- Top of page
Objective: Although the benefits of coronary stenting have been demonstrated in several large clinical trials, controversy remains as to whether stenting results in long-term cost savings compared to percutaneous transluminal coronary angioplasty (PTCA). The objective of this study was to evaluate the resource use and cost (Medicare payment) of PTCA versus bare stent in actual practice over a 2-year period.
Methods: The data for this study came from the 1996 through 1998 Standard Analytic Files that contain 5% of Medicare claims. The rates of repeat revascularization procedures and hospitalizations were reported at 1 and 2 years. Costs associated with inpatient admission, outpatient procedures, physician services, skilled nursing facility admissions, and home health-care services were included to perform a comprehensive assessment. Regression analysis was performed to test for cost differences controlling for case-mix variation between the patient groups.
Results: The selection process yielded 3782 PTCA patients and 2690 stent patients for analysis. The rate of revascularization was 26.7% for the PTCA group and 22.2% for the stent group at 2 years. The mean total cost for the initial procedure was $13,724 for PTCA and $15,021 for stenting. At 2 years, the total cumulative cost was $32,654 for the PTCA group and $32,102 for the stent group, a difference that was not statistically significant.
Conclusion: Although the difference in the rate of repeat revascularization procedures between PTCA and stenting is not as large as those reported in clinical trials, bare stents are cost-neutral when compared to PTCA for the Medicare population.
- Top of page
Coronary stent placements have grown rapidly over the past decade, and currently over 80% of all angioplasties performed in the United States involve stent placement. Although the benefits of coronary stenting have been demonstrated in several large clinical trials, controversy remains as to whether stenting results in long-term cost savings compared to percutaneous transluminal coronary angioplasty (PTCA). A search of the published literature revealed studies that showed stenting and PTCA to have similar long-term costs [1–3] and those that conclude that stenting remained more expensive [4–6].
Proponents of coronary stenting have argued that although stenting is initially more expensive, the reduced need for repeat procedures produces comparable cost between stent and PTCA over the long term. Therefore, the lower cost after discharge makes up for the higher initial procedural cost. Several studies based on clinical trials have reported on the reduced need for repeat revascularization rates associated with stent placement [7–9], which lends support for lower cost among stent patients in the period after discharge compared to PTCA patients. These findings have been questioned by some clinicians, Topol, for example , and recent studies analyzing data from clinical practice setting concluded that there were no significant differences in revascularization rates between stent placement and PTCA [11,12]. These authors conclude that stent placement results in higher overall cost compared to PTCA and question the cost-effectiveness of stent placement.
Therefore, although the effectiveness and cost of stent placements have been extensively studied in clinical trials, there still remains much debate over the long-term cost of coronary stent placement in everyday practice setting. Previous studies that have evaluated stenting in clinical practice have had short follow-up and therefore may not have captured the entire cost differential associated with the procedures in the period after discharge [11–13]. In addition, no study to date has performed an assessment including costs of all health-care services provided to these patients. Despite this lack of cost assessment, coronary stenting has been adopted rapidly with over 70% of percutaneous coronary intervention involving stent placement. The cost-effectiveness of this practice needs to be evaluated and with the imminent launch of coated stents such an assessment is even more critical. Coated stents have shown to dramatically reduce restenosis rates in recent clinical trials but are expected to cost significantly more than bare stents. Coated stents are projected to cannibalize the market for bare stents when introduced and therefore it is important to evaluate the cost-effectiveness of current practice, PTCA versus bare stents, to compare coated stents against the most cost-effective option.
The objective of this study was to evaluate the resource use and cost of PTCA versus bare stent in actual practice over a 2-year period and discuss implications for coated stents. The analysis was performed from the perspective of the largest payer for health-care services in the United States, the Center for Medicare & Medicaid Services (CMS). Cost to the Medicare program, therefore, payment made to hospitals, was analyzed and the focus of the analysis was on the elderly population eligible for Medicare, those 65 years and older. Health-care claims were analyzed to obtain resource use and cost information from normal practice rather than clinical trial setting. Costs associated with inpatient admission, outpatient procedures, physician services, skilled nursing facility admissions, and home health-care services were included to perform a comprehensive assessment.
- Top of page
The data for this study came from the 1996 through 1998 Standard Analytic Files created by the CMS. These files contain a 5% nationwide random sample of Medicare claims for inpatient, outpatient, physician office, skilled nursing facility, and home health-care services. We excluded patients who were less than 65 years old or who lived outside the United States. In addition, to ensure complete record of all services, we excluded patients who were enrolled in an HMO, who did not have both Part A and Part B coverage, and for whom Medicare was not the primary payer. When transfer occurred between hospitals, the records where combined to indicate a single admission to avoid duplication. A programmer performed all data manipulation in SAS with significant experience in Medicare claims data analyses. The data were manually reviewed after each step in the analytic file construction process to ensure accuracy in the programming performed.
The International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes were used to identify angioplasty (36.01, 36.02, and 36.05) and stent (36.01, 36.02, and 36.05) procedures that were performed in 1996. Data were abstracted for these patients for a 2-year period. Because we did not always have information on date of service, the 2-year follow-up period was defined as eight consecutive quarters starting with the quarter in which the index admission occurred. To ensure that patients were continuously enrolled in fee-for-service plans, only those with at least one contact with the medical system in 1997 and 1998 were included in the analysis. A very small proportion, 2% of stenting and 3% of the angioplasties, occurred in the outpatient setting. Patients with these outpatient procedures were eliminated from the study sample, partly to avoid the uncertainty on whether these were truly outpatient or whether they lacked full data.
To compare case-mix differences between the two groups, the Charlson index adapted for use with administrative databases was calculated [14,15]. The Charlson index is a summation of weighted comorbid conditions demonstrated to be strong predictors of cardiovascular-related morbidity and mortality [16,17]. The variables contained in the Charlson index and their weights (given in parenthesis) are as follows: peripheral vascular disease (1); chronic lung disease (1); dementia (1); chronic liver disease (1); peptic ulcer disease (1); diabetes mellitus with no sequelae (1); diabetes mellitus with sequelae (2); renal failure (2); leukemia, lymphoma, or solid cancer (2); liver disease with sequelae (3); and metastatic cancer or multiple cancers (6). The presence of these conditions was identified for each individual during the index admission and over the 2-year follow-up period. In addition to the Charlson index, the proportion of patients admitted with myocardial infarction (MI) and those who underwent multivessel procedures was abstracted. Demographic information, including age, sex, and race, was summarized for patients in both groups.
The short-term resource use parameters analyzed are same-admission coronary artery bypass graft (CABG), index admission length of stay, and discharge to skilled nursing facility (SNF) or home health-care services. The rate of repeat revascularization procedures (PTCA, stent, CABG) is reported at 1- and 2-year follow-up for the PTCA and stent groups. The percentage of individuals with hospitalizations for MI and other reasons, outpatient visits, physician office visits, SNF stays, and home health-care services usage is also reported. As appropriate, the average number of visits and the length of stay are provided.
Index admission and 2-year cumulative costs were also estimated. Medicare payments are used as a proxy for cost and the estimates provided, therefore, reflect the cost to the Medicare program. In 1996, the Medicare inpatient payments for PTCA and stent placement were the same, and payment changes to reflect differences in resource use between the two procedures were not implemented until 1998. In 1998, the average reimbursement for stent placement was $1259 more than for PTCA. We increased the 1996 Medicare reimbursement to hospitals for stent placement by $1219 (1998 difference adjusted for inflation using hospital services component of Consumer Price Index) to reflect the higher cost associated with the stent procedure. Inpatient, outpatient, physician, skilled nursing facility, home health-care services, and total costs were estimated.
All case-mix, resource use, and cost measures generated were tested for statistically significant differences between the PTCA and stent group using either a Student's t test or chi-square test as appropriate. Statistical significance at the 5% level is generally reported, unless otherwise noted. Regression analysis was performed to test for differences in 2-year revascularization rate and cost controlling for patient case mix. Logistic regression was estimated to assess differences in revascularization, and ordinary least squares regression was utilized to study impact on cumulative 2-year cost. Log cost was also attempted to identify whether this provided a better fit. The independent variables were age, sex, race, Charlson index, MI at index admission, and treatment group (stent/PTCA).
- Top of page
The selection process yielded 3782 PTCA patients and 2690 stent patients for analysis. A comparison of the patient characteristics and risk factors are provided in Table 1. There are some differences in age and race across the groups. The PTCA patients were older (43.1% vs. 39.2% were 74 years or older) and more likely to be nonwhite (7.8% vs. 5.7%). It should be noted that although these differences are statistically significant, the magnitude of the differences is not large. PTCA cases also had a significantly greater proportion of MIs (30.5% vs. 25.1%) and diabetes (31.5% vs. 28.1%). On the other hand, stent patients had significantly greater number of cases with unstable angina (44.0% vs. 48.9%). Stent patients also had a larger number of multivessel procedures (11.7% vs. 13.2%), but this was not a statistically significant result at the 5% level; significant at the 10% level. There were no significant differences in the proportion with peripheral vascular disease or renal disease and in the overall comorbidity as measured by the Charlson index.
Table 1. Patient characteristics and risk factors
| ||PTCA (n= 3782)||Stent (n= 2690)|
|Sex (% women)||44.1||41.9|
|Multivessel procedure (%)||11.7||13.2|
|Number of selected comorbid conditions|
| Unstable angina*||44.0||48.9|
| Peripheral vascular disease||12.2||11.7|
| Renal disease||5.1||5.2|
|Charlson index (number of conditions)†|
| 4 or more||13.9||12.6|
Resource use indicators for the acute phase, or index admission, are reported in Table 2. Same-admission CABG, not controlling for patient differences, is much higher for the PTCA group than the stent group. Overall, 3.4% of patients in the PTCA group had a CABG performed during the index admission compared to 1.3% in the stent group. The PTCA group had a significantly higher rate of in-hospital mortality compared to the stent group (3.3% vs. 1.8%). Although a higher proportion of PTCA patients were discharged to SNFs (3.2% vs. 2.5%), this is not a statistically significant difference. Both groups have the same proportion of discharges to home health-care services (8.5%) and no difference in average index admission length of stay of 5 days.
Table 2. Acute-phase outcomes and resource use parameters
| ||PTCA (n= 3782)||Stent (n= 2690)|
|Same-admission CABG (%)*||3.4||1.3|
|In-hospital mortality (%)*||3.3||1.8|
|Discharge to SNF (%)||3.2||2.5|
|Discharge to home health-care services (%)||8.5||8.5|
|Length of stay (days)||5.0 (4.4)||5.0 (4.2)|
The 1- and 2-year resource use parameters shown in Table 3 reveal several differences between the PTCA and the stent groups. Any revascularization, which includes PTCA, stent placement, or CABG, at 2 years was 26.7% for the PTCA group and 22.2% for the stent group. This difference is statistically significant and the gap between PTCA and stent groups increased from 2% at 1 year to 4.5% at 2 years. The difference between the two groups is largely driven by the higher rate of CABG surgery among the PTCA patients compared to the stented patients. At the 2-years follow-up, 10.3% in the PTCA group had CABG surgery versus 6.9% in the stent group. In addition, in the 2-year postprocedure period, stent patients had a slightly lower rate of PTCA (11.0% vs. 10.4%) and stent placement (9.4% vs. 8.4%).
Table 3. Long-term resource use
| ||1 year||2 years|
|Any revascularization (%)*†||18.3||16.3||26.7||22.2|
| MI hospitalization (%)||3.8||3.8||5.6||5.5|
| Percentage with any admission||38.0||41.3||58.4||56.1|
| Average length of stay (days)*§||4.0||4.1||6.8||5.9|
| Percentage with outpatient visit||64.0||70.7||92.1||91.7|
| Number of outpatient visits§||4.3||5.0||7.7||7.6|
| Percentage with SNF use*||4.2||3.9||7.0||5.7|
| SNF length of stay (days)*§||0.9||0.8||1.6||1.2|
|Home health-care services|
| Percentage using home health-care services*||17.6||17.7||27.1||24.2|
| Number of home health-care visits*§||0.8||0.8||1.5||1.1|
The rate of hospital admissions over the 2-year follow-up, whether MI-related or an admission for any condition, was very similar between the two groups. The average length of stay, however, was higher for the PTCA group compared to the stent group (6.8 days vs. 5.9 days), and this difference was statistically significant at the 5% level. Overall, about half of the patients with hospitalizations, a total of 58.4% for PTCA group and 56.1% for stent group, were admitted either for a revascularization procedure or as a result of a MI. There were no differences in either the rate or the number of outpatient clinic visits. Nevertheless, PTCA patients used more SNF and home health-care services. Among PTCA patients, the rate of SNF and home health-care services usage was 7.0 and 27.1%, respectively, and for stent patients it was 5.7 and 24.2%, respectively.
As expected, PTCA patients had on average lower cost associated with the index admission than stent patients. The mean total cost for the PTCA group was $13,724 and for the stent group it was $15,021, a statistically significant difference (Table 4). This difference does not persist in the follow-up period and the 2-year total cumulative cost was $32,654 for the PTCA group compared to $32,102 for the stent group. Cost of inpatient services makes up a large proportion, over 70% of the total 2-year cost. The cost of outpatient services, SNF stays, and home health-care services account for 10%, or approximately $3500, of the total cumulative cost. Although there were statistically significant differences in SNF and home health-care usage, these differences were not reflected in costs related to these services between the two groups. Significant differences were found in cost of outpatient care and physician services, although the magnitude of these differences was small, approximately $250 in both cases.
Table 4. Index admission and 2-year cost *
| ||Index admission||Follow-up period||Two-year cumulative†|
|Outpatient??||—|| $2,247|| $2,001 (4,375)|| $2,247 (3,506)|| $2,001 (4,375)||(3,506)|
|Physician§??¶|| $1,996|| $2,096|| $4,706|| $4,365|| $6,702||$6,461|
|SNF||—||—|| $497 (2,649)|| $389 (2,387)|| $497 (2,649)|| $389 (2,387)|
|Home health care||—||—|| $1,078 (4,151)|| $948 (4,529)|| $1,078 (4,151)|| $948 (4,529)|
Multivariate analyses of differences in cumulative 2-year revascularization rate and cost of PTCA versus stenting are presented in Table 5. Patients with stent placement were significantly less likely to require revascularization compared to PTCA patients after controlling for case-mix differences, with an odds ratio of 0.77. Although age was not consistently significant, patients 65 to 74 years were more likely to undergo repeat revascularization procedures that those > 80 years. White patients were about 1.5 times more likely to have a revascularization procedure in the 2-year follow-up period than nonwhite persons. Revascularization rates increased with the number of comorbidities. Patients with four or more clinical conditions as measures by the Charlson index were 1.4 times more likely to have a procedure than those with any conditions. Sex and acute MI (AMI) at admission did not impact repeat revascularization rate.
Table 5. Differences in 2-year revascularization rate and cost controlling for risk factors
|Variables||Two-year cumulative revascularization rate*||Two-year cumulative cost†|
|Parameter estimate (SE)||Odds ratio||Chi-square||Parameter estimate (SE)||t statistic|
|Stent (vs. PTCA)||−0.26||0.77|| 19.01‡|| −$84||−0.14|
| (0.06)|| || || (583)|| |
|Age (vs. those >80 years)|
| 65–69‡|| 0.39||1.48|| 17.41‡||−$2,460||−2.75*|
| (0.09)|| || || (895)|| |
| 70–74‡|| 0.35||1.42|| 14.95‡||−$2,309||−2.64*|
| (0.09)|| || || (874)|| |
| 75–79§|| 0.11||1.12|| 1.42||−$1,883||−2.08§|
| (0.10)|| || || (907)|| |
|Female (vs. male)‡|| 0.03||1.03|| 0.28||−$2,274||−3.9‡|
| (0.06)|| || || (583)|| |
|White (vs. other races)‡|| 0.44||1.56|| 12.6*||−$3,454||−3.05‡|
| (0.13)|| || ||(1,134)|| |
|Charlson index (vs. those without conditions)|
| One condition‡|| 0.19||1.22|| 6.31¶|| $3,965|| 5.28‡|
| (0.08)|| || || (751)|| |
| Two conditions‡|| 0.23||1.26|| 7.02‡|| $8,395|| 9.85‡|
| (0.09)|| || || (852)|| |
| Three conditions‡|| 0.33||1.39|| 10.14‡||$13,715|| 13.3‡|
| (0.10)|| || ||(1,033)|| |
| Four or more conditions‡|| 0.36||1.44|| 14.77‡||$23,996|| 25.49‡|
| (0.09)|| || || (942)|| |
|AMI (versus no AMI at admission)||−0.04||0.97|| 0.29|| −$793||−1.24|
| (0.06)|| || || (638)|| |
|Intercept||−1.86|| ||142.58*||$31,828|| 22.30‡|
| || (0.16)|| || ||(1,445)|| |
After controlling for case-mix differences the PTCA and stent groups did not differ in cumulative 2-year cost. As shown in the regression results presented in Table 5, age, sex, race, and Charlson index impacted cost significantly, whereas AMI at admission and the indicator of stent versus PTCA did not result in any difference. Younger patients were less expensive ($1883 to $2460), women had lower cost than men ($2274), and white patients had lower cost than other races ($3454). The more the number of clinical conditions identified by the Charlson index, the greater the cost. A patient with one condition had a total cost of $3965 more than a patient with no conditions, and similarly a patient with four or more conditions was $23,996 more expensive. Overall, stent placement was cost-neutral. Although initially more expensive, stenting resulted in significantly lower cost in the period after discharge, which helped offset the high index procedural cost.
To test the robustness of the results presented above, we performed both sensitivity and subgroup analyses. We had increased the cost of stenting in 1996 by $1219 based on the difference between PTCA and stenting in 1998. Increasing the cost of stenting by up to an additional $550 would not change the conclusions reached but costs greater than this will result in lower 2-year cumulative costs for PTCA versus stenting. As indicated in Table 1, there were differences in patient characteristics and risk factors between the groups. Specifically, PTCA patients had a higher proportion of MI and diabetes, which may place PTCA patients at a higher risk for complications and restenosis. We had included MI in the multivariate analysis to control for differences between PTCA and stent patients, but diabetes was only included as a component of the Charlson index. To eliminate any bias that may result from the higher levels of diabetes in the PTCA group, we performed the same regression excluding all patients with diabetes in both groups. The results remain the same with no differences in the 2-year cost between PTCA and stent patients.
- Top of page
This study evaluated the 2-year resource use and cost of bare stent placement versus PTCA in the actual practice setting. At 2-year follow-up, stent patients were significantly less likely to undergo repeat revascularization procedures after controlling for case-mix differences. Much of this difference is due to the lower rate of both emergent and elective CABG surgery among stent patients. Although stent patients had a significantly lower rate of repeat revascularization procedures, the difference of 4.5% at 2 years is much smaller than those reported in many clinical trials. One possible explanation is that stent patients in clinical trials may have been different from those who undergo stent placement in actual clinical practice. If this is the case, then we can expect the repeat revascularization rates between PTCA and stent patients in this study to differ from those reported in controlled clinical trials. In addition, stent technology has evolved over the years, and therefore the outcomes associated with newer stents and deliver systems could be better.
Stent and PTCA patients did not differ in the rate of hospitalization or outpatient visits during the follow-up period but stent patients stayed fewer days on average. Stent patients also had a significantly lower rate of SNF admissions and home health-care service usage. Not controlling for patient differences, the 2-year cost for stent patients was $32,102, and for PTCA patients it was $32,654. After controlling for differences in patient characteristics and risk factors, the regression estimates indicate that there are no cost differences between PTCA and bare stent patients. The index procedure cost was $1297 higher for stent patients but the significantly lower cost for stent patients compared to PTCA patients in the follow-up period offsets the higher initial cost. Although about 50% of the cost savings resulted from lower inpatient cost, a substantial 25% was from lower cost associated with outpatient, SNF, and home health-care services. Therefore, the inclusion of cost for provision of services in the nonhospital setting is important in analyzing the differences between PTCA and stent patients. These costs have not been included in previous assessments and should be included especially when studying the elderly population.
The results presented in this study focused on the resource use and costs associated with stenting versus PTCA. In evaluating these two procedures, it is important to consider the findings from this study in relation to the outcomes such as mortality and health-related quality of life associated with each procedure. For this study population, the in-hospital mortality, not controlling for patient differences, was 3.3% for PTCA and 1.8% for stents. We did not have mortality information for the follow-up period and therefore cannot ascertain whether this difference persisted over the long term and after controlling for patient differences.
This study analyzed claims data and therefore clinical information related to location, size, and type of lesion treated were not available. The absence of clinical data results in several limitations that need to be highlighted. First, because we do not have details on lesion treated, our revascularization rates are not directly comparable to the target lesion revascularization reported in clinical studies. The revascularization rate reported in this study could include interventions performed on lesions other than those targeted during the index intervention. Second, we do not have information on the restenosis rate, and therefore comparisons were solely based on rate of repeat interventions. Third, we tried to control for differences in patient risk factors between the two groups but may not have been able to account for all factors because they are often underreported in claims data. The PTCA group may have included more high-risk patients than the stent group and therefore a comparable group of PTCA patients to stent patients may have resulted in lower costs associated with PTCA than those reported in this study.
The Medicare program perspective was adopted in this study and therefore payment information and not hospital cost was reported. A potential limitation is that although Medicare payments are supposed to represent hospital cost, they may not always accurately reflect the true cost of providing care in a hospital setting. In addition, the data did not allow for costs associated with other treatments that may have been provided in addition to PTCA or stenting during the inpatient admission, for instance, thrombolytic therapy, to be reported separately. We therefore could not account for cost differences between the two groups that may have resulted from the use of adjunct interventions.
In conclusion, bare stents appear to be cost-neutral when compared to PTCA for the Medicare program. Although stents have a significantly higher index procedure cost, cost savings in the follow-up period offset the higher initial cost. Costs associated with use of skilled nursing facility, home health care, and outpatient services are a significant contributor to overall cost of these patients. The differences in repeat revascularization procedures although lower for stent patients were not as large as those reported in clinical trials. It is therefore important that studies outside of the clinical trial setting, controlling for differences in practice patterns, type of lesion treated, and patient risk factors, be performed to evaluate the long-term costs and outcomes associated with new stent technology such as coated stents.