Relation Between Hospital Specialization With Primary Percutaneous Coronary Intervention and Clinical Outcomes in ST-Segment Elevation Myocardial Infarction
National Registry of Myocardial Infarction-4 Analysis
Background— Hospitals with primary percutaneous coronary intervention (PPCI) capability may choose to predominately offer PPCI to their patients with ST-segment elevation myocardial infarction (STEMI), or they may selectively offer PPCI or fibrinolytic therapy based on patient and hospital-level factors. Whether a greater level of hospital specialization with PPCI is associated with better quality of care is unknown.
Methods and Results— We analyzed data from the National Registry of Myocardial Infarction-4 to compare in-hospital mortality and times to treatment in STEMI across different levels of hospital specialization with PPCI. We divided 463 hospitals into quartiles of PPCI specialization based on the relative proportion of reperfusion-treated patients who underwent PPCI (≤34.0%, >34.0 to 62.5%, >62.5 to 88.5%, >88.5%). Hierarchical multivariable regression assessed whether PPCI specialization was associated with better outcomes, after adjusting for patient and hospital characteristics, including PPCI volume. We found that greater PPCI specialization was associated with a lower relative risk of in-hospital mortality in patients treated with PPCI (adjusted relative risk comparing the highest and lowest quartiles, 0.64; P=0.006) but not in those treated with fibrinolytic therapy. Compared with patients at hospitals in the lowest quartile of PPCI specialization, adjusted door-to-balloon times in the highest quartile were significantly shorter (99.6 versus 118.3 minutes; P<0.001), and the likelihood of door-to-balloon times exceeding 90 minutes was significantly lower (relative risk, 0.78; P<0.001). Adjusting for PPCI specialization diminished the association between PPCI volume and clinical outcomes.
Conclusions— Greater specialization with PPCI is associated with lower in-hospital mortality and shorter door-to-balloon times in STEMI patients treated with PPCI.
Received July 22, 2005; revision received November 10, 2005; accepted November 11, 2005.
Hospitals with primary percutaneous coronary intervention (PPCI) capability must make strategic decisions about delivering reperfusion therapy to patients with ST-segment elevation myocardial infarction (STEMI). These facilities may try to exclusively provide PPCI, or they may provide either PPCI or fibrinolytic therapy on an individual basis according to patient- and hospital-level factors.1,2 There are few data to inform this decision. It is possible that hospitals that routinely provide both PPCI and fibrinolytic therapy are able to deliver reperfusion faster and more effectively than hospitals that use a single approach. Alternatively, it may be that hospitals committed to PPCI provide better care by eliminating additional clinical decision-making steps and streamlining processes of care.
Clinical Perspective p 229
Our goal was to provide evidence about the potential advantages of these different strategies. Accordingly, we assessed in-hospital mortality rates and treatment times in patients who received reperfusion therapy as a function of hospital PPCI specialization. We use the term “specialization” to refer to the relative proportion of reperfusion-treated patients (PPCI plus fibrinolytic therapy) treated with PPCI, a concept that is distinct from the overall number of STEMI patients treated or PPCIs performed at a hospital (ie, hospital volume). Our findings may have important policy implications for US hospitals, as the growing availability of PPCI makes it likely that an increasing number of STEMI patients will be treated at hospitals with the capacity to provide both types of reperfusion therapy.
The National Registry of Myocardial Infarction (NRMI) is an ongoing observational study of acute myocardial infarction patients admitted to participating hospitals in the United States. We included patients in the NRMI-4 cohort admitted between January 1, 2000, and December 31, 2002. Details of NRMI hospitals, cohorts, and data collection methods have been described elsewhere.3
In brief, the NRMI protocol specifies that all consecutive patients with the diagnosis of acute myocardial infarction are enrolled. Cases are screened according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) discharge diagnosis code of 410.X1. Myocardial infarction is confirmed on the basis of a suggestive patient history and 1 or more of the following criteria: (1) Cardiac biomarker elevation; (2) ECG evidence; and (3) scintigraphic, echocardiographic, or autopsy evidence.
The validity of NRMI data collection has been established through a comparison with the Cooperative Cardiovascular Project.4 Institutional review board approval for data collection was obtained at participating centers in NRMI as required. Data on hospital characteristics were supplemented by using the 2000 American Hospital Association Annual Survey5 and the SMG Marketing Group dataset.6
Patients with STEMI who were treated with PPCI or fibrinolytic therapy were eligible for this study. We included patients (1) with ST-segment elevation in 2 or more leads or left bundle branch block on their first ECG; (2) who presented within 12 hours of symptom onset; and (3) who were not transferred in from another acute-care hospital. So as to consider only patients who could have received either type of reperfusion, we excluded any patients with contraindications to fibrinolytic therapy or PPCI. These included active internal bleeding or known bleeding diathesis; recent surgery, major trauma, or traumatic cardiopulmonary resuscitation; chest pain resolution, known intracranial neoplasm, arteriovenous malformation, or aneurysm; recent cerebrovascular accident; severe uncontrolled hypertension; severe comorbid disease; patient or family refusal; and unsuitable coronary anatomy.
We also excluded patients (1) who presented to hospitals that did not perform PPCI (n=23 215 patients at 709 hospitals) or to hospitals that treated ≤5 patients per year with reperfusion therapy during the study period (n=539 patients at 70 hospitals) and (2) those with times from hospital arrival to treatment that were unknown or >6 hours (n=1034). Patients transferred to another acute-care hospital at the time of discharge (n=2627, or 7.1%) were considered as discharged alive and were included in our analysis. Because transfer rates were generally higher at discharge among hospitals with less PPCI specialization, we presumed that all of those patients survived their acute hospitalization period, an assumption that biased against hospitals with greater degrees of PPCI specialization. We also repeated our analysis after excluding those patients, and the results were not substantially different. The final study population comprised 37 233 patients treated at 463 hospitals.
Clinical Data and Hospital Specialization With PPCI
Patient records included data on demographics and cardiovascular and noncardiovascular comorbidities (history of myocardial infarction, coronary revascularization, chronic renal insufficiency, and stroke). Clinical data were available on presenting characteristics, including chest pain, symptoms of congestive heart failure, systolic blood pressure and heart rate, time of symptom onset, and time of hospital arrival. Available data from the diagnostic ECG included location of myocardial infarction and extent of lead involvement.
PPCI specialization at each hospital was determined by the percentage of reperfusion-treated patients who received PPCI. For example, a hospital that used PPCI to treat 30 of 40 reperfusion-treated patients had a rate of PPCI specialization of 75%, which was higher than that of a hospital that used PPCI to treat 75 of 150 reperfusion-treated patients (ie, a 50% PPCI specialization rate) despite the former having a lower overall number of PPCI and reperfusion-treated cases.
The study population for our analyses was divided into 2 cohorts based on whether the patients were treated with PPCI or fibrinolytic therapy. Our primary outcome was in-hospital mortality. As a secondary outcome, we also assessed time to reperfusion. Time to reperfusion for PPCI was defined as the time from hospital arrival to first balloon inflation (ie, door-to-balloon time). Time to reperfusion for fibrinolytic therapy was the time from hospital arrival to initiation of drug infusion (ie, door-to-needle time). In both cohorts, we assessed for absolute differences in time to treatment as well as the likelihood of “delayed” reperfusion. Consistent with treatment guidelines,1 we considered PPCI to have been delayed when door-to-balloon times were >90 minutes and fibrinolytic therapy to have been delayed when door-to-needle times were >30 minutes.
For the main analysis, we categorized hospitals into 4 quartiles on the basis of the relative proportion of reperfusion-treated patients treated with PPCI: ≤34.0%, >34.0 to 62.5%, >62.5 to 88.5%, and >88.5%. Unadjusted analyses evaluated differences in patient and hospital characteristics across quartiles of patients in the PPCI and fibrinolytic cohorts, with ANOVA tests used for continuous variables and χ2 tests used for categorical variables. Multivariable hierarchical logistic-regression models examined the independent relation between PPCI specialization and the likelihood of (1) in-hospital mortality and (2) delayed reperfusion. Multivariable hierarchical linear-regression models evaluated the independent relation between PPCI specialization and (1) door-to-balloon times and (2) door-to-needle times.
Models were adjusted for the following patient and hospital characteristics: Demographics, comorbidities, prior myocardial infarction or revascularization, chest pain, systolic blood pressure, heart rate, Killip classification, performance of a prehospital ECG, ECG findings, time from symptom onset to hospital arrival, time and day of presentation (weekdays between 8 am and 6 pm or other), hospital location and teaching status, and the presence of on-site cardiac surgery.
Given its strong correlation with PPCI specialization (R2=0.34, P<0.001), we adjusted for hospital volume in several ways. First, we categorized low- and high-volume hospitals according to a threshold of 50 PPCI cases per year. This categorization was based on the median number of PPCI cases at hospitals in our study population and earlier reports from NRMI.7 Next, we performed additional analyses with different cutoffs for PPCI volume (<30, 30 to 70, >70) and estimates for hospital volume based on the number of STEMI cases per year (<50, 50 to 150, >150). Because there were no substantial differences in outcomes across these analyses, we report results of the first analysis for ease of interpretation. In addition, we examined the association between PPCI specialization and clinical outcomes before and after adjusting for PPCI volume, as well as the association between PPCI volume and clinical outcomes before and after adjusting for PPCI specialization. Finally, we constructed models to test for interaction effects between PPCI specialization and volume as well as repeating our analyses separately in low- and high-volume hospitals.
We used hierarchical models to account for clustering of patients both within hospitals (the unit of enrollment in NRMI) and reporting study periods. In all models, the intercept and calendar time at admission were modeled as random effects to account for hospital-specific effects. Odds ratios from the multivariable logistic-regression models were converted into risk ratios.8 Before the analysis, door-to-balloon and door-to-needle times in the multivariable linear-regression models were logarithmically transformed because of their skewed distributions. To improve clinical interpretation of the results, we transformed these values back into their original units, ie, minutes, by using geometric means9,10 and simulation techniques.11
We used bootstrapping techniques that drew random samples of 1000 replicas from our dataset to internally validate several of our findings; overall, we found consistent results across our unadjusted and adjusted analyses. All analyses were performed with SAS version 9.1 (SAS Institute Inc) and Stata SE version 8.0 (Stata Corporation).
We identified 37 233 patients treated with PPCI or fibrinolytic therapy at 463 hospitals. Of these, 22 387 (60.1%) patients were treated with PPCI, and 14 846 (39.9%) were treated with fibrinolytic therapy. PPCI specialization varied widely across the 463 hospitals, with the mean percentage of patients who received PPCI ranging from <1% to 100% (median, 63.0%; interquartile range, 34.5% to 88.5%). Because of the large size of the dataset, statistically significant differences were found in patients for several variables across quartiles of PPCI specialization; however, for many variables, absolute differences were small and were not deemed clinically significant.
Table 1 and Table 2 display key patient- and hospital-level characteristics for which substantial differences were found across quartiles of PPCI specialization. Patients treated with PPCI at hospitals in the lowest quartile of PPCI specialization were often at higher risk and were more likely to present during usual working hours. In general, patients treated at hospitals in the highest quartile of PPCI specialization were more frequently cared for at higher-volume hospitals and hospitals with on-site cardiac surgery, compared with those at hospitals in the lowest quartile.
Greater PPCI specialization was associated with lower in-hospital mortality for patients treated with PPCI (Table 3). For example, patients treated with PPCI at hospitals in the highest quartile of PPCI specialization had a nearly 36% lower adjusted relative risk (RR) of in-hospital death compared with patients treated with PPCI at hospitals in the lowest quartile (RR, 0.64; 95% confidence interval [CI], 0.46 to 0.88; P=0.006). PPCI specialization was not associated with in-hospital mortality for patients treated with fibrinolytic therapy (Table 4).
Table 3 and Table 4 also display times to reperfusion and the likelihood of receiving delayed reperfusion across quartiles of PPCI specialization. Overall, door-to-balloon times were shorter and door-to-needle times were longer for hospitals with greater PPCI specialization. The likelihood of a door-to-balloon time exceeding 90 minutes was significantly lower in patients treated with PPCI at hospitals in the highest quartile of PPCI specialization compared with the lowest quartile (RR, 0.78; 95% CI, 0.67 to 0.88; P<0.001), but the likelihood of door-to-needle times exceeding 30 minutes was significantly higher in patients treated with fibrinolytic therapy (RR, 1.43; 95% CI, 1.31 to 1.54; P<0.001).
Finally, PPCI specialization was associated with in-hospital death in patients treated with PPCI before (RR for highest quartile versus lowest quartile, 0.67, 95% CI, 0.51 to 0.88; P=0.004) and after adjusting for PPCI volume (RR for highest quartile versus lowest quartile, 0.64, 95% CI, 0.46 to 0.88; P=0.006). In contrast, when we adjusted for all patient and hospital characteristics except PPCI specialization in the PPCI cohort, an annual PPCI volume of >50 was associated with a lower RR for in-hospital death (RR, 0.86; 95% CI, 0.75 to 1.01; P=0.071) and a door-to-balloon time exceeding 90 minutes (RR, 0.86; 95% CI, 0.80 to 0.92; P<0.001) compared with a PPCI volume of ≤50 in patients treated with PPCI. However, the relation between PPCI volume and clinical outcome diminished after adjusting for PPCI specialization (RR, 0.96; 95% CI, 0.79 to 1.14; P=0.58 for in-hospital death; and RR, 0.96; 95% CI, 0.89 to 1.02; P=0.19 for door-to-balloon times exceeding 90 minutes).
We also found that there were no significant interaction effects (P>0.05) between PPCI specialization and PPCI volume on in-hospital mortality, door-to-balloon times, or door-to-needle times. When low- and high-volume hospitals were analyzed separately, the relation between greater PPCI specialization and improved clinical outcomes demonstrated similar trends in both categories of PPCI volume (Table 5).
We found that among hospitals with PPCI capability, their relative utilization of PPCI over fibrinolytic therapy as a reperfusion strategy in STEMI was directly related to their ability to provide PPCI in an effective and timely manner. We believe our findings have important clinical implications for hospitals providing reperfusion therapy in the United States. Although earlier studies have evaluated the influence of PPCI volume on clinical outcomes with PPCI in STEMI patients,7,12 none have specifically examined the effect of PPCI specialization. Our findings suggest that the relation between PPCI specialization and clinical outcomes is independent of PPCI volume. In fact, adjusting for PPCI specialization diminished the relation between PPCI volume and clinical outcomes. PPCI specialization also appeared to be equally important at low- and high-volume hospitals, with no significant interaction effects noted between PPCI specialization and PPCI volume.
Based on recent clinical trial data, there is growing interest in expanding the use of PPCI as the principal strategy for reperfusion therapy in STEMI.13 Many experts have promoted this strategy as a means to improve short- and long-term clinical outcomes,14 and as a result, the use of fibrinolytic therapy in the United States has decreased in hospitals during the last decade while PPCI rates have nearly tripled.15 In addition, hospitals without on-site cardiac surgery that previously performed diagnostic cardiac catheterization only are now performing PPCI in patients with STEMI.16
However, given the intensive resources required for 24-hour PPCI availability, it is likely that many hospitals will be unable to provide PPCI exclusively. In our study population, for example, only 1% of hospitals exclusively used PPCI, and >50% of hospitals used it in less than two thirds of patients. When hospitals utilize both strategies, it is possible that times to treatment may be compromised. Doorey and colleagues17 found that the introduction of PPCI at a single hospital that had previously used only fibrinolytic therapy led to substantial delays in delivering both types of reperfusion therapy. In many cases, confusion over which therapy to institute was cited as a reason for the delays by emergency department physicians in that study.
Overall, the challenges of providing PPCI in an effective and timely manner may be greater than those for fibrinolytic therapy. With PPCI, streamlining the processes of care involves several groups of physicians and support staff.18 It also typically requires incorporating system-wide measures outside the emergency department, such as early activation of the cardiac catheterization laboratory. These complex factors may explain why specialization was related to better clinical outcomes for PPCI but not for fibrinolytic therapy, which is provided as a drug infusion and requires fewer resources for delivery. However, we did note that door-to-needle times were worse at hospitals in the highest quartile of PPCI specialization. Although this may be due to selection bias against the use of fibrinolytic therapy at these hospitals, hospitals with greater PPCI specialization may need to be vigilant for treatment delays in patients treated with fibrinolytic therapy.
The significance of our findings includes their potential for policy application at the hospital level. As opposed to PPCI volume, increasing the degree of PPCI specialization may be achievable for many hospitals and could translate into lower in-hospital mortality rates and shorter door-to-balloon times. In this regard, PPCI specialization reflects the commitment of a hospital to maximize procedure benefit rather than increasing experience through a larger number of cases. From a healthcare system perspective, an additional benefit is that increasing PPCI specialization does not require reallocating large numbers of patients across hospitals.19 Of course, the ability to improve PPCI specialization may be unattainable for some hospitals, given that substantial resources may be required to provide 24-hour availability. Hospitals operating under this scenario may need to consider approaches for better coordinating the use of both types of reperfusion therapy, such as the development of explicit protocols to rapidly guide patients based on their risk and time of presentation.
Importantly, our use of the term specialization is distinct and specific. It should not be confused with previous studies that have linked clinical outcomes to certain provider or hospital characteristics. Particularly in the cancer literature, specialization has had a wide variety of meanings, from greater knowledge or technical expertise among clinicians to specialized equipment or resources at large hospitals to unique organizational structures in healthcare systems.20–22 It is also not uncommon for specialization to refer to the absolute number of procedures at a hospital or by a provider, both of which have been shown to be associated with clinical outcomes in several surgical and medical conditions.23,24
Our analysis should be interpreted in the context of the following study design issues. First, NRMI collects voluntarily reported data from participating hospitals. Although NRMI hospitals do include a wide variety of acute-care hospitals (1 in 4 hospitals in the United States contributed to NRMI-4), these centers generally tend to be larger than nonparticipating hospitals. Second, this was an observational study, and there is the potential for residual confounding despite adjustment for several patient- and hospital-level covariates. For example, PPCI specialization may simply reflect a greater degree of physician or hospital resources at that institution for which we were unable to account, such as higher individual provider volumes. Our findings therefore need to be confirmed in other registries. Finally, although we have demonstrated a strong association between PPCI specialization and improved clinical outcomes, this analysis is unable to determine whether the relation is causal.
Hospitals and healthcare systems will face several challenges as the use of PPCI continues to expand to more facilities across the United States. These challenges may be especially important for hospitals that are unable to provide PPCI exclusively for their patients with STEMI. Because PPCI specialization is associated with lower in-hospital mortality and shorter door-to-balloon times, hospitals that perform PPCI may improve clinical outcomes by more exclusively committing to PPCI as a reperfusion strategy. However, if greater PPCI specialization is not possible owing to logistical constraints, innovative approaches may be needed to maximize the clinical benefit of reperfusion therapy.
This project was supported by the National Heart, Lung, and Blood Institute (R01HS10407-01). Dr Nallamothu is supported as a clinical scholar under a K12 grant from the National Institutes of Health (RR017607-01). Genentech, Inc (South San Francisco, Calif), approved the study protocol before the analysis and provided access to NRMI data without charge.
Dr Pollack serves on the advisory board for NRMI. The other authors report no conflicts.
Antman EM, Anbe DT, Armstrong PW, Bates ER, Green LA, Hand M, Hochman JS, Krumholz HM, Kushner FG, Lamas GA, Mullany CJ, Ornato JP, Pearle DL, Sloan MA, Smith SC Jr. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction—executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1999 Guidelines for the Management of Patients with Acute Myocardial Infarction). J Am Coll Cardiol. 2004; 44: 671–719.
Nallamothu BK, Bates ER, Herrin J, Wang Y, Bradley EH, Krumholz HM. Times to treatment in transfer patients undergoing primary percutaneous coronary intervention in the United States: National Registry of Myocardial Infarction (NRMI)-3/4 analysis. Circulation. 2005; 111: 761–767.
American Hospital Association. The Annual Survey of Hospitals Database: Documentation for 2000 Data. Chicago, Ill: American Hospital Association; 2000.
SMG Marketing Group. SMG Dataset. Available at http://www.smg.com. Accessed January 28, 2002.
Bland JM, Altman DG. Transformations, means, and confidence intervals. BMJ. 1996; 312: 1079.
Bland JM, Altman DG. The use of transformation when comparing two means. BMJ. 1996; 312: 1153.
King G, Tomz M, Wittenberg J. Making the most of statistical analyses: improving interpretation and presentation. Am J Pol Sci. 1998; 44: 341–355.
Canto JG, Every NR, Magid DJ, Rogers WJ, Malmgren JA, Frederick PD, French WJ, Tiefenbrunn AJ, Misra VK, Kiefe CI, Barron HV. The volume of primary angioplasty procedures and survival after acute myocardial infarction: National Registry of Myocardial Infarction 2 Investigators. N Engl J Med. 2000; 342: 1573–1580.
Weaver WD. All hospitals are not equal for treatment of patients with acute myocardial infarction. Circulation. 2003; 108: 1768–1771.
Rogers WJ, Canto JG, Lambrew CT, Tiefenbrunn AJ, Kinkaid B, Shoultz DA, Frederick PD, Every N. Temporal trends in the treatment of over 1.5 million patients with myocardial infarction in the US from 1990 through 1999: the National Registry of Myocardial Infarction 1, 2 and 3. J Am Coll Cardiol. 2000; 36: 2056–2063.
Doorey A, Patel S, Reese C, O’Connor R, Geloo N, Sutherland S, Price N, Gleasner E, Rodrigue R. Dangers of delay of initiation of either thrombolysis or primary angioplasty in acute myocardial infarction with increasing use of primary angioplasty. Am J Cardiol. 1998; 81: 1173–1177.
Bradley EH, Roumanis SA, Radford MJ, Webster TR, McNamara RL, Mattera JA, Barton BA, Berg DN, Portnay EL, Moscovitz H, Parkosewich J, Holmboe ES, Blaney M, Krumholz HM. Achieving door-to-balloon times that meet quality guidelines: how do successful hospitals do it? J Am Coll Cardiol. 2005; 46: 1236–1241.
Grilli R, Minozzi S, Tinazzi A, Labianca R, Sheldon TA, Liberati A. Do specialists do it better? The impact of specialization on the processes and outcomes of care for cancer patients. Ann Oncol. 1998; 9: 365–374.
Grilli R. Specialization and cancer: words with too many meanings should be handled with care. Can Med Assoc J. 2001; 164: 210–211.
Hospitals with PPCI capability may choose to predominately offer this treatment modality to their patients with STEMI, or they may selectively offer PPCI or fibrinolytic therapy based on patient- and hospital-level factors. Whether a greater degree of hospital specialization with PPCI is related to better quality of care and outcomes in PPCI had not been previously reported. Using data from the NRMI-4 cohort, we found that greater PPCI specialization was associated with a lower relative risk of in-hospital mortality in patients treated with PPCI (adjusted relative risk comparing the highest and lowest quartiles, 0.64; P=0.006) but not in those treated with fibrinolytic therapy. Patients at hospitals in the higher quartiles of PPCI specialization also had shorter door-to-balloon times and more often met guideline-based recommendations for time to treatment. Importantly, these relations were independent of PPCI volume. Hospitals performing PPCI for their patients with STEMI may be able to improve clinical outcomes by more exclusively committing to PPCI as a reperfusion strategy. However, if greater PPCI specialization is not possible, innovative approaches may be needed to maximize the clinical benefit of reperfusion therapy.
Guest Editor for this article was Thomas Henry Lee, MD, SM, MSc.