Introduction of drug-eluting stents resulted in improved clinical outcomes for patients and reduced overall procedural costs.
Background: In clinical trials, drug-eluting stents (DES) improve clinical outcomes but are more expensive than bare-metal stents (BMS).
Objective: To assess clinical and economic outcomes of all percutaneous coronary intervention (PCI) procedures in a general interventional cardiology practice before and after DES introduction in 2003.
Methods: We identified all patients undergoing PCI in 2000-2002 early cohort, pre-DES era) and from 2004 through April 31, 2006 (late cohort, DES era) in a large PCI registry. Logistic and Cox proportional hazard models estimated the risk of adverse events; generalized linear modeling predicted economic outcomes.
Results: We compared 4303 early-cohort patients with 3422 late-cohort patients. Most early-cohort patients (90%) had BMS implanted; the rest had atherectomy or balloon angioplasty only. Among late-cohort patients, 83% had DES, 14% BMS, and 6% balloon angioplasty or atherectomy only. In-hospital adverse-event rates and incidence of death or myocardial infarction (during a median follow-up of 22 months) were similar. Follow-up procedures were significantly fewer in the later era (hazard ratio for target lesion revascularization: 0.58; 95% confidence interval [CI], 0.50-0.68). Although catheterization lab supply costs were higher in the DES era, length of stay following index PCI and overall practice costs were reduced, on average, 0.40 days and $2053 in the late cohort (95% bootstrapped CI of adjusted mean difference, −$2937 to −$1197). Follow-up cardiac hospitalization costs were similar at 1 year.
Conclusions: Patients undergoing PCI following DES introduction experienced improved clinical outcomes during follow-up and reduced overall procedural costs, despite higher stent acquisition costs.
(Am J Manag Care. 2010;16(8):580-587)
We assessed clinical and economic outcomes of all percutaneous coronary intervention procedures before and after introduction of drug-eluting stents (DES) in 2003.
Adoption of innovations in medical technologies presents numerous challenges. Pivotal randomized trials are performed in highly selected patients under tightly controlled circumstances. As new technologies diffuse into general practice, assessment of clinical efficacy and safety as well as expenses is necessary. Drug-eluting stents (DES) present a particular challenge given their rapidity of adoption by the cardiology community as the preferred percutaneous coronary intervention (PCI) treatment modality.1-8 Acquisition costs of DES are higher than those of the previously used bare-metal stents (BMS), but may be offset by fewer repeat procedures in clinical trials. The impact of DES on the overall interventional cardiology practice is not well understood, particularly from the provider’s perspective, and controversy continues over the efficacy, safety, and cost-effectiveness of this technology.1, 8-13
Trial-based economic analyses have assessed the cost-effectiveness of DES, but these analyses have significant limitations.14-17 For example, trials focus on DES versus BMS comparisons, whereas in clinical practice patients and devices are selected by operators based on clinical circumstances. Trial-based analyses restricted to DES versus BMS fail to take into account the proportion of patients with lesions who do not receive stents, and do not include patients that receive DES for so-called “offlabel” indications in day-to-day clinical practice.1 Although analyses of use among patients at higher risk in practice settings are emerging,18-24 economic analyses of the impact of the introduction of new technologies such as DES on overall clinical practice are limited. Whether DES can be introduced into a busy interventional practice in a cost-effective manner remains uncertain. Therefore, we assessed clinical and economic outcomes of all PCI procedures in a general interventional practice encompassing time periods before and after introduction of DES. We hypothesized that expensive new technologies such as DES can be introduced while managing and maintaining overall costs and expenses.
Study Design and Sample
All patients undergoing PCI at the Mayo Clinic are prospectively followed according to a well-established protocol, the Mayo Clinic PCI registry.25 This database contains demographic, clinical, angiographic, and outcome data, as well as follow-up hospitalization information. Patients are interviewed in person or by telephone at 6 and 12 months after the procedure and yearly thereafter to assess follow-up events. For cardiac-related hospitalizations that occur outside our local setting, discharge summaries are obtained and abstracted.
Drug-eluting stents became commercially available in the United States in March 2003. Because 2003 was a transitional year from BMS to DES, we compared outcomes among all patients undergoing PCI from January 1, 2004, through April 31, 2006 (late cohort, DES era) with all patients treated from January 1, 2000, through December 31, 2002 (early cohort, pre-DES era). To gain the best perspective on the impact of DES introduction on the overall practice, no specific patient subsets were excluded, such as those treated with balloon angioplasty alone or those with high-risk disease such as myocardial infarction (MI) and cardiogenic shock. Procedures were performed as previously described, all with overnight stays.26 Patients who underwent multiple procedures had the initial procedure within each time period analyzed, and follow-up procedures were classified as adverse events. As required by State of Minnesota statute, we excluded patients who refused consent for medical records research.
The Mayo Clinic is an integrated healthcare delivery system, in which physicians are salaried and physician and administrative leaders are accountable for both professional and technical costs. We have previously described our management responses to economic pressures caused by falling reimbursements. These led to a number of active ongoing cost-containment strategies.27 Briefly, these included development of preferred supplier relationships, critical analyses of utilization (particularly high-cost supply items), analyses of processes of care, and elimination of non—value-added utilization. For example, use of smaller vascular access sheaths, increased use of transradial access, fewer prolonged glycoprotein inhibitor infusions, and virtual elimination of routine postprocedure heparin infusions led to earlier ambulation. Case mix also changed, with gradual elimination of vascular brachytherapy and expansion of primary PCI for MI. Acquisition costs for ancillary supplies such as balloons, wires, and guiding catheters are continuously and vigorously negotiated using competitive bidding processes.
Clinical outcomes of interest included procedural success (defined as <50% residual stenosis and without in-hospital death, Q-wave MI, or coronary artery bypass surgery), in-hospital death, any MI, stroke, or target lesion revascularization; and target lesion failure rates during follow-up (defined as death, MI, or target lesion revascularization). We defined MI as the presence of any 2 of the following: an episode of angina lasting >20 minutes, a rise in the serum creatine kinase or creatine kinase MB isoenzyme concentration greater than 2-fold, or ST-segment changes or new Q-waves on serial electrocardiograms indicative of myocardial damage. The Mayo Clinic Risk Score was used to assess the risk profile of treated patients.28
Economic analyses were conducted from the provider perspective and focused on direct costs associated with the index PCI (procedural costs) as well as postprocedural length of stay (LOS) (days from procedure to hospital discharge date). The rate of cardiac-related hospitalizations and associated costs also was assessed during 1 year following initial PCI (follow-up costs).
Procedural Costs. We used administrative data to track hospital and physician service use and related expenditures for index PCI episodes. Utilization was valued using standard methods by grouping services into the Medicare Part A and Part B classification. Part A billed charges were adjusted using department level hospital cost-to-charge ratios and wage indexes. Part B physician service costs were proxied based on Medicare reimbursement rates.
Follow-up Costs. To value cardiac hospitalizations during follow-up, we matched observed procedure and diagnosis combinations to episodes of similar case mix occurring in the 2004 Nationwide Inpatient Sample. Sponsored by the US Agency for Healthcare Research and Quality, the Nationwide Inpatient Sample contains data from approximately 8 million hospital stays each year, including estimated hospital (nonphysician) costs.29 All costs presented have been adjusted to reflect 2007 constant dollars.
Continuous data are summarized as mean ± standard deviation. Discrete data are presented as frequency (percentage). Patient characteristics and observed in-hospital clinical outcomes were compared using the t test for continuous data, the Mann-Whitney rank sum test for ordinal data, and Pearson’s χ2 test for categorical data, as appropriate. Kaplan-Meier estimated adverse-event rates during follow-up were compared using the log rank test. Follow-up began at the time of PCI and included in-hospital events.
We used propensity score methods to account for observed potential confounding factors.30,31 Logistic regression estimated the probability (propensity score) that a patient was from the DES-era group given clinical and angiographic characteristics. We included variables whose unadjusted associations with time era were significant at the .15 level. Patients with propensity scores outside the range common to both eras were excluded from propensity-score—adjusted analyses. Five propensity score strata of nearly equal size were defined.
We used logistic regression to estimate adjusted odds ratios (ORs) for in-hospital events and Cox proportional hazard models to estimate the hazard for adverse events during follow-up within stratum. These 5 estimates were then combined using an inverse-variance weighted average. The resulting estimate and its confidence interval (CI) were exponentiated back to the OR or hazard ratio (HR) scale.
Observed costs and LOS were compared using t tests and nonparametric bootstrapped CIs.32,33 Generalized linear modeling assessed the impact of treatment era on in-hospital costs34 and assumed a logarithmic link function and a gamma, inverse Gaussian, or Poisson distribution based on the modified Park test recommended by Manning and Mullahy.35 We assumed a negative binomial distribution function with log link in the model assessing LOS.
The effect of time era on hospitalization rates and costs during the year of follow-up were assessed using 2-part models to account for the overabundance of zero values.36 For the 2-part model specifications, logistic regression was used in part 1 to estimate the probability of experiencing hospitalization (or costs) conditional on patient characteristics; part 2 used either a Poisson specification to assess hospitalization rates (among those with events) or an inverse Gaussian distribution function with log link in cost models. Models adjusted for person weights to account for right censoring of data. All statistical tests were 2-sided, and P values less than .05 were considered significant. SAS version 9.1 (SAS Institute Inc, Cary, NC) was used in the analyses.
We identified 4303 early-cohort patients and 3422 latecohort patients undergoing PCI. Baseline characteristics are outlined in Table 1. The mean age of patients undergoing PCI was about 67 years, and 70% were male. More than half (57%) had previously experienced an MI, including a substantial proportion in the week prior to the index procedure 4% had preprocedural shock; and roughly 16% had a history of congestive heart failure. About a quarter of patients were diabetic, and 18% were current smokers. The Mayo Clinic Risk Scores were similar between cohorts.
The vast majority of PCIs (70%) were performed for urgent or emergent indications on inpatients. Bare-metal stents were used in 90% of cases of early-cohort patients; the remainder had balloon angioplasty alone. In the late cohort, however, DES were used in 83% of cases, BMS in 14%, and balloon angioplasty alone in 5%. Utilization of glycoprotein 2b/3a inhibitors fell from 65% in the early cohort to 56% in the late cohort (P <.001) as the rate of dual oral antiplatelet therapy increased.
Observed in-hospital clinical outcomes by treatment era are shown in Table 2. Procedural success was obtained in 95% of cases in both time periods, and postprocedural thrombolysis in myocardial infarction 3 flow (normal antegrade flow) in all treated lesions was obtained in 94% and 95% of the early and late cohorts, respectively (P = .019). In-hospital deaths rates were similar between groups, with death occurring in fewer than 2% of cases. In-hospital MI occurred in 5% and 4% of the early and late cohorts, respectively (P = .005), and in-hospital bypass surgery was required in 1% of cases in both time periods. The incidence of in-hospital death, MI, and target lesion revascularization was 7% of cases in both cohorts.
The median duration of follow-up for adverse clinical events was 28 months and 14 months for the early and late cohorts, respectively; 93% of patients had been contacted in the year prior to freezing the data for analysis and 98.5% had been contacted in the prior 2 years.
The Kaplan-Meier estimated all-cause mortality at 24 months was 9.0% in both groups, and the cumulative incidence of death or MI was 17% in both (P = .41). As shown in Figure 1, target lesion revascularization rates were reduced among patients treated in the later era (16% vs 9%; P <.001).
Incomplete data and late research authorization denials forced the exclusion of 48 patients from economic analysis, leaving 4271 early-cohort and 3412 late-cohort patients for cost analyses. Observed hospital costs were similar between eras ($13,724 early vs $13,029 late cohort; P = .08). Mean costs associated with catheterization lab—related supplies (which includes stents) were a mean of $684 higher in the cohort ($3115 vs $3799; P <.001). Higher catheterization lab supply costs were partially offset by nonsupply catheterization lab costs which were, on average, $389 lower in the late cohort ($6078 vs $5689; P <.001). Because of the overall impact of expense management strategies, the total mean direct costs and hospital LOS were reduced by approximately $1000 and 0.23 days, respectively, for patients treated in the DES era (total costs: $17,381 vs $16,270; bootstrapped 95% CI, −$1958 to −$397).
Excluded from the propensity score classification because of extreme scores were 8 late-cohort and 21 early-cohort patients, leaving 3414 late-cohort and 4282 early-cohort PCIs for strata-adjusted comparison. Patient characteristics were similar between cohorts within strata.
The adjusted incidence of in-hospital death was similar in both cohorts (OR, 1.07; 95% CI, 0.74-1.56; P = .71), although a trend toward lower in-hospital MI in the late cohort was noted (OR, 0.79; 95% CI, 0.62- 1.00; P = .055). The incidence of in-hospital death, Q-wave MI, emergency bypass surgery, or stroke was similar between early and late cohorts (OR, 1.09; 95% CI, 0.81-1.46; P = .58). In subgroup analysis, lower rates of in-hospital MI were observed in the late cohort among individuals age >70 years (OR, 0.65; 95% CI, 0.47-0.90; P = .01) and diabetics (OR, 0.58; 95% CI, 0.37- 0.93; P = . 023).
Similar death rates during follow-up were observed in both cohorts HR, 0.97; 95% CI, 0.81-1.15; P = .70). The incidence of death or MI during follow-up also was similar (HR, 0.93; 95% CI, 0.82-1.05; P = .24). The long-term incidence of death, MI, or target lesion revascularization, however, was significantly reduced in the late cohort (HR, 0.80; 95% CI, 0.72-0.88; P <.001). As shown in Figure 2, every patient subgroup considered had significantly reduced rates of target lesion revascularization during follow-up except those with a final device size greater than 3.5 mm. There were no significant differences in frequency of staged procedures that might have influenced results. There were 59 staged PCIs in the early cohort (1.4%) and 64 (1.9%) in the late cohort (P = .084).
With the exception of catheterization lab supply costs, all adjusted procedural cost analyses suggest savings in the late compared with the early cohort (Table 3). Models predicted an incremental hospital cost savings of $1375 and a physician cost savings of $619, on average, in the late cohort. Total costs associated with the index PCI were estimated to be approximately $2000 lower for late-cohort patients ($19,111 vs $17,058; 95% CI, −$2937 to −$1197). Procedural cost savings were predicted for all patient subgroups considered except those with a maximum final device size less than 2.5 mm.
The predicted rate of cardiac-related hospitalizations during 1-year followup was similar in both cohorts (0.36 vs 0.36; 95% CI, −0.019 to 0.0002). Hospital costs associated with these episodes also were similar in adjusted analyses ($6817 early cohort vs $6984 late cohort; 95% CI, −$59 to $387).
Our analysis attempts to address whether expensive new technologies, such as DES, can be introduced into clinical practice while maintaining overall procedural cost structure. A practice-based analysis is needed for a number of reasons. Patients enrolled in randomized trials frequently are not representative of the population in whom technologies are subsequently applied. As a result, these trials fail to capture the clinical and economic impacts of new technologies on the overall population of patients ultimately treated with new devices. By including all patients in the pre-DES and post-DES eras and assessing clinical as well as economic outcomes, we attempted to minimize biases and to gauge the complete impact of DES technology introduction on interventional practice. We found that the incidence of in-hospital major adverse cardiac events was similar between eras, and the incidence of death or MI during long-term follow-up also was similar between eras. However, need for target lesion revascularization was significantly reduced in the post-DES era. Total in-hospital cost of the index coronary interventional procedures was lower in the post-DES era, despite higher stent acquisition costs reflecting aggressive management strategies to contain costs. The rate of cardiac hospitalizations during follow-up and the estimated associated costs were statistically similar between the pre-DES and post-DES eras.
Two recent observational analyses also assessed clinical outcomes following coronary artery stenting in eras before and after DES availability and similarly found the widespread adoption of DES to be associated with reduced revascularization rates and comparable death rates.21,24 However, neither study investigated the economic impact of DES uptake.21,24 A number of trialbased economic analyses have been conducted comparing patients receiving DES versus BMS.14-17 In general, these studies suggest that although use of DES is not cost saving compared with the use of BMS, this technology provides good value for money spent. Recently, Groeneveld and colleagues observationally compared total costs of care among Medicare beneficiaries and found increased costs initially with DES compared with BMS use. However, this cumulative cost difference narrowed during 1 year following PCI.23 It is important to note that we focused on the economic implications of DES availability for the entire PCI patient population, not just on pure stent subsets. Despite the higher acquisition cost of DES compared with BMS, we found initial procedural cost savings, suggesting that new technologies can be adopted if overall practice-based expenses are carefully managed.
The fact that procedural costs fell despite the higher DES acquisition costs was noteworthy, and there are multiple reasons for this change. Stent acquisition costs are merely one component of in-hospital costs. Advances in technology and increasing experience have resulted in a temporal improvement in the outcomes of PCI procedures and reduction in complications, such as need for emergency coronary artery bypass surgery and declining LOS.37 Simultaneous with the introduction of DES, supply acquisition costs for other items, such as BMS, angioplasty balloons, sheaths, and wires, also fell in response to market pressures and competitive bidding. Our institution actively engages in practice evaluation and process improvement aimed at optimizing resource consumption for PCI procedures.27 Resource utilization clinically felt not to be contributory to outcomes is targeted for elimination or reduction. The net effect of this comprehensive practice management was to offset the higher acquisition costs of DES.
Our analysis has a number of important limitations. Because of its retrospective nature, patients obviously could not be randomly assigned to time eras, and the usual limitations of this design are relevant. It is possible, though not likely, that selection bias may have resulted in a significant shift in the type of patients being selected for PCI. If a higher proportion of lower risk patients were selected for PCI in the DES era, that may have influenced the results. The data, specifically the Mayo Clinic Risk Score, however, do not suggest this, and indeed case mix has been steadily evolving toward higher risk and older patients.38 Additionally, we had detailed resource utilization data for the initial PCI episode, but relied on secondary administrative data and gross costing methods to value cardiac-related hospitalizations during follow-up. It remains unclear to what extent this valuation approach impacted our findings. Outpatient costs, particularly for prolonged dual antiplatelet therapy following DES implantation, were not considered. Finally, our study reflects the experience of a single, high-volume referral center. Patients and results may differ in other centers and practice settings.
In a large consecutive series, DES technology was introduced while maintaining or reducing the overall costs of PCI. The DES era was associated with improved clinical outcomes during follow-up and with reduced in-hospital costs. These data suggest that costly new technologies can be introduced into a general practice setting while maintaining and improving patient outcomes and overall cost structures.
Author Affiliations: From the Division of Cardiovascular Diseases (CSR, JLR, MS, JFB, BJG, HHT, DRH), Division of Biomedical Statistics and Informatics (RJL), and the Division of Health Care Policy & Research (JTL, KHL), College of Medicine, Mayo Clinic, Rochester, MN.
Funding Source: There was no outside funding for this research.
Author Disclosures: Dr Gersh reports serving as a paid consultant to Abbott Laboratories and Boston Scientific. The other authors (CSR, JLR, MS, RJL, JFB, JTL, HHT, DRH, KHL) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (CSR, JLR, MS, RJL, HHT, KHL); acquisition of data (CSR, RJL, JTL, HHT, KHL); analysis and interpretation of data (CSR, JLR, RJL, JFB, JTL, BJG, HHT, KHL); drafting of the manuscript (CSR, MS, JFB, HHT, DRH, KHL); critical revision of the manuscript for important intellectual content (CSR, MS, RJL, JFB, BJG, HHT, DRH, KHL); statistical analysis (RJL, JTL, BJG, KHL); administrative, technical, or logistic support (CSR, JLR, KHL); and supervision (CSR).
Address correspondence to: Charanjit S. Rihal, MD, MBA, Division of Cardiovascular Diseases, Mayo Clinic, 200 First St SW, Rochester, MN 55905. E-mail: firstname.lastname@example.org.
1. Farb A, Boam AB. Stent thrombosis redux—the FDA perspective. N Engl J Med. 2007;356(10):984-987.
2. Morice MC, Serruys PW, Sousa JE, et al; RAVEL Study Group. A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med. 2002;346(23):1773-1780.
3. Moses JW, Leon MB, Popma JJ, et al; SIRIUS Investigators. Sirolimus-eluting stents versus standard stents in patients with stenosis in a native coronary artery. N Engl J Med. 2003;349(14):1315-1323.
4. Stone GW, Ellis SG, Cox DA, et al; TAXUS-IV Investigators. A polymer-based, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med. 2004;350(3):221-231.
5. Valgimigli M, Percoco G, Malagutti P, et al; STRATEGY Investigators. Tirofiban and sirolimus-eluting stent vs abciximab and bare-metal stent for acute myocardial infarction: a randomized trial. JAMA. 2005;293(17):2109-2117.
6. Laarman GJ, Suttorp MJ, Dirksen MT, et al. Paclitaxel-eluting versus uncoated stents in primary percutaneous coronary intervention. N Engl J Med. 2006;355(11):1105-1113.
7. Spaulding C, Henry P, Teiger E, et al; TYPHOON Investigators. Sirolimus-eluting versus uncoated stents in acute myocardial infarction. N Engl J Med. 2006;355(11):1093-1104.
8. Stone GW, Moses JW, Elllis SG, et al. Safety and efficacy of sirolimus- and paclitaxel-eluting coronary stents. N Engl J Med. 2007;356(10):998-1008.
9. Lagerqvist B, James SK, Stenestrand U, Lindbäck J, Nilsson T, Wallentin L; SCAAR Study Group. Long-term outcomes with drug-eluting stents versus bare-metal stents in Sweden. N Engl J Med. 2007;356(10):1009-1019.
10. Shuchman M. Trading restenosis for thrombosis? New questions about drug-eluting stents. N Engl J Med. 2006;355(19):1949-1952.
11. Kong DF, Eisenstein EL, Sketch MH, et al. Economic impact of drug-eluting stents on hospital systems: a disease-state model. Am Heart J. 2004;147(3):449-456.
12. Hodgson JM, Bottner RK, Klein LW, et al. Drug-eluting stent task force: final report and recommendations of the working committees on cost-effectiveness/economics, access to care, and medicolegal issues. Catheter Cardiovasc Interv. 2004;62(1):1-17.
13. Gunn J, Morton AC, Wales C, Newman CM, Crossman DC, Cumberland DC. Drug eluting stents: maximising benefit and minimising cost. Heart. 2003;89(2):127-131.
14. Cohen DJ, Bakhai A, Shi C, et al; SIRUS Investigators. Cost-effectiveness of sirolimus-eluting stents for treatment of complex coronary stenoses: results from the Sirolimus-Eluting Expandable Stent in the Treatment of Patients With De Novo Native Coronary Artery Lesions (SIRIUS) Trial. Circulation. 2004;110(5):508-514.
15. Van Hout BA, Serruys PW, Lemos PA, et al. One year cost-effectiveness of sirolimus eluting stents compared with bare metal stents in the treatment of single native de novo coronary lesions: an analysis from the RAVEL trial. Heart. 2005;9(4):507-512.
16. Bakhai A, Stone GW, Mahoney E, et al; TAXUS-IV Investigators. Cost effectiveness of paclitaxel-eluting stents for patients undergoing coronary revascularization: results from the TAXUS-IV Trial. J Am Coll Cardiol. 2006;48(2):253-261.
17. Kaiser C, Brunner-La Rocca P, Buser PT, et al; BASKET Investigators. Incremental cost-effectiveness of drug-eluting stents compared with a third-generation bare-metal stent in a real-world setting: randomised Basel Stent Kosten Effektivitats Trial (BASKET) [published correction appears in Lancet. 2005;366(9503):2086]. Lancet. 2005;366(9489):921-929.
18. Marroquin OC, Selzer F, Mulukutla SR, et al. A comparison of bare-metal and drug-eluting stents for off-label indications. N Engl J Med. 2008;358(4):342-352.
19. Kindermann M, Adam O, Werner N, Böhm M. Clinical Trial Updates and Hotline Sessions presented at the European Society of Cardiology Congress 2007: (FINESSE, CARESS, OASIS 5, PRAGUE-8, OPTIMIST, GRACE, STEEPLE, SCAAR, STRATEGY, DANAMI-2, ExTRACT-TIMI-25, ISAR-REACT 2, ACUITY, ALOFT, 3CPO, PROSPECT, EVEREST, COACH, BENEFiT, MERLIN-TIMI 36, SEARCH-MI, ADVANCE, WENBIT, EUROASPIRE I-III, ARISE, getABI, RIO). Clin Res Cardiol. 2007;96(11):767-786.
20. Tu JV, Bowen J, Chiu M, et al. Effectiveness and safety of drug-eluting stents in Ontario. N Engl J Med. 2007;357(14):1393-1402.
21. Hannan EL, Racz M, Holmes DR, et al. Comparison of coronary artery stenting outcomes in the eras before and after the introduction of drug-eluting stents. Circulation. 2008;117(16):2071-2078.
22. Groeneveld PW, Matta MA, Greenhut AP, Yang F. Drug-eluting compared with bare-metal coronary stents among elderly patients. J Am Coll Cardiol. 2008;51(21):2017-2024.
23. Groeneveld PW, Matta MA, Greenhut AP, Yang F. The costs of drug-eluting coronary stents among Medicare beneficiaries. Am Heart J. 2008;155(6):1097-1105.
24. Malenka DJ, Kaplan AV, Lucas FL, Sharp SM, Skinner JS. Outcomes following coronary stenting in the era of bare-metal vs the era of drug-eluting stents. JAMA. 2008;299(24):2868-2876.
25. Hasdai D, Garratt KN, Grill DE, Lerman A, Holmes DR Jr. Effect of smoking status on the long-term outcome after successful percutaneous coronary revascularization. N Engl J Med. 1997;336(11):755-761.
26. Holmes DR Jr, Vlietstra RE. Percutaneous transluminal coronary angioplasty: current status and future trends. Mayo Clin Proc. 1986;61(11):865-876.
27. Rihal CS, Kamath CC, Holmes DR Jr, et al. Economic and clinical outcomes of a physician-led continuous quality improvement intervention in the delivery of percutaneous coronary intervention. Am J Manag Care. 2006;12(8):445-452.
28. Singh M, Rihal CS, Lennon RJ, Garratt KN, Holmes DR Jr. Comparison of Mayo Clinic risk score and American College of Cardiology/American Heart Association lesion classification in the prediction of adverse cardiovascular outcome following percutaneous coronary interventions. J Am Coll Cardiol. 2004;44(2):357-361.
29. Healthcare Cost and Utilization Project (HCUP). Overview of the Nationwide Inpatient Sample (NIS). 2004. http://www.hcup-us.ahrq.gov/nisoverview.jsp.
30. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8 pt 2):757-763.
31. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17(19):2265-2281.
32. Efron B, Tibshirani RJ. An Introduction to the Bootstrap. Boca Raton, FL: Chapman & Hall/CRC; 1998.
33. Briggs A, Gray A. The distribution of health care costs and their statistical analysis for economic evaluation. J Health Serv Res Policy. 1998;3(4):233-245.
34. Barber J, Thompson S. Multiple regression of cost data: use of generalised linear models. J Health Serv Res Policy. 2004;9(4):197-204.
35. Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ. 2001;20(4):461-494.
36. Afifi AA, Kotlerman JB, Ettner SL, Cowan M. Methods for improving regression analysis for skewed continuous or count responses. Annu Rev Public Health. 2007;28:95-111.
37. Yang EH, Gumina RJ, Lennon RJ, Holmes DR Jr, Rihal CS, Singh M. Emergency coronary artery bypass surgery for percutaneous coronary interventions: changes in the incidence, clinical characteristics, and indications from 1979 to 2003. J Am Coll Cardiol. 2005;46(11):2004-2009.
38. Singh M, Rihal CS, Gersh BJ, et al. Twenty-five year trends in in-hospital and long-term outcome after percutaneous coronary intervention: a single-institution experience. Circulation. 2007;115(22):2835-2841.