0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Original Investigation |

Hospital Variation in Time to Defibrillation After In-Hospital Cardiac Arrest FREE

Paul S. Chan, MD, MSc; Graham Nichol, MD, MPH; Harlan M. Krumholz, MD, SM; John A. Spertus, MD, MPH; Brahmajee K. Nallamothu, MD, MPH ; American Heart Association National Registry of Cardiopulmonary Resuscitation (NRCPR) Investigators
[+] Author Affiliations

Author Affiliations: Saint Luke's Mid-America Heart Institute, Kansas City, Missouri (Drs Chan and Spertus); University of Washington–Harborview Center for Prehospital Emergency Care, Seattle (Dr Nichol); Section of Cardiovascular Medicine and the Robert Wood Johnson Clinical Scholars Program, Department of Medicine, and Section of Health Policy and Administration, Department of Epidemiology and Public Health, Yale University School of Medicine, and Center for Outcomes Research and Evaluation, Yale–New Haven Hospital, New Haven, Connecticut (Dr Krumholz); and Veterans Affairs Ann Arbor Health Services Research and Development Center of Excellence, and Division of Cardiovascular Medicine, University of Michigan, Ann Arbor (Dr Nallamothu).


Arch Intern Med. 2009;169(14):1265-1273. doi:10.1001/archinternmed.2009.196.
Text Size: A A A
Published online

Background  Delays to defibrillation are associated with worse survival after in-hospital cardiac arrest, but the degree to which hospitals vary in defibrillation response times and hospital predictors of delays remain unknown.

Methods  Using hierarchical models, we evaluated hospital variation in rates of delayed defibrillation (>2 minutes) and its impact on survival among 7479 adult inpatients with cardiac arrests at 200 hospitals within the National Registry of Cardiopulmonary Resuscitation.

Results  Adjusted rates of delayed defibrillation varied substantially among hospitals (range, 2.4%-50.9%), with hospital-level effects accounting for a significant amount of the total variation in defibrillation delays after adjusting for patient factors. We found a 46% greater odds of patients with identical covariates getting delayed defibrillation at one randomly selected hospital compared with another. Among traditional hospital factors evaluated, however, only bed volume (reference category: <200 beds; 200-499 beds: odds ratio [OR], 0.62 [95% confidence interval {CI}, 0.48-0.80]; ≥500 beds: OR, 0.74 [95% CI, 0.53-1.04]) and arrest location (reference category: intensive care unit; telemetry unit: OR, 1.92 [95% CI, 1.65-2.22]; nonmonitored unit: OR, 1.90 [95% CI, 1.61-2.24]) were associated with differences in rates of delayed defibrillation. Wide variation also existed in adjusted hospital rates of survival to discharge (range, 5.3%-49.6%), with higher survival among hospitals in the top-performing quartile for defibrillation time (compared with the bottom quartile: OR for top quartile, 1.41 [95% CI, 1.11-1.77]).

Conclusions  Rates of delayed defibrillation vary widely among hospitals but are largely unexplained by traditional hospital factors. Given its association with improved survival, future research is needed to better understand best practices in the delivery of defibrillation at top-performing hospitals.

Figures in this Article

Given that in-hospital cardiac arrests are common and are associated with poor survival and neurological outcomes,1,2 efforts to minimize delays in defibrillating eligible patients are increasingly being recognized as an opportunity to improve the quality of care. A recent study found that as many as 30% of in-hospital cardiac arrests from ventricular arrhythmias are not treated within the American Heart Association's recommendation1 of 2 minutes, a delay that was associated with a 50% lower rate of in-hospital survival. In that study, patient-level factors, such as a noncardiac admitting diagnosis and after-hours cardiac arrests (evenings and weekends), were identified as clinically significant predictors of delayed defibrillation.

In contrast, much less is known about hospital-level variation in delays to defibrillation, a critical step toward sharing best practices so that all centers can improve their care of patients with cardiac arrest. Prior studies have generally been limited in assessing hospital-level factors, although smaller hospital size has been previously linked to longer defibrillation times.1 Thus, the extent of variation in defibrillation times across different hospitals and the potential factors that explain this variation remain largely unknown in the United States. An improved understanding of current hospital performance in defibrillation times and factors associated with delayed defibrillation is critical for developing effective interventions that could be implemented at hospitals to improve outcomes after in-hospital cardiac arrests.

Recently, the National Registry of Cardiopulmonary Resuscitation (NRCPR), a large national quality improvement resuscitation registry, completed a detailed survey of facility characteristics among its participating hospitals. The objectives of this investigation were to use the results of this survey to better understand hospital-level variation in rates of delayed defibrillation at acute-care hospitals in the United States and to quantify the contribution of patient- and hospital-level factors to hospital performance in defibrillation time.

STUDY DESIGN

The NRCPR is a large, prospective, voluntary registry of in-hospital cardiac arrests in the United States that was initiated in 2000. Its study design has been previously described in detail.2 Cardiac arrests in the NRCPR are defined as cessation of cardiac mechanical activity, determined by the absence of a palpable central pulse, apnea, and unresponsiveness. Consecutive patients with cardiac arrests and without do-not-resuscitation orders are screened by specially trained quality improvement personnel at participating hospitals. To ensure adequate capture, cases are identified by multiple methods, including centralized collection of cardiac arrest flow sheets, reviews of hospital paging system logs, routine checks for code cart usage, pharmacy tracer drug records for resuscitation medications, and screening for code cart charges from hospital billing offices.1,2

Data on cardiac arrests are collected using standardized Utstein-style definitions, which are a template of precisely defined variables for uniform reporting developed by international experts.37 Data accuracy in the NRCPR is ensured by certification of research staff, use of case-study methods for newly enrolled hospitals to ensure operational definition compliance prior to data acceptance, and a periodic reabstraction process, which has been demonstrated to have a mean error rate of 2.4% for all data.2 Moreover, the software for data submission has more than 250 built-in data checks to ensure data completeness and to alert the data entry person of outlying responses.2,8

All patients are assigned a unique code during a single hospitalization, and deidentified data are transmitted to a central repository (Digital Innovation, Forest Hill, Maryland), in compliance with the Health Insurance Portability and Accountability Act. Oversight for the entire process of data collection, integrity, and research is provided by the American Heart Association.

NRCPR SURVEY OF HOSPITAL FACILITIES AND STUDY POPULATION

In an effort to explore potential reasons for hospital-level variation in resuscitation care, hospitals actively participating in the NRCPR were asked in 2006 to complete a detailed survey that provided information on several facility characteristics. Data elements included structural characteristics, such as geographic location, hospital teaching status (nonteaching, residency programs only, residency and fellowship programs), number of adult inpatient beds (<200, 200-499, ≥500), intensive care unit (ICU) bed and monitored bed to total adult bed ratio, mean annual in-hospital cardiac arrests per 1000 admissions in 2005 (analyzed by tertiles), mean daily ICU and non-ICU admission volumes, and annual hospital-wide mortality rate. Key process-of-care measures available from the survey included whether medical emergency teams (METs) or automatic external defibrillators (AEDs) were in place at the participating hospital at the time of the survey.

The survey completion rate was 79% (240 of 304 general nonpediatric hospitals). Our analysis was limited to the 200 acute-care hospitals that (1) completed the facility survey (64 hospitals were excluded), (2) had mean annual cardiac arrest volumes of 5 or greater (40 hospitals comprising 288 cardiac arrests were excluded), and (3) provided at least 6 months of data between January 1, 2000, and January 31, 2008. Within these hospitals, we identified 13 506 patients 18 years or older with an index in-hospital cardiac arrest in which the first identifiable rhythm was ventricular fibrillation (VF) or pulseless ventricular tachycardia (VT).

We limited our study population to patients located in inpatient units at the time of the cardiac arrest. We excluded cardiac arrests occurring in the emergency department, operating suites, procedure areas (cardiac catheterization, electrophysiology, and angiography suites), and postprocedural areas (n = 3192) owing to the distinctive clinical circumstances and subsequent resuscitation responses associated with these environments. We also excluded patients with (1) implantable cardioverter-defibrillators (n = 246); (2) intravenous infusions of acute cardiac life support protocol medications for pulseless VT and VF (epinephrine, amiodarone, lidocaine, or procainamide) at the time of cardiac arrest (n = 1467); and (3) missing (n = 1099) or inconsistent (n = 23) data on the time of the cardiac arrest or defibrillation. Patients who were excluded because of missing data were more frequently of black race and were less likely to have a cardiac admitting diagnosis, a history or admission diagnosis of myocardial infarction, and VF as their initial presenting rhythm (eTable 1, http://www.archinternmed.com). The final study sample comprised 7479 patients from the 200 hospitals.

CLINICAL DATA AND OUTCOME MEASURES

Patient records included data on demographics (age, sex, and race), initial cardiac rhythm (VF, pulseless VT), admitting diagnosis category (cardiac, noncardiac), time of cardiac arrest (work hours, 8 AM to 5 PM; after hours, 5 PM to 8 AM; or weekends), cardiac arrest location (ICU, telemetry unit, or nonmonitored unit) and the use of a hospital-wide alert during cardiac arrest. Clinical data were also available on arrest characteristics such as presence of congestive heart failure or myocardial infarction during admission, comorbidities or medical conditions present prior to cardiac arrest (history of congestive heart failure, myocardial infarction, or diabetes mellitus; renal or respiratory insufficiency; baseline evidence of motor, cognitive, or functional deficits [central nervous system depression]; acute stroke; pneumonia; hypotension; sepsis; or major trauma), and use of therapeutic interventions at the time of the cardiac arrest (mechanical ventilation, intra-aortic balloon pump, and pulmonary artery catheter).

The primary outcome for our analysis was the hospital rate of delayed defibrillation, which we defined based on prior work1 and current guidelines9,10 as a time to defibrillation of greater than 2 minutes. Time to defibrillation was calculated as the reported time from initial recognition of the cardiac arrest to the reported time of first attempted defibrillation. Both of these times were identified from cardiac arrest documentation in the patient's medical records and recorded at the minute level. In sensitivity analyses, time to defibrillation was also assessed as 4 discrete categories (≤2 minutes, 3 or 4 minutes, 5 or 6 minutes, and >6 minutes). Finally, we evaluated the relationship between hospital rates of delayed defibrillation with outcomes such as survival of initial code (return of spontaneous circulation for at least 20 minutes) and survival to hospital discharge.

STATISTICAL ANALYSES

To examine the extent of hospital-level variation in defibrillation times, we first calculated the rates of delayed defibrillation for each hospital. We then compared patient and hospital-level characteristics across quartiles of hospital performance in rates of delayed defibrillation using Mantel-Haenszel trend tests.

Two-level multivariable hierarchical models were used to assess the relationship of delayed defibrillation with the patient and hospital-level factors described in the previous subsection. Hierarchical models would adjust for the nonindependence (ie, clustering) of outcomes at the hospital level so as to avoid overestimation of the significance of statistical associations.11 Because hospitals may begin and end participation in the NRCPR at different time points throughout the study, hierarchical models would also permit adjustment for the nonindependence of outcomes within reporting years.11 Therefore, both the hospital site (random intercept of the model) and study calendar year were modeled as random effects in our analyses. Patient and hospital-level characteristics previously described were modeled as fixed effects, as was the duration of hospital participation within the NRCPR for each cardiac arrest.

We used a 2-stage approach to the evaluation of hospital variation in defibrillation time. We first used the median odds ratio (OR) to quantify the extent to which variations in rates of delayed defibrillation were explained at the cluster site level (ie, by differences across hospitals).12,13 The median OR is always 1 or greater (a median OR of 1 suggests no variation between clusters), does not have a confidence interval (CI), and is determined from hierarchical models with only patient-level factors included. In our analysis, the median OR can be interpreted as the odds that 2 patients with identical patient-level covariates from separate, randomly chosen hospitals will receive delayed defibrillation for a cardiac arrest. For example, a median OR of 1.50 suggests a 50% higher odds of receiving delayed defibrillation at one randomly selected hospital when compared with another for the same patient. In addition to precisely measuring the degree of variation explained by hospital-level effects, the median OR permits meaningful comparisons with the effect sizes of patient factors (eg, sex) included in the hierarchical models, thus overcoming interpretational limitations that are inherent with the intraclass correlation coefficient.13,14 After quantifying the contribution of site variation, we then constructed hierarchical models with both patient and hospital-level characteristics to identify which facility characteristics were associated with differences in rates of delayed defibrillation across hospitals. In a sensitivity analysis, we constructed hierarchical ordinal logistic regression models and evaluated time to defibrillation as 4 discrete categories as described in the previous subsection.

Finally, we examined the extent to which variations in rates of delayed defibrillation across hospitals were associated with survival. In these analyses, survival of initial code and survival to hospital discharge were dependent variables, and hospital quartiles of delayed defibrillation were included as an additional covariate with the previously described patient and hospital-level characteristics in multivariable hierarchical models. For all analyses, data on study covariates and survival outcomes were 100% complete. The null hypothesis was evaluated at a 2-sided significance level of .05 with 95% CIs. All analyses were conducted with SAS statistical software (version 9.1; SAS Institute Inc, Cary, North Carolina) and R open-source statistical software (version 2.6.2; SAS).

The institutional review board of the Mid-America Heart Institute approved this study and waived the requirement for informed consent.

The mean (SD) age in the study cohort of 7479 patients was 67 (15) years. A cardiac admitting diagnosis was noted at presentation in 55% of patients, and 1 in 4 patients was hospitalized with either a diagnosis of myocardial infarction or heart failure (Table 1). The initial cardiac arrest rhythm was VF in 64% and pulseless VT in 36% of patients, and over 70% of cardiac arrests occurred during nonworking hours or on weekends. Approximately one-quarter of patients were diagnosed as having hypotension prior to cardiac arrest and 29% were on mechanical ventilation. Finally, most cardiac arrests occurred in the ICU, whereas 1 in 6 arrests occurred in nonmonitored units.

Table Graphic Jump LocationTable 1. Description of the Patient Samples, Stratified by Hospital Quartiles for Rates of Delayed Defibrillationa

The mean (SD) number of eligible cardiac arrests contributed by each hospital was 37 (31) (median, 28; interquartile range [IQR], 14-51; range, 8-192). Most cardiac arrest cases were submitted from 2003 through 2007 (eFigure 1), and 18 hospitals submitted cases throughout the entire study period. Hospitals were distributed relatively evenly across US geographical locations (Table 2). Most hospitals were intermediate in size (200-499 beds), 27% had fewer than 200 beds, and 19% had 500 or more beds. Most were teaching hospitals (23% with residency programs only and 45% with both residency and fellowship programs) and had implemented MET or AED programs. The distribution of hospitals by daily ICU and non-ICU volumes, annual hospital-wide mortality rate, cardiac arrests per 1000 admissions, and ratio of monitored beds to total beds is also described in Table 2.

Table Graphic Jump LocationTable 2. Description of the Hospital Samples, Stratified by Hospital Quartiles for Rates of Delayed Defibrillationa
HOSPITAL PERFORMANCE

The mean (SD) time to defibrillation was 1 (2) minutes, with a skewed right-tail distribution (median time of <1 minute; IQR, <1 to 2 minutes). Overall, 1369 of 7479 patients (18.3%) had a defibrillation time of more than 2 minutes, with 396 (5.3%) treated at 3 minutes, 254 (3.4%) treated at 4 minutes, 282 (3.8%) treated at 5 minutes, 164 (2.2%) at 6 minutes, and 273 (3.7%) treated at more than 6 minutes.

Hospital performance in defibrillation time varied widely. Unadjusted rates of delayed defibrillation lasting more than 2 minutes ranged from 2.0% to 71.4% across hospitals and exceeded 20% of all arrests in 45% of hospitals (91 of 200) (Figure 1A). Variations in rates of delayed defibrillation persisted when we examined only those hospitals at or above the median arrest case volume (range, 2.0%-37.1%), exceeding 20% in 35% of hospitals (36 of 101), or when the analysis was restricted to cardiac arrests occurring only within the ICU (eFigure 2). When examined by quartiles of hospital performance (rate of delayed defibrillation <12%, 12% to <18%, 18% to <25%, and >25%), there was a trend for hospitals with higher bed volumes (>200 beds) and with higher proportion of arrests in ICUs to be in quartiles with lower rates of delayed defibrillation time (Table 2). In addition, patient characteristics were generally similar across quartiles of hospital performance in defibrillation time with the exception for rates of myocardial infarction before and during index admission and the presence of hypotension or central nervous system depression before the cardiac arrest (Table 1).

Place holder to copy figure label and caption
Figure 1.

Distribution of hospital rates of delayed defibrillation. Wide variation in hospital rates of delayed defibrillation time (>2 minutes) was seen in the study cohort (A) and persisted after adjustment for patient and hospital characteristics (B).

Graphic Jump Location

After adjusting for patient characteristics only, a substantial amount of the variation in the rates of delayed defibrillation time continued to be explained by differences at the hospital level. The median OR for the hospital effect was 1.46, suggesting a 46% greater odds of patients with identical covariates receiving delayed defibrillation at one randomly selected hospital compared with another. Importantly, the median OR was larger than the adjusted ORs of any of the patient-level factors (Table 3), which suggests that hospital factors explained a greater proportion of the variation in rates of delayed defibrillation than any individual patient factor. When the model was adjusted further for individual hospital-level characteristics, a larger hospital size was found to be associated with lower rates of delayed defibrillation (reference category: hospitals with <200 beds; for hospitals with 200-499 beds: OR, 0.62 [95% CI, 0.48-0.80]; for hospitals with ≥500 beds: OR, 0.74 [95% CI, 0.53-1.04]; P value for trend, <.001), whereas non-ICU arrest location was associated with higher rates of delayed defibrillation (reference category: ICU; for telemetry unit arrests: OR, 1.92 [95% CI, 1.65-2.22]; for nonmonitored unit arrests: OR, 1.90 [95% CI, 1.61-2.24]; P value for trend, <.001) (Table 4). In contrast, most other hospital characteristics were not associated with delayed defibrillation, including academic status, geographical location, arrest volume per 1000 admissions, daily ICU and non-ICU admission volume, duration of hospital participation within NRCPR, and hospitals with MET or AED programs. Even after adjustment for patient and hospital-level characteristics, hospital performance in defibrillation time continued to vary widely with adjusted rates of delayed defibrillation of more than 2 minutes (range, 2.4%-50.9%) (Figure 1B).

Table Graphic Jump LocationTable 3. Site Effect on Defibrillation Time
Table Graphic Jump LocationTable 4. Association of Hospital Factors With Defibrillation Timea

In sensitivity analyses, we found that these results were essentially unchanged when we evaluated time to defibrillation as 4 discrete time categories of less than 2 minutes, 3 or 4 minutes, 5 or 6 minutes, and more than 6 minutes. Significant hospital variation in defibrillation times remained (median OR, 1.46) and were largely unexplained by facility characteristics (eTables 2, 3, and 4).

SURVIVAL OUTCOMES

Overall, 4770 patients (63.8%) experienced a return of spontaneous circulation, and 2555 (34.2%) survived to hospital discharge. There was a large variation among hospitals in adjusted rates of survival to hospital discharge, ranging from 5.3% to 49.6% (Figure 2A), with higher rates of survival among hospitals in the top-performing quartiles of defibrillation time (Figure 2B). Defibrillation time was a stronger predictor of survival than overall differences at the hospital level. For instance, the median OR for the hospital effect was 1.29 for return of spontaneous circulation compared with an adjusted OR of 1.52 (95% CI, 1.35-1.72) for rapid defibrillation within 2 minutes (ie, no delayed defibrillation). Similarly, the median OR for the hospital effect was 1.35 for survival to discharge compared with an adjusted OR of 1.54 (95% CI, 1.33-1.75) for rapid defibrillation.

Place holder to copy figure label and caption
Figure 2.

Distribution of hospital rates of survival to discharge. A, Significant variation in risk-adjusted hospital rates of survival to discharge after cardiac arrest was observed for the overall cohort. B, Hospitals in the top-performing quartiles for defibrillation time (ie, those with the lowest rates of delays in defibrillation time) had higher unadjusted rates of survival to discharge (P = .05).

Graphic Jump Location

After adjustment for patient and hospital-level characteristics, hospitals in the top-performing quartiles for defibrillation time had higher rates of survival to discharge (reference category: quartile 4 [delayed defibrillation, >25% of arrests]; quartile 3 [18% to <25%]: OR, 1.18 [95% CI, 0.93-1.49]; quartile 2 [12% to <18%]: OR, 1.26 [95% CI, 1.00-1.60]; quartile 1 [<12%]: OR, 1.41 [95% CI, 1.11-1.77]; P value for trend, .03). Otherwise, the only other hospital-level factor associated with differences in rates of survival to hospital discharge was hospital arrest location (reference category: ICU unit; for telemetry unit: OR, 1.35 [95% CI, 1.19-1.54]; for nonmonitored unit: OR, 0.73 [95% CI, 0.62-0.86]; P value for trend, <.001) (eTable 5).

Substantial variation in hospital performance in defibrillation times exists across the United States, with overall rates of delayed defibrillation ranging from 2% to 51%. Importantly, hospital-level differences explain a sizable amount of this variation and were a stronger predictor than any specific patient characteristic. However, many of the individual hospital characteristics that we explored—such as volume, academic status, and hospital-wide mortality rate—were unrelated to hospital performance in defibrillation time. This lack of correlation between “conventional” hospital-level factors and defibrillation time suggests that other unmeasured characteristics are responsible for certain institutions achieving extremely low rates of delayed defibrillation. We also found that lower hospital rates of delayed defibrillation was a major predictor of survival after cardiac arrests, with survival to discharge being 41% higher in facilities in the top quartile (1) of defibrillation time performance compared with those in the bottom quartile (4). Given extensive differences in defibrillation time across institutions and the recognized impact of delayed defibrillation on survival, new approaches to improve hospital performance in defibrillation time could represent a critical area for quality improvement.

Although higher volume has been shown to be associated with improved outcomes for a number of surgical and medical procedures,1518 we did not find a relationship between in-hospital cardiac arrest volume and time to defibrillation. We also did not find an association between the presence of AED or MET programs at a hospital and rates of delayed times to defibrillation. However, this analysis does not distinguish between arrests occurring before and after implementation of an AED or MET program in a given hospital and therefore could not address whether their actual implementation improved outcomes.

Like other complex clinical care and logistical processes, rapid responses to resuscitation raise substantial challenges. They require rapid hospital staff recognition of cardiac arrest, rhythm identification, and prompt mobilization of defibrillator therapy to the patient's bedside. Therefore, differences in hospital performance of defibrillation time are likely to be explained by how hospitals approach the resuscitation process—whether hospitals learn from prior resuscitations, develop innovations to reduce the time to defibrillation, and are willing to invest in quality improvement and leadership. The paucity of conventional hospital-level characteristics associated with performance of defibrillation time may reflect a true opportunity for future improvement because it suggests the importance of process of care measures in achieving best practices. Indeed, many of the characteristics we evaluated are not even readily amenable to modification (eg, admission volume, academic status, and annual mortality rate). The association between bed volume and defibrillation time also seen in this study may actually reflect the impact of quality improvement efforts or even active intervention trials to improve resuscitation outcomes at larger hospitals rather than the hospital size itself.

Clearly, given the large variation in defibrillation times observed across hospitals and its implications for survival, there is an imperative for a systematic approach in identifying what processes top-performing hospitals have implemented successfully to shorten time to defibrillation. Achieving rapid resuscitation response times may require several novel approaches. These could include innovations related to (1) improving early access to defibrillation (eg, the use of portable automatic external defibrillators, empowerment of allied staff to initiate defibrillation before physician arrival), (2) multifaceted quality improvement initiatives (eg, routine conduct of mock codes, systematic and immediate debriefing after cardiopulmonary arrests), and (3) dynamic leadership structures (eg, prioritization of in-hospital resuscitation performance, multidisciplinary teams for quality improvement). As our findings have shown, however, the actual qualities that distinguish top hospital performers are unlikely to be captured with traditional facility characteristics from data registries, which focus on hospital structure. Instead, a more thorough and detailed assessment, driven primarily by site surveys within large registries like the NRCPR, or the use of qualitative or mixed-methods approaches is required to better understand the reasons why some hospitals are able to consistently achieve rapid defibrillation response times while others do not. Once these key hospital interventions are identified, validation studies are needed to demonstrate that their implementation across all hospitals leads to improvements in patient outcomes. Recently, this approach has been used successfully for door-to-balloon times for percutaneous coronary intervention, another time-based quality metric that has been tightly linked to patient outcome.1921 In the meanwhile, all hospitals should vigilantly monitor their own times to defibrillation, identify barriers that lead to delays at their institution, and begin to develop strategies to eliminate these barriers.

Although few in number, preliminary studies of process-of-care measure interventions that may influence hospital performance for in-hospital arrests have been encouraging. One single-center study22 trained non-ICU nurses to perform defibrillation hospital-wide, which resulted in nurses administering defibrillation in 46% of arrests prior to the arrival of a cardiac arrest team member. This resulted in a nonsignificant increase in survival to hospital discharge from 41% to 55% after intervention.22 A recent systematic review23 suggests that the emergence of AED technology also supports the concept that nurses provide patients who have undergone cardiac arrest with prompt and safe defibrillation. More recently, data feedback through a performance debriefing intervention among hospital staff immediately after each cardiopulmonary arrest resuscitation event was found to improve return of spontaneous circulation from 45% to 59%.24 These preliminary studies suggest that defibrillation times and resuscitation performance at hospitals are dynamic and can be improved with innovation and committed leadership. Given that it has been estimated that 500 000 patients experience in-hospital cardiac arrests annually in the United States,25 of which 100 000 are caused by VF or pulseless VT, as many as 2000 more patients could survive to discharge each year if hospitals in the lower 3 quartiles of defibrillation time performance achieved the performance of hospitals in the top quartile.

Our study should be interpreted in the context of the following limitations. Although we evaluated a number of facility-level characteristics, our analysis lacked information on some key hospital characteristics (eg, nurse-to-patient ratio, use of hypothermia) and process of care measures (eg, data feedback), as well as other patient factors (eg, cardiopulmonary resuscitation quality and adequacy). In addition, we assessed only the extent of variation in delayed defibrillation among 200 hospitals that voluntarily contributed data within a quality improvement registry. Both factors may have limited our ability to detect significant associations among some hospital-level factors and performance in defibrillation times. Moreover, performance at NRCPR hospitals may be different than non-NRCPR hospitals. Nevertheless, we believe the use of the NRCPR was a key strength of our study. The NRCPR remains the largest contemporary registry of in-hospital cardiac arrests, and its use of standardized Utstein-style definitions and built-in data entry checks facilitated data accuracy, reliability, and completeness.

Second, because the documentation of time within the NRCPR was at the minute level, a defibrillation time of 2 minutes may potentially represent any time greater than 1 minute and up to 3 minutes. However, we have no reason to believe that the approximation of time would not be normally distributed in this large population (ie, the time at each minute level was rounded up as frequently as it was rounded down) such that any misclassification of time, if present, would bias the relationship between hospital performance of defibrillation time and survival toward the null. Moreover, this study focused on hospital-level variation rather than patient-level effects, so this concern is less likely to have an impact on our findings.

Third, despite adjustment for a number of patient and hospital-level factors, our study used an observational design and is subject to unmeasured confounding from either unmeasured variables or inadequate adjustment for existing covariates. Fourth, our analysis was limited to those hospitals that completed the facility survey in 2006 and those patients with available times to defibrillation. It is also possible that some factors in the survey, such as the use of METs or AEDs, varied over the study period, although most characteristics were related to hospital structure and are unlikely to have changed over time. Fifth, the NRCPR does not collect information on whether resuscitations were “slow codes,” in which patients perceived by the medical staff to have a poor prognosis received perfunctory resuscitation efforts. However, we believe that slow codes represent a small fraction of all cardiac arrests within the NRCPR and are unlikely to explain the observed hospital variation in defibrillation time and survival. Finally, time to defibrillation in the NRCPR is determined from reported times of cardiac arrest and defibrillation. The lack of time synchronization between cardiac monitors and defibrillators and the use of multiple clocks may lead to discrepancies in determining the time to defibrillation.26 However, we have no reason to believe that such discrepancies would have occurred differently across hospitals.

In conclusion, we found clinically significant and important variations in hospital performance of defibrillation times for patients with in-hospital cardiac arrests. Differences across hospitals explained a substantial degree of the variation in rates of delayed defibrillation time, but few facility characteristics were found to explain this variation. New approaches for identifying hospital innovations in process-of-care measures and resuscitation quality that are associated with improved performance in defibrillation times are needed so that hospitals can minimize delays in defibrillation and improve survival for patients with cardiac arrests.

Correspondence: Paul S. Chan, MD, MSc, Mid-America Heart Institute, Fifth Floor, 4401 Wornall Rd, Kansas City, MO 64111 (pchan@cc-pc.com).

Accepted for Publication: March 30, 2009.

Author Contributions: Dr Chan had full access to all of the data and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Chan and Nallamothu. Acquisition of data: Nichol. Analysis and interpretation of data: Chan, Nichol, Krumholz, Spertus, and Nallamothu. Drafting of the manuscript: Chan. Critical revision of the manuscript for important intellectual content: Chan, Nichol, Krumholz, Spertus, and Nallamothu. Statistical analysis: Chan. Study supervision: Chan.

NRCPR Investigators: The American Heart Association NRCPR Investigators include Drs Chan and Nichol, as well as Emilie Allen, RN, BSN; Robert Berg, MD; Scott Braithwaite, MD; Kathy Duncan, RN; Brian Eigel, PhD; Romergryko Geocadin, MD; Elizabeth Hunt, MD; Karl Kern, MD; Tim Mader, MD; David Magid, MD; Mary Mancini, RN, PhD; Vinay Nadkarni, MD; Thomas Noel, MD; Joseph Ornato, MD; Mary Ann Peberdy, MD; Jerry Potts, PhD; Tanya Truitt, RN, MS; and Sam Warren, MD.

Financial Disclosure: Dr Nichol has served as a consultant to Northfield Laboratories; has received travel compensation from INNERcool Inc and Radiant Medical Inc; has received research grant funding from Medtronic and the National Heart Lung and Blood Institute; has received equipment donations for overseas medical missions from Medtronic Physio-Control and Laerdal Inc; and has served on advisory boards with the American Heart Association and the Medic One Foundation.

Funding/Support: The American Heart Association provides operational funding for the NRCPR.

Additional Information: The final manuscript draft was approved by the American Heart Association"s Scientific Committee. The eTables and eFigures are available at http://www.archinternmed.com.

Chan  PSKrumholz  HMNichol  GNallamothu  BKAmerican Heart Association National Registry of Cardiopulmonary Resuscitation Investigators, Delayed time to defibrillation after in-hospital cardiac arrest. N Engl J Med 2008;358 (1) 9- 17
PubMed Link to Article
Peberdy  MAKaye  WOrnato  JP  et al.  Cardiopulmonary resuscitation of adults in the hospital. Resuscitation 2003;58 (3) 297- 308
PubMed Link to Article
Jacobs  INadkarni  VBahr  J  et al. International Liaison Committee on Resuscitation; American Heart Association; European Resuscitation Council; Australian Resuscitation Council; New Zealand Resuscitation Council; Heart and Stroke Foundation of Canada; InterAmerican Heart Foundation; Resuscitation Councils of Southern Africa; ILCOR Task Force on Cardiac Arrest and Cardiopulmonary Resuscitation Outcomes, Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004;110 (21) 3385- 3397
PubMed Link to Article
Cummins  ROChamberlain  DHazinski  MF  et al. American Heart Association, Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation. Circulation 1997;95 (8) 2213- 2239
PubMed Link to Article
Cummins  ROSanders  AMancini  EHazinski  MF In-hospital resuscitation: executive summary. Ann Emerg Med 1997;29 (5) 647- 649
PubMed Link to Article
Zaritsky  ANadkarni  VHazinski  MF  et al. Writing Group, Recommended guidelines for uniform reporting of pediatric advanced life support. Circulation 1995;92 (7) 2006- 2020
PubMed Link to Article
Zaritsky  ANadkarni  VHazinski  MF  et al.  Recommended guidelines for uniform reporting of pediatric advanced life support. Resuscitation 1995;30 (2) 95- 115
PubMed Link to Article
Peberdy  MAOrnato  JPLarkin  GL  et al. National Registry of Cardiopulmonary Resuscitation Investigators, Survival from in-hospital cardiac arrest during nights and weekends. JAMA 2008;299 (7) 785- 792
PubMed Link to Article
Ewy  GAOrnato  JP 31st Bethesda Conference: emergency cardiac care: task force 1: cardiac arrest. J Am Coll Cardiol 2000;35 (4) 832- 846
PubMed Link to Article
Cummins  ROOrnato  JPThies  WHPepe  PE Improving survival from sudden cardiac arrest: the “chain of survival” concept. Circulation 1991;83 (5) 1832- 1847
PubMed Link to Article
Goldstein  H  Multilevel Statistical Models.  London, England Edward Arnold1995;
Larsen  KPetersen  JHBudtz-Jorgensen  EEndahl  L Interpreting parameters in the logistic regression model with random effects. Biometrics 2000;56 (3) 909- 914
PubMed Link to Article
Larsen  KMerlo  J Appropriate assessment of neighborhood effects on individual health. Am J Epidemiol 2005;161 (1) 81- 88
PubMed Link to Article
Goldstein  HBrowne  WRasbash  J Partitioning variation in multilevel models. Underst Stat 2002;1223- 232
Link to Article
Birkmeyer  JDStukel  TASiewers  AEGoodney  PPWennberg  DELucas  FL Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349 (22) 2117- 2127
PubMed Link to Article
Dudley  RAJohansen  KLBrand  RRennie  DJMilstein  A Selective referral to high-volume hospitals. JAMA 2000;283 (9) 1159- 1166
PubMed Link to Article
Canto  JGEvery  NRMagid  DJ  et al. National Registry of Myocardial Infarction 2 Investigators, The volume of primary angioplasty procedures and survival after acute myocardial infarction. N Engl J Med 2000;342 (21) 1573- 1580
PubMed Link to Article
Halm  EALee  CChassin  MR Is volume related to outcome in health care? Ann Intern Med 2002;137 (6) 511- 520
PubMed Link to Article
Bradley  EHHerrin  JWang  Y  et al.  Strategies for reducing the door-to-balloon time in acute myocardial infarction. N Engl J Med 2006;355 (22) 2308- 2320
PubMed Link to Article
Bradley  EHCurry  LAWebster  TR  et al.  Achieving rapid door-to-balloon times. Circulation 2006;113 (8) 1079- 1085
PubMed Link to Article
Bradley  EHRoumanis  SARadford  MJ  et al.  Achieving door-to-balloon times that meet quality guidelines. J Am Coll Cardiol 2005;46 (7) 1236- 1241
PubMed Link to Article
Coady  EM A strategy for nurse defibrillation in general wards. Resuscitation 1999;42 (3) 183- 186
PubMed Link to Article
Kenward  GCastle  NHodgetts  TJ Should ward nurses be using automatic external defibrillators as first responders to improve the outcome from cardiac arrest? a systematic review of the primary research. Resuscitation 2002;52 (1) 31- 37
PubMed Link to Article
Edelson  DPLitzinger  BArora  V  et al.  Improving in-hospital cardiac arrest process and outcomes with performance debriefing. Arch Intern Med 2008;168 (10) 1063- 1069
PubMed Link to Article
Eisenberg  MSMengert  TJ Cardiac resuscitation. N Engl J Med 2001;344 (17) 1304- 1313
PubMed Link to Article
Kaye  WMancini  METruitt  TL When minutes count: the fallacy of accurate time documentation during in-hospital resuscitation. Resuscitation 2005;65 (3) 285- 290
PubMed Link to Article

Figures

Place holder to copy figure label and caption
Figure 1.

Distribution of hospital rates of delayed defibrillation. Wide variation in hospital rates of delayed defibrillation time (>2 minutes) was seen in the study cohort (A) and persisted after adjustment for patient and hospital characteristics (B).

Graphic Jump Location
Place holder to copy figure label and caption
Figure 2.

Distribution of hospital rates of survival to discharge. A, Significant variation in risk-adjusted hospital rates of survival to discharge after cardiac arrest was observed for the overall cohort. B, Hospitals in the top-performing quartiles for defibrillation time (ie, those with the lowest rates of delays in defibrillation time) had higher unadjusted rates of survival to discharge (P = .05).

Graphic Jump Location

Tables

Table Graphic Jump LocationTable 1. Description of the Patient Samples, Stratified by Hospital Quartiles for Rates of Delayed Defibrillationa
Table Graphic Jump LocationTable 2. Description of the Hospital Samples, Stratified by Hospital Quartiles for Rates of Delayed Defibrillationa
Table Graphic Jump LocationTable 3. Site Effect on Defibrillation Time
Table Graphic Jump LocationTable 4. Association of Hospital Factors With Defibrillation Timea

References

Chan  PSKrumholz  HMNichol  GNallamothu  BKAmerican Heart Association National Registry of Cardiopulmonary Resuscitation Investigators, Delayed time to defibrillation after in-hospital cardiac arrest. N Engl J Med 2008;358 (1) 9- 17
PubMed Link to Article
Peberdy  MAKaye  WOrnato  JP  et al.  Cardiopulmonary resuscitation of adults in the hospital. Resuscitation 2003;58 (3) 297- 308
PubMed Link to Article
Jacobs  INadkarni  VBahr  J  et al. International Liaison Committee on Resuscitation; American Heart Association; European Resuscitation Council; Australian Resuscitation Council; New Zealand Resuscitation Council; Heart and Stroke Foundation of Canada; InterAmerican Heart Foundation; Resuscitation Councils of Southern Africa; ILCOR Task Force on Cardiac Arrest and Cardiopulmonary Resuscitation Outcomes, Cardiac arrest and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004;110 (21) 3385- 3397
PubMed Link to Article
Cummins  ROChamberlain  DHazinski  MF  et al. American Heart Association, Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation. Circulation 1997;95 (8) 2213- 2239
PubMed Link to Article
Cummins  ROSanders  AMancini  EHazinski  MF In-hospital resuscitation: executive summary. Ann Emerg Med 1997;29 (5) 647- 649
PubMed Link to Article
Zaritsky  ANadkarni  VHazinski  MF  et al. Writing Group, Recommended guidelines for uniform reporting of pediatric advanced life support. Circulation 1995;92 (7) 2006- 2020
PubMed Link to Article
Zaritsky  ANadkarni  VHazinski  MF  et al.  Recommended guidelines for uniform reporting of pediatric advanced life support. Resuscitation 1995;30 (2) 95- 115
PubMed Link to Article
Peberdy  MAOrnato  JPLarkin  GL  et al. National Registry of Cardiopulmonary Resuscitation Investigators, Survival from in-hospital cardiac arrest during nights and weekends. JAMA 2008;299 (7) 785- 792
PubMed Link to Article
Ewy  GAOrnato  JP 31st Bethesda Conference: emergency cardiac care: task force 1: cardiac arrest. J Am Coll Cardiol 2000;35 (4) 832- 846
PubMed Link to Article
Cummins  ROOrnato  JPThies  WHPepe  PE Improving survival from sudden cardiac arrest: the “chain of survival” concept. Circulation 1991;83 (5) 1832- 1847
PubMed Link to Article
Goldstein  H  Multilevel Statistical Models.  London, England Edward Arnold1995;
Larsen  KPetersen  JHBudtz-Jorgensen  EEndahl  L Interpreting parameters in the logistic regression model with random effects. Biometrics 2000;56 (3) 909- 914
PubMed Link to Article
Larsen  KMerlo  J Appropriate assessment of neighborhood effects on individual health. Am J Epidemiol 2005;161 (1) 81- 88
PubMed Link to Article
Goldstein  HBrowne  WRasbash  J Partitioning variation in multilevel models. Underst Stat 2002;1223- 232
Link to Article
Birkmeyer  JDStukel  TASiewers  AEGoodney  PPWennberg  DELucas  FL Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349 (22) 2117- 2127
PubMed Link to Article
Dudley  RAJohansen  KLBrand  RRennie  DJMilstein  A Selective referral to high-volume hospitals. JAMA 2000;283 (9) 1159- 1166
PubMed Link to Article
Canto  JGEvery  NRMagid  DJ  et al. National Registry of Myocardial Infarction 2 Investigators, The volume of primary angioplasty procedures and survival after acute myocardial infarction. N Engl J Med 2000;342 (21) 1573- 1580
PubMed Link to Article
Halm  EALee  CChassin  MR Is volume related to outcome in health care? Ann Intern Med 2002;137 (6) 511- 520
PubMed Link to Article
Bradley  EHHerrin  JWang  Y  et al.  Strategies for reducing the door-to-balloon time in acute myocardial infarction. N Engl J Med 2006;355 (22) 2308- 2320
PubMed Link to Article
Bradley  EHCurry  LAWebster  TR  et al.  Achieving rapid door-to-balloon times. Circulation 2006;113 (8) 1079- 1085
PubMed Link to Article
Bradley  EHRoumanis  SARadford  MJ  et al.  Achieving door-to-balloon times that meet quality guidelines. J Am Coll Cardiol 2005;46 (7) 1236- 1241
PubMed Link to Article
Coady  EM A strategy for nurse defibrillation in general wards. Resuscitation 1999;42 (3) 183- 186
PubMed Link to Article
Kenward  GCastle  NHodgetts  TJ Should ward nurses be using automatic external defibrillators as first responders to improve the outcome from cardiac arrest? a systematic review of the primary research. Resuscitation 2002;52 (1) 31- 37
PubMed Link to Article
Edelson  DPLitzinger  BArora  V  et al.  Improving in-hospital cardiac arrest process and outcomes with performance debriefing. Arch Intern Med 2008;168 (10) 1063- 1069
PubMed Link to Article
Eisenberg  MSMengert  TJ Cardiac resuscitation. N Engl J Med 2001;344 (17) 1304- 1313
PubMed Link to Article
Kaye  WMancini  METruitt  TL When minutes count: the fallacy of accurate time documentation during in-hospital resuscitation. Resuscitation 2005;65 (3) 285- 290
PubMed Link to Article

Correspondence

CME
Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
Submit a Comment

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

See Also...
Articles Related By Topic
Related Collections
PubMed Articles
JAMAevidence.com

Users' Guides to the Medical Literature
Clinical Resolution

Users' Guides to the Medical Literature
Clinical Scenario