In 2013, The Center for Medicare and Medicaid Innovation launched the Bundled Payments for Care Improvement (BPCI) initiative, a program that proponents hoped could rein in health care costs by “bundling” payment for the full gamut of services that comprise an episode of care. The model certainly seemed like a good bet, as it would reward hospitals for reducing the cost of soup-to-nuts care for any of 48 conditions and penalize them for overruns. Indeed, bundled care for hip and knee replacement has been a dramatic success with clear savings and no increase in emergency department visits, readmissions, or 30-day mortality.
However, our research suggests that bundles may not work as well for other types of conditions. That doesn’t necessarily mean that the problem lies with the bundled payment model itself. Rather, we think it could lie with the fragmented nature of the patient journey in a dysfunctional system, which is exposed by medical bundles’ lack of impact. But before delving into that diagnosis, let’s step back and look at the research.
Sponsored by Medtronic
Creating better outcomes at reduced cost.
With our colleagues John Orav and Jie Zheng, we used Medicare claims from 2013 through 2015 to identify admissions for the five most common medical conditions covered under the Medicare bundled payment initiative: heart attack, heart failure, pneumonia, chronic obstructive pulmonary disease, and sepsis. We calculated the costs of each “episode” — the hospitalization plus all costs in the 90 days post discharge — for hospitals that joined BPCI as well as hospitals that didn’t join (our control group). We then looked to see whether costs dropped more in the BPCI hospitals than the control hospitals after the program started. Overall, the average Medicare payment per episode of care across the five conditions — about $24,000 — dipped just a few hundred dollars, a statistically insignificant amount. In addition, there was no difference in the change over time based on whether hospitals were or weren’t participating in the program. We also didn’t find any differences in clinical complexity, length of stay, emergency department use or readmission within 30 or 90 days after hospital discharge, or death within 30 or 90 days after admission between the intervention and control hospitals. In short, for these five common medical conditions, bundled payment had no impact on costs or clinical outcomes — at least in the program’s first year.
We don’t know exactly why bundles were successful for hip and knee replacements, but not for medical conditions. Perhaps all the hospitals that signed up for the hip and knee bundles were much more motivated than the hospitals that signed up for the medical bundles. Or perhaps we just need to wait longer to see an impact of bundling on a wider range of conditions.
However, we suspect that these patterns reflect the complexity and fragmentation of the patient journey for common medical conditions as a cause and shed light on not only how we might help hospitals succeed under bundles, but how we might also improve patient experience.
Hip and knee replacements are discrete, pre-planned events with a fairly standardized and consistent patient journey, from pre-operative evaluation to scheduled OR date to post-operative rehabilitation, and with a single “captain” of the ship. The surgeon performing the operation is ultimately responsible for the entirety of that patient’s clinical course. For the patient too, there is an obvious point person before the admission, during the hospitalization, and after discharge. Most patients who have had a total joint replacement could tell you the name of their surgeon even years after the procedure, often with great fondness.
Medical admissions follow an entirely different course. Consider what happens to a patient coming to the hospital with a heart failure exacerbation. She certainly did not plan the admission, and may have no pre-existing relationship with the clinician she sees in the hospital. She sees the emergency department physician who happens to be on duty that day, and depending on clinical severity, bed availability, and the call schedule of the residents and interns if she’s in a teaching hospital, could end up admitted to a general medical service, hospitalist service, cardiology service, medical intensive care unit, or cardiac intensive care unit. Her care team could change daily, including the nurses and nursing assistants. During her stay in the hospital she may see cardiologists, internists, and nephrologists, as well as physical therapists, pharmacists, social workers, and discharge planners. At discharge, she may or may not be scheduled to come back to see anyone that was involved in her inpatient care, depending on where she wishes to establish or maintain cardiovascular follow-up, and there is likely no post-discharge protocol for follow-up and rehabilitation, nor formal relationships with the nursing facility or home health agency to which she is discharged.
To improve this patient’s journey though the health care system — and increase the chance that bundled payments can help her achieve better outcomes and help the system lower its costs — we first need to understand the journey. Patterns of care are heterogeneous for medical conditions compared with discrete elective surgeries. And perhaps the overwhelming complexity and fragmentation of the patient journey for most unplanned medical admissions explains both why medical bundles are hard, and why they are so very important.
The early failure of medical bundles is a window into the disjointed, piecemeal health system most of our patients (and clinicians themselves) experience. Even the first step for hospitals electing to participate in BPCI for a medical condition — trying to figure out who needs to be around the table to discuss improving care across the clinical episode — is not an easy one. But rather than discourage us from using bundles as a way to improve care, this complexity makes it all the more critical to use mechanisms like bundled payment programs to incent health systems to change the paradigm. Hospitals can and should design and implement standardized clinical pathways and provide more coordinated, efficient care for medical conditions, and bundling may be a powerful policy mechanism to help get us there.
The five years of experience we have had with BPCI seems like a sufficient time to have learned a lot about it. And perhaps with more time and greater incentives hospitals will be better able and more willing to make the changes needed in care delivery for medical bundles to be effective and to create a better experience for patients. But for now, we are only scratching the surface, and there is much more to learn about the use of bundling for different conditions and different patients. Indeed, bundling is just one in a series of policy innovations that Medicare and others are experimenting with to move us past traditional fee for service. The road to better policy will be long and winding. We can only hope that on the way we will discover much about more efficient care, and most importantly, ways to both control costs and improve outcomes that matter to patients.