[Skip to Navigation]
Sign In
JAMA Forum Archive, 2012-2019: Health policy commentary from leaders in the field
JAMA Forum

Will Episode Payment Models Show How to Better Pay for Hospital Care?

What’s the right way to pay for hospital care? Believe it or not, even in 2016, we don’t really know. In many advanced economies, hospital care is paid with a yearly budget for a given population. In other countries with advanced health care systems, hospital care is paid piecemeal; every inpatient day generates new bills for the room, the nursing staff, and the procedures.

Image: Aubrey LaMedica/Harvard T.H. Chan School of Public Health

Piecemeal was how Medicare paid for hospital care until the mid-1980s when it switched, for a large majority of hospitals, to what is known as prospective payment. Paying prospectively was simple: identify why someone was admitted and provide a lump sum for that hospitalization based on our best understanding of what it costs to care for that condition. Although the formula for payments is complex, the idea underlying it is straightforward. We pay more for more complicated conditions and for patients with more complex issues, and then let hospitals figure out how best to spend the chunk of money from an “inpatient bundle.”

Most policy makers believe prospective payment has worked well. Seen in one light, we know that bundling payments for inpatient care has reduced hospital spending growth. For example, inpatient spending on acute myocardial infarction (AMI) grew just 0.9 % annually from 1997 through 20 10, far below the overall growth in health care spending. However, the number and types of hospitalizations have changed, such that per capita hospital spending has grown at nearly 5 % between 2000 and 2010.

But the news is worse. Much of the spending growth around inpatient care, which has grown slowly, has shifted to postacute care—long-term hospital care, rehabilitation care, and skilled nursing facility care—where spending ha s skyrocketed. Inpatient bundles may appear to have successfully curtailed spending growth, but like pushing on a balloon, the spending growth bulges elsewhere.

Given these challenges, the Centers for Medicare & Medicaid Services (CMS) has experimented with 30-day bundles for the last several years, allowing hospitals to voluntarily choose which conditions they want to bundle. The hospital receives a single payment for the expected costs of a patient’s hospitalization for the condition and first 30 days after discharge. If the hospital uses fewer services, it gets to keep some of the savings; if it spends extra, it typically has to pay Medicare back. The effect of this program, which has had only modest participation, is largely unknown.

To move toward bundles that extend beyond hospital care, CMS introduced 2 new programs over the past year. The Comprehensive Care for Joint Replacement, introduced in April 2016, focuses on hip and knee replacement surgery for all acute-care hospitals in 67 metropolitan areas. On July 25th, CMS announced a substantial expansion, the Episode Payment Model (EPM) program, for 3 conditions: AMI, coronary artery bypass graft (CABG) surgery, and hip and femur fracture surgery. All health care services for these conditions that occur within 90 days of discharge, rather than within 30 days, are included in a single bundled payment that a hospital must manage. The rationale is that most of the health care services that occur soon after discharge are related to the condition for which the patient was hospitalized, and holding hospitals accountable for that care makes sense.

Will EPM Programs Improve Care or Lower Costs?

Careful assessment of the effect of EPM programs on both the costs and outcomes of care is critically important. The EPM has several features that should give us reasons for optimism and other that should be addressed to make the program better.

Whether the EPM will improve care or not will be driven in large part by having the right set of quality measures. The easiest way to make money in any bundled payment program is to skimp on care. To guard against this risk, such programs need metrics that ensure patients are getting good care, not just less care.

For AMI, CMS will use mortality, patient experience, and the number of days a patient spends in the hospital as quality metrics. For CABG surgery, CMS will focus on mortality and patient experience. For hip fracture surgeries, CMS will assess quality using a mix of complication rates, patient experience, and voluntary reporting of patient-reported outcome measures (PROMs).

Although these measures are a good start, small but important changes would allow us to do better. Mortality rates surely are most important outcome measures for AMI and CABG, but functional status, measured through PROMs, matter immensely to patients. However, progress on measuring PROMs has been exceedingly slow. Counting complications that cause substantial morbidity, such as health care–associated infections, falls, or thromboembolic events, is also  important for AMI and CABG surgery, not just for hip fracture surgery (for which they are set to be measured). Expanding the list of quality metrics to measure more of what matters is vital.

Timing of Bundles

A second issue is the time line of postacute care. Many of us have called for timing of bundles based on clinical rationale. For some conditions, bundling 90 days of health care services makes sense because most of the spending in that period is likely due to the index hospitalization. For other conditions, the right time line might be 60 days or even less.

There are 2 reasons to be more clinically nuanced about timing. First, it signals to clinicians that CMS understands that not all medical and surgical conditions are the same and that clinical trajectory has been taken into account. Second, making bundles longer than necessary creates little additional clinical efficiency but incentivizes skimping on care that is unrelated to the original condition. Being more clinically nuanced can be done by examining the data and seeing how long spending continues for a given condition and what types of health care services are being used as patients get further away from the index hospitalization.

Finally, CMS should consider implementing the EPM in ways that will allow all of us to learn the best way to bundle and pay for hospital and postacute care. Right now, CMS has randomly chosen 98 communities and is proposing implementing the EPM in the same way across all of these communities. This approach will allow us to study whether the EPM worked, but that is not enough. A more useful approach might be to try several different versions of the EPM. For instance, in some places, CMS could bundle for 90 days while in others, 45 or 60 days. Some bundles could include physician visits while other bundles, not.

Although several new models of health care delivery have been under study, we have learned too little about which models optimize clinical outcomes and reduce waste. We need to treat the EPM as an experiment and implement different versions across different communities in ways that allow us to better understand how best to pay for hospital and postacute care. We cannot afford to miss this opportunity.

About the author:Ashish K. Jha, MD, MPH, is K. T. Li Professor of Global Health and Health Policy at the Harvard T. H. Chan School of Public Health and a practicing internist at the Veterans Affairs Boston Healthcare System. He received his doctor of medicine from Harvard Medical School and was trained in internal medicine at the University of California, San Francisco. He received his master’s in public health from Harvard School of Public Health. Dr Jha’s major research interests lie in improving the quality and costs of health care. His work has focused on 4 primary areas—public reporting, pay for performance, health information technology, and leadership—and the roles they play in fixing the US health care system.