CORR Insights®: Is There an Association Between Bundled Payments and “Cherry Picking” and “Lemon Dropping” in Orthopaedic Surgery? A Systematic Review

Where Are We Now?

The intersection of patient care, economics, and health policy, not surprisingly, is rife with debate in the orthopaedic community [1]. As healthcare costs in the United States continue to trend upward, there is a critical need to find alternative reimbursement models that limit costs without diminishing care quality or access. Bundled payments have been touted as a potential solution moving forward, and the idea behind them certainly is appealing. By providing a lump-sum reimbursement for the entire episode of care based on the average cost of perioperative treatment, surgeons are theoretically incentivized to limit unnecessary costs to receive increased compensation for their services.

But what happens when the patients being served do not fall into the same category as the “average” patients on which the cost model was designed? This is a prominent criticism of the bundled payment model—that it disincentivizes surgeons from taking care of the patients who have a greater chance of postoperative complications, such as infection, and are more likely to require a higher level of perioperative management, thus incurring greater costs [5, 6]. Left unchecked, this can lead to situations where safety net hospitals are the only ones willing to take care of the patients who lack adequate healthcare coverage or have the most complex and difficult-to-treat conditions, often incurring a huge loss in the process. A sad case in point is Hahnemann University Hospital in Philadelphia, PA, USA, which had to close its doors in 2019 and disenfranchise an entire population of patients that relied on its care, while other private institutions in the city logged record profits in the same year [4, 7]. Prior studies have commented on this phenomenon, asserting that more nuance is required in reimbursement models to preemptively risk-stratify and account for the patients who are likely to incur the highest healthcare costs, so as not to limit their access to care [3, 8-10].

In their systematic review, Bernstein and colleagues [2] examine the impact of bundled care payments on patient “cherry picking” in orthopaedics. They examined 10 studies to determine whether the implementation of bundled care payments changed the sociodemographic, case-complexity, and healthcare resource utilization characteristics of the patients treated, and found that the differences before and after bundle payment implementation were minor but with some notable concerning trends. These included a decreased percentage of treated patients with dual public insurance eligibility, as well as fewer patients with disabilities or higher comorbidity scores.

While all these changes appeared relatively minor in the data studied in this paper, they raise questions about the future of these payment models and the research questions needed to investigate them in more detail.

Where Do We Need To Go?

As the bundled payment model moves beyond its nascent stages and into an established economic entity, it is important that we continue to critically assess the impact of this model and other cost containment initiatives on patient care. While it may sound promising in theory, we need to ensure that it does not deny care for patients who lack the financial means to decide where to get their care, such that they can only receive care in safety-net hospitals, and risk causing economic collapse for inner city hospitals.

Studies such as those examined in this systematic review [2] are a good starting point. But to take it further, we need more detail. Future research must look at the impact on entire sociodemographic groups in a defined geographic area to determine whether certain groups are getting preferential care at the expense of others. We need to look at both operative and nonoperative care of patients to examine whether there are differences in the treatment plans and options offered to patients based on their socioeconomic and risk profiles. Additionally, we need to compare institutions within these areas to see whether some are profiting while others fail, particularly if the reason for success and failure is the cherry picking by some hospitals of patients with less-complex medical problems and more-remunerative insurance, leaving the more-challenging balance to safety-net hospitals. The findings of this study [2] tell us that there may be subtle trends emerging, but we need to get into the details to really see the broader socioeconomic and healthcare implications of these changes.

How Do We Get There?

Registry data provide the perfect vehicle for investigating socioeconomic and reimbursement changes on a larger scale, and thus are perfect for this topic. But to truly be helpful, the data need to be more granular; we need to study not merely the patients getting care, but also the patients being denied care. To begin with, existing registries should be altered to include more hyper-specific regional data. While this may be difficult in areas with smaller populations and fewer hospitals, it could lend new insight when geographic areas with wide variability in patient demographics and with numerous treating institutions for a large catchment area are examined. Researchers then could examine the average socioeconomic and health profiles of the patients seen and treated at each institution. Are the patients incurring the highest healthcare costs all clustered at certain institutions? Did they originally seek care at other institutions, only to be turned away and forced to follow-up elsewhere? Did these patients have satisfactory patient-reported outcomes after surgery when compared with regional norms and were the treating institutions able to still make a profit overall? The data need to be detailed enough to examine all of these questions over time and their temporal associations with the increasing implementation of bundled care. Are these patients migrating more to the safety-net institutions? Are they becoming less prevalent among operative patients overall? Existing databases such as the National Surgical Quality Improvement Program (NSQIP) database could likely be modified to include these extra data points without too much overhaul of existing infrastructure, thus increasing their utility for epidemiological studies on this topic. By using detailed, reliable, large-scale databases to examine these questions, we can start to fully understand the long-term implications of bundled care models on the care of the patients with less healthcare insurance coverage and more complex preoperative comorbidity profiles.

To truly be useful, these data must fuel policy initiatives, including modified bundles for underinsured patients with higher case complexity at catchment hospitals. In the long run, it appears we may be migrating toward a model where the patients that require the most complex surgical management are funneled to safety-net hospitals that take on the most challenging diagnoses. Implementing a two-tiered payment system that accounts for these differences in patient characteristics may be a way to ensure that hospitals can earn enough money to continue to perform their missions, without denying care to patients who need it the most. In order to make these changes, a concerted effort between researchers and national orthopaedic societies must be made to convey these findings to political leaders who have the power to lobby for policy changes that translate these ideas into reality.