CMS RELEASES BPCI SECOND ANNUAL EVALUATION REPORT

Transcription

1 CMS RELEASES BPCI SECOND ANNUAL EVALUATION REPORT Introduction On September 19, 2016, CMS released the second annual evaluation report for Models 2-4 of the Bundled Payments for Care Improvement (BPCI) Initiative. The Initiative is designed to test whether linking the payments for all providers involved in delivering an episode of care can reduce Medicare costs while maintaining or improving quality of care. The BPCI models align with the Department of Health and Human Service s goal to shift Medicare payments from rewarding quantity to rewarding quality by creating strong incentives for hospitals and clinicians to deliver better care to patients at a lower cost. The report describes the characteristics of the participants and includes quantitative analyses of Phase 2 participants from the first year of BPCI (October 2013-September 2014) and qualitative analyses of participants that joined during the first seven quarters (October 2013-June 2015). 1 BPCI INITIATIVE OVERVIEW The BPCI Initiative was launched under the authority of the Center for Medicare and Medicaid Innovation. It is designed to reward and incentivize BPCI Awardees, which can include hospitals, physician groups, PAC providers and other entities, to adopt practices that reduce Medicare payments for the bundle of services in the episode relative to a target price that CMS determines based on the provider s historical payments for the same type of episode. When Awardees episode payments are below the target price, they are eligible to receive net payment reconciliation amounts. When Awardees episode payments are above the target price, they may have to return amounts to CMS. As with other alternative payment models, BPCI is designed to reward clinicians and facilities that deliver care more efficiently and effectively. The BPCI Initiative was designed with multiple implementation options, including the threshold choice of whether to participate or not. Beyond that fundamental decision, participants must select their BPCI Model, episodes, and episode length. Four payment models and 48 clinical episodes were offered. The three BPCI Models evaluated in the Report vary as to the bundle definition and payment approach. Each is described briefly below: Model 2: The most comprehensive bundle, which includes the anchor hospital stay, all concurrent professional services, post-discharge services, and readmissions within the chosen episode length of 30, 60, or 90 days. Providers are paid on a fee-for-service basis and total episode payments are reconciled retrospectively. Model 3: Includes services after the anchor hospital discharge, including professional services, post-discharge services, and readmissions within the chose episode length of 30, 60, or 90 days. Providers are paid on a feefor-service basis and total episode payments are reconciled retrospectively against the target price. Model 4: Includes the anchor hospitalization, all concurrent professional services and any readmissions and associated professional services that occur within 30 days of discharge that are not explicitly excluded. Awardees are paid a prospectively determined amount and the Awardees then pay the providers. REPORT OVERVIEW The Report describes the participant characteristics and provides a summative and formative evaluation of the BPCI Initiative based on quantitative analyses of Phase 2 participants that joined BPCI during the first year and qualitative analyses of participants that joined during the first seven quarters. Across the three Models, 94 Awardees entered 1 The first year refers to October 2013 through September 2014, the first full year of the active phase of the initiative, as referenced in the Annual Report, pg. 3.

2 into agreements with CMS, and 227 providers (130 hospitals, 63 SNFs, 28 HHAs, and 4 PGPs) initiated 58,410 episodes of care during the first year of BPCI. The study is observational and the analyses relied on difference-in-difference (DiD) models to evaluate outcomes of beneficiaries associated with BPCI providers compared with beneficiaries who received care from comparison providers. The BPCI intervention population included all episodes initiated in BPCI-participating providers from October 2013 through September The comparison group is similar to the BPCI providers with respect to organization characteristics, market share, and volume of BPCI qualifying admissions, market concentration and size, availability of PAC providers, and case mix for each Model and Episode Initiating (EI) provider type from Medicare providers that were not in Phase 1 or Phase 2 of BPCI. The DiD estimate compares the change in outcomes between the baseline and intervention period for beneficiaries receiving care from BPCI-participating providers relative to that same change for beneficiaries receiving care from providers in the comparison group. The report was prepared by The Lewin Group, and was released on September 19, REPORT RESULTS Participation has spanned across the range of the 48 clinical episode groups, although orthopedic surgery episodes have dominated. BPCI-participating hospitals tended to be larger, urban facilities, who generally operated in more affluent urban areas, had higher episode costs, and differed in other ways from non-participating providers. Many revealed that commitment from their leadership and financial investment in consultants or other resources were key to implementing the BPCI changes. Further, Hospital EIs indicated that episode selection decisions were based on opportunities to reduce spending and perceived financial risk, as well as whether they had a sufficient volume of cases within the episode type. This indicates that the availability of financial and other resources may be an important distinguishing factor between those providers that participated and those that did not. The analyses indicate that providers have responded to the incentives of BPCI, although statistically significant differences in episode costs or quality between BPCI participating providers and comparison providers have been few. There have been modest reductions in Medicare episode payments for select clinical episode groups with isolated instances of quality declines and fewer instances of increased quality. The lack of widespread impact of BPCI on Medicare payments across clinical episodes may be due to several factors. The quantitative results are based on less than one full year of BPCI experience for the majority of participants. The short period may not have been enough to first redesign care, and then see the impact of these changes on care. In addition, there is limited sample size for most clinical episodes, which limits the Report s ability to detect changes in payments. The use of aggregation analyses may also mask payment changes for particular types of episodes. Over the years, increased sample sizes and extended time and experiences under BPCI will diminish these limitations, expand understanding of the impact of BPCI, and strengthen conclusions. Model 2 Results Model 2 was the most widely adopted model, accounting for approximately three-quarters of the episodes and half of provider participants with 61 Awardees, 110 hospital EIs, and 3 PGP EIs. Almost one-third of the EIs were participating under two large Awardee Conveners, accounting for over 50% of Model 2 episodes. A total of 45,572 episodes were initiated across all 48 clinical episodes during the first year.

3 BPCI participating hospitals differed from hospitals that did not participate. Generally, participating hospitals were likely to be larger, urban facilities, located in more densely populated areas with higher median incomes. The large majority were also not-for-profit. BPCI-participating hospitals had a lower Medicare share in total Medicare inpatient stays than non-participants, though BPCI markets had higher Medicare Advantage penetration. The participating markets also tended to have more primary care physicians, specialists, and nurse practitioners. Each of the 48 clinical episodes was chosen by some Model 2 Awardees, with the average Model 2 EI participating in 5 clinical episodes. The most common episode was major joint replacement of a lower extremity, which was selected by 81 hospital EIs (74%) and accounted for 17,004 of the hospital-initiated episodes (44%). Congestive heart failure, chronic obstructive pulmonary disease, and pneumonia were also popular, with 35%, 26%, and 20% of EIs, respectively, selecting these episodes. Twenty-eight clinical episodes were selected by fewer than 10% of EIs. Though there were not statistically significant changes in episode costs or quality between BPCI and comparison group episodes for most episodes, total standardized payments declined for clinical groups that constitute most of the BPCI episodes. For orthopedic surgery, cardiovascular surgery, and spinal surgery episodes, there was a statistically significant decline (at least 5%) in total Medicare episode payments. These three clinical episode groups accounted for 22,761 episodes during the first year of BPCI, or 53% of all Model 2 episodes, and are discussed in more detail below. Further, for orthopedic and cardiovascular episodes, there was a statistically significant shift from more expensive institutional PAC to less expensive home health care among discharged beneficiaries. This shift will tend to increase patient complexity in both home and institutional PAC settings, and warrants further examination. Orthopedic Surgery Cardiovascular Surgery Spinal Surgery Participation 82 hospitals 30 hospitals 20 hospitals Participation 75% of Model 2 Hospital EIs 27% of Model 2 Hospital EIs 18% of Model 2 hospital EIs Percentage Episodes 18,936 episodes 2,859 episodes 966 episodes Avg Payment Change (BPCI v. Comparison) $864 (3%) less for BPCI providers than comparison No statistically significant difference, except for 30-day episodes with PAC use $3,447 increase for BPCI providers than comparison Quality of Care Remained constant Remained constant No statistically significant change Orthopedic Surgery o Reduction in Medicare payments was primarily due to reduced use of institutional PAC Share of institutional PAC use for the BPCI population decreased from 64% to 57% between the baseline and intervention periods, which was 4.9 percentage points greater than the decline in the comparison population (63.2% to 61.2%) Average SNF length of stay was 1.3 days shorter during the intervention period than the baseline period for participants discharged from BPCI participating hospitals; while the SNF stay in comparison hospitals remained virtually unchanged

4 o MJRLE episodes: BPCI participants had greater improvement than the comparison population in two functional measures improved ability to walk without resting (65.7% to 57.5%) and improved ability to walk up and down 12 stairs (65.4% to 57.9%) o Total episode payments declined for 89% of the BPCI hospitals, and total per-episode payments for the anchor hospitalization and services through the 90-day PDP declined $2,137 on average, ranging from a decline of $7,867 to an increase of $4,163. o BPCI hospitals with the greatest decline in payments were likelier to have obtained the hospital 3-day waiver and be located in areas with fewer SNF beds per population Cardiovascular Surgery o For 30-day cardiovascular surgery episodes with PAC use, total payments for the episode was estimated to have declined $4,149 more for BPCI episodes than for episodes at comparison providers o Institutional PAC use declined more in BPCI episodes than in episodes initiated by comparison group Share of institutional PAC use for the BPCI population decreased from 55.1% to 44.2%; statistically different from the 47.2% to 46.2% decline in episodes by the comparison group For HH users, there was a statistically significant increase of 1.5 home health visits in BPCI episodes relative to those in comparison episodes Spinal Surgery o Model 3 Results Payments for the anchor stay increased for BPCI episodes, but there was no statistically significant change in Part A payments by setting or Part B payments by service relative to comparison episodes, except for Part B payments for imaging and lab increased $53 more in BPCI episodes than in comparison episodes There were 20 Model 3 Awardees and 94 EIs by the end of September 2014, comprised of 63 SNFs, 28 HHAs, 1 IRF, 1 LTCH, and 1 PGP. All Model 3 EIs participated under a Convener and over two-thirds of the EIs participated under one of three Awardee Conveners, accounting for 74% of the Model 3 episodes. BPCI participating SNFs and HHAs differed from non-participating SNFs and HHAs. SNF EIs were larger than nonparticipants (148 beds vs. 110 beds), were more likely to be for-profit than non-participants (83% to 69%), and all of the participating SNFs were in urban locations (versus 69% of non-participating SNFs). Ninety-three percent of HHA EIs were for-profit entities, compared with 79% of non-participating HHAs. Participating HHAs had more employed nurses, an indication that they were larger. For both SNF and HHA participants, they were located in more densely populated areas, with higher average Medicare Advantage penetration and higher median household income, as well as more primary care physicians, specialists, and nurse practitioners for their populations compared with their comparison non-bpci markets. Model 3 EIs participated in 46 out of 48 potential clinical episodes, with the average EI participating in 19 clinical episodes. The most common clinical episode was congestive heart failure, which was selected by 95% of EIs, though participants later noted challenges with this episode, causing some to drop it from their program. Chronic obstructive pulmonary disease and simple pneumonia were the next most common clinical episodes, both chosen by 74% of EIs during the first year of BPCI.

5 For most clinical episode groups, there were no statistically significant differences between BPCI and comparison episodes in total Medicare standardized allowed payments. However, there were statistically significant declines in standardized payments for SNF services across all clinical groups, except for non-surgical respiratory. This decline in payment was consistent with a decline in utilization of SNF services. There were no statistically significant differences in payment for HHA-initiated episodes, except for a statistically significant decrease in the total amount of payments included in the pre-bundle period for non-surgical episodes. Quality outcomes generally remained constant or improved with BPCI participants, with a few exceptions where BPCI participant quality outcomes declined relative to the comparison group. For non-surgical cardiovascular clinical episodes, there was a statistically significant increase of 7.0 percentage points in the unplanned readmission rate during the first 30 days of the episode, as well as a statistically significant decline of 13.9 percentage points in the share of beneficiaries with improvement in self-care function for orthopedic surgery episodes. Model 4 Results There were 13 active Model 4 Awardees in the third quarter of 2014, with 20 EI hospitals. Ten of the EIs terminated their participation in BPCI during the first year of BPCI. The results included episodes initiated at the terminated EIs until the date of termination. Compared with non-participating hospitals, Model 4 EIs were larger (427 beds vs. 188 beds), were more likely to be non-profit (65% vs. 59%), and had more intensive teaching programs. All of the EIs were located in an urban area (100% vs. 71%). In addition, EIs had over twice as many BPCI episode admissions in 2011 than non-participating hospitals (4,516 vs. 2,140). EI hospitals averaged a lower share of Medicare days than non-participating hospitals (29% vs. 41%), but had similar Medicare disproportionate share percentages (29% vs. 28%). The Report interviewed the EIs that left BPCI, who indicated that they had faced significant challenges using the monthly data files that they received from CMS. These information and payment issues interfered with and delayed their ability to reimburse physicians, leading to the key reason they withdrew from BPCI. The limited participation in Model 4 limited the Report s analyses and findings. Model 4 EIs participated in 17 of the 48 clinical episodes, with the most common being MJRLE (approximately 70% of EIs). The two next most common clinical episodes were coronary artery bypass graft and double joint replacement of the lower extremity, with each being chosen by 45% of the EIs during the first year of BPCI. The Report calculated the impact of BPCI on payment and quality for orthopedic surgery and cardiovascular surgery clinical episode groups, which comprised 81% of all episodes with 3,021 total. There was a statistically significant increase in utilization and payments during the first 30 days post-hospital discharge for the cardiovascular surgery clinical episode group. For those cardiovascular surgery patients who received their first PAC treatment at an HHA, there was also a statistically significant decline of BPCI patients who demonstrated improvement in bathing, ambulation, and bed transferring relative to the comparison group (0.05). For the orthopedic surgery clinical episode group, there were no statistically significant changes in Medicare standardized allowed payments, quality, or utilization.