Author + information
- Received December 2, 2011
- Revision received March 15, 2012
- Accepted March 20, 2012
- Published online July 3, 2012.
- Florence Dumas, MD, MPH⁎,†,‡,⁎ (, )
- Lindsay White, MPH⁎,
- Benjamin A. Stubbs, MPH⁎,
- Alain Cariou, MD†,§ and
- Thomas D. Rea, MD, MPH⁎∥
- ↵⁎Reprint requests and correspondence:
Dr. Florence Dumas, Emergency Medical Services Division of Public Health, 401 5th Avenue, Suite 1200, Seattle, Washington 98104
Objectives The aim of the study was to assess the influence of percutaneous coronary intervention (PCI) and therapeutic hypothermia (TH) on long-term prognosis.
Background Although hospital care consisting of TH and/or PCI in particular patients resuscitated following out-of-hospital cardiac arrest (OHCA) can improve survival to hospital discharge, there is little evidence regarding how these therapies may impact long-term prognosis.
Methods We performed a cohort investigation of all persons >18 years of age who suffered nontraumatic OHCA and were resuscitated and discharged alive from the hospital between January 1, 2001, and December 31, 2009, in a metropolitan emergency medical service (EMS) system. We reviewed EMS and hospital records, state death certificates, and the national death index to determine clinical characteristics and vital status. Survival analyses were conducted using Kaplan-Meier estimates and multivariable Cox regression. Analyses of TH were restricted to those patients who were comatose at hospital admission.
Results Of the 5,958 persons who received EMS-attempted resuscitation, 1,001 (16.8%) were discharged alive from the hospital. PCI was performed in 384 of 1,001 (38.4%), whereas TH was performed in 241 of 941 (25.6%) persons comatose at hospital admission. Five-year survival was 78.7% among those treated with PCI compared with 54.4% among those not receiving PCI and 77.5% among those treated with TH compared with 60.4% among those not receiving TH (both p < 0.001). After adjustment for confounders, PCI was associated with a lower risk of death (hazard ratio [HR]: 0.46 [95% confidence interval [CI]: 0.34 to 0.61]; p < 0.001). Likewise, TH was associated with a lower risk of death (HR: 0.70 [95% CI: 0.50 to 0.97]; p = 0.04).
Conclusions The findings suggested that effects of acute hospital interventions for post-resuscitation treatment extend beyond hospital survival and can positively influence prognosis following the arrest hospitalization.
Out-of-hospital cardiac arrest (OHCA) is a leading cause of death worldwide (1,2). Opportunities to improve OHCA survival, and in turn, public health depend in part on the chain of survival. The links in the chain include early activation of emergency response, early cardiopulmonary resuscitation (CPR), early defibrillation, expert advanced therapies, and integrated post-resuscitation care (3). Increasing evidence supports a role for hospital-based interventions that can reverse the underlying cause of arrest and limit ischemia-reperfusion injury that often occurs in OHCA (4). Evidence-based guidelines indicate that hospital care consisting of therapeutic hypothermia (TH) (5–8) and/or percutaneous coronary intervention (PCI) (9,10) in particular patients can improve functional hospital survival (5–8,10,11).
However, there is little evidence regarding how these therapies may impact long-term prognosis. Ultimately, an optimal public health assessment of therapy reflects both short- and long-term treatment effects. We hypothesized that the potential short-term survival benefit of hospital-based interventions of TH and PCI following OHCA would be amplified following hospital discharge and so would be associated with a reduction in mortality following hospital discharge.
Study design, population, and setting
We performed a cohort investigation of all patients aged ≥18 years who experienced nontraumatic OHCA and were resuscitated and discharged alive from the hospital between January 1, 2001, and December 31, 2009, in a large metropolitan emergency medical service (EMS) system. The study was approved by the appropriate review boards. The EMS system serves a population of approximately 1.3 million persons residing in urban, suburban, and rural settings covering an area of about 2,000 square miles. The EMS is a 2-tiered system composed of basic life support provided by emergency medical technician–trained firefighters equipped with automated external defibrillators and advanced life support delivered by paramedics who are trained in rhythm recognition and provide intubation, manual defibrillation, and intravenous medications (12). Patients are brought to 1 of 13 hospitals. Each hospital has a coronary catheterization laboratory and intensive care services. Hospital-based patient care is at the direction and discretion of the treating physician team.
Data collection and definitions
The EMS system maintains an ongoing registry of all EMS-treated cardiac arrests. Information about demographics, circumstances, presenting arrest rhythm, arrest etiology, clinical characteristics, care, and outcome is collected using emergency dispatch, EMS, and hospital records. The information is organized according to the Utstein Guidelines for reporting OHCA (13). We collected data about hospital interventions and their indications. Specifically, we collected information about TH, PCI, implantable cardioverter defibrillator (ICD), neurological status upon hospital arrival, and ST-segment elevation myocardial infarction (MI) on initial hospital electrocardiogram. Comatose status was based on review of the EMS, emergency department, and admitting records. ST-segment elevation infarction was defined as elevation of ≥1 mm in 2 contiguous limb leads and ≥2 mm in 2 contiguous precordial leads. Characteristics were abstracted without knowledge of outcome status.
The primary outcome was vital status following discharge from the hospital. We ascertained vital status by linking participant information with both state vital records (death certificates) through December 31, 2009, and the National Social Security Death Index through December 31, 2010.
We summarized categorical variables with proportions and compared them using the Pearson chi-square test or Fisher exact text if appropriate. Continuous variables were described with either their median (25th to 75th percentile) or their mean ± SD and compared using the nonparametric Wilcoxon test.
Entry into the cohort began at hospital discharge. Follow-up was censored at the date of death or up to December 31, 2010. Analyses specifically evaluating PCI and survival included all patients. Analyses investigating the relationship between TH and survival were restricted to those who were comatose at hospital admission and would therefore be considered eligible for TH. Survival was estimated by the Kaplan-Meier product-limit method. Comparisons between groups were performed using the log-rank or stratified log-rank test when appropriate. We evaluated survival according to TH status and PCI status separately and then evaluated survival simultaneously according to the 4 exclusive TH and PCI groups.
To evaluate the independent association of TH and PCI with long-term survival, we used Cox regression to perform a multivariable analysis that included potential confounders. The model initially included demographic characteristics. Subsequent models progressively included pre-hospital and clinical characteristics to test the robustness of the observed hazard ratios (HRs). Models also adjusted for calendar year or truncated after 5 years of follow-up to assess whether general temporal patterns could account for PCI or TH associations.
We assessed subgroups defined by initial rhythm (shockable/nonshockable) and ST-segment elevation MI (yes/no) by including an interaction term between the intervention and subgroup in the multivariable model. We also evaluated the PCI association among those who received the PCI urgently, defined as PCI within 6 h of hospital arrival.
In a separate set of analyses designed to address potential confounding in the relationships between TH and survival and between PCI and survival, we conducted a nested case–control cohort study with counter-matching design. This analytic design matches those with and without the exposure of interest across confounders adapted on survival analysis and then compares survival (14,15). Finally, we addressed potential confounding by using multivariable logistic regression to derive a propensity score to predict treatment with PCI or TH. We adjusted the association between the intervention and survival with the propensity to receive the intervention. We also stratified the patients according to the propensity score, matched patients according to their intervention status (intervention+/intervention−), and then compared survival. All tests were 2-sided with p ≤ 0.05 considered statistically significant. We performed analyses using STATA/SE 11.0 software (College Station, Texas).
During the study period, 16.8% (1,001 of 5,958) of patients ≥18 years of age were discharged alive from the hospital following EMS-treated, nontraumatic OHCA (Fig. 1). The length of hospitalization was 8 (5 to 14) days, and 85% of the cohort had a favorable outcome of cerebral performance category 1 or 2 at the time of hospital discharge. Among this cohort of OHCA hospital survivors, 384 of 1,001 (38.4%) received a PCI during the arrest hospitalization. Six percent (60 of 1,001) were conscious at hospital admission and so were not eligible for TH. Among the 941 eligible for TH, 245 of 941 (26.0%) received TH during the arrest hospitalization. A smaller subset—9.1% (86 of 941)—received both PCI and TH. PCI occurred within 6 h of hospital arrival in 306 of 384 patients (80%). Among those who received PCI within 6 h, 217 of 306 (70.9%) had evidence of ST elevation on the initial hospital 12-lead electrocardiogram. For TH-treated patients, the mean time to reach target temperature <34°C was 5.7 ± 5.5 h, the mean length of treatment was 21.0 ± 9.3 h, and the lowest mean temperature was 32.8°C ± 1.2°C.
Overall, two-thirds of the study cohort were male, and the median age was 61 years (Table 1). Nearly 80% (791 of 1,001) had a cardiac etiology, and nearly 70% (688 of 1,001) presented with an initial shockable rhythm. Compared with those who did not receive PCI, those who received PCI were more likely to be male, be younger, have an arrest due to a cardiac etiology, have a public location arrest, receive bystander CPR, and present with an initial shockable rhythm (Table 1). In addition, those who received PCI were more likely to have a witnessed arrest as well as clinically recognized cardiovascular risk factors and coronary disease. In the comparison according to TH status, those who received TH demonstrated similar relationships with regard to sex, age, arrest etiology, arrest location, bystander CPR status, and presenting rhythm. There were no observed differences in PCI status according to TH status or TH status according to PCI status.
During 3,800.7 person-years of follow-up, there were 348 deaths. Overall, survival was 86.9% (869 of 1,001) at 6 months, 82.2% (822 of 1,001) at 1 year, and 64.1% (641 of 1,001) at 5 years. Survival was better for those who received PCI compared with those who did not receive PCI (log-rank p < 0.001) (Fig. 2A). Similarly, survival was better for those receiving TH compared with those who did not receive TH (log-rank p < 0.001) (Fig. 2B). When stratified into 4 exclusive groups according to TH and PCI status, those who received neither treatment had the lowest 1-year and 5-year survival estimates, whereas those who received both TH and PCI had the highest 1-year and 5-year survival estimates (stratified log-rank p < 0.001) (Fig. 2C).
In the multivariable model, PCI, TH, and ICD were each independently associated with a better prognosis (Table 2). In the full model, PCI was associated with a lower risk of death (HR: 0.46 [95% confidence interval [CI]: 0.34 to 0.61]) as was TH (HR: 0.70 [95% CI: 0.50 to 0.98]). Associations of TH and PCI did not appear to be a consequence of overall temporal changes in long-term survival. Associations for each intervention were similar when adjusted for calendar year (model 5) or when the cohort was truncated at 5 years of follow-up (model 6). Interaction terms between clinical subsets (ST-segment elevation MI and initial rhythm) and primary exposures (PCI and TH) were not significant and so did not indicate evidence of differences in association according to subgroups. However, the confidence intervals crossed 1 for some of the subgroup associations. The HR of death associated with PCI was 0.41 (95% CI: 0.21 to 0.77) among those with ST-segment elevation MI and 0.55 (95% CI: 0.38 to 0.78) among those without ST-segment elevation MI. The HR of death associated with TH was 0.62 (95% CI: 0.41 to 0.93) among those with a shockable arrest rhythm and 0.89 (95% CI: 0.51 to 1.55) among those with a nonshockable rhythm. When the PCI exposure was restricted to those who underwent PCI within 6 h of hospital admission, the HR of death associated with “urgent” PCI was 0.36 (95% CI: 0.25 to 0.50).
The nested cohort was built for each intervention to enable a comparison matched on potential confounders and time at risk. There were 45 pairs for TH/no TH and 46 pairs for PCI/ no PCI. The analytic approach demonstrated favorable outcome associations for TH and PCI (Table 3).
Similarly, the beneficial association was evident when adjusting for the propensity score for PCI (HR: 0.60 [95% CI: 0.44 to 0.81]) and for TH (HR: 0.70 [95% CI: 0.50 to 0.98]) and when the comparison was stratified by propensity score and matched on intervention status.
An optimal assessment of medical therapy attempts to measure both short- and long-term treatment effects (16). Although evidence supports short-term survival benefits of TH and PCI in specific post-arrest populations, there is little information about how these hospital-based interventions may affect long-term prognosis. In this population-based cohort of patients resuscitated from OHCA who survived to hospital discharge, 5-year survival was 79% among those who received PCI and 54% among those who did not receive PCI. Similar survival relationships were observed for TH; 5-year survival was 78% among those who received TH and 60% among those who did not receive TH. These long-term protective associations were robust and evident in multivariable models and nested case–cohort analyses designed to account for confounding in this observational cohort study.
Long-term survival was substantial, with an overall 5-year survival rate of 64% among a group whose mean age was 61 years at cohort entry. Importantly, the cohort included all patients who were resuscitated from OHCA and discharged from the hospital regardless of etiology or functional status at discharge in an effort to provide a representative community experience that would maximize generalizability. Taken together, the overall long-term survival is encouraging and indicates that initial efforts that help achieve survival to hospital discharge can result in considerable long-term survival and measurable public health reduction in mortality.
We specifically evaluated the prognostic role of interventional hospital therapies. The placement of an ICD was associated with improved long-term survival, a finding consistent with the results of randomized controlled trials (17). The comparability between the association observed in the current study and the survival effects determined in previous ICD trials of secondary prevention suggested that the treatment associations observed in the current study are valid.
We focused on the long-term prognostic influence of PCI and TH and observed an independent survival benefit associated with each intervention. PCI in patients with transmural infarction can potentially provide long-term survival benefits through a variety of mechanisms, such as a reduction of recurrent ischemia and preservation of myocardial function (18–20). However the long-term survival influence of PCI in a large comprehensive cohort of OHCA survivors is less established. Prior studies have indicated survival benefit at 6 months or among select smaller groups of OHCA survivors (21). We observed a beneficial PCI survival association that was evident early after discharge and continued to accrue over 5 years of follow-up.
Although approximately three-quarters of PCIs occurred among those with evidence of ST-elevation MI, a quarter of PCIs was performed among those without ST-elevation MI. A beneficial survival association was evident among those with and without ST-elevation MI. This finding is provocative given the current debate about whether patients without evidence of ST elevation following resuscitation can benefit from PCI and should undergo early and routine coronary catheterization (10,22,23). The current study suggested that PCI—performed at the discretion of the cardiologist—can provide long-term survival benefits. Importantly, the current study cannot definitively characterize those patients without ST elevation who are most likely to benefit from PCI.
We also observed a long-term survival benefit associated with TH. The mechanism by which hypothermia provides early benefit is not fully defined, although cooling appears to reduce systemic ischemia-reperfusion injury and help preserve critical organ function including the brain (4). Prior studies have suggested that recovery from neurological injury after resuscitation from OHCA can take many months (24,25). Given this protracted time course of neurological recovery, one hypothesis is that TH in the acute phase may benefit more convalescent phases of physiological recovery (26).
We also stratified survival into 4 treatment groups based on the combination of PCI and TH and found that patients treated with both interventions had the highest long-term survival, those treated with one or the other had an intermediate long-term prognosis, and those treated with neither had the lowest survival. Prior studies have indicated that the 2 therapies can be safely combined and may provide additive short-term survival benefit (27–30). The current results suggested that this additive benefit may extend long term. The finding that distinct interventions improved long-term outcome supports a multidisciplinary hospital approach that could coordinate multiple treatments (31–33).
The study population was an inception cohort of those who survived to hospital discharge following OHCA resuscitation. As a consequence, we were not able to evaluate the potential in-hospital effects of PCI and TH. However, multiple studies have evaluated the short-term outcomes associated with these therapies, although few have evaluated the long-term prognostic implications. The study was observational; therefore, we cannot be sure that the associations were causal. For example, we were not able to adjust for medication treatments, which can influence prognosis. To help address potential confounding, models included a range of demographic and clinical covariates including the common indications for medication treatments (i.e., coronary artery disease, congestive heart failure). We found that the beneficial associations of PCI and TH were robust to a variety of analyses designed to account for confounding, including the matched nested cohort analysis. We did not have detailed information about the type of hypothermia or the extent of PCI—information that might help delineate how these interventions provided benefit. We did not detect a difference in the intervention-outcome relationship according to clinical subgroups. However, these analyses had limited power to detect subgroup differences. Although the investigation included all patients who survived to hospital discharge and involved a large metropolitan region served by a mix of academic and community-based hospitals, we cannot be certain that the results will generalize to other communities or health systems. For example, all of the receiving hospitals in the study community have a catheterization laboratory and intensive care services. We do not have information about functional status or quality of life during the follow-up period, although the majority was discharged with a favorable cerebral performance category score of 1 or 2.
In this population-based cohort of patients discharged alive from the hospital following OHCA, we observed a long-term survival benefit associated with PCI and TH therapies provided during the OHCA hospitalization. The findings suggested that the effects of acute hospital interventions for post-resuscitation treatment of OHCA extend beyond initial hospital survival and influence prognosis following the arrest hospitalization. Thus, the optimal measure of these therapies may incorporate short- and long-term outcomes. The challenge then is to identify those who will and those who will not benefit from such interventions to maximize health benefit, minimize risk, and use resources efficiently.
All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiopulmonary resuscitation
- emergency medical service
- implantable cardioverter defibrillator
- myocardial infarction
- out-of-hospital cardiac arrest
- percutaneous coronary intervention
- therapeutic hypothermia
- Received December 2, 2011.
- Revision received March 15, 2012.
- Accepted March 20, 2012.
- American College of Cardiology Foundation
- Travers A.H.,
- Rea T.D.,
- Bobrow B.J.,
- et al.
- Nolan J.P.,
- Neumar R.W.,
- Adrie C.,
- et al.
- Dumas F.,
- Grimaldi D.,
- Zuber B.,
- et al.
- Dumas F.,
- Cariou A.,
- Manzo-Silberman S.,
- et al.
- Peberdy M.A.,
- Callaway C.W.,
- Neumar R.W.,
- et al.
- Rea T.D.,
- Helbock M.,
- Perry S.,
- et al.
- Jacobs I.,
- Nadkarni V.,
- Bahr J.,
- et al.
- Becker L.B.,
- Aufderheide T.P.,
- Geocadin R.G.,
- et al.
- Connolly S.J.,
- Hallstrom A.P.,
- Cappato R.,
- et al.
- De Luca G.,
- Suryapranata H.,
- Zijlstra F.,
- et al.
- Antman E.M.,
- Anbe D.T.,
- Armstrong P.W.,
- et al.
- Garot P.,
- Lefevre T.,
- Eltchaninoff H.,
- et al.
- Wolfrum S.,
- Pierau C.,
- Radke P.W.,
- Schunkert H.,
- Kurowski V.
- Stub D.,
- Bernard S.,
- Duffy S.J.,
- Kaye D.M.