The Top Ten Things You Need to Know Now That the Medicare CJR Program is Final

by Gloria Kupferman, Vice President, DataGen Group

In November 2015, the Centers for Medicare and Medicaid Services (CMS) issued its Comprehensive Care for Joint Replacement (CJR) payment model, a pilot bundled payment program for the most common inpatient surgeries for Medicare beneficiaries—hip and knee replacements, known as lower extremity joint replacements (LEJRs). CJR will be mandatory for hospitals in 67 Metropolitan Statistical Areas (MSAs). The CJR model will be mandatory for about 800 hospitals.

The CJR program is a five-year(ish) program, beginning with discharges occurring on April 1, 2016 and ending with episodes completed by December 31, 2020.   However, this timeframe does not mean that the rest of the healthcare field has a few years before bundled payments apply to them.  CMS can—and probably will—exert its authority to expand this program or introduce another, similar program at any time.    Given the rapidly evolving innovation in healthcare payment and delivery driven by Medicare and other payers, it is critical that healthcare executives understand and stay abreast of the implications on their organizations.

This white paper examines the top ten things every hospital executive should know about the CJR program, its relationship to the BPCI demonstration, and the implications of this model for other innovative payment models.

1.CJR is not mandatory for everyone, but it might be for you.

CMS identifies 67 MSAs that have a high volume of Medicare fee-for-service (FFS) major joint replacements and low participation in the BPCI program.  Hospitals that are located in one of these MSAs are required to participate in CJR. There are very few exceptions:  hospitals that are already at risk for LEJRs under BPCI, and Critical Access Hospitals.

2.Hospitals own the episodes—not physicians or post-acute care providers.

Hospitals are only risk-bearing entities under this model and will be responsible for the Medicare cost of care for services provided during the entire 90-day episode.  CMS states that it would not be appropriate to mandate physician or post-acute episode owners in the CJR model because:  1) there can only be one party responsible for an episode and including the upstream and downstream providers would create overlap; and 2) physicians and post-acute providers generally do not have adequate financial reserves for managing risk without the underwriting of a third-party, which would not be permitted under the CJR model.  CMS also notes that physicians and post-acute providers rarely have the infrastructure necessary to provide the care coordination required under bundled payment programs.

Since the inpatient component of the episode is based on DRGs, the Medicare spending for the inpatient care is almost identical across the country.  It is clear that the variation in care and Medicare spending for LEJR episodes falls primarily in the post-discharge (i.e., post-acute) time period.  Post-acute care providers are the most at risk here and will need to demonstrate their value to referring physicians and hospitals.

3.Episodes are triggered by inpatient discharges and cover the inpatient admission and 90 days post-discharge services.

As in the BPCI program, episodes are defined by the Diagnosis Related Group (DRG) of the index inpatient admission. The proposed CJR model applies to all hospital discharges (in the mandatory MSAs) for DRGs 469 (major joint replacement or reattachment of lower extremity with major complications and comorbidities [MCC]) and DRG 470 (major joint replacement or reattachment of lower extremity without MCC).

These two DRGs were selected because they are the most popular in the BPCI program, have significant volume, involve relatively few physicians during the initial admission, have relatively few readmissions, and present significant opportunities to reduce spending across the episode. Unlike the BPCI program, participant providers cannot choose the episode length; all CJR episodes cover 90-day episodes of care. This may cause some concern among the mandatory hospitals, but it should not.  Our analysis of the 90-day episodes for LEJRs indicates that the preponderance of care occurs during the first 30 days—expanding the episode length to 90 days does not appreciably increase risk. 

The 90-day episodes include Medicare payments for all related covered services (other than Part D drugs) for all providers of care including physicians, skilled nursing facilities, home health agencies, inpatient rehabilitation facilities, and other outpatient services. The episode also includes all related readmissions during the 90-day period. Similar to BPCI, the CJR model excludes relatively few unrelated diagnoses/services from the episode.  Among the excluded services are unrelated surgical procedures such as appendectomies, and acute disease diagnoses such as head injuries and cancer.  There are still many diagnoses/services that clinicians may believe are unrelated to joint replacements that are included in the CJR episodes.

CMS has responded to industry comments regarding the higher PAC costs for hip fracture episodes.  There will be separate pricing for these episodes.

4.Target prices will be set prospectively and will reflect a blend of hospital-specific and regional data.

Just as in the BPCI program, all providers will continue to bill and receive payments according to current Medicare FFS rules.  At the end of each performance year, the average Medicare spending for each participating hospital will be compared to its calculated targets for the two LEJR DRGs.  The difference between the average spending and the target amount will be multiplied by the episode volume to arrive at a gross total—either positive or negative. 

Target prices for the CJR episodes will be established using Medicare claims data from a three-year baseline period.  For the first two performance years, target prices will be a blend of two-thirds hospital-specific and one-third regional average Medicare spending for the baseline period. In the third performance year, the target prices will be a blend of one-third hospital-specific and two-thirds regional average spending. In the fourth and fifth performance years, the target prices will be based entirely on regional averages. There will be exceptions for low-volume hospitals and new providers if there is insufficient data at the hospital level. 

The three-year baseline period will reflect actual Medicare FFS claims experience and will be re-based every other year to capture utilization changes over time.  The three years of baseline data will be trended forward to one common year—the most recent of the three—using a factor that reflects both price and utilization changes. The baseline regional averages will then be trended forward to the performance year, using known update factors for each of the payment settings, to set prospective target prices.  This is a significant difference (and improvement) from the BPCI program, where the targets are adjusted retrospectively to account for actual utilization and payment changes.

As in the BPCI program, under the CJR model, Medicare will reduce target prices by a discount percentage. This guarantees that there will be Medicare program savings even if the participating hospitals are unable to reduce episode spending. The maximum discount is 3% of the target price, but that discount may be reduced for hospitals that perform well quality-wise.

5.There is financial risk involved and it could impact cash flow, but there are protections against downside risk and limits on upside gains.

For the first program year, participants are eligible to receive payments for creating savings, but are not financially responsible for losses. After that, participants are responsible for both gains and losses. Note that there is no shared savings component to this model, unlike the Medicare Shared Savings Program and some other similar arrangements. Under CJR, hospitals will receive back all of the program savings they create (after the CMS discount) and are responsible for 100% of the losses. 

The reconciliation process is retrospective and will not begin until two months after the end of each calendar year.  This may present cash flow issues for those hospitals that end up in a pay-back situation.  It will also prove difficult for gainsharing arrangements, particularly with physicians, because the rewards will lag significantly.

The CJR model employs several mechanisms to limit participating hospitals’ losses—and gains. First, high outlier limits have been established at two standard deviations above the regional average episode amount. Individual episodes where the Medicare payments exceed the outlier limit will be truncated to that limit; unlike BPCI where excess amounts are not truncated, they are reduced to 20% of the excess over the limit. These outlier limits will be applied to baseline target price calculations as well as performance period reconciliations.  This differs from the BPCI program, in which participating providers are able to select outlier limits based on their own risk analysis (the 99th, 95th, or 75th percentile of the national average). 

In addition to no downside risk in the first performance year, CMS is limiting the downside risk in subsequent years by reducing the discount percentage based on both the program year and the hospital’s quality performance.  In program years 2 and 3, the discount percentage for payback calculations will range from 2% to 0.5% depending upon quality performance.  The fully discounted target (3%) will apply for repayments in the fourth and fifth years of the model for those hospitals that are the poorest quality performers; while top performers will be held to a discount percentage of 1.5% if they are in a payback situation by years 4 and 5.

Finally, CMS is applying stop-loss limits on the total amount that hospitals will be required to repay if their Medicare average payment per episode exceeds the target amount. The maximum repayment amount for the second year of the CJR program is limited to 5% of the aggregate target amount; that maximum repayment increases to 10% in year 3; and 20% in years 4 and 5.  The stop-loss limits are lower and more protective for Sole Community, Medicare Dependent, and Rural Referral Center hospitals.  However, CMS is also limiting the total amount that hospitals can gain under the CJR model to 5% of target in years 1 and 2; 10% in year 3; and 20% of the aggregate target in years 4 and 5. 

6.Gainsharing is allowed, within limits, and waivers are available.

The rule concedes that hospitals may wish to create financial relationships with other providers to coordinate financial incentives across the episode and identifies two funding streams for gainsharing: savings generated by reducing Medicare spending to below the target amount and internal cost savings.

CMS is very clear as to limits on the amounts that can be shared with collaborating providers and how much those collaborators can be asked to contribute toward repayments for excess spending.  Hospitals are required to retain at least 50% of any payments or repayments for amounts under/over the targets and cannot pass along more than 25% of repayment losses to any single collaborating provider. Gainsharing payments made to physician group practices (PGPs) are capped at 50% of total Medicare amounts approved under the fee schedule for services furnished by physician members of the PGP to the hospital’s CJR beneficiaries during a performance year.

CMS also makes allowances to provide certain items of value to beneficiaries to facilitate their care; e.g., post-acute surgical monitoring equipment.  However, gifts or services unrelated to patient care cannot be provided and any tangible items with significant monetary value must be returned at the end of the care episode.

CMS will provide waivers only for beneficiaries that are part of a CJR episode of care and only when the waiver is used to bill for a service that is furnished to the beneficiary even if the episode is later cancelled. This differs from the BPCI program in which waivers do not apply if the episode is later cancelled.  The  Medicare rule waivers include the Home Health PPS "incident to" rule; the geographic site requirement for telehealth services; the requirement that the eligible telehealth individual be in one of 8 eligible types of sites; and the 3-day hospital stay required for SNF payment. 

CMS and OIG have issued a joint statement that waives the federal anti-kickback statute and the physician self-referral law with respect to certain financial arrangements.  These waivers will protect payments made under gainsharing and share risk agreements that comply with the CJR program requirements.  The joint statement also waives the federal anti-kickback statute and civil monetary penalty law with respect to certain incentives that participating providers may offer to Medicare beneficiaries during an episode.

7.Quality is still job one.

The CJR program establishes a composite quality performance measure with a standard that must be met in order to receive reconciliation payments. This composite measure will also be used to as a performance incentive by reducing the target price discount for top performers.  The composite measure is comprised of the following measures: surgical complications within 90 days, patient satisfaction scores, and reporting of a new outcomes measure for LEJRs.  The first measure is already used in other Medicare quality payment programs and is currently reported on Hospital Compare.  The HCAHPS measure uses the same data source as is used for the Medicare Value Based Purchasing (VBP) program, but the metric calculation is different.  For CJR quality measurement, CMS will use the linear HCAHPS composite calculation that is used to assign the Hospital Compare star ratings.

Hospitals will be grouped into one of four quality categories based upon their composite score: Below Acceptable, Acceptable, Good or Excellent.  Hospitals that are deemed Below Acceptable will not be eligible to reconciliation payments, even if their financial performance is below the target amount.  Hospitals in the other three categories will be eligible for reconciliation payments, but the Good and Excellent hospitals will be compared to higher targets and thereby will receive a larger portion of the savings than the Acceptable hospitals.

Similarly, the quality categories will be applied for the hospitals that are determined to owe money back (in years 2 through 5).  Hospitals deemed to be Below Acceptable or Acceptable, will be held to the lowest target (maximum discount percentage) when calculating their repayment amounts, while Excellent hospitals’ repayment will be based on a higher target.

8.Claims data will be provided, but you have to ask for it. 

CMS recognizes the need for claim-level data to effectively manage the clinical, operational, and financial components of bundled payments. In the BPCI program, participants are provided with monthly updates on performance, as well as all of the baseline data from which targets were calculated. In the CJR model, CMS will provide baseline data some time prior to the start of the program, and performance period data will be provided and updated on at least a quarterly basis (monthly if practical). CMS will also provide summary data reports on hospital performance and aggregated regional data.

Rather than providing the claim-level data for all episodes on a routine basis, CJR participating hospitals must request their data on an annual basis.  This is significantly more burdensome on the participating hospital than the BPCI process. 

9.Medicare beneficiaries cannot opt out of the bundle.

Medicare beneficiaries who receive care for a LEJR in a mandatory hospital must be made aware that they are part of the CJR program; however, the beneficiary cannot opt out of the program.   The only way to opt out would be to choose another provider that is not in a mandatory MSA.  CMS has wisely rescinded its proposal to allow beneficiaries to opt out of sharing their data; that would have been disadvantageous for the hospitals at-risk under the program.

Some patients are excluded from episodes: Medicare Advantage beneficiaries, end-stage renal disease (ESRD) patients, and beneficiaries for whom Medicare is not the primary payer.  There is no episode if the patient dies at any point during the episode.  

10.Look out for interactions with other payment innovation models.

There are a growing number of Medicare payment demonstrations, pilots, and experiments and some of them may coincide with the CJR model.  CMS defines each of these programs and details how each would interact with the CJR model.  CMS will not allow any savings or repayments to be counted twice; therefore, any interactions between programs that have a financial impact will be reconciled.  To put it succinctly: the BPCI program always trumps CJR—in all other cases, the CJR model trumps other programs (e.g., Medicare Shared Savings Program, Pioneer ACO, etc.).

The table below compares some of the salient program components between Model 2 of the BPCI program and the CJR model:

 

 

BPCI Model 2

CJR

Episode initiators

Hospitals, physicians, post-acute providers

Hospitals only

Patient eligibility

Excludes Medicare Advantage, end-stage renal disease (ESRD), non-primary payer

No difference: Excludes Medicare Advantage, ESRD, non-primary payer

Episode definition

DRGs

No difference: DRGs

Episode length

30, 60, or 90 days

90 days

Awardee (risk-bearing) conveners

Allowed

Not allowed

Included payments

All Medicare Part A and Part B for IPPS, SNF, LTCH, IRF, HH, OPPS, Physicians, DME, Part B Drugs

All Medicare Part A and Part B, same as BPCI plus IPPS capital

Excluded services

Specified DRGs and ICD-9 codes

No difference: Specified DRGs and ICD-9 codes

Reconciliation frequency

Quarterly

Annual

Target price calculation

Hospital-specific; three-year baseline period; trended forward to quarterly performance periods; trend includes national changes in utilization as well as prices

Blend of hospital-specific and regional (nine census divisions); transition to 100% regional in years 4 and 5; three-year baseline period; re-basing every other year; trended forward to annual performance periods; trend includes price changes only

Retrospective vs. prospective

Retrospective payment reconciliation; retrospective target prices

Retrospective payment reconciliation; prospective target prices

Baseline period

July 2009 - June 2012

Years 1 and 2: January 2012 - December 2014

Years 3 and 4: January 2014 - December 2016

Year 5: January 2016 - December 2018

Reconciliation true-ups

Three true-ups after initial reconciliation

One true-up after initial reconciliation

Gainsharing

Explicitly allowed

No difference: Explicitly allowed

Participation discount

2% or 3%, depending upon episode length

Between 3% and 0.5% depending upon program year, quality performance, and whether the facility is receiving reconciliation payment or making a repayment

Upside and downside risk

Both in all years

Upside only in Year 1, then both

Limits on aggregate payments/repayments

+/- 20% of aggregate target (i.e., target price multiplied by volume)

Downside: -5% in Year 2; -10% in Year 3; -20% thereafter

Upside: +5% in Years1 and 2; +10% in Year 3; +20% thereafter

Episode payment exclusions (i.e., payments  are outside of the episode)

Indirect Medical Education (IME), Disproportionate Share Hospital (DSH), capital, VBP, readmissions penalty, HAC penalty, quality reporting penalty, Low Volume add-on, new technology add-on, hospital-specific payment rates, low income percentage add-on, HIV add-on

No difference: IME, DSH, capital, VBP, readmissions penalty, hospital-acquired conditions penalty, quality reporting penalty, Low Volume add-on, new technology add-on, hospital-specific payment rates, low income percentage add-on, HIV add-on

Outlier limits

Risk selection of 75%, 95% or 99% of national experience; 20% of payments above the risk limit are included in target and reconciliation

Two standard deviations of regional experience; all payments above the limit are excluded

 It is important to keep in mind that these two bundled payment models are strictly for Medicare beneficiaries.  Simultaneously, hospitals are also engaged in new and evolving innovative payment models, beyond bundled payment, with state Medicaid programs and commercial insurers for different populations. 

It is extremely important that today’s healthcare executives are aware of and understand the implications and interactions between all of the new and varied valued-based reimbursement constructs; new payment strategies will affect internal resources and financial health, relationships among providers within the healthcare system, and, ultimately, the way in which providers will define their businesses. 

Gloria Kupferman is Vice President and Chief of Healthcare Data Analytics for DataGen. She can be reached at GKupferm@hanys.org.