After a slow start in 2011, the Center for Medicare & Medicaid Innovation (CMMI) at the Centers of Medicare & Medicaid Services (CMS) seems to have gone into overdrive. As part of the Department of Health and Human Services’ pledge to move the majority of Medicare payments away from undifferentiated fee-for-service payments to some type of value-based care, CMMI has been sponsoring a wide variety of models. These include models that feature fee-for-service payments with incentives added, bundled or episode-based payments, and population-based payments.
Some of ongoing or scheduled demonstration models featuring fee-for-service with incentives are the patient-centered medical home model, the Comprehensive Primary Care initiative, and the Comprehensive Primary Care Plus initiatives (which I discussed in a recent JAMA Forum). Some episode-based payment models are procedure-based models, such as the voluntary Bundled Payment for Improvement Initiative (BPCI), and the mandatory Comprehensive Joint Replacement (CJR) program, which applies to all Medicare patients receiving a joint replacement in areas where the pilot project. A mandatory cardiovascular program, which will cover all services that occur with 90 days of the hospital discharge for patients who experience acute myocardial infarction and coronary artery bypass graft surgery (discussed in a recent JAMA Forum), is scheduled to begin in 2017. Other episode-based payment models feature condition-based episode payments, such as the Oncology Care Model, which began in July 2016, and the Medicare Diabetes Prevention Program, which has not yet started.
Finally, there are a variety of population-based models. These include the Medicare Shared Savings Program—the original version of accountable care organization (ACO) model that began in 2012. It now includes 434 ACOs; the Pioneer ACOs (down from 32 to 16), groups that have been in existence previously and were prepared to take on greater levels of financial risk in return for greater potential rewards; and 18 Next Generation ACOs, which feature prospectively set payment levels that offers 2 levels of shared savings and losses.
It’s not hard to imagine the potential for unintended consequences that could arise when clinicians or institutions providing Medicare services who participate in 1 ACO model interact with the same or overlapping participants in other models. Payouts resulting from savings incurred under bundled payments have been designed to prevent the same savings from being paid multiple times to different groups. However, the rationale for which group gets the savings is not always clear. This has led to what seems to be a peculiar, if not undesirable, interaction between ACOs, which are population-based and focus on all of the care for each patient for the year, and providers (usually a hospital) participating in mandatory or voluntary episode-based payment models.
The voluntary BPCI, which is already under way, has financial and performance accountability for episodes of care covering 4 bundled payment models that include both inpatient and postacute care. Participating organizations can take part in only 1 bundled payment model. The BPCI currently covers 48 diagnosis related group (DRG) bundles in 415 acute-care hospitals, 305 physician groups, and 723 skilled nursing facilities. The clinicians and hospitals can participate both in ACOs and a bundled payment model.
The concern that's been raised is that organizations working under these bundled payment models will compete with ACOs that focus on the whole care of the patient for the entire year—for example, as occurs when a patient who customarily receives care from an ACO is admitted to a hospital participating in the BPCI for a condition related to one of the DRG bundles. In such cases, Medicare payments for the patient’s care is carved out of the ACO delivery system whenever he or she is admitted for one of the BPCI DRG bundles. The organizing group under the BPCI (usually a hospital) agrees to give up a discount of 2% to 3% off of all of the Medicare payments to all entities providing services to the patient for the bundle, including the time after discharge. Thus, the payment bundle target for a particular DGR is the historical Medicare payment for that DRG minus the discount. If the payments to the various institutional providers and clinicians end up being less than the payment bundle, the BCPI organizing group keeps the difference; if the payments are greater than the target, the organizing group must repay Medicare the difference.
In this circumstance, it's not obvious that BCPI hospitals and specialists who not part of the ACO will have any incentive to work with the ACOs organizations, which mostly focus on being primary care networks. When Medicare payments are lower than the targeted amount for an ACO, the savings are shared between the ACO and Medicare, but there are no shared savings with BCPI. Lower readmission rates among patients receiving care from an ACO normally would produce savings for ACOs. However, savings produced by lower readmission rates for patients with conditions covered by the BCPI will now go directly to the organizing BCPI entity (usually the hospital).
To say this changes incentives is an understatement. The ACO has less incentive to reduce hospitalizations for the covered conditions. And bundled payments, which encourage less payment within the bundle, carry the incentive to create more bundles—which increases the number of payments. This is hardly a good trade-off, especially for CMS, if hospital readmissions for the covered conditions increase as a result of altered incentives.
Assessing the Demonstration Projects
This is but one example of the unintended consequences of the myriad demonstration projects that have been or are in the process of being launched. Pilot projects can be useful. Most frequently, they seem to show what strategies don’t work, but even this apparent failure may instead reflect the unwillingness of the affected parties to change their behavior for an institutional change that they doubt will be permanent. This was an issue raised by some of the large physician practice groups that participated in a CMS demonstration project a decade ago, and it’s even likelier to be a factor today, given the multiplicity of pilot projects under way.
In January, a new administration will be in place. Although it’s still unclear how best to design Medicare payment systems that will produce better-value care, I would suggest that the next administration assess what has been tried and what appears to have been learned so far from the various projects. Convening a group of clinicians, hospital and nursing home administrators, and beneficiaries' representatives to join the assessment would be politically and strategically important. After appropriate consultation, the administration should make the best decisions it can about the next generation of Medicare reimbursement. Modifications can and will be made in the future, just as they have been in the past. Enough demonstrations, already.