Magid DJ, Calonge BN, Rumsfeld JS, Canto JG, Frederick PD, Every NR, Barron HV, for the National Registry of Myocardial Infarction 2 and 3 Investigators . Relation Between Hospital Primary Angioplasty Volume and Mortality for Patients With Acute MI Treated With Primary Angioplasty vs Thrombolytic Therapy. JAMA. 2000;284(24):3131–3138. doi:10.1001/jama.284.24.3131
Author Affiliations: Clinical Research Unit, Colorado Permanente Medical Group, Denver (Drs Magid and Calonge); Department of Preventive Medicine and Biometrics (Drs Magid and Calonge), Division of Emergency Medicine (Dr Magid), Department of Family Medicine (Dr Calonge), and Division of Cardiology (Dr Rumsfeld), University of Colorado Health Sciences Center, Denver; Denver VA Medical Center (Dr Rumsfeld); University of Alabama Medical Center, Birmingham (Dr Canto); University of Washington Cardiovascular Outcomes Research Center, Seattle(Dr Every and Mr Frederick); VA Puget Sound Healthcare System, Seattle, Wash (Dr Every); University of California at San Francisco, and Genentech Inc, South San Francisco, Calif (Dr Barron).
Context Institutional experience with primary angioplasty has been suggested
as a factor in selecting a reperfusion strategy for patients with acute myocardial
infarction (AMI). However, no large studies have directly compared outcomes
of primary angioplasty vs thrombolytic therapy as a function of institutional
Objective To compare outcomes among patients with AMI who were treated with primary
angioplasty vs thrombolytic therapy at hospitals with different volumes of
Design Retrospective cohort.
Setting A total of 446 acute care hospitals with 112 classified as low volume
(≤16 procedures), 223 as intermediate volume (17-48 procedures), and 111
as high volume (≥49 procedures) based on their annual primary angioplasty
Patients A total of 62 299 patients with AMI treated with primary angioplasty
or thrombolytic therapy from June 1, 1994, through July 31, 1999.
Main Outcome Measure In-hospital mortality.
Results Mortality was lower among patients who received primary angioplasty
compared with those who received thrombolysis at hospitals with intermediate
volumes (4.5% vs 5.9%; P<.001) and high volumes
(3.4% vs 5.4%; P<.001) of primary angioplasty.
At low-volume hospitals, there was no significant difference in mortality
between patients treated with primary angioplasty vs those treated with thrombolysis
(6.2% vs 5.9%; P = .58). Adjusting for differences
in demographic, medical history, clinical presentation, treatment, and hospital
characteristics did not significantly alter these findings.
Conclusions In this study, patients with AMI treated at hospitals with high or intermediate
volumes of primary angioplasty had lower mortality with primary angioplasty
than with thrombolysis, whereas patients with AMI treated at hospitals with
low angioplasty volumes had similar mortality outcomes with primary angioplasty
Reperfusion therapy with either primary angioplasty or thrombolytic
therapy reduces mortality for eligible patients with acute myocardial infarction
Several randomized clinical trials conducted at selected medical centers have
demonstrated lower mortality for patients with AMI who received primary angioplasty,
compared with those who received thrombolytic therapy.5- 8
In contrast, some large community-based observational studies found no survival
benefit for patients who received primary angioplasty, compared with those
who received thrombolytic therapy.9- 12
These divergent study results may reflect the fact that the randomized trials
were conducted at medical centers highly experienced in performing primary
angioplasty, whereas the observational studies were conducted at an unselected
group of hospitals having varying levels of experience with primary angioplasty.
The American College of Cardiology (ACC) and American Heart Association
(AHA) recommend that "primary angioplasty should be used as an alternative
to thrombolytic therapy only if performed in a timely fashion by individuals
skilled in the procedure and supported by experienced personnel in high-volume
centers."13,14 Despite the inherent
logic in the ACC/AHA guidelines, we are aware of no large studies that have
directly compared outcomes of primary angioplasty vs thrombolytic therapy
as a function of hospital primary angioplasty volume.
To address this issue, we examined the association between hospital-specific
primary angioplasty volume and short-term mortality for patients with AMI
treated with primary angioplasty vs thrombolytic therapy. We hypothesized
that the higher the volume of cases of primary angioplasty performed at an
institution, the lower the relative risk of death during hospitalization among
patients with AMI receiving primary angioplasty compared with similar patients
receiving thrombolytic therapy.
The National Registry of Myocardial Infarction (NRMI) is a voluntary,
prospective, observational database that collects data on patients admitted
with AMI throughout the United States. Characteristics of the NRMI registry,
data gathering procedures, and reliability have been previously described.15- 17 For this study, we
used data for patients enrolled in the registry from June 1, 1994, through
July 31, 1999.
Hospitals were eligible for this study if they performed 5 or more primary
angioplasty and 5 or more intravenous thrombolysis procedures per year. Prior
studies have noted that patients initially treated with thrombolytic therapy
are more likely to be subsequently transferred to another facility than patients
initially treated with primary angioplasty.10
To reduce the potential bias associated with differential patient transfer,
we excluded hospitals with patient transfer rates exceeding 15%. We also excluded
hospitals that did not regularly report data to the NRMI registry and those
that had participated in the registry for fewer than 6 months, because of
concerns regarding the completeness of data from these hospitals.
We classified hospitals into primary angioplasty volume quartiles based
on the annual number of procedures each had performed. Hospitals in the lowest
quartile were defined as the low-volume group (≤16 procedures per year),
whereas hospitals in the highest quartile were defined as the high-volume
group (≥49 procedures per year). Because the unadjusted and adjusted mortality
rates for both angioplasty and thrombolysis were similar for patients in the
middle 2 quartiles, and were distinctly different from those in the lowest
and highest quartiles, we combined the middle 2 quartiles into a single intermediate-volume
group (17-48 procedures per year).
The following clinical criteria were used to identify patients with
AMI who were optimal candidates for both thrombolytic therapy and primary
angioplasty: (1) arrival at the hospital within 12 hours of AMI symptom onset;
(2) initial electrocardiogram demonstrating ST-segment elevation or left bundle-branch
block; (3) absence of cardiogenic shock; and (4) no contraindications to thrombolytic
therapy.13,14 We excluded patients
who did not complete their hospital stay at a single hospital because data
on baseline characteristics, clinical outcomes, or both were not consistently
available for these patients.
The study population was divided into 2 cohorts based on whether the
patients received intravenous thrombolysis or primary angioplasty. Complete
data were available for more than 96% of patients for all baseline demographic,
medical history, clinical presentation, medical therapy, and hospital variables.
The primary outcome for this study was in-hospital mortality. Secondary
outcomes included nonfatal stroke (both hemorrhagic and thromboembolic), other
major bleeding following reperfusion therapy (defined as bleeding other than
intracranial that resulted in substantial hemodynamic compromise), and subsequent
revascularization during the index hospitalization (defined as coronary artery
bypass graft surgery or angioplasty that was performed for reasons other than
as the initial reperfusion strategy). Process of care measures included the
time to treatment, defined as the time from hospital arrival to the initiation
of thrombolytic therapy for patients receiving intravenous thrombolysis and
as the time from hospital arrival to first balloon inflation for patients
undergoing primary angioplasty.
For this study the unit of analysis was the individual patient. For
each patient we included a variable for the total number of primary angioplasty
procedures performed at the admitting hospital. We used the generalized estimating
equation (GEE) for risk modeling to adjust for the effects of within hospital
clustering.18,19 Separate models
were constructed for each of the primary angioplasty volume strata (low, intermediate,
and high). The outcome variable for all risk models was in-hospital mortality.
An indicator variable with thrombolytic therapy as the referent group was
included in the models to assess the relation of reperfusion modality with
hospital mortality. All patient and hospital variables listed in Table 1, Table 2, and Table 3
were included as covariates in the risk models. Odds ratios (ORs) and 95%
confidence intervals (CIs) were calculated for all covariates in the models.
We developed additional regression models to test for an interaction
between hospital primary angioplasty volume and reperfusion modality. The
initial model included terms for reperfusion modality, hospital-specific primary
angioplasty volume, and the interaction between reperfusion modality and primary
angioplasty volume. Subsequent models adjusted for all baseline patient and
hospital characteristics. In separate analyses, primary angioplasty volume
was coded as a categorical and as a continuous variable.
We performed several subgroup analyses. To evaluate potential selection
bias associated with different patient transfer rates, we repeated our analysis
assuming that all transferred patients survived. To test for possible referral
bias resulting from the influence of patient clinical status on the physician's
choice of reperfusion strategy, we performed an analysis that included only
patients classified as high risk using the definition of the Primary Angioplasty
in Myocardial Infarction (PAMI) trial, ie, those with either an anterior location
of infarct, age older than 70 years, or heart rate greater than 100/min.6 The relationship between primary angioplasty volume
and outcomes of reperfusion therapy also was assessed in patient cohorts stratified
by sex, age older than or younger than 65 years, and in the subset of patients
treated with primary angioplasty with coronary stent placement. Additional
subgroup analyses were conducted for patient groups stratified by total hospital
AMI volume (≤ or >300 AMI patients per year) and total hospital thrombolysis
volume (≤ or >65 patients per year).
Of the 1799 hospitals reporting data to NRMI during the study period,
we identified 616 hospitals that provided thrombolytic therapy and primary
angioplasty to 5 or more patients per year. We excluded hospitals that had
participated in the registry for fewer than 6 months (43), those that did
not regularly report data to the registry (91), and institutions with patient
transfer rates exceeding 15% (36). Overall, 446 hospitals met the criteria
for the study; there were 112 low-volume, 223 intermediate-volume, and 111
high-volume primary angioplasty hospitals.
A total of 488 758 patients with AMI were treated at the 446 eligible
hospitals during the 62-month study period. We excluded 187 141 patients
because they did not complete their hospital stay at a single hospital. Of
the remaining 301 617 patients, 87 873 were optimal candidates,
by our criteria, for thrombolytic therapy or primary angioplasty. A total
of 62 299 (70.9%) of these patients received reperfusion therapy. The
final study population included 21 973 in the primary angioplasty group
and 40 326 in the thrombolytic therapy group.
The thrombolytic therapy and primary angioplasty groups were similar
with regard to age, sex, and race/ethnicity (Table 1). Given the large patient sample size, most baseline comparisons
between the 2 cohorts were statistically different at the P<.01 level.20 Baseline characteristics
with relative differences of 10% or more between the 2 cohorts are reported
Patients in the group receiving thrombolytic therapy were more likely
to have a history of myocardial infarction (MI), heart failure, and coronary
artery bypass graft surgery, but less likely to have had a prior angioplasty
procedure. Patients treated with thrombolytic therapy arrived at the hospital
sooner after onset of symptoms and were less likely to present with an anterior
MI (Table 2).
Treatment with thrombolytic therapy was evenly distributed during the
day, evening, and night, whereas treatment with primary angioplasty was more
common during the daytime. Compared with patients who received primary angioplasty,
patients who received thrombolysis were more likely to have been given aspirin
and β-blockers and less likely to have received angiotensin-converting
enzyme inhibitors during the first 24 hours of hospitalization. Treatment
with primary angioplasty was more common in teaching hospitals and during
the later years of the study (Table 3).
For patients receiving care at low-volume primary angioplasty centers,
unadjusted in-hospital case fatality rates were similar for the primary angioplasty
and thrombolytic therapy groups (6.2% vs 5.9%, respectively; P = .58) (Figure 1). In the
separate GEE regression models that adjusted for all baseline demographic,
medical history, clinical presentation, medical therapy, and hospital characteristics
(Table 4), there was no association
between reperfusion strategy and in-hospital mortality in low-volume institutions
(OR, 1.06; 95% CI, 0.80-1.40).
At intermediate-volume primary angioplasty centers, patients treated
with primary angioplasty had significantly lower unadjusted mortality rates
during hospitalization than those treated with thrombolysis (4.5% vs 5.9%; P<.001). The survival advantage of primary angioplasty
compared with thrombolysis persisted after adjusting for all baseline patient
and hospital characteristics (OR, 0.65; 95% CI, 0.57-0.75).
At high-volume primary angioplasty centers, unadjusted in-hospital mortality
rates were also lower for patients treated with primary angioplasty than those
for patients treated with thrombolysis (3.4% vs 5.4%, P<.001). The association between treatment with primary angioplasty
and lower in-hospital mortality remained significant in a regression analysis
that adjusted for all patient and hospital characteristics (OR, 0.61; 95%
The difference in crude hospital mortality rates between the primary
angioplasty group and the thrombolysis group increased as hospital primary
angioplasty volume increased (Figure 1; P<.001 for the overall interaction between hospital
primary angioplasty volume and reperfusion modality). The interaction between
primary angioplasty volume and reperfusion modality was also significant in
analyses in which primary angioplasty was coded as a continuous variable (P<.01) and in analyses that adjusted for all baseline
patient and hospital characteristics (P<.01).
The time to treatment for patients who received thrombolytic therapy
(the time from hospital arrival to delivery of intravenous thrombolysis) was
shorter than the comparable time to treatment for patients who received primary
angioplasty (the time from hospital arrival to balloon inflation) (Table 5). Although the time to treatment
for thrombolytic therapy was relatively constant across hospital volume strata,
the time to treatment for primary angioplasty decreased as the hospital primary
angioplasty volume increased. After adding the time to treatment variable
to our regression model, the association between primary angioplasty volume
and mortality was slightly attenuated (approximately 7% of the mortality reduction
at higher-volume hospitals was explained by differences in time to reperfusion).
Overall, patients treated with thrombolytic therapy were more likely
to have a nonfatal stroke than patients treated with primary angioplasty (1.1%
vs 0.4%; P<.001). There was no significant difference
in the rates of major bleeding between the 2 treatment groups (4.0% in the
thrombolysis group vs 3.7% in the primary angioplasty group; P = .10). These findings were consistent across primary angioplasty
volume strata (Table 5) and persisted
after adjustment for all baseline patient and hospital characteristics.
The incidence of subsequent revascularization was higher for patients
treated with thrombolysis than for those treated with primary angioplasty
(overall incidence, 59.0% vs 15.0%; P<.001). For
patients treated with primary angioplasty, the incidence of subsequent revascularization
decreased as the hospital primary angioplasty volume increased (P<.001). These findings persisted after adjusting for all baseline
patient and hospital characteristics.
More patients treated with thrombolysis (3.6%) than those treated with
primary angioplasty (1.4%) were excluded from the primary analysis because
they were transferred from the initial hospital to a second hospital. To assess
the potential for bias resulting from differential patient transfer rates,
we repeated the analysis assuming that all transfer patients survived. In
this analysis, the adjusted risk of death among patients who received primary
angioplasty compared with that of those who received thrombolysis was similar
to the adjusted risk found in the primary analysis (low-volume centers: OR,
1.10, 95% CI, 0.87-1.38; intermediate-volume centers: OR, 0.66, 95% CI, 0.58-0.76;
high-volume centers: OR, 0.61, 95% CI, 0.52-0.73).
To test for selection bias in the physicians' choice of reperfusion
strategy, we also performed a subgroup analysis that included only high-risk
patients as defined in the PAMI trial. In this analysis, the adjusted risk
of death among patients who received primary angioplasty compared with that
of those who received thrombolysis was similar to the adjusted risk found
in the primary analysis (low-volume centers: OR, 1.03, 95% CI, 0.76-1.40;
intermediate-volume centers: OR, 0.64, 95% CI, 0.55-0.75; high-volume centers:
OR, 0.61, 95% CI, 0.51-0.74). The results of subgroup analyses stratified
by age, sex, hospital AMI volume, and hospital thrombolysis volume were similar
to the results of the primary analysis. The results of the subset analysis
of patients who received coronary stent placement showed an inverse association
between primary angioplasty volume and in-hospital mortality that was similar
to that found in the primary analysis.
In this study, we compared outcomes among patients with AMI treated
with primary angioplasty vs thrombolytic therapy at hospitals having varying
levels of experience with primary angioplasty. At intermediate- and high-volume
hospitals, we found that mortality was lower among patients who received primary
angioplasty compared with those who received thrombolytic therapy. In contrast,
the risk of death among patients treated with primary angioplasty and thrombolysis
at low-volume hospitals was similar. Adjusting for differences in patient
and clinical variables did not significantly alter these findings. Results
were also similar in age, sex, and high-risk subgroups. Patients treated with
primary angioplasty were less likely to have a nonfatal stroke or to undergo
subsequent revascularization than patients treated with thrombolysis. There
was no significant difference in rates of other major bleeding between the
By using the NRMI registry we were able to assess AMI outcomes at geographically
diverse hospitals with a wide range of primary angioplasty experience. An
additional strength was the availability of extensive clinical and demographic
data that allowed for adjustment of patient and hospital differences (including
overall hospital MI volume)21 that might influence
This study may provide an explanation for the discordant results of
previous studies comparing primary angioplasty with thrombolytic therapy.
Our finding of improved outcomes for primary angioplasty at higher-volume
centers is consistent with previous clinical trials, mostly conducted at experienced
However, our finding of no difference in outcomes by reperfusion strategy
at low-volume centers is consistent with previous community-based studies
that included institutions with low primary angioplasty volumes.9- 12
We noted a progressive decline in mortality associated with primary
angioplasty at hospitals with higher procedural volume. This finding supports
the ACC/AHA guideline recommendations and is consistent with prior studies
documenting an inverse association between hospital procedural volume and
outcomes following coronary artery bypass graft surgery, elective angioplasty,
and primary angioplasty.13,14,22- 32
Since primary angioplasty is a complex procedure, improved organizational
performance may contribute to superior outcomes at high-volume centers. Our
finding of shorter times between hospital arrival and balloon inflation at
higher-volume centers is consistent with this hypothesis. High-volume centers
also may experience more favorable outcomes because they have more experienced
operators or perform more total procedures (including elective angioplasty
procedures). However, since the NRMI registry does not contain data on physician
procedure volume or total hospital angioplasty volume, we were not able to
assess the degree to which these factors contributed to our findings.
Another possible explanation for our finding of lower mortality with
primary angioplasty at higher-volume centers might be that physicians at these
centers selected lower-risk patients for primary angioplasty. However, patients
who received primary angioplasty and thrombolytic therapy were similar with
regard to baseline factors. If anything, the primary angioplasty cohort may
have been at greater risk of death given their significantly larger proportion
of anterior MIs and longer time from onset of symptoms to hospital arrival.
The lower mortality observed for primary angioplasty at higher-volume centers
persisted after adjustment for important clinical predictors of mortality
and was also noted in subgroup analyses stratified by mortality risk.
Two prior studies found no relationship between primary angioplasty
volume and the relative outcomes of primary angioplasty and thrombolytic therapy.
In a subgroup analysis of patients in the Myocardial Infarction Triage Intervention
study, Every et al9 noted no difference in
outcome by reperfusion strategy for 995 patients treated at 3 high-volume
primary angioplasty centers compared with 1394 patients treated at 9 low-volume
centers. Similarly, Danchin et al,11 in a cohort
of 721 patients, found no difference in relative mortality between hospitals
that performed more than 4 primary angioplasty procedures per month and those
that performed 3 or fewer procedures per month. However, the relatively small
number of hospitals and patients in these studies limited the investigators'
ability to assess the influence of primary angioplasty volume on mortality.
Several potential limitations of our study should be acknowledged. Most
important, patients were not randomly assigned to undergo primary angioplasty
or receive thrombolytic therapy. To minimize patient selection bias we restricted
our analysis to patients who were optimal candidates for both thrombolytic
therapy and primary angioplasty. Although the results did not significantly
change after adjustment for known baseline differences in the 2 study cohorts,
there may be residual bias from unmeasured differences between the 2 groups.
In addition, the NRMI registry does not include long-term follow-up;
therefore, we cannot be certain that the mortality differences observed in
this study will persist. In the Gusto IIB trial,33
an early mortality benefit of primary angioplasty over thrombolysis was not
evident at 6-month follow-up. However, in a more recent study, Zijlstra et
al34 reported a survival benefit among patients
treated with primary angioplasty that was sustained for 5 years.
Most US medical centers do not have the facilities and staff to provide
primary angioplasty.13,14 Our
findings do not support a policy of routine patient transfer from hospitals
offering only thrombolysis to high-volume interventional hospitals, because
potential benefits from primary angioplasty may well be offset by delays in
achieving reperfusion.13,14 Alternatively,
emergency medical service systems may consider out-of-hospital triage of eligible
AMI patients to higher-volume primary angioplasty centers if triage can be
accomplished without significant delays to reperfusion. Since only 71% of
eligible AMI patients received reperfusion therapy in this study, efforts
also should be made to increase reperfusion rates for eligible patients.
Our study may be relevant for hospitals contemplating offering primary
angioplasty as a reperfusion option. Our findings suggest that, at hospitals
that perform higher volumes of primary angioplasty procedures, patients with
AMI who are eligible for reperfusion may experience reduced mortality from
the use of primary angioplasty compared with thrombolysis. In contrast, hospitals
that perform low volumes of primary angioplasty procedures should weigh the
economic and logistical burdens of providing primary angioplasty for patients
with AMI against our observation that, at low-volume centers, primary angioplasty
offered no significant reduction in mortality compared with thrombolytic therapy.