Hospitalization Costs of Primary Stenting Versus Thrombolysis in Acute Myocardial Infarction
Cost Analysis of the Canadian STAT Study
Background— We previously showed that primary stenting was more effective than accelerated tPA in reducing the 6-month composite of death, reinfarction, stroke, or repeat revascularization for ischemia. This study looks at the hospitalization costs of primary stenting compared with accelerated tPA.
Methods and Results— Initial and 6-month hospitalization costs were computed for all patients randomly assigned to primary stenting (n=62) or accelerated tPA (n=61) in the Stenting versus Thrombolysis in Acute myocardial infarction Trial (STAT). Costs and resource usage were collected in detail for each patient. Physician fees were obtained directly from billings to the Ontario Health Insurance Plan. The length of initial hospitalization was 6.7±11.3 days in the stent group and 8.7±6.7 days in the tPA group (P<0.001). Total hospitalization days at 6 months were 8.3±13 days in the stent group and 12.1±14.0 days in the tPA group (P=0.001). Hospitalization costs were less in the stent group for the initial hospitalization, $6354±6382 versus $7893±4429 (P=0.001), and at 6 months, $7100±7111 versus $9559±6933 (P=0.001).
Conclusions— In centers in which facilities and experienced interventionists are available, primary stenting is less costly and more effective than thrombolysis.
Received November 14, 2002; de novo received June 30, 2003; revision received August 22, 2003; accepted August 26, 2003.
Several randomized trials have reported that primary percutaneous coronary intervention (PCI), compared with thrombolysis, yields superior clinical outcomes in patients with ST-segment elevation myocardial infarction (STEMI).1–9 A recent meta-analysis showed that PCI was better than thrombolysis in reducing death, nonfatal reinfarction, and stroke at both short-term and long-term follow-up.10 Primary angioplasty with stent implantation (primary stenting) has clinical benefits beyond those of primary angioplasty alone.11–13 Although we recently demonstrated a clinical benefit of primary stenting compared with thrombolysis in the Stenting versus Thrombolysis in Acute myocardial infarction Trial (STAT),9 concerns about cost may reduce the enthusiasm for this strategy. Randomized studies comparing the costs of PCI with thrombolysis are few.14–16 Only 2 studies measured costs beyond the initial hospitalization,14,15 and none assessed the impact of primary stenting. This study therefore compares 6-month hospitalization costs associated with primary stenting with those for accelerated tissue plasminogen activator (tPA) in STAT.
The methodology and results of the STAT study have been described previously.9 Briefly, 123 patients presenting with STEMI of ≤12 hours duration to a single tertiary cardiovascular health center were randomized to primary stenting, n=62, or accelerated tPA, n=61. Enrollment began in August 1997 and ended in June 1999. Patients with cardiogenic shock, contraindications to thrombolysis, PCI within 6 months, previous stenting of the infarct-related artery, previous bypass surgery (CABG), or inability to provide informed consent were excluded. The primary end point was the 6-month composite of death, reinfarction, stroke, or repeat target-vessel revascularization for ischemia. Recurrent unstable ischemia was defined as recurrent chest pain at rest associated with new ST-segment or T-wave changes, hypotension, or pulmonary edema. An institutional review board approved the study, and written consent was obtained from all participating patients.
Patients randomized to primary stenting underwent coronary angiography as soon as possible, followed by stenting, according to protocol guidelines. Patients randomized to thrombolysis received accelerated tPA and heparin as reported previously.17 Routine coronary angiography was discouraged, and risk stratification was performed according to published guidelines.18
In Ontario, the province allocates a budget to provide services and purchase medical and surgical supplies. Costs reported in this study are the direct hospitalization costs of patient care and do not include indirect costs such as depreciation of equipment, electricity, maintenance, and medical records. All costs are expressed in US dollars using the 1999 exchange rate for Canadian dollars ($US/$CDN=0.67).
Detailed resource usage was recorded for each patient during the study period. The actual cost of cardiac catheterization was determined by keeping a detailed inventory of balloons, stents, and catheters used during each PCI, and the associated costs were added to an average cost per patient for basic medical and surgical supplies. The cost of an ACS Multilink stent (Guidant, Advanced Cardiovascular Systems) was $637, and the cost of abciximab was $1005 per procedure. Labor costs were based on the actual catheterization time per patient and were calculated from the average hourly rate for technicians, nurses, porters, and clerks. Labor costs during off-hours were calculated from appropriate overtime rates.
The actual costs incurred for thrombolysis, laboratory medicine, electrocardiography, nuclear medicine, and radiology were obtained from direct costs to the hospital. Costs for medical supplies and labor associated with CABG, ie, cardiac operating room and perfusion, were calculated by use of the average cost per case for our hospital during the study period.
To compute the costs associated with supplies and labor in the coronary care unit, intermediate care unit, recovery room, surgical intensive care unit, and medical and surgical wards, we multiplied the average daily cost in 1999 for each unit by the length of stay in that unit. Respiratory therapy costs were calculated according to patient-days on the ventilator. Cost of meals was based on an average daily cost per patient. Emergency room labor costs were calculated from the actual number of hours each patient spent there. Hospitalization costs included physician fees obtained from actual billings to the Ontario Health Insurance Plan. Technical and professional fees for performing stress tests and echocardiograms were included in the cardiologists’ Ontario Health Insurance Plan billings.
Comparisons were performed on an intention-to-treat basis. Categorical variables were compared by χ2 analysis or the Fisher exact test and continuous variables by Student’s t test. The primary analysis for costs included professional fees. The distributions of costs and hospitalization days were skewed and were log-transformed to normalize their distribution. Data presented in tables are expressed as mean±SD of untransformed data. We identified univariate associations between clinical variables and hospitalization costs using linear regression analysis. Those with a probability value of P≤0.1 were entered into a stepwise regression analysis to establish whether they were independently associated with cost. To ascertain the sensitivity of cost comparisons to clinical events, analyses were performed with and without patients who had experienced these events.
Bootstrapping was done using untransformed costs to determine confidence intervals for mean cost differences between the 2 treatment groups and to relate costs to clinical outcomes.19 This was done by randomly drawing 1000 replications with a sample size of 123 with replacement from the study population. Mean cost differences between groups and odds ratios for clinical outcomes were calculated for each replication, and the bootstrapped pairs were represented graphically on a cost-effectiveness plane.
Analyses were done with Systat version 10.2 (Systat Inc).
Groups were well matched for baseline characteristics.9 In the stent group, the mean age was 60.9±2.3 versus 60.2±11.3 years in the tPA group, diabetes mellitus was present in 21.0% versus 23.0% of patients, and anterior STEMI was noted in 41.9% versus 37.7%. In patients assigned to primary stenting, 61 (98.4%) had immediate angiography performed, 1 received tPA, 50 (81%) had successful deployment of stents, and 2 had PCI with balloon alone. Abciximab was used in 12 patients (19.4%) treated with primary stenting. All patients assigned to thrombolysis received accelerated tPA.
Follow-up was available in all surviving patients. The primary end point at 6 months was reduced in the stent group: 24.2% versus 55.7% (P<0.001), mainly because of less repeat target-vessel revascularization for ischemia: 14.5% versus 49.2%, P<0.001 (Figure 1). Recurrent unstable ischemia was also lower in the stent group, 9.7% versus 26.2% (P=0.019). Between 6 months and 2 years, there was 1 accidental death in the stent group, 2 reinfarctions, 2 PCIs, and 1 CABG in each group. At 2 years, the composite end point remained lower in the stent group, 32.2% versus 57.3% in the tPA group (P=0.005).
Per patient, costs for the initial hospitalization were lower in the stent group than in the tPA group, $6354±6382 versus $7893±4429 (P=0.001), and at 6 months, $7100±7111 versus $9559±6933 (P=0.001) (Table 1).
There were fewer hospitalization days in the stent group: 6.7±11.3 versus 8.7±6.7 days for the initial hospitalization and 8.3±13.0 versus 12.1±14.0 at 6 months (Table 2). During the initial hospitalization, stress testing was performed in 3.2% of the stent group and 41.0% of the tPA group. The frequency of cardiac catheterization and intervention is shown in Table 3. During the initial hospitalization, nonprotocol catheterization was performed in 9.7% of the stent group compared with 63.9% in the tPA group, and at 6 months, in 22.6% versus 72.1%. For patients in the thrombolytic group needing catheterization during the initial hospitalization, the median time from admission to catheterization was 4 days. Catheterization procedures averaged 86±36 minutes in the stent group and 76±31 minutes in the thrombolytic group.
The results of the univariate and multivariate analysis appear in Table 4. Days in hospital were driven by clinical events and were highly correlated with 6-month costs (r=0.90, P<0.001). Because of this high degree of collinearity, we excluded days in hospital from our final models. Regression analyses including baseline variables only revealed primary stenting to be an independent predictor of cost. Subsequent analysis showed that CABG, ischemia-driven PCI, recurrent unstable ischemia, major bleed, and stroke were independent correlates of cost. When these were added to the multivariate models, randomization group was no longer statistically significant, indicating that differences in clinical outcomes are the predominant reason why primary stenting was cheaper.
Analyses were repeated excluding patients experiencing CABG, major bleed, and stroke, outcomes associated with increased days in hospital; primary stenting was still significantly cheaper (Table 5). At 6 months, costs associated with CABG averaged $21 460; with major bleed, $21 660; and with stroke, $20 643.
Figure 2 shows the results of bootstrapping. Primary stenting was cheaper than thrombolysis in 96.4% of replications. For the composite outcome of death, reinfarction, or stroke, primary stenting was better than thrombolysis (odds ratio >1.0) in 88.1% of replications. For death, reinfarction, stroke, or recurrent unstable ischemia, primary stenting was better than thrombolysis in 99.9% of replications.
Our study is the first to show that primary stenting is cost saving compared with thrombolysis. At 6 months, hospitalization costs with primary stenting averaged $2459 less per patient. Initial length of hospital stay and cumulative 6-month hospitalization days were also reduced. We previously found that primary stenting was more effective than accelerated tPA in reducing the 6-month composite of death, reinfarction, stroke, or repeat revascularization for ischemia.9 Taken together, these data indicate that primary stenting is highly cost-effective.
Reports from randomized trials on the economics of primary PCI versus thrombolysis are few.14–16 In the Zwolle trial, combined hospital and outpatient costs were similar at 12 months, $17 306 with primary PCI versus $16 681 with streptokinase.14 For US patients enrolled in the Primary Angioplasty in Acute Myocardial Infarction (PAMI)-1 trial,16 combined charges and professional fees for the initial hospitalization favored PCI but were not significantly different, $27 653 versus $30 227. The Mayo Clinic study calculating costs at 80% of charges reported a trend toward a lower 6-month cost with PCI compared with tPA, $17 292 versus $24 129.15 Because stents were not yet available, these comparisons were performed against balloon angioplasty.
Two randomized trials evaluated the cost-effectiveness of primary stenting versus primary balloon angioplasty.20,21 The 1-year costs in StentPAMI20 were higher with stenting, $20 571 versus $19 595. The Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial found that the 1-year costs were $18 021 for primary PCI alone, $19 331 for PCI plus abciximab, $18 278 for stenting alone, and $19 466 for stenting plus abciximab (P=0.02 for pooled abciximab, P=0.75 for stenting).21 Primary stenting was deemed cost-effective because it was associated with superior clinical outcomes and the use of abciximab reasonably cost-effective because it facilitated earlier hospital discharge. Extrapolating from these results, primary stenting would still have been cost saving if we had used abciximab. The additional cost of $1005 per patient is less than the difference between groups in per-patient costs ($2459) and may have been further offset by reduced length of stay.
In our study, primary stenting reduced clinical events that prolong hospitalization days, which correlated highly with costs (r=0.90). However, the results of bootstrapping analysis and the observation that primary stenting remained less costly when patients experiencing CABG, major bleed, and stroke were excluded make it extremely unlikely that our results were due to random differences in clinical events.
Our results complement those reported in the PAMI-1 trial16 and extend them to include the influence of stents. The initial length of stay in our stent group averaged almost 1 day less than the PAMI-1 angioplasty group (6.7 versus 7.6 days), whereas the length of stay for the thrombolysis group was slightly longer (8.7 versus 8.4 days). Our rate of catheterization in the thrombolysis group (63.9%) during the initial hospitalization was also comparable to that reported in PAMI-1.5
There are several reasons why our costs seem to be lower than in previous studies.15,16 First, we report actual costs and not charges. Studies that use the latter as a proxy for cost typically report higher figures.22 Second, we did not include indirect costs. However, indirect costs would also be expected to be less with primary stenting, because this strategy reduced hospitalization days. Thus, our analysis is proportionally correct but has probably underestimated the differences in absolute costs between the 2 strategies. Our results also may not apply to low-volume centers, in which fixed costs would make up a larger proportion of total costs. However, in high-volume PCI centers, it is not likely that the costing method would bias a comparison between primary stenting and thrombolysis, and therefore our results should be applicable in these centers.
The cost of a stent used in this study, $637, may be lower than in the United States and other countries. Because patients randomized to stenting required ≈0.5 additional stents per patient during the 6-month study period, doubling the stent price would add only $300 to $400 per patient; therefore, stenting would still remain at least $2000 less expensive than thrombolysis.
A cost-effectiveness analysis of streptokinase versus tPA reported that tPA costs $2000 more per patient.23 Because this difference approximates the cost saving in our study, our results may not apply to centers using streptokinase. However, any cost-saving with streptokinase would probably be offset by worse clinical outcomes, because streptokinase is clinically less effective.17
There are potential limitations of this study that need to be considered. First, the sample size was limited. However, this comparison was conducted in the context of a prospective randomized trial with complete follow-up, and our analyses revealed that the results were unlikely to be caused by instability in the rates of critical clinical events.
Second, the study was conducted in a single Canadian center, and it might be argued that the cost determination in this study applies only to centers within the Canadian healthcare system. However, our clinical decisions concerning acute coronary revascularizations were based on the same principles that guide therapy throughout the developed world, and our effectiveness results are in keeping with other randomized trials comparing PCI to thrombolysis. In other countries, the details of costing would be expected to differ, but the relative cost difference between primary stenting and thrombolysis will probably be maintained.
Third, a determination of quality-adjusted life years was not performed. This would have been important had we found primary stenting to be more expensive than thrombolysis. Because primary stenting was cheaper, this is not required to establish cost-effectiveness.
Finally, our study did not address the costs associated with training additional interventionists and building the new cardiac facilities that would be needed to implement a strategy of primary stenting for all STEMI patients. This is an important issue for policy makers that is beyond the scope of this article.
Compared with accelerated tPA, primary stenting is associated with significantly lower hospitalization costs than accelerated tPA. It is also more effective, making it a dominant strategy in centers in which facilities and experienced interventionists are available.
This work was supported by an unrestricted grant from Guidant, Advanced Cardiovascular Systems Inc, Temecula, Calif. We are indebted to Sheila Ryan for her excellent administrative assistance.
A clinical trial comparing primary coronary angioplasty with tissue plasminogen activator for acute myocardial infarction. The Global Use of Strategies to Open Occluded Coronary Arteries in Acute Coronary Syndromes (GUSTO IIb) Angioplasty Substudy Investigators N Engl J Med. 1997; 336: 1621–1628.
Widimsky P, Groch L, Zelizko M, et al. Multicentre randomized trial comparing transport to primary angioplasty vs immediate thrombolysis vs combined strategy for patients with acute myocardial infarction presenting to a community hospital without a catheterization laboratory. The PRAGUE Study. Eur Heart J. 2000; 21: 823–831.
Suryapranata H, van’t Hof AW, Hoorntje JC, et al. Randomized comparison of coronary stenting with balloon angioplasty in selected patients with acute myocardial infarction. Circulation. 1998; 97: 2502–2505.
Gibbons RJ, Holmes DR, Reeder GS, et al. Immediate angioplasty compared with the administration of a thrombolytic agent followed by conservative treatment for myocardial infarction. The Mayo Coronary Care Unit and Catheterization Laboratory Groups. N Engl J Med. 1993; 328: 685–691.
Stone GW, Grines CL, Rothbaum D, et al. Analysis of the relative costs and effectiveness of primary angioplasty versus tissue-type plasminogen activator: the Primary Angioplasty in Myocardial Infarction (PAMI) Trial. The PAMI Trial Investigators. J Am Coll Cardiol. 1997; 29: 901–907.
O’Brien BJ, Briggs AH. Analysis of uncertainty in health care cost-effectiveness studies: an introduction to statistical issues and methods. Stat Methods Med Res. 2002; 11: 455–468.
Cohen DJ, Taira DA, Berezin R, et al. Cost-effectiveness of coronary stenting in acute myocardial infarction: results from the Stent Primary Angioplasty in Myocardial Infarction (StentPAMI) Trial. Circulation. 2001; 104: 3039–3045.
Bakhai A, Stone GW, Murphy SA, et al. Cost-effectiveness of coronary stenting and abciximab for patients with AMI: results from the CADILLAC Trial. Circulation. 2002; 106 (suppl II): II–518.Abstract.