Myocardial Infarction

Transcription

1 Myocardial Infarction Nationwide Analysis of Patients With ST-Segment Elevation Myocardial Infarction Transferred for Primary Percutaneous Intervention Findings From the American Heart Association Mission: Lifeline Program Harold L. Dauerman, MD; Eric R. Bates, MD; Michael C. Kontos, MD; Shuang Li, MS; J. Lee Garvey, MD; Timothy D. Henry, MD; Steven V. Manoukian, MD; Matthew T. Roe, MD, MHS Background Current American College of Cardiology/American Heart Association guidelines recommend transfer and primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) patients within the time limit of first contact to device 120 minutes. We determined the hospital-level, patient-level, and process characteristics of timely versus delayed primary PCI for a diverse national sample of transfer patients confined to a travel distance that facilitates the process. Methods and Results We studied patients transferred from non PCI-capable hospitals for primary PCI to 398 National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines hospitals between July 2008 and December Patients with estimated transfer times >60 minutes (by Google Maps driving times) were excluded from the analysis. Patients achieving first door-to-device time 120 minutes were compared with patients with delayed treatment; independent predictors of timely treatment were determined using generalized estimating equations logistic regression models. The median estimated transfer distance was 26.5 miles. First door-to-device 120 minutes was achieved in 65% of patients (n=9380); only 37% of the hospitals were high-performing hospitals (defined as risk-adjusted rate, 75% of transfer STEMI patients with 120-minute first door-to-device time). In addition to known predictors of delay (cardiogenic shock, cardiac arrest, and prolonged door-in door-out time), STEMI referral hospitals rural location and longer estimated transfer time were identified as predictors of delay. In this diverse national sample, regional and racial variations in care were observed. Finally, lower PCI hospital annual STEMI volume was a potent predictor of delay. Conclusions More than one third of US STEMI patients transferred for primary PCI fail to achieve first door-todevice time 120 minutes, despite estimated transfer times <60 minutes. Delays are related to process variables, comorbidities, and lower annual PCI hospital STEMI volumes. (Circ Cardiovasc Interv. 2015;8:e DOI: /CIRCINTERVENTIONS ) Key Words: regional systems of care ST-segment elevation myocardial infarction transfer Multiple strategies have been used to improve the safety and efficacy of regional primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) programs. 1 3 Strategies that may improve first medical contact to device times include prehospital cardiac catheterization laboratory activation and bypass of non PCI-capable hospitals or emergency departments. 4,5 Although rerouting patients directly to PCI centers may eliminate potential delays associated with transfer of STEMI patients for primary PCI, 5 50% of patients with STEMI do not arrive to an emergency department via ambulance. 6 Thus, regional systems of care face the ongoing challenge of determining best practices for timely transfer of patients from STEMI referring centers to STEMI PCI hospitals. Although significant progress has been made in achieving short ischemic time intervals for patients presenting directly to PCI centers, transfer STEMI time intervals remain a challenge. 7,8 Both European and US STEMI guidelines recommend first medical contact to device time of 120 minutes for the transfer STEMI population 9,10 : STEMI referring facilities unable to achieve this 2-hour interval should consider fibrinolytic therapy as the primary reperfusion strategy. 11 Timely first door-to-device intervals have been achieved for transfer patients in selected regional programs. 12,13 However, recent Received November 14, 2014; accepted March 2, From the University of Vermont Cardiovascular Research Institute, Burlington (H.L.D.); University of Michigan, Ann Arbor (E.R.B.); Virginia Commonwealth University, Richmond (M.C.K.); Duke Clinical Research Institute, Durham, NC (S.L., M.T.R.); Carolinas Medical Center, Charlotte, NC (J.L.G.); Cedars-Sinai Heart Institute, Los Angeles, CA (T.D.H.); and Hospital Corporation of America, Nashville, TN (S.V.M.). Correspondence to Harold L. Dauerman, MD, Division of Cardiology, McClure 1, University of Vermont Medical Center, 111 Colchester Ave, Burlington, VT harold.dauerman@uvmhealth.org 2015 American Heart Association, Inc. Circ Cardiovasc Interv is available at DOI: /CIRCINTERVENTIONS

2 2 Dauerman et al Transfer for Acute Myocardial Infarction WHAT IS KNOWN Current American College of Cardiology/American Heart Association guidelines recommend transfer and primary percutaneous coronary intervention for ST-segment elevation myocardial infarction patients within the time limit of first contact to device 120 minutes. At least one third of transfer ST-segment elevation myocardial infarction patients do not achieve timely reperfusion. WHAT THE STUDY ADDS This is the first study to confine analysis of reperfusion times to transfer patients who are within 60 minutes of a percutaneous coronary intervention center. Still, more than one third of US ST-segment elevation myocardial infarction patients transferred for primary percutaneous coronary intervention fail to achieve first door-to-device time 120 minutes, despite estimated transfer times <60 minutes. Delays are related to process variables, comorbidities, and lower annual percutaneous coronary intervention hospital ST-segment elevation myocardial infarction volumes. national estimates of first door (at STEMI referral centers) to device times (at the PCI center) demonstrate considerable opportunity for improvement: less than two thirds of transfer STEMI patients achieve first door-to-device times 120 minutes 7,8 although these previous analyses have had unrestricted estimated transfer distances. Previous publications have focused on prolonged door-in door-out (DIDO) times as both an opportunity for process improvement and a predictor of mortality among the STEMI transfer population Other studies have emphasized the impact of transfer time and distance in determining timely primary PCI. 18 These analyses were not confined specifically to patients within a potentially realistic estimated ground transfer time ( 60 minutes) and thus may underestimate the potential for timely transfer of appropriate patients. To further understand the characteristics and processes of care associated with achieving first door-to-device times of 120 minutes, we studied all US patients undergoing transfer for primary PCI within a 60-minute estimated drive time between the non PCI-capable hospital and the PCI hospital in the National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines. Methods Study Population Using the National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines database, we identified a sample of consecutive transfer STEMI patients at 398 sites over the period July 1, 2008, to December 31, Patients were excluded if they did not have data forms representing details for transfer STEMI patients, if transfer time was >60 minutes (as estimated by Google Maps using address of both STEMI referral center and PCI hospital), if they presented initially to a PCI center, or if they were determined to have NSTEMI as their primary diagnosis. Patients were also excluded if they had fibrinolysis, did not receive primary PCI, or had a nonsystem reason for delay Nonsystem reasons for delay included 2567 patients with non shock/non arrest system-related delays including difficulty in obtaining consent, vascular access, or crossing the culprit lesion. Approximately 20% of the sample population (n=3776) were beyond the 60-minute estimated transfer time limit or had inadequate information to demonstrate this transfer time window. The resulting analysis population included patients who were clearly within 60-minute transfer time from STEMI referral center to PCI hospital. The initial model analysis excluded patients presenting with cardiogenic shock or cardiac arrest (n=931) for a sample of The rationale for the 60-minute transfer time limit is shown in Figure 1 and was based on 30 (door-in door out), 60 (transfer time), and 30 (second door-to-device) minute divisions of time comprising the 120-minute total time limit for timely transfer. Definitions and End Points An estimated transfer time 60 minutes was calculated from addresses of STEMI referring center and PCI center using Google Maps. DIDO time was calculated as time of initial presentation to the STEMI referral center to transfer out time from the STEMI referral center. Off hours was defined as admission between 6 pm and 8 am (during week days) and 6 pm Friday to 8 am Monday, as well as on national holidays. Patients were characterized as rural versus urban location based on patient s residential zip code. Hospital rural location was based on hospital s zip code. The primary analysis of this study was the comparison of characteristics for patients achieving first door-to-device 120 minutes compared with patients not achieving this time interval and identify independent risk factors associated with first door-to-device 120 minutes. A secondary goal of this study was the comparison of characteristics of high-performing PCI centers when compared with non high-performing PCI centers: highperforming PCI centers were identified based on achieving initial door-to-device time of <120 minutes in risk adjustment rate 75% of transfer STEMI patients similar to a previous metric set for nontransfer primary PCI performance 11,19 : PCI receiving hospitals with fewer than 25 patients were excluded from the comparison of highperforming versus low-performing hospitals: thus, 253 PCI primary PCI centers were excluded to achieve data stability. This registry was either approved by an institutional review board or considered quality assurance data and not subject to institutional review board approval based on individual site determinations. The Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines serves as the hospital data collection and evaluation mechanism for the American Heart Association s Mission: Lifeline STEMI program. The Duke Clinical Research Institute serves as the data analysis center and has an agreement to analyze the aggregate deidentified data for research purposes. Statistical Analysis Continuous variables are presented as median and interquartile range (IQR), and categorical variables were summarized as percentage. Characteristics of patients between 2 groups were compared using Mantel Haenszel χ 2 test for categorical variables and Wilcoxon ranksum tests for continuous variables. All potential risk factors were selected from univariate analysis of baseline characteristics. These potential risk factors consisted of demographics (age, sex, and race), medical history (hypertension, dialysis, chronic lung disease, diabetes mellitus, previous MI, previous revascularization, previous congestive heart failure, previous atrial fibrillation, previous stroke, and previous peripheral arterial disease), signs and symptoms at presentation (ECG findings with STEMI versus left bundle branch block versus posterior MI), hospital characteristic (region of PCI hospital, teaching versus nonteaching PCI hospital, PCI center with no surgical backup versus coronary artery

3 3 Dauerman et al Transfer for Acute Myocardial Infarction bypass grafting capability hospitals, PCI hospital annual STEMI volume, PCI hospital bed size, referral non-pci hospital bed size, and rural status), patient insurance status, estimated transfer time by Google Maps, time of hospital presentation, transport method to first facility (self versus ambulance or mobile intensive care unit or air), transport method from first referral facility to PCI hospital (ambulance, air, versus mobile intensive care unit), or use of prehospital ECG. The dependent outcome variable was a binary indicator of first door-to-balloon time 120 minutes (ie, 1=yes, 0=no). To identify independent factors associated with first door-to-device time of <120 minutes, multivariable generalized estimating equations logistic regression models with backward model selection at 0.05 significance level of stay were used. The generalized estimating equation method was implemented with a compound symmetrical working correlation matrix and empirical (sandwich) SE estimates and is adjusted for clustering of observations from the same hospital. 20 The linearity of all continuous risk factors with respect to outcomes was determined: the linear term or linear spline terms with knots allowing different slopes were used for these continuous factors. 21,22 To assess cardiac arrest and cardiogenic shock, which were known to be associated with delays, the model described above (main model) was modified to include patients with cardiogenic shock and cardiac arrest. Similarly, to assess the influence of DIDO time on the associated outcome, DIDO was added to the main model: we only included patients with DIDO between 0 and 60 minutes in this model to remove outliers because outliers were unlikely to achieve the optimal first door-to-device times. Furthermore, to compare characteristics of high-performing PCI centers with non high-performing PCI centers, risk-adjusted rates of first door-to-device time of <120 minutes were calculated using the following formula for each hospital: Risk-adjusted rate of specific hospital = Observed no in specific hospital of events Overall observed rate Expected no of events of all patients in specific hospital frommain model Analyses were conducting by National Cardiovascular Data Registry data analysis center at the Duke Clinical Research Institute (Durham, NC) using SAS version 9.3 software (SAS Institute, Cary, NC). Results First Door-to-Device Times First door-to-device time 120 minutes was achieved in 65% of primary PCI patients within an estimated 60-minute transfer distance between first referral hospital and PCI hospital. Figure 1. Study population and rationale using the National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines database. The analysis focused on primary percutaneous coronary intervention patients with transfer times 60 minutes. Transfer ST-segment elevation myocardial infarction (STEMI) for primary PCI: n= (excludes transfer patients treated initially with thrombolytics, no PCI performed, any nonsystem reason for delay or missing information). Major exclusion: 3776 patients without Google map estimated transfer time 60 min. Total study population: n= transfer STEMI PCI patients. Patients with cardiac arrest or cardiogenic shock (n=931) on first medical contact excluded on initial analysis. Primary study population: n= Study period: July 1, 2008 to December, 31, 2012; number of sites: 398. DIDO indicates door-in door-out. Of note, first medical contact times were not distinct from first door times in over three fourths of the patient population as 75% of patients presented to the non PCI-capable hospital via self-transport mechanisms (Figure 2). Only 11.3% of patients received a prehospital ECG. As per study design, estimated transfer times (median, 33 minutes; IQR, minutes) and estimated transfer distances (median, 26.5 miles; IQR, miles) were not excessive and 70% of interhospital transport was achieved via ground. DIDO times were long and varied widely (median, 46.0 minutes; IQR, minutes), even when excluding patients with cardiogenic shock and cardiac arrest. Most patients presented off hours (64.5%) and were affiliated with urban residential zip codes (62.0%). Characteristics of Patients and Hospitals With Delay Patients with first door-to-device times >120 minutes were older, more likely to be women and nonwhite race, and had an increased prevalence of comorbid factors including previous MI, previous revascularization, peripheral arterial disease, chronic lung disease, diabetes mellitus, and previous stroke (Table 1). Despite excluding patients with cardiogenic shock and cardiac arrest from the primary analysis, patients with delay were still sicker: heart failure on presentation was twice as likely (6.27% versus 2.81%) and resting heart rate was higher (median, 80 bpm; IQR, bpm versus median 77 bpm; IQR, bpm; P<0.001). On univariate analysis, there was no significant difference in off-hour presentation, but patients transferred from rural hospitals were significantly more likely to have delay in first door-to-device times. We observed statistically significant regional differences in achievement of a first door-to-device time 120 minutes (48% versus 69% comparing the Northeast with the Midwest; P<0.001) although the number of patients in each region varied from 923 in the Northeast to 7831 in the South because of large differences in regional hospital participation. When hospitals were stratified according to high-performance status (as defined by risk-adjusted rate, 75% of transfer patients with door-to-device times 120 minutes), only 37% of US hospitals reached this performance standard. Among patients in the high-performing hospitals, median first doorto-device time was 93 minutes (IQR, minutes). DIDO times were shorter in patients in the high-performing hospitals

4 4 Dauerman et al Transfer for Acute Myocardial Infarction Table 1. Time Characteristics Stratified by First Door-to-Device Figure 2. Three fourths of patients presented to the non percutaneous coronary intervention (PCI) hospital via self-transport mechanisms; transport from the non PCI-capable hospital to the PCI center was primarily by ground. First medical contact available in only 25% of patients so first door-to-device metric was used in all patients. (median, 38 minutes; IQR, minutes) when compared with patients in the lower performing group (median, 54 minutes; IQR, minutes; Table 2). Patients in the high-performing hospitals were twice as likely to receive prehospital electrocardiograms. In the analysis, the characteristics of the high-performing PCI-capable hospital were not clearly predictive of high-performing status with the exception of academic affiliation (33% versus 19%; P=0.05). Independent Predictors of Delayed First Door-to- Device Times In the first multivariable analysis, we excluded 3 known predictors of delay DIDO time, cardiac arrest, and cardiogenic shock (Table 3). Patients with longer transfer times (per 5-minute increase and confined to patients with minute transfer times) were 20% less likely to achieve guideline recommended first door-to-device times. Similarly, older patients (per 5-year increase for those >55 years old), women, and patients with diabetes mellitus or off hours presentation were less likely to achieve first door-to-device times 120 minutes. Two of the most potent predictors of longer time intervals to treatment were left bundle branch block on the qualifying ECG and previous coronary artery bypass grafting. Although lack of prehospital ECG activation was not independently predictive of delay, self-transport to the first hospital conferred an 18% lesser chance of achieving optimal timing when compared with ambulance transport. Referral hospital bed size was associated with optimal door-to-device times; the 2 most potent positive predictors of achieving timely first door-todevice times were white race and higher PCI hospital annual STEMI volume ( 18% increased likelihood per 50 STEMI patients/y annual increment in volume). Finally, a significant regional difference in achieving timely first door-to-device times was still observed after controlling for the other predictors of delay. C statistic index of the main model was with a 95% confidence interval of to We repeated the multivariable analysis to determine the impact of cardiogenic shock or cardiac arrest on factors associated with Variable First Door-to- Device Time >120 min (n=5138) First Door-to- Device Time 120 min (n=9380) P Value First door-to-device time, 149 ( ) 93 (81 106) <0.001 min, median (interquartile range) Door-in door-out time, min, 76 (59 103) 37 (28 48) <0.001 median (interquartile range) Estimated transfer distance, 32 (19 42) 23 (11 34) <0.001 miles, median (interquartile range) Estimated transfer time, min, 37 (24 47) 29 (18 41) <0.001 median (interquartile range) Off-hours presentation, % Rural info of patients <0.001 residential zip code, % Age 75 y, % <0.001 Female sex, % <0.001 Diabetes mellitus, % <0.001 Previous MI, % <0.001 Previous stroke, % 5 3 <0.001 Peripheral arterial disease 5 4 <0.001 Previous revascularization, % <0.001 Chronic lung disease, % 10 8 <0.001 Heart failure on presentation, % <0.001 MI indicates myocardial infarction. first door-to-device delay: such critically ill patients with shock or cardiac arrest (n=931) comprised only 6% of the total transfer STEMI PCI population (n=15449). Cardiogenic shock and arrest were both associated with a 20% to 30% lower likelihood of achieving optimal door-to-device timing cardiogenic shock: odds ratio, 0. 72; 95% confidence interval, 0.62 to 0.84; P<0.001 and cardiac arrest: odds ratio, 0.79; 95% confidence interval, 0.64 to 0.98; P=0.03. However, the factors previously identified remained potent negative predictors (including older age, female sex, longer transfer time, diabetes mellitus, off-hour presentation, left bundle branch block on initial ECG) and positive predictors (white race, PCI hospital annual STEMI volume, and regional location in the Midwest) in this expanded model. Finally, a model was performed including a previously emphasized predictor of delay DIDO time and longer DIDO time was confirmed to be a potent independent predictor of delay even when excluding the 30% of transfer patients with DIDO>60 minutes (Table 4). Patients with prolonged DIDO time (per 5-minute increments between 0- and 60-minute total DIDO time) were 34% less likely to achieve optimal door-to-device timing (odds ratio, 0.66; 95% confidence interval, ; P<0.001). Even with inclusion of prolonged DIDO time, older age, previous coronary artery bypass grafting, rural STEMI referring center, and longer transfer times remained significantly associated with delay, whereas white race and Midwest regional location remained associated with timely treatment. There continued to be a positive association

5 5 Dauerman et al Transfer for Acute Myocardial Infarction Table 2. Patient Characteristics According to Hospital Performance Status Variable Optimal Timing in <75% of Transfers Optimal Timing in 75% of Transfers P Value First door-to-device time, 118 (97 148) 93 (80 112) <0.001 min, median (interquartile range) Door-in door-out time, min, 54 (33 77) 38 (28 53) <0.001 median (interquartile range) First ECG prehospital, % 8 15 <0.001 Transfer time, min 34 (21 43) 33 (20 42) <0.001 Congestive heart failure on admission, % Self-transport to first <0.001 hospital, % Peripheral arterial disease, % Chronic lung disease, % Diabetes mellitus, % between volume and quality: PCI hospital annual STEMI volume remained associated with a 25% increased likelihood of achieving optimal first door-to-device time for every increment of 50 patients for annual STEMI volume (odds ratio, 1.25; 95% confidence interval, ; P<0.001). We further investigated whether the relationship between treatment delay and lower PCI hospital annual STEMI volume was explained by any association with prolonged DIDO times. When stratifying PCI-capable hospitals by tertiles of annual STEMI volumes, the median DIDO time for highest tertile STEMI volume hospitals was the same as for lower tertiles: 37 minutes across all 3 tertiles; Spearman rank order correlation coefficient, Thus, we were unable to demonstrate an interaction between door-in door-out times at transfer hospital and volume of STEMI patients treated at the PCI centers. Discussion In this report, we have confirmed that approximately one third of patients requiring transfer for primary PCI do not achieve first door-to-device times of 120 minutes; previously identified predictors of delay (cardiogenic shock, cardiac arrest, and prolonged DIDO times) are significant factors in this process. Our study provides the following unique insights: (1) even when confining the analysis to patients within a 60-minute estimated transfer time, most US PCI hospitals in our study fail to achieve a performance standard of 75% of transfer patients treated in a timely fashion, (2) significant racial and regional disparities in timely transfer process are observed even when controlling for the influence of comorbidities and prolonged DIDO times, (3) the volume of STEMI patients treated by the PCI center has a strong relationship Table 3. Independent Predictors of First Door-to-Device Time 120 Minutes Transfer Patients Excluding Cardiogenic Shock/Arrest and Door-In Door-Out Times Adjusted Odds Ratio 95% CI χ 2 P Value Estimated transfer time (per 5-min increment for transfers time min) Estimated transfer time (per 5-min increment for transfers time <0.001 between 20 and 60 min) Age (per 5-y increment, and age 55 y) <0.001 Age (per 5-y increment, age >55 y) <0.001 Female sex <0.001 Previous CABG <0.001 LBBB on presentation (vs ST-segment elevation) Isolated posterior MI (vs ST-segment elevation) Diabetes mellitus <0.001 Region of presentation 19 Northeast vs South Midwest vs South West vs South Self-transport to first hospital <0.001 PCI hospital annual STEMI volume (per 50-patient increment) <0.001 White race Off-hours hospital presentation Referral hospital bed size (per 50-bed increment) Previous CHF Peripheral arterial disease CABG indicates coronary artery bypass grafting; CHF, congestive heart failure; CI, confidence interval; LBBB, left bundle branch block; MI, myocardial infarction; PCI, percutaneous coronary intervention; and STEMI, ST-segment elevation myocardial infarction.

6 6 Dauerman et al Transfer for Acute Myocardial Infarction with transfer process success and this relationship is independent of the DIDO time achieved by the referring hospitals, and (4) despite previous data on improved reperfusion times with prehospital electrocardiograms, emergency medical service systems were used in a minority of initial transports and prehospital electrocardiograms were obtained in only 11% of the entire sample. Mode and Time of Transfer Our primary end point was first door-to-device time 120 minutes rather than first medical contact to device time 120 minutes. The practicality of our primary end point relates to the observation that self/family transport to the first hospital occurred in 75% of patients. This finding is consistent with previous analyses showing that at least 50% of the STEMI patients do not use the emergency medical system (and the potential for prehospital electrocardiograms) to initiate the process. 6,16 Selftransport to the first hospital was associated in our analysis with an 18% less likelihood of achieving optimal first door-todevice time (P<0.001), which offers an opportunity for quality improvement by education of patients on the importance of emergency medical services. On the contrary, our study found no association between mode of interhospital transport and optimal door-to-device times. This is consistent with previous studies 18 and suggests that regional primary PCI transfer systems need not be based on availability of air transport (among patients with an estimated drive time of 60 minutes). Table 4. Independent Predictors of First Door-to-Device Times 120 Minutes (Model Including DIDO Times) Predictors Adjusted Odds Ratio 95% CI χ 2 P Value DIDO time (per 5-min <0.001 increase for DIDO time between 0 and 60 min) Longer transfer time <0.001 (per 5-min increment for transfers between 20 and 60 min) Previous CABG <0.001 Older age (per 5-y <0.001 increment, beginning at age 55 y) Region of presentation 20 Northeast vs South Midwest vs South West vs South PCI hospital annual STEMI <0.001 volume (per 50-patient increment) PCI hospital bed size (per bed increase and 350 total beds) Rural STEMI referral center White race CABG indicates coronary artery bypass grafting; CI, confidence interval; DIDO, door-in door-out; PCI, percutaneous coronary intervention; and STEMI, ST-segment elevation myocardial infarction. We hypothesized that a 60-minute transfer time limit would provide a rational approach to achieving the first door-todevice goal of 120 minutes. Our findings confirm that 80% of US transfer STEMI patients are within a relatively short distance for transfer, consistent with previous studies analyzing density of US PCI centers. 23 Despite this 60-minute transfer time limit, first door-to-device times 120 minutes were achieved in only two third of patients and only a minority of hospitals achieved this benchmark in >75% of their transfer STEMI population. A previous study of a state-wide STEMI program in North Carolina has suggested that the transfer time should optimally be <30 minutes to achieve timely first door-to-device times: in their study, transfer times >45 minutes were associated with timely first door-to-device times of 120 minutes in only approximately one third of patients. 18 Our results confirm that a shorter transfer time is associated with improved first door-to-device times: there is an 20% less likelihood of achieving timely first door-to-device times for every 5-minute increment in transfer time if estimated transfer time id between 20 and 60 minutes. The failure to achieve timely initial door-to-balloon times in approximately one third of patients in this national registry has implications for clinical care and quality improvement: one interpretation is that a significant minority of transfer patients should undergo initial fibrinolysis rather than primary PCI. 24 An alternative interpretation is that focused improvement is required in use of prehospital activation and transfer to higher volume PCI centers to achieve guideline mandated primary PCI practice. Race, Region, and Volume We identified a 19% increased risk of exceeding optimal first door-to-device times for patients of nonwhite race. Racial disparities in STEMI care have been previously identified 25,26 : for example, in a national study of DIDO times using CMS data, blacks had a 9.1-minute longer DIDO time when compared with white Americans. 17 Our findings are significant in that racial disparity remains persistent even when controlling for the DIDO time differential in multivariable analysis, thus highlighting racial disparities in the entire process of care for transfer STEMI patients. Although state, regional, provincial, and national analyses have provided clear strategies to improve the transfer STEMI process, 7,8,13,16,27 our national analyses confirm that STEMI volume of the referring hospital may be related to achievement of timely first door-to-device times. 8,17 Furthermore, our findings emphasize that one of the most powerful predictors of a first door-to-device time 120 minutes is the volume of STEMI patients treated at the PCI center. High-volume STEMI PCI centers were 20% more likely (per 50 STEMI patient increments) to achieve a timely first door-to-device time when compared with lower volume centers. In multivariable analysis, this relationship persists even after controlling for the influence of transfer time, comorbidities, and region. We hypothesized that larger volume STEMI PCI programs might influence the key quality metric of DIDO time via relationships and feedback between the PCI and the STEMI referring center: this relationship could not be demonstrated, however, because distribution of DIDO times at the associated

7 7 Dauerman et al Transfer for Acute Myocardial Infarction STEMI referral centers were identical for PCI hospitals in lowest versus highest tertile of STEMI volume. Limitations Although the Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines database provides robust prospective information on patients hospitalized with STEMI at broadly inclusive centers across the United States, there were limited data available that would have allowed further analysis of the STEMI referral centers. Multivariable analysis was used to control for confounding among multiple potential predictors of delay and it is possible that other or unknown variables may also play a role in the process of care. Our observations of regional variations in achieving optimal first door-to-device times should be interpreted cautiously given the variability in hospital participation in different US regions. Recently, Shah et al 28 have studied a cohort of STEMI patients from a US hospital discharge database and documented dynamic changes in STEMI PCI volumes and outcomes at individual centers during the past decade. We recognize that hospital STEMI PCI volume and performance may change over time, but this was not specifically evaluated in our cohort because of limited annual sample sizes. We did test for temporal variations in hospital performance over time by fitting a mixed model for patient-level door-to-balloon time <120 minutes related to hospital and time: we found the temporal variation within hospitals over time was small (variance 0.2; χ 2 =40). A fixed effect for trend over time was nonsignificant. Conclusions In this large national analysis of patients undergoing primary PCI for STEMI with a 60-minute transfer time, we observed an ongoing challenge in achieving first door-to-device time 120 minutes. Although our study confirms the significance of both DIDO times and transfer times in achieving a timely STEMI transfer PCI, this analysis identifies 2 other opportunities for quality improvement: (1) volume quality relationships are clear and suggest a need for regional STEMI PCI centers with adequate volumes to ensure expertise and (2) emergency medical service activated transport to the first hospital and associated use of prehospital electrocardiograms remains uncommon and, therefore, represents a significant opportunity to improve transfer times in this challenging population. Disclosures Dr Dauerman is a consultant to Medtronic, Boston Scientific, The Medicines Company, Daichi Sankyo, and Abbott Vascular and has research grants from Abbott Vascular and Medtronic. The other authors report no conflicts. References 1. Jacobs AK, Antman EM, Ellrodt G, Faxon DP, Gregory T, Mensah GA, Moyer P, Ornato J, Peterson ED, Sadwin L, Smith SC; American Heart Association s Acute Myocardial Infarction Advisory Working Group. Recommendation to develop strategies to increase the number of STsegment-elevation myocardial infarction patients with timely access to primary percutaneous coronary intervention. Circulation. 2006;113: doi: /CIRCULATIONAHA Jollis JG, Granger CB, Henry TD, Antman EM, Berger PB, Moyer PH, Pratt FD, Rokos IC, Acuña AR, Roettig ML, Jacobs AK. Systems of care for ST-segment-elevation myocardial infarction: a report From the American Heart Association s Mission: Lifeline. Circ Cardiovasc Qual Outcomes. 2012;5: doi: /CIRCOUTCOMES Terkelsen CJ, Jensen LO, Tilsted HH, Thaysen P, Ravkilde J, Johnsen SP, Trautner S, Andersen HR, Thuesen L, Lassen JF. Primary percutaneous coronary intervention as a national reperfusion strategy in patients with ST-segment elevation myocardial infarction. Circ Cardiovasc Interv. 2011;4: doi: /CIRCINTERVENTIONS Bagai A, Al-Khalidi HR, Muñoz D, Monk L, Roettig ML, Corbett CC, Garvey JL, Wilson BH, Granger CB, Jollis JG. Bypassing the emergency department and time to reperfusion in patients with prehospital ST-segmentelevation: findings from the reperfusion in acute myocardial infarction in Carolina Emergency Departments project. Circ Cardiovasc Interv. 2013;6: doi: /CIRCINTERVENTIONS Fosbol EL, Granger CB, Jollis JG, Monk L, Lin L, Lytle BL, Xian Y, Garvey JL, Mears G, Corbett CC, Peterson ED, Glickman SW. The impact of a statewide pre-hospital STEMI strategy to bypass hospitals without percutaneous coronary intervention capability on treatment times. Circulation. 2013;127: doi: /CIRCULATIONAHA Boothroyd LJ, Lambert LJ, Segal E, Ross D, Kouz S, Maire S, Harvey R, Xiao Y, Brown KA, Nasmith J, Bogaty P. Comparison of outcomes of ambulance users and nonusers in ST elevation myocardial infarction. Am J Cardiol. 2014;114: doi: /j.amjcard Miedema MD, Newell MC, Duval S, Garberich RF, Handran CB, Larson DM, Mulder S, Wang YL, Lips DL, Henry TD. Causes of delay and associated mortality in patients transferred with ST-segment-elevation myocardial infarction. Circulation. 2011;124: doi: / CIRCULATIONAHA Wang TY, Peterson ED, Ou FS, Nallamothu BK, Rumsfeld JS, Roe MT. Door-to-balloon times for patients with ST-segment elevation myocardial infarction requiring interhospital transfer for primary percutaneous coronary intervention: a report from the national cardiovascular data registry. Am Heart J. 2011;161:76 83.e1. doi: /j.ahj Steg PG, James SK, Atar D, Badano LP, Blomstrom-Lundqvist C, Borger MA, Di MC, Dickstein K, Ducrocq G, Fernandez-Aviles F, Gershlick AH, Giannuzzi P, Halvorsen S, Huber K, Juni P, Kastrati A, Knuuti J, Lenzen MJ, Mahaffey KW, Valgimigli M, van t HA, Widimsky P, Zahger D. ESC Guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation. Eur Heart J. 2012;33: O Gara PT, Kushner FG, Ascheim DD, Casey DE Jr, Chung MK, de Lemos JA, Ettinger SM, Fang JC, Fesmire FM, Franklin BA, Granger CB, Krumholz HM, Linderbaum JA, Morrow DA, Newby LK, Ornato JP, Ou N, Radford MJ, Tamis-Holland JE, Tommaso CL, Tracy CM, Woo YJ, Zhao DX, Anderson JL, Jacobs AK, Halperin JL, Albert NM, Brindis RG, Creager MA, DeMets D, Guyton RA, Hochman JS, Kovacs RJ, Kushner FG, Ohman EM, Stevenson WG, Yancy CW; American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines ACCF/AHA guideline for the management of ST-elevation myocardial infarction: a report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines. Circulation. 2013;127:e362 e425. doi: /CIR.0b013e cf Dauerman HL, Sobel BE. Toward a comprehensive approach to pharmacoinvasive therapy for patients with ST segment elevation acute myocardial infarction. J Thromb Thrombolysis. 2012;34: doi: / s x. 12. Ahmed B, Lischke S, Straight F, Gogo P, Leffler S, Kutler M, Schneider DJ, Dauerman HL. Consistent door-to-balloon times of less than 90 minutes for STEMI patients transferred for primary PCI. J Invasive Cardiol. 2009;21: Wilson BH, Humphrey AD, Cedarholm JC, Downey WE, Haber RH, Kowalchuk GJ, Rinaldi MJ, Miller DA, Sarafin JL, Garvey JL. Achieving sustainable first door-to-balloon times of 90 minutes for regional transfer ST-segment elevation myocardial infarction. JACC Cardiovasc Interv. 2013;6: doi: /j.jcin Wang TY, Nallamothu BK, Krumholz HM, Li S, Roe MT, Jollis JG, Jacobs AK, Holmes DR, Peterson ED, Ting HH. Association of door-in to door-out time with reperfusion delays and outcomes among patients transferred for primary percutaneous coronary intervention. JAMA. 2011;305: doi: /jama Terkelsen CJ, Sørensen JT, Maeng M, Jensen LO, Tilsted HH, Trautner S, Vach W, Johnsen SP, Thuesen L, Lassen JF. System delay and mortality among patients with STEMI treated with primary percutaneous coronary intervention. JAMA. 2010;304: doi: /jama

8 8 Dauerman et al Transfer for Acute Myocardial Infarction 16. Lambert LJ, Brown KA, Boothroyd LJ, Segal E, Maire S, Kouz S, Ross D, Harvey R, Rinfret S, Xiao Y, Nasmith J, Bogaty P. Transfer of patients with ST-elevation myocardial infarction for primary percutaneous coronary intervention: a province-wide evaluation of door-in to door-out delays at the first hospital. Circulation. 2014;129: doi: / CIRCULATIONAHA Herrin J, Miller LE, Turkmani DF, Nsa W, Drye EE, Bernheim SM, Ling SM, Rapp MT, Han LF, Bratzler DW, Bradley EH, Nallamothu BK, Ting HH, Krumholz HM. National performance on door-in to door-out time among patients transferred for primary percutaneous coronary intervention. Arch Intern Med. 2011;171: doi: /archinternmed Muñoz D, Roettig ML, Monk L, Al-Khalidi H, Jollis JG, Granger CB. Transport time and care processes for patients transferred with ST-segmentelevation myocardial infarction: the reperfusion in acute myocardial infarction in Carolina emergency rooms experience. Circ Cardiovasc Interv. 2012;5: doi: /CIRCINTERVENTIONS Bradley EH, Nallamothu BK, Herrin J, Ting HH, Stern AF, Nembhard IM, Yuan CT, Green JC, Kline-Rogers E, Wang Y, Curtis JP, Webster TR, Masoudi FA, Fonarow GC, Brush JE Jr, Krumholz HM. National efforts to improve door-to-balloon time results from the Door-to-Balloon Alliance. J Am Coll Cardiol. 2009;54: doi: /j.jacc DeLong ER, Peterson ED, DeLong DM, Muhlbaier LH, Hackett S, Mark DB. Comparing risk-adjustment methods for provider profiling. Stat Med. 1997;16: Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42: Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med. 2010;29: doi: /sim Concannon TW, Nelson J, Goetz J, Griffith JL. A percutaneous coronary intervention lab in every hospital? Circ Cardiovasc Qual Outcomes. 2012;5: doi: /CIRCOUTCOMES Vora AN, Holmes DN, Rokos I, Roe MT, Granger CB, French WJ, Antman E, Henry TD, Thomas L, Bates ER, Wang TY. Fibrinolysis Use Among Patients Requiring Interhospital Transfer for ST-Segment Elevation Myocardial Infarction Care: A Report From the US National Cardiovascular Data Registry. JAMA Intern Med. 2015;175: doi: /jamainternmed Cavender MA, Rassi AN, Fonarow GC, Cannon CP, Peacock WF, Laskey WK, Hernandez AF, Peterson ED, Cox M, Grau-Sepulveda M, Schwamm LH, Bhatt DL. Relationship of race/ethnicity with door-to-balloon time and mortality in patients undergoing primary percutaneous coronary intervention for ST-elevation myocardial infarction: findings from Get With the Guidelines-Coronary Artery Disease. Clin Cardiol. 2013;36: doi: /clc Curtis JP, Herrin J, Bratzler DW, Bradley EH, Krumholz HM. Trends in race-based differences in door-to-balloon times. Arch Intern Med. 2010;170: doi: /archinternmed Glickman SW, Lytle BL, Ou FS, Mears G, O Brien S, Cairns CB, Garvey JL, Bohle DJ, Peterson ED, Jollis JG, Granger CB. Care processes associated with quicker door-in-door-out times for patients with ST-elevationmyocardial infarction requiring transfer: results from a statewide regionalization program. Circ Cardiovasc Qual Outcomes. 2011;4: doi: /CIRCOUTCOMES Shah RU, Henry TD, Rutten-Ramos S, Garberich RF, Tighiouart M, Bairey Merz CN. Increasing Percutaneous Coronary Interventions for ST- Segment Elevation Myocardial Infarction in the United States: Progress and Opportunity. JACC Cardiovasc Interv. 2015;8: