Innovations in hospital payment models: the pros and cons of bundled payments

January 27, 2017

Innovations in hospital payment models: the pros and cons of bundled payments

Michigan Health Lab


As health care costs for the nation’s growing population of older adults continue to rise, the federal government has been considering different measures to help rein in spending while ensuring incentives to improve the quality and efficiency of care remain in place.

One such approach is by “bundling” Medicare payments to hospitals for a range of health care services required to treat a particular condition. These alternative payment models are intended to spur coordinated care between hospitals and other providers to reduce potentially unnecessary services that may be contributing to big variations in costs.

A University of Michigan team of researchers is informing the conversation about these models, and their work is likely to help shape how they are refined and implemented in the near future.

The team’s analysis of a newly instituted payment model for joint replacement, published last year in Health Affairs, determined that the way the incentive program was structured could end up unfairly penalizing hospitals that treat more medically “complex” patients, who may be sicker, older, or have multiple illnesses, and whose appropriate care may require more expensive services.

In a Final Rule published on Jan. 3, 2017, the Centers for Medicare and Medicaid Services modified the existing joint replacement program and created new mandatory payment systems to cover care for certain heart conditions. The agency also stated, for the first time, that it would explore and plan to implement additional adjustments to these programs to account for patient complexity, citing the stakeholder response to the 2016 Health Affairs paper as part of the reason for their decision.

Here, Chad Ellimoottil, M.D., M.S., a U-M health policy researcher and practicing physician, and lead author on the paper CMS referenced, explains more.


Chad Ellimoottil, M.D., M.S.

What is the impact of this new Final Rule?
Ellimoottil: The rule is significant in the sense that it is a testament to Medicare’s plan to move away from traditional fee-for-service, which reimburses health care providers for medications, rehabilitation, tests, devices and other services as individual expenses, and toward alternative payment models such as bundled payments, in which providers are reimbursed through a single payment for all services to treat a particular condition (or a care “episode”). The new programs target Medicare enrollees after heart attacks, coronary artery bypass grafting (CABG), and joint replacement, and includes payments for up to 90 days after a patient leaves the hospital.

Other alternative payment models have mostly been voluntary in nature, but these new programs are mandatory. Medicare believes that participation needs to be mandated for the programs to have the intended impact.

How are bundled payments intended to achieve higher quality and more efficient care?
Ellimoottil: These models aim to improve inefficiencies in the current payment system and reduce variation in spending. Sources of variation can include the use of skilled facilities after uncomplicated joint replacement surgeries, the use of outpatient rehabilitation or the number of readmissions.

Unfortunately, it can be hard to know in each case whether services like post-acute care were unwarranted, or whether a readmission was unavoidable and contributed to better quality care for that patient. The overall idea behind bundled payments is that there is some waste in the system, and by tying the events that occur after the hospitalization to payment, hospitals will be motivated to coordinate with the other providers involved in the patient’s care to make these episodes of care leaner. Medicare believes the hospital actually has the greatest influence over these events that occur after a hospitalization though that’s not always the case.

Why is it important to consider patient complexity when considering costs?
Ellimoottil: Medicare assigns a target price for each episode and compares the hospital’s actual spending to the target to determine whether a hospital receives a payment penalty or a bonus. The target price is calculated by considering the average spending per care episode in the hospital’s U.S. Census Division. That is a very large comparison group!

But while differences in hospital episode spending can certainly be influenced by variation in the use of services, differences in patients’ medical complexity can also be an important driver of spending. When I refer to patient complexity, I am referring to patient characteristics that the hospital does not have control over, but may affect an outcome. Examples include other health conditions that a patient has (or "comorbidities"), the patient’s age and other factors like socioeconomic status that are a little more controversial.

Our Health Affairs paper takes the position that considering some of these additional patient factors is absolutely necessary to compare patient costs across different hospitals. Our analysis examined the feasibility of using a process called risk-adjustment as a way to do this. While many different risk-adjustment models exist, we used a particular risk-adjustment model that has a low administrative burden and that CMS is already using for several existing programs.

We found that without risk-adjustment, the hospitals that treat highly complex patents, some older patients, and patients with more comorbidities end up losing in this program just because of the fact that they’re treating sicker patients, which is unfair to patients and can lead to access issues.

Some of these patient characteristics can be obtained using claims-based databases. But complexity goes much deeper than that, and a lot of work that’s being done at the Institute for Healthcare Policy & Innovation (IHPI) and through Blue Cross Blue Shield of Michigan’s unique Collaborative Quality Initiative (CQI) programs is trying to understand complexity beyond what’s available in an administrative database. For example, many of the CQIs have a registry with robust clinical variables.

It can be administratively burdensome to capture granular clinical information, but I do believe that in the long run these registries may do a better job of adjusting for risk. I think ultimately if CMS is open to exploring replicating the work that’s being done in Michigan through the CQI models, they may consider requiring hospitals to fill out an additional form with variables that are known to impact payment.  

What happens next with these programs and how we think about them?
Ellimoottil: Although the future of bundled payment systems is a bit uncertain under the new administration, hospitals have already invested a lot of money into redesigning care to accommodate them, so I think it’s unlikely these programs would go away completely.  

As CMS considers additional risk-adjustment measures to adjust payments for these programs, I think there are many opportunities for policymakers to tap into our team’s expertise. Our primary goal is to objectively understand the role of risk-adjustment in bundled payments, and we’re currently doing several analyses related to this larger goal. For example, we are exploring linkages with CQIs and socioeconomic status as risk-adjustment approaches. Within the Michigan Value Partnership (MVC), a consortium of 77 Michigan hospitals, we’re also doing a lot of work exploring episode payments for over 20 conditions.

You’re a urologist. How did you get interested in studying health care policy?
Ellimoottil: At this point, urological conditions do not make up a big part of Medicare spending. The questions that our research group explores are relevant across all fields. My interest in bundled payments started early in my research career and was accelerated during my postdoctoral fellowship from the Agency for Healthcare Research and Quality (AHRQ). Our team is always looking for new ideas related to risk-adjustment and episode payments, and so we welcome conversations with and contributions from anyone who is interested in these areas. 

Featured IHPI Members