This article describes research related to the design of payment models for ambulatory gastroenterology and cardiology services for possible testing by the Center for Medicare and Medicaid Innovation (CMMI) at the Centers for Medicare and Medicaid Services (CMS). Gastroenterology and cardiology services are common and costly among Medicare beneficiaries. Episode-based payment, which aims to create incentives for high-quality, low-cost care, has been identified as a promising alternative payment model (Calsyn and Emanuel, 2014; Miller et al., 2011; Mechanic, 2011; Mechanic and Altman, 2009).
Based on evidence from environmental scans of gastroenterology and cardiology care payment reform options (McClellan et al., 2014a; McClellan et al., 2014b) and feedback from stakeholder interviews and technical expert panels convened by the Brookings Institution and the MITRE Corporation (MITRE), CMS chose to investigate the possible development of episode-based payment models for outpatient gastroenterology and cardiology procedures. The models would be designed for testing in the traditional Medicare fee-for-service (FFS) program (Parts A and B). The details of the models have yet to be determined.
CMS asked MITRE and RAND to conduct analyses to inform design decisions related to episode-based gastroenterology and cardiology models for Medicare beneficiaries undergoing selected procedures. In particular, this study focuses on analyses of Medicare FFS claims data related to the care settings of gastrointestinal (GI) episodes for colonoscopy and upper endoscopy and cardiology episodes for percutaneous coronary intervention (PCI) and catheterization; patterns of spending during and surrounding the episodes; and the characteristics of patients receiving and practices providing the services. The analyses are intended to support decision-making related to model design and are not specific to any particular design or feature of a payment model.
CMS faces a series of decisions as it considers new payment models for gastroenterology and cardiology services. The premise of our study—informed by input from CMS—is that the payment models under consideration would take the form of an episode-based payment rather than a case management payment targeted to specialists or another approach. With this in mind, key CMS decisions include:
- Episode definition: Which gastroenterology and cardiology procedures will serve as index procedures that anchor an episode?
- Index procedure payment rate adjustments: To what extent should payment rates vary across service settings? How can the payment model accommodate multiple eligible index procedures?
- Payment model scope: Which services are included in the episode-based payment model?
- Provider eligibility for the payment model: Which providers should be eligible to participate in the payment model?
- Patient eligibility for the payment model: Which patients should be eligible to participate in the payment model?
The study population included Medicare FFS beneficiaries receiving at least one of the gastroenterology and cardiology study procedures identified by CMS and RAND, which included colonoscopy, upper GI endoscopy, cardiac catheterization, and PCI. The study sample was drawn from a 2012 100-percent sample of national Medicare FFS claims files, including the Carrier, Outpatient hospital, and Medicare Provider Analysis and Review (MedPAR) files. We defined a single study procedure per beneficiary as the “index procedure” that would potentially trigger an episode of care; for beneficiaries with more than one eligible index procedure during a nine-day window, we created a category for other “eligible procedures.” We matched facility and professional claim lines for all index procedures and assigned a place of service based on the facility claim (or Carrier claim for office-based services). We identified the practice providing each index procedure using the Tax Identification Number (TIN) on the professional claim. The final analytic data sets included 3,333,814 gastroenterology index procedures and 453,843 cardiology index procedures.
We then identified health care services provided to patients in a nine-day episode around each index procedure. We classified all Carrier, Outpatient, and MedPAR claims provided during these episodes to categories including inpatient care, emergency department care, imaging, laboratory tests, pathology, anesthesiology, evaluation and management visits, and others. We calculated the number of claims and total payments in each service category for each episode, both totaled across the entire nine-day episode and for each day in the episode. We also calculated the proportion of episodes that experienced at least one claim in each service category, again over the entire nine-day episode and for each day individually. We selected a nine-day maximum episode duration in consultation with CMS and members of technical expert panels convened by the Brookings Institution. Most services related to the study procedures are expected to fall within these nine-day periods. We also report results by day in order to allow for calculations of spending and utilization over episode lengths varying from one to nine days.
Summary of Main Gastroenterology Results
Colonoscopy procedures were more common than endoscopy procedures as index procedures. A significant share (18 percent) of colonoscopy index procedures were coded as screening procedures (as defined by the authors based on Healthcare Common Procedure Coding System [HCPCS] code), and an additional 15 percent were initiated as screening procedures but were converted to diagnostic and therapeutic procedures, as indicated by the “-PT” HCPCS modifier on the diagnostic/therapeutic procedure claim.
We found that the majority of gastroenterology index procedures were delivered in the hospital outpatient (HOPD) and ambulatory surgical center (ASC) settings. On average, ASC payment rates were significantly lower than HOPD payment rates. We found that Medicare payments for gastroenterology index procedures would be about 16 percent lower overall if HOPD procedures were reimbursed at the lower ASC payment rates ($1.80 versus $2.15 billion). Some states used ASCs much less frequently than other states, and ASC use was less common in rural areas.
We described several patterns in the utilization of and payments for non-index procedure services in the days immediately before and following index procedures. One consistent observation is that utilization and payments for non-index services are concentrated on the date of service of the index procedure itself. We identified differences between the patterns of utilization of non-index services for episodes involving a single eligible gastroenterology procedure and multiple eligible gastroenterology index procedures. Payments for non-index services were lower for episodes with a single eligible index procedure. We also found differences in utilization patterns between screening and non-screening episodes. Screening episodes involved lower per-episode utilization and payments for non-index services and fewer non-index services other than anesthesiology. Episodes involving diagnostic and therapeutic index procedures involved higher utilization rates and payments for non-index services, including surgical pathology services.
We identified 11,140 practices that provided gastroenterology index procedures. Many of these had low volumes; 7,808 practices provided at least 20 index procedures, and these accounted for 99.4 percent of total index procedures. In 60 percent of practices, a single physician provided all of the index procedures. Nearly 38 percent of practices used HOPDs exclusively, while fewer than 7 percent of practices provided these procedures entirely in non-hospital settings. There was substantial variation between practices in the average total payments for services during episodes (interquartile range for upper GI endoscopy, $989–$1,469; for colonoscopy, $921–$1,232). Larger practices tended to have lower average total episode payments.
We found limited variation in payments for the index procedure and other multiple eligible index procedures occurring within an episode for colonoscopy and endoscopy episodes. However, Medicare payments for other services were considerably higher for certain subsets of beneficiaries included in the analysis—particularly, those with Medicare eligibility through end stage renal disease (ESRD) or ESRD and disability, as determined from the beneficiary summary file. Such beneficiaries are likely to require more extensive health services due to ESRD and potentially other conditions and could be considered for exclusion from the payment model.
Summary of Main Cardiology Results
We found that catheterization procedures were more common than PCI procedures as index procedures. While PCI index procedures accounted for only 20 percent of total index procedures by volume, they were associated with nearly 40 percent of spending on index procedures. Almost all cardiology index procedures were performed in the HOPD setting.
We described several patterns in the utilization of and payments for non-index procedure services in the days immediately before and following index procedures. As in our gastroenterology analysis, utilization and payments for non-index services were concentrated on index procedure dates of service. This suggests that a narrow episode definition is appropriate for an episode-based payment model for catheterization and PCI procedures.
We observed relatively high levels of non-index utilization and spending across a few service categories. Eligible (non-index) procedures, imaging, laboratory tests, ambulatory procedures not otherwise classified (NOC), and ambulatory non-procedural services NOC (such as physician-administered drugs) were the largest non-index payment categories on the index procedure date of service. Cardiology episodes were associated with higher rates and payments for emergency department care on the day before the index procedure compared with gastroenterology episodes, as might be expected with cardiac symptoms. The most noticeable difference in non-index utilization and spending in cardiology episodes compared with gastroenterology episodes was a significant spike in inpatient facility spending on the index procedure date of service, followed by a decline from 1 to 3 days after the index procedure, and then a marked increase from day 4 through day 7. We discuss the utilization of and payments for inpatient facility services separately from other non-index services.
We identified 4,466 practices that provided cardiology index procedures, but many of these had low volumes; 2,868 practices provided at least 20 index procedures, and these accounted for 97 percent of total index procedures. In approximately one-third of practices, a single physician provided all of the index procedures. There was substantial variation between practices in the average total payments for services during episodes (interquartile range for catheterization, $2,851–$4,521; for PCI, $7,434–$9,409). Larger practices tended to have lower average total episode payments.
We found that payments for non-index services during a catheterization episode varied by gender, race and ethnicity, age, and reason for current eligibility. Payments for the index procedure in PCI episodes varied across race and ethnicity. In addition, beneficiaries eligible for Medicare through ESRD incurred 30 to 50 percent higher payment for non-index services during PCI episodes relative to beneficiaries eligible through age or disability alone.
Results of our analyses may inform CMS decisions regarding episode definition, patient and provider eligibility for potential gastroenterology and cardiology payment models, and payment rate adjustments.
Our analyses highlighted several cases where special considerations may be needed in using index procedures to define an episode. First, for colonoscopy procedures, episode definitions would need to account for conversions from screening to diagnostic/therapeutic procedures. Second, procedures in an HOPD that lead directly to inpatient admission may not be identified as index procedures using the methods we applied, since the services may be billed on the inpatient claim. In contrast, a service initiated in an ASC leading directly to inpatient admission would be included.
Analyses of utilization patterns during episodes indicated that extending episode definitions beyond the day of the index procedure could increase administrative complexity and financial risk for potential model participants and would not include a substantial amount of additional services in the payment model. However, despite low frequency and payment, it may be important to monitor specific types of services, such as inpatient care, that could result from complications. This could be a component of quality measurement accompanying episode payment.
In analyses of practice characteristics, we found that a substantial percentage of practices that performed index procedures had a very low volume. CMS could impose a minimum practice volume threshold for participation and include most patients with relevant gastroenterology and cardiology procedures.
In analyses of patient characteristics, we found that patients with ESRD had much higher average payments than other patients; these patients could be excluded from the model or subject to payment adjustments. We did not analyze other types of patients with complex conditions; there may be other patient groups that would require either exclusion from the model or risk adjustment.
Payment Rate Adjustments
Payment differentials by service setting are relevant for gastroenterology procedures but not for the cardiology services we studied, which were provided almost exclusively in HOPDs. We found that the majority of gastroenterology index procedures were delivered in the HOPD and ASC settings. ASC payment rates were significantly lower than HOPD payment rates, which is an intended result of the design of the current ASC and HOPD Medicare payment systems. An episode-based payment model could preserve this payment differential to reflect the higher costs of providing hospital-based care. Alternatively, a new payment model could reduce or eliminate the differential, as recommended by MedPAC (Medicare Payment Advisory Commission, 2013). The argument for reduced differentials is that for some services, such as office visits, the costs are really not different. Much of the debate comes down to whether it is appropriate to pay for hospital overhead costs related to activities such as standby capacity, access for disadvantaged populations, and community outreach through higher prices for HOPD services (Medicare Payment Advisory Commission, 2013).
We also found some differences in utilization of services during episodes between settings. For example, anesthesiology services were used more frequently in ASCs than in HOPDs. Colonoscopy procedures that initiated as screening procedures were converted to diagnostic or therapeutic procedures more frequently in ASCs than in HOPDs or offices. These differences could be due to differences in patient characteristics between settings or differences in practice patterns.
We found that it was common for multiple eligible index procedures to be performed on the same day or in the same episode of care for both gastroenterology and cardiology. Episodes with multiple eligible index procedures had higher spending than episodes with a single index procedure, due to both the payments associated with the multiple eligible index procedures and higher payments for other, non-index services. Under current Medicare payment policy, multiple related procedures performed by the same provider during the same patient visit are subject to discounted payment. In an episode-based payment model, one option would be to develop a single payment rate that, on average, compensates providers for multiple index procedures and all non-index services provided in the episode. This would create incentives to reduce, on the margin, the frequency of additional index procedures and ancillary services. Another option is to develop separate payment rates for episodes with one or multiple index procedures.
The results of this study provide one source of information for consideration in the design of gastroenterology and cardiology payment models. Claims data can provide important information on patterns of health care utilization, but it is crucial to augment such analysis with clinical evidence and practice guidelines. The analyses presented in this report describe the frequency and characteristics of gastroenterology and cardiology index procedures, the practices that delivered index procedures, and the patients that received index procedures. We also describe the volume and payments for services that are delivered in a nine-day episode anchored on index procedures. The results can be used to inform CMS decision-making about the definition of episodes in an episode-based payment model; payment adjustments for service setting, multiple procedures, or other factors; and eligibility for the payment model.
Calsyn, M., and E. J. Emanuel, “Controlling Costs by Expanding the Medicare Acute Care Episode Demonstration,” JAMA Internal Medicine, Vol. 174, No. 9, 2014, pp. 1438–1439. As of November 10, 2014:
McClellan, Mark, Kavita Patel, John O'Shea, Jeffrey Nadel, Andrea Thoumi, and Judith Tobin, Specialty Payment Model Opportunities Assessment and Design: Environmental Scan for Cardiology, Washington, D.C.: Brookings Institution, 2014a. As of November 10, 2014:
McClellan, Mark, Kavita Patel, John O'Shea, Andrea Thoumi, Chelsey Crim, Jeffrey Nadel, Elise Presser, and Judith Tobin, Specialty Payment Model Opportunities Assessment and Design: Environmental Scan for Gastroenterology, Washington, D.C.: Brookings Institution, 2014b. As of November 10, 2014:
Mechanic, Robert E., “Opportunities and Challenges for Episode-Based Payment,” New England Journal of Medicine, Vol. 365, No. 9, 2011, pp. 777–779. As of November 10, 2014:
Mechanic, Robert E., and Stuart H. Altman, “Payment Reform Options: Episode Payment Is a Good Place to Start,” Health Affairs, January 27, 2009, p. hlthaff.28.22.w262.
Medicare Payment Advisory Commission, Report to the Congress: Medicare and the Health Care Delivery System, Washington, D.C., 2013.
Miller, David C., Cathryn Gust, Justin B. Dimick, Nancy Birkmeyer, Jonathan Skinner, and John D. Birkmeyer, “Large Variations in Medicare Payments for Surgery Highlight Savings Potential from Bundled Payment Programs,” Health Affairs, Vol. 30, No. 11, November 1, 2011, pp. 2107–2115. As of November 10, 2014:
The research addressed in this article was conducted in RAND Health, a division of the RAND Corporation, under a subcontract to MITRE.