Delayed Time to Defibrillation after Intraoperative and Periprocedural Cardiac Arrest

• ❖ For patients with ventricular cardiac arrest, delayed defibrillation is associated with worse survival in numerous hospital and nonhospital settings.

• ❖ Data from the National Registry of Cardiopulmonary Resuscitation suggest that delayed defibrillation is associated with lower hospital survival for patients with periprocedural, but not intraoperative ventricular cardiac arrest.

DEFIBRILLATION within 2 min has been shown to predict survival after ventricular fibrillation or pulseless ventricular tachycardia among patients whose cardiac arrest occurred while they were in intensive care units or inpatient beds.1Current recommendations are that hospitalized patients receive defibrillation therapy within 2 min after recognition of cardiac arrest.2,3 

Intraoperative and periprocedural cardiac arrests occur in a unique clinical setting where resuscitation response times and underlying etiology may differ from cardiac arrests occurring elsewhere in the hospital.4A monograph from the American Society of Anesthesiology Committee on Critical Care Medicine notes that perioperative cardiac arrest is usually witnessed, frequently anticipated, and relatively rare when compared with arrest in other locations. Given a paucity of data specific to the perioperative setting, the monograph calls for “systematic study of how to manage these rare events.”§Little is known regarding the relationship between defibrillation response times and outcomes of intraoperative and periprocedural cardiac arrest. Patients who develop pulseless ventricular tachycardia or ventricular fibrillation arrest in the cardiac intervention suites may be more responsive to rapid defibrillation because of the nature of the ongoing procedure (e.g ., catheter ablation, coronary angiography). Alternatively, patients who arrest in the operating room as a result of intraoperative hemorrhage may have a poor prognosis regardless of resuscitation measures such as defibrillation time.5A recent study that examined the relationship between time to defibrillation and survival for pulseless ventricular tachycardia or ventricular fibrillation arrests excluded patients experiencing cardiac arrest in the operating room, perianesthesia care unit (PACU), and interventional procedure areas.1 

Accordingly, the primary purpose of this analysis was to examine the association between time to defibrillation and survival for intraoperative and periprocedural patients with cardiac arrest whose first pulseless rhythm is identified as pulseless ventricular tachycardia or ventricular fibrillation.

In 2000, the American Heart Association established the National Registry of Cardiopulmonary Resuscitation (NRCPR) to assist participating hospitals with systematic data collection on resuscitative efforts. The NRCPR is a prospective, observational, multicenter registry of in-hospital cardiac arrests. Hospitals join voluntarily and pay a fee for data support and quarterly quality improvement reports. The registry is currently the largest repository of information on in-hospital cardiopulmonary arrest,6with close to 175,000 cardiopulmonary arrest events recorded as of September 30, 2009.∥The members of the American Heart Association National Registry of Cardiopulmonary Resuscitation Investigators are listed in  appendix 1.

In-hospital resuscitation teams typically assign a clinical provider to serve in a data-recording role during the resuscitation event. These individuals log the event for clinical documentation purposes, using local documentation processes, and the event logs are stored with the patient's hospital medical record.

Hospital medical records of sequential cardiac arrests are abstracted electronically at each institution by trained NRCPR-certified research coordinators who are typically not involved in the resuscitation event. Data elements use standardized Utstein nomenclature to facilitate consistent definitions of variables across multiple sites.7Data abstractors are required to complete a certification exam successfully. These exams comprise evaluation of data abstraction, entry accuracy, and operational definition compliance before a site is allowed to submit data to the NRCPR. In addition, the data entry software (provided by Digital Innovation, Inc., Forest Hill, MD, 2008) has more than 300 checks and smart skips to assist with accurate data entry. Participating hospitals are required to submit 10% of their resuscitation forms so that reabstraction can be performed by NRCPR personnel. The NRCPR has completed a periodic audit of the accuracy of the data elements, including time variables, and the most recent error rate for all variables was less than 3%.8Other efforts to ensure data accuracy include ongoing abstractor training with monthly users' group calls and an annual users' group conference.

Each patient is assigned a unique, deidentified database number before data entry and transmission to the central NRCPR database in compliance with the Health Insurance Portability and Accountability Act. On behalf of the NRCPR, Digital Innovation, Inc., provides data management and quarterly reports summarizing individual hospital data and comparative group reports. Oversight is provided by the American Heart Association, represented by a scientific advisory board and an executive database steering committee.

Six major categories of variables for each patient with a cardiac arrest are collected: (1) facility data; (2) patient demographic data; (3) preevent data; (4) event data; (5) outcome data; and (6) process quality improvement data.

The initial study population included adults 18 yrs and older who experienced a pulseless cardiac arrest caused by pulseless ventricular tachycardia or ventricular fibrillation in the operating room, PACU, diagnostic suite, or cardiac intervention suite. All events were reported from 1 of the 413 facilities in the United States and Canada that provided at least 6 months of data to the NRCPR between January 1, 2000, and February 11, 2008. To be included in the registry, the event had to elicit a resuscitation response by facility personnel and to have had a resuscitation record completed. To evaluate the impact of delayed defibrillation, only patients whose first identified rhythm was pulseless ventricular tachycardia or ventricular fibrillation were included in the analysis (n = 1,131).

For patients having multiple cardiac arrests during the same hospitalization, only the first in-hospital cardiac arrest event was analyzed, and recurrent arrests (n = 59) were excluded (fig. 1). Further exclusions comprised patients with an implanted cardioverter-defibrillator (n = 25), prior Do-not-resuscitate orders (n = 9), and missing or inconsistent data on the time of the cardiac arrest (n = 173). The final study cohort included 865 patients who experienced cardiac arrest in 259 facilities.

Time to defibrillation, the primary process of care variable, was calculated as the interval from the reported time of initial recognition of cardiac arrest to the reported time of first attempted defibrillation. Both times were determined from cardiac arrest documentation in the patient's medical record and were recorded in minutes. These data were used to determine the proportion of study subjects with delayed defibrillation, which was defined, based on prior work and current guidelines,1,6as a time to defibrillation greater than 2 min.

The primary outcome was survival to hospital discharge. Unadjusted analyses evaluated baseline differences between patients with and without missing time data and between patients with and without delayed defibrillation using the Student t  test for continuous variables and the Pearson chi-square test or the Fisher exact test for categorical variables, as appropriate. To distinguish between random and informative missing time data, a multivariable logistic regression model was built to predict time to defibrillation, controlling for age, sex, and race; bivariate analyses of potential predictor variables with and without missing time to defibrillation were entered into this regression model based on an α level of 0.05.

A multivariable logistic regression model was then constructed to investigate the relationship between delayed defibrillation and survival to discharge. Collinearity diagnostics were performed for all study covariates before model development. A generalized estimating equation with an exchangeable correlation matrix was used to control for clustering at the facility level. Independent variables included age, sex, race (white vs . nonwhite), presenting rhythm (ventricular fibrillation or pulseless ventricular tachycardia), location of arrest (the operating room, PACU, cardiac catheterization laboratory, or diagnostic suite), arrest in the operating room (vs . all other locations), admitting diagnosis (cardiac vs . noncardiac), and weekday arrests (7:00 am until 11:00 pm Monday through Friday vs . arrests during all other times).9 

Event survival, defined as a return of spontaneous circulation for at least 20 min from the initial resuscitation event, was a secondary outcome. Following the analytic steps described above, a multivariable logistic regression model was constructed to examine the relationship between delayed defibrillation and event survival.

Additional multivariable logistic regression models were used to evaluate statistical interactions between delayed defibrillation time and cardiac arrest location, between delayed defibrillation and a diagnosis of myocardial infarction during the current hospitalization, and between delayed defibrillation and preexisting hemorrhage on survival outcomes. In cases where a significant interaction (P < 0.05) was found, the final models explored the relationship between delayed defibrillation and survival outcomes within these subgroups. Hemorrhage is not recorded in the NRCPR. We therefore defined hemorrhage as existing if a patient received any blood product transfusion or aortic cross-clamping or had preexisting hypotension or hypoperfusion but no acute nonstroke event, acute stroke, sepsis, pneumonia, myocardial infarction, or congestive heart failure in the current admission. The models were constructed as above using generalized estimating equations, with separate models constructed for all patients arresting in the operating room, all patients arresting in the cardiac catheterization laboratory, those with and those without preexisting hemorrhage, and those with and those without a diagnosis of myocardial infarction.

Finally, a simple logistic regression model was used to estimate the effect size of each additional minute of delay in defibrillation on survival to discharge.

For all analyses, the null hypothesis was evaluated at a two-sided significance level of 0.05, and effect sizes were reported as odds ratios (OR) or adjusted odds ratios (AOR), both with 95% confidence intervals (CIs). All statistical analyses were performed using Stata, Version 10, 2006 (StataCorp LP, College Station, TX).

Exclusions are illustrated in figure 1. The 91 patients (9.4%) who were excluded because their records were missing time to defibrillation had baseline characteristics that were similar to those patients in the final study cohort, except that the excluded patients were younger, more likely to present in the operating room as opposed to other locations, with a cardiac surgical or traumatic illness category, with preexisting hemorrhage or major trauma, or arrest attributed to conscious sedation or hypotension ( appendix 2). Excluded patients were less likely to survive the event (55.9% vs . 70.8%, OR 0.48, 95% CI 0.31, 0.75, P = 0.001) or to survive until hospital discharge (33.0% vs . 51.2%, OR 0.47, 95% CI 0.30, 0.74, P = 0.001). Multivariable analysis confirmed that three variables were associated with missing time information: arrests in the operating room, arrests attributed to conscious sedation, and arrests attributed to arrhythmia ( appendix 3).

Of the 865 patients included in the study cohort, 373 (43.1%) arrested in an operating room, 145 (16.7%) in the PACU, 264 (30.5%) in a cardiac catheterization laboratory, and 83 (9.6%) in a diagnostic suite. The majority of patients had cardiac rhythm monitoring in place at the time of the arrest (358 [96.0%] in the operating room, 137 [94.5%] in the PACU, 263 [99.6%] in the cardiac catheterization laboratory, and 62 [74.7%] in the diagnostic suite, P < 0.001). The overall median time to defibrillation was less than 1 min (interquartile range, <1 to 1 min), with 119 patients (13.8%) noted to have delayed defibrillation (more than 2 min from cardiac arrest to defibrillation). A higher proportion of patients with delayed defibrillation were admitted with noncardiac diagnoses and presented in pulseless ventricular tachycardia in the operating room or diagnostic suite (table 1). In adjusted analyses, a presenting rhythm of pulseless ventricular tachycardia and a noncardiac admitting diagnosis were independently associated with delays in defibrillation (table 2).

In bivariate analyses, delayed defibrillation was associated with lower event survival (60% vs . 72.5%, OR 0.62, 95% CI 0.42, 0.92, P = 0.02) and lower survival to discharge (39.5% vs . 53.1%, OR 0.67, 95% CI 0.46, 0.98, P = 0.04).

In predicting event survival, significant interactions were found between delayed defibrillation and arrest in the operating room as opposed to other locations (adjusted P = 0.022), between delayed defibrillation and preexisting diagnosis of myocardial infarction during the current hospitalization (P = 0.009), and between delayed defibrillation and preexisting diagnosis of hemorrhage during the current admission (P = 0.028). In predicting survival to discharge, significant interactions were found between delayed defibrillation and location of arrest (arrest in the operating room as opposed to other locations) (adjusted P = 0.003), between delayed defibrillation and a preexisting diagnosis of myocardial infarction during the current hospitalization (P = 0.008), and between delayed defibrillation and a preexisting diagnosis of hemorrhage during the current admission (P = 0.003). Because of the significant interactions, we evaluated the relationships of delayed defibrillation with survival outcomes separately for periprocedural and intraoperative cardiac arrests, for patients with and without preexisting myocardial infarction, and for patients with and without preexisting hemorrhage.

Among patients arresting outside of the operating room (in the cardiac catheterization laboratory, the diagnostic suites, or the PACU), delayed defibrillation was independently associated with lower event survival (52.6% vs . 78.1%, AOR 0.44, 95% CI 0.25, 0.79, P = 0.006) and lower survival to discharge (31.6% vs . 62.1%, AOR 0.49, 95% CI 0.27, 0.88, P = 0.018) (table 3). Other significant predictors of survival to discharge for periprocedural arrests included patient age, male sex, and a cardiac admission diagnosis.

The effect on event survival was also evident in a subanalysis of patients arresting in the cardiac catheterization laboratory, where 20 patients (6.6%) received delayed defibrillation. Event survival for these patients was reduced (45.0% vs . 81.7%, AOR 0.27, 95% CI 0.1, 0.69, P = 0.006) when compared with survival among catheterization patients with prompt defibrillation. The effect on survival to discharge was also significant (35.0% vs . 68.7%, AOR 0.38, 95% CI 0.15, 0.98, P = 0.05). Because of collinearity, cardiac admission diagnosis was excluded from the list of independent variables for the model specific to the cardiac catheterization laboratory.

Among patients who arrested in the operating room, there was no difference between delayed defibrillation and event survival (66.1% vs . 64.1%, AOR 0.95, 95% CI 0.52, 1.75, P = 0.88) or survival to discharge (46.8% vs . 39.6%, OR 1.23, 95% CI 0.70, 2.19, P = 0.47). For these arrests, independent predictors of survival to discharge included patient age, a cardiac admission diagnosis, and time of arrest, but not delayed defibrillation (table 3).

In areas outside of the operating room, each additional minute of delay in defibrillation produced a nonsignificant trend toward reduced survival. This trend was not apparent among arrests in the operating room (fig. 2).

Among patients with a preexisting diagnosis of myocardial infarction during the current admission, delayed defibrillation was independently associated with lower event survival (36.8% vs . 79.1%, AOR 0.19, 95% CI 0.07, 0.52, P = 0.001) and lower survival to discharge (15.8% vs . 59.5%, AOR 0.16, 95% CI 0.05, 0.59, P = 0.005) (table 4).

For arrests among the 136 patients with hemorrhage, none of the independent variables predicted survival to discharge in multivariable analysis (table 5). In bivariate analysis, delayed defibrillation was not associated with a change in event survival (65.0% vs . 53.5%, P = 0.34) or survival to discharge (45.0% vs . 24.1%, P = 0.05).

A preexisting diagnosis of myocardial infarction during the current admission was most common in the cardiac catheterization laboratory (140 [53.0%], compared with 18 [21.7%] in a diagnostic suite, 31 [13.3%] in the PACU, and 45 [12.1%] in the operating room, P < 0.001). A preexisting diagnosis of hemorrhage was most common in the operating room (101 [27.1%], compared with 20 [13.8%] in the PACU, 6 [7.2%] in the diagnostic suite, and 9 [6.6%] in the cardiac catheterization laboratory, P < 0.001).

In this large, multisite registry of in-hospital cardiac arrests, we found that delays in defibrillation occurred even in the intraoperative and periprocedural arenas where multiple medical and nursing staff are typically present. Delays in therapeutic defibrillation impacted 13.8% of resuscitations and were more frequent in patients presenting in pulseless ventricular tachycardia and in patients admitted for a noncardiac diagnosis. For patients arresting in periprocedural areas, including the cardiac catheterization laboratory, the diagnostic suites, and the PACU, these delays were associated with lower rates of survival through the initial resuscitation and to hospital discharge.

A number of studies have demonstrated that increased time to defibrillation decreases survival for patients suffering from out-of-hospital cardiac arrest with ventricular fibrillation or pulseless ventricular tachycardia.10–17This relationship has more recently been confirmed among patients experiencing in-hospital ventricular arrest, with events drawn primarily from hospital wards and intensive care units1,18–21Skrifvars et al . included 3 patients suffering from pulseless ventricular tachycardia in the operating area among a sample of 57 patients with shockable rhythms throughout the hospital, and found that survivors had a shorter time interval to defibrillation compared with nonsurvivors (median 1 min vs . 5 min, P = 0.046). Twelve of 206 arrests reported by Hajbaghery et al . were identified in operating rooms or surgical wards, but further details, such as the presenting rhythm for these 12 arrests, were not provided.21 

Using data from the NRCPR, Chan et al . examined the relationship between time to defibrillation for ventricular fibrillation and pulseless ventricular tachycardia arrests and survival to discharge among patients in intensive care units and general care beds.1A graded inverse association was found between time to defibrillation and unadjusted survival across a broad range of time thresholds. The accompanying editorial concluded that survival after cardiac arrest caused by ventricular tachycardia or ventricular fibrillation requires prompt defibrillation, regardless of the setting in which it occurs. The editorial went further to assert that “it is clear that timely defibrillation in the hospital is an important determinant of the quality of cardiovascular care.”22 

The current analysis failed to confirm a relationship between delayed defibrillation and survival for patients arresting in the operating room. We speculate several potential reasons why delayed defibrillation may not be associated with survival for cardiac arrest occurring in the operating room. Baseline characteristics of patients may differ in the operating room compared with other locations. For example, patients who arrested in the operating room were less likely to have a preexisting diagnosis of myocardial infarction, a diagnosis shown to predict sensitivity to delayed defibrillation in multivariate analysis of our population (table 4). Conversely, hemorrhage was more commonly identified in the operating room, a diagnosis that did not predict sensitivity to delayed defibrillation in multivariable analysis (table 5). In fact, patients with hemorrhage demonstrated a trend toward improved survival with delay in defibrillation. Hemorrhage may be lethal regardless of rhythm. In the hemorrhaging patient, an initial rhythm of ventricular fibrillation or pulseless ventricular tachycardia may rapidly deteriorate to asystole or pulseless electrical activity, rhythms that would be unresponsive to defibrillation and would carry a worse prognosis. A focus on advanced cardiac life support and rapid defibrillation may conflict with optimal management of surgical hemorrhage if the effort inhibits surgical access to the operative site.23,24More broadly, with intraoperative arrest, if providers are engaged in activity that is expected to reverse the etiology of the arrest, they may defer defibrillation to focus on performing the critical actions as rapidly and reliably as possible.

Alternatively, unique processes of care may explain the lack of relationship between defibrillation and survival in the operating room. For example, some intraoperative patients may be treated with cardiopulmonary bypass before defibrillation. In addition, the operating room is the only location where a resuscitation expert with no surgical or procedural responsibilities (i.e ., the anesthesia provider) is guaranteed to be at the patient's bedside. Therefore, in the operating room, delays in defibrillation may be less likely to signal other quality gaps in patient care. Finally, our inability to detect an association between delayed defibrillation and survival for patients arresting in the operating room may have been due to an inadequate sample size of intraoperative arrests.

In many locations of the hospital, including periprocedural areas, prompt defibrillation can improve outcomes for patients presenting with ventricular fibrillation or pulseless ventricular tachycardia. The relationship is sufficiently robust that defibrillation within 2 min may be considered a useful process measure to indicate high-quality resuscitation care. This relationship was not confirmed in the operating room. We suggest that a very large database of intraoperative cardiopulmonary arrest is needed to evaluate the relative contributions of surgical variables and process variables on survival outcomes.

This analysis focused on patients with cardiac arrest caused by ventricular fibrillation or pulseless ventricular tachycardia and excluded other potentially inappropriate patients, such as those with preexisting implantable cardioverter-defibrillators. Drawing data from more than 250 hospitals, the large size of NRCPR, and its use of standardized definitions, suggests a representative sample of relatively reliable data.

Limitations of the NRCPR dataset include: (1) registry hospitals may not be representative of all US hospitals; (2) the accuracy of documented times may be problematic, particularly in the diagnostic suite where 25% of patients were not monitored at the time of their arrest; and (3) potentially important data elements were not available (e.g ., American Society of Anesthesiologists' functional status and emergency designation, the diagnosis of hemorrhage, the estimated blood loss, and the type of operative or nonoperative procedure). The variable used to designate hemorrhage in this analysis was a composite variable that may not accurately represent true hemorrhage. The variables designating cause of arrest were abstracted from the medical record by NRCPR research coordinators not involved in clinical care of the patient, and the accuracy of these recorded causes has not been validated. Because the registry only captures events that elicit a hospital response or require use of a crash cart, “arrests” in the cardiac catheterization laboratory and the cardiac operating room may be treated per routine and may not have been reported in the NRCPR.

Results may have differed for the 91 patients excluded because the time to defibrillation was missing. The patients excluded for missing time information were more likely to arrest in the operating room, and this subgroup likely includes a mixed population of relatively young trauma patients for whom prognosis was poor regardless of defibrillation times as well as cardiac surgical patients for whom other interventions (cardiopulmonary bypass) dictated survival. A total of 41 patients arrested because of conscious sedation, and of these 8 were missing information on time to defibrillation. Although this rate was significantly increased, the numbers are relatively small, and this finding needs to be confirmed in future analyses.

Finally, a relatively large number of analyses were conducted when compared with the number of actual events observed. Therefore, the associations presented in this paper need to be verified in future populations of patients with cardiopulmonary arrest.

This is the first paper to use the American Heart Association NRCPR to explore the effect of a process measure (delayed defibrillation) on survival in the perioperative environment. The current analysis suggests that this database may be improved for addressing questions about perioperative care by including procedural-specific variables such as the operative or nonoperative procedure, the primary anesthetic technique, the estimated blood loss before arrest, and the American Society of Anesthesiology physical examination status. Alternatively, questions of perioperative and periprocedural resuscitation quality may best be addressed by linking the NRCPR with more detailed databases of perioperative and periprocedural care.25 

Delays in defibrillation follow 13.8% of in-hospital intraoperative and periprocedural cardiac arrests, despite the presence of numerous medical and nursing personnel. These delays were associated with lower hospital survival for patients with periprocedural, but not intraoperative, cardiac arrests. More research is needed to define and explain the relationship between time to defibrillation and survival for intraoperative cardiac arrests.

The American Heart Association National Registry of CPR investigators includes: Paul S. Chan, M.D., M.Sc., Assistant Professor of Medicine, Bridgeport Hospital and Yale University School of Medicine, Bridgeport, Connecticut; Timothy Mader, M.D., Associate Professor, Department of Emergency Medicine, Baystate Medical Center/Tufts University School of Medicine, Springfield, Massachusetts; David J. Magid, M.D., M.P.H., Director of Research for CPMG in the Institute for Health Research, Kaiser Permanente Colorado Institute for Health Research, Denver, Colorado; Karl B. Kern, M.D., F.S.C.A.I., Professor of Medicine, Sarver Heart Center, University of Arizona, Tuscan, Arizona; Sam A. Warren, M.D., Acting Instructor, Harborview Medical Center and University of Washington-Harborview Center for Prehospital Emergency Care, University of Washington, Seattle, Washington; Nichol Graham, M.D., Professor, University of Washington-Harborview Center for Prehospital Emergency Care, Seattle, Washington; Joseph Ornato, M.D., F.A.C.P., F.A.C.C., F.A.C.E.P., Department Chairman, Medical Director of Richmond Ambulance Authority, Department of Emergency Medicine, Virginia Commonwealth University, Richmond, Virginia; Mary Ann Peberdy, M.D., Associate Professor of Medicine and Emergency Medicine Internal Medicine and Emergency Medicine, Virginia Commonwealth University, Richmond, Virginia; Romergryko Geocadin, M.D., Associate Professor of Neurology, Neurosurgery, and Anesthesiology & Critical Care Medicine, Director, The Johns Hopkins Hospital, Neu-rosciences Critical Care Unit Director, Division of Neurosciences Critical Care, The Johns Hopkins Medical Institutions, Baltimore, Maryland; R. Scott Braithwaite, M.D., M.Sc., F.A.C.P., Associate Professor of Internal Medicine, New York University School of Medicine, New York, New York; Mary E. Mancini, R.N., Ph.D., Associate Dean School of Nursing, The University of Texas Arlington, Arlington, Texas; Robert A. Berg, M.D., Division Chief, Critical Care Medicine, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania; Emilie Allen, B.S.N., Physician's Assistant, Rush University Medical Center, Chicago, Illinois; Elizabeth A. Hunt, M.D., M.P.H., Ph.D., Director of the Johns Hopkins Simulation Center, The Johns Hopkins Hospital, Baltimore, Maryland; Vinay M. Nadkarni, M.D., Medical Director, Center for Simulation, Advanced Education and Innovation, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania; Kathy Duncan, R.N., Field Manager, Faculty Institute for Healthcare Improvement, Cambridge, Massachusetts; Tanya Lane Truitt, R.N., M.S., Manager of Critical Care Education, Parkland Health and Hospital System, Dallas, Texas; Jerry Potts, Ph.D., Director of Science for the Emergency Cardiovascular Care Programs, American Heart Association National Center, Dallas, Texas; Brian Eigel, Ph.D., Assistant Director of Emergency Cardiovascular Care Programs, American Heart Association National Center, Dallas Texas