Association of Hospital-Level Acute Resuscitation and Postresuscitation Survival With Overall Risk-Standardized Survival to Discharge for In-Hospital Cardiac Arrest

Key Points Question Are rates of acute resuscitation and postresuscitation survival associated with rates of overall risk-standardized survival to discharge for in-hospital cardiac arrest? Findings In this cohort study of 86 426 patients with in-hospital cardiac arrest from 290 hospitals, a hospital’s overall risk-standardized survival rate was more strongly correlated with its risk-adjusted postresuscitation survival than with acute resuscitation survival. There was no correlation between risk-adjusted acute resuscitation and postresuscitation survival. Meaning The findings suggest that, because current quality improvement initiatives focus largely on acute resuscitation care, efforts to strengthen postresuscitation care may offer additional opportunities to improve survival after in-hospital cardiac arrest.


Introduction
There is substantial variation between hospitals for survival of in-hospital cardiac arrest (IHCA). 1,2 To date, most quality improvement initiatives have focused on delivering timely chest compressions, early defibrillation, and epinephrine during an acute resuscitation response. [3][4][5][6][7] However, what has been underappreciated is that IHCA survival depends on 2 distinct phases of care. 8,9 Survival may depend on care during the initial resuscitation, which is largely associated with the responsiveness and quality of the hospital resuscitation or code team (ie, acute resuscitation phase). Survival may also depend on care after return of spontaneous circulation, driven largely by the quality and expertise of intensive and specialty care at a hospital (ie, postresuscitation phase).
Previous studies 1,8 of IHCA have not defined the association of acute resuscitation and postresuscitation phases with overall survival. Although studies have shown that overall survival for IHCA varies by more than 3-fold across hospitals, 1,2 it remains unknown whether high survival at top-performing hospitals is associated with high rates of acute resuscitation survival, postresuscitation survival, or both. This is important to understand because current initiatives for improving resuscitation care quality and reducing variation in IHCA survival largely focus on incentivizing acute resuscitation care delivery, such as reducing time to defibrillation and delivering effective chest compressions. However, such initiatives will have the strongest association with survival if hospitals that excel in acute resuscitation care also excel in postresuscitation care.
To address this gap in knowledge, we used contemporary data from the American Heart Association Get With The Guidelines (GWTG)-Resuscitation registry 10 to examine site-level variation in IHCA survival to identify hospitals that had high overall survival rates among patients with IHCA after adjustment for patient case mix. We further examined the extent of correlation between a hospital's overall IHCA survival with its risk-adjusted rate of acute resuscitation survival and postresuscitation survival. We believe that a better understanding of the association of overall IHCA survival with acute resuscitation and postresuscitation survival will have important implications for designing future initiatives for improving resuscitation care quality.

Study Design and Data Sources
We designed a cohort study within the GWTG-Resuscitation registry, a prospective multisite registry of IHCA events in the US. The design of this registry has been described previously. 10 The study was reviewed by the University of Iowa institutional review board, Iowa City, which waived the requirement for informed consent because of the use of deidentified data and approved the study.
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The design of the GWTG-Resuscitation registry has been described previously. 10 In brief, all hospitalized patients with confirmed IHCA, defined as absence of a palpable central arterial pulse, apnea, and unresponsiveness, and without do-not-resuscitate orders, were enrolled by trained personnel at participating hospitals. Multiple case-finding approaches were used, including review of centralized collection of cardiac arrest flow sheets, routine review of code cards, pharmacy tracer drug records, review of hospital paging system logs, and hospital billing charges for resuscitation medications. Hospital participation was voluntary with data collected using standardized Utsteinstyle definitions for all patient variables and outcomes to facilitate uniform reporting across hospitals. 11,12 Accuracy and completeness of the data were ensured by rigorous training and certification of medical staff at participating hospitals along with use of standardized software for internal checks and periodic reabstractions and audits of collected data. 10

Study Population
Using contemporary data from GWTG-Resuscitation, we identified 90 343 patients aged 18 years or older who experienced an index IHCA event from January 1, 2015, through December 31, 2018. From this sample, we excluded patients who were missing data on comorbidities (n = 546), arrest location (n = 53), and survival (n = 1042). To ensure that the estimates of hospital survival for IHCA obtained from multivariable models were statistically reliable, we excluded 2276 patients from hospitals with fewer than 50 cases during the study period. Our final cohort comprised 86 426 patients at 290 hospitals (eFigure 1 in the Supplement).

Study Variables and Outcomes
The main outcomes of our study were a hospital's overall risk-standardized survival rate (RSSR) to discharge and its 2 components: acute resuscitation survival and postresuscitation survival. Acute resuscitation survival was defined as return of spontaneous circulation for at least 20 minutes among patients with an initial cardiac arrest. Postresuscitation survival was defined as survival to discharge among patients who achieved return of spontaneous circulation.
Patient level data included (1) demographics (age, sex, and race/ethnicity); (2) comorbidities and preexisting medical conditions (current or previous heart failure; current or previous myocardial infarction; diabetes; renal, hepatic, or respiratory insufficiency; baseline evidence of motor, cognitive, or functional deficits [central nervous system depression]; acute stroke; pneumonia; hypotension; sepsis; major trauma; metabolic or electrolyte abnormalities; or metastatic or hematologic malignant neoplasm); (3) cardiac arrest characteristics (initial rhythm [asystole, pulseless electrical activity, ventricular fibrillation, and pulseless ventricular tachycardia]); (4) the use of a hospital-wide cardiopulmonary arrest alert; (5) time of cardiac arrest (weekday: 8:00 AM to 5:00 PM from Monday to Friday, weeknight: 5:00 PM to 8:00 AM from Monday to Friday, and weekend: all day on Saturday and Sunday); (6) cardiac arrest location (intensive care unit [ICU], monitored unit, nonmonitored unit, emergency department, procedural or surgical area, and other); and (7) interventions in place at the time of cardiac arrest (mechanical ventilation, use of intravenous vasoactive vasopressors, intraarterial lines, and dialysis). Hospital-level variables included number of beds, number of ICU and cardiac ICU beds, academic status, urban or rural location, and geographic census region.

Statistical Analyses
Our primary objective was to assess the extent of correlation between a hospital's overall rate of survival to discharge for IHCA with its acute resuscitation and postresuscitation survival. For the outcome of hospital rate of survival to discharge, we calculated the overall RSSR for each hospital in the cohort by using a previously validated method. 1 Specifically, we built a 2-level multivariable hierarchical regression model to relate the log odds of survival with patient variables. Hierarchical models account for clustering of patients within a hospital and separate within-hospital variation from between-hospital variation as well as model the assumption that underlying differences in hospital quality explain the between-hospital differences in survival. 13 Patient variables included in this model were based on the previous validation study and included age, initial cardiac arrest rhythm, location of arrest, hypotension, sepsis, metastatic or hematologic malignanc neoplasm, hepatic insufficiency, mechanical ventilation, and use of intravenous vasopressors before the cardiac arrest. The hospital site was included as a random effect in these models.
Using regression coefficients from this model, we estimated each hospital's risk-standardized survival as the ratio of predicted to expected survival multiplied by the overall unadjusted survival rate for patients with IHCA. Compared with the observed to expected ratio, the predicted to expected ratio does not unfairly penalize small-volume hospitals by accounting for the lower precision in survival estimates from such volume hospitals. 14 This model had excellent discrimination (C statistic, 0.74) and calibration in the previous validation study. 1 We also quantified variation in survival rates across hospitals using median odds ratios from the the hierarchical models described using the variance estimate of the random hospital intercept. 15 Because validated models to risk standardize acute resuscitation and postresuscitation survival have not been developed, we calculated each hospital's risk-adjusted rate of acute resuscitation survival and postresuscitation survival. Risk adjustment was performed using the same variables that were included in the model for overall survival.
Next, we categorized study hospitals into quartiles based on their overall RSSR (Q1, lowest quartile; Q4, highest quartile) and used descriptive statistics to compare hospital-level and patient characteristics using a χ 2 test for categorical variables and analysis of variance for linear variables. We then compared rates of acute resuscitation survival and postresuscitation survival across hospital quartiles of RSSR and calculated the absolute difference between Q4 and Q1 RSSR quartiles. We also examined the extent to which hospital performance on the RSSR metric was concordant with performance on the acute resuscitation and postresuscitation survival. Finally, we calculated the Pearson correlation between hospital's RSSR and its risk-adjusted rates of acute resuscitation survival and postresuscitation survival. The significance level was set at P < .05 using a 2-sided test. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc).

Results
A total of 290 hospitals and 86 426 patients with IHCA were included. Table 1

JAMA Network Open | Cardiology
Association of Resuscitation-Related Care With Overall Survival of In-Hospital Cardiac Arrest

JAMA Network Open | Cardiology
Association of Resuscitation-Related Care With Overall Survival of In-Hospital Cardiac Arrest

JAMA Network Open | Cardiology
Association of Resuscitation-Related Care With Overall Survival of In-Hospital Cardiac Arrest Q4 hospitals had more cardiac beds and were more likely to be located in the North Central region compared with Q1 hospitals.  Table 4). The

Discussion
In this contemporary study of 290 GWTG-Resuscitation hospitals, we found an approximately 3-fold variation in overall rates of IHCA survival (14.1%-40.8%). Although we found that a hospital's rate of  The development of quality metrics for postresuscitation care has been substantially hampered by the lack of evidence from randomized clinical trials for existing postarrest treatments. For example, clinical trials have largely shown a benefit of targeted temperature management (TTM) only in patients with out-of-hospital cardiac arrest. [17][18][19] Observational studies of TTM in adults with IHCA have also yielded mixed results, with the largest one showing no survival benefit. [20][21][22] A dedicated randomized clinical trial of TTM in patients with IHCA was conducted in children and did not show a benefit. 23 However, a recent randomized clinical trial of patients with cardiac arrest due to a nonshockable rhythm that included 27% of patients with IHCA found higher rates of favorable neurologic survival in patients treated with moderate therapeutic hypothermia compared with targeted normothermia. 24 Likewise, a strategy of routine coronary angiography, to date, has not been shown to be associated with improved survival in patients successfully resuscitated from out-of-hospital cardiac arrest, 25 but remains to be studied in patients with IHCA.
Although TTM and routine coronary angiography remain therapeutic options that require further study, hospitals that excel in postresuscitation care are more likely to structure and deliver high-quality care to successfully resuscitated patients in the ICU. The American Heart Association recommends a multipronged strategy focused on optimization of hemodynamics, gas exchange, neurologic and metabolic parameters with care guided by specialists in intensive care, neurocritical care, and cardiology. 26 Although best practices for maximizing postresuscitation survival have not been clearly delineated, a few medical centers have developed highly specialized postcardiac arrest care teams that provide consultation 24 hours per day for 7 days per week to all patients with cardiac arrest throughout the hospital. 27 Such a team-based structure ensures that management of these patients needing complex care is concentrated among a small group of physicians with appropriate expertise and that care is standardized according to protocols. It is important to determine how the use of innovative postarrest strategies such as the use of specialized cardiac arrest teams is associated with postresuscitation and overall IHCA survival. Because existing registries such as GWTG-Resuscitation do not capture these data, identifying best practices for postresuscitation survival will require a combination of quantitative and qualitative approaches (ie, mixed methods) to identify best practices for improving postresuscitation and overall IHCA survival.

Limitations
This study has limitations. First, hospitals participating in GWTG-Resuscitation are predominantly large, urban hospitals with an interest in resuscitation quality improvement, which may limit the generalizability of our findings. Second, although GWTG-Resuscitation collects rich data on patientlevel variables for case-mix adjustment, the potential for residual confounding because of unmeasured clinical or socioeconomic variables remains. Third, we lacked information on postresuscitation treatment strategies at individual hospitals, which limited our ability to identify the specific hospital practices that may be associated with hospital performance on postresuscitation.
Fourth, our study was primarily limited to in-hospital survival outcomes and data on quality of life; data on physical and mental functioning after hospital discharge were not available.

Conclusions
The findings suggest that hospitals with high overall survival rates for IHCA, in general, excel in either acute resuscitation or postresuscitation care but not both. Since most hospital-based quality improvement initiatives largely focus on acute resuscitation survival, our findings suggest that efforts to strengthen postresuscitation intensive care may offer additional opportunities to improve IHCA survival.

JAMA Network Open | Cardiology
Association of Resuscitation-Related Care With Overall Survival of In-Hospital Cardiac Arrest