Documente Academic
Documente Profesional
Documente Cultură
Amisha V. Barochia, MBBS; Xizhong Cui, MD, PhD; David Vitberg, MD; Anthony F. Suffredini, MD;
Naomi P. O’Grady, MD; Steven M. Banks, PhD;† Peter Minneci, MD; Steven J. Kern, BS;
Robert L. Danner, MD; Charles Natanson, MD; Peter Q. Eichacker, MD
Context: Sepsis bundles have been developed to improve pa- data, there were consistent (I2 ⴝ 0%, p > .64) decreases in time
tient outcomes by combining component therapies. Valid bundles to antibiotics, and increases in the appropriateness of antibiotics
require effective components with additive benefits. Proponents (p < .0002 for both). In contrast, significant heterogeneity was
encourage evaluation of bundles, both as a whole and based on seen across trials for all other treatments (antibiotic use within a
the performance of each component. specified time period; administration of fluids, vasopressors, ino-
Objective: Assess the association between outcome and the tropes, and packed red blood cells titrated to hemodynamic goals;
utilization of component therapies in studies of sepsis bundles. corticosteroids and human recombinant activated protein C use)
Data Source: Database searches (January 1980 to July 2008) of (all I2 > 67%, p < .002). Except for antibiotics, sepsis bundle
PubMed, Embase, and the Cochrane Library, using the terms components are still being investigated for efficacy in randomized
sepsis, bundles, guidelines, and early goal directed therapy. controlled trials.
Data Extraction: Inclusion required comparison of septic adults Conclusion: Bundle use was associated with consistent and
who received bundled care vs. nonprotocolized care. Survival and significant improvement in survival and antibiotic use. Use of
use rates for individual interventions were abstracted. other bundle components changed heterogeneously across stud-
Main Results: Eight unblinded trials, one randomized and ies, making their impact on survival uncertain. However, this
seven with historical controls, were identified. Sepsis bundles analysis should be interpreted cautiously as these studies were
were associated with a consistent (I2 ⴝ 0%, p ⴝ .87) and unblinded, and only one was randomized. (Crit Care Med 2010; 38:
significant increase in survival (odds ratio, 1.91; 95% confidence 668 – 678)
interval, 1.49 –2.45; p < .0001). For all studies reporting such KEY WORDS: sepsis; septic shock; bundles; bundled care; treatment
C are bundles or protocols that lence of catheter-related infection or Although promising, care bundles
combine several medical prac- mortality in mechanically ventilated pa- pose challenges. Several care bundles
tices have been proposed as tients supported this approach (2, 3). posted on the Internet have not under-
tools to promote rapid adop- Care bundles have also been proposed gone rigorous peer review (11, 12). Some
tion of proven therapies, benchmark per- based on the holistic principle that the bundles addressing the same problem—
formance, and improve patient outcomes whole is greater than the sum of its parts sepsis— differ in content and compliance
(1). Reports that several practices insti- (4). Based on this therapeutic approach, rates (13). Many bundles lack strong ev-
tuted together could reduce the preva- the Institute for Healthcare Improvement idence for the efficacy of one or more of
and the Centers for Medicare and Medic- their individual components (14). Impor-
aid Services recently proposed instituting tantly, adoption of all bundle elements as
*See also p. 733. “all or none” performance measures (5, a single intervention limits the ability to
From the Critical Care Medicine Department (AVB, 6). Hospital performance, and possibly re- test the interdependent and independent
DV, XC, AFS, NPO, SJK, RLD, CN, PQE), Clinical Center,
National Institutes of Health, Bethesda, MD; and the imbursement in the future, may be based efficacy of individual components (15).
Department of Surgery (PM), The Children’s Hospital of on the frequency with which all elements Therefore, the introduction of care bun-
Philadelphia, University of Pennsylvania School of of a care bundle were administered to- dles may mandate changes in standard
Medicine, Philadelphia, PA. gether (7, 8). care without the ability to fully monitor
†Deceased.
This study was supported, in part, by The Intra- The Joint Commission and Institute the impact of component parts. Resolving
mural Research Program of the National Institutes of for Healthcare Improvement have recom- these issues has become critical as care
Health and the NIH Clinical Center, Bethesda, MD. mended that components of a bundle bundles have evolved from preventing
Dr. Banks, senior mathematical statistician, Critical should individually have proven benefit complications (e.g., catheter-related in-
Care Medicine Department, National Institutes of Health,
died unexpectedly during the final stages of writing this
and wide acceptance, and together have fections) to treating life-threatening con-
manuscript. Dr. Banks dedicated his life’s work to ad- even greater benefit (1, 9, 10). Impor- ditions (e.g., sepsis).
vancing science; his colleagues mourn his loss. tantly, bundle proponents encourage de- As bundle development and applica-
The authors have not disclosed any potential con- termination that individual components tion lack clear standards but are increas-
flicts of interest.
For information regarding this article, E-mail:
add to aggregate beneficial effects on out- ing in frequency, we examined this ap-
barochiaav@mail.nih.gov come (6). However, there is currently no proach for the treatment of sepsis. We
Copyright © 2010 by the Society of Critical Care consensus on methods and standards for performed a meta-analysis of clinical tri-
Medicine and Lippincott Williams & Wilkins the development and testing of valid care als, testing the impact of sepsis bundles
DOI: 10.1097/CCM.0b013e3181cb0ddf bundles. compared with nonprotocolized care.
Year published 2001 2006 2006 2006 2006 2007 2007 2008
Study designa Prospective, Before-afterb Before-afterb Before-afterb Before-after Before-afterb,c Before-after Before-afterb
randomized
Data collection
Controls 3/97–3/00 1/03–1/04 1/02–8/02 2/00–2/01 12/04–11–05 10/03–9/05 8/04–9/05 3/01–4/04
Bundled care 3/97–3/00 1/04–1/05 8/02–12/03 11/03–11/04 4/04–9/05 11/05–10/06 5/04–2/07
Setting ED ED/ICU ICU ED/ICU ED/ICU ED/ICU ED ED/ICU
Aids to facilitate Yes No No Yes Yes Yes Yes Yes
protocol cared
Study group imbalance No Yes Yes Yes No Yes NA⬁ No
in baseline criteria
APACHE II scorese
Control 20.4 ⫾ 7.4 25 ⫾ 10 31 (26/35) 25 ⫾ 10 22 ⫾ 7 30 ⫾ 11 NA 40 ⫾ 16
Bundled care 21.4 ⫾ 6.9 23 ⫾ 11 35 (30/37) 24 ⫾ 10 23 ⫾ 10 29 ⫾ 11 NA 42 ⫾ 18
NA, not available; ICU, intensive care unit; ED, emergency department
a
All trials were unblinded; bemployed subjects defined retrospectively by chart review; calso included nonrandomized contemporary controls;
d
specialized education of health workers, dedicated areas of care, sepsis medication carts, tool kits, flow, sheets, order sets, dedicated lines of communication
to sepsis experts, quality improvement programs or department head communications to increase compliance; eAcute Physiology and Chronic health
Evaluaiton (APACHE) II scores are reported as mean ⫾ SD, except in the Kortgen study, which reported median (25th/75th percentile) values.
Rivers et al (22) Trzeciak et al (23) Kortgen et al (24) Shapiro et al (25) Micek et al (26) Nguyen et al (27) Jones et al (28) El Solh et al (29)
ⱖ2 SIRS criteriaa Suspected or Patients with Suspected Suspected infection Suspected infection with Suspected infection Sepsisa
confirmed septic shocka infection with with ⱖ2 SIRS ⱖ2 SIRS criteriaa with ⱖ2 SIRS
sepsisa ⱖ2 SIRS criteriaa criteriaa
criteriaa
AND AND AND AND AND AND AND
SBP ⱕ90 mm Hg SBP ⬍90 mm Hg SBP ⬍90 mm Hg Vasodilatory shock with SBP ⬍90 mm Hg or MAP SBP ⱕ90 mm Hg SBP ⬍90 mm Hg
after 20–30 after 1500 mL after 20–30 MAP ⱕ65 mm Hg ⬍65 mm Hg after 20 mL/kg after adequate
mL/kg fluid fluid challenge mL/kg fluid requiring fluids and fluid challenge fluid challenge
challenge challenge vasopressors
OR OR OR OR OR
Lactate ⱖ4 Lactate ⬎4 Lactate ⱖ4 Lactate ⱖ4 and Decrease in SBP
need for ICU by ⬎40 mm
admission Hg from
baseline
SIRS, systemic inflammatory response syndrome; SBP, systolic blood pressure; MAP, mean arterial pressure; ICU, intensive care unit.
a
As defined by the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference (17).
Initial treatment time 0–6 Hrs Time in ED 0–6 Hrs 0–6 Hrs Time in ED Time in ED 0–6 Hrs 0–6 Hrs
Analyzed
Protocol treatments
Antibiotics No No Yes Yes Yes Yes Yes Yes
EGDT Yes Yes Yes Yes Yes Yes Yes Yes
Corticosteroids No No Yes Yes Yes Yes No Yes
rhAPC No No Yes Yes Yes No No Yes
Low tidal volume No No Yesa Yesa No No No No
Intensive insulin No No Yesb Yesb No No No Yesb
Nonprotocol treatments
reported
Antibiotics Yes Yes — — — — — —
Corticosteroids No Yes — — — — Yes —
rhAPC No Yes — — — Yes Yes —
Low tidal volume No No — — No No No Yesa
DVT prophylaxis No Yes No No No No No Yesc
Mortality end point In hospital In hospital 28 day 28 day 28 day In hospital In hospital 28 day
EGDT parameters MAP, CVP, Hct, NA NA Hct, ScvO2, NA CVP, Hb, ScvO2, CVP, ScvO2, NA
reported at baseline ScvO2, Lactate Lactate Lactate Lactated
EGDT parameters MAP, CVP, Hct, NA NA NA NA NA NA NA
reported after ScvO2, Lactate
treatment
ED, emergency department; EGDT, early goal-directed therapy; rhAPC, recombinant human activated protein C; DVT, deep venous thrombosis; MAP,
mean arterial pressure; CVP, central venous pressure; Hct, hematocrit; ScvO2, central venous oxygen saturation; NA, not available; —, treatment was
included in the protocol.
a
Patients were reported as being ventilated with lung protective ventilation, but actual mean tidal volumes were not reported; bdata not provided regarding
adequacy of glucose control, or mean blood glucose values in control and treatment groups; call patients were reported as receiving DVT prophylaxis, but no details
regarding dose or duration of medication used was provided; dScvO2 and CVP data not available for controls.
Antibiotics received by (time perioda) ⬍6 hrs — ⬍6 hrs ⬍6 hrs ⬍3 hrs ⬍4 hrs — ⬍4 hrs
Controls, % 92.4 NA 100 94.1 60 87.4 NA 91
Bundled care, % 86.3 NA 100 98.7 86.7 100 NA 95
Mean ⫾ SD time, hr to antibiotics
Controls NA 3.0 ⫾ 2.7 NA 2.7 ⫾ 2.0 NA 1.5 ⫾ 1.8 2.4 ⫾ 2.0 NA
Bundled care NA 2.5 ⫾ 1.7 NA 2.0 ⫾ 1.4 NA 1.0 ⫾ 1.2 1.7 ⫾ 1.2 NA
Appropriate antibiotics receiveda
Controls, % 94.3 NA 77.8b 88.2 71.7 NA NA 84
Bundled care, % 96.7 NA 85.7b 97.5 86.7 NA NA 97
Definition of appropriate antibiotics providedc No No No Yes Yes No No Yes
triggers for interventions (22, 25, 27, 28). (test for heterogeneity: I2 ⫽ 0%, p ⫽ .97) administration (in hrs) between bundle
Furthermore, for only one trial were the base- (Fig. 2). Overall, there was a statisti- and control patients was consistent (I2 ⫽
line data complete, and this study alone of the cally significant increase in the odds of 0%, p ⫽ .89). Time to antibiotics (hours
eight reported data showing how targeted pa- surviving with bundled care compared from time of admission) significantly de-
rameters changed with treatment (22). Three with controls (OR, 1.91; 95% CI, 1.49 – creased (weighted mean difference,
studies reported on the use of intensive insu- 2.45; p ⬍ .0001). ⫺0.58 hrs [⫺0.85 to ⫺0.33]; p ⬍ .0001)
lin therapy, and no study reported tidal vol- with bundled care. There was also across
umes or airway pressures during mechanical Components of Care five studies a consistent (I2 ⫽ 0%, p ⫽
ventilation (Table 4). .76) (Table 5 and Fig. 3) and significant
Consistently Increased With
Bundled Care Across Studies increase in the odds of receiving appro-
Survival priate antibiotics with bundled care com-
Across the eight studies, the effect of Across the four studies reporting such pared with controls (OR, 3.06; 95% CI,
bundled care on survival was consistent data, the difference in time to antibiotic 1.69 –5.53; p ⫽ .0002).
DISCUSSION
Figure 2. Effect of bundled care on the odds ratio of survival (95% confidence interval [CI]) for the Sepsis care bundles were associated
seven studies analyzed and overall. with consistent and significant increases
Figure 3. Effect of bundled care on antibiotic therapy including: A, antibiotics within a prespecified time period; B, mean time to antibiotics; and C,
percentage of patients receiving appropriate antibiotics. CI, confidence interval.
Fluidsa
Controls 3499 ⫾ 2438 3509 ⫾ 2312 2750 (1750/3750) 2871 ⫾ 1773 2825 ⫾ 1624 2807 ⫾ 2091 2540 ⫾ 2400 2490 ⫾ 1020
Bundled care 4981 ⫾ 2984 5685 ⫾ 3021 2450 (1625/3870) 4107 ⫾ 2590 3789 ⫾ 1730 2755 ⫾ 1477 4660 ⫾ 1800 3960 ⫾ 1990
Vasopressorsb
Controls, % 30.3 43.8 NA 45.1 100 43.9 34.2 NA
Bundled care, % 27.4 59.1 NA 79.7 71.7 50.6 68.8 NA
Inotropesb
Controls, % 0.8 0 0 1.9 NA 26.5 1.3 NA
Bundled care, % 13.7 9.1 20 7.6 NA 23.4 2.6 NA
PRBCb
Controls, % 18.5 0 16.7 5.8 6.7 12.6 1.3 13.0
Bundled care, % 64.1 13.6 16.7 10.1 20 14.3 5.2 14.0
Corticosteroidsc
Controls, % NA 31.3 43.3 23.5 50 16.2 6.3 16.0
Bundled care, % NA 36.4 100 29.1 21.7 29.9 40.3 95.0
rhAPCc
Controls, % NA 14.3 0 0 11.7 1.6 3.8 2.0
Bundled care, % NA 33.3 23.3 3.8 3.3 29.9 3.9 13.0
NA, not available; PRBC, packed red blood cells; rhAPC, recombinant human activated protein C.
a
Mean ⫾ SD volume in mL of fluids administered during the initial period analyzed, except Kortgen et al who reported median (25th/75th percentile)
values; b% of patients receiving these therapies during initial period analyzed; c% of patients receiving these therapies during similarly monitored periods.
Figure 4. Effect of bundled care on therapies based on set central venous pressure, mean arterial blood pressure, and central venous oxygen saturation goals
including: A, total crystalloid use; and B–D, the percentage of patients receiving vasopressors, inotropes, and packed red blood cells (PRBC), respectively.
CI, confidence interval.
Table 7. Summary for the therapies analyzed including the association between bundle use and their increase in mortality (32). Based on such
administration, the strength of recommendation and quality of evidence supporting their use in septic evidence supporting the critical need for
shock based on the GRADE system,a and whether randomized controlled trials (RCT) are under way early appropriate antibiotics in treating se-
or planned to determine the effectiveness of these therapies in septic shock
rious infections, the Surviving Sepsis Cam-
Consistent Association SSC 2008 Guidelines Grade paign gave this treatment a strong rating
Between Bundle Use (Grade 1B) in their updated 2008 guide-
and Administration Quality of On-going or lines (31). Importantly, use of early appro-
Therapy of Therapy Recommendation Evidence Planned RCTb priate antibiotics for sepsis meets the stated
requirement of the Joint Commission and
Antibiotics Institute for Healthcare Improvement that
Rapid antibiotics Yes Strong Moderate None
Appropriate Yes Strong Moderate None bundle components be proven and well-
antibiotics accepted interventions (1, 9).
Fluids for CVP No Strong Low ProCESS, ARISE, Resuscitation fluid volumes and the
8–12 mm Hg ProMISE percentage of patients receiving vasopres-
Vasopressors for MAP No Strong Low ProCESS, ARISE, sors were not consistently altered by bun-
ⱖ65 mm Hg ProMISE dled care. This heterogeneity was not re-
Inotropes for Scvo2 No Weak Low ProCESS, ARISE,
ⱖ70% ProMISE lated to any individual study. Although
PRBC for ScvO2 No Weak Low ProCESS, ARISE, bundles in each of these studies targeted
ⱖ70% ProMISE a central venous pressure goal of 8
Corticosteroids No Weak Low HYPRESS, mm Hg to 12 mm Hg for fluid adminis-
APROCCHS tration and an MAP of ⱖ65 mm Hg for
rhAPC No Weak Moderate PROWESS-SHOCK,
vasopressors, levels outside of this range
APROCHHS
were likely employed clinically in some
CVP, central venous pressure; MAP, mean arterial pressure; PRBC, packed red blood cells; rhAPC, patients and may have contributed to het-
recombinant human activated protein C. erogeneity. Notably, only one trial pro-
a
The GRADE system as employed in the SSC guidelines (31); bsee Table 8 for details regarding vided data regarding the levels of central
these trials. venous pressure and MAP actually
reached with bundled treatment and
these exceeded the stipulated goals in
in survival across eight studies. Two of Numerous studies over the past 20 yrs many patients (22).
three measures of antibiotic use were also have demonstrated improved outcomes Hemodynamic support with fluids and
consistently and significantly improved in life-threatening infections with early vasopressors is as important as antibiot-
across the studies reporting such data. In administration of appropriate antibiotics, ics in reducing mortality from septic
contrast, there was significant heteroge- and multiple clinical guidelines empha- shock (58). Nonetheless, differences in
neity in the effect of bundled care on the size such care (31–57). In a recent, large, physician practice and among patient
use of all remaining bundle components retrospective study in septic patients with populations could lead to heterogeneity
analyzed. Insulin therapy and lung pro- hypotension (n ⫽ 2731), every hour of in application of these interventions.
tective strategies were insufficiently re- delay in appropriate antibiotic adminis- There is considerable variation in the
ported and, therefore, not analyzed. tration was associated with a significant ranges of central venous pressure and
Protocolized Care for Multicentered, randomized, Infection, sepsis, refractory EGDT vs. Protocolized standard January 2010
Early Septic Shock open-label, active control, shock, hypotension care vs. standard care
(ProCESS)a NCT00510835 parallel assignment safety/
efficacy study
Australian Resuscitation Multicentered, randomized Early severe sepsis NA NA
Sepsis Evaluation controlled study
(ARISE)b
Protocolized Management Multicentered, randomized Early severe sepsis NA NA
of Sepsis, UK Based controlled study
(ProMISE)b
Hydrocortisone for Multicentered, randomized, Infection, sepsis, Hydrocortisone vs. placebo May 2011
Prevention of double-blind, placebo-controlled vasopressor dependent
Septic Shock parallel assignment safety/ shock
(HYPRESS)a NCT00670254 efficacy study
Efficacy and Safety of Multicentered, randomized, Infection, SIRS, rhAPC vs. placebo NA
rhAPC in Adults with double-blind, active controlled, vasopressor dependent
Septic Shock (PROWESS- placebo-controlled parallel shock
SHOCK)a NCT00604214 assignment safety/efficacy study
rhAPC and Corticosteroids Multicentered, randomized, double- Infection, vasopressor P-HC/FC ⫹ P-rhAPC vs. December 2010
for Human Septic Shock blind, placebo-controlled parallel dependent shock HC/FC ⫹ P-rhAPC vs.
(APROCCHS)a NCT00625209 assignment safety/efficacy P-HC/FC ⫹ rhAPC vs.
study HC/FC ⫹ rhAPC
NA, not available; SIRS, systemic inflammatory response syndrome; P, placebo; HC, hydrocortisone; HC/FC, hydrocortisone and fludrocortisone.
a
Ongoing; bplanned.
MAP which physicians believe should be Administration of PRBC and inotropes relate to variations in practice or patient
targeted in septic patients (59). Further- to obtain an ScvO2 of ⱖ70% was also not populations (70 –76). However, as ques-
more, a central venous pressure goal of 8 consistently altered with bundles over the tions persist regarding the risks and ben-
mm Hg to 12 mm Hg may be too low in eight studies. Although PRBC use be- efits of these therapies for sepsis, either
some septic patients or unnecessarily came consistent after removal of one trial when administered individually or to-
high in others (60 – 62). Many experts (22), it was not significantly different gether, they continue to undergo investi-
recommend fluid titration based on the with bundled care. In contrast, removal gation (77–79) (Tables 7 and 8). The Sur-
response to carefully monitored boluses, of another trial (27) did consistently in- viving Sepsis Campaign guidelines gave
rather than using arbitrary targets (63). crease inotrope use significantly. Of note, these therapies a weak recommendation
Although the Surviving Sepsis Campaign however, the efficacy of administering for use in patients with severe sepsis and
guidelines provide a strong recommenda- PRBC and inotropes to achieve an ScvO2 septic shock (Grade 2C for steroids, and
tion for these target numbers (Grade 1C), of ⱖ70% in patients with sepsis is un- 2B/2C for rhAPC). Although these agents
they also stipulate the importance of in- clear. How often this goal was reached may benefit some septic patients, until
dividualizing care (31). A single center with bundled care in seven of the eight such subgroups are clear, their inclusion
trial that used the same central venous trials analyzed is not reported. The 2008 in care bundles is inappropriate.
pressure and MAP targets in both study Surviving Sepsis Campaign guidelines Consistent use of earlier and appropri-
arms to titrate crystalloids and vasopres- give such treatment during the early re- ate antibiotics with care bundles could
sors is cited as the primary evidence to suscitation of septic shock a weak recom- plausibly have contributed to the consis-
support these targets (22). As such, these mendation (Grade 2C). Inotropic support tent increases in survival noted across
targets were not tested and their use is and PRBC transfusions targeted to ScvO2 these studies. However, other factors may
questioned (64 – 68). Importantly, the ef- are being further tested in ongoing ran- have also contributed, independent of
fectiveness of these targets is now being domized controlled trials to determine component therapies. Importantly, six of
reevaluated in large, multicentered ran- their efficacy in sepsis (69) (Tables 7 and 8). the trials described education or treat-
domized controlled trials (69) (Tables 7 Accordingly, at present, these interventions ment aids to improve bundle utilization
and 8). Although rapid fluid and vaso- do not meet Joint Commission and Insti- (22, 25–28). Consequently, unmeasured
pressor resuscitation is indispensable for tute for Healthcare Improvement criteria effects (e.g., earlier recognition of pa-
sepsis, optimal goals for such therapy for inclusion in a bundle (1, 9). tients requiring surgical intervention or
may differ in patients based on underly- Bundled care did not uniformly more readily available nonbundled ther-
ing medical conditions, and their use change low-dose corticosteroid and apies, such as respiratory support) may
should be individualized (67). rhAPC use across trials and this may also have changed outcomes. A large, multi-