Resuscitation 80 (2009) 30–34
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/resuscitation
Inter-hospital variability in post-cardiac arrest mortality?
Brendan G. Carra,b,c,d,e,∗, Jeremy M. Kahnd,e,f, Raina M. Merchanta,b,c,d,
Andrew A. Kramerg, Robert W. Neumarb,c
aThe Robert Wood Johnson Clinical Scholars Program, University of Pennsylvania School of Medicine (UPENN SOM), United States
bDepartment of Emergency Medicine, UPENN SOM, United States
cCenter for Resuscitation Science, UPENN SOM, United States
dThe Leonard Davis Institute of Health Economics, University of Pennsylvania, United States
eCenter for Clinical Epidemiology & Biostatistics, UPENN SOM, United States
fDivision of Pulmonary & Critical Care Medicine, Department of Medicine, UPENN SOM, United States
gCerner Corporation, Vienna, VA, United States
a r t i c l ei n f o
Received 14 May 2008
Received in revised form 25 August 2008
Accepted 3 September 2008
a b s t r a c t
Aim: A growing body of evidence suggests that variability in post-cardiac arrest care contributes to differ-
hospital-level variation in mortality of patients admitted to United States intensive care units (ICUs) with
a diagnosis of cardiac arrest.
Methods: Patients with a primary ICU admission diagnosis of cardiac arrest were identified in the
2002–2005 Acute Physiology and Chronic Health Evaluation (APACHE) IV dataset, a multicenter clinical
registry of ICU patients.
Results: We identified 4674 patients from 39 hospitals. The median number of annual patients was 33
per hospital (range: 12–116). Mean APACHE score was 94 (±38), and overall mortality was 56.8%. Age,
severity of illness (acute physiology score), and admission Glasgow Coma Scale were all associated with
department vs. the inpatient floor. Among institutions, unadjusted in-hospital mortality ranged from 41%
to 81%. After adjusting for age and severity of illness, institutional mortality ranged from 46% to 68%.
Patients treated at higher volume centers were significantly less likely to die in the hospital.
Conclusions: We demonstrate hospital-level variation in severity adjusted mortality among patients
admitted to the ICU after cardiac arrest. We identify a volume–outcome relationship showing lower
mortality among patients admitted to ICUs that treat a high volume of post-cardiac arrest patients.
Prospective studies should identify hospital-level and patient care factors that contribute to post-cardiac
© 2008 Elsevier Ireland Ltd. All rights reserved.
Overall survival after cardiac arrest is poor,1,2and has remained
largely unchanged over time.3,4Furthermore, despite international
efforts to develop and disseminate cardiac arrest treatment guide-
lines, significant outcome variability persists among individual
emergency medical service (EMS) systems and hospital. Reported
in the final online version at doi:10.1016/j.resuscitation.2008.09.001.
∗Corresponding author at: University of Pennsylvania School of Medicine, 932
Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021,
United States. Tel.: +1 215 573 3976; fax: +1 215 573 2265.
E-mail address: Brendan.Carr@uphs.upenn.edu (B.G. Carr).
of-hospital cardiac arrest2,5and 0% to 42% after in-hospital cardiac
tors contribute to this variability, much less is known about the
relative contribution of post-cardiac arrest care.
Post-cardiac arrest care is now recognized as a critical link in
the chain of survival.16,17Therapeutic hypothermia and formal-
ized post-cardiac arrest treatment protocols decrease morbidity
and mortality.18–21However, substantial inter-hospital variability
in mortality of patients that achieve initial return of spontaneous
been associated with hospital-based factors as well as patient care
factors.23,24Although likely to be a universal phenomenon, all of
these data are from sites outside the United States (US).
The goal of this study was to examine within the US health care
system the variability in mortality of patients that achieve initial
ROSC after cardiac arrest. Documenting inter-hospital variability of
0300-9572/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.
B.G. Carr et al. / Resuscitation 80 (2009) 30–34
post-cardiac arrest mortality is the first step in understanding the
role of post-cardiac arrest care in the overall chain of survival. Ulti-
care unit (ICU) database to determine site-dependent variability in
adjusted in-hospital mortality and the association between mor-
tality and hospital case volume.
Study design and patients
We conducted a retrospective cohort study using the Acute
Physiology and Chronic Health Evaluation (APACHE) IV database
(Cerner Corporation, Kansas City, MO). APACHE is an ICU clinical
information system used by participating hospitals within the US
ics, admission source, primary admission diagnosis, and detailed
laboratory and physiologic variables are collected in the first 24h
of ICU admission. Trained clinical coordinators supervised data
collection at each site. Data quality is ensured with extensive on-
site training, automated software that prevents implausible values,
and data audits that occur regularly at both the local and central
level. Hospitals participating in APACHE are diverse in size, owner-
ship, academic status and region of the country. APACHE contains
data from all types of ICUs within participating hospitals, includ-
ing medical, surgical, mixed medical–surgical, coronary care units,
and other specialty ICUs. The APACHE IV dataset has been utilized
in several recent ICU outcomes studies.25–27
All patients admitted to an APACHE hospital ICU with the diag-
nosis of cardiac arrest from 2002 to 2005 were eligible for this
analysis. ICU admission diagnosis is determined by the admit-
ting team in conjunction with the local APACHE data coordinator,
and represents the single diagnosis most responsible for the ICU
admission. We excluded patients less than 18 years of age, ICU
readmissions, and patients transferred into the ICU from outside
hospitals. Not all hospitals participated in APACHE for the entire
study period. To ensure that hospitals with short lengths of par-
ticipation did not overly influence the analysis, we also excluded
hospitals with less than 12 cardiac arrest patients per year or less
than 20 cardiac arrest patients total.
Variables and risk adjustment
The primary outcome variable was in-hospital mortality. Clini-
cal variables for risk adjustment included age, Glasgow Coma Scale
(GCS), whether or not the patient was mechanically ventilated,
and acute physiology score (APS), a composite severity of illness
measure strongly associated with in-hospital mortality.28The APS
is recorded upon admission to the ICU and includes physiologic
variables such as temperature, mean arterial pressure, heart rate,
respiratory rate, oxygen delivery, pO2, arterial pH, serum sodium,
serum potassium, serum creatinine, hematocrit, and white blood
cell count. Hospital-level variables included region of the country,
state into northeast, south, midwest and west. Teaching status was
defined by membership in the American Association of Medical
College’s Council of Teaching Hospitals (COTH) and was catego-
rized into academic (residents in the ICU plus COTH member),
community with housestaff (residents in the ICU without COTH
membership), and community (no residents in the ICU). Hospital
case volume was defined as the number of cardiac arrest patients
admitted to the ICU per year; hospital case volume was categorized
into natural cut-points based on the observed distribution.
Community with housestaff
Number of hospital beds384 [280–650]
Annual ICU admissions for cardiac arrest
Hospital volume (cases/year)
tile range], or range.
Patient and hospital characteristics are presented as fre-
quency (percent), mean±standard deviation, or median [range or
interquartile range], as appropriate. Multivariate logistic regres-
sion was used to determine factors independently associated with
in-hospital mortality and compare outcomes across hospitals as
described above. Generalized estimating equations with robust
ship to the primary outcome (in-hospital mortality). To determine
factors associated with mortality, all covariates were retained in
the final model. To compare outcomes between hospitals, only
patient-level covariates were included. Calibration and discrimi-
nation of this model were examined using the Hosmer–Lemeshow
test and the C statistic, respectively. Hospital-specific risk-adjusted
mortality was obtained from the regression results using condi-
tional standardization based on mean and modal values of the
model covariates.30All analyses were performed with Stata 9.2
(College Station, TX), and a two-tailed p-value of <0.05 was consid-
ered significant. This research used a de-identified dataset and was
considered exempt from human subjects review by the University
of Pennsylvania Institutional Review Board.
pitals. We excluded 215 patients in 18 very low volume hospitals,
resulting in 39 hospitals in the final sample (Table 1). Just over half
of the hospitals were academic hospitals (26%) or community hos-
pitals with housestaff (33%). Most hospitals were located in the
southeast (44%), west (28%), or midwest (21%). The median num-
ber of hospital beds at each facility was 384 (IQR 280–650), and the
median number of cardiac arrest admissions per hospital was 33
Of 5171 remaining patients, 4674 patients met inclusion cri-
teria. The average patient was aged 66 years, and just under half
(46.5%) were female. The average GCS on admission was 7 (3–15),
B.G. Carr et al. / Resuscitation 80 (2009) 30–34
APACHE III score
Acute physiology score
GCS on admission
Admission source (%)
ICU type (%)
Requiring mechanical ventilation (%) 89.3
Hospital discharge location (%)
Skilled nursing facility
APACHE=acute physiology and chronic health evaluation; GCS=Glasgow Coma
Scale; variables presented as percents, mean±standard deviation, or frequency (%).
and the average APACHE score was 94 (±38). Equal percentages of
patients (44.7%) were admitted from the ED and from the hospital
floor. Most patients (89.3%) required mechanical ventilation. Over-
all in-hospital mortality was 56.8%. 23.2% of the population were
discharged home, and 14% were discharged to a skilled nursing
facility. Additional demographic data are detailed in Table 2.
(Figure 1a). In-hospital mortality was similar for patients admitted
from the ED and from the hospital floor (56.8% vs. 56.2%, p=0.26).
Median length of stay for all patients was 2.8 days. Patients who
died during the admission had a median length of ICU stay of less
than 2 days (1.6). ICU length of stay was significantly longer for
survivors than nonsurvivors (3.8 days vs. 1.6 days; Table 3).
mortality variedwidely betweenhospitals
Unadjusted patient outcomes.
In-hospital mortality (%)
Admitted from ED
Admitted from other location
ICU length of stay (days)
ICU=intensive care unit; ED=emergency department; variables presented median
[interquartile range] or percents.
Figure 1. Frequency of in-hospital mortality among APACHE Hospitals: (a) unad-
justed rates; (b) rates adjusted for age, severity of illness and ventilation status.
tic=0.81) and calibration (p value for goodness of fit test=0.52).
Using the standardized adjusted mortality, inter-hospital variation
in mortality decreased somewhat but was still significant, ranging
from 46% to 68% (Figure 1b). Patient-level factors associated with
in-hospital mortality included age (OR for every 10 year increase
Figure 2. In-hospital mortality rate (mortality rates standardized by age, acute
physiology score, Glasgow Coma Scale on admission and ventilation status) vs.
annualized post-arrest volume: scatter plot of APACHE ICUs.
B.G. Carr et al. / Resuscitation 80 (2009) 30–34
Results of the logistic regression model for in-hospital mortality for patients admit-
ted to the ICU after cardiac arrest.
VariableOR 95% CIs
Age (per 10 units)
Acute physiology score (per 10 units)
GCS on ICU admission (per 1 point)
Mechanical ventilation on ICU admission
Hospital teaching status
Hospital volume (cases/year)
GCS=Glasgow Coma Scale; ICU=intensive care unit; ED=emergency department;
OR=odds ratio; CI=confidence interval. Odds ratios are adjusted for all listed vari-
ables; confidence intervals and p values adjusted for clustering by center using
generalized estimating equations.
increase 1.33, 95% CI 1.29–1.37), and GCS on ICU admission (OR for
every point increase 0.94, 95% CI 0.92–0.96) (Table 4). The need
for mechanical ventilation on admission was also associated with
mortality (1.85, 95% CI 1.33–2.54).
Standardized mortality as a function of hospital volume is
demonstrated graphically in Figure 2. Increasing annual ICU case
by poor outcomes at some small hospitals. Relative to hospitals
that treated fewer than 20 cardiac arrest patients per year, a non-
significant trend towards decreasing mortality was observed for
hospitals that treated 20–34 (OR 0.78, 95% CI 0.55–1.11), and 35–50
(OR 0.71, 95% CI 0.45–1.11) patients per year. Hospitals that treated
more than 50 cardiac arrest patients per year had a significantly
lower mortality than hospitals that treated fewer than 20 patients
per year (OR 0.62, 95% CI 0.45–0.86). No association was observed
between in-hospital mortality and type of hospital (academic vs.
community with housestaff vs. community).
We demonstrate inter-hospital variability in severity adjusted
post-cardiac arrest mortality in patients admitted to the ICU
after successful resuscitation from cardiac arrest. In addition, we
observed an inverse relationship between the volume of cardiac
arrest patients treated in the ICU and in-hospital mortality. These
findings are consistent with previous work that has demonstrated
and the relationship between experience (volume) and outcomes
in other critically ill patient populations.27,31–33
Increasingly, attention has focused on the importance of post-
cardiac arrest care on patient outcomes. Specific critical care
interventions, most notably induced hypothermia, have been
demonstrated to decrease mortality and morbidity,18–21and are
recommended in current treatment guidelines.34However, imple-
mentation remains poor.35,36Additional post-arrest therapies
that have been associated with improved outcome include early
percutaneous coronary intervention,20,37fever prevention,38and
been implicated as a source of variable outcomes.16This study is
the first to demonstrate inter-ICU variability in severity adjusted
post-cardiac arrest mortality within the US health care system.
The variability in post-cardiac arrest mortality observed in our
been reported in a number of studies outside the US over the past
vs. 67% (p<0.001) for patients admitted to two different Swedish
hospitals within the same EMS system between 1980 and 1996.23
Langdhalle et al. reported in-hospital mortality ranged from 44% to
66% for patients admitted to four Norwegian hospitals after out-of-
hospital cardiac arrest between 1995 and 1999.8It is interesting to
note that the hospital with the lowest in-hospital mortality rate in
Most recently, Herlitz et al. reported that 1-month mortality after
hospital admission following out-of-hospital cardiac arrest ranged
from 58% to 86% among 21 Swedish hospitals.22
Our analysis has several limitations. We use a large proprietary
database that provides detailed clinical information. This dataset
allows us to perform high-quality severity adjustment, but the
included hospitals uniformly have demonstrated a commitment to
quality assurance by participating in the APACHE clinical informa-
tion system. If anything, this might serve to reduce inter-hospital
variation in outcome, making it more likely that the observed vari-
ation is real. While this may influence the overall survival in our
cohort, there is no reason to believe that bias of the included ICUs
would differ by volume of patients with cardiac arrest. An addi-
tional selection bias may exist in that we had no details about the
individual resuscitation or the organization of code teams within
A number of pre-arrest and intra-arrest variables known to be
associated with outcome were not controlled for due to limitations
tributed to outcome variability. However, we believe that we were
able to adequately adjust for case-mix using age, GCS, acute phys-
iology score, and mechanical ventilation, all of which were found
to be independent predictors of mortality. In addition, we had no
access to procedures performed in the post-cardiac arrest period.
apies associated with improved outcomes including therapeutic
hypothermia or cardiac catheterization. While these interventions
could potentially explain some of the variability between hospi-
tals demonstrated in our analysis, the goal of our analysis was to
describe the variability that exists in outcomes and to demonstrate
the relationship between volume and outcome. The causal path-
way for post-arrest mortality is complex and the contribution of
individual interventions to decrease mortality will need to be rig-
orously tested to explain the variability that we describe. Future
work should be directed at examining processes of care that might
explain the observed variation, such as therapeutic hypothermia
and other advances in critical care for these patients.
Our outcomes are limited to in-hospital mortality, length of
hospital stay, and hospital disposition. We could not differentiate
the effect of do not resuscitate orders on in-hospital mortality. It
is unknown if post-arrest management practices with respect to
withdrawing care differ by hospital type, ICU type, or region of the
country, and this is fertile ground for future research. Local and
regional discharge practices could also contribute to variability in
rates of in-hospital death. That is, post-arrest patients who die in
who are transferred to inpatient or home hospice would not have
34 Download full-text
B.G. Carr et al. / Resuscitation 80 (2009) 30–34
been identified. This variability may be accounted for by insurance,
family resources, or other factors. Data were not available regard-
ing long term outcomes and we cannot comment on the long-term
morbidity and mortality associated with cardiac arrest.
arrest, and one potential source of outcome variability is patient
volume. These results suggest the need for additional research to
delineate best-practices and to optimize post-cardiac arrest care. If
variability in patient care is found to be causal, then development,
dissemination and implementation of comprehensive post-cardiac
arrest care guidelines should be considered. Furthermore, the
tive evaluation of regional post-arrest care centers.40
Conflict of interest
BGC & JK have no pertinent disclosures. RMM serves on the BLS
Subcommittee of the American Heart Association Emergency Car-
diac Care Committee as a fellow. AAK is an employee of Cerner
Corporation and owns shares of Cerner stock. RWN serves on the
ACLS Subcommittee of the American Heart Association Emergency
Cardiac Care Committee.
2. Rosamond W, Flegal K, Furie K, et al. Heart disease and stroke statistics—2008
update: a report from the American Heart Association Statistics Committee and
Stroke Statistics Subcommittee. Circulation 2008;117:e25–146.
3. Herlitz J, Bang A, Gunnarsson J, et al. Factors associated with survival to hospital
discharge among patients hospitalised alive after out of hospital cardiac arrest:
change in outcome over 20 years in the community of Goteborg, Sweden. Heart
4. Niskanen M, Reinikainen M, Kurola J. Outcome from intensive care after car-
diac arrest: comparison between two patient samples treated in 1986–87 and
1999–2001 in Finnish ICUs. Acta Anaesthesiol Scand 2007;51:151–7.
for victims of out-of-hospital cardiac arrest. Ann Emerg Med 1999;34:517–25.
6. Sandroni C, Nolan J, Cavallaro F, Antonelli M. In-hospital cardiac arrest: inci-
dence, prognosis and possible measures to improve survival. Intens Care Med
7. Juchems R, Wahlig G, Frese W. Influence of age on the survival rate of out-of-
hospital and in-hospital resuscitation. Resuscitation 1993;26:23–9.
8. Langhelle A, Tyvold SS, Lexow K, Hapnes SA, Sunde K, Steen PA. In-hospital fac-
tors associated with improved outcome after out-of-hospital cardiac arrest. A
comparison between four regions in Norway. Resuscitation 2003;56:247–63.
9. Iwami T, Hiraide A, Nakanishi N, et al. Outcome and characteristics of out-of-
population-based study in Osaka, Japan. Resuscitation 2006;69:221–8.
10. Eisenburger P, Sterz F, Haugk M, et al. Cardiac arrest in public locations—an
independent predictor for better outcome? Resuscitation 2006;70:395–403.
11. Vukmir RB. Survival from prehospital cardiac arrest is critically dependent upon
response time. Resuscitation 2006;69:229–34.
12. Finn JC, Jacobs IG, Holman CD, Oxer HF. Outcomes of out-of-hospital car-
diac arrest patients in Perth, Western Australia, 1996–1999. Resuscitation
in Sweden. Eur Heart J 2001;22:511–9.
MS, Mengert TJ.Cardiac resuscitation.N EnglJ Med
14. Nadkarni VM, Larkin GL, Peberdy MA, et al. First documented rhythm and clin-
ical outcome from in-hospital cardiac arrest among children and adults. JAMA
15. Eisenberg MS, Copass MK, Hallstrom AP, et al. Treatment of out-of-hospital car-
diac arrests with rapid defibrillation by emergency medical technicians. N Engl
J Med 1980;302:1379–83.
of Survival? Resuscitation 2005;64:135–7.
17. Adrie C, Laurent I, Monchi M, Cariou A, Dhainaou JF, Spaulding C. Postresusci-
tation disease after cardiac arrest: a sepsis-like syndrome? Curr Opin Crit Care
18. Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors
of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med
19. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac
arrest. N Engl J Med 2002;346:549–56.
20. Sunde K, Pytte M, Jacobsen D, et al. Implementation of a standardised treat-
ment protocol for post resuscitation care after out-of-hospital cardiac arrest.
tice: effective implementation of therapeutic hypothermia to improve patient
outcome after cardiac arrest. Crit Care Med 2006;34:1865–73.
22. Herlitz J, Engdahl J, Svensson L, Angquist KA, Silfverstolpe J, Holmberg S. Major
differences in 1-month survival between hospitals in Sweden among initial
survivors of out-of-hospital cardiac arrest. Resuscitation 2006;70:404–9.
23. Engdahl J, Abrahamsson P, Bang A, Lindqvist J, Karlsson T, Herlitz J. Is hospital
the same Emergency Medical Service and admitted to one of two hospitals over
24. Keenan SP, Dodek P, Martin C, Priestap F, Norena M, Wong H. Variation in length
of intensive care unit stay after cardiac arrest: where you are is as important as
who you are. Crit Care Med 2007;35:836–41.
Health Evaluation (APACHE) IV: hospital mortality assessment for today’s criti-
cally ill patients. Crit Care Med 2006;34:1297–310.
length of stay: benchmarking based on Acute Physiology and Chronic Health
Evaluation (APACHE) IV. Crit Care Med 2006;34:2517–29.
27. Kahn JM, Goss CH, Heagerty PJ, Kramer AA, O’Brien CR, Rubenfeld GD. Hos-
pital volume and the outcomes of mechanical ventilation. N Engl J Med
28. Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system.
Risk prediction of hospital mortality for critically ill hospitalized adults. Chest
29. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous out-
comes. Biometrics 1986;42:121–30.
30. Altman DG. Practical statistics for medical research. London: Chapman & Hall;
31. Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The
empirical relation between surgical volume and mortality. N Engl J Med
32. Halm EA, Lee C, Chassin MR. Is volume related to outcome in health care? A
systematic review and methodologic critique of the literature. Ann Intern Med
33. Dudley RA, Johansen KL, Brand R, Rennie DJ, Milstein A. Selective refer-
ral to high-volume hospitals: estimating potentially avoidable deaths. JAMA
34. Postresuscitation support, Circulation 2005;112:84–8.
is underused after resuscitation from cardiac arrest: a current practice survey.
36. Merchant RM, Soar J, Skrifvars MB, et al. Therapeutic hypothermia utiliza-
tion among physicians after resuscitation from cardiac arrest. Crit Care Med
37. Spaulding CM, Joly LM, Rosenberg A, et al. Immediate coronary angiography in
survivors of out-of-hospital cardiac arrest. N Engl J Med 1997;336:1629–33.
38. Zeiner A, Holzer M, Sterz F, et al. Hyperthermia after cardiac arrest is associ-
39. Oksanen T, Skrifvars MB, Varpula T, et al. Strict versus moderate glucose control
after resuscitation from ventricular fibrillation. Intens Care Med 2007.
40. Lurie KG, Idris A, Holcomb JB. Level 1 cardiac arrest centers: learning from the
trauma surgeons. Acad Emerg Med 2005;12:79–80.