Publication
Article
Economic and Clinical Outcomes of a Physician-led Continuous Quality Improvement Intervention in the Delivery of Percutaneous Coronary Intervention
Objective: To compare clinical and economic outcomes associated with percutaneous coronary intervention (PCI) in cohorts before and after continuous quality improvement (CQI) was instituted.
Study Design: Observational study.
Methods: Clinical, angiographic, procedural, and outcome data on 1441 pre-CQI and 1760 post-CQI PCIs (performed in 1997 and 1998, respectively) were derived from an institutional PCI registry. Administrative data were used to estimate total procedural and postprocedural costs and length of stay (LOS). Logistic and generalized linear modeling was used to adjust in-hospital clinical and economic outcomes, respectively, for differences in patient characteristics.
P
P
Results: The 2 cohorts were similar in terms of age, sex, and rate of diabetes. Post-CQI patients more often received intracoronary stents, had urgent PCIs, had a history of prior PCI, and received glycoprotein IIb/IIIa inhibitors. Procedural success without in-hospital complications occurred in 90% of both cohorts and did not differ statistically in adjusted analyses. Compared with patients treated pre-CQI, those treated post-CQI had a reduced adjusted odds ratio for in-hospital death or any myocardial infarction (odds ratio = 0.66; 95% confidence interval = 0.46, 0.95). Models predicted a mean postprocedural LOS difference of 0.8 days (2.8 days pre-CQI vs 2.0 days post-CQI; <.001) and an average post-CQI cost savings of $5430 (<.001).
Conclusion: Physician-led, multidisciplinary practice management efforts were successful at significantly reducing PCI-related costs in an era of rapid technological advances while maintaining and perhaps improving quality of care.
(Am J Manag Care. 2006;12:445-452)
An estimated 664 000 percutaneous coronary interventions (PCIs) were performed in the United States in 2003, an estimated increase of 326% since 1987.1 This growth has been driven by innovations such as intracoronary stents and parenteral platelet receptor antagonists. These technologies have allowed wide applicability of PCI to nearly all groups of patients with coronary artery disease–including the rapidly expanding elderly population. However, this growth has been associated with concerns over the costs and outcomes of these procedures.2,3
New technology has been accompanied by a sharp increase in the acquisition costs of consumable and disposable supply items for provider institutions. At the same time, reimbursement pressures and increases in labor costs have exerted pressure on margins–in some cases rendering previously strong tertiary-care institutions insolvent. Shrinking margins have brought expense management, rather than mere volume increases, into focus as a tool for ensuring financial sustainability of technologically intensive practices.4 The realization of shrinking margins and even financial losses for new medical technologies is particularly evident in cardiovascular services, which account for approximately 17% of total medical care expenditure in the United States.5,6
This article describes the impact of a continuous quality improvement (CQI) effort implemented by a physician-led, multidisciplinary team at a large academic medical center located in the Midwest. The objectives of this effort were to identify opportunities for expense reduction to maintain the long-term financial viability of the cardiovascular service line, while simultaneously ensuring access to new technology, good procedural outcomes, and patient satisfaction. This effort included a critical evaluation of current practice to develop process improvements, reduce practice variation, and optimize resource consumption for PCI procedures. The effectiveness of this intervention was examined in terms of specific clinical and economic outcomes. The purpose of this report is to present the process and outcomes of the CQI.
METHODS
Intervention
In 1996, Cardiac Catheterization Laboratory leadership was charged with critically evaluating the PCI practice to identify and pursue opportunities for expense reduction while improving current evidence-based clinical practice. In response, a multidisciplinary group consisting of clinicians (physicians, nurses, technicians), administrators (management accounting, revenue systems, administration), and continuous-improvement experts was assembled to address the issue.
This group used the Plan-Do-Study-Act (PDSA) cycle model for improvement. This model initially focuses on 3 important questions:
- What are we trying to accomplish?
- What changes can we make?
The PDSA cycle helps develop, test, and implement change with an efficient trial and learning process.7,8 Cycles are initiated for tests of change on a small scale, which eventually can lead to important improvements. Details about this PDSA model of improvement in the context of expense reduction and quality enhancement in PCI delivery are provided in the Appendix. (Note to readers: this appendix is available online at: www.ajmc.com.) The leadership and involvement of physicians in the group and process were critical.
This CQI process resulted in improved management of cardiac supply expenses and development of a final clinical pathway for optimal care processes. This pathway included a detailed resource protocol (or internal benchmark) for appropriate tests and supplies for a median PCI case under diagnosis-related group 112. The protocol included details on labor, supply, and hospital resources deemed necessary by the group to perform uncomplicated PCI procedures, with an expense goal below the aggregate Medicare reimbursement. Unnecessary inventory and resource use deemed of low clinical benefit (eg, routine complete blood count and electrolyte measurement after successful PCI) were eliminated. Resource use deemed of high clinical benefit (eg, use of platelet glycoprotein receptor inhibitors) was retained for quality of care (Table 1). Only bare metal stents were used during the study time frame. As changes were identified and tested, physicians became increasingly aware of and developed interest in resource utilization patterns and associated costs in the course of doing procedures. A significant "spillover" effect of this effort on all procedures occurred.
After a series of PDSA cycles, specific changes to the practice were recommended. The new guidelines and changes were implemented in a stepwise fashion and were complete by early 1998.
Study Population
The study population consisted of pre-CQI and post- CQI cohorts of patients who underwent PCI between January 1 and December 31, 1997, and January 1 and December 31, 1998, respectively. We included multiple qualifying procedures per patient during the study; however, target vessel revascularization was considered a complication of initial PCI and not a qualifying "index" PCI event. In accordance with state statute, we excluded all procedures for patients who did not authorize use of their records for research.
A total of 3379 PCIs were performed on 2992 unique patients during the study duration; 73 PCIs (65 patients) were excluded because of lack of research authorization. Sixty-three nonindex procedures were identified and excluded either because they were target vessel revascularizations or because they were part of an elective, staged PCI during the same hospital episode. Incomplete billing data forced the exclusion of an additional 42 PCI episodes (33 patients) from analysis. The result was a sample size of 3201 PCIs (2894 patients), of which 1441 were pre-CQI PCIs and 1760 were post-CQI PCIs. The study was approved by the institutional review board.
Data Sources
All clinical data were obtained from a prospective PCI registry where all patient-specific data are entered at the time of an interventional procedure. This registry includes clinical, procedural, angiographic, in-hospital, and follow-up outcome data on all patients receiving a PCI locally.9
Administrative data sources were used to track medical resource utilization, related expenditures, and length of stay (LOS) for these PCI episodes. Utilization was valued using standard methods by graphing services into the Medicare Part A and Part B classification: Part A billed charges were adjusted using hospital cost-to-charge ratios at the departmental level and wage indexes. Costs associated with Part B physician services were proxied based on 2000 Medicare reimbursement rates by current procedural terminology code. All costs presented were adjusted to reflect 2004 constant dollars, using the hospital services and physicians' services components of the Consumer Price Index, as appropriate.10 Accounting practices remained unchanged for the duration of the study.
Outcomes
Primary clinical outcomes of interest included procedural success (defined as achieving a final luminal diameter stenosis of <50%, without in-hospital complications of death, myocardial infarction [MI], or target vessel revascularization); in-hospital death or MI; and a secondary composite end point of in-hospital death, MI, or target vessel revascularization. Myocardial infarction was defined as any 2 of the following criteria: an episode of prolonged angina lasting ≥20 minutes, a rise in the serum concentration of the creatine kinase MB isoenzyme of more than 2-fold above normal, or ST-segment/T-wave changes or new Q waves on serial electrocardiograms indicative of myocardial damage. Routine measurement of cardiac biomarkers was not standard practice during the entire study duration.
Economic outcomes of interest consisted of total direct procedural and postprocedural costs for PCI-related hospitalizations and postprocedural LOS (defined as the number of days from the date of the procedure to the hospital discharge date). Total cost assessments included costs associated with hospital as well as physician services.
Statistical Analyses
t
t
We examined differences in baseline characteristics and unadjusted clinical outcomes between the 2 patient cohorts with tests and chi-square tests for continuous and categorical variables, respectively. Observed mean costs and LOS were compared using tests and nonparametric bootstrap confidence intervals (CIs).11 Multivariate regression techniques were used to evaluate the independent effect of intervention, controlling for baseline differences in cohort characteristics. Models were developed separately for in-hospital clinical outcomes, total costs, and LOS.
International
Classification of Diseases, Ninth Revision, Clinical
Modification
ICD-9-CM
A well-validated summary patient risk score was used in outcomes adjustment.12,13 Specific variables considered in this Mayo Clinic Risk Score include 5 clinical variables (age; congestive heart failure [CHF], defined as New York Heart Association functional class ≥III; urgent/emergent PCI; chronic renal disease; preprocedural cardiogenic shock) and 3 angiographic characteristics (left main stenosis of ≥70%, multivessel disease, presence of thrombus in any lesion). Final models included an indicator variable for the post-CQI period; demographic, clinical, angiographic, and procedural characteristics; and the Mayo Clinic Risk Score. Cost and LOS prediction models also included summary economic- severity measures (Overall Resource Demand Scale and LOS scale) obtained by using () diagnosis codes (which were noted in the administrative data) processed through Medstat Disease Staging software.14 Resource demand and LOS scales are measures of expected resource consumption and LOS, respectively, conditional on known comorbidities and severity of illness, scaled to average 100 across all patients in the Disease Staging development database. An estimated Overall Resource Demand Scale score of 125, therefore, implies that resource consumption (charges) was 25% higher than expected compared with the average of predicted charges taken across all patients in the database.
P
Logistic regression modeling was used to assess in-hospital clinical outcomes. Generalized linear modeling techniques were used to assess costs and LOS to account for the nonnegative and typically skewed nature of these economic end points.15 Cost models assumed a logarithmic link function and a gamma or inverse Gaussian distribution based on the modified Park test recommended by Manning and Mullahy.16 We assumed a negative binomial distribution function with log link in the model assessing LOS. All statistical tests were 2-sided, and values less than .05 were considered significant. SAS version 8.2 (SAS Institute Inc., Cary, NC) was used in the analyses.
RESULTS
P
P
P
P
Table 2 presents the baseline clinical characteristics of patients undergoing PCI before and after the CQI. The 2 cohorts were similar in terms of age (mean of 66 years), sex (70% male), body mass index (29), and most clinical characteristics including the composite Mayo Clinic Risk Score (6.3 vs 6.4; = .18). However, a greater proportion of patients treated post-CQI were hypertensive (64% vs 60%; = .015); had a greater number of MIs 1 to 7 days before the procedure; had more cases of prior PCI (31% vs 27%; = .036); had more cases of urgent PCI (48% vs 36%; <.001); and had more high-risk angiographic characteristics (eg, an ulcerated plaque, moderate or severe angulated bends at treatment sites). A higher rate of use of glycoprotein IIb/IIIa receptor inhibitors, a greater number of vessels treated (including the left anterior descending), and a greater number of stents placed also were observed in the post-CQI cohort.
Clinical Outcomes After CQI
P
Observed clinical outcomes after PCI are presented in Table 3. Similar rates of in-hospital major complications (death, any MI, any Q-wave MI, coronary artery bypass graft surgery) were seen pre-CQI and post-CQI. Rates of the composite end points of in-hospital death or MI, and in-hospital death, MI, or target vessel revascularization also were similar in the 2 cohorts. The rates of overall procedural success did not differ: procedural success was obtained in 1305 (90.6%) and in 1591 (90.4%) of PCIs performed pre-CQI and post-CQI, respectively (= .88).
Table 4 presents the estimated effects of the CQI on clinical outcomes with adjustment for baseline cohort characteristics. Logistic model results for in-hospital death, in-hospital any MI, and in-hospital Q-wave MI suggest that CQI implementation did not significantly affect these adverse clinical event rates. However, patients treated post-CQI were less likely to experience the composite end point of death or any MI during hospitalization (odds ratio [OR] = 0.66; 95% CI = 0.46, 0.95) compared with the patients treated pre- CQI. Similarly, the adjusted OR for in-hospital death, any MI, or target vessel revascularization favored the post-CQI implementation cohort (OR = 0.56; 95% CI = 0.39, 0.78). The likelihood of overall procedural success did not differ in the pre-CQI and post-CQI cohorts (OR = 1.16; 95% CI = 0.88, 1.53).
Economic Outcomes After CQI
P
P
P
Table 5 summarizes the observed costs of care and LOS associated with PCI. Total direct medical costs were, on average, approximately $2031 less for patients undergoing procedures post-CQI than for patients undergoing procedures pre-CQI ($16 815 vs $14 784; 95% CI of difference = -$3006, -$1020). Significantly reduced hospital costs ($999; = .03) and physician costs ($1033; <.001) also were observed after CQI, comprising 80% and 20% of total PCI-related costs, respectively. Mean costs associated with supply use in the catheterization laboratory were nonsignificantly reduced by $148 post-CQI ($3861 vs $3712; 95% CI of difference = -$370, $67). Observed postprocedural LOS was similar between study periods (2.66 days pre-CQI vs 2.44 days post-CQI; = .12).
P
P
P
In adjusted analyses, the CQI was a strong independent predictor of all economic outcomes of interest (LOS, and hospital, physician, and total costs of care). Table 6 shows the estimated independent impact of select covariates on total costs and LOS. Cost models predict total in-hospital PCI costs to be, on average, $5430 lower post-CQI than total in-hospital PCI costs pre-CQI ($19 520 vs $14 090; <.001). The CQI intervention also significantly affected the LOS in multivariate analysis, with an adjusted mean LOS reduction of 0.82 days after the CQI (95% CI of difference = -1.05, -0.6). Similar results confirming an economic advantage with the intervention were seen in adjusted analyses of hospital and physician costs (model results not shown), with predicted average cost savings of $3979 ($15 251 pre-CQI vs $11 272 post-CQI; <.001) and physician cost savings of nearly $1500 ($4357 pre-CQI vs $2885 post-CQI; <.001).
Adjusted analyses also identified clinical factors that had a significant independent impact on economic outcomes (Table 6). For example, patients who presented with a recent MI, CHF, or a C-type lesion had significantly higher predicted costs of care than patients who presented without these clinical characteristics (adjusted mean cost differences of $7129, $2997, and $1398, respectively). Similarly, our adjusted LOS analysis suggested that diabetic patients incurred, on average, a 0.4- day longer postprocedural LOS than nondiabetic patients.
DISCUSSION
Our objective was to examine the impact of a physician- led multifactor CQI intervention in the catheterization laboratory on economic and clinical outcomes. Study results indicate that CQI implementation favorably affected overall costs of providing care. Despite demonstrably higher clinical risk factors and greater utilization of new high-cost technologies, observed costs were not only contained but were 12% lower for patients undergoing PCI post-CQI compared with patients undergoing PCI pre-CQI. With adjustment for potential confounding factors, estimated costs were nearly 28% lower for PCI patients treated post-CQI. An expected LOS reduction of nearly 1 day contributed to this predicted cost differential. Results of clinical analyses indicate that these cost savings were obtained while maintaining and perhaps even improving quality of care. Similar rates of procedural success without in-hospital major complications (90%) were observed pre-CQI and post-CQI, and did not differ statistically in adjusted analyses. Interestingly, the adjusted ORs for in-hospital death or any MI, and for death, any MI, or target vessel revascularization were significantly reduced post-CQI.
Academic medical centers face unique challenges. Research and education are vital components of the academic medical enterprise, and access to new technology often is perceived as a critical component of maintaining a state-of-the-art delivery system capable of providing high-quality care and successfully competing for patient referrals. To maintain the financial viability of their tertiary-care cardiac services, many academic medical centers have focused on CQI and other process measures aimed at quality improvement and expense management through a reduction in complication rates.17-21 Similar to our efforts, some academic medical centers have complemented these strategies with cost reduction activities such as competitive bidding for vendors of cardiac supplies, reducing pharmacy costs, and optimizing allocation of personnel.22,23 Our results suggest that ongoing attention to expense management (rather than just volume increases) should be the cornerstone of successful practice management in technologically intensive practices.
Early research by Hashimoto et al indicated that CQI focused on postprocedural anticoagulation protocols reduced complication rates and LOS for patients undergoing coronary interventions.20 More recently, researchers at the University of Michigan reported an annualized saving of $1.3 million in coronary intervention and electrophysiological treatments as a result of competitive bidding for high-cost items.22 Furthermore, an unadjusted hospital cost savings of $1223 per case and an adjusted hospital cost savings of $460 per PCI case were estimated as a result of competitive bidding combined with pathway implementation focused on postprocedural use of heparin and arterial sheath removal.23 Assuming that competitive bidding resulted in a similar price reduction, the greater hospital cost savings realized at our center may reflect a more comprehensive CQI intervention and/or greater clinical pathway acceptance.
In our experience, the process of developing and implementing change was important to the outcome. Physician involvement in the form of leadership of the multidisciplinary team was critical to the success of the CQI effort. The importance of physician involvement is validated and evidenced by the guidelines developed by the Laboratory Performance Standards Committee of the Society for Cardiac Angiography & Interventions,24 which emphasize the need for leadership by clinicians. Although an essential contribution is made by administrative disciplines (quality process experts, management accountants, data experts, statisticians), critical evaluation of old practices and initiation, integration, and championing of new processes in clinical practice are best done by practicing clinicians. Further, the process of implementing the resource protocol brought utilization patterns and associated costs to the attention of practicing physicians, resulting in a culture of sensitivity to expense management while ensuring the same high standards of care. It is plausible that the magnitude of success with cost containment may not extend to other contexts, where such commitment and buy-in by physicians are lacking.
Study Limitations
Because this was an observational study, spurious correlations cannot be totally excluded. For example, the post-CQI cohort may have been systematically different in an unaccounted-for manner, which caused them to incur fewer costs than the pre-CQI cohort. We had the advantage, however, of a large sample size and comprehensive clinical and administrative data, affording adjustment for a wealth of potential confounding factors. Our study design allowed for examination of the effect of the CQI and cohort characteristics on actual clinical care in a large unselected population. Our multivariate analyses highlighted clinical characteristics (eg, indication for PCI, comorbid conditions) that significantly contributed to costs and prolonged LOS. Future practice evaluations and CQI efforts may focus on these identified patient populations of interest.
We recognize that our study is restricted to the experience of a single high-volume referral center in the late 1990s and that current practice has evolved to include drug-eluting stents and new pharmacotherapies. The fundamental dynamics of new technology introduction during times of financial stress remain unchanged, however. The advent of these new high-cost technologies highlights the fact that critical practice evaluation to identify optimal care processes and cost containment still is very applicable to current interventional practice. Other centers may have different economic and quality-of-care concerns, and the specific components of our CQI intervention may not be appropriate in all settings. However, the pressures to contain costs are not restricted to academic medical centers and high-volume referral centers. The process of change we describe may serve as a general guide to CQI effort by way of successful clinical pathway development and implementation, as well as effective vendor management.
Other limitations include the fact that temporal changes in PCI-related costs and outcomes may have influenced our results, although the relatively short study duration makes this much less of a concern. We also recognize a potential bias in ascertainment of MI because routine measurement of cardiac biomarkers was not standard practice during the study. Finally, we did not directly consider the effect of operator volume and provider characteristics in the analyses, despite literature suggesting that facility and provider volume are associated with adverse clinical events after PCI.25-28 Facility and provider volume have been shown to be inversely related to adverse event rates (ie, higher volume, decreased adverse events). It is possible, though, that high-volume institutions experience a learning process similar to the one we explicitly describe and acknowledge in our study. Further research is needed to assess the effect of facility and provider volumes on costs and clinical outcomes before better comparisons can be made between these factors and CQI efforts.
CONCLUSION
Physician-led, multidisciplinary practice management efforts were successful at constraining growth of PCI-related costs of care in an era of rapid introduction of new technology, while maintaining and likely improving quality of care. This case study on the process and outcome of cost containment efforts has important implications for cardiovascular settings, as well as other settings characterized by expensive new healthcare strategies.
Acknowledgments
The authors thank Ryan Lennon, MS, and Marlené Boyd for data and editorial assistance, respectively.
Appendix. Continuous Quality Improvement Process in the Context of Delivery of Percutaneous Coronary Intervention
Charged with expense reduction and quality enhancement, the multidisciplinary group used the Plan- Do-Study-Act (PDSA) cycle model for improving percutaneous coronary intervention (PCI) delivery. The PDSA model begins with improvement teams asking themselves 3 important questions: (1) What are we trying to accomplish? (2) How will we know that a change is an improvement? (3) What changes can we make? The steps in the process for improvement in PCI delivery that were followed by the group are described below.
What are we trying to accomplish?
The teams began by evaluating the potential for improvement and reached agreement on the aim. They established a general goal of identifying and pursuing opportunities for expense reduction while improving current evidence-based practice in PCI delivery. A more specific aim was also defined: to evaluate PCI-related resource utilization in order to understand how costs are generated and to realize the potential to reduce costs of care.
How will we know that a change is an improvement?
To determine whether an implemented change was actually an improvement required background on current PCI clinical practice, associated costs, and established measures of quality. In this regard, the group analyzed inpatient PCIs performed from January to September 1996 (identified under diagnosis- related group [DRG] 112 using diagnosis code 414.01 and procedure codes of 36.01, 36.05, or 36.06) for a breakdown of costs and utilization. Data were extracted from administrative databases for these episodes of interest and were supplemented with data from the PCI registry. The focus of the intervention was limited to uncomplicated PCIs and elective PCIs (defined as procedures done on the day of a non-emergency department admission), due to the high variability and unpredictability associated with urgent and emergent cases. Median-cost cases were used for benchmarking purposes.
Observed utilization patterns, including interoperator variability in utilization and cost, were analyzed and critically reviewed by the work group. Inpatient costs relating to labor (physician, nurse, anesthesia, technologist), facility (catheterization laboratory facility, hospital floor), and pharmaceutical drugs were analyzed to understand the components and relative weighting of the cost structure associated with these cases.
Resource utilization was stratified into a cost-benefit matrix (Table 1), where utilization found to be of little incremental clinical benefit was targeted for elimination or modification. Thus, tests yielding little incremental information and unnecessary inventory were candidates for elimination. Utilization considered of high clinical benefit but also costly (eg, intracoronary stents) was identified as an important candidate for more aggressive expense reduction efforts.
What changes can we make that will result in improvement?
This stage of the improvement process focused on PDSA cycles for learning and improvement, where teams began by increasing knowledge to develop change, developed and tested change on a small scale, and then implemented the change if the tests were favorable. The group identified specific strategies to achieve expense reduction. These included expense management of cardiac supplies via negotiation of share-based contracts with leading vendors, inventory management such as consignment purchasing, minimizing waste due to expiration dates, and electronic tracking of supplies. The teams also focused on the optimization of personnel allocation and operating costs associated with the clinical practice. A clinical pathway for resource use was developed, including an appropriate staffing model with attention to the number and mix of physicians and allied health personnel. Cost containment strategies not in the patients' best interest (eg, reprocessing of single-use items, switching to ionic contrast for radiographic imaging) were not considered.
PDSA cycles of testing were accomplished through education, audit, feedback, and evaluation with a focus on interoperator variability in utilization and costs. These learning cycles and evaluation resulted in more competitive bidding for vendors of cardiac supplies, as well as the development of a final clinical pathway for care and an associated resource protocol for a median DRG 112 PCI case.
From the Division of Cardiovascular Diseases (CSR, DRH), the Division of Endocrinology, Diabetes, Nutrition & Metabolism (CCK), the Department of Health Sciences Research (SSA, EKM, KHL), and the Department of Patient Support Services (MKR), Mayo Clinic College of Medicine, Mayo Clinic, Rochester, Minn.
Abstract presented in part at the 26th Annual Meeting of the Society for Medical Decision Making, Atlanta, Ga, October 17-20, 2004.
Address correspondence to: Charanjit S. Rihal, MD, MBA, Director, Cardiac Catheterization Laboratory, Mayo Clinic, 200 First St SW, Rochester, MN 55905. E-mail: rihal@mayo.edu.
