Impact of Medicares Bundled Payments for Care Improvement (BPCI) - - PowerPoint PPT Presentation

HEALTH CARE AND HUMAN SERVICES POLICY, RESEARCH, AND CONSULTING - WITH REAL-WORLD PERSPECTIVE.

Impact of Medicare’s Bundled Payments for Care Improvement (BPCI) Initiative on Cost and Quality of Care

Laura Dummit, MSPH, Grecia Marrufo, PhD, Jaclyn Marshall, MS, Eleonora Tan, PhD, Aylin Bradley, MA, Karla López de Nava, PhD, The Lewin Group June 25, 2017

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the

• fficial views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Outline of Presentation

 Overview of BPCI  Analytical Approach



Methods and Data Sources

 Participants  Impact of BPCI on Payments and Quality of Care, Model 2



Orthopedic Surgery



Cardiovascular Surgery

 Conclusion

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Overview of BPCI

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Overview: Bundled Payments for Care Improvement (BPCI) Initiative Models 2, 3, 4

The BPCI initiative has multiple layers of complexity that the monitoring and evaluation project must consider in the design and analysis.

Model 2: Retrospective hospital + post acute care (PAC) bundle Inpatient stay through 30 days post hospital discharge Inpatient stay + 60 days post hospital discharge Inpatient stay + 90 days post hospital discharge Model 3: Retrospective PAC only bundle Initiation of PAC to day 30 Initiation of PAC to day 60 Initiation of PAC to day 90 Model 4: Prospective acute care bundle Inpatient stay + related readmissions within 30 days

Models Episode length

48 clinical episodes as identified by MS-DRGs

Range of approaches by awardees (including ability to use waivers)

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Major Evaluation Questions

Nature of the initiative and participants at baseline and over the course of the initiative

• Participants
• Market characteristics
• Patient population

characteristics

• Model incentive

structure characteristics

• Care redesign and cost

saving strategy characteristics Impact of the initiative

• Impact on utilization and

payment

• Impact on quality
• Other unintended

consequences

• Market and spill-over

impact Explanatory factors associated with impact of the initiative

• Characteristics of the

model

• Characteristics of the

participating providers’ approach to their chosen model

• Characteristics of the

participating providers’ specific features and ability to carry out their proposed intervention(s)

• Characteristics of the

market

• Patient population and

case mix characteristics

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Analytical Approach

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Methods and Data Sources

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

 Mixed methods approach consisting of:

• Descriptive Analysis
• Difference in Differences
• Qualitative Analysis

 Data Sources

• Primary Data: Beneficiary survey, site visits, focus groups,

expert interviews, awardee-submitted data, technical expert panels

• Secondary Data: Medicare claims and enrollment data, Patient

Assessment data (OASIS, MDS, IRF-PAI), Provider of Services (POS) file, Medicare Provider Enrollment, Chain, and Ownership System (PECOS), and other data

www.lewin.com

Study Population for Claim-Based Analysis

 BPCI Intervention Population: defined by model and clinical

• episode. Includes all episodes initiated in BPCI-participating

providers from October 2013 through September 2014.

 Comparison Population: selected in 4 steps:

1.

Excluded non-participating hospitals whose characteristics were not represented in the BPCI population (e.g., not paid under Medicare’s IPPS) or who could have been influenced by BPCI (e.g. located in markets with high BPCI participation,

• wned by a BPCI participating organization, or preparing to join BPCI).

2.

Each BPCI provider was matched with up to 15 comparison providers based on market and provider characteristics and baseline claims outcomes using propensity score matching or Mahalanobis matching.

3.

Constructed episodes for BPCI and matched comparison providers between October 2010 and September 2014.

4.

Among the comparison episodes identified in step 3, we drew a random sample to match the distribution of MS-DRGs and quarters of episode start in the BPCI population.

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Participants

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

BPCI-participating Hospitals Differ from All Hospitals

Source: Lewin analysis of 2013 Provider of Service (POS) and 2011 Medicare claims. BPCI participating hospitals are defined as Model 2 EIs, Q4 2013 – Q3 2014. Non-participant hospitals are all other hospitals and exclude Model 4 hospitals participating in BPCI during the first year. BPCI Hospitals (N=110) Non-participating Hospitals (N=3,056)

N % N % Ownership For Profit 13 12% 672 22% Government 4 4% 550 18% Non-Profit 93 85% 1,803 59% Urban/Rural Rural 6 5% 886 29% Urban 104 95% 2,170 71% Part of Chain Yes 54 49% 1,528 52% Mean Mean Bed Count 359 188 Number of Admissions for BPCI Episode MS-DRGs, 2011 4,060 2,140 Medicare Days Percent 37% 41% Resident-to-bed ratio 0.18 0.06 Disproportionate Share Percent 30% 28% Standardized Part A Allowed Payment inpatient stay plus 90 day PDP, 2011 Clinical Episode Group (N, BPCI discharges) Mean Mean Orthopedic surgery (38,718) $29,439 $28,882 Non-surgical other medical (14,577) $26,595 $24,897 Non-surgical neuro (2,690) $25,235 $24,040 Non-surgical respiratory (19,818) $24,310 $23,183 Non-surgical cardiac (30,290) $22,191 $21,677 Surgical and non-surgical GI (5,854) $21,935 $20,835 Cardiovascular surgery (18,452) $33,664 $33,513 Spinal surgery (2,504) $29,781 $28,929

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Impact of BPCI on Payments and Quality of Care, Model 2, Hospital, Orthopedic Surgery

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Total Payments Declined for Orthopedic Surgery Episodes at BPCI Hospitals relative to Comparison Hospitals

*Denotes statistical significance at the 5% level. LCI=lower confidence interval at the 5% level: UCI= upper confidence interval at the 5% level. Note: Baseline is defined as episodes that began Q4 2011 through Q3 2012. Intervention is defined as episodes that began Q4 2013 through Q3 2014. Sample sizes reflect number of episodes initiated during the intervention period that met inclusion criteria for the

• utcome. Medicare payments are trended and expressed in 2014 dollars.

Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.

Measure BPCI (N=17,672) Comparison (N=17,688) Diff-in-Diff estimate

Baseline Intervention Baseline Intervention Diff-in-Diff LCI UCI Total Standardized Allowed Payment, Inpatient Anchor Hospitalization and 90-day PDP $30,239 $28,232 $29,814 $28,670

• $864*
• $1,353
• $375

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Changes in the Use of Post-Acute Care (PAC) Account for Decline in Total Payments for Orthopedic Surgery Episodes among BPCI Participants

Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.



Patients discharged from BPCI hospitals reduced average SNF number

• f days by 1.3 days (from 26.1 to 24.9) relative to comparison

patients (25.5 to 25.7)



Greater decline in Medicare Inpatient Rehabilitation Facility (IRF) payments for BPCI episodes than comparison episodes (DiD= -$459)

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

64% 63% 57% 61% BPCI Hospitals Comparison Hospitals

Proportion of Beneficiaries Discharged to Institutional PAC

Baseline Intervention

www.lewin.com

No Difference in the Quality of Care for Medicare Beneficiaries Receiving Orthopedic Surgery at BPCI-Hospitals Relative to Those at Comparison Hospitals (Claim-based Measures)

LCI=lower confidence interval at the 5% level: UCI=upper confidence interval at the 5% level. Note: Baseline is defined as episodes that began Q4 2011 through Q3 2012. Intervention is defined as episodes that began Q4 2013 through Q3 2014. Sample sizes reflect number of episodes initiated during the intervention period that met inclusion criteria for the outcome. Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

Measure BPCI (N=17,927) Comparison (N=17,929) Diff-in-Diff estimate

Baseline Intervention Baseline Intervention Diff-in-Diff LCI UCI Unplanned Readmission Rate, within 30 days 6.4% 6.2% 6.9% 6.6% 0.0

• 0.7

0.8 Unplanned Readmission Rate, within 90 days 10.9% 10.5% 11.3% 10.9%

• 0.1
• 1.1

0.9 Emergency Department Visit Rate, within 30 days 7.3% 7.7% 7.5% 7.7% 0.2

• 0.6

1.0 Emergency Department Visit Rate, within 90 days 13.8% 14.4% 13.7% 14.2% 0.1

• 1.0

1.2 Mortality, within 30 days 0.9% 1.0% 0.8% 1.0% 0.0

• 0.3

0.2

www.lewin.com

Improvement in Functioning Was Similar for Beneficiaries Who Had Orthopedic Surgery at BPCI Hospitals and Comparison Beneficiaries (Assessment-based Quality Measures)

 Improvement in functional status outcomes among patients

discharged to SNF , IRF , or HHA (e.g. self-care, mobility) were similar between beneficiaries whose episodes started in a BPCI participating hospital and those whose episode started in comparison hospitals.

 One exception: the share of patients that exhibited

improvement in upper body dressing declined among BPCI beneficiaries and increased among comparison beneficiaries who received their first PAC treatment at a HHA.

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Impact of BPCI on Payments and Quality of Care, Model 2, Cardiovascular Surgery Episodes

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

The Change in Total Payments Was Not Statistically Different for Cardiovascular Surgery Episodes at BPCI Hospitals Relative to Comparison Hospitals

LCI=lower confidence interval at the 5% level; UCI=upper confidence interval at the 5% level. Note: Baseline is defined as episodes that began Q4 2011 through Q3 2012. Intervention is defined as episodes that began Q4 2013 through Q3 2014. Sample sizes reflect number of episodes initiated during the intervention period that met inclusion criteria for the outcome. Medicare payments are trended and expressed in 2014 dollars. Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.

Measure BPCI (N=2,641) Comparison (N=2,653) Diff-in-Diff estimate

Baseline Intervention Baseline Intervention Diff-in-Diff LCI UCI Total Standardized Allowed Payment, Inpatient Anchor Hospitalization and 90-day PDP $46,805 $46,282 $46,272 $46,628

• $880
• $2,584

$825 The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Changes in the Use of PAC Account for Potential Decline in Total Payments for BPCI Cardiovascular Surgery Episodes among BPCI Participants

Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.



Number of HH visits increased for BPCI beneficiaries (from 15.8 to 16.6) and decreased for the comparison group (16.6 to 15.9). This resulted in a significant increase of 1.5 visits for BPCI relative to comparison group.

• The contents of this presentation are solely the responsibility of the authors

and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

55% 47% 44% 46% BPCI Hospitals Comparison Hospitals

Proportion of Beneficiaries Discharged to Institutional PAC

Baseline Intervention

www.lewin.com

Mixed Results Regarding Quality of Care Measured by Emergency Department Visit and Mortality Rates, for BPCI Cardiovascular Surgery Episodes (Claims-based Quality Measures)

*Denotes statistical significance at the 5% level. LCI= lower confidence interval at the 5% level; UCI=upper confidence interval at the 5% level Note: Baseline is defined as episodes that began Q4 2011 through Q3 2012. Intervention is defined as episodes that began Q4 2013 through Q3 2014. Sample sizes reflect number of episodes initiated during the intervention period that met inclusion criteria for the outcome. Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI and comparison providers.

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.



None of the changes in assessment-based quality measures from baseline to intervention period were significantly different between BPCI and comparison group beneficiaries with cardiovascular surgery episodes who received their first PAC treatment at a SNF , HHA or an IRF .

Measure BPCI (N=2,712) Comparison (N=2,714) Diff-in-Diff estimate

Baseline Interventio n Baseline Intervention Diff-in-Diff LCI UCI Unplanned Readmission Rate, within 30 days 15.7% 15.4% 14.9% 16.0%

• 1.3
• 3.7

1.2 Unplanned Readmission Rate, within 90 days 25.0% 24.2% 24.8% 24.7%

• 0.7
• 3.2

1.8 Emergency Department Visit Rate, within 30 days 9.3% 12.0% 10.8% 11.3% 2.2* 0.1 4.4 Emergency Department Visit Rate, within 90 days 17.8% 20.9% 19.7% 22.3% 0.4

• 2.3

3.0 Mortality, within 30 days 1.6% 1.9% 2.1% 1.4% 0.9* 0.0 1.8

www.lewin.com

Conclusion

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Key Findings



BPCI providers are not representative of all health care providers. They tend to be larger, operate in urban areas and have other characteristics that might suggest they have more resources than

• ther providers to initiate care redesign



Limited evidence that BPCI initiative reduced Medicare payments:

• Significant decrease in payments for orthopedic surgery

episodes, likely due to significant decreases in use of institutional PAC

• Payments for cardiovascular surgery episodes were not

statistically different between participants and non-

• participants. However, there were significant decreases in use
• f institutional PAC among participants



Few indications that quality of care declined:

• No difference in quality of care for Medicare beneficiaries

receiving orthopedic surgery relative to comparison beneficiaries

• Cardiovascular surgery episodes showed mixed evidence

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Appendix

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

BPCI Participants

 Single Awardee (SA)– Under Models 2, 3 and 4, SAs are individual Medicare providers that

assume financial risk under the Model for episodes initiated at their institution. These SAs are also Episode Initiators.

 Awardee Convener (AC) - Parent companies, health systems, or other organizations that

assume financial risk under the Model for Medicare beneficiaries that initiate episodes at their respective Episode Initiating Bundled Payments Provider Organization (EI-BPPO) are ACs. An AC may or may not be a Medicare provider or initiate episodes.

 Facilitator Convener (FC) - An entity that submits a BPCI application and serves an

administrative and technical assistance function on behalf of one or more Designated Awardees (DA) or Designated Awardee Conveners (DAC) is a Facilitator Convener (FC). Designated Awardees and DACs function as SAs and ACs, respectively, but may join the initiative under a

• FC. Facilitator Conveners do not have an agreement with CMS, nor do they bear financial risk

under the Model, or receive payment from CMS. The DA or DAC would have an agreement with CMS and assume financial risk under the Model for episodes initiated at their institution.

 Episode Initiating Bundled Payments Provider Organization (EI-BPPO) — Under Models 2, 3

and 4, EI-BPPOs are Medicare providers that deliver care to beneficiaries. Episodes start at EI-

• EPPOs. EI-EPPOs do not assume financial risk. They are associated with an AC or a DAC that

assumes the financial risk.

 Episode Initiators (EI) — Under Model 2, an EI is the participating hospital where the BPCI

episode begins or a participating physician group practice (PGP) if one of its physicians is the patient’s admitting physician or surgeon for the anchor hospitalization. Under Model 3 an EI may be a participating PGP or a participating SNF , HHA, IRF , or LTCH that admits the patient within 30 days following a hospital discharge for a MS-DRG for the relevant clinical episodes (anchor hospitalization). Under Model 4 an EI is the participating hospital where the BPCI episode begins. Single Awardees and DAs are EIs. Awardee Conveners and DACs may or may not be EIs themselves, and may also have one or more EIs under their Awardee structure. The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Analytical Methods: Mixed Methods Approach Quantitative and Qualitative Analysis

Crosswalk of Research Questions to Analytical Methods Research Question Descriptive Analysis DiD Cross-section comparisons between BPCI and non-BPCI survey samples Before-After Comparisons Across BPCI participants Market Dynamic Analysis Qualitative Analysis

A. What are the characteristics of the program and participants at baseline and how have they changed during the course of the initiative? X X

• B. What is the impact of the BPCI

initiative on the costs of episodes, the Medicare program, and the quality of care for Medicare beneficiaries? Impact on payment and utilization X Impact on quality of care X X X Impact on provider referral an market share X

• C. What program, provider,

beneficiary, and environmental factors contributed to the various results of the BPCI initiative? X X X

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

DiD Estimator

 DiD estimator



Compares changes in average outcomes over time between participants and matched controls



Controls for differences in risk factors and utilization trends during the months prior to the start of the episode

Provider Period Pre BPCI Post BPCI BPCI initiator site

YBPCI-Pre YBPCI-Post

Control Site

YControl-Pre Ycontrol-Pre

Example: Yij = Total predicted costs following hip replacement among beneficiaries who received services from provider type “i” in period “j”

DiD= (YBPCI-Post-YBPCI-Pre) – (YControl-Post-YControl-Pre)

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Primary and Secondary Data Sources Used in BPCI Evaluation and Monitoring Activities

Use in Evaluation Primary Data Secondary Data

Provide context to the BPCI initiative: BPCI participants, BPCI markets, care redesign approaches, and partnerships

• Case study site visits
• Focus groups
• Expert interviews
• Quarterly Awardee

interviews

• Awardee-submitted

data

• Salesforce BPCI Participant and Episode Reports
• Medicare Provider Enrollment, Chain, and

Ownership System (PECOS) files

• Provider of Services (POS) files
• Area Health Resource Files (AHRF)
• Implementation protocols

Assess implementation strategies and challenges

• Case study site visits
• Focus groups
• Expert interviews
• Quarterly Awardee

interviews Construct BPCI population and comparison groups

• Case study site visits
• Expert interviews
• Quarterly Awardee

interviews

• Salesforce BPCI Participant and Episode Reports
• Medicare Provider Enrollment, Chain, and

Ownership System (PECOS) files

• Provider of Services (POS) files
• Area Health Resource Files (AHRF)
• Master Data Management (MDM)
• Medicare Part A and B claims
• The Master Beneficiary Summary File
• Episode files from Reconciliation contractor

Evaluate BPCI impact on health and functional outcomes, utilization, and payments

• Beneficiary survey
• Medicare Part A and B claims
• The Master Beneficiary Summary File
• MDS, OASIS, and IRF-PAI patient assessments

Evaluate patient-centered

• utcomes
• Beneficiary survey
• Technical expert panel

Evaluate factors explaining differential outcomes

• Awardee-submitted

data

• Medicare Provider Enrollment, Chain, and

Ownership System (PECOS) files

• Provider of Services (POS) files
• Area Health Resource Files (AHRF)
• Medicare Part A and B claims
• The Master Beneficiary Summary File
• Implementation protocols

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Orthopedic and Cardiovascular Surgery Clinical Episodes

Orthopedic Surgery Cardiovascular Surgery

Major joint replacement of the upper extremity Coronary artery bypass graft Amputation Percutaneous coronary intervention Major joint replacement of the lower extremity Pacemaker Double joint replacement of the lower extremity Cardiac defibrillator Revision of the hip or knee Pacemaker device replacement or revision Hip & femur procedures except major joint AICD generator or lead Other knee procedures Cardiac valve Lower extremity and humerus procedure except hip, foot, femur Major cardiovascular procedure Removal of orthopedic devices Other vascular surgery The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Clinical Episode Groups and Representative Clinical Episodes

Clinical Episode Groupings Representative Clinical Episodes included in the Clinical Episode Group

Non-surgical and surgical Gastrointestinal Esophagitis, gastroenteritis and other digestive disorders Non-surgical cardiovascular Congestive heart failure Non-surgical neurovascular Stroke Non-surgical orthopedic Fractures of the femur and hip or pelvis Non-surgical other medical Sepsis Non-surgical respiratory Chronic obstructive pulmonary disease, bronchitis, asthma Cardiovascular surgery Coronary artery bypass graft Orthopedic surgery Major joint replacement of the lower extremity Spinal surgery Cervical spinal fusion The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Growth in BPCI Participants Over Time

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

Model Participant Role and Type of Episode Initiators Count of participants as of Q3 2014 Count of participants as of Q3 2015 N N

Model 2 Awardees 61 225 Episode initiators 113 718 Acute care hospitals 100 426 Physician Group Practices 3 292 Model 3 Awardees 20 138 Episode initiators 94 1,347 Skilled nursing facilities 63 1,075 Home health agencies 28 116 Inpatient rehabilitation facilities 1 9 Long term care hospitals 1 1 Physician Group Practices 1 146 Model 4 Awardees 13 19 Episode initiators 20 23 Acute care hospitals 20 23 All Models Awardees 94 382 Episode initiators 227 2,088

Source: Count of Awardees and EIs based on Lewin analysis of Salesforce data, May and August 2015. Notes: In Q3 2015, one ACH Single Awardee is counted in both Model 2 and Model 4, because it switched models during the initiative. The counts include Awardees and EIs who have terminated their participation in the program prior to October 2015 given our analysis is cumulative and includes providers during the time in which they were participating in BPCI.

www.lewin.com

Number of BPCI-participating Hospitals by CBSA, Model 2 and Model 4, Q4 2013-Q32014

Source: Lewin analysis of Salesforce data for all Q4 2013 – Q3 2014 BPCI participating hospital EIs. The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Characteristics of BPCI Markets and Non-BPCI Markets, Models 2 and 4, Q4 2013-Q3 2014

Source: Lewin analysis of Medicare claims and 2011 AHRF .

Market Characteristics Models 2 & 4 BPCI Markets N=60; 6.4% of Markets Non-BPCI Markets N=882; 93.6% of Markets

Mean Median 25th 75th Mean Median 25th 75th Hospital Penetration 38.5% 31.5% 16.0% 46.5% 0.0% 0.0% 0.0% 0.0% Herfindahl Index – hospital 0.29 0.25 0.11 0.34 0.69 0.95 0.43 1.00 Herfindahl Index – SNF 0.07 0.04 0.02 0.07 0.33 0.28 0.15 0.41 Herfindahl Index – HHA 0.22 0.14 0.07 0.27 0.54 0.52 0.23 1.00 Herfindahl Index – IRF 0.38 0.26 0.00 1.00 0.11 0.00 0.00 0.00 Medicare Advantage Penetration 26.9% 25.0% 17.9% 36.9% 17.6% 14.7% 8.3% 23.7% Population 1,830,486 698,835 360,072 1,842,713 206,728 67,698 38,885 150,811 Median Household Income $51,069 $50,101 $46,222 $55,076 $43,741 $42,570 $38,276 $48,029 % Age 65+ 14% 13% 12% 15% 15% 15% 13% 17% PCPs Per 10,000 8.2 8.0 7.3 8.8 6.3 6.1 4.7 7.5 Specialists Per 10,000 11.2 10.2 7.5 13.1 5.1 4.3 2.5 6.5 PA/NPs Per 10,000 7.8 7.2 5.5 9.1 6.0 5.5 3.8 7.5 SNF Beds Per 10,000 58.2 57.8 40.8 75.0 71.4 65.1 43.8 91.5 LTCH Beds Per 10,000 1.1 0.8 0.3 1.6 0.5 0.0 0.0 0.0 IRF Beds Per 10,000 0.7 0.2 0.0 1.1 0.3 0.0 0.0 0.0 CAH Beds Per 10,000 0.4 0.0 0.0 0.6 1.8 0.0 0.0 1.5 The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Model and EI Types Included in Evaluation Analyses

EI Type Number of BPCI EIs, Q4 2013 – Q3 2014 Characteristics

• f the Program

Impact of BPCI (DiD estimates) Beneficiary survey Provider referral and market share Factors Contributing to Differences across BPCI Providers

Model 2 Hospital 110 X X X X X PGP 3 X Model 3 SNF 63 X X X X X HHA 28 X X X X X IRF 1 X X LTCH 1 X PGP 1 X Model 4 Hospital 20 X X X X

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Characteristics of BPCI Beneficiaries and All Medicare Beneficiaries with a Hospitalization for an Orthopedic Surgery MS-DRG, Model 2, Q4 2013-Q3 2014

Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI providers and for all admissions for the same MS-DRG to non-participating hospitals in Q3 2014.

Characteristics Model 2 BPCI Beneficiaries with Orthopedic Surgery Episodes, Q4 2013 – Q3 2014 All Medicare Beneficiaries With Same MS-DRG Admission, Q3 2014 (N=17,934) (N=196,694) N % N %

Age 20-64 1,472 8.2% 22,694 11.5% 65-79 11,556 64.4% 115,834 58.9% 80+ 4,906 27.4% 58,166 29.6% Gender Female 11,843 66.0% 127,405 64.8% Male 6,091 34.0% 69,289 35.2% Medicaid and Disability Percent Eligible for Medicaid 1,926 10.7% 29,707 15.1% Percent Disability, no ESRD 1,946 10.9% 26,129 13.3%

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Proportion of Beneficiaries who Improved in Functional Status was Significantly Higher for BPCI Beneficiaries than for Comparison Beneficiaries (Survey-based Quality Measures)



Proportion of beneficiaries who reported improved in the “walk by yourself without resting” measure was significantly higher for BPCI beneficiaries with MJRLE episodes than for the comparison beneficiaries

• 65.7% of BPCI respondents with MJRLE improved their ability to

walk without resting as compared with 57.5% of comparison respondents

 Proportion of beneficiaries who reported improved in the “walk up

and down 12 stairs” measure was significantly higher for BPCI beneficiaries with MJRLE episodes than for the comparison beneficiaries

• 65.4% of BPCI respondents with MJRLE improved their ability to

use stairs as compared with 57.9% among comparison respondents

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

No Differences in Health Care Experience for BPCI MJRLE Beneficiaries and Comparison Beneficiaries (Survey-based Quality Measures)



Only Exception: BPCI beneficiaries were less likely than comparison respondents to report that they agree or strongly agree that “medical staff clearly explained what follow-up appointments or treatment would be needed when prepared to go home”

• 96.9% of BPCI beneficiaries with MJRLE agreed/strongly agreed

with the above statement compared to 99.7% of comparison respondents

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Characteristics of BPCI Beneficiaries and All Medicare Beneficiaries with a Hospitalization for a Cardiovascular Surgery MS-DRG, Model 2, Q4 2013-Q3 2014

Source: Lewin analysis of Medicare claims and enrollment data for episodes that began Q4 2011 through Q3 2014 for BPCI providers and for all admissions for the same MS-DRG to non-participating hospitals in Q3 2014.

Characteristics Model 2 BPCI Beneficiaries with Cardiovascular Surgery Episodes, Q4 2013 – Q3 2014 All Medicare Beneficiaries With Same MS-DRG Admission, Q3 2014 (N=2,718) (N=62,576)

N % N % Age 20-64 247 9.1% 8,023 12.8% 65-79 1,578 58.1% 34,213 54.7% 80+ 893 32.9% 20,340 32.5% Gender Female 1,083 39.8% 26,340 42.1% Male 1,635 60.2% 36,236 57.9% Medicaid and Disability Percent Eligible for Medicaid 445 16.4% 11,177 17.9% Percent Disability, no ESRD 327 12.0% 6,931 11.1%

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.

www.lewin.com

Team

Contractor Brief description of role The Lewin Group

Responsible for overall project, including timely and high quality

• deliverables. Create website for quarterly awardee data submission

and feedback reports. Analyze awardee-submitted and claims data. Conduct a subset of case studies. Lead creation of all reports.

Optum

Provide payment, market, and clinical expertise.

Abt

Analyze patient assessment and quality data. Conduct and analyze

• survey. Conduct a subset of case studies.

Telligen

Conduct clinical technical expert panels (TEPs). Provide clinical expertise on case studies and overall.

GDIT

Create CMS claims, enrollment, and assessment data extracts.

The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies.