THERAPEUTIC HYPOTHERMIA AND TEMPERATURE MANAGEMENT Volume 5, Number 1, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/ther.2014.0023
National Trends in the Use of Postcardiac Arrest Therapeutic Hypothermia and Hospital Factors Influencing Its Use Scott M. Dresden, MD, MS,1,2 Lanty M. O’Connor, MS,3 Charles G. Pearce, MD,1 D. Mark Courtney, MD, MSCI,1 and Emilie S. Powell, MD, MS, MBA1,2
Background: Therapeutic hypothermia (TH) in cardiac arrest continues to be underused in the United States. A better understanding of its utilization could inform future efforts and policies to improve utilization. This study investigates trends in TH use for in and out-of-hospital cardiac arrest, and hospital factors associated with its use. Methods: We performed a cross-sectional analysis using the Nationwide Inpatient Sample (NIS), 2007–2010, of US adult hospitalizations with cardiac arrest. Annual rates of TH use and proportions of hospitals using TH were calculated using NIS weighting. Potential hospital factors associated with increased likelihood of TH utilization were assessed using logistic regression. Results: Across 2007–2010, 1.35% of cardiac arrest patients received TH; increasing from 0.34% (2007) to 2.49% (2010). The proportion of hospitals using TH was 13.63%, increasing from 4.63% (2007) to 22.16% (2010). Significant hospital factors associated with TH utilization were: large hospitals, urban location, Northeast or West regions, teaching hospitals, non-safety net hospitals, increasing year, and hospitals with higher annual cardiac arrest volume. Conclusion: Utilization of TH in cardiac arrest remains low, but increased sevenfold from 2007 to 2010. The significant variability in implementation of TH, argues for nationwide best practices or regionalization of postcardiac arrest care hospitals.
Cardiopulmonary Resuscitation and Emergency Cardiovascular Care’’ to include the induction of postresuscitation hypothermia in all comatose survivors of out-of-hospital cardiac arrest with ventricular fibrillation or ventricular tachycardia, and to consider it for survivors of in-hospital cardiac arrest with additional cardiac rhythms (Nolan et al., 2003; American Heart Association, 2005a, 2005b). The guidelines are supported by the Cochrane Review (Arrich et al., 2012); however, the quality of evidence has been questioned by other researchers, calling for further study of the recommended treatment strategy (Nielsen et al., 2011). Debate has centered on the defining of the inclusion criteria and optimal temperature for TH after cardiac arrest, but not if TH should be considered at all (Holzer, 2010; Dumas et al., 2011; Nichol et al., 2013; Nielsen et al., 2013). Despite the AHA recommendations, TH is used infrequently (Abella et al., 2005; Laver et al., 2006; Merchant et al., 2006; Jena et al., 2012; Mikkelsen et al., 2013). In
Introduction
A
nnually 300,000 cardiac arrests occur in the United States with a mortality of greater than 90% (Go et al., 2013). Patients who survive cardiac arrest often have significant morbidity and lifelong neurologic impairment. In patients who have return of spontaneous circulation (ROSC) after cardiac arrest, induction of postresuscitation therapeutic hypothermia (TH) has repeatedly been shown to be neuroprotective and reduce morbidity for survivors (Bernard et al., 2002; The Hypothermia after Cardiac Arrest Study Group, 2002; Delhaye et al., 2012; Kim et al., 2013). Further studies have demonstrated TH to be cost-effective and easy to implement with minimal additional resources (Rittenberger et al., 2008; Gaieski et al., 2009; Merchant et al., 2009). In 2003 a formal recommendation for the use of TH after cardiac arrest was issued, and in 2005, the American Heart Association (AHA) published the updated ‘‘Guidelines for
1 Department of Emergency Medicine, 2Center for Healthcare Studies, and 3Center for Education in Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois.
48
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
multiple studies, the majority of physicians surveyed have never or rarely used TH in the management of cardiac arrest (Abella et al., 2005; Haque et al., 2006; Laver et al., 2006; Merchant et al., 2006; Sugerman and Abella, 2009). Physicians report not using TH because of perceived lack of evidence for its efficacy, technical and logistical difficulties in initiating therapy, and a lack of financial or personnel resources to initiate TH (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). Use of TH for in-hospital cardiac arrest has been increasing slowly, from 0.7% in 2003 to 3.3% in 2009 (Mikkelsen et al., 2013). Changes in the use of TH for out-of-hospital cardiac arrest since the publication of the 2005 AHA guidelines are unknown. Additionally, the relationships between hospital factors and the use of TH after cardiac arrest are unclear. In particular, it is unknown if there is a relationship between hospital cardiac arrest volume and TH use. While a recent study demonstrated that there is not a clear relationship between hospital cardiac arrest volume and survival after out-of-hospital cardiac arrest (Cudnik et al., 2012), we hypothesize that patients who have ROSC might receive higher quality of care (i.e., TH use) at higher volume centers. Improved outcomes have been demonstrated in many other time-sensitive high-acuity scenarios, such as sepsis and surgery, in higher volume centers, likely secondary to increased clinical experience (Luft et al., 1987; Birkmeyer et al., 2002; Elixhauser et al., 2003; Powell et al., 2010). The goal of this study is to use population-based national hospital discharge data to examine trends in the use of TH in both in- and out-of-hospital cardiac arrest patients after implementation of the 2005 AHA guidelines, and evaluate hospital factors associated with its use, specifically hospital cardiac arrest case volume. We hypothesized that while use of TH in cardiac arrest would be low, use would increase over time, and hospitals with relatively higher cardiac arrest case volumes would be more likely to use TH after cardiac arrest. Materials and Methods
49
further identified all hospitalizations with a diagnosis or procedure code consistent with the use of TH (ICD-9 780.65, procedure code 99.81) (Centers for Disease Control and Prevention, 2008; Reaven and Rosenbloom, 2009). Data synthesis
We aggregated hospitalization discharge data by hospital ID and year so that hospitals included in the NIS for multiple years were analyzed separately each year. The NIS includes several hospital-level descriptive variables: hospital bed size (small, medium, large), teaching status, location (rural, urban), hospital region (Northeast, Midwest, South, West), and hospital ownership (public, not-for-profit, investor owned, private, not specified). We created a binary variable, safety net hospital, defined by a modified version of the CDC definition of a safety net hospital: any hospital with over 40% of hospitalizations by uninsured or Medicaid patients (Burt and Arispe, 2004). We created an annual hospital cardiac arrest case volume variable, defined as the total number of hospitalizations with any discharge diagnosis consistent with cardiac arrest in a given year; the annual hospital cardiac arrest volume was calculated separately for each year. Hospitals were empirically divided into four volume-level quartiles for ease in interpretation. Hospitals were classified as ‘‘users of TH’’ or ‘‘non-users of TH’’ if there was any documentation of TH in that hospital in a given year or not. Outcome measures
The primary outcome was any use of TH by a hospital for in- or out-of-hospital cardiac arrest patients in a given year (‘‘user of TH’’). This primary outcome supports the three analyses: (1) overall use of TH, (2) the use of TH over time, and (3) evaluation of hospital factors associated with TH use, specifically annual cardiac arrest volume. The unit of analysis was the hospital in a given year. Additionally, annual cardiac arrest volume and annual TH use were defined with the patient and hospitalization as the unit of analyses.
Study design
We performed a cross-sectional analysis of the Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS) for years 2007–2010 to allow dissemination and uptake of the 2005 AHA guidelines before analysis. The NIS is the largest publicly available all-payer inpatient dataset in the US and is provided by the US Agency for Healthcare Research and Quality (AHRQ). The NIS includes data from *1000 hospitals and 8 million inpatient hospital stays per year, representing approximately a 20% stratified sample of US community hospitals (www.hcup-us.ahrq.gov/). Data are weighted to result in a sample that is representative of all admissions to nonfederal US hospitals on an annual basis. The Institutional Review Board found this study exempt from review. Sample selection
We included all US adult hospitalizations over the study period with any discharge diagnosis of cardiac arrest as defined by the International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM), code for cardiac arrest (427.5) (Carr et al., 2009). Within this sample, we
Statistical analysis
Analyses were conducted with the STATA version 12.0 (STATA Corp., College Station, TX). National population estimates for annual cardiac arrest volume and TH utilization were made using the NIS discharge-level weighting. The proportion of hospitals using TH nationally across the study period was calculated using NIS hospital-level weighting. Differences between years (2007–2010) were evaluated with the Wald test for linear hypothesis. We evaluated the unadjusted association of hospital factors with the hospital use of TH using the chi-squared test. Hospital factors included bed size, ownership, urban versus rural location, region, teaching status, safety net status, annual cardiac arrest volume (by volume quartile), and year of inclusion in the NIS (2007–2010). We used a multivariate logistic regression model to evaluate the simultaneous effects of hospital factors found to be significant in the unadjusted analyses, exclusive of hospital ownership (data were missing from 39% of the sample). Results are reported as odds ratios (ORs) and 95% confidence intervals (95% CI). Significant associations were interpreted as p < 0.05. To approximate patients with ROSC and TH eligibility after cardiac arrest, we
50
DRESDEN ET AL.
conducted an exploratory analysis, including only patients who survived to discharge and, therefore, must have had ROSC and time for TH implementation. This serves only as an approximation as there are likely other patients who had ROSC after in-hospital cardiac arrest, but then died during the hospitalization, and some patients with ROSC who survived to hospital discharge, but were not eligible for TH, but there is no way to discern this in the NIS. We again evaluated the use of TH and the simultaneous effects of the same hospital factors on TH use. Results The use of TH
Between 2007 and 2010, there were 110,386 hospitalizations with cardiac arrest included in the NIS. Using NIS discharge-level weighting, this represents an estimated 548,604 hospitalizations with cardiac arrest in the US. National estimates of hospitalizations with cardiac arrest remained stable throughout the study period: 130,935 in 2007 (95% CI 117,685–144,183), 138,651 in 2008 (95% CI 125,056–152,246), 137,941 in 2009 (95% CI 124,913– 150,968), and 141,077 in 2010 (95% CI 127,249–154,906) ( p = 0.74). Between 2007 and 2010, use of TH in cardiac arrest was documented in 1490 hospitalizations, representing an estimated 7421 hospitalizations nationwide. The estimated number of hospitalizations with cardiac arrest and documented TH use in the NIS increased over the study period: 443 in 2007 (95% CI 214–673), 1024 in 2008 (95% CI 587–1463), 2465 in 2009 (95% CI 1903–3027), and 3487 in 2010 (2793–4182) ( p < 0.001). The proportion of hospitalizations with cardiac arrest that had documented TH use was 1.35% (95% CI 1.17–1.56%) over 2007–2010. This proportion increased annually: 0.34% (95% CI 0.21–0.56%) in 2007, 0.74% (95% CI 0.49–1.11%) in 2008, 1.79% (95% CI 1.47–2.18%) in 2009, and 2.49% (95% CI 2.10–2.90%) in 2010 ( p < 0.001) (Fig. 1). When limiting the sample to only patients who survived to discharge, utilization of TH remains low: 0.35% (95% CI 0.21–0.58%) in 2007, 0.81% (95% CI 0.51–1.31%) in 2008, 1.44% (95% CI 0.11–1.89%) in 2009, and 2.03% (95% CI 1.66–2.48%) in 2010, with 1.21% overall (95% CI 1.01–1.44%). Hospital use of TH: unadjusted analysis
Using NIS hospital level weighting, the proportion of hospitals with at least one cardiac arrest who used TH at least once in a given year (TH users) increased annually from 2007 to 2010: 4.64% in 2007 (199 out of 4293), to 8.99% in 2008 (377/4192), to 18.22% in 2009 (759/4166), to 23.16% in 2010 (983/4244) ( p < 0.001) (Fig. 1). All hospital factors considered were associated with cardiac arrest volume quartile in the unadjusted analysis. Hospitals with large bed size, private ownership, urban location, located in the west region, teaching hospitals, and non-safety net hospitals were more likely to be in the top quartile of cardiac arrest volume (Table 1). Hospital characteristics significantly associated with TH use included large hospital size, private hospitals, urban hospitals, hospitals in the Northeast and West, teaching hospitals, non-safety net hospitals, increasing year of inclusion in the NIS, and higher annual cardiac arrest case volume (Table 2). When limiting the sample to only discharged pa-
FIG. 1. National estimates of percent of patients receiving therapeutic hypothermia (TH) after cardiac arrest and percent of hospitals providing TH in cardiac arrest. tients, the significant unadjusted associations remain the same (data not shown). Hospital characteristics and use of TH: multivariable analysis
Large hospitals, urban hospitals, hospitals in the West, Northeast, or Midwest, teaching hospitals, and non-safety net hospitals were all independently more likely to use TH in cardiac arrest (Table 3). Increasing year also had a significant independent association with TH use: in reference to 2007, the OR was 2.3 (95% CI 1.5–3.5, p < 0.05) in 2008, 6.9 (95% CI 4.6–10.3, p < 0.001) in 2009, and 10.8 (95% CI 7.3–16.0, p < 0.001) in 2010. Increasing cardiac arrest volume quartiles had the strongest independent association with TH use with an OR of 88.6 when compared with the lowest volume quartile (95% CI 26.8–292.3, p < 0.001). When limiting the sample to only patients who survived to discharge (this analysis included 262 hospitals as users of TH versus 476 hospitals in the larger analysis), the independent association between increasing year and increasing cardiac arrest volume quartile and TH use remained statistically significant. In addition, hospitals in the Northeast or West and teaching hospitals were significantly more likely to use TH in cardiac arrest, but the association did not remain for Midwestern, non-safety net, large, or urban hospitals (data not shown). Discussion
This is the first study to evaluate the use of TH in both inand out-of-hospital cardiac arrest patients nationwide over time since the introduction of the AHA Guidelines supporting its use in 2005. Encouragingly, we demonstrate a sevenfold
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
51
Table 1. Baseline Characteristics Annual cardiac arrest cases, n (%) Hospital rank (quartile) Hospital characteristics Bed size (n = 3434)a Small Medium Large Hospital ownership (n = 2124)a Public Private Hospital location (n = 3434)a Rural Urban Hospital region (n = 3463) Northeast Midwest South West Teaching status (n = 3463) Non-teaching Teaching Safety net status (n = 3463) Non-safety net hospital Safety net hospital
Low
2nd
3rd
High
686 (55) 200 (21) 103 (8)
347 (28) 216 (23) 182 (15)
182 (14) 330 (35) 339 (27)
44 (3) 186 (20) 619 (50)
263 (50) 493 (31)
140 (27) 415 (26)
80 (15) 407 (25)
43 (8) 283 (18)
601 (50) 388 (17)
359 (30) 386 (17)
191 (16) 660 (30)
48 (4) 801 (36)
91 356 397 151
100 168 362 124
164 193 330 170
128 192 333 204
p < 0.001
< 0.001 < 0.001 < 0.001 (19) (39) (28) (23)
(21) (18) (25) (19)
(34) (21) (23) (26)
(27) (21) (23) (31) < 0.001
944 (35) 45 (6)
706 (26) 39 (6)
662 (24) 189 (27)
422 (15) 427 (61)
862 (27) 133 (58)
706 (22) 48 (21)
839 (26) 18 (8)
828 (26) 29 (13)
< 0.001
Hospital characteristics across cardiac arrest volume quartiles. a Data on bed size, hospital ownership, and hospital location not available for all hospitals.
increase in use nationally and an increase in the proportion of hospitals with any TH use, across the study years. But, the overall use remains extremely low with TH use documented in only 2.49% of hospitalizations with either inpatient or outof-hospital cardiac arrest in 2010. Even when limiting the sample to only patients who must have had ROSC as they survived to hospital discharge, utilization remains low at 2.03% in 2010. This low rate of utilization may have lasting effects on functional outcome, disability, and use of health services for the 98% of patients who survive cardiac arrest, but are not offered TH. Many hospital characteristics were associated with TH use: primarily high cardiac arrest case volume. Hospital characteristics associated with lack of use include safety net status, rural location, nonteaching status, and hospital region (Midwestern and Southern regions). Our results are consistent with the established literature in the US. The low, but rising rates of TH use confirm a previous study that examined in-hospital cardiac arrest only (2% utilization over 2003–2009), another study of a single year (0.32% utilization in 2007), and a regional study (0.44% of in- and out-of-hospital cardiac arrest in California from 1999 to 2008, with 87.3% of TH use coming after 2005) (Patel et al., 2011; Jena et al., 2012; Mikkelsen et al., 2013). However, our results are markedly different from a recent observational study of out-of-hospital cardiac arrest patients treated in 22 Finnish ICUs (61.7% of all patients with ROSC and 85.8% of all patients with ROSC after ventricular tachycardia or ventricular fibrillation received TH) (Vaahersalo et al., 2013). Additionally, we demonstrate a striking difference from a recent US study examining in-hospital cardiac arrest in the American Heart Association’s Get with the Guidelines–Resuscitation (GWTG-R) database, which
reported that 52% of hospitals used TH at least once (Mikkelsen et al., 2013), versus 13.58% in our study. This discrepancy may be a function of our use of a representative sample of all US hospitals operating outside of the influence of a study or registry and that relies on billing and coding data, rather than a voluntary registry such as the GWTG-R that relies on self-report. We sought to evaluate hospital characteristics associated with TH use to better discern the primary contributors to overall low utilization and identify where improvement efforts could be directed. In the adjusted analysis, hospitals with the greatest odds of using TH at least once in a given year were those with the highest annual cardiac arrest case volume (OR 88.6, 95% CI 26.8–292.3). This is consistent with the extensive literature base on the volume-outcome relationship in other disciplines, particularly elective surgery, where higher case volume correlates with superior quality of care (Luft et al., 1987; Birkmeyer et al., 2002; Cross et al., 2003; Durairaj et al., 2005; Schull et al., 2006; Powell et al., 2010). Centers with high volume of cardiac arrest patients may be better able to establish TH protocols and have the appropriate resources and personnel in place, addressing perceived difficulties to initiating TH at other institutions (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). We demonstrate several other hospital characteristics associated with a lack of reported use of TH: safety net status, rural location, nonteaching status, and hospital region. As safety net hospitals may be more sensitive to cost issues, we hypothesize that the disparity in TH use by safety net status is likely secondary to the perceived financial barriers to providing TH (Merchant et al., 2006). However, this perceived
52
DRESDEN ET AL.
Table 2. Unadjusted Analyses Hospital characteristic Bed size Small Medium High Ownership Public Private Hospital location Rural Urban Hospital region South Midwest Northeast West Teaching status Nonteaching Teaching Safety net status Non-safety net hospital Safety net hospital Year 2007 2008 2009 2010 Cardiac arrest volume (quartile) Low 2nd 3rd High
Hospitals using TH, n (%)
Table 3. Multivariable Logistic Regression Hospital characteristic p < 0.001
51 (4.1) 117 (12.6) 302 (24.3) < 0.001 30 (5.7) 148 (9.2) < 0.001 36 (3.0) 434 (19.4) < 0.001 142 107 91 136
(10.0) (11.8) (18.9) (21.0) < 0.001
238 (8.7) 232 (33.1) < 0.001 463 (14.3) 13 (5.7) < 0.001 41 78 156 201
(4.7) (9.1) (18.3) (23.2) < 0.001
3 26 115 332
(0.3) (3.5) (13.4) (38.7)
Unadjusted univariate association of hospital use of TH and hospital characteristics, including year and annual cardiac arrest volume. TH, therapeutic hypothermia.
barrier is not supported, while there are expensive TH options available, TH can also be done for very little expense and has been shown to be cost effective with a ratio of $47,168 per quality-adjusted life-year (Merchant et al., 2009). The lower odds of TH use by hospitals in rural locations and in nonteaching hospitals have been demonstrated in two prior studies as well (Carr et al., 2009; Patel et al., 2011). These findings could relate to literature supporting that the use of advanced technologies and decreased mortality are positively correlated with teaching intensity (Shahian et al., 2012). While this literature cannot be directly applied to TH utilization, it is suggestive of a root cause of lack of implementation due again to a perception that TH requires advanced technology to be effective. Finally and perhaps most unexpectedly, a significant hospital regional variation also exists. The most consistent nonusers were hospitals located in the Midwestern and Southern regions. This could be attributable to the lack of teaching hospitals in these regions. However, when controlling for teaching status, this regional difference still remains. The density of teaching institutions, particularly in the Northeast, may render a halo effect on the
Bed size Small Medium Large Hospital location Rural Urban Hospital region South Midwest Northeast West Teaching status Nonteaching Teaching Safety net status Safety net hospital Non-safety net hospital Year 2007 2008 2009 2010 Number of cardiac arrests Low 2nd 3rd High
Odds ratio
p
Ref 1.2 (0.8–1.8) 1.9 (1.3–2.8)
0.333 0.001
Ref 2.0 (1.3–2.9)
0.001
Ref 1.4 (1.0–1.9) 1.7 (1.2–2.4) 2.4 (1.7–3.3)
0.049 0.004 < 0.001
Ref 2.1 (1.6–2.7)
< 0.001
Ref 2.8 (1.4–5.6)
0.005
Ref 2.3 (1.5–3.5) 6.9 (4.6–10.3) 10.8 (7.3–16.0) 2007–2010 (quartile) Ref 9.5 (2.8–31.7) 29.0 (9.0–93.9) 88.6 (26.8–292.3)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
Multivariable logistic regression (adjusted) analyses of the likelihood of TH use at least once in a given year (i.e., hospital use of TH).
region as a whole, an effect that may be less apparent in more rural regions. The results of this study demonstrate that the use of TH has been low despite the 2005 and 2010 AHA Guidelines (American Heart Association, 2005a, 2005b; Peberdy et al., 2010). The reasons for the overall low implementation of TH are beyond the scope of this study. However, in previous studies, providers have cited perceived lack of evidence for its efficacy, technical and logistical difficulties in initiating therapy, and a lack of financial or personnel resources to initiate TH (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). The debate in the literature about the optimal inclusion criteria and temperature for TH may also play a role in providers’ hesitancy to use TH (Holzer, 2010; Dumas et al., 2011; Nielsen et al., 2011, 2013; Nichol et al., 2013). However, the low percentage of hospitals that use TH at all suggests that the barriers to TH implementation may be at the hospital level, rather than at the patient or provider level. Future studies of the operations and practices of high volume cardiac arrest centers may inform the implementation of TH across all centers. Nationwide educational interventions disseminating best practices and supporting the benefits of TH could be directed toward hospitals in the Midwestern and Southern regions of the country. Alternatively, given the influence of location and hospital type, the solution may lie in
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
regionalization of care and the development of postcardiac arrest care centers. Providers at dedicated TH centers may be more familiar with the literature on TH, and the technical and logistical process to perform TH. The low use of TH demonstrated in this study also calls for a close examination of how TH use is documented. It is possible that TH is underreported in billing and coding data (as used in this study with the NIS) and actual use might be higher and efforts should be made to optimize consistent coding and reporting across institutions for more accurate research study, quality reporting by hospitals, outcome analysis, and optimal national reporting. Despite the use of a well-established and widely used dataset with nationwide scope, our study has several limitations. The most significant and difficult to circumvent limitation is the reliance on appropriate coding of diagnoses and procedures. It is possible that patients could have received TH without appropriate documentation and, therefore, are categorized as nonusers. Additionally, the NIS is a representative sampling of US hospitals, including different hospitals each year, making it impossible to trend the same hospitals over time. With an intervention whose use is so infrequent, a single institution with a high rate of utilization might confound the patient level data from 1 year to the next. Finally, we were limited by the data provided by the NIS. The NIS does not contain granular clinical data for patient-level risk adjustment or to determine ROSC and strict eligibility for TH. However, in limiting the analysis to only those hospitalizations that survived to hospital the results were similar to the original larger analysis. Conclusion
This US nationwide study of TH utilization demonstrates that while use is increasing, the overall utilization of TH relative to all cardiac arrests remains very low. The etiology is likely multifactorial: low overall rates of ROSC in cardiac arrest, low rates of TH use in cardiac arrest with ROSC, and under-reporting of TH use. Despite a sevenfold increase in the use of TH since 2007, there continues to be significant geographic and hospital characteristics associated with disparities and variability in implementation. Acknowledgment
Research for this article was done while Scott Dresden was a National Research Service Award postdoctoral fellow at the Center for Healthcare Studies under an institutional award from the Agency for Healthcare Research and Quality, T-32 HS 000078 (PI: Jane L. Holl, MD MPH). Author Disclosure Statement
The authors have nothing to disclose. References
Abella BS, Rhee JW, Huang KN, et al. Induced hypothermia is underused after resuscitation from cardiac arrest: a current practice survey. Resuscitation 2005;64:181–186. American Heart Association. 2005 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2005a;112(24 Suppl): IV1-203.
53
American Heart Association. 2005 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science with Treatment Recommendations. Part 4: Advanced Life Support. Resuscitation 2005b;67:213–247. Arrich J, Holzer M, Havel C, et al. Hypothermia for neuroprotection in adults after cardiopulmonary resuscitation. The Cochrane database of systematic reviews 2012;9:CD004128. Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med 2002;346:557–563. Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128–1137. Burt CW, Arispe IE. Characteristics of emergency departments serving high volumes of safety-net patients: United States, 2000. Vital Health Stat 13 2004;155:1–16. Carr BG, Goyal M, Band RA, et al. A national analysis of the relationship between hospital factors and post-cardiac arrest mortality. Intensive Care Med 2009;35:505–511. Centers for Disease Control and Prevention. 2008, Icd-9-Cm Coordination and Maintenance Committee Meeting Diagnosis Agenda. www.cdc.gov/nchs/data/icd9/agendaMa08.pdf Cross DT, 3rd, Tirschwell DL, Clark MA, et al. Mortality rates after subarachnoid hemorrhage: variations according to hospital case volume in 18 states. J Neurosurg 2003;99:810–817. Cudnik MT, Sasson C, Rea TD, et al. Increasing hospital volume is not associated with improved survival in out of hospital cardiac arrest of cardiac etiology. Resuscitation 2012; 83:862–868. Delhaye C, Mahmoudi M, Waksman R. Hypothermia therapy: neurological and cardiac benefits. J Am Coll Cardiol 2012; 59:197–210. Dumas F, Grimaldi D, Zuber B, et al. Is hypothermia after cardiac arrest effective in both shockable and nonshockable patients?: insights from a large registry. Circulation 2011; 123:877–886. Durairaj L, Torner JC, Chrischilles EA, et al. Hospital volumeoutcome relationships among medical admissions to ICUs. Chest 2005;128:1682–1689. Elixhauser A, Steiner C, Fraser I. Volume thresholds and hospital characteristics in the United States. Health Aff (Millwood) 2003;22:167–177. Gaieski DF, Fuchs B, Carr BG, et al. Practical implementation of therapeutic hypothermia after cardiac arrest. Hosp Pract (1995) 2009;37:71–83. Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation 2013;127:e6–e245. Haque IU, Latour MC, Zaritsky AL. Pediatric critical care community survey of knowledge and attitudes toward therapeutic hypothermia in comatose children after cardiac arrest. Pediatr Crit Care Med 2006;7:7–14. Holzer M. Targeted temperature management for comatose survivors of cardiac arrest. N Engl J Med 2010;363:1256–1264. Jena AB, Romley JA, Newton-Cheh C, et al. Therapeutic hypothermia for cardiac arrest: real-world utilization trends and hospital mortality. J Hosp Med 2012;7:684–689. Kim JY, Shin SD, Ro YS, et al. Post-resuscitation care and outcomes of out-of-hospital cardiac arrest: a nationwide propensity score-matching analysis. Resuscitation 2013;84: 1068–1077. Laver SR, Padkin A, Atalla A, et al. Therapeutic hypothermia after cardiac arrest: a survey of practice in intensive care units in the United Kingdom. Anaesthesia 2006;61:873–877.
54
Luft HS, Hunt SS, Maerki SC. The volume-outcome relationship: practice-makes-perfect or selective-referral patterns? Health Serv Res 1987;22:26. Merchant RM, Becker LB, Abella BS, et al. Cost-effectiveness of therapeutic hypothermia after cardiac arrest. Circulation 2009;2:421–428. Merchant RM, Soar J, Skrifvars MB, et al. Therapeutic hypothermia utilization among physicians after resuscitation from cardiac arrest. Crit Care Med 2006;34:1935–1940. Mikkelsen ME, Christie JD, Abella BS, et al. Use of therapeutic hypothermia after in-hospital cardiac arrest. Crit Care Med 2013;41:1385–1395. Nichol G, Huszti E, Kim F, et al. Does induction of hypothermia improve outcomes after in-hospital cardiac arrest? Resuscitation 2013;84:620–625. Nielsen N, Friberg H, Gluud C, et al. Hypothermia after cardiac arrest should be further evaluated—a systematic review of randomised trials with meta-analysis and trial sequential analysis. Int J Cardiol 2011;151:333–341. Nielsen N, Wetterslev J, Cronberg T, et al. Targeted temperature management at 33 degrees C versus 36 degrees C after cardiac arrest. N Engl J Med 2013;369:2197–2206. Nolan JP, Morley PT, Vanden Hoek TL, et al. Therapeutic hypothermia after cardiac arrest: an advisory statement by the advanced life support task force of the International Liaison Committee on Resuscitation. Circulation 2003;108:118–121. Patel PV, John S, Garg RK, et al. Therapeutic hypothermia after cardiac arrest is underutilized in the United States. Ther Hypothermia Temp Manag 2011;1:199–203. Peberdy MA, Callaway CW, Neumar RW, et al. Part 9: Postcardiac arrest care: 2010 American Heart Association Guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation 2010;122(18 Suppl 3):S768–S786. Powell ES, Khare RK, Courtney DM, et al. Volume of emergency department admissions for sepsis is related to inpatient
DRESDEN ET AL.
mortality: results of a nationwide cross-sectional analysis. Crit Care Med 2010;38:2161–2168. Reaven NL, Rosenbloom J. Commentary on the reimbursement paradox. Crit Care Med 2009;37(7 Suppl):S285–S289. Rittenberger JC, Guyette FX, Tisherman SA, et al. Outcomes of a hospital-wide plan to improve care of comatose survivors of cardiac arrest. Resuscitation 2008;79:198–204. Schull MJ, Vermeulen MJ, Stukel TA. The risk of missed diagnosis of acute myocardial infarction associated with emergency department volume. Ann Emerg Med 2006;48: 647–655. Shahian DM, Nordberg P, Meyer GS, et al. Contemporary performance of U.S. teaching and nonteaching hospitals. Acad Med 2012;87:701–708. Sugerman NT, Abella BS. Hospital-based use of therapeutic hypothermia after cardiac arrest in adults. J Neurotrauma 2009;26:371–376. The Hypothermia after Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002;346:549–556. Vaahersalo J, Hiltunen P, Tiainen M, et al. Therapeutic hypothermia after out-of-hospital cardiac arrest in Finnish intensive care units: the Finnresusci Study. Intensive Care Med 2013;39:826–837.
Address correspondence to: Scott M. Dresden, MD, MS Department of Emergency Medicine Center for Healthcare Studies Feinberg School of Medicine Northwestern University 211 E. Ontario St. Suite 200 Chicago, IL 60611 E-mail: [email protected]
National Trends in the Use of Postcardiac Arrest Therapeutic Hypothermia and Hospital Factors Influencing Its Use Scott M. Dresden, MD, MS,1,2 Lanty M. O’Connor, MS,3 Charles G. Pearce, MD,1 D. Mark Courtney, MD, MSCI,1 and Emilie S. Powell, MD, MS, MBA1,2
Background: Therapeutic hypothermia (TH) in cardiac arrest continues to be underused in the United States. A better understanding of its utilization could inform future efforts and policies to improve utilization. This study investigates trends in TH use for in and out-of-hospital cardiac arrest, and hospital factors associated with its use. Methods: We performed a cross-sectional analysis using the Nationwide Inpatient Sample (NIS), 2007–2010, of US adult hospitalizations with cardiac arrest. Annual rates of TH use and proportions of hospitals using TH were calculated using NIS weighting. Potential hospital factors associated with increased likelihood of TH utilization were assessed using logistic regression. Results: Across 2007–2010, 1.35% of cardiac arrest patients received TH; increasing from 0.34% (2007) to 2.49% (2010). The proportion of hospitals using TH was 13.63%, increasing from 4.63% (2007) to 22.16% (2010). Significant hospital factors associated with TH utilization were: large hospitals, urban location, Northeast or West regions, teaching hospitals, non-safety net hospitals, increasing year, and hospitals with higher annual cardiac arrest volume. Conclusion: Utilization of TH in cardiac arrest remains low, but increased sevenfold from 2007 to 2010. The significant variability in implementation of TH, argues for nationwide best practices or regionalization of postcardiac arrest care hospitals.
Cardiopulmonary Resuscitation and Emergency Cardiovascular Care’’ to include the induction of postresuscitation hypothermia in all comatose survivors of out-of-hospital cardiac arrest with ventricular fibrillation or ventricular tachycardia, and to consider it for survivors of in-hospital cardiac arrest with additional cardiac rhythms (Nolan et al., 2003; American Heart Association, 2005a, 2005b). The guidelines are supported by the Cochrane Review (Arrich et al., 2012); however, the quality of evidence has been questioned by other researchers, calling for further study of the recommended treatment strategy (Nielsen et al., 2011). Debate has centered on the defining of the inclusion criteria and optimal temperature for TH after cardiac arrest, but not if TH should be considered at all (Holzer, 2010; Dumas et al., 2011; Nichol et al., 2013; Nielsen et al., 2013). Despite the AHA recommendations, TH is used infrequently (Abella et al., 2005; Laver et al., 2006; Merchant et al., 2006; Jena et al., 2012; Mikkelsen et al., 2013). In
Introduction
A
nnually 300,000 cardiac arrests occur in the United States with a mortality of greater than 90% (Go et al., 2013). Patients who survive cardiac arrest often have significant morbidity and lifelong neurologic impairment. In patients who have return of spontaneous circulation (ROSC) after cardiac arrest, induction of postresuscitation therapeutic hypothermia (TH) has repeatedly been shown to be neuroprotective and reduce morbidity for survivors (Bernard et al., 2002; The Hypothermia after Cardiac Arrest Study Group, 2002; Delhaye et al., 2012; Kim et al., 2013). Further studies have demonstrated TH to be cost-effective and easy to implement with minimal additional resources (Rittenberger et al., 2008; Gaieski et al., 2009; Merchant et al., 2009). In 2003 a formal recommendation for the use of TH after cardiac arrest was issued, and in 2005, the American Heart Association (AHA) published the updated ‘‘Guidelines for
1 Department of Emergency Medicine, 2Center for Healthcare Studies, and 3Center for Education in Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois.
48
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
multiple studies, the majority of physicians surveyed have never or rarely used TH in the management of cardiac arrest (Abella et al., 2005; Haque et al., 2006; Laver et al., 2006; Merchant et al., 2006; Sugerman and Abella, 2009). Physicians report not using TH because of perceived lack of evidence for its efficacy, technical and logistical difficulties in initiating therapy, and a lack of financial or personnel resources to initiate TH (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). Use of TH for in-hospital cardiac arrest has been increasing slowly, from 0.7% in 2003 to 3.3% in 2009 (Mikkelsen et al., 2013). Changes in the use of TH for out-of-hospital cardiac arrest since the publication of the 2005 AHA guidelines are unknown. Additionally, the relationships between hospital factors and the use of TH after cardiac arrest are unclear. In particular, it is unknown if there is a relationship between hospital cardiac arrest volume and TH use. While a recent study demonstrated that there is not a clear relationship between hospital cardiac arrest volume and survival after out-of-hospital cardiac arrest (Cudnik et al., 2012), we hypothesize that patients who have ROSC might receive higher quality of care (i.e., TH use) at higher volume centers. Improved outcomes have been demonstrated in many other time-sensitive high-acuity scenarios, such as sepsis and surgery, in higher volume centers, likely secondary to increased clinical experience (Luft et al., 1987; Birkmeyer et al., 2002; Elixhauser et al., 2003; Powell et al., 2010). The goal of this study is to use population-based national hospital discharge data to examine trends in the use of TH in both in- and out-of-hospital cardiac arrest patients after implementation of the 2005 AHA guidelines, and evaluate hospital factors associated with its use, specifically hospital cardiac arrest case volume. We hypothesized that while use of TH in cardiac arrest would be low, use would increase over time, and hospitals with relatively higher cardiac arrest case volumes would be more likely to use TH after cardiac arrest. Materials and Methods
49
further identified all hospitalizations with a diagnosis or procedure code consistent with the use of TH (ICD-9 780.65, procedure code 99.81) (Centers for Disease Control and Prevention, 2008; Reaven and Rosenbloom, 2009). Data synthesis
We aggregated hospitalization discharge data by hospital ID and year so that hospitals included in the NIS for multiple years were analyzed separately each year. The NIS includes several hospital-level descriptive variables: hospital bed size (small, medium, large), teaching status, location (rural, urban), hospital region (Northeast, Midwest, South, West), and hospital ownership (public, not-for-profit, investor owned, private, not specified). We created a binary variable, safety net hospital, defined by a modified version of the CDC definition of a safety net hospital: any hospital with over 40% of hospitalizations by uninsured or Medicaid patients (Burt and Arispe, 2004). We created an annual hospital cardiac arrest case volume variable, defined as the total number of hospitalizations with any discharge diagnosis consistent with cardiac arrest in a given year; the annual hospital cardiac arrest volume was calculated separately for each year. Hospitals were empirically divided into four volume-level quartiles for ease in interpretation. Hospitals were classified as ‘‘users of TH’’ or ‘‘non-users of TH’’ if there was any documentation of TH in that hospital in a given year or not. Outcome measures
The primary outcome was any use of TH by a hospital for in- or out-of-hospital cardiac arrest patients in a given year (‘‘user of TH’’). This primary outcome supports the three analyses: (1) overall use of TH, (2) the use of TH over time, and (3) evaluation of hospital factors associated with TH use, specifically annual cardiac arrest volume. The unit of analysis was the hospital in a given year. Additionally, annual cardiac arrest volume and annual TH use were defined with the patient and hospitalization as the unit of analyses.
Study design
We performed a cross-sectional analysis of the Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS) for years 2007–2010 to allow dissemination and uptake of the 2005 AHA guidelines before analysis. The NIS is the largest publicly available all-payer inpatient dataset in the US and is provided by the US Agency for Healthcare Research and Quality (AHRQ). The NIS includes data from *1000 hospitals and 8 million inpatient hospital stays per year, representing approximately a 20% stratified sample of US community hospitals (www.hcup-us.ahrq.gov/). Data are weighted to result in a sample that is representative of all admissions to nonfederal US hospitals on an annual basis. The Institutional Review Board found this study exempt from review. Sample selection
We included all US adult hospitalizations over the study period with any discharge diagnosis of cardiac arrest as defined by the International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM), code for cardiac arrest (427.5) (Carr et al., 2009). Within this sample, we
Statistical analysis
Analyses were conducted with the STATA version 12.0 (STATA Corp., College Station, TX). National population estimates for annual cardiac arrest volume and TH utilization were made using the NIS discharge-level weighting. The proportion of hospitals using TH nationally across the study period was calculated using NIS hospital-level weighting. Differences between years (2007–2010) were evaluated with the Wald test for linear hypothesis. We evaluated the unadjusted association of hospital factors with the hospital use of TH using the chi-squared test. Hospital factors included bed size, ownership, urban versus rural location, region, teaching status, safety net status, annual cardiac arrest volume (by volume quartile), and year of inclusion in the NIS (2007–2010). We used a multivariate logistic regression model to evaluate the simultaneous effects of hospital factors found to be significant in the unadjusted analyses, exclusive of hospital ownership (data were missing from 39% of the sample). Results are reported as odds ratios (ORs) and 95% confidence intervals (95% CI). Significant associations were interpreted as p < 0.05. To approximate patients with ROSC and TH eligibility after cardiac arrest, we
50
DRESDEN ET AL.
conducted an exploratory analysis, including only patients who survived to discharge and, therefore, must have had ROSC and time for TH implementation. This serves only as an approximation as there are likely other patients who had ROSC after in-hospital cardiac arrest, but then died during the hospitalization, and some patients with ROSC who survived to hospital discharge, but were not eligible for TH, but there is no way to discern this in the NIS. We again evaluated the use of TH and the simultaneous effects of the same hospital factors on TH use. Results The use of TH
Between 2007 and 2010, there were 110,386 hospitalizations with cardiac arrest included in the NIS. Using NIS discharge-level weighting, this represents an estimated 548,604 hospitalizations with cardiac arrest in the US. National estimates of hospitalizations with cardiac arrest remained stable throughout the study period: 130,935 in 2007 (95% CI 117,685–144,183), 138,651 in 2008 (95% CI 125,056–152,246), 137,941 in 2009 (95% CI 124,913– 150,968), and 141,077 in 2010 (95% CI 127,249–154,906) ( p = 0.74). Between 2007 and 2010, use of TH in cardiac arrest was documented in 1490 hospitalizations, representing an estimated 7421 hospitalizations nationwide. The estimated number of hospitalizations with cardiac arrest and documented TH use in the NIS increased over the study period: 443 in 2007 (95% CI 214–673), 1024 in 2008 (95% CI 587–1463), 2465 in 2009 (95% CI 1903–3027), and 3487 in 2010 (2793–4182) ( p < 0.001). The proportion of hospitalizations with cardiac arrest that had documented TH use was 1.35% (95% CI 1.17–1.56%) over 2007–2010. This proportion increased annually: 0.34% (95% CI 0.21–0.56%) in 2007, 0.74% (95% CI 0.49–1.11%) in 2008, 1.79% (95% CI 1.47–2.18%) in 2009, and 2.49% (95% CI 2.10–2.90%) in 2010 ( p < 0.001) (Fig. 1). When limiting the sample to only patients who survived to discharge, utilization of TH remains low: 0.35% (95% CI 0.21–0.58%) in 2007, 0.81% (95% CI 0.51–1.31%) in 2008, 1.44% (95% CI 0.11–1.89%) in 2009, and 2.03% (95% CI 1.66–2.48%) in 2010, with 1.21% overall (95% CI 1.01–1.44%). Hospital use of TH: unadjusted analysis
Using NIS hospital level weighting, the proportion of hospitals with at least one cardiac arrest who used TH at least once in a given year (TH users) increased annually from 2007 to 2010: 4.64% in 2007 (199 out of 4293), to 8.99% in 2008 (377/4192), to 18.22% in 2009 (759/4166), to 23.16% in 2010 (983/4244) ( p < 0.001) (Fig. 1). All hospital factors considered were associated with cardiac arrest volume quartile in the unadjusted analysis. Hospitals with large bed size, private ownership, urban location, located in the west region, teaching hospitals, and non-safety net hospitals were more likely to be in the top quartile of cardiac arrest volume (Table 1). Hospital characteristics significantly associated with TH use included large hospital size, private hospitals, urban hospitals, hospitals in the Northeast and West, teaching hospitals, non-safety net hospitals, increasing year of inclusion in the NIS, and higher annual cardiac arrest case volume (Table 2). When limiting the sample to only discharged pa-
FIG. 1. National estimates of percent of patients receiving therapeutic hypothermia (TH) after cardiac arrest and percent of hospitals providing TH in cardiac arrest. tients, the significant unadjusted associations remain the same (data not shown). Hospital characteristics and use of TH: multivariable analysis
Large hospitals, urban hospitals, hospitals in the West, Northeast, or Midwest, teaching hospitals, and non-safety net hospitals were all independently more likely to use TH in cardiac arrest (Table 3). Increasing year also had a significant independent association with TH use: in reference to 2007, the OR was 2.3 (95% CI 1.5–3.5, p < 0.05) in 2008, 6.9 (95% CI 4.6–10.3, p < 0.001) in 2009, and 10.8 (95% CI 7.3–16.0, p < 0.001) in 2010. Increasing cardiac arrest volume quartiles had the strongest independent association with TH use with an OR of 88.6 when compared with the lowest volume quartile (95% CI 26.8–292.3, p < 0.001). When limiting the sample to only patients who survived to discharge (this analysis included 262 hospitals as users of TH versus 476 hospitals in the larger analysis), the independent association between increasing year and increasing cardiac arrest volume quartile and TH use remained statistically significant. In addition, hospitals in the Northeast or West and teaching hospitals were significantly more likely to use TH in cardiac arrest, but the association did not remain for Midwestern, non-safety net, large, or urban hospitals (data not shown). Discussion
This is the first study to evaluate the use of TH in both inand out-of-hospital cardiac arrest patients nationwide over time since the introduction of the AHA Guidelines supporting its use in 2005. Encouragingly, we demonstrate a sevenfold
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
51
Table 1. Baseline Characteristics Annual cardiac arrest cases, n (%) Hospital rank (quartile) Hospital characteristics Bed size (n = 3434)a Small Medium Large Hospital ownership (n = 2124)a Public Private Hospital location (n = 3434)a Rural Urban Hospital region (n = 3463) Northeast Midwest South West Teaching status (n = 3463) Non-teaching Teaching Safety net status (n = 3463) Non-safety net hospital Safety net hospital
Low
2nd
3rd
High
686 (55) 200 (21) 103 (8)
347 (28) 216 (23) 182 (15)
182 (14) 330 (35) 339 (27)
44 (3) 186 (20) 619 (50)
263 (50) 493 (31)
140 (27) 415 (26)
80 (15) 407 (25)
43 (8) 283 (18)
601 (50) 388 (17)
359 (30) 386 (17)
191 (16) 660 (30)
48 (4) 801 (36)
91 356 397 151
100 168 362 124
164 193 330 170
128 192 333 204
p < 0.001
< 0.001 < 0.001 < 0.001 (19) (39) (28) (23)
(21) (18) (25) (19)
(34) (21) (23) (26)
(27) (21) (23) (31) < 0.001
944 (35) 45 (6)
706 (26) 39 (6)
662 (24) 189 (27)
422 (15) 427 (61)
862 (27) 133 (58)
706 (22) 48 (21)
839 (26) 18 (8)
828 (26) 29 (13)
< 0.001
Hospital characteristics across cardiac arrest volume quartiles. a Data on bed size, hospital ownership, and hospital location not available for all hospitals.
increase in use nationally and an increase in the proportion of hospitals with any TH use, across the study years. But, the overall use remains extremely low with TH use documented in only 2.49% of hospitalizations with either inpatient or outof-hospital cardiac arrest in 2010. Even when limiting the sample to only patients who must have had ROSC as they survived to hospital discharge, utilization remains low at 2.03% in 2010. This low rate of utilization may have lasting effects on functional outcome, disability, and use of health services for the 98% of patients who survive cardiac arrest, but are not offered TH. Many hospital characteristics were associated with TH use: primarily high cardiac arrest case volume. Hospital characteristics associated with lack of use include safety net status, rural location, nonteaching status, and hospital region (Midwestern and Southern regions). Our results are consistent with the established literature in the US. The low, but rising rates of TH use confirm a previous study that examined in-hospital cardiac arrest only (2% utilization over 2003–2009), another study of a single year (0.32% utilization in 2007), and a regional study (0.44% of in- and out-of-hospital cardiac arrest in California from 1999 to 2008, with 87.3% of TH use coming after 2005) (Patel et al., 2011; Jena et al., 2012; Mikkelsen et al., 2013). However, our results are markedly different from a recent observational study of out-of-hospital cardiac arrest patients treated in 22 Finnish ICUs (61.7% of all patients with ROSC and 85.8% of all patients with ROSC after ventricular tachycardia or ventricular fibrillation received TH) (Vaahersalo et al., 2013). Additionally, we demonstrate a striking difference from a recent US study examining in-hospital cardiac arrest in the American Heart Association’s Get with the Guidelines–Resuscitation (GWTG-R) database, which
reported that 52% of hospitals used TH at least once (Mikkelsen et al., 2013), versus 13.58% in our study. This discrepancy may be a function of our use of a representative sample of all US hospitals operating outside of the influence of a study or registry and that relies on billing and coding data, rather than a voluntary registry such as the GWTG-R that relies on self-report. We sought to evaluate hospital characteristics associated with TH use to better discern the primary contributors to overall low utilization and identify where improvement efforts could be directed. In the adjusted analysis, hospitals with the greatest odds of using TH at least once in a given year were those with the highest annual cardiac arrest case volume (OR 88.6, 95% CI 26.8–292.3). This is consistent with the extensive literature base on the volume-outcome relationship in other disciplines, particularly elective surgery, where higher case volume correlates with superior quality of care (Luft et al., 1987; Birkmeyer et al., 2002; Cross et al., 2003; Durairaj et al., 2005; Schull et al., 2006; Powell et al., 2010). Centers with high volume of cardiac arrest patients may be better able to establish TH protocols and have the appropriate resources and personnel in place, addressing perceived difficulties to initiating TH at other institutions (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). We demonstrate several other hospital characteristics associated with a lack of reported use of TH: safety net status, rural location, nonteaching status, and hospital region. As safety net hospitals may be more sensitive to cost issues, we hypothesize that the disparity in TH use by safety net status is likely secondary to the perceived financial barriers to providing TH (Merchant et al., 2006). However, this perceived
52
DRESDEN ET AL.
Table 2. Unadjusted Analyses Hospital characteristic Bed size Small Medium High Ownership Public Private Hospital location Rural Urban Hospital region South Midwest Northeast West Teaching status Nonteaching Teaching Safety net status Non-safety net hospital Safety net hospital Year 2007 2008 2009 2010 Cardiac arrest volume (quartile) Low 2nd 3rd High
Hospitals using TH, n (%)
Table 3. Multivariable Logistic Regression Hospital characteristic p < 0.001
51 (4.1) 117 (12.6) 302 (24.3) < 0.001 30 (5.7) 148 (9.2) < 0.001 36 (3.0) 434 (19.4) < 0.001 142 107 91 136
(10.0) (11.8) (18.9) (21.0) < 0.001
238 (8.7) 232 (33.1) < 0.001 463 (14.3) 13 (5.7) < 0.001 41 78 156 201
(4.7) (9.1) (18.3) (23.2) < 0.001
3 26 115 332
(0.3) (3.5) (13.4) (38.7)
Unadjusted univariate association of hospital use of TH and hospital characteristics, including year and annual cardiac arrest volume. TH, therapeutic hypothermia.
barrier is not supported, while there are expensive TH options available, TH can also be done for very little expense and has been shown to be cost effective with a ratio of $47,168 per quality-adjusted life-year (Merchant et al., 2009). The lower odds of TH use by hospitals in rural locations and in nonteaching hospitals have been demonstrated in two prior studies as well (Carr et al., 2009; Patel et al., 2011). These findings could relate to literature supporting that the use of advanced technologies and decreased mortality are positively correlated with teaching intensity (Shahian et al., 2012). While this literature cannot be directly applied to TH utilization, it is suggestive of a root cause of lack of implementation due again to a perception that TH requires advanced technology to be effective. Finally and perhaps most unexpectedly, a significant hospital regional variation also exists. The most consistent nonusers were hospitals located in the Midwestern and Southern regions. This could be attributable to the lack of teaching hospitals in these regions. However, when controlling for teaching status, this regional difference still remains. The density of teaching institutions, particularly in the Northeast, may render a halo effect on the
Bed size Small Medium Large Hospital location Rural Urban Hospital region South Midwest Northeast West Teaching status Nonteaching Teaching Safety net status Safety net hospital Non-safety net hospital Year 2007 2008 2009 2010 Number of cardiac arrests Low 2nd 3rd High
Odds ratio
p
Ref 1.2 (0.8–1.8) 1.9 (1.3–2.8)
0.333 0.001
Ref 2.0 (1.3–2.9)
0.001
Ref 1.4 (1.0–1.9) 1.7 (1.2–2.4) 2.4 (1.7–3.3)
0.049 0.004 < 0.001
Ref 2.1 (1.6–2.7)
< 0.001
Ref 2.8 (1.4–5.6)
0.005
Ref 2.3 (1.5–3.5) 6.9 (4.6–10.3) 10.8 (7.3–16.0) 2007–2010 (quartile) Ref 9.5 (2.8–31.7) 29.0 (9.0–93.9) 88.6 (26.8–292.3)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
Multivariable logistic regression (adjusted) analyses of the likelihood of TH use at least once in a given year (i.e., hospital use of TH).
region as a whole, an effect that may be less apparent in more rural regions. The results of this study demonstrate that the use of TH has been low despite the 2005 and 2010 AHA Guidelines (American Heart Association, 2005a, 2005b; Peberdy et al., 2010). The reasons for the overall low implementation of TH are beyond the scope of this study. However, in previous studies, providers have cited perceived lack of evidence for its efficacy, technical and logistical difficulties in initiating therapy, and a lack of financial or personnel resources to initiate TH (Abella et al., 2005; Haque et al., 2006; Merchant et al., 2006). The debate in the literature about the optimal inclusion criteria and temperature for TH may also play a role in providers’ hesitancy to use TH (Holzer, 2010; Dumas et al., 2011; Nielsen et al., 2011, 2013; Nichol et al., 2013). However, the low percentage of hospitals that use TH at all suggests that the barriers to TH implementation may be at the hospital level, rather than at the patient or provider level. Future studies of the operations and practices of high volume cardiac arrest centers may inform the implementation of TH across all centers. Nationwide educational interventions disseminating best practices and supporting the benefits of TH could be directed toward hospitals in the Midwestern and Southern regions of the country. Alternatively, given the influence of location and hospital type, the solution may lie in
NATIONAL TRENDS IN THERAPEUTIC HYPOTHERMIA
regionalization of care and the development of postcardiac arrest care centers. Providers at dedicated TH centers may be more familiar with the literature on TH, and the technical and logistical process to perform TH. The low use of TH demonstrated in this study also calls for a close examination of how TH use is documented. It is possible that TH is underreported in billing and coding data (as used in this study with the NIS) and actual use might be higher and efforts should be made to optimize consistent coding and reporting across institutions for more accurate research study, quality reporting by hospitals, outcome analysis, and optimal national reporting. Despite the use of a well-established and widely used dataset with nationwide scope, our study has several limitations. The most significant and difficult to circumvent limitation is the reliance on appropriate coding of diagnoses and procedures. It is possible that patients could have received TH without appropriate documentation and, therefore, are categorized as nonusers. Additionally, the NIS is a representative sampling of US hospitals, including different hospitals each year, making it impossible to trend the same hospitals over time. With an intervention whose use is so infrequent, a single institution with a high rate of utilization might confound the patient level data from 1 year to the next. Finally, we were limited by the data provided by the NIS. The NIS does not contain granular clinical data for patient-level risk adjustment or to determine ROSC and strict eligibility for TH. However, in limiting the analysis to only those hospitalizations that survived to hospital the results were similar to the original larger analysis. Conclusion
This US nationwide study of TH utilization demonstrates that while use is increasing, the overall utilization of TH relative to all cardiac arrests remains very low. The etiology is likely multifactorial: low overall rates of ROSC in cardiac arrest, low rates of TH use in cardiac arrest with ROSC, and under-reporting of TH use. Despite a sevenfold increase in the use of TH since 2007, there continues to be significant geographic and hospital characteristics associated with disparities and variability in implementation. Acknowledgment
Research for this article was done while Scott Dresden was a National Research Service Award postdoctoral fellow at the Center for Healthcare Studies under an institutional award from the Agency for Healthcare Research and Quality, T-32 HS 000078 (PI: Jane L. Holl, MD MPH). Author Disclosure Statement
The authors have nothing to disclose. References
Abella BS, Rhee JW, Huang KN, et al. Induced hypothermia is underused after resuscitation from cardiac arrest: a current practice survey. Resuscitation 2005;64:181–186. American Heart Association. 2005 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2005a;112(24 Suppl): IV1-203.
53
American Heart Association. 2005 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science with Treatment Recommendations. Part 4: Advanced Life Support. Resuscitation 2005b;67:213–247. Arrich J, Holzer M, Havel C, et al. Hypothermia for neuroprotection in adults after cardiopulmonary resuscitation. The Cochrane database of systematic reviews 2012;9:CD004128. Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med 2002;346:557–563. Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128–1137. Burt CW, Arispe IE. Characteristics of emergency departments serving high volumes of safety-net patients: United States, 2000. Vital Health Stat 13 2004;155:1–16. Carr BG, Goyal M, Band RA, et al. A national analysis of the relationship between hospital factors and post-cardiac arrest mortality. Intensive Care Med 2009;35:505–511. Centers for Disease Control and Prevention. 2008, Icd-9-Cm Coordination and Maintenance Committee Meeting Diagnosis Agenda. www.cdc.gov/nchs/data/icd9/agendaMa08.pdf Cross DT, 3rd, Tirschwell DL, Clark MA, et al. Mortality rates after subarachnoid hemorrhage: variations according to hospital case volume in 18 states. J Neurosurg 2003;99:810–817. Cudnik MT, Sasson C, Rea TD, et al. Increasing hospital volume is not associated with improved survival in out of hospital cardiac arrest of cardiac etiology. Resuscitation 2012; 83:862–868. Delhaye C, Mahmoudi M, Waksman R. Hypothermia therapy: neurological and cardiac benefits. J Am Coll Cardiol 2012; 59:197–210. Dumas F, Grimaldi D, Zuber B, et al. Is hypothermia after cardiac arrest effective in both shockable and nonshockable patients?: insights from a large registry. Circulation 2011; 123:877–886. Durairaj L, Torner JC, Chrischilles EA, et al. Hospital volumeoutcome relationships among medical admissions to ICUs. Chest 2005;128:1682–1689. Elixhauser A, Steiner C, Fraser I. Volume thresholds and hospital characteristics in the United States. Health Aff (Millwood) 2003;22:167–177. Gaieski DF, Fuchs B, Carr BG, et al. Practical implementation of therapeutic hypothermia after cardiac arrest. Hosp Pract (1995) 2009;37:71–83. Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation 2013;127:e6–e245. Haque IU, Latour MC, Zaritsky AL. Pediatric critical care community survey of knowledge and attitudes toward therapeutic hypothermia in comatose children after cardiac arrest. Pediatr Crit Care Med 2006;7:7–14. Holzer M. Targeted temperature management for comatose survivors of cardiac arrest. N Engl J Med 2010;363:1256–1264. Jena AB, Romley JA, Newton-Cheh C, et al. Therapeutic hypothermia for cardiac arrest: real-world utilization trends and hospital mortality. J Hosp Med 2012;7:684–689. Kim JY, Shin SD, Ro YS, et al. Post-resuscitation care and outcomes of out-of-hospital cardiac arrest: a nationwide propensity score-matching analysis. Resuscitation 2013;84: 1068–1077. Laver SR, Padkin A, Atalla A, et al. Therapeutic hypothermia after cardiac arrest: a survey of practice in intensive care units in the United Kingdom. Anaesthesia 2006;61:873–877.
54
Luft HS, Hunt SS, Maerki SC. The volume-outcome relationship: practice-makes-perfect or selective-referral patterns? Health Serv Res 1987;22:26. Merchant RM, Becker LB, Abella BS, et al. Cost-effectiveness of therapeutic hypothermia after cardiac arrest. Circulation 2009;2:421–428. Merchant RM, Soar J, Skrifvars MB, et al. Therapeutic hypothermia utilization among physicians after resuscitation from cardiac arrest. Crit Care Med 2006;34:1935–1940. Mikkelsen ME, Christie JD, Abella BS, et al. Use of therapeutic hypothermia after in-hospital cardiac arrest. Crit Care Med 2013;41:1385–1395. Nichol G, Huszti E, Kim F, et al. Does induction of hypothermia improve outcomes after in-hospital cardiac arrest? Resuscitation 2013;84:620–625. Nielsen N, Friberg H, Gluud C, et al. Hypothermia after cardiac arrest should be further evaluated—a systematic review of randomised trials with meta-analysis and trial sequential analysis. Int J Cardiol 2011;151:333–341. Nielsen N, Wetterslev J, Cronberg T, et al. Targeted temperature management at 33 degrees C versus 36 degrees C after cardiac arrest. N Engl J Med 2013;369:2197–2206. Nolan JP, Morley PT, Vanden Hoek TL, et al. Therapeutic hypothermia after cardiac arrest: an advisory statement by the advanced life support task force of the International Liaison Committee on Resuscitation. Circulation 2003;108:118–121. Patel PV, John S, Garg RK, et al. Therapeutic hypothermia after cardiac arrest is underutilized in the United States. Ther Hypothermia Temp Manag 2011;1:199–203. Peberdy MA, Callaway CW, Neumar RW, et al. Part 9: Postcardiac arrest care: 2010 American Heart Association Guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation 2010;122(18 Suppl 3):S768–S786. Powell ES, Khare RK, Courtney DM, et al. Volume of emergency department admissions for sepsis is related to inpatient
DRESDEN ET AL.
mortality: results of a nationwide cross-sectional analysis. Crit Care Med 2010;38:2161–2168. Reaven NL, Rosenbloom J. Commentary on the reimbursement paradox. Crit Care Med 2009;37(7 Suppl):S285–S289. Rittenberger JC, Guyette FX, Tisherman SA, et al. Outcomes of a hospital-wide plan to improve care of comatose survivors of cardiac arrest. Resuscitation 2008;79:198–204. Schull MJ, Vermeulen MJ, Stukel TA. The risk of missed diagnosis of acute myocardial infarction associated with emergency department volume. Ann Emerg Med 2006;48: 647–655. Shahian DM, Nordberg P, Meyer GS, et al. Contemporary performance of U.S. teaching and nonteaching hospitals. Acad Med 2012;87:701–708. Sugerman NT, Abella BS. Hospital-based use of therapeutic hypothermia after cardiac arrest in adults. J Neurotrauma 2009;26:371–376. The Hypothermia after Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002;346:549–556. Vaahersalo J, Hiltunen P, Tiainen M, et al. Therapeutic hypothermia after out-of-hospital cardiac arrest in Finnish intensive care units: the Finnresusci Study. Intensive Care Med 2013;39:826–837.
Address correspondence to: Scott M. Dresden, MD, MS Department of Emergency Medicine Center for Healthcare Studies Feinberg School of Medicine Northwestern University 211 E. Ontario St. Suite 200 Chicago, IL 60611 E-mail: [email protected]
