sepsis and being small enough to encourage site participation and data reliability.We enrolled relatively few patients per site on a single study day and for many countries only a few sites participated.This reduces the external generalizability of our data set.This 'point' estimate reduces the external validity as there is likely to be significant variance in both admission numbers of patients presenting to hospital and clinical practice on a day to day basis and also does not compensate for the known seasonal variations in incidence of the condition in the different regions of the world.In addition we only followed our patients up until hospital discharge, therefore we have little understanding into what happened to the patients following discharge and to where the patients went.This is likely to be very different between countries in the study.
We have limited data describing other quality metrics of the participating institutions.It is possible that the association we have found between bundle compliance and outcome improvement may be nothing more than a surrogate of how well that institution performs.It would be unwise to infer causality from this relationship.In deed results from several recent large randomized controlled trials [13,14,26] have questioned the need for some of the elements that are included in the 6 hour bundle.These new data are currently being assimilated into an update of the evidence based guidelines and the quality improvement metrics will also change to reflect the new data, in particular on the emphasis placed on measurement of central venous oxygen saturations as part of the overall protocolized resuscitation strategy.
Our study has confirmed the reports of others that compliance with sepsis improvement metrics are not good although when performed are associated with outcome improvements.This study is the first report of compliance with the 2012 Surviving Sepsis Campaign bundles [7] and as such adds to the literature supporting this methodology for quality improvement [4,17,[27][28][29].Our study confirms previous reports that the rate of bundle compliance is different between regions [4,17] and confirms the ability of sites in North America to perform the initial (3-hour) resuscitation bundle elements better that other regions, but also suggests that Western Europe has higher compliance with the 6-hour elements.In addition we have confirmed the reports that compliance with these tools improves outcome even when taking into account all presenting differences [4, 17, 23, 25 , In conclusion we have observed in a large multi-national observational study that compliance with evidence-based bundle metrics designed to improve outcomes from septic shock remains low, varies significantly between different geographical regions and when performed is associated with improvements in outcome.

Figure 1
Figure 1Estimated mortality and its associated 95% CI by region where the number represents the observations within each region.

Table 1
Presenting characteristics and outcomes for patients enrolled into the IMPreSS Study split by geographic region.All numbers are presented as n (%) unless otherwise stated.

Table 2 Surviving
Sepsis Campaign bundle compliance and associated hospital mortality for patients enrolled into the IMPreSS Study.All numbers are presented as n (%) unless otherwise stated.* represents a p value of < than 0.0001 by the Fishers exact test for the mortality of bundle compliance versus non compliance.

Table 3 Surviving
Sepsis Campaign bundle compliance and hospital outcome for patients enrolled into the IMPreSS Study split by geographic region.All numbers are presented as n (%) unless otherwise stated.