The effect of hospital discharge price increases on publicly reported measures of quality.

OBJECTIVE: To determine if increases in hospital discharge prices are associated with improvements in clinical quality or patient experience. DATA SOURCES: This study used Medicare cost report data and publicly available Medicare.gov Care Compare quality measures for approximately 3000 short-term care general hospitals between 2011 and 2018. STUDY DESIGN: We separately regressed quality measure scores on a lag of case mix adjusted discharge price, hospital fixed effects, and year indicators. Clinical quality measures included 30-day readmission rates for acute myocardial infarction, chronic obstructive pulmonary disease, heart failure, hip and knee replacement, and pneumonia; risk-adjusted 30-day mortality rates for acute myocardial infarction, chronic obstructive pulmonary disease, heart failure, and stroke; and 90-day complication rate for hip and knee replacement. Patient experience measures included the summary star rating and 10 domain measures reported through the Hospital Consumer Assessment of Healthcare Providers and Systems survey. We tested for heterogeneous effects by hospital ownership, number of beds, the commercial share of overall discharges, and market concentration. DATA COLLECTION/EXTRACTION METHODS: We linked hospitals identified in Medicare cost reports to Medicare.gov Care Compare quality measures. We excluded hospitals for which we could not identify a discharge price or that had an unrealistic price. PRINCIPAL FINDINGS: There was no positive association between lagged discharge price and any clinical quality measure. For patient experience measures, a 2% increase in discharge price was not associated with overall patient satisfaction but was associated with small, statistically significant increases ranging from 0.01% to 0.02% (relative to mean scores) for seven of ten domain measures. There was a positive association for five of ten patient experience measures in competitive markets and one measure in both moderately concentrated and heavily concentrated markets. CONCLUSIONS: We found no evidence that hospitals use higher prices to make investments in clinical quality; patient experience improved, but only negligibly.

Claims-Based Reporting of Post-Operative Visits for Procedures with 10- or 90-Day Global Periods: Updated Results Using Calendar Year 2018 Data.

Medicare payments for most surgical procedures cover both procedures and post-operative visits occurring within a global period of either ten or 90 days following procedures. There have been concerns that fewer post-operative visits are provided than the number of post-operative visits considered when the procedure was valued. To help inform accurate valuation of procedures with global periods, the Centers for Medicare & Medicaid Services (CMS) required select practitioners to report on post-operative visits after select procedures with 10- or 90-day global periods. The authors of this article summarize patterns of post-operative visits for procedures furnished during calendar year 2018, building on prior research that analyzed data for procedures with July 1, 2017, through June 30, 2018, service dates. During calendar year 2018, 96.5 percent of procedures with 10-day global periods did not have an associated post-operative visit. Approximately two-thirds of procedures with 90-day global periods had an associated post-operative visit; however, the ratio of observed to expected post-operative visits provided for 90-day global period procedures was only 0.38. Underreporting of post-operative visits might be driving these low rates. However, in sensitivity analyses limited to practitioners who were actively reporting their post-operative visits, post-operative patterns were largely similar to the main analysis. Collectively, these findings suggest that a large share of expected post-operative visits are not delivered, and that underreporting is unlikely to fully explain the low ratio of expected post-operative visits provided.

Claims-Based Reporting of Post-Operative Visits for Procedures with 10- or 90-Day Global Periods: Updated Results Using Calendar Year 2019 Data.

Medicare payments for most surgical procedures cover both procedures and post-operative visits occurring within a global period of either 10 or 90 days following procedures. There have been concerns that fewer post-operative visits are provided than the number of post-operative visits considered when the procedure was valued. To help inform accurate valuation of procedures with global periods, the Centers for Medicare & Medicaid Services (CMS) required select practitioners to report on post-operative visits after select procedures with 10- or 90-day global periods. The authors of this article summarize patterns of post-operative visits for procedures furnished during calendar year 2019, building on prior research that analyzed data for procedures furnished from July 1, 2017, through June 30, 2018, and for the entire 2018 calendar year. During calendar year 2019, 96.5 percent of procedures with 10-day global periods did not have an associated post-operative visit. Approximately two-thirds of procedures with 90-day global periods had an associated post-operative visit; however, the ratio of observed to expected post-operative visits provided for 90-day global period procedures was only 0.38. Underreporting of post-operative visits might be driving these low rates. However, in sensitivity analyses limited to practitioners who were actively reporting their post-operative visits, post-operative patterns were largely similar to the main analysis. Collectively, these findings suggest that many expected post-operative visits are not delivered and that underreporting is unlikely to fully explain the low ratio of expected post-operative visits provided.

Practice Expense Data Collection and Methodology: Phase II Final Report.

Each year, Medicare allocates tens of billions of dollars for indirect practice expense (PE) across services on the basis of data from the Physician Practice Information (PPI) Survey, which reflects 2006 expenses. Because these data are not regularly updated, and because there have been significant changes in the U.S. economy and health care system since 2006, there are concerns that continued reliance on PPI Survey data might result in PE payments that do not accurately capture the resources that are typically required to provide services. In this final phase of a study on PE methodology, the authors address how the Centers for Medicare & Medicaid Services (CMS) might improve the methodology used in PE rate-setting, update data that inform PE rates, or both. The authors conclude that this information is best provided by a survey; therefore, they focus on the advantages and disadvantages of survey-based approaches. They also describe the use of a lean model survey instrument, as well as partnering with another agency to collect data. Finally, the authors describe a virtual town hall meeting held in June 2021 to give stakeholders an opportunity to provide feedback on PE data collection and rate-setting. The system of data and methods that CMS uses to support PE rate-setting is complex; thus, CMS must take into account a number of competing priorities when considering changes to the system. With this in mind, the authors offer a number of near- and longer-term recommendations.

Using Claims-Based Estimates of Post-Operative Visits to Revalue Procedures with 10- and 90-Day Global Periods: Updated Results Using Calendar Year 2019 Data.

Medicare payment for many health care procedures covers not only the procedure itself but also most post-operative care over a fixed period of time (the ""global period""). The Centers for Medicare & Medicaid Services (CMS) sets payment rates assuming that a certain number and type of post-operative visits specific to each procedure typically occur. This article describes how CMS might use claims-based data on the number of post-operative visits to adjust valuation for procedures with 10- and 90-day global periods. There are links between the number of bundled post-operative visits and the components of valuation addressed in this study: work, practice expense (PE), and malpractice relative value units (RVUs). There is some ambiguity regarding how a reduction in post-operative visits translates into changes in work RVUs. In contrast, a reduction in post-operative visits has clear implications on physician time and direct PE. Changes in physician work, physician time, and direct PE will, in turn, affect the allocation of pools of PE and malpractice RVUs to individual services. The idiosyncrasies of the resource-based relative value scale system used to determine payment for Medicare services result in some ambiguity about how procedures should be revalued to reflect reductions in post-operative visits. These results may inform further policy development around revaluation for global procedures.

Variation in Estimated Surgical Procedure Times Across Patient Characteristics and Surgeon Specialty.

Importance: The time involved in performing a procedure is a key factor in determining physician payments by Medicare. However, there are long-standing concerns regarding the accuracy of the time estimates generated by the American Medical Association/Specialty Society Relative Value Scale Update Committee surveys that are used in the valuation process, and there have been calls to use other data sources to estimate procedure times. Objective: To compare estimated procedure times that come from claims with the times used in Medicare's valuation process. Design and Setting: Building off prior work using Medicare fee-for-service claims, procedure times were estimated from linked anesthesia claims data for 1349 different Current Procedure Terminology codes that are typically performed with anesthesia. All procedures in the nation performed in 2018 for Medicare fee-for-service beneficiaries were included in the analysis. These estimated times were compared with the times used in the valuation process. Analysis took place from February to November 2021. Main Outcomes and Measures: Estimated procedure times using anesthesia claims were compared with the procedure time used in valuation by calculating an estimated-to-valuation procedure time ratio for each code. The valuation procedure time is publicly reported by Medicare. The mean and median ratio are presented over all procedures and for select high-volume codes as well as by patient characteristics (age, sex, and risk score) and specialty of the physician performing the procedure. Results: Across 4.9 million procedures in this analysis, the mean estimated procedure time was 27% lower than the time used in the valuation process. There were notable exceptions, for which the mean estimated procedure time equaled or exceeded the valuation time including total hip arthroplasty (5% longer) and total knee arthroplasty (equal duration). Within a given code, older patients and those with more illness had longer procedure times. There was substantial variation across specialties in the percent difference between mean estimated and valuation procedure times ranging from gastroenterology (36% shorter) and ophthalmology (35% shorter) to cardiac surgery (2% longer) and thoracic surgery (7% longer). Conclusions and Relevance: Claims-based procedure times could be used to improve the accuracy of valuations for procedures.

Smartphone Application Allowing Physicians to Call Patients Associated with Increased Physician Productivity.

BACKGROUND: Telehealth and other technologies that enable remote patient-physician communication technologies have widespread use among physicians and other health care providers, but the impacts of these technologies on physician productivity are not well known. OBJECTIVE: To determine whether a HIPAA-compliant application that allows physicians to call patients from their personal cell phones is associated with an increase in physician productivity. DESIGN, SETTING, AND PARTICIPANTS: We used a 100% sample of Medicare claims and longitudinal physician-level data to examine whether physician use of a smartphone application that enables physician-patient phone calls is associated with changes in Medicare patient volume and services. We compared early adopters of the application, 31,577 physicians providing Part B services who initiated use of the application between January 2014 and December 2017, with later adopters, 22,988 physicians who initiated use between January 2018 and July 2019. MAIN MEASURES: Physician productivity was measured as total Medicare Part B beneficiaries, total Part B services provided, the number of Part B beneficiaries with any evaluation and management (E&M) service, the total number of E&M services provided, and the average number of E&M services provided per beneficiary. KEY RESULTS: Following application use, there was a 0.52 increase (95% CI: 0.19 to 0.85) in the monthly number of Part B beneficiaries seen. This difference translates to a 0.8% increase in Part B beneficiaries. Similar increases were observed for the number of unique beneficiaries for which the physician provided E&M services-a 0.50 increase (95% CI: 0.27 to 0.73) or 1.2%. There was a 0.43 increase (95% CI: 0.07 to 0.78) in monthly E&M services (0.7% increase). CONCLUSIONS: Physicians who used a freely available smartphone application modestly increased their total Medicare beneficiary volume and total number of E&M services provided, suggesting potential improvements in physician productivity.

Do health systems respond to the quality of their competitors?

OBJECTIVES: Some large employers and healthcare analysts have advocated for retail competition that relies on providers competing on performance metrics to improve care quality. Using publicly available performance measures, we determined whether health systems increased the quality of diabetes care provided by their clinics based on performance relative to competitors. STUDY DESIGN: Our analysis examined publicly reported performance measures of diabetes care from 2006 to 2013 for clinics in Minnesota health systems. METHODS: We obtained data for 654 clinics, of which 572 publicly reported diabetes care performance. Because some clinics did not report performance, we estimated a Heckman selection model. First, we predicted whether or not clinics reported performance. Second, we estimated the effect of relative performance (a clinic's performance minus the mean performance of clinics in competing health systems) on clinic performance using the results of the reporting model to control for selection into the sample of reporting clinics. RESULTS: Although diabetes care performance improved during our study, health systems did not differentially improve the diabetes care performance of their clinics performing worse than clinics in competing systems. This result indicates divergence between high-performing and low-performing clinics. This result does not appear to be due to risk selection. CONCLUSIONS: Publicly reporting quality information did not incentivize health systems to increase the performance of their clinics with lower performance than competitors, as would be expected under retail competition. Our results do not support strategies that rely on competition on publicly reported performance measures to improve quality in diabetes care management.

Do Health Systems Have Consistent Performance Across Locations and Is Consistency Associated With Higher Performance?

This study addresses whether health systems have consistent diabetes care performance across their ambulatory clinics and whether increasing consistency is associated with improvements in clinic performance. Study data included 2007 to 2013 diabetes care intermediate outcome measures for 661 ambulatory clinics in Minnesota and bordering states. Health systems provided more consistent performance, as measured by the standard deviation of performance for clinics in a system, relative to propensity score-matched proxy systems created for comparison purposes. No evidence was found that improvements in consistency were associated with higher clinic performance. The combination of high performance and consistent care is likely to enhance a health system's brand reputation, allowing it to better mitigate the financial risks of consumers seeking care outside the organization. These results suggest that larger health systems are most likely to deliver the combination of consistent and high-performance care. Future research should explore the mechanisms that drive consistent care within health systems.

Health System Consolidation and Diabetes Care Performance at Ambulatory Clinics.

OBJECTIVE: We addressed two questions regarding health system consolidation through the acquisition of ambulatory clinics: (1) Was increasing health system size associated with improved diabetes care performance and (2) Did the diabetes care performance of acquired clinics improve postacquisition? DATA SOURCES/STUDY SETTING: Six hundred sixty-one ambulatory clinics in Minnesota and bordering states that reported performance data from 2007 to 2013. STUDY DESIGN: We employed fixed effects regression to determine if increased health system size and being acquired improved clinics' performance. Using our regression results, we estimated the average effect of consolidation on the performance of clinics that were acquired during our study. DATA COLLECTION/EXTRACTION METHODS: Publicly reported performance data obtained from Minnesota Community Measurement. PRINCIPAL FINDINGS: Acquired clinics experienced performance improvements starting in their third year postacquisition. By their fifth year postacquisition, acquired clinics had 3.6 percentage points (95 percent confidence interval: 2.0, 5.1) higher performance than if they had never been acquired. Increasing health system size was associated with slight performance improvements at the end of the study. CONCLUSIONS: Health systems modestly improved the diabetes care performance of their acquired clinics; however, we found little evidence that systems experienced large, system-wide performance gains by increasing their size.

The effect of participation in an incentive-based wellness program on self-reported exercise.

Employers are increasingly trying to promote healthy behaviors, including regular exercise, through wellness programs that offer financial incentives. However, there is limited evidence that these types of programs affect exercise habits within employee populations. In this study, we estimate the effect of participation in an incentive-based wellness program on self-reported exercise. Since 2008, the University of Minnesota's Fitness Rewards Program has offered a $20 monthly incentive to encourage fitness center utilization among its employees. Using 2006 to 2010 health risk assessments and university administrative files for 2972 employees, we conducted a retrospective cohort study utilizing propensity score methods to estimate the effect of participation in the Fitness Rewards Program on self-reported exercise days per week from 2008 to 2010. On average, participation in the program led to an increase of 0.59 vigorous exercise days per week (95% Confidence Interval: 0.42, 0.78) and 0.43 strength-building exercise days per week (95% Confidence Interval: 0.31, 0.58) in 2008 for participants relative to non-participants. Increases in exercise persisted through 2010. Employees reporting less frequent exercise prior to the program were least likely to participate in the program, but when they participated they had the largest increases in exercise compared to non-participants. Offering an incentive for fitness center utilization encourages higher levels of exercise. Future policies may want to concentrate on how to motivate participation among individuals who are less frequently physically active.

Initiation and Maintenance of Fitness Center Utilization in an Incentive-Based Employer Wellness Program.

OBJECTIVE: To investigate the initiation and maintenance of participation in an employer-based wellness program that provides financial incentives for fitness center utilization. METHODS: Using multivariate analysis, we investigated how employees' demographics, health status, exercise-related factors, and lifestyle change preferences affect program participation. RESULTS: Forty-two percent of eligible employees participated in the program, and 24% earned a $20 incentive at least once by utilizing a gym eight times or more in a month. On average, participants utilized fitness centers 7.0 months each year and earned credit 4.5 months. Participants' utilization diminished after their first year in the program. CONCLUSIONS: Factors associated with initiation and maintenance of fitness center utilization were similar. Declining utilization over time raises concern about the long-run effectiveness of fitness-focused wellness programs. Employers may want to consider additional levers to positively reinforce participation.

Acupuncture provides short-term pain relief for patients in a total joint replacement program.

OBJECTIVE: Given the risks of opioid medications, nonpharmacological strategies should be considered for total joint replacement patients. We investigated acupuncture as an adjunct therapy for postsurgical pain management in a total joint replacement program by examining which total hip and knee replacement patients elected to receive acupuncture and the effect of acupuncture on short-term pain. DESIGN: A total joint replacement program using fast-track physiotherapy offered elective postsurgical acupuncture to all patients, at no additional cost, as an adjunct therapy to opioids for pain management. SETTING: The Joint Replacement Center at Abbott Northwestern Hospital, a 630-bed teaching and specialty hospital in Minneapolis, Minnesota from 2010 to 2012. SUBJECTS: Our sample included 2,500 admissions of total hip (THR) and total knee replacement (TKR) patients. METHODS: Self-reported pain was assessed before and after acupuncture using a 0-10 scale and categorized as none/mild (0-4) and moderate/severe pain (5-10). RESULTS: Seventy-five percent of admissions included acupuncture. Women (Odds Ratio: 1.48, 95% Confidence Interval (CI): 1.22, 1.81) had higher odds of receiving acupuncture compared to men, and nonwhite patients (Odds Ratio: 0.55, 95% CI: 0.39, 0.78) had lower odds of receiving acupuncture compared to white patients. Average short-term pain reduction was 1.91 points (95% CI: 1.83, 1.99), a 45% reduction from the mean prepain score. Forty-one percent of patients reported moderate/severe pain prior to receiving acupuncture, while only 15% indicated moderate/severe pain after acupuncture. CONCLUSIONS: Acupuncture may be a viable adjunct to pharmacological approaches for pain management after THR or TKR.

Public reporting and the evolution of diabetes quality.

We address three questions related to public reports of diabetes quality. First, does clinic quality evolve over time? Second, does the quality of reporting clinics converge to a common standard? Third, how persistent are provider quality rankings across time? Since current methods of public reporting rely on historic data, measures of clinic quality are most informative if relative clinic performance is persistent across time. We use data from the Minnesota Community Measurement spanning 2007-2012. We employ seemingly-unrelated regression to measure quality improvement conditional upon cohort effects and changes in quality metrics. Basic autoregressive models are used to measure quality persistence. There were striking differences in initial quality across cohorts of clinics and early-reporting cohorts maintained higher quality in all years. This suggests that consumers can infer, on average, that non-reporting clinics have poorer quality than reporting clinics. Average quality, however, improves slowly in all cohorts and quality dispersion declines over time both within and across cohorts. Relative clinic quality is highly persistent year-to-year, suggesting that publicly-reported measures can inform consumers in choice of clinics, even though they represent measured quality for a previous time period. Finally, definition changes in measures can make it difficult to draw appropriate inferences from longitudinal public reports data.

The effectiveness of integrative medicine interventions on pain and anxiety in cardiovascular inpatients: a practice-based research evaluation.

BACKGROUND: Pain and anxiety occurring from cardiovascular disease are associated with long-term health risks. Integrative medicine (IM) therapies reduce pain and anxiety in small samples of hospitalized cardiovascular patients within randomized controlled trials; however, practice-based effectiveness research has been limited. The goal of the study is to evaluate the effectiveness of IM interventions (i.e., bodywork, mind-body and energy therapies, and traditional Chinese medicine) on pain and anxiety measures across a cardiovascular population. METHODS: Retrospective data obtained from medical records identified patients with a cardiovascular ICD-9 code admitted to a large Midwestern hospital between 7/1/2009 and 12/31/2012. Outcomes were changes in patient-reported pain and anxiety, rated before and after IM treatments based on a numeric scale (0-10). RESULTS: Of 57,295 hospital cardiovascular admissions, 6,589 (11.5%) included IM. After receiving IM therapy, patients averaged a 46.5% (p-value < 0.001) decrease in pain and a 54.8% (p-value < 0.001) decrease in anxiety. There was no difference between treatment modalities on pain reduction; however, mind-body and energy therapies (p-value < 0.01), traditional Chinese medicine (p-value < 0.05), and combination therapies (p-value < 0.01) were more effective at reducing anxiety than bodywork therapies. Each additional year of age reduced the odds of receiving any IM therapy by two percent (OR: 0.98, p-value < 0.01) and females had 96% (OR: 1.96, p-value < 0.01) higher odds of receiving any IM therapy compared to males. CONCLUSIONS: Cardiovascular inpatients reported statistically significant decreases in pain and anxiety following care with adjunctive IM interventions. This study underscores the potential for future practice-based research to investigate the best approach for incorporating these therapies into an acute care setting such that IM therapies are most appropriately provided to patient populations.

What is the cost of quality for diabetes care?

Increasing the quality of care and reducing cost growth are core objectives of numerous private- and public-sector performance improvement initiatives. Using a unique panel data set for a commercially insured population and multivariate regression analysis, this study examines the relationship between medical care spending and diabetes-related quality measures, including provider-initiated processes of care and patient-dependent quality activities. Empirical evidence generated from this analysis of the relationship between a comprehensive set of diabetes quality measures and diabetes-related spending does not lend support for the assumption that high-quality preventive and primary care combined with effective patient self-management can lead to lower costs in the near term. Finally, we find no relationship between adjusted spending and intermediate clinical outcomes (e.g., HbA1c level) measured at the clinic level.

Effects of integrative medicine on pain and anxiety among oncology inpatients.

BACKGROUND: Few studies have investigated the effectiveness of integrative medicine (IM) therapies on pain and anxiety among oncology inpatients. METHODS: Retrospective data obtained from electronic medical records identified patients with an oncology International Classification of Diseases-9 code who were admitted to a large Midwestern hospital between July 1, 2009 and December 31, 2012. Outcomes were change in patient-reported pain and anxiety, rated before and after individual IM treatment sessions, using a numeric scale (0-10). RESULTS: Of 10948 hospital admissions over the study period, 1833 (17%) included IM therapy. Older patients had reduced odds of receiving any IM therapy (odds ratio [OR]: 0.97, 95% confidence interval [95% CI] = 0.96 to 0.98) and females had 63% (OR: 1.63, 95% CI = 1.38 to 1.92) higher odds of receiving any IM therapy compared with males. Moderate (OR: 1.97, 95% CI = 1.61 to 2.41), major (OR: 3.54, 95% CI = 2.88 to 4.35), and extreme (OR: 5.96, 95% CI = 4.71 to 7.56) illness severity were significantly associated with higher odds of receiving IM therapy compared with admissions of minor illness severity. After receiving IM therapy, patients averaged a 46.9% (95% CI = 45.1% to 48.6%, P <.001) reduction in pain and a 56.1% (95% CI = 54.3% to 58.0%, P <.001) reduction in anxiety. Bodywork and traditional Chinese Medicine therapies were most effective for reducing pain, while no significant differences among therapies for reducing anxiety were observed. CONCLUSIONS: IM services to oncology inpatients resulted in substantial decreases in pain and anxiety. Observational studies using electronic medical records provide unique information about real-world utilization of IM. Future studies are warranted and should explore potential synergy of opioid analgesics and IM therapy for pain control.

Resource use trajectories for aged medicare beneficiaries with complex coronary conditions.

OBJECTIVE: To use coronary revascularization choice to illustrate the application of a method simulating a treatment's effect on subsequent resource use. DATA SOURCES: Medicare inpatient and outpatient claims from 2002 to 2008 for patients receiving multivessel revascularization for symptomatic coronary disease in 2003-2004. STUDY DESIGN: This retrospective cohort study of 102,877 beneficiaries assessed survival, days in institutional settings, and Medicare payments for up to 6 years following receipt of percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG). METHODS: A three-part estimator designed to provide robust estimates of a treatment's effect in the setting of mortality and censored follow-up was used. The estimator decomposes the treatment effect into effects attributable to survival differences versus treatment-related intensity of resource use. PRINCIPAL FINDINGS: After adjustment, on average CABG recipients survived 23 days longer, spent an 11 additional days in institutional settings, and had cumulative Medicare payments that were $12,834 higher than PCI recipients. The majority of the differences in institutional days and payments were due to intensity rather than survival effects. CONCLUSIONS: In this example, the survival benefit from CABG was modest and the resource implications were substantial, although further adjustments for treatment selection are needed.

Evaluating the effectiveness of a rapidly adopted cardiovascular technology with administrative data: the case of drug-eluting stents for acute coronary syndromes.

BACKGROUND: Instrumental variable (IV) methods can correct for unmeasured confounding when using administrative (claims) data for cardiovascular outcomes research, but difficulties identifying valid IVs have limited their use. We evaluated the safety and efficacy of drug-eluting coronary stents (DES) compared with bare-metal stents (BMS) for Medicare beneficiaries with acute coronary syndromes using the rapid uptake of DES in clinical practice as an instrument. We compared results from IV with those from propensity score matching (PSM) and multivariable regression models. METHODS: This is a retrospective cohort study involving 62,309 fee-for-service beneficiaries 66 years and older treated with coronary stenting between May 2003 and February 2004. Outcomes were measured for 46 months after revascularization using claims data. RESULTS: Recipients of DES were younger, had a lower prevalence of myocardial infarction, and had fewer comorbidities compared with BMS recipients. Use of DES was associated with lower rates of mortality by PSM (hazard ratio [HR] 0.80, CI 0.77-0.83) but not by IV (HR 0.99, CI 0.87-1.11). Instrumental variable models estimated a larger reduction in repeat revascularization (HR 0.76, CI 0.63-0.89) than did PSM (HR 0.90, CI 0.87-0.93). CONCLUSIONS: Based on IV analysis, the increased utilization of DES relative to BMS among Medicare beneficiaries with acute coronary syndrome is associated with reduced rates of repeat revascularization and no difference in mortality. Instrumental variable approaches provide a useful complement to conventional approaches to cardiovascular outcomes research with administrative data.

Stress imaging use and repeat revascularization among medicare patients with high-risk coronary artery disease.

The optimal use of stress testing after coronary revascularization remains unclear, and overuse of stress testing might increase the rates of repeat revascularization. We analyzed the association at both the patient and regional level between the use of stress testing and repeat revascularization for a cohort of Medicare beneficiaries receiving revascularization within 30 days of an admission for symptomatic coronary artery disease. The sample consisted of 219,748 Medicare beneficiaries aged >65 years who received percutaneous coronary intervention or cardiac bypass artery grafting after hospital admission for symptomatic coronary artery disease in 2003 to 2004. Medicare claims data through 2008 identified the use of stress testing and repeat revascularization. The associations between the cumulative incidence of stress testing and repeat revascularization were analyzed using linear regression analysis. Within 6 years of the initial revascularization, the cumulative incidence of events was 0.61 for stress testing and 0.23 for repeat revascularization. Most (53.1%) repeat revascularizations were preceded by a stress test. Only 10.3% of repeat revascularization procedures were preceded by myocardial infarction. The 4-year cumulative incidence of repeat revascularization and stress testing varied between the Hospital Referral Regions represented by the sample, and the positive correlation between the rates by the health referral region accounted for only a small portion of the total health referral region variation in revascularization rates. In conclusion, stress testing is commonly performed among Medicare patients after the initial revascularization, and most repeat procedures are performed for stable coronary artery disease. The variation in stress testing patterns only explained a modest fraction of the regional variation in the repeat revascularization rates.