Guideline Bundles Adherence and Mortality in Severe Sepsis and Septic Shock Arthur R. H. van Zanten, MD, PhD1; Sylvia Brinkman, PhD2,3; M. Sesmu Arbous, MD, PhD2,4; Ameen Abu-Hanna, PhD3; Mitchell M. Levy, MD5; Nicolette F. de Keizer, PhD2,3; for The Netherlands Patient Safety Agency Sepsis Expert Group
Department of Intensive Care, Gelderse Vallei Hospital, Ede, The Netherlands. Department of Medical Informatics, Academic Medical Center, Amsterdam, The Netherlands. 3 National Intensive Care Evaluation (NICE) Foundation, The Netherlands. 4 Department of Intensive Care, Leiden University Medical Center, Leiden, The Netherlands. 5 Division of Pulmonary and Critical Care Medicine, Warren Alpert Medical School of Brown University, Providence, RI. Members of the Netherlands Patient Safety Agency Sepsis Expert Group are mentioned in Addendum 1 (Supplemental Digital Content 1, http:// links.lww.com/CCM/A923). Dr. van Zanten contributed to conception and design of the VMS Sepsis program, holds the position of Chair of the Sepsis Expert Group, promoted acquisition of data, was involved in analysis and interpretation of data, drafted and finalized the manuscript and revised it critically for the important intellectual content and comments made by other authors, and approved the final version of the article. Dr. Brinkman had full access to all the data in the study, takes responsibility for the integrity of the data and the accuracy of the data analysis, performed the statistical analyses, was involved in the interpretation of data, contributed to drafting and finalizing the manuscript and revised it critically for the important intellectual content and comments made by other authors, and approved the final version of the manuscript. Dr. Arbous contributed to conception and design of the VMS Sepsis program, holds the position of member of the Sepsis Expert Group, promoted acquisition of data, was involved in analysis and interpretation of data, contributed to drafting and finalizing the manuscript and revised it critically for the important intellectual content and comments made by other authors, and approved the final version of the manuscript. Dr. Abu-Hanna supervised the statistical analyses and contributed to drafting and finalizing the manuscript and revised it critically for the important intellectual content and comments made by other authors and approved the final version of the manuscript. Dr. Levy contributed to drafting and finalizing the manuscript and revised it critically for the important intellectual content and comments made by other authors and approved the final version to be submitted of the manuscript. Dr. Keizer had full access to all the data in the study, takes responsibility for the integrity of the data and the accuracy of the data analysis, was responsible for the NICE registry and involved in statistical analyses performed and interpretation of data, contributed to drafting and finalizing the manuscript and revised it critically for the important intellectual content and comments made by other authors, and approved the final version of the manuscript. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ ccmjournal). 1 2
Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000297
1890
www.ccmjournal.org
Presented, in part, at the 33rd International Symposium on Intensive Care and Emergency Medicine Meeting, Square-Brussels Meeting Center, Brussels, Belgium, March 21, 2013, by the coordinating investigator and corresponding author Dr. van Zanten, titled “Resuscitation from severe sepsis: Do we need care bundles? Yes”—SSC guidelines (Gold Hall). Dr. van Zanten received support for travel (Safety Management System [VMS] initiation meetings were reimbursed by the VMS program); consulted and lectured for Eli Lilly (received honoraria for advisory board meetings and lectures, < $1,000 before 2008); and disclosed other: for inclusion of patients in the Prowess Shock trial from Eli Lilly, research costs were paid to the local ICU research foundation. He was not involved in the international Surviving Sepsis guideline development; however, he has been involved with the Surviving Sepsis Campaign in The Netherlands from its beginning. Dr. Brinkman is employed by National Intensive Care Evaluation (NICE) foundation and is data manager and researcher of the NICE registry. (The National Patient Safety Agency (VMS) asked the NICE foundation to host the registration on sepsis. Dr. Brinkman is a postdoc working for the NICE registry.) Her institution received grant support from the National Patient Safety Agency (VMS) (the NICE registry received a grant of $55,000 of the VMS sepsis program to process, report, and analyze the data). Dr. Arbous is employed by Leiden University Medical Center (LUMC), Leiden, The Netherlands (works as anesthesiologist-intensivist on the department of intensive care of the LUMC). Travel costs for VMS initiation meetings were reimbursed by the VMS program. Dr. Abu Hanna’s institution received grant support from National Patient Safety Agency (VMS). (Dr. Abu-Hanna is head of the Department of Medical Informatics which hosts the NICE registry. The NICE registry received a grant of $55,000 of the VMS sepsis program to process, report, and analyze the data.) Dr. Levy received grant support from Eisai (Ocean State Clinical Coordinating Center to fund clinical trial [$500K]), and he has been involved with the Surviving Sepsis Campaign guideline from its beginning. Dr. de Keizer is employed by the NICE foundation and is manager and principal investigator of the NICE registry. (The National Patient Safety Agency [VMS] asked the NICE foundation to host the registration on sepsis. Dr. de Keizer is a member and the managing director of the NICE board for which the NICE registry pays the department of medical informatics.) Her institution received grant support from National Patient Safety Agency (VMS) (Dr. de Keizer is a principal investigator at the department of medical informatics which hosts the NICE registry. The NICE registry received a grant of $55,000 of the VMS sepsis program to process, report, and analyze the data) and served as a board member for the NICE foundation (the National Patient Safety Agency (VMS) asked the NICE foundation to host the registration on sepsis). For information regarding this article, E-mail:
[email protected]
Objective: Surviving Sepsis Campaign bundles have been associated with reduced mortality in severe sepsis and septic shock August 2014 • Volume 42 • Number 8
Clinical Investigations patients. Case-mix adjusted mortality evaluations have not been performed to compare hospitals participating in sepsis bundle programs with those not participating. We aimed to achieve an individual bundle target adherence more than 80% and a relative mortality reduction of at least 15% (absolute mortality reduction 5.2%) at the end of 2012. Design: Prospective multicenter cohort study in participating and nonparticipating centers. Setting: Eighty-two ICUs in The Netherlands. Patients: In total, 213,677 adult ICU patients admitted to all ICUs among which 8,387 severe sepsis patients at 52 participating ICUs and 8,031 severe sepsis patients at 30 nonparticipating ICUs. Interventions: A national program to screen patients for severe sepsis and septic shock and implement Surviving Sepsis Campaign bundles to complete within 6 and 24 hours after ICU admission. Measurements and Main Results: Bundle target adherence and case-mix adjusted in-hospital mortality were evaluated through odds ratios of time since program initiation by logistic generalized estimating equation analyses (July 2009 through January 2013). Outcomes were adjusted for age, gender, admission type, severity of illness, and sepsis diagnosis location. Participation duration was associated with improved bundle target adherence (adjusted odds ratio per month = 1.024 [1.016– 1.031]) and decreased in-hospital mortality (adjusted odds ratio per month = 0.992 [0.986–0.997]) equivalent to 5.8% adjusted absolute mortality reduction over 3.5 years. Mortality reduced in screened patients with other diagnoses (1.9% over 3.5 yr, adjusted odds ratio per month = 0.995 [0.9906– 0.9996]) and did not change in nonscreened patients in participating ICUs, nor in patients with sepsis or other diagnoses in nonparticipating ICUs. Conclusions: Implementation of a national sepsis program resulted in improved adherence to sepsis bundles in severe sepsis and septic shock patients and was associated with reduced adjusted in-hospital mortality only in participating ICUs, suggesting direct impact of sepsis screening and bundle application on in-hospital mortality. (Crit Care Med 2014; 42:1890–1898) Key Words: adherence; bundles; guidelines; intensive care medicine; mortality; quality improvement; screening; septic shock; severe sepsis; Surviving Sepsis Campaign
S
evere sepsis is a major cause of morbidity and mortality and a leading reason for ICU admission (1–3). In 2002, the Surviving Sepsis Campaign (SSC) was launched to improve outcome in sepsis (4). Evidence-based guidelines were developed to optimize care of patients with severe sepsis and septic shock and reduce mortality (5, 6). A multifaceted model was developed using guideline bundles to change bedside practice. Furthermore, an international SSC database was developed (4). Data from this database showed sustained, continuous quality improvements in sepsis care. Participation up to 2 years was associated with reductions in in-hospital mortality rates (5.4% over 2 yr [95% CI, 2.5–8.4%]) (7). In addition, other studies have
Critical Care Medicine
been published reporting positive effects on bundle adherence and mortality (8–11). Most related studies have important limitations such as lack of correction for severity of illness of patients included and not having control groups. Recently, unadjusted inhospital mortality differences were reported in ICUs participating in SSC initiatives in Europe versus the United States that disappeared after case-mix adjustment, indicating the importance of c ase-mix adjustment (12). Positive effects of previous reports could be due to inclusion of patients with lower risk of mortality caused by increased awareness and recognition of sepsis or to general quality improving initiatives. Therefore, causality of observed effects could not be demonstrated. As part of a national patient safety program initiated in 2008 in The Netherlands to improve patient care in hospitals, a program was developed based on SSC bundles to improve early diagnosis and treatment of patients with severe sepsis and septic shock. The program started in July 2009. In-hospital mortality among severe sepsis and septic shock ICU patients in The Netherlands was 34.8% in the first half of 2009. We aimed to achieve a relative mortality reduction of 15%, corresponding to an absolute mortality reduction of 5.2% at the end of 2012. We evaluated effects of screening for severe sepsis and implementation of SSC bundles on adherence to the 6- and 24-hour sepsis bundle targets and adjusted in-hospital mortality comparing ICUs that participated in the program with ICUs that did not participate.
MATERIALS AND METHODS The National Patient Safety Agency (Safety Management System [VMS]) initiated the sepsis program. An expert group (Addendum 1, Supplemental Digital Content 1, http://links. lww.com/CCM/A923) developed educational materials based on SSC bundles. All ICUs in The Netherlands could participate (n = 93). Participation was voluntary, nonfunded, and time of entry varied among hospitals. The Dutch government funded initiation of VMS. The program promoted local physician and nurse champions training through national meetings. During regional training, educational materials were introduced. A database as part of the National Intensive Care Evaluation (NICE) registry was developed. The existing infrastructure of NICE allowed us to collect, process, and analyze data and adjust mortality for c ase-mix (13) and offered a means for providing practice audit and feedback to participating centers. Study Design and Population We designed a prospective multicenter cohort study to evaluate effects of the program in patients with sepsis and septic shock and other diagnoses both in participating and nonparticipating centers. Severe sepsis and/or septic shock definitions were based on modified international severe sepsis criteria (Addendum 2, Supplemental Digital Content 1, http://links.lww.com/CCM/ A923) (14). Only adult patients (≥ 18 yr) were eligible (first ICU www.ccmjournal.org
1891
van Zanten et al
Figure 1. Enrollment and outcomes. In nonparticipating ICUs, severe sepsis and septic shock Acute Physiology and Chronic Health Evaluation IV diagnosis categories (15) were used to select severe sepsis patients, and in participating centers, positive screened patients for severe sepsis were used to categorize patients.
Demographic Characteristics of Patients Over Study Time,a According to Study Groups in Control and Intervention ICUs Table 1.
All ICUsa
Nonparticipating ICUs
All Patientsa
All Patients
Patients Without Severe Sepsis or Septic Shock
Patients With Severe Sepsis or Septic Shock
213,667
115,602
107,571
8,031
63.6 (15.6)
63.2 (15.6)
63.0 (15.6)
66.1 (14.7)
59.3
60.2
60.5
56.3
Medical, %
40.6
38.3
35.4
77.1
Urgent surgery, %
13.8
14.1
13.9
17.4-
Elective surgery, %
45.6
47.6
50.7
5.5 79.0 (62.0–104.0)
Patients (n) Age (sd) Gender (male, %) ICU admission type
Severity of illness APACHE-IV score (median, 25–75%) APACHE-IV predicted mortality (%, 25–75%)
51.0 (36.0–70.0)
51.0 (36.0–70.0)
49.0 (35.0–67.0)
5.3 (1.7–19.2)
5.2 (1.7–18.8)
4.6 (1.6–15.4)
30.7 (14.3–57.9)
APACHE-IV = Acute Physiology and Chronic Health Evaluation IV prognostic model (15). a Patient data were collected for the first participating centers in July 2009 and data collection ended for all centers on January 1, 2013.
1892
www.ccmjournal.org
August 2014 • Volume 42 • Number 8
Clinical Investigations
admission during hospital stay), with a sepsis diagnosis less than 24 hours after ICU admission. Only patients with recorded Acute Physiology and Chronic Health Evaluation (APACHE)-IV (15) data were included, as APACHE-IV scores were used to adjust for severity of illness. Patient data and bundle adherence data were collected from July 2009 until January 1, 2013. Interventions and Outcome Measures To improve estimation of associations between duration of participation in the program and outcomes and to increase power, we included time explicitly in our models. Change in adjusted in-hospital mortality per time unit (month and program duration) was used to address the primary outcome. The secondary outcome measures were changes in adherence to individual and combined bundle targets over time and their association with in-hospital mortality. Two sets of bundle targets were used: six targets to be accomplished within 6 hours of diagnosis (resuscitation bundle) and four targets within 24 hours (management bundle) (Addendum 3, Supplemental Digital Content 1, http://links.lww.com/CCM/A923). We defined adherence as evidence that guideline bundle targets were implemented within indicated time frames (for time of diagnosis definitions, see Addendum 3, Supplemental Digital Content 1, http://links.lww.com/CCM/A923). Failure to adhere might occur either because physicians consciously did not attempt to meet targets or due to failure to reach targets despite the clinician’s attempt. We defined three scoring options: 1) target applied and achieved, 2) target not applicable, and 3) target not applied or target applied and not achieved. Not applicable was scored in case (relative) contraindications or clinical reasons not to apply targets were present (e.g., no refractory shock
and glucocorticoid administration). The category “targets not applied or not achieved” refers to patients with failed or too late attempts according to the attending physician. After withdrawal of activated protein C (aPC) based on recent trial results (16), ICUs were instructed to register this target as not applicable. An individual target adherence more than 80% (applied and not applicable combined) and a relative reduction of in-hospital mortality of 15% at the end of the study were aimed for (17). Data Acquisition Sepsis screening and bundle target information in participating ICUs and case-mix and in-hospital mortality data from both participating and nonparticipating ICUs were electronically sent to the NICE registry. All data prior to enrolment in participating ICUs were classified as nonparticipating data and were classified as participating data after enrolment. Data quality checks, onsite audits, and intensive data collector training sessions ensured high quality of the case-mix and mortality data. Quality of sepsis screening and bundle target data were not checked specifically. All data were entered into the database locally into predefined standardized fields documenting performance data. Identification fields were removed or encrypted before uploading to the NICE registry. Benchmark information was provided 2–4 times annually to participating ICUs. In The Netherlands, consent is not needed when registry data obtained from routine care and without patient-identifying data are used. The NICE registry is officially registered according to the Dutch Personal Data Protection Act.
Participating ICUs
All Patients
All Nonscreened Patients
All Screened Patients
Screened Patients Without Severe Sepsis or Septic Shock
Screened Patients With Severe Sepsis or Septic Shock
98,065
22,821
75,244
66,857
8,387
64.1 (15.6)
64.6 (14.7)
63.9 (15.8)
63.6 (16.0)
66.7 (14.2)
58.2
61.1
57.3
57.5
56.0
43.3
32.9
46.5
43.1
73.7
13.4
12.2
13.8
12.8
21.7
43.2
54.9
39.7
44.1
4.6 83.0 (65.0–106.0)
51.0 (36.0–71.0)
46.0 (32.0–64.0)
52.0 (37.0–73.0)
49.0 (35.0–67.0)
5.5 (1.6–19.6)
3.5 (1.0–12.9)
6.2 (1.9–21.8)
4.9 (1.6–16.0)
Critical Care Medicine
34.5 (17.0–61.7)
www.ccmjournal.org
1893
van Zanten et al
Statistical Analysis We used logistic generalized estimating equation (GEE) models to analyze effects of participation duration on in-hospital mortality and individual and combined target adherence. Odds ratios and 95% CIs of participation time in months were adjusted for age, gender, admission type, and severity of illness (APACHE-IV score). Analyses of effects of participation duration on target adherence were additionally adjusted for location of sepsis diagnosis (ICU, general ward, or emergency department [ED]). We transformed adjusted odds ratios (aORs) to observed absolute drops in in-hospital mortality by calculating average predictive differences in probability over time (18), reflecting average differences between predicted mortality estimates at two moments in time. To rule out general mortality reductions in all ICU patients, we compared time effects in screened severe sepsis patients at participating ICUs with time effects in all other groups (Fig. 1). To address changes in mortality risk, potentially induced by increased sepsis awareness during participation, we evaluated changes in severity of illness (APACHE-IV predicted mortality) of severe sepsis and septic shock patients admitted over time to participating ICUs. To analyze the effects of bundle adherence on in-hospital mortality, odds ratios and CIs were calculated while adjusting for age, gender, admission type, severity of illness (APACHE-IV score), and location of sepsis diagnosis. A separate logistic GEE model was used for (combined) bundles and for each target separately. We performed propensity score analysis for sepsis patients in which a propensity score, indicating the probability of belonging to participating or nonparticipating ICUs, was used as covariate in a GEE regression model along with APACHE-IV score, gender, age, admission type (i.e., medical or surgical), time, and a variable indicating the sepsis group. The propensity score itself was based on these same covariates, except for time and sepsis group indicator. Associations were considered statistically significant if 95% CIs did not contain 1.00. Data were analyzed using SPSS 19 (SPSS, Chicago, IL) and statistical environment R version 2.15.1.
RESULTS Study Group Characteristics From July 1, 2009, to January 1, 2013, in total 256,093 ICU patients who are 18 years old and older were admitted to one of the 83 ICUs participating in The NICE registry of in total 93 ICUs in the Netherlands. In total, 213,667 ICU patient records (83.4%) met the APACHE-IV inclusion criteria. Of these, 98,065 patients (45.9%) were admitted to an ICU that participated in the program at the time of admission. The first participating ICU was enrolled at July 1, 2009, and the last participating ICU was enrolled at September 9, 2012. In total, 52 ICUs participated for a median duration of 17.4 months (range, 5.5–42.0 mo) and included a median of 177 sepsis patients (range, 13–509). 1894
www.ccmjournal.org
Participating ICUs were mixed medical-surgical units located in university and teaching hospitals (n = 19) and nonteaching hospitals (n = 33). Nonparticipating ICUs were located in university and teaching hospitals (n = 14) and nonteaching hospitals (n = 16). Table 1 shows site and patient demographic characteristics of study groups. In total, 75,244 patients (76.7%) in participating ICUs were screened for severe sepsis of whom 8,387 (11.1%) were diagnosed with severe sepsis or septic shock (Fig. 1). Reasons for not screening patients at participating centers were not recorded. Predicted mortality in nonscreened patients was very low (3.5%). Not considered to have sepsis, failure to record, or missing variables were most likely reasons for nonscreening (A. R. H. van Zanten, personal communication, 2013). Locations of diagnosis were the ICU for 3,180 patients (38.5%), the ED for 2,808 patients (34.0%), and general wards for 2,278 patients (27.5%). Sepsis patient characteristics seem similar in participating and nonparticipating ICUs (propensity score nonsignificant [p = 0.47]). Changes in the In-Hospital Mortality Over Time In-hospital mortality of sepsis or septic shock patients decreased during participation in the program (aOR per month, 0.992 [0.986–0.997]), corresponding to an average adjusted absolute drop of 0.4% per quarter and 5.8% over 3.5 years (relative mortality reduction of 16.7% compared with baseline). A smaller (0.1% per quarter and 1.9% over 3.5 yr) significant effect was found among screened patients that were not diagnosed with sepsis (aOR per month, 0.995 [0.9906–0.9996]). In-hospital mortality did not decrease over time in patients treated in nonparticipating ICUs, patients with APACHE-IV diagnosis of severe sepsis in nonparticipating ICUs, and patients without severe sepsis in nonparticipating ICUs. Furthermore, there was also no reduction in in-hospital mortality of patients not screened for severe sepsis in participating ICUs (Table 2). Severity of illness expressed by APACHE-IV-predicted mortality of patients screened and diagnosed with septic shock did not change significantly during the study period (odds ratio per month, 0.999 [0.993–1.005]) (Fig. 2). Odds ratios for in-hospital mortality of sepsis patients at participating ICUs compared with APACHE-IV sepsis patients were still significant (aOR, 0.863 [0.772–0.964]), after adjusting for the propensity score along with other potential confounders. Change in Adherence to Bundle Targets Over Time Figure 2 depicts a significant increase in adherence to resuscitation bundle and management bundle targets. Adherence to individual targets within bundles increased over time, although the magnitude of improvement varied across targets. Table 3 shows detailed information on adherence in the first quarter of participation to individual and combined bundle targets, aORs of participation time (in months) for bundle adherence, and increase of target adherence rate over 3.5 years. Adherence to the resuscitation and management bundle significantly increased during participation time (aOR August 2014 • Volume 42 • Number 8
Clinical Investigations
Table 2. Target Mortality Rate and Odds Ratio of In-Hospital Mortality per Participation Montha Over Study Time and Effect on Absolute Mortality Rate in Participating and Nonparticipating ICUs
Patients at Risk
Mortality Rate in First Quarter of Study Period/Participation Period (%)
213,667
13.0
0.995 (0.994–0.997)
–1.8b
All patients
115,602
13.0
0.999 (0.996–1.002)
–0.4
Patients without diagnosis severe sepsis or septic shock
107,571
11.4
0.999 (0.996–1.002)
–0.3
8,031
33.8
0.997 (0.993–1.002)
–1.7
All patients
98,065
13.1
0.995 (0.992–0.997)
–2.0b
All nonscreened patients
22,821
13.0
1.004 (0.996–1.013)
1.2
All screened patients
75,244
13.1
0.994 (0.991–0.998)
–2.3b
Screened patients without diagnosis severe sepsis or septic shock
66,857
10.0
0.995 (0.9906–0.9996)
–1.9b
8,387
31.1
0.992 (0.986–0.997)
–5.8b
All patientsa
Odds Ratioa (95% CI) of Mortality per Month of Study Period/Participation Period
Absolute Mortality Rate Change Over 3.5 Yr (%)
Nonparticipating ICUs
Patients with severe sepsis or septic shock Participating ICUs
Screened patients with severe sepsis or septic shock
Patient data were collected for the first participating centers in July 2009 and data collection ended for all centers on January 1, 2013. In nonparticipating ICUs, data presented concern the period of July 2009 until January 1, 2013. Mortality rate change is calculated during the period from July 2009 to January 2013 for nonparticipating ICUs and during the participating period (which differs between ICUs) for the participating ICUs and adjusted for age, gender, admission type, and severity of disease (Acute Physiology and Chronic Health Evaluation IV score). b Significant change in mortality rate over 3.5 yr. a
per month, 1.018 [1.011–1.025] and 1.041 [1.027–1.054], respectively). Similar effects were also observed in all individual targets, except for optimization of fluid status. Average target adherence rates of individual targets over time are shown in Supplementary Appendix Table S1 (Supplemental Digital Content 1, http://links.lww.com/CCM/A923). All target adherence rates were above the goal of 80%, except for the mixed venous saturation measurement target performed in only 77.8% of patients. Relationship Between Bundle Targets and Hospital Mortality Adherence to the management bundle was associated with a decrease of in-hospital mortality (aOR, 0.82 [0.71–0.94]), whereas adherence to the resuscitation bundle and combined adherence to both resuscitation and management bundle were not significantly associated with in-hospital mortality (Table 4). Odds ratios of in-hospital mortality concerning individual bundle targets were nonsignificant, except for glucocorticoid administration, associated with a decrease in in-hospital mortality (aOR, 0.66 [0.46–0.93]) (Table 4). Critical Care Medicine
DISCUSSION This prospective national multicenter study, implementing a sepsis screening program and bundles based on guidelines according to the SSC, aimed to reduce mortality in patients with severe sepsis and/or septic shock by at least 15%. We showed a relative in-hospital mortality reduction of 16.7% over 3.5 years in patients admitted to participating centers. This effect was not observed in nonparticipating centers. We show that evidence against results was not due to inclusion of patients with lower mortality risk, as predicted mortality in screened severe sepsis patients did not change over time. Furthermore, we used propensity scores to adjust for possible differences between patients with severe sepsis and septic shock in participating and nonparticipating ICUs. In addition, we showed that mortality among patients in participating ICUs that were not screened according to the program did not reduce over time. This approach maximally reduced the limitations of an observational study and underlines the impact of the program on in-hospital mortality. Size of Mortality Reduction Observed mortality reductions are comparable to those found in other sepsis bundle implementation publications emphasizing the reliability and validity of the current study www.ccmjournal.org
1895
van Zanten et al
guideline target was not performed in 22.2% of patients over the study period, possibly due to no or later insertion (> 6 hr) of supradiaphragmal central venous catheters. In patients achieving the combined management bundle targets, we found significant positive effects on adjusted odds of mortality. Effect of Bundle Target Glucocorticoid Administration A significant effect of applying the glucocorticoids administration target in refractory septic shock was observed. This may fuel the ongoing debate on use of glucocorticoids in septic shock, as results reported are contradictory. Inclusion criteria of studies varied, and meta-analyses on studies performed draw different conclusions based on different methodological selection criteria (20–26). The glucocorticoid target in our study was introduced after publication of the Corticosteroid Therapy of Septic Shock trial (23), and ICUs were advised to use steroids only in patients with refractory septic shock, possibly selecting patients with higher risk of mortality. This might explain that 42.9% of patients were considered not applicable for use of glucocorticoids by attending physicians.
Figure 2. Proportion of adherence to the sepsis resuscitation, management and combined bundles, and predicted and observed actual and expected-observed mortality difference during 42 study months in 8,387 severe sepsis patients admitted to participating ICUs in The Netherlands from July 2009 to January 2013. A, The proportion of adherence to the resuscitation bundle, the management bundle, and both bundles in severe sepsis and septic shock patients as a function of participation duration in months for participating ICUs. For bundles target definitions, see Addendum 2 (Supplemental Digital Content 1, http:// links.lww.com/CCM/A923). B, The predicted and observed actual and expected-observed mortality difference in severe sepsis and septic shock patients as a function of participation duration in months for participating ICUs. Predicted mortality was estimated using Acute Physiology and Chronic Health Evaluation IV prediction models (15).
(7–11). Earlier data from The Netherlands, reporting on 863 Dutch patients from the SSC database, showed a decrease of in-hospital mortality from 52% at baseline to 35% after 2 years (p = 0.049) (19). These mortality data were not corrected for baseline mortality risk, and results were concerning periods before 2009. Mortality rate before the start of our present study was 34.8% and further reductions in mortality were achieved to a mean in-hospital mortality of 29% in participating hospitals. Bundle Target Adherence An improved bundle target adherence was noticed in participating centers with an average rate of adherence with guideline targets during the study varying from 77.8% to 99.1%, where we aimed for more than 80% for individual bundle targets. The mixed venous saturation measurement 1896
www.ccmjournal.org
Effect of Severe Sepsis Screening We observed positive effects of screening on in-hospital mortality both in patients with severe sepsis and to a lesser extent (1.9% lower mortality in 3.5 yr) in patients screened but without severe sepsis. This effect was not demonstrated in unscreened patients. This suggests that screening for severe sepsis potentially could benefit ICU patients with an ultimate other diagnosis. Possibly early resuscitation or increased awareness for elements in guidelines positively affect outcome in other diagnoses. Early goal–directed therapy (27) has been shown to reduce morbidity and hospital length of stay when applied to major surgery patients (28). Often these patients are treated in ICUs. We lack information on bundles in nonsepsis diagnoses, as targets were only recorded in sepsis patients. It is, however, unlikely that sepsis bundles have been applied to many patients with nonsepsis diagnoses. The beneficial effect of sepsis screening on in-hospital mortality in these patients warrants further study. Limitations ICUs entered the program on a voluntary basis, and this could potentially have introduced selection bias in terms of patient categories and/or delivered care. However, we were able to analyze data on severe sepsis and septic shock patients in nonparticipating ICUs using various covariates and were able to select severe sepsis and septic shock cases based on APACHE-IV diagnosis categories. We believe that the effects of selection bias are mitigated by several factors: first, we included 83 (all ICUs participating in the NICE registry) of, in total, 93 ICUs in The Netherlands in our analyses reflecting more than 90% of all ICU patients admitted to ICUs during the observation time. Second, ICUs were geographically distributed over The Netherlands. Third, we confirmed that patient mix was comparable in participating August 2014 • Volume 42 • Number 8
Clinical Investigations
Table 3. Guideline Target Adherence and Adherence Odds Ratio per Month of Participation in the Program and Increase of Absolute Adherence Rate Over Total Program Duration in Severe Sepsis Patients (n = 8,387) Target Adherence Rate in First Quarter of Participation (%)
Target Adherence Odds ratio (95% CI) per Month of Participation in the Program
Increase of Absolute Target Adherence Rate Over 3.5 Yr (%)a
Combined resuscitation and management targets
38.2
1.024 (1.016–1.031)
23.6
Combined resuscitation bundle targets
50.3
1.018 (1.011–1.025)
17.5
Combined management bundle targets
69.8
1.041 (1.027–1.054)
20.0
Blood cultures
83.4
1.012 (1.005–1.020)
6.6
Antibiotics
93.3
1.012 (1.004–1.020)
3.7
Lactate measurement
95.3
1.020 (1.002–1.038)
2.3
Mixed venous saturation measurement
65.0
1.019 (1.010–1.028)
14.6
Optimization fluid status
97.5
1.033 (0.990–1.079)
1.0
Vasopressors in fluid refractory hypotension
96.9
1.051 (1.031–1.072)
3.2
Protective mechanical ventilation
93.2
1.025 (1.008–1.043)
3.3
Activated protein C
80.5
1.061 (1.043–1.080)
17.7
Normoglycemia
91.6
1.032 (1.021–1.043)
5.8
Glucocorticoids
91.9
1.034 (1.020–1.048)
5.3
Resuscitation bundle
Management bundle
The absolute increase in target adherence rate is depicted compared with start of participation in the program (which differs between ICUs). Target adherence at the end of the observation period may be estimated by adding the first quarter data and absolute increase in adherence rate. Odds ratios per month of participation are adjusted for age, gender, admission type, severity of disease (Acute Physiology and Chronic Health Evaluation IV score), and location of sepsis diagnosis. All increases in target adherence rate were statistically significant except for optimization of fluid status. For bundles target definitions, see Addendum 2 (Supplemental Digital Content 1, http://links.lww.com/CCM/A923).
a
versus nonparticipating ICUs. Fourth, we performed rigorous case-mix adjustments. Fifth, propensity score analysis to adjust for selection bias due to lack of randomization showed that observed effects on mortality reduction in participating ICUs were not due to differences in patient characteristics in participating versus nonparticipating ICUs. In nonscreened patients in participating hospitals, there was no effect of the program on mortality, suggesting only effects in screened patients that were exposed to the studies intervention program. The effects of the program on mortality reduction demonstrated in participating ICUs might, in part, be attributable to the Hawthorne effect. This effect is potentially caused by sepsis screening, outcome and bundles performance feedback, or other unmeasured factors. Therefore, interpretation of the impact of individual bundle interventions should be done with caution. We believe, however, that the relatively long duration of the program mitigates the Hawthorne effect. Future Work New SSC guidelines have been published. The resuscitation bundle has been redesigned and comprises 3- and 6-hour targets. The management bundle has been dropped (29). Critical Care Medicine
Although we have demonstrated that elements of the management bundle may have been of relevance to the observed effects, we advise not to put major efforts on attempting to identify and justify individual components of sepsis bundles. We acknowledge, as mortality in sepsis still remains high, that early recognition and prompt resuscitation to maintain homeostasis—long standing mainstays of intensive care medicine—are the critical issues. Therefore, obtaining greater institutional participation rather than generating more complex and detailed protocols should now be our priority to improve outcome in patients with severe sepsis and septic shock admitted to our ICUs.
CONCLUSIONS Implementation of a national sepsis program resulted in improved adherence to sepsis guideline bundles in patients with severe sepsis and septic shock and was associated with reduced adjusted in-hospital mortality. Such reduction was not observed in patients in nonparticipating ICUs, suggesting direct impact of sepsis screening and bundle application on in-hospital mortality. www.ccmjournal.org
1897
van Zanten et al
Odds Ratio of In-Hospital Mortality Related to Bundle Target Adherence in Severe Sepsis Patients (n = 8,387) Table 4.
Odds Ratio (95% CI)a
Combined resuscitation and management targets
0.95 (0.84–1.08)
Combined resuscitation bundle targets
0.97 (0.85–1.11)
Combined management bundle targets 0.82 (0.71–0.94) Resuscitation bundle Blood cultures
0.99 (0.80–1.22)
Antibiotics
0.92 (0.73–1.17)
Lactate measurement
0.84 (0.66–1.07)
Mixed venous saturation measurement
0.97 (0.85–1.11)
Optimization fluid status
1.14 (0.70–1.83)
Vasopressors in fluid refractory hypotension
0.86 (0.52–1.43)
Management bundle Protective mechanical ventilation
0.88 (0.65–1.19)
Activated protein C
0.96 (0.79–1.15)
Normoglycemia
0.91 (0.69–1.21)
Glucocorticoids
0.66 (0.47–0.93)
Adjusted for age, gender, admission type, severity of disease (Acute Physiology and Chronic Health Evaluation IV score), and location of sepsis diagnosis. Odds ratios were calculated comparing percentages applied and not applicable versus not applied. For bundles target definitions, see Addendum 2 (Supplemental Digital Content 1, http://links.lww.com/CCM/A923).
a
ACKNOWLEDGMENTS We thank all participating hospitals, ICU physicians, ICU nurses, and other allied healthcare professionals, the initiators of VMS in The Netherlands, and the Dutch NICE registry for providing data.
REFERENCES
1. Angus DC, Linde-Zwirble WT, Lidicker J, et al: Epidemiology of severe sepsis in the United States: Analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001; 29:1303–1310 2. Moreno R, Afonso S, Fevereiro T: Incidence of sepsis in hospitalized patients. Curr Infect Dis Rep 2006; 8:346–350 3. Vincent JL, Sakr Y, Sprung CL, et al; Sepsis Occurrence in Acutely Ill Patients Investigators: Sepsis in European intensive care units: Results of the SOAP study. Crit Care Med 2006; 34:344–353 4. Townsend SR, Schorr C, Levy MM, et al: Reducing mortality in severe sepsis: The Surviving Sepsis Campaign. Clin Chest Med 2008; 29:721–733, x 5. Dellinger RP, Carlet JM, Masur H, et al; Surviving Sepsis Campaign Management Guidelines Committee: Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock. Crit Care Med 2004; 32:858–873 6. Dellinger RP, Levy MM, Carlet JM, et al: Surviving Sepsis Campaign: International guidelines for management of severe sepsis and septic shock: 2008. Intensive Care Med 2008; 34:17–60 7. Dellinger RP, Townsend SR, Linde-Zwirble WT, et al: The Surviving Sepsis Campaign: Results of an international guideline-based performance improvement program targeting severe sepsis. Crit Care Med 2010; 38:367–374
1898
www.ccmjournal.org
8. Ferrer R, Artigas A, Levy MM, et al; Edusepsis Study Group: Improvement in process of care and outcome after a multicenter severe sepsis educational program in Spain. JAMA 2008; 299:2294–2303 9. Barochia AV, Cui X, Vitberg D, et al: Bundled care for septic shock: An analysis of clinical trials. Crit Care Med 2010; 38:668–678 10. Castellanos-Ortega A, Suberviola B, García-Astudillo LA, et al: Impact of the Surviving Sepsis Campaign protocols on hospital length of stay and mortality in septic shock patients: Results of a three-year follow-up quasi-experimental study. Crit Care Med 2010; 38:1036–1043 11. Jeon K, Shin TG, Sim MS, et al: Improvements in compliance with resuscitation bundles and achievement of end points after an educational program on the management of severe sepsis and septic shock. Shock 2012; 37:463–467 12. Levy MM, Artigas A, Phillips GS, et al: Outcomes of the Surviving Sepsis Campaign in intensive care units in the USA and Europe: A prospective cohort study. Lancet Infect Dis 2012; 12:919–924 13. Bakhshi-Raiez F, Peek N, Bosman RJ, et al: The impact of different prognostic models and their customization on institutional comparison of intensive care units. Crit Care Med 2007; 35:2553–2560 14. Levy MM, Fink MP, Marshall JC, et al; SCCM/ESICM/ACCP/ATS/ SIS: 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31:1250–1256 15. Zimmerman JE, Kramer AA, McNair DS, et al: Acute Physiology and Chronic Health Evaluation (APACHE) IV: Hospital mortality assessment for today’s critically ill patients. Crit Care Med 2006; 34:1297–1310 16. Ranieri VM, Thompson BT, Barie PS, et al; PROWESS-SHOCK Study Group: Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med 2012; 366:2055–2064 17. VMS praktijkgids: Voorkomen van lijnsepsis en behandeling van ernstige sepsis. VMS Veiligheidsprogramma, 2009. Available at: http:// www.vmszorg.nl/_library/5539/web_2009.0101_praktijkgids_sepsis.pdf. Accessed March 18, 2013 18. Gelman A, Hill J: Chapter 5: Logistic regression. In: Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge, UK, Cambridge University Press, 2007, pp 101–104 19. Tromp M, Tjan DH, van Zanten AR, et al: The effects of implementation of the Surviving Sepsis Campaign in the Netherlands. Neth J Med 2011; 69:292–298 20. Bollaert PE, Charpentier C, Levy B, et al: Reversal of late septic shock with supraphysiologic doses of hydrocortisone. Crit Care Med 1998; 26:645–650 21. Briegel J, Forst H, Haller M, et al: Stress doses of hydrocortisone reverse hyperdynamic septic shock: A prospective, randomized, double-blind, single-center study. Crit Care Med 1999; 27:723–732 22. Annane D, Sébille V, Charpentier C, et al: Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA 2002; 288:862–871 23. Sprung CL, Annane D, Keh D, et al; CORTICUS Study Group: Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008; 358:111–124 24. Annane D, Bellissant E, Bollaert PE, et al: Corticosteroids in the treatment of severe sepsis and septic shock in adults: A systematic review. JAMA 2009; 301:2362–2375 25. Sligl WI, Milner DA Jr, Sundar S, et al: Safety and efficacy of corticosteroids for the treatment of septic shock: A systematic review and meta-analysis. Clin Infect Dis 2009; 49:93–101 26. Patel GP, Balk RA: Systemic steroids in severe sepsis and septic shock. Am J Respir Crit Care Med 2012; 185:133–139 27. Rivers E, Nguyen B, Havstad S, et al; Early Goal-Directed Therapy Collaborative Group: Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001; 345:1368–1377 28. Pearse R, Dawson D, Fawcett J, et al: Early goal-directed therapy after major surgery reduces complications and duration of hospital stay. Crit Care 2005; 9:R687–R693 29. Dellinger RP, Levy MM, Rhodes A, et al; Surviving Sepsis Campaign Guidelines Committee including the Pediatric Subgroup: Surviving sepsis campaign: International guidelines for management of severe sepsis and septic shock: 2012. Crit Care Med 2013; 41:580–637 August 2014 • Volume 42 • Number 8
