Post–Cardiac Arrest Mortality Is DecliningClinical Perspective
A Study of the US National Inpatient Sample 2001 to 2009
Background—Despite several advances in postresuscitation care over the past decade, population-based mortality rates for patients hospitalized with cardiac arrest in the United States have not been studied over this time period. The aim of this study was to determine the annual in-hospital mortality rates of patients with cardiac arrest from 2001 to 2009.
Methods and Results—The US mortality rates for hospitalized patients with cardiac arrest were determined using the 2001 to 2009 US National Inpatient Sample, a national hospital discharge database. Using the International Classification of Diseases, 9th Edition, code 427.5, we identified patients hospitalized in the United States with cardiac arrest from 2001 to 2009. The main outcome measure was in-hospital mortality. A total of 1 190 860 patients were hospitalized with a diagnosis of cardiac arrest in the United States from 2001 to 2009. The in-hospital mortality rate decreased each year from 69.6% in 2001 to 57.8% in 2009. In multivariable analysis, when controlling for age, sex, race, and comorbidities, earlier year was a strong independent predictor of in-hospital death. The mortality rate declined across all analyzed subgroups, including sex, age, race, and stratification by comorbidity.
Conclusions—The in-hospital mortality rate of patients hospitalized with cardiac arrest in the United States decreased by 11.8% from 2001 to 2009.
Out-of-hospital cardiac arrest in the United States is frequently lethal. The overall case fatality rate for all patients assessed by emergency medical technicians is ≈94%.1 Yet, recent mortality rates have been reported in the range of 40% to 60% for patients who survive the initial resuscitation and get admitted to an intensive care unit.2–4
Over the past decade there have been several advances in cardiopulmonary resuscitation (CPR) protocols and immediate postresuscitation care. New guidelines put forth by the American Heart Association in 2005 emphasize high-quality CPR and the importance of early use of automated external defibrillators (AEDs) for shockable cardiac arrhythmias.5 In addition, results from observational studies have suggested that some interventions, such as immediate coronary intervention, may improve survival in certain populations.6,7
Over the past decade, the highest level of evidence for a treatment that improves outcomes came from 2 randomized trials published in 2002 that showed that induced hypothermia reduces mortality and improves neurological outcomes in patients with out-of-hospital ventricular fibrillation cardiac arrest.8,9 Primarily on the basis of these trials, therapeutic hypothermia (TH) is now considered standard-of-care in the United States for patients who remain comatose after resuscitation for this particular patient population.5
Clinical Prospective on p 550
To our knowledge, in-hospital mortality rates for patients with cardiac arrest in the US population have not been studied over the previous decade after the establishment of these therapies. We hypothesized that the mortality of patients hospitalized with cardiac arrest has decreased since 2001, before the era of routine use of TH in post–cardiac arrest patients. The aim of our study was to assess the yearly mortality rates from 2001 to 2009 of US patients hospitalized with cardiac arrest using the US National Inpatient Sample (NIS) database.
The NIS is a hospital discharge database that represents 20% of all inpatient admissions to nonfederal, nonrehabilitation hospitals in the United States. Using the NIS database from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality, Rockville, MD for the years 2001 to 2009, we identified all patients with a diagnosis of cardiac arrest (code 427.5 in the International Classification of Diseases, 9th Edition, Clinical Modification [ICD-9-CM]). We stratified patients by age (1, patients <65 years of age; 2, patients 65 to 79 years of age; and 3, patients ≥80 years of age), sex, and race (white, black, Hispanic, Asian/Pacific Islander). Comorbidities were assessed using the Charlson Comorbidity Index (stratified 1, 2, 3, ≥4).10 We also performed a search of the NIS for the ICD-9-CM diagnostic codes of cardiac catheterization; coronary arteriography (37.21, 37.22, 37.23, 88.52, 88.53, 88.54, 88.55 88.56, and 88.57) and ICD-9-CM procedural codes for percutaneous transluminal coronary angioplasty (00.66, 17.55, 36.01, 36.02, 36.05, 36.06, 36.07) for patients hospitalized with cardiac arrest each year from 2001 to 2009. The primary end point was in-hospital mortality.
χ2 tests were used to compare categorical variables, and the Student t test was used to compare continuous variables. To obtain national estimates and account for sampling, weights were applied as indicated in the Healthcare Cost and Utilization Project-NIS Calculating NIS Variances Guide. Weights for each discharge are based on hospital characteristics (ie, bed-size, location, teaching status, and ownership). We performed multivariate logistic regression analysis to determine predictors of in-hospital mortality. In this analysis we included the following variables: year of cardiac arrest, age group, sex, race, and Charlson Comorbidity Index. When considering odds of mortality by year of cardiac arrest, we used the year 2005 as the reference year because data after that year would reflect the practice after publication of the American Heart Association guidelines recommending the use of TH along with most other current standards of care for the management of patients who need cardiopulmonary resuscitation. All variables in the multivariate logistic regression analysis were categorical variables. An assessment of the mortality pattern across each subcategory of age group, sex, race, and comorbidity was performed within logistic regression models by adding interaction terms (ie, interaction between calendar year and dummy variables created for respective variables of interest) to a model containing main effects (ie, calendar year and dummy variables for respective variables of interest). A statistically significant p-value for interaction term would indicate that the mortality pattern is different for various categories of a variable over the 9-year period. A p-value in this study needs to be interpreted with caution given the very large sample size allowing for a detection of minor deviations in mortality patterns. Statistical analysis was performed using the SAS-based statistical package JMP 9.0 (www.jmp.com).
From 2001 to 2009, a total of 1 190 860 patients were hospitalized with a cardiac arrest, with a mean of 132 318 patients affected per year (SD 5336, range 127 205–141 256). There were 642 907 men (54%) and 547 739 women (46%). In-hospital mortality declined steadily from 69.6% in 2001 to 57.8% in 2009. The trend in mortality rates over the study period is shown in Figure 1 and Table 1. In-hospital mortality rate was 59.5% (279 826 of 470 244) for patients <65 years of age, 64.3% (266 082 of 413 554) for patients 65 to 79 years of age, and 74.6% (228 942 of 306 752) for patients ≥80 years of age. Mortality rates declined across all age groups, sexes, races, and comorbiditiy levels. These data are shown in Table 2 and in Figure I in the online-only Data Supplement. The decline in mortality across subgroups was not statistically significantly different for age groups (P=0.6226) or sex (P=0.7582). There were statistically significant differences in the mortality decline for the subgroups of race (P<0.0001) and comorbidity index (P<0.0001). The mortality declined despite a rise in the proportion of patients with high comorbidity indices over time, and the decline in mortality was more pronounced in patients with more comorbidities (Table I in the online-only Data Supplement).
The proportion of patients who were treated at large volume centers increased slightly over the study time period, from 64.4% in 2001 to 67.4% in 2009, but this was not a consistent year-to-year increase. The percentage of patients who underwent coronary catheterization and percutaneous transluminal coronary angioplasty also inconsistently trended upward by approximately 3% and 2%, respectively, over the 9 years. These data are depicted inFigure 2.
In multivariate analysis, when controlling for age, sex, race, and comorbidities, the odds of in-hospital mortality associated with cardiac arrest in 2009 were significantly lower than the odds of mortality in 2005 (odds ratio [OR], 0.69; 95% confidence interval [CI], 0.67–0.70; P<0.0001). The odds of in-hospital mortality associated with cardiac arrest in 2001 were significantly higher than the odds of mortality in 2005 (OR, 1.11; 95% CI, 1.09–1.14; P<0.0001). The associations identified in the multivariate analysis are summarized inTable 3.
The mortality rate of hospitalized patients with cardiac arrest in the United States has consistently decreased since 2001. After controlling for medical comorbidity and age, the year of cardiac arrest remained a strong predictor of in-hospital mortality. We found that patients hospitalized with cardiac arrest had a decreasing likelihood of in-hospital mortality each year from 2001 to 2009. Chance of survival was increased by 40% in the year 2009 compared with 2005. There are several possibilities to explain these findings, and we cannot know with certainty from these data which specific factor (or more likely, combination of factors) drove this reduction in mortality. Changes in the care of these patients, rather than a change in the population itself, likely led to this decline in mortality. In fact, over the study time period the proportion of patients with high comorbidity indices increased, indicating that patients undergoing resuscitation in the later years were less healthy compared with the population in the earlier years of the study period.
Although we cannot prove a causal relationship because of the retrospective observational design of this study, it is plausible that the implementation of TH for comatose post–cardiac arrest patients over this time period has contributed to the improved survival rates. Results from 2 recent studies using large intensive care unit databases in Finland and in the Netherlands indicate that TH is contributing to declining cardiac arrest mortality rates. These studies found an absolute risk reduction of mortality of cardiac arrest patients of ≈6.7% from pre-TH to post-TH time periods.3,11 In a study of 3958 patients with cardiac arrest in Finnish intensive care units, mortality decreased from 57.8% to 51.1% when comparing 2 time periods (2000–2002 versus 2003–2008). TH was used in approximately 36.2% of the second group.3 In a retrospective study using a large Dutch database, mortality decreased from 72% to 65.4% before and after the introduction of TH protocols.11
Although it is likewise possible that the use of TH for post–cardiac arrest patients contributed to the 11.8% absolute decrease in mortality in our study, this interpretation is limited because we do not know the number of US hospitals sampled by the NIS that routinely use TH for post–cardiac arrest care. Unfortunately it is not possible to reliably identify patients in the NIS treated with hypothermia over this time period because the diagnostic ICD-9 code 780.65 for hypothermia (not associated with low environmental temperature) was not implemented until October 2008. A procedural hypothermia code (99186) has existed but was proposed for deletion by the American Medical Association at a Current Procedural Terminology Editorial meeting in June 2008 because it was not payable, outdated, and rarely used.10 Thus it is difficult to estimate how frequently and consistently TH for post–cardiac arrest is used in US hospitals. Overall, the implementation of TH in the United States. has been relatively slow, particularly before the publication of the 2005 American Heart Association guidelines.
Other factors may have contributed to the decreased mortality rate over this time period, and trying to determine the relative contribution of each factor may be a difficult–if not impossible–task. For instance, it is possible that the increased survival could be attributable in part to the regular use of post–cardiac arrest management protocols, which leave less chance for human errors.12 We do not have information on how many hospitals in our study had protocols introduced during this time period. Other changes in post–cardiac arrest care also could have contributed to the reduction in mortality. Although there has not been a prospective randomized trial evaluating its effect, immediate percutaneous coronary intervention has been associated with improved survival in some studies of patients with out-of-hospital cardiac arrests.6,7 In an observational study using data from a Parisian registry, in which 86% of patients were treated with TH and 68% had ventricular fibrillation or ventricular tachycardia as initial rhythm, successful coronary angioplasty was associated with a 2-fold increase in the likelihood of survival to hospital discharge. The proportion of patients with cardiac arrest who underwent percutaneous transluminal coronary angioplasty in our study increased by 2% over the entire study period, and although this certainly may have contributed to the mortality reduction, we do not think that the magnitude of change in the use of emergency percutaneous transluminal coronary angioplasty is sufficient to be considered the primary contributor to the more pronounced and consistent decline in mortality.
Other potential contributing factors also deserve consideration. In 2005, new American Heart Association guidelines for CPR emphasized the importance of uninterrupted chest compressions and high-quality CPR. However, whether this has translated to improved survival rates is unclear, especially because, since the guideline publication, there is evidence showing that the quality of CPR delivered by trained laypersons often does not achieve guideline parameters.13 The guidelines also highlighted the importance of rapid AED use for patients with shockable arrhythmias (ventricular fibrillation or pulseless ventricular tachycardia), including adequate training of lay rescuers to use AEDs. The annual number of AEDs sold increased nearly 3-fold from 2001 to 2005, so there was likely a considerable increase in availability of public AEDs, at least during the early years of our study period.14 The overall impact of this, however, is unknown because >75% of all out-of-hospital cardiac arrests occur in homes.15 In addition, data from a large US cardiac arrest registry funded by the Centers for Disease Control and Prevention indicate that only 3.7% of patients with out-of-hospital cardiac arrest (OHCA) from 2005 to 2010 had an AED used before arrival of emergency medical service personnel. Data from the Resuscitation Outcomes Consortium showed that application of an AED does increase the odds of survival after OHCA by nearly 2-fold, and 7% of all survivors received a shock from an AED before arrival of emergency medical services. However, only 2% of all patients with OHCA in this study had an AED applied by bystanders before arrival of emergency medical services, and 1.2% had an AED-delivered shock.16 Given these findings, it is unlikely that an increasing number of public AEDs are the sole cause of the decline in mortality seen in our study, but it remains a possible contributor.
Our study is limited because we do not have information about initial cardiac rhythm, location of cardiac arrest, and the duration or quality of CPR. The effect of other unavailable factors, including some of those mentioned before, cannot be assessed. It is also important to notice that the decreased mortality reported in this analysis only applies to patients who survived the initial cardiac arrest and got to be hospitalized and does not account for the patients who died in an emergency department or before reaching a hospital.
A major strength of our study is the large national population included and the reflection of real-world practice. We did not limit inclusion to patients admitted to an intensive care unit, which may have been a reason that the improvement in survival in our study was greater than that previously reported in other series.
In summary, this retrospective observational study shows that mortality of patients hospitalized in the United States with cardiac arrest has decreased substantially and consistently from 2001 to 2009. This decrease in mortality has coincided with advances in postresuscitation care, including the introduction of TH, but the cause of this decreased mortality cannot be established with certainty.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.111.088807/-/DC1.
- Received December 22, 2011.
- Accepted June 8, 2012.
- © 2012 American Heart Association, Inc.
- Dumas F,
- Cariou A,
- Manzo-Silberman S,
- Grimaldi D,
- Vivien B,
- Rosencher J,
- Empana JP,
- Carli P,
- Mira JP,
- Jouven X,
- Spaulding C
- Grasner JT,
- Meybohm P,
- Caliebe A,
- Bottiger BW,
- Wnent J,
- Messelken M,
- Jantzen T,
- Zeng T,
- Strickmann B,
- Bohn A,
- Fischer H,
- Scholz J,
- Fischer M
- Weisfeldt ML,
- Sitlani CM,
- Ornato JP,
- Rea T,
- Aufderheide TP,
- Davis D,
- Dreyer J,
- Hess EP,
- Jui J,
- Maloney J,
- Sopko G,
- Powell J,
- Nichol G,
- Morrison LJ
Improving the postresuscitation care of patients with cardiac arrest has received much attention over the past decade. Landmark randomized trials were published in 2001 showing a benefit in mortality and morbidity with the use of therapeutic hypothermia in selected patients. Modifications to national cardiopulmonary resuscitation guidelines emphasizing high quality chest compressions, early use of automated external defibrillators, and early coronary interventions have also been advocated. We used the National Inpatient Sample database to examine the annual population-based mortality rates of 1 190 760 patients hospitalized in the United States from 2001 to 2009. The in-hospital mortality rate decreased consistently each year from 69.6% in 2001% to 57.8% in 2009. In multivariable analysis, when controlling for age, sex, race, and comorbidities, earlier year was a strong independent predictor of in-hospital death. The odds of in-hospital mortality associated with cardiac arrest in 2001 were significantly higher than the odds of mortality in 2005 (odds ratio, 1.11; 95% confidence interval, 1.09–1.14; P<0.0001). The odds of mortality associated with cardiac arrest in 2009 were significantly lower than the odds of mortality in 2005 (odds ratio, 0.69; 95% confidence interval, 0.67–0.70; P<0.0001). This decline in mortality rate was similar across all analyzed subgroups, including sex, age, race, and stratification by comorbidity. Although we cannot definitively conclude which specific factor is responsible for the decline in mortality, our results suggest that advances in postresuscitation care have positively impacted survival rates of patients hospitalized with cardiac arrest in the United States from 2001 to 2009.