Continuing Medical Education Article

Signicance of arterial hypotension after resuscitation from cardiac arrest*

Stephen Trzeciak, MD, MPH; Alan E. Jones, MD; J. Hope Kilgannon, MD; Barry Milcarek, PhD; Krystal Hunter, MBA; Nathan I. Shapiro, MD, MPH; Steven M. Hollenberg, MD; R. Phillip Dellinger, MD; Joseph E. Parrillo, MD

LEARNING OBJECTIVES After participating in this educational activity, the participant should be better able to: 1. Dene post-resuscitation hypotension. 2. Explain impact of post-return of spontaneous circulation hypotension on prognosis. 3. Use this information in a clinical setting. Dr. Trzeciak has disclosed that he was the recipient of research funds/grants from Novo Nordisk. Dr. Kilgannon has disclosed that she is the recipient of a research fund/grant from the American College of Emergency Physicians Career Development. Dr. Shapiro has disclosed that he was the recipient of research funds/grants from Eli Lilly, Hutchinson Technologies, Istat, and Biosite; is the recipient of research funds/grants from Eli Lilly and Hutchinson Technologies; and was on the speakers bureau for Biosite and Hutchinson. Dr. Parrillo has disclosed that he was the recipient of research funds/grants from the Robert Wood Johnson Foundation and Salem Health Wellness Foundation; is a consultant/advisor for Artisan; and is a stock shareholder for DeepBreeze. The remaining authors have disclosed that they have no nancial relationships with or interests in any commercial companies pertaining to this educational activity. All faculty and staff in a position to control the content of this CME activity have disclosed that they have no nancial relationship with, or nancial interests in, any commercial companies pertaining to this educational activity. Lippincott CME Institute, Inc., has identied and resolved all faculty conicts of interest regarding this educational activity. Visit the Critical Care Medicine Web site (www.ccmjournal.org) for information on obtaining continuing medical education credit.

Objective: Expert guidelines advocate hemodynamic optimization after return of spontaneous circulation (ROSC) from cardiac arrest despite a lack of empirical data on prevalence of postROSC hemodynamic abnormalities and their relationship with outcome. Our objective was to determine whether post-ROSC arterial hypotension predicts outcome among postcardiac arrest patients who survive to intensive care unit admission. Design: Cohort study utilizing the Project IMPACT database (intensive care unit admissions from 120 U.S. hospitals) from 20012005. Setting: One hundred twenty intensive care units. Patients: Inclusion criteria were: 1) age >18 yrs; 2) nontrauma; and 3) received cardiopulmonary resuscitation before intensive care unit arrival. Interventions: None. Measurements and Main Results: Subjects were divided into two groups: 1) Hypotension Present one or more documented systolic blood pressure <90 mm Hg within 1 hr of intensive care unit arrival; or 2) Hypotension Absentall systolic blood pressure

>90 mm Hg. The primary outcome was in-hospital mortality. The secondary outcome was functional status at hospital discharge among survivors. A total of 8736 subjects met the inclusion criteria. Overall mortality was 50%. Post-ROSC hypotension was present in 47% and was associated with signicantly higher rates of mortality (65% vs. 37%) and diminished discharge functional status among survivors (49% vs. 38%), p < .001 for both. On multivariable analysis, post-ROSC hypotension had an odds ratio for death of 2.7 (95% condence interval, 2.53.0). Conclusions: Half of postcardiac arrest patients who survive to intensive care unit admission die in the hospital. Post-ROSC hypotension is common, is a predictor of in-hospital death, and is associated with diminished functional status among survivors. These associations indicate that arterial hypotension after ROSC may represent a potentially treatable target to improve outcomes from cardiac arrest. (Crit Care Med 2009; 37:28952903) KEY WORDS: heart arrest; cardiopulmonary resuscitation; resuscitation; shock; hemodynamics

Crit Care Med 2009 Vol. 37, No. 11

2895

udden cardiac arrest is the most common lethal manifestation of cardiovascular disease and a major public health burden (1, 2). Although most cardiac arrest research to date has focused on cardiopulmonary resuscitation (CPR) and methods to achieve return of spontaneous circulation (ROSC), recent research advances have identied that post-ROSC interventions can also have an important impact on outcome (3, 4). Accordingly, the post-ROSC phase of therapy is now considered to be a crucial link in the Chain of Survival paradigm for treating cardiac arrest (5, 6). However, the precise determinants of outcome in the postROSC period remain incompletely understood, and this knowledge gap represents a promising new arena for cardiac arrest research (7). After ROSC, the postcardiac arrest syndrome (8, 9) is characterized by profound but reversible myocardial stunning (10 12) and a robust systemic proinammatory response (13), which collectively can result in major hemodynamic perturbations. Although expert guidelines currently advocate hemodynamic optimization after ROSC from cardiac arrest (14), these recommendations are extrapolated from studies in other shock states (15) and are based solely on expert opinion (16). Furthermore, the strength of the association between post-ROSC hemodynamic alterations and outcome has not yet been quantied in a large cohort (16). Arterial hypotension, a hallmark of circulatory insufciency, is an important predictor of long-term mortality in heterogeneous critically ill populations (17, 18) as well as in patients with specic high-risk cardiovascular conditions, such as acute myocardial infarction (19) and pulmonary embolism (20), where the development of hypotension signies the need for escalation of therapy with emer*See also p. 2986. Assistant Professor (ST), Division of Critical Care Medicine and Emergency Medicine, Robert Wood Johnson Medical School, Cooper University Hospital, Camden, NJ; Assistant Research Director (AEJ), Department of Emergency Medicine, Carolinas Medical Center, Charlotte, NC; Assistant Professor of Emergency Medicine (JHK), UMDNJRobert Wood Johnson Medical School at Camden, Cooper University Hospital, Camden, NJ; Biostatistician (BM), Cooper University Hospital, Camden, NJ; Statistician II (KH), Cooper University Hospital, Camden, NJ; Research Director (NIS), Department of Emergency Medicine, Beth Israel Deaconess Medical Center, Boston, MA; Professor of Medicine (SMH), UMDNJ-Robert Wood Johnson Medical School, Camden, NJ; Director (SMH), Coronary Care,

gent interventions. In cardiac arrest, experimental data from Safar and colleagues identied that post-ROSC arterial pressure is a critical determinant of the degree of microcirculatory cerebral hypoperfusion that typically occurs in the post-ROSC state (2123); however, there is a paucity of data on the association between post-ROSC arterial pressure and outcome in humans. To date, the only available clinical data on this relationship has been limited to small single-center studies and the ndings have been conicting (11, 24 26), with two studies reporting that post-ROSC arterial hypotension is associated with worse neurologic outcome (25) or survival (24) and two studies reporting no association (11, 26). In this study, we quantify the prevalence of post-ROSC arterial hypotension and report the subsequent outcomes in a large multicentered cohort of postcardiac arrest patients who survive to intensive care unit (ICU) admission. The hypotheses are: 1) Among postcardiac arrest subjects alive at ICU arrival, the presence of post-ROSC hypotension (dened as a systolic blood pressure (SBP) 90 mm Hg within 1 hr of ICU arrival) is a common occurrence. 2) The presence of postROSC hypotension is associated with both an increased risk of in-hospital death and a decline in functional status between admission and discharge among survivors. 3) When adjusted for a predened set of confounding variables in a multivariable analysis, post-ROSC hypotension will remain signicantly associated with in-hospital death.

cine and specically designed for the critically ill patient. Currently, 201 adult ICUs from 131 U.S. hospitals participate in Project IMPACT. Participating institutions upload data quarterly to a central repository. Data elds include hospital and ICU organizational characteristics, admission source (e.g., Emergency Department vs. inpatient), demographics, physiologic data (including hemodynamic indices and laboratory values), procedures, complications, length of stay, and outcomes. All data are collected by dedicated on-site personnel who must be trained and certied by Project IMPACT in advance including passing a written certication examination to ensure uniformity in both database denitions and entry. On-site data collectors must also receive additional certication from Project IMPACT as a prerequisite to collating and uploading data.

Study Setting and Subjects

The ICUs in Project IMPACT represent a wide scope of practice environments including medical, surgical, and multidisciplinary ICUs. The institutions are heterogeneous in terms of hospital size, type (community vs. academic; private vs. public), and location (urban, suburban, or rural). Participating hospitals are not restricted to any particular geographic region. We included adult subjects who suffered nontraumatic cardiac arrest and were admitted to the ICU at a participating center from 20012005. Inclusion criteria were: 1) age 18 yrs; 2) nontrauma; and 3) received CPR 24 hrs before ICU arrival.

Data Collection
We abstracted the following variables: demographics, comorbidities, preadmission functional status, site of origin before ICU arrival, hospital characteristics, admission time of day (27), most abnormal physiologic parameters over the rst 24 hrs in the ICU, Simplied Acute Physiology Score (SAPS) II (28), life-support interventions (e.g., vasopressor and/or inotrope utilization), vital status at hospital discharge (alive/dead), and functional

METHODS
This study was approved by the Institutional Review Board at Cooper University Hospital, Camden, NJ. We analyzed Project IMPACT (Cerner, Bel Air, MD), a large administrative database developed by the Society of Critical Care Medi-

Cooper University Hospital, Camden, NJ; Professor of Medicine (RPD, JEP), UMDNJ-Robert Wood Johnson Medical School, Camden, NJ; Head (RPD), Division of Critical Care Medicine, Cooper University Hospital, Camden, NJ; Chief (JEP), Department of Medicine, Cooper University Hospital, Camden, NJ; Edward D. Viner Chair (JEP), Department of Medicine, Cooper University Hospital, Camden, NJ; and Director (JEP), Cooper Heart Institute, Cooper University Hospital, Camden, NJ. Dr. Trzeciaks effort on this project was supported, in part, by Grant K23GM83211 from the National Institutes of Health/National Institute of General Medical Sciences. Dr. Joness effort on this project was supported, in part, by Grant K23GM76652 from the National Institutes of Health/National Institute of General Medical Sciences. Dr.

Kilgannons effort on this project was supported, in part, by a grant from the Emergency Medicine Foundation. Dr. Trzeciak has previously received research support from Novo Nordisk, Eli Lilly, and Biosite, but he receives no personal remuneration from any commercial interest. The remaining authors have not disclosed any potential conicts of interest. For information regarding this article, E-mail: trzeciak-stephen@cooperhealth.edu Copyright 2009 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0b013e3181b01d8c

2896

Crit Care Med 2009 Vol. 37, No. 11

Table 1. Denitions for the Project IMPACT functional status categories Discharge Functional Status Category Independent Partially dependent

Denition Able to live at home requiring no assistance to complete activities of daily living Able to live at home, in a group home, or in a care facility and requiring some assistance to complete activities of daily living; the limitation(s) requiring assistance may be physical or mental Able to live at home or in a care facility and unable to perform activities of daily living; must be cared for by others; the limitation(s) requiring assistance may be physical or mental

Fully dependent

The functional status is assigned at the time of hospital admission and hospital discharge.

status at hospital discharge. Denitions for functional status categories appear in Table 1. We received the data in Access (Microsoft Corporation, Redmond, WA) and exported it to SPSS version 15.0 (SPSS, Chicago, IL) for analysis.

Outcome Measures and Data Analysis

For purposes of analysis, we dened two groups a priori according to the presence or absence of arterial hypotension at the time of ICU arrival: a) Hypotension Present one or more documented systolic blood pressure (SBP) value 90 mm Hg within 1 hr of ICU arrival; and b) Hypotension Absentall documented SBP values 90 mm Hg within 1 hr of ICU arrival. To provide mechanistic data on subjects in the Hypotension Present group, we abstracted the initial hemodynamic indices for patients who had a pulmonary artery catheter inserted. We dened cardiac index values as normal ( 2.5 L/min/m2) or low ( 2.5 L/min/m2), and we subclassied low cardiac index as being severely depressed if it was 2.0 L/min/m2 (29). The primary outcome measure was inhospital mortality. The secondary outcome measure was change in functional status (Table 1) from preadmission to hospital discharge among survivors. The change in functional status outcome measure was determined for all survivors who were not classied as fully dependent before admission. We compared the occurrence of the primary and secondary outcomes between the Hypotension Present and Hypotension Absent groups, using binomial test and 95% condence intervals (CIs) for difference in proportions. Continuous data are presented as mean standard deviation values or medians and interquartile ranges, and categorical data are reported as proportions and 95% CI where applicable. The KaplanMeier survival estimates and log-rank test for comparison were used for days-to-primary outcome analysis. Odds ratios were calculated to determine independent predictors of in-

hospital mortality, using stepwise multiple logistic regression. Given a dichotomous outcome (in-hospital death), a logistic regression model was t using maximum likelihood estimation. We considered all patient-oriented data to be potential candidate variables for the model. We did not include differences in hospital characteristics in the model. First, we performed bivariate analysis comparing survivors versus nonsurvivors. For all continuous variables we used independent samples t test or Wilcoxon rank-sum test as appropriate. For dichotomous variables we used Fishers exact test for difference in proportions. For multinomial variables we used Pearson 2. Candidate variables that had p .05 in bivariate analysis between survivors and nonsurvivors were entered into a multivariable logistic regression model. The regression model was run in four steps. The fourth step provides a signicance test, odds ratio, and a 95% condence interval around the odds ratio for the primary covariate of interest, post-ROSC hypotension. This yields an effect that is adjusted for all other variables included in earlier steps of the model. Steps one and two adjusted for pre-arrest factors. Step one included demographics age, sex, and race. Age and sex had signicant associations with mortality. For entry into the model, we coded age by deciles and coded race as white vs. nonwhite. Step two included prearrest patient characteristics other than demographics. Comorbidities were entered in the model as a total number (i.e. count) of comorbid conditions. Preadmission functional status was coded as independent vs. nonindependent. Site of origin prior to ICU admission was entered as Emergency Department vs. inpatient. Time of admission to the ICU was dichotomous and coded as night if the arrival time was between 11:00 P.M. and 6:59 A.M. Step three included postcardiac arrest physiologic data other than blood pressure. For vital signs, such as body temperature and heart rate, we coded each subject as being

above or below the median for each parameter. We did not include respiratory rate in the regression model because for mechanicallyventilated patients the respiratory rate is typically set by the clinician. Requirement of an inotropic agent was treated as a dichotomous variable in the model. On the grounds that multiple variables could be collinear with post-ROSC hypotension, we omitted the following variables from consideration for the model: Lowest systolic blood pressure over the rst 24 hours, lowest mean arterial pressure over the rst 24 hours, and vasopressor agent utilization. We also omitted the SAPS II from the model on the grounds that: 1) blood pressure is a component of the SAPS II score and therefore could be collinear with post-ROSC hypotension, 2) multiple other components of the SAPS II score were captured in the available data already entered into the model in steps 1-3, and 3) 16% of cases had missing SAPS II scores (because severity of illness score calculation is one of the few components of Project IMPACT database that are not mandatory elds for data abstractors) and we did not want to limit the sample size for the model. We performed three secondary analyses. First, to test the primary hypothesis in a subpopulation that had more subtle post-ROSC cardiovascular dysfunction and demonstrate that the ndings of the primary analysis were not only driven by the presence of overt shock, we compared the outcomes among subjects meeting the same criteria for post-ROSC hypotension (one or more documented SBP 90 mm Hg within an hour of ICU arrival vs. all SBP 90 mm Hg) after excluding all patients who had vasopressor-dependent shock (dened as vasopressor requirement at any time). Second, to assess whether or not effective cardiovascular support for the avoidance of additional hypotension over the rst 24 hrs in the ICU was associated with outcome, we analyzed the subgroup of Hypotension Present subjects who were treated with vasopressor agents and we compared in-hospital mortality among the subjects that had no additional SBP recordings 90 mm Hg over the rst 24 hrs in the ICU vs. subjects who had one or more additional SBP recordings 90 mm Hg in the ICU (based on the lowest SBP value recorded). Third, to test whether a hypertensive response in the post-ROSC phase was associated with outcome, we classied subjects according to the maximum mean arterial pressure (MAP) over the rst 24 hrs in the ICU. We dened a hypertensive response as an MAP 100 mm Hg (25). We divided the cohort into three MAP groups for analysis: 1) MAP 100 mm Hg; 2) MAP 80 to 99 mm Hg; and 3) MAP 80 mm Hg. We compared in-hospital mortality among the three groups by chi-square test, and performed pairwise comparisons between

Crit Care Med 2009 Vol. 37, No. 11

2897

Table 2. Baseline characteristics of the study subjects and hospital characteristics Hypotension Present (n 4092) 66 16 1874 (46) 3060 (75) 695 (17) 155 (4) 28 ( 1) 154 (4) Hypotension Absent (n 4644) 63 16 2141 (46) 3521 (76) 681 (15) 175 (4) 43 ( 1) 224 (5)

All Subjects (n 8736) Age, yrs (mean SD) Gender, female, n (%) Race, n (%) White/caucasian Black/African-American Latino/Hispanic Asian/Pacic Islander Other Preadmission functional status,b n (%) Independent Partially dependent Fully dependent Chronic comorbidities, n (%) Severe cardiovascular diseasec End-stage renal disease Hepatic cirrhosis (with portal hypertension) Cancer with metastatic disease Active chemotherapy Hematologic malignancy Acquired immune deciency syndrome Site of origin immediately before ICU arrival, n (%) Emergency department Inpatient Admitted to ICU at night (11 PM 6:59 AM), n (%) Hospital characteristics, n (%) Hospital sized Small to medium ( 300 beds) Large (301500 beds) Extra large ( 500 beds) Hospital type Community, nonacademic Academic, university-based Public Military Hospital location Urban Suburban Rural On-site cardiac catheterization laboratory 64 16 4015 (46) 6581 (75) 1376 (16) 330 (4) 71 (1) 378 (4)

pa .01 .80 .29 .01 .96 .21 .02

5882 (67) 1814 (21) 1040 (12) 1000 (11) 707 (8) 198 (2) 363 (4) 181 (2) 37 ( 1) 52 ( 1)

2562 (63) 981 (24) 549 (13) 541 (13) 394 (10) 121 (3) 226 (6) 122 (3) 22 ( 1) 21 ( 1)

3320 (71) 833 (18) 491 (11) 459 (10) 313 (7) 77 (2) 137 (3) 59 (1) 15 ( 1) 31 ( 1)

.01 .01 .01 .01 .01 .01 .01 .01 .12 .35

3792 (43) 4944 (57) 2184 (25)

1481 (36) 2611 (64) 1053 (26)

2311 (50) 2333 (50) 1131 (24)

.01 .01 .14

1455 (17) 3615 (41) 3666 (42) 6939 (79) 1561 (19) 178 (2) 58 ( 1) 4616 (53) 2765 (32) 1355 (15) 8342 (95)

630 (15) 1790 (44) 1672 (41) 3228 (79) 749 (18) 88 (2) 27 ( 1) 2070 (51) 1280 (31) 742 (18) 3940 (96)

825 (18) 1825 (39) 1994 (43) 3711 (80) 812 (18) 90 (2) 31 ( 1) 2546 (55) 1485 (32) 613 (13) 4402 (95)

.01 .01 .05 .24 .32 .48 .97 .01 .49 .01 .01

ICU, intensive care unit. a Comparing Hypotension Present and Hypotension Absent groups; bas dened in Table 1; cDened as baseline symptoms, such as angina or shortness of breath at rest or on minimal exertion (New York Heart Association Class IV) plus one or more of the following diagnoses: severe coronary artery disease; severe valvular heart disease; or severe cardiomyopathy; das dened according to the criteria from Halpern et al (41).

MAP groups with Bonferroni correction for multiple comparisons (i.e., for three groups, 0.05/3 0.017). We performed a sample size calculation to detect a 10% difference in the primary outcome between the Hypotension Present and Hypotension Absent groups. To detect a signicant difference with 0.05 and a 0.20, a minimum of 376 subjects in each group was needed.

RESULTS
A total of 8736 subjects from 120 hospitals met the inclusion criteria. The median number of cardiac arrest cases per hospital was 54 (interquartile range 23100). Subjects were predominantly Caucasian and from community nonacademic hospitals. The most common comorbid condition was severe cardiovascu-

lar disease (e.g., New York Heart Association Class IV) (Table 2). Of the 8736 subjects, 4092 (47%) were in the Hypotension Present group and 4644 (53%) were in the Hypotension Absent group. Of the 4092 subjects with Hypotension Present on ICU admission, 95% also had evidence of arterial hypotension (SBP 90 mm Hg, MAP 65 mm Hg, or vasopressor administration) over the rst 24 hrs in the ICU. Physiologic data in the ICU for both groups are displayed in Table 3. A total of 333 subjects in the Hypotension Present group underwent pulmonary artery catheterization. Initial median (interquartile range) hemodynamic indices were: central venous pressure 14 mm Hg (10 18 mm Hg); pulmonary capillary occlusion pressure 17 mm Hg (1323 mm Hg); cardiac index 2.5 L/min/m2 (1.9 3.5 L/min/m2). The initial cardiac index was low ( 2.5 L/min/m2) in 49% of all subjects and severely depressed ( 2.0 L/min/m2) in 28% of all subjects. Overall, 4375 (50%) of 8736 subjects met the primary outcome of in-hospital mortality. The Hypotension Present group had signicantly higher in-hospital mortality as compared with the Hypotension Absent group (65% vs. 37%; risk difference 28%; 95% CI, 26%30%; p .001). Among all survivors, 1653 (42%) of 3976 had a decline in functional status from preadmission to discharge. Survivors in the Hypotension Present group were signicantly more likely to experience a decline in functional status from preadmission to discharge (49%) as compared with Hypotension Absent (38%; risk difference 11%; 95% CI, 8%14%; p .001) (Table 4). The survival fractions for the two groups diverged signicantly over time by Kaplan-Meier survival analysis (p .001) (Fig. 1). Subjects in the Hypotension Present group were also more likely to receive repeat CPR in the ICU compared with the Hypotension Absent group (7% vs. 3%; risk difference 4%; 95% CI, 3% 5%; p .001). Ten risk factors for in-hospital death proved to be signicantly associated on multivariable logistic regression. Signicant demographic and prearrest factors entered in the rst two steps were age, male gender, total number of comorbid conditions, preadmission functional status, Emergency Department origin, and nighttime ICU admission. Signicant physiologic factors in the third step were high heart rate, low body temperature,
Crit Care Med 2009 Vol. 37, No. 11

2898

Table 3. Most abnormal physiologic parameters over the rst 24 hrs in the ICU and hemodynamic support requirements All Subjects (n 8736) Temperature, high C, median 37.7 (37.038.4) (IQR) Temperature, low C, median 36.1 (35.536.7) (IQR) Heart rate, high, beats/min, 112 (95130) median (IQR) Respiratory rate, high, breaths/ 24 (2029) min, median (IQR) Systolic blood pressure, low, 87 (73100) mm Hg, median (IQR) Mean arterial pressure, low, 60 (5170) mm Hg, median (IQR) Hemodynamic support, n (%) Vasopressor agentb 4536 (52) Dobutamine 675 (8) c SAPS II, median (IQR) 56 (3972) Hypotension Present Hypotension Absent (n 4092) (n 4644) 37.7 (36.938.4) 36.0 (35.136.6) 112 (95130) 24 (2029) 75 (6385) 53 (4461) 37.7 (37.138.3) 36.2 (35.736.7) 112 (95130) 24 (2030) 96 (85109) 66 (5876)

pa .01 .01 .01 .44 .01 .01

3030 (74) 450 (11) 56 (4073)

1506 (32) 225 (5) 56 (3972)

.01 .01 .01

IQR, interquartile range; SAPS, Simplied Acute Physiology Score. Comparing Hypotension Present and Hypotension Absent groups; bdened as initiation of a continuous infusion of any of the following drugs: dopamine, norepinephrine, epinephrine, phenylephrine; cseverity scoring was performed for 7031 (80%) of 8736 of all subjects, 3304 (81%) of 4092 of the Hypotension Present group, and 3726 (80%) of 4644 of the Hypotension Absent group.
a

tered and no additional SBP recordings 90 mm Hg for 24 hrs after ICU admission. The remaining 2759 (91%) experienced additional hypotension. Mortality was signicantly lower in the group with avoidance of additional hypotension (150 [55%] of 271 vs. 2035 [74%] of 2759; p .001). In the secondary analysis of patients with a hypertensive response (dened as a maximum MAP 100 mm Hg), 8503 (97%) of 8736 of subjects had maximum MAP data available. Of these, the inhospital mortality rates were 1925 (44%) of 4425 for MAP 100 mm Hg, 1440 (48%) of 2974 for MAP 80 to 99 mm Hg, and 822 (75%) of 1104 for MAP 80 mm Hg; p .001. On pairwise comparisons between groups, the mortality in the hypertensive group (MAP 100 mm Hg) was signicantly lower than in the MAP 80 to 99 mm Hg group; p .001.

DISCUSSION
This large, multicentered cohort study documents the frequency and signicance of post-ROSC arterial hypotension among cardiac arrest victims, measured at the time of ICU admission. We found that 47% of patients who survive cardiac arrest and are admitted to the ICU have post-ROSC hypotension, and two thirds of these patients do not survive to hospital discharge. The presence of post-ROSC hypotension at the time of ICU admission is associated with an approximate twofold risk of in-hospital death. Among survivors, patients with post-ROSC hypotension were signicantly more likely to experience a decline in functional status between hospital admission and discharge compared with patients without post-ROSC hypotension. This is the rst large multicentered study documenting the association between post-ROSC arterial pressure and outcome. The post-ROSC phase of therapy has recently emerged as a critical window of opportunity for impact on survival from cardiac arrest (3, 4, 6, 30). However, the particular factors associated with survival in the postcardiac arrest syndrome have been insufciently studied and remain poorly understood (7). Persistent circulatory insufciency is common in the postROSC state. Specically, arterial hypotension can result from a combination of factors including myocardial stunning, neurally mediated cardiac dysfunction, relative adrenal insufciency, and a profound proinammatory response related
2899

Table 4. Outcomes Hypotension Present (n 4092) 2672 (65) 1420 Hypotension Absent (n 4644) 1703 (37) 2941

All Subjects (n 8736) In-hospital mortality, n (%) Survivors, n Functional status of survivors at hospital discharge, n (%) Independent Partially dependent Fully dependent Decline in functional status on discharge compared with preadmission, n (%)b
a

pa .001

4375 (50) 4361

1840 (42) 1652 (38) 869 (20) 1653 (42)

460 (32) 616 (44) 344 (24) 627 (49)

1380 (47) 1036 (35) 525 (18) 1026 (38)

.001 .001 .001 .001

Comparing Hypotension Present and Hypotension Absent groups; bdetermined for all survivors who were not classied as fully dependent before admission (Hypotension Present, n 1288; Hypotension Absent, n 2688).

and inotrope requirement. Presence of post-ROSC hypotension was tested last and was found to be a signicant predictor of in-hospital death and have the highest numeric odds ratio among all the candidate variables in the model (odds ratio 2.69; 95% CI, 2.452.96) (Table 5). This is a unique predictive effect, observed after adjusting risks associated with all variables previously included in the model. In the secondary analysis excluding patients with vasopressor-dependent shock, 1287 (31%) of 4200 met the primary outcome of in-hospital mortality. The Hypotension Present group had signicantly higher rates of in-hospital morCrit Care Med 2009 Vol. 37, No. 11

tality as compared with the Hypotension Absent group (46% vs. 25%; risk difference 21%; 95% CI, 18%24%; p .001). Among all survivors, 976 (36%) of 2674 had a decline in functional status from preadmission to discharge. Survivors with Hypotension Present were signicantly more likely to experience a decline in functional status from preadmission to discharge (43%) as compared with Hypotension Absent (35%) (risk difference 8%; 95% CI, 3%13%; p .001) (Table 6). In the secondary analysis of Hypotension Present subjects who required vasopressor agent support (n 3030), 271 (9%) had vasopressor agents adminis-

Figure 1. Kaplan-Meier survival curves for patients with Hypotension Present and Hypotension Absent after return of spontaneous circulation from cardiac arrest (with censoring). The survival fractions diverged signicantly by log-rank test (p .001).

Table 5. Multivariable logistic regression analysis with in-hospital mortality as the dependent variable Odds Ratio (95% Condence Interval) 1.14 (1.111.18) 1.10 (1.011.21) 1.41 (1.281.56) 1.37 (1.251.50) 1.36 (1.241.50) 1.21 (1.101.35) 2.18 (1.992.40) 1.74 (1.591.90) 1.19 (1.001.41) 2.69 (2.452.96)

Variable Age, decile Male gender Nonindependent functional status on admission Total number of comorbidities Emergency Department origin Admission to ICU at night Heart rate in ICU, higha Temperature in ICU, lowb Inotrope requirement in ICU Post-ROSC hypotension

Coefcient 0.13 0.10 0.34 0.31 0.31 0.19 0.78 0.55 0.17 0.99

Sig .001 .04 .001 .001 .001 .001 .001 .001 .05 .001

ICU, intensive care unit; ROSC, return of spontaneous circulation; Sig, signicance. Highest value for rst 24 hrs in ICU (1 exceeds median; 0 median or lower); blowest value for rst 24 hrs in ICU (1 below median; 0 median or higher).
a

Table 6. Subgroup analysis of outcomes for subjects without vasopressor-dependent shock (dened as not requiring vasopressors at any time) All Subjects Hypotension Present Hypotension Absent (n 4200) (n 1062) (n 3138) In-hospital mortality, n (%) Total survivors, n Decline in functional status on discharge compared with preadmission, n (%)b
a

pa 0.001 0.001

1287 (31) 2913 976 (36)

487 (46) 575 227 (43)

800 (25) 2338 749 (35)

Comparing Hypotension Present and Hypotension Absent groups; bdetermined for survivors who were not classied as fully dependent before admission (total n 2674; Hypotension Present, n 527; Hypotension Absent, n 2147).

to global ischemia/reperfusion injury that induces a sepsis-like state (9, 13). Although consensus treatment recommendations advocate global hemodynamic optimization after ROSC (14), few empirical data exist in humans regarding the incidence and prognostic signicance of persistent circulatory insufciency after ROSC is achieved. Experimental data from Safar and colleagues identied that the post-ROSC state is characterized by patchy microcirculatory cerebral hypoperfusion, and post-ROSC arterial pressure is a major determinant of the degree of cerebral perfusion impairment (2123). A hypertensive surge was found to be protective, and a corollary to these ndings is that low arterial pressure in the early post-ROSC phase may be harmful (2123). Although post-ROSC hemodynamic alterations are known to be common in the postcardiac arrest syndrome (11, 31), few human studies have specically tested the hypothesis that post-ROSC arterial pressure is associated with outcome since these sentinel studies in animal models were performed more than a decade ago. Mullner et al reported that early post-ROSC hypotension was associated with worse neurologic outcome at 6 mos (25). Kilgannon et al reported that post-ROSC hypotension was associated with lower survival to hospital discharge (24). In contrast, Oddo et al recently reported that post-ROSC shock was not independently associated with outcome in a multivariable prediction model (26), and Laurent et al reported no signicant difference in the incidence of severe brain damage among subjects with and without post-ROSC hemodynamic instability (11). Importantly, all of these previous reports were single center studies with a relatively small sample. Our study is the rst large multicentered study to quantify the incidence of post-ROSC hemodynamic abnormalities and strength of association with outcome. By virtue of the potentially rapidly reversible nature of myocardial dysfunction in the early post-ROSC period (11), one could believe that the presence of postROSC hypotension is a benign transitory event in the course of recovery from a global ischemic insult with little bearing on long-term outcome. Alternatively, and similar to what has been reported in other cardiovascular emergencies (e.g., acute myocardial infarction or pulmonary embolism) in which the presence of arterial hypotension signies a high risk
Crit Care Med 2009 Vol. 37, No. 11

2900

of death and the need for emergent interventions (19, 20), the present study conrms our hypothesis that post-ROSC arterial hypotension is a strong predictor of in-hospital death among cardiac arrest victims who survive to ICU admission. We recognize that two philosophies might explain why arterial hypotension can precede death. First, reduced arterial pressure can potentially function as a mechanism of injury contributing to poor outcome (i.e., reduced perfusion pressure). Second, arterial hypotension may function as a signal for the presence of another underlying pathophysiologic derangement (e.g., impaired cardiac function). We submit that whether hypotension causes or marks the presence of poor perfusion, our results clearly establish that arterial hypotension is a critically important physiologic signal after resuscitation from cardiac arrest. Our nding that post-ROSC hypotension was also associated with diminished functional status among subjects who survived to hospital discharge suggests that post-ROSC hypotension could potentially be an important contributing factor to insults, such as the severity of anoxic brain injury, perhaps akin to the concept of secondary brain injury in hypotensive victims of head trauma (32). To our knowledge, this is the rst large multicentered study to demonstrate an association between hypotension after ROSC and lower functional status among survivors. Our subgroup analysis in patients who did not require vasopressors at any time indicates that the association between arterial pressure and outcome is not limited to patients with vasopressordependent shock, and is also present in those subjects with less obvious cardiovascular dysfunction. We acknowledge important limitations in interpreting our results. Although Project IMPACT is a large, robust, and clinically oriented database, this was a nonexperimental study and the observed epidemiologic associations with poor outcome do not necessarily indicate causality. One possibility is that the observed associations between post-ROSC hypotension and outcome are a reection of a more severe ischemic injury during the cardiac arrest event (10). In addition, the Project IMPACT database was designed from an ICU perspective, and thus it does not capture variables in the Utstein style (33) (e.g., initial cardiac rhythm, down time, CPR quality) specic to the cardiac arrest event that preCrit Care Med 2009 Vol. 37, No. 11

ceded the admission to the ICU. Although other multicentered cardiac arrest registries that are designed around the cardiac arrest event could provide more pertinent data for the study of CPR, the Project IMPACT database contains comprehensive data pertaining to the post-ROSC ICU course. The determinants of survival at the time of cardiac arrest and during CPR have already been studied extensively and are well established; however, the factors associated with survival after ROSC, particularly after arrival in the ICU, are poorly understood. Therefore, we submit that an ICU-based perspective is not only a novel approach to analysis but also can yield unique and valuable information. Another limitation worthy of mention is that our study did not capture whether or not therapeutic hypothermia was attempted as part of the post-ROSC treatment strategy. There are several suggestions, albeit indirect ones, that therapeutic hypothermia was likely not a signicant factor in the present cohort. First, only a small minority of subjects (6%) had a body temperature during the rst 24 hrs in the ICU at or below the typical target range (32C34C) for therapeutic hypothermia. Second, our study period began before the landmark studies of the efcacy of therapeutic hypothermia were published (3, 4). Third, published data indicated that therapeutic hypothermia had very poor penetration into routine clinical practice during the remainder of our study period that followed publication of the efcacy trials (34, 35). Additionally, our multivariable analysis indicated that lower body temperature (which would have theoretically been present in every patient treated with therapeutic hypothermia) was actually associated with higher mortality, and therefore the presence of very low body temperature in this study may have been a reection of postcardiac arrest impaired thermoregulation (which would likely indicate an unfavorable prognosis) rather than the effect of a therapeutic strategy. Additionally, we do not know if effective provision of therapeutic hypothermia would have modulated the association between post-ROSC hypotension and outcome observed in this study. Importantly, we also do not report which patients received coronary interventions that could have modulated the observed association between hypotension and outcome. We also acknowledge that subjects thought to be nonsalvageable by clini-

cians could have potentially received less aggressive care and thus be more likely to experience post-ROSC hypotension; however, this potential concern is attenuated by the fact that the Hypotension Present group was more than twice as likely to receive repeat CPR in the ICU. In addition, there may have been signicant heterogeneity in the time from ROSC to arrival in the ICU, where the experimental measurements were made. As early circulatory optimization has been shown to improve survival in other critically ill populations, most notably in patients with sepsis (15, 36), it is conceivable that circulatory optimization could also improve outcome in patients resuscitated from cardiac arrest (14). Recently, multiple studies have reported favorable outcomes with use of a standardized postresuscitation care protocol that includes a goal for optimization of post-ROSC arterial pressure (among multiple other post-ROSC interventions). The study from Sunde et al targeted an MAP 65 mm Hg whereas Oddo et al and Skulec et al used MAP target ranges of 75 to 80 mm Hg and 80 to 100 mm Hg, respectively (3739). Similarly, for patients with cardiogenic shock after ROSC, protocol-directed cardiovascular interventions (including intra-aortic balloon counterpulsation, if necessary) can improve post-ROSC hemodynamics; in conjunction with therapeutic hypothermia, such a strategy has been associated with favorable outcome (38, 40). Importantly, these recent papers demonstrate the feasibility of a treatment strategy for circulatory optimization after cardiac arrest. However, an experimental trial has not been performed; therefore, whether or not such a strategy actually confers a survival benet and the optimal range for post-ROSC arterial pressure are currently unknown (16). The epidemiologic associations found in the present study indicate that arterial hypotension after ROSC may represent a treatable target to improve outcomes from cardiac arrest and underscore the need for experimental trials of circulatory optimization after ROSC has been achieved.

CONCLUSIONS
This is the rst large multicentered study to document systematically the incidence and prognostic signicance of arterial hypotension after resuscitation from cardiac arrest. We found that the presence of post-ROSC hypotension is strongly associated with in-hospital death and that,
2901

among survivors, post-ROSC hypotension is associated with diminished functional status at hospital discharge. These associations raise the possibility that arterial hypotension after ROSC may represent a potentially treatable target to improve outcomes from cardiac arrest.

REFERENCES
1. Rosamond W, Flegal K, Friday G, et al: Heart disease and stroke statistics2007 update: A report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2007; 115: e69 e171 2. Zheng ZJ, Croft JB, Giles WH, et al: Sudden cardiac death in the United States, 1989 to 1998. Circulation 2001; 104:2158 2163 3. Bernard SA, Gray TW, Buist MD, et al: Treatment of comatose survivors of out-ofhospital cardiac arrest with induced hypothermia. N Engl J Med 2002; 346:557563 4. Hypothermia after Cardiac Arrest Study Group: Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002; 346:549 556 5. Cummins RO, Ornato JP, Thies WH, Pepe PE: Improving survival from sudden cardiac arrest: the chain of survival concept. A statement for health professionals from the Advanced Cardiac Life Support Subcommittee and the Emergency Cardiac Care Committee, American Heart Association. Circulation 1991; 83:18321847 6. Peberdy MA, Ornato JP: Post-resuscitation care: is it the missing link in the Chain of Survival? Resuscitation 2005; 64:135137 7. Gazmuri RJ, Nadkarni VM, Nolan JP, et al: Scientic knowledge gaps and clinical research priorities for cardiopulmonary resuscitation and emergency cardiovascular care identied during the 2005 International Consensus Conference on E and CPR Science With Treatment Recommendations: A consensus statement from the International Liaison Committee on Resuscitation (American Heart Association, Australian Resuscitation Council, European Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Council of Southern Africa, and the New Zealand Resuscitation Council); the American Heart Association Emergency Cardiovascular Care Committee; the Stroke Council; and the Cardiovascular Nursing Council. Circulation 2007; 116: 25012512 8. Negovsky VA: The second step in resuscitationThe treatment of the post-resuscitation disease. Resuscitation 1972; 1:17 9. Neumar RW, Nolan JP, Adrie C, et al: Postcardiac arrest syndrome: Epidemiology, pathophysiology, treatment, and prognostication. A consensus statement from the International Liaison Committee on Resuscitation (American Heart Association, Australian

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

and New Zealand Council on Resuscitation, European Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Council of Asia, and the Resuscitation Council of Southern Africa); the American Heart Association Emergency Cardiovascular Care Committee; the Council on Cardiovascular Surgery and Anesthesia; the Council on Cardiopulmonary, Perioperative, and Critical Care; the Council on Clinical Cardiology; and the Stroke Council. Circulation 2008; 118: 24522483 Chang WT, Ma MH, Chien KL, et al: Postresuscitation myocardial dysfunction: correlated factors and prognostic implications. Intensive Care Med 2007; 33:88 95 Laurent I, Monchi M, Chiche JD, et al: Reversible myocardial dysfunction in survivors of out-of-hospital cardiac arrest. J Am Coll Cardiol 2002; 40:2110 2116 Ruiz-Bailen M, Aguayo de Hoyos E, RuizNavarro S, et al: Reversible myocardial dysfunction after cardiopulmonary resuscitation. Resuscitation 2005; 66:175181 Adrie C, Adib-Conquy M, Laurent I, et al: Successful cardiopulmonary resuscitation after cardiac arrest as a sepsis-like syndrome. Circulation 2002; 106:562568 2005 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care: Part 7.5: Postresuscitation Support. Circulation 2005; 112(Suppl):IV-84 IV-88 Rivers E, Nguyen B, Havstad S, et al: Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001; 345:1368 1377 Jones AE, Shapiro NI, Kilgannon JH, et al: Goal-directed hemodynamic optimization in the post-cardiac arrest syndrome: A systematic review. Resuscitation 2008; 77:26 29 Jones AE, Aborn LS, Kline JA: Severity of emergency department hypotension predicts adverse hospital outcome. Shock 2004; 22: 410 414 Jones AE, Yiannibas V, Johnson C, et al: Emergency department hypotension predicts sudden unexpected in-hospital mortality: A prospective cohort study. Chest 2006; 130:941946 Lee KL, Woodlief LH, Topol EJ, et al: Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction. Results from an international trial of 41,021 patients. GUSTO-I Investigators. Circulation 1995; 91:1659 1668 Goldhaber SZ, Visani L, De Rosa M: Acute pulmonary embolism: clinical outcomes in the International Cooperative Pulmonary Embolism Registry (ICOPER). Lancet 1999; 353:1386 1389 Leonov Y, Sterz F, Safar P, et al: Hypertension with hemodilution prevents multifocal cerebral hypoperfusion after cardiac arrest in dogs. Stroke 1992; 23:4553 Safar P, Xiao F, Radovsky A, et al: Improved

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

cerebral resuscitation from cardiac arrest in dogs with mild hypothermia plus blood ow promotion. Stroke 1996; 27:105113 Sterz F, Leonov Y, Safar P, et al: Hypertension with or without hemodilution after cardiac arrest in dogs. Stroke 1990; 21:1178 1184 Kilgannon JH, Roberts BW, Reihl LR, et al: Early arterial hypotension is common in the post-cardiac arrest syndrome and associated with increased in-hospital mortality. Resuscitation 2008; 79:410 416 Mullner M, Sterz F, Binder M, et al: Arterial blood pressure after human cardiac arrest and neurological recovery. Stroke 1996; 27:59 62 Oddo M, Ribordy V, Feihl F, et al: Early predictors of outcome in comatose survivors of ventricular brillation and non-ventricular brillation cardiac arrest treated with hypothermia: A prospective study. Crit Care Med 2008; 36:2296 2301 Peberdy MA, Ornato JP, Larkin GL, et al: Survival from in-hospital cardiac arrest during nights and weekends. JAMA 2008; 299: 785792 Le Gall JR, Lemeshow S, Saulnier F: A new Simplied Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270:29572963 Hollenberg SM, Ahrens TS, Annane D, et al: Practice parameters for hemodynamic support of sepsis in adult patients: 2004 update. Crit Care Med 2004; 32:1928 1948 Langhelle A, Tyvold SS, Lexow K, et al: Inhospital factors associated with improved outcome after out-of-hospital cardiac arrest. A comparison between four regions in Norway. Resuscitation 2003; 56:247263 Laver S, Farrow C, Turner D, et al: Mode of death after admission to an intensive care unit following cardiac arrest. Intensive Care Med 2004; 30:2126 2128 American College of Surgeons Committee on Trauma: Advanced Trauma Life Support for Doctors Student Manual. 7th ed. Chicago, IL, American College of Surgeons, 2004 Jacobs I, Nadkarni V, Bahr J, et al: Cardiac arrest and cardiopulmonary resuscitation outcome reports: Update and simplication of the Utstein templates for resuscitation registries: A statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004; 110:33853397 Abella BS, Rhee JW, Huang KN, et al: Induced hypothermia is underused after resuscitation from cardiac arrest: A current practice survey. Resuscitation 2005; 64:181186 Merchant RM, Soar J, Skrifvars MB, et al: Therapeutic hypothermia utilization among physicians after resuscitation from cardiac arrest. Crit Care Med 2006; 34:19351940

2902

Crit Care Med 2009 Vol. 37, No. 11

36. Jones AE, Brown MD, Trzeciak S, et al: The effect of a quantitative resuscitation strategy on mortality in patients with sepsis: A metaanalysis. Crit Care Med 2008; 36:2734 2739 37. Oddo M, Schaller MD, Feihl F, et al: From evidence to clinical practice: effective implementation of therapeutic hypothermia to improve patient outcome after cardiac arrest. Crit Care Med 2006; 34:18651873 38. Skulec R, Kovarnik T, Dostalova G, et al:

Induction of mild hypothermia in cardiac arrest survivors presenting with cardiogenic shock syndrome. Acta Anaesthesiol Scand 2008; 52:188 194 39. Sunde K, Pytte M, Jacobsen D, et al: Implementation of a standardised treatment protocol for post resuscitation care after out-ofhospital cardiac arrest. Resuscitation 2007; 73:29 39 40. Hovdenes J, Laake JH, Aaberge L, et al: Ther-

apeutic hypothermia after out-of-hospital cardiac arrest: Experiences with patients treated with percutaneous coronary intervention and cardiogenic shock. Acta Anaesthesiol Scand 2007; 51:137142 41. Halpern NA, Pastores SM, Thaler HT, et al: Changes in critical care beds and occupancy in the United States 19852000: Differences attributable to hospital size. Crit Care Med 2006; 34:21052112

Crit Care Med 2009 Vol. 37, No. 11

2903