Author + information
- Received January 18, 2010
- Revision received May 31, 2010
- Accepted June 6, 2010
- Published online January 18, 2011.
- James C. Blankenship, MD⁎,⁎ (, )
- Thomas D. Scott, DO⁎,
- Kimberly A. Skelding, MD⁎,
- Thomas A. Haldis, DO†,
- Karen Tompkins-Weber, RN⁎,
- Marie Y. Sledgen, RN⁎,
- Michael A. Donegan, DO‡,
- Jeremy W. Buckley, MD§,
- Jennifer A. Sartorius, MS∥,
- John McB. Hodgson, MD⁎ and
- Peter B. Berger, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. James C. Blankenship, Department of Cardiology 21-60, Geisinger Medical Center, 100 North Academy Drive, Danville, Pennsylvania 17822
Objectives The purpose of this study was to demonstrate the feasibility of routine transfer of ST-segment elevation myocardial infarction (STEMI) patients to achieve percutaneous coronary intervention (PCI) in less than 90 min from presentation.
Background Many PCI hospitals have achieved routine door-to-balloon times under 90 min for patients with STEMI presenting directly to the hospital. However, few patients transferred from a non-PCI center undergo PCI within 90 min of presentation.
Methods Our rural PCI hospital implemented a program in 2005 for rapid triage, transfer, and treatment of STEMI patients and made additional improvements in 2006 and 2007. Intervals between milestones in the STEMI triage/transfer/treatment process were assessed before and after implementation of the program.
Results During the 5-year study period, 676 patients with 687 STEMIs were transferred from 19 community hospitals and underwent PCI. Median door-to-balloon time decreased from 189 min to 88 min (p < 0.001). The time intervals reflecting efficiency of the referring hospitals, transfer services, and PCI hospital all significantly improved. In 2008, median door-to-balloon times were <90 min for 6 of the 7 most frequently referring hospitals. Delays during off-hours presentation in 2004 were abolished after the program was implemented in 2005. In-hospital mortality decreased from 6% before to 3% after implementation of the program. In multivariate modeling, presentation before initiation of the STEMI program predicted increased risk of in-hospital mortality (odds ratio: 3.74, 95% confidence interval: 1.22 to 11.51, p = 0.021).
Conclusions A program of rapid triage, transfer, and treatment of STEMI patients presenting to non-PCI hospitals can reduce in-hospital mortality and produce progressive improvements in door-to-balloon time such that median door-to-balloon times under 90 min are feasible.
Primary percutaneous coronary intervention (PCI) is the preferred treatment for patients presenting with ST-segment elevation myocardial infarction (STEMI) (1). Time to treatment is a critical determinant of outcome (2,3). Recent guidelines recommend reperfusion within 90 min of first medical contact, even for patients transferred from non-PCI hospitals (4). At present, many patients presenting directly to PCI hospital emergency departments do not meet this goal (5), and most patients transferred for PCI STEMI do not meet it despite development of innovative attempts to shorten time to treatment (6–10). Only 1 system has reported median door-to-balloon time under 90 min for transferred patients; this was for only 37 STEMI patients transferred from a single hospital for PCI (11). We analyzed our experience as part of a regional STEMI network involving many hospitals in a rural setting, and the impact of a quality initiative on door-to-balloon times over a 5-year period.
The PCI center in this study is a 437-bed tertiary care hospital in rural central Pennsylvania serving 37 counties with a population of 2.4 million. It has provided primary PCI services since 1985 and treated STEMI patients exclusively with PCI since 1995. The PCI center is part of a rural health system that owned and operated 4 to 5 helicopters based at 4 to 5 sites throughout central Pennsylvania during the study period (Fig. 1). Ambulances were used when helicopter transport was impossible due to weather.
All patients presenting within 12 h of onset of symptoms of STEMI and transferred to our PCI hospital from January 1, 2004, to December 31, 2008, for PCI were considered for inclusion in this analysis. We excluded from this analysis patients transferred after having received thrombolytic therapy. Patients with relative contraindications to PCI, including allergy to contrast dye, major comorbidities, and warfarin anticoagulation, were accepted in transfer and evaluated in the emergency department by the interventional cardiologist. Patients in whom the anticipated benefits of emergency PCI exceeded risk underwent emergency PCI and were included in all analyses reported here. The study was approved by the institutional investigational review board.
Transfer and treatment protocols
During 2004, patients were treated according to standard practice at the time. Pre-hospital electrocardiography (ECG) was not available to any local emergency medical services. Patients with STEMI presenting to a referring hospital were evaluated there by emergency physicians, primary care physicians, or cardiologists. Most referring hospitals routinely transferred STEMI patients for PCI; a small minority of patients received on-site thrombolytic therapy. Transfer of STEMI patients was arranged by calling the emergency physician or the general cardiologist on call at the PCI hospital. The PCI hospital emergency physician screened the patient for appropriateness for PCI, checked cardiac intensive care unit bed availability, and then requested helicopter transfer. On arrival at the PCI hospital, the patient was taken to the emergency department where the emergency physician and the general cardiologist on call assessed the patient, checked laboratory results, and consulted the interventional cardiologist. If appropriate, the interventional cardiologist directed the paging operator to page the catheterization team. All members of the team arrived in the laboratory within 30 min of being paged.
A STEMI rapid triage and transfer protocol was developed in the last half of 2004 and fully implemented on January 1, 2005. It encouraged referring hospital emergency physicians to directly call the PCI hospital emergency department. After reviewing a faxed copy of the ECG and screening the patient with a 9-question checklist, the emergency department physician dispatched a helicopter and alerted the interventional cardiologist, who then called the operator to group-page the catheterization laboratory team. In most cases, the interventional cardiologist and catheterization team met the patient and helicopter crew at the doors of the catheterization laboratory suite. After obtaining a brief history, examination, and informed consent, PCI was performed. Additional improvements to the program were made yearly (Table 1).
During the time period of this study, few local ambulances performed pre-hospital ECGs, and diversion from referring hospitals on the basis of pre-hospital ECGs was extremely rare. In general, interhospital transport by helicopter rather than ambulance was preferred by community emergency physicians due to reluctance of ambulance crews to travel outside of their community, frequent unavailability of ambulances for interhospital transport, and availability of trained emergency medical technicians and nurses with medical control on the helicopter.
Coronary intervention at the PCI hospital during the study period was performed by 3 to 5 staff interventionists employed by the health care system. All operators performed >200 PCIs per year, and during the study period, 4,782 elective, urgent, and emergent PCIs were performed at the PCI hospital. PCI was typically performed with 6-F guide catheters via a femoral approach, heparin or bivalirudin anticoagulation, and vascular closure with intra-arterial collagen plug or suture device. Details of diagnostic catheterization and PCI (e.g., imaging of the noninfarct vessel before versus after PCI, post-dilation, glycoprotein 2b/3a inhibitor treatment) varied among the interventionists.
In the rare circumstance that 2 patients with STEMI presented together after hours, a catheterization laboratory technician, nurses from the 24-h catheterization laboratory recovery suite, and the interventional cardiology fellow prepared the second patient while the first patient was revascularized. When the first patient was stable, the second patient was revascularized. Revascularization for the second patient was almost never delayed by more than 15 min.
Data on all patients were entered into the American College of Cardiology National Cardiovascular Data Registry. Data were collected immediately after PCI by the operator, and information about the remainder of the hospitalization was obtained at the time of discharge by cardiology advanced practitioners.
Transfer times were collected retrospectively for the first 6 months of 2004 using logs from the helicopter service, cardiac catheterization laboratories, and patient charts. Precise times were available for all key data points for all patients. Starting July 1, 2004, transfer times were collected immediately after PCI on standardized forms and entered into a dedicated database. Definitions of times and intervals are listed in Table 2.
Vital status for each patient 1 year after the index STEMI was ascertained from the Social Security Death Index maintained by the U.S. Social Security Administration.
Demographic and procedural data are reported as percentages or mean (± SD). Times are reported as medians since all time distributions were skewed. Only patients transferred by helicopter yielded times from ECG to helicopter dispatch and for helicopter dispatch to arrival at the PCI center. All patients were used for all other analyses. Categorical data were compared using Pearson chi-square statistic or Fisher exact test (in case of small expected numbers). The Kruskal-Wallis test (nonparametric since time variables were skewed) was used for continuous variables.
The nonparametric Jonckheere-Terpstra test was used to check for trends of the time variables by year, to analyze data in which the variable (i.e., times) is at least of an ordinal scale and the underlying hypothesis is of an increasing or decreasing monotonic trend. It was hypothesized that over time (year of presentation), the median times tended to monotonically decrease. Cochran-Armitage trend test was used to check for trends over time of the percentage of patients with total times <90 or <120 min. All tests were 2-sided, and p values <0.05 were considered significant.
Relationships between clinical characteristics that varied over the study period were examined to determine whether they were independently related to treatment intervals. For characteristics that both varied by year of presentation and were associated with significant differences in median times, the bivariate test results reported in the tables were checked after adjusting for the related baseline characteristic(s). To obtain adjusted p values, simple regressions were performed using times as the dependent variables (log transformed since highly skewed). Adjusted p values (adjusted for other covariates in the model) were obtained for the independent variable of interest (such as year of presentation). If the resulting adjusted p value changed the interpretation of results, the p value was reported in the tables as well.
To identify predictors of death, potential variables were bivariately associated with death using Fisher exact test for categorical variables and the Wilcoxon-Mann-Whitney rank sum test for continuous predictors. Variables checked as possible predictors included all of the time segments and procedural characteristics prior to intervention (as listed in Table 3), referring hospital, year of MI, presentation during workday versus off-hours, transport method, and interventional operator. Variables that were bivariately associated with death with a p value <0.1 were kept for logistic regression modeling. Logistic modeling was performed with backward elimination to retain significant predictors in the final model. Ejection fraction was excluded from multivariate models since it was determined post-PCI.
During the 5-year study period, 676 patients with 687 STEMIs were transferred from 19 community hospitals and underwent PCI. Helicopter transport was used for 571 (83%) of patients transferred for STEMI PCI; the rest were transferred by ambulance. During this same period, 278 STEMI patients presented directly to the PCI center emergency department and are not included in this analysis.
Several demographic and procedural characteristics (Table 3) fluctuated significantly from year to year during the study, but the only consistent trend over time was in the incidence of prior MI, which rose from 6% to 8% in 2004 to 2006, to 17% to 18% in 2007 to 2008 (p = 0.001).
Median times for intervals in the STEMI triage/transfer/treatment process are shown by year in Table 4. Symptom onset-to-initial presentation times did not change over the study period. All other time intervals decreased significantly and progressively, with an equilibrium reached only in the fifth year of the study. Median door-to-balloon time decreased by 101 min over the 5-year study period, with three-fourths of the decrease in the first year of the program (Fig. 1). Two-thirds of the 101-min decrease was attributable to efficiencies at the PCI hospital; one-third were due to efficiencies before the patient reached the PCI hospital. The percentage of patients with door-to-balloon times under 90 min increased from 1% in 2004 to 53% in 2008.
Six of these 7 referring hospitals most commonly referring STEMI patients had median door-to-balloon times <90 min in 2008, and all 7 hospitals had significantly shorter door-to-balloon time at the end of the study period (Table 5). Time from initial presentation to helicopter dispatch, the interval most closely reflecting referring hospital efficiency, decreased numerically at 6 of the 7 hospitals and significantly so at 4 of the 7. The least efficient hospital in 2004 became the most efficient in 2008.
Workday versus off-hours presentation
Throughout the study period, workday and off-hour patients had similar median times from presentation to ECG, and from ECG to dispatch (Table 6). No baseline characteristics were associated with on- versus off-hours presentation. In 2004, before implementation of the STEMI program, median time from arrival at the PCI center to arrival in the catheterization laboratory was shorter for on-hours patients (38 min [interquartile range (IQR): 22, 50 min]) than off-hours patients (52 min [IQR: 37, 73 min], p = 0.001). Door-to-balloon time tended to be shorter for workday patients (180 min [IQR: 135, 205 min]) than off-hours patients (191.5 min [IQR: 155, 254 min], p = 0.063). After implementation of the STEMI program in 2005, differences between workday and off-hours median time intervals narrowed to 1 to 2 min (p = NS) (Table 6).
In-hospital mortality was 3.4% (6% for patients with MI in 2004, 4% in 2005, 3% in 2006, 2% in 2007, and 3% in 2008; differences not statistically significant, p = 0.50). One-year mortality was 10.1% for patients with MI in 2004, 8.1% in 2005, 4.9% in 2006, 6.8% in 2007, and 4.4 % in 2008 (p = 0.4).
Correlates of mortality
Variables checked as possible predictors of mortality included all time segments, procedural characteristics, referring hospital, year of MI, presentation during workday versus off-hours, transport method, and interventional operator. In the final logistic model, 3 variables were significantly associated with death: shock (odds ratio [OR]: 26.98, 95% confidence interval [CI]: 9.77 to 74.48, p < 0.001), age (each increasing year) (OR: 1.09, 95% CI: 1.05 to 1.14, p < 0.001), and presentation in 2004 (before initiation of the STEMI program) compared with 2005 to 2008 (OR :3.74, 95% CI: 1.22 to 11.51, p = 0.021).
Three variables were significantly associated with 1-year mortality in the final multivariate logistic model: shock (OR: 19.56, 95% CI: 7.24 to 52.84, p < 0.001), age (each increasing year, OR: 1.09, 95% CI: 1.06 to 1.13, p < 0.001), and prior congestive heart failure (OR: 8.00, 95% CI: 2.58 to 24.58; p < 0.001). There was a trend toward higher mortality in patients presenting in 2004 (before initiation of the STEMI program) compared with 2005 to 2008 (OR: 2.9, 95% CI: 0.98 to 8.45, p = 0.054).
Patients presenting directly to PCI center
During the period covered in this study, median door-to-balloon times for 263 patients presenting to the angioplasty center improved from 110 min in 2004 to 76 min in 2005, 60 min in 2006, 42 min in 2007, and 43 min in 2008. In-hospital mortality changed from 5.6% in 2004 (n = 36) to 0.4% in 2005 to 2008 (n = 227), and cumulative 1-year mortality changed from 8.3% for patients with MI in 2004 to 2.1% for patients with MI in 2005 to 2008.
The most important finding of this study is that rapid diagnosis, triage, transfer, and treatment is possible for some STEMI patients in a rural setting in the U.S. In the last year of our study, median door-to-balloon time for patients transferred for PCI was 88 min, and 53% achieved a door-to-balloon time under 90 min. Our system has produced door-to-balloon times over an hour shorter than the 171 min reported by Nallamothu et al. from the National Registry of Myocardial Information database (12), and shorter than those recently reported by other systems (8–10).
A second important finding is that sequential improvements in all components of the system—the referring hospitals, transport services, and the PCI hospital—contributed to decreasing delays to treatment. Referring hospitals decreased the time from arrival at the referring hospital to ECG from 6.5 to 3 min (p = 0.012), and the time from ECG to helicopter dispatch from 20.5 min to 12 min (p < 0.001). Transfer services decreased time from helicopter dispatch to arrival at the PCI hospital from 53 min to 47 min (p = 0.003). Within the PCI hospital, time from arrival at the hospital to arrival in the catheterization laboratory decreased from 46 to 4 min (p < 0.001), and time from arrival in the catheterization laboratory to balloon inflation decreased from 36 to 17 min (p < 0.001). These improvements occurred gradually. Door-to-balloon time decreased from 189 min in 2004 (before implementation of the program) to 113 min in 2005 (after implementation of the program). However, additional refinements to the program led to further improvement in door-to-balloon time to 104.5 min in 2006, to 95 min in 2007, and to 88 min in 2008. These subsequent improvements were partly due to implementing proven strategies reported by Bradley et al. (13) and to additional efforts by all members of the system to increase efficiency of the system.
A third important finding of this study is that the results described above were achieved without full implementation of formal rigid systems advocated by others (14–16). Formal transfer agreements were not signed between referring hospitals and the PCI hospital. Instead, a simple 1-call-for-transfer program was developed by the PCI hospital and offered to emergency department personnel at local referring hospitals. Formal referring hospital treatment protocols were suggested but not required. There were no administrative relationships among the hospitals participating in the system reported in this study. Participation in the STEMI program was informal and completely voluntary.
Other investigators have reported faster door-to-balloon times during workday hours compared to off-hours (17,18). However, after implementation of our system, median door-to-balloon times for workday hours were similar to and within a few minutes of those for off-hours. Because all of the interventionists and at least 1 member of each catheterization laboratory staff call team live within 5 min driving time of the PCI hospital, there was always ample time for the catheterization team to reach the laboratory before the patient arrived.
Prior to implementation of this STEMI program, referring hospitals demonstrated wide differences in the efficiency with which they identified and triaged STEMI patients. Initially, median door-to-dispatch times among the 7 most frequently referring hospitals ranged from 21 to 80 min. By 2008, these times had decreased to 12 to 27 min, demonstrating that community hospitals can dramatically improve their own processes for evaluation and treatment of STEMI patients. We attribute these improvements to consistent communication with the PCI hospital physicians, including feedback after individual STEMI patient transfers, and the efforts of health care providers at referring hospitals to improve their processes. However, differences remaining among referring hospitals, even after 4 years, demonstrate that hospitals embrace the challenge of establishing and improving STEMI programs with varying levels of enthusiasm and success.
Median door-to-balloon times in 2008 were under 90 min for all 5 referring hospitals with flight times to the PCI hospital <10 min (and with driving times of <40 min). Because most patients in the U.S. live within this proximity to a PCI hospital (19), PCI in <90 min should be feasible for the majority of STEMI patients in the U.S. These data also raise the issue of whether initiating low-volume no-surgery-on-site STEMI programs are of any benefit when efficient STEMI transfer systems are available.
Several studies have found that door-to-balloon time is an independent correlate of death (2,3), but ours did not. The vast majority of our patients had door-to-balloon times of 70 to 110 min (Fig. 2), and variation in door-to-balloon times within this relatively narrow range is likely not sufficiently influential to produce differences in mortality large enough to be identified in a study of this size. Larger studies with larger ranges of door-to-balloon times would be more likely to demonstrate a correlation between door-to- balloon time and mortality. Although door-to-balloon time did not correlate with mortality, presentation before initiation of the STEMI transfer program did correlate with higher mortality (OR: 3.74, p = 0.021).
We did not track STEMI patients with absolute contraindications to PCI, nor did we track patients presenting with STEMI more than 12 h after the onset of symptoms. The generalizability of the results of this study may be limited by several unique features of the region and health system in which it was performed. The PCI hospital catheterization laboratory is a closed laboratory, staffed by a single group of employed interventionists. The helicopters used for transport are owned by the same health system that operates the PCI hospital and are based in areas that frequently refer patients, an arrangement that favors efficient transfers. Pre-hospital ECGs were not routinely available to local ambulance units during the study period, and their use may further improve transfer efficiency. Standardized community hospital order sets and transfer strategies, if implemented, might have provided further improvements in efficiency. This study did not compare a helicopter transport system to an efficient ambulance transport system, and for short transport distances, it is possible that an efficient ambulance transfer system would achieve comparable transfer times and better cost efficiency compared with helicopter transport. Due to the small number of mortality events in the multivariate logistic regression model, results regarding correlates of mortality are presented with caution.
We used wire-crossing time prior to 2007 instead of balloon inflation time because: 1) balloon inflation is an arbitrary end point that often does not reflect time of reperfusion; 2) many procedures started with rheolytic or aspiration thrombectomy, not balloons; and 3) there was no consensus that balloon inflation or device activation was the correct surrogate for reperfusion time. Median time from wire crossing lesion to balloon inflation in 2007 and 2008 was 2 min [Q1, Q3 = 1, 3], representing a small increase in times when balloon inflation was substituted for wire crossing in 2007.
A program of rapid triage, transfer, and treatment of STEMI patients presenting to non-PCI hospitals can reduce in-hospital mortality and produce progressive improvements in door-to-balloon time such that median door-to-balloon times under 90 min are feasible.
The past and current success of the STEMI program at the authors' PCI center could not have been achieved without the efforts of the dedicated emergency physicians and nurses at our center and at referring hospitals, Lifeflight personnel, and staff and physicians of our catheterization laboratories.
Funding for statistical analyses was provided by the Clinical Endowment for Research of the Geisinger Health System. Dr. Skelding has DSMB and Educator relationships with Medtronic. Dr. Berger has served as a consultant to AstraZeneca, Boehringer Ingelheim, and Eli Lilly/Daiichi Sankyo; and has received research funding from AstraZeneca, Corgenix/Aspirinworks, Eli Lilly/Daiichi Sankyo, Haemoscope, Helena, The Medicines Company, and Thrombovision. All other authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- Received January 18, 2010.
- Revision received May 31, 2010.
- Accepted June 6, 2010.
- American College of Cardiology Foundation
- McNamara R.L.,
- Wang Y.,
- Herrin J.,
- et al.
- Brodie B.R.,
- Hansen C.,
- Stuckey T.D.,
- et al.
- Antman E.M.,
- Hand M.,
- Armstrong P.W.,
- et al.
- Sejerstein M.,
- Sillesen M.,
- Hansen P.R.,
- et al.
- Henry T.,
- Sharkey S.,
- Nicholas B.,
- et al.
- Ahmed B.,
- Lischke S.,
- Straight F.,
- et al.
- Nallamothu B.K.,
- Bates E.R.,
- Herrin E.R.,
- et al.
- Henry T.D.,
- Atkins J.M.,
- Cunningham M.S.,
- et al.
- Jacobs A.K.
- Nallamothu B.K.,
- Bates E.R.,
- Wang Y.,
- Bradley E.H.,
- Krumholz H.M.