Bundled payment initiatives are a growing form of value-based payment. The use of bundled payments can align reimbursement with the health care triple aim of improving experience of care, improving population health, and reducing total costs of care. Successful bundled payment initiatives have demonstrated an ability to both lower costs and improve health care quality.
However, bundled payments also change financial incentives because the model shifts risk from payers to providers. This may result in unintended consequences, including underutilization of needed but costly services or avoiding caring for the sickest patients.
To address these concerns, we identified three principles that are designed to both maximize the benefits and minimize negative consequences of bundled payments:
- Bundled payments should be adequate for the care needed to achieve optimal patient outcomes. This includes setting an appropriate time frame, providing sufficient reimbursement for services and technology, and targeting a homogenous population.
- Evidence-based treatment variability should be incorporated into the bundled payment, including risk adjustment as needed and allowances for patient choice.
- Quality metrics should be used to ensure appropriate care in a bundled payment program. Quality metrics should encourage appropriate care regardless of health status and should have financial implications.
Bundled payments should be adequate for the care needed to achieve optimal patient outcomes
Bundled payments should sufficiently reimburse the process elements and resources needed to achieve the desired patient outcome(s). The ultimate goal should be improving patient health by delivering the right services to the right patient at the right time. Fee-for-service reimburssment can incentivize providers to deliver unnecessary, low value, and potentially harmful care. Conversely, bundled payments, may incentivize underuse and result in patients not receiving needed care.
To find the appropriate balance and prevent underuse, financial bundles should be designed so that providers receive adequate compensation for clinically necessary services. For example, the 2016 Medicare Comprehensive Care for Joint (CJR) Program, which is focused on hip and knee replacement, takes into account all related medical services including but not limited to inpatient rehabilitation facility, skilled nursing facility, home health agency, hospital outpatient, outpatient therapy, clinical laboratory, durable medical equipment, and Part B drugs.
The cost impact of evidence-based clinical practices changes should also be incorporated in a timely manner. When payment bundles are not adjusted to compensate providers for new technologies, providers are incentivized to use older less expensive technologies irrespective of value. The net result may be harm to the patient and a reduction in innovation.
For example, a payment bundle for Hepatitis C treatment in 2014 would have needed to evolve as new oral therapies such as Sovaldi and Harvoni became available. Although these drugs have been shown to be cost effective in the long term, their use would likely have been discouraged by an outdated Hepatitis C bundle. As standard practice, Medicare updates its outpatient prospective payment amounts on an annual basis. In addition, Medicare routinely authorizes an additional drug payment under its outpatient prospective payment system for new devices/biopharmaceuticals that both present significant clinical benefit and incremental costs.
The time frame of services should also reflect the full cycle of care. Consistent with reimbursing for outcomes rather than individual services, the bundled payment time period should be aligned with the clinical outcomes of interest. A Harvard Business Review article highlighted a bundled payment design for rotator cuff repair, where the bundle covered all related services through one year post surgery including post-surgical complications, clinic visits, and physical therapy. This approach incentivizes providers to focus upon the desired outcomes rather than just short-term process elements.
The services and medications required to achieve the desired evidence-based outcomes should be relatively homogenous across the target patient population. The challenge is that health care costs are typically concentrated among a small portion of the population. This phenomenon creates the potential for under treatment when individual patient costs greatly exceed the average. Therefore, the population selected for the bundle should minimize the number of these atypical patients, or find analytic approaches to identify them via risk adjustment.
When atypical patients cannot be identified, the clinical area may not be appropriate for bundled payment. For illustration, we reviewed the Medicare CJR patient population for homogeneity by reviewing the operating room procedures that can trigger an episode under this program. What we found is that ankle replacement is included along with hip and knee replacement.
However, Medicare reimbursement does not differentiate between hip, knee, and ankle replacement. Figure 1 shows why this is a concern as 2013 median hospital costs (not reimbursement) for patients with ankle replacement were 33 percent higher than hip or knee replacement, which creates the potential for under treatment among this group of patients.
Evidence-based treatment variability should be incorporated into the bundled payment
Bundled payment should reduce treatment variability by incentivizing providers to consider which resources are necessary. However, a portion of this variability is clinically determined and driven by factors such as age, genetics, comorbid conditions, disease severity, gender, environment, and personal preferences. If payment bundles use a “one-size-fits-all” approach, providers may lose money on the sickest patients or meet financial constraints; as a result those patients may not receive all of the care that they need.
Bundled payment compensation should be adjusted to account for variation of patient health status and required resources. A flat or average payment may be appropriate when providers care for a large population with the clinical condition of interest or the patient population is relatively clinically homogeneous. More typically, providers care for smaller patient populations where significant clinical variability may result in either over- or underpayment.
For example, some patients with rheumatoid arthritis respond well to relatively inexpensive generic drugs like methotrexate. However, other patients fail generic agents and require more costly biologic therapies. While flat payments may sometimes suffice, the care of patients with differing resource needs may require payment adjustment.
Risk adjustment leverages health status to calculate a score that prospectively or retrospectively adjusts compensation. Application of risk adjustment is complex and data intensive and will challenge the ability of many insurers to accurately predict costs. Despite many years of learning, the Hierarchical Condition Categories methodology, which is used to adjust Medicare payment to payers, still over and under predicts costs for patients with the same condition. Complexity also reduces transparency to providers and may present organizational challenges.
Exclusion of services/medications that are highly variable, high cost, and unpredictable from a bundle (e.g. prior rheumatoid arthritis example) reduces the risk of provider financial loss and under treatment. Risk of overutilization may be mitigated through performance metrics linked to payment. Clinical outliers, which are difficult to pre-identify and eliminate, may be handled via an exception process or additional retrospective payment that reduces incentives for under treatment. Exclusion of outliers is standard practice for all Medicare prospective payment programs. The best approach to payment adjustment will depend on organizational capabilities, disease epidemiology, and the availability of accurate cost predictors.
Payment bundles should allow for patient choice among appropriate therapeutic options. Patients make therapeutic choices that fit their utilities; outcomes have been shown to be worse when patient preferences are ignored. Given this trade-off, services/medications could either be excluded from the bundled payment or separated into different bundles when patient preferences are highly variable and the associated treatment options have markedly different costs.
Based on National Comprehensive Cancer Network (NCCN) guidelines, women with stage I or II breast cancer may choose either breast conserving therapy or total mastectomy. In this clinical scenario, the payment bundle should provide the flexibility to the patient to decide her preferred therapy.
Quality metrics should be used to ensure appropriate care in a bundled payment program
Bundled payment requires oversight to ensure that the incentive to use resources wisely does not shift care from potential overuse to significant underuse of needed care. Simply paying providers adequately does not guarantee they will provide the appropriate services.
Also complicating matters is that the bundling of services into one payment may reduce visibility into the clinical services provided to the patient when information is not collected. To address these concerns, quality metrics should be used to provide the right mix of incentives and disincentives to optimize care.
Designing Quality Metrics
Quality metrics should be used to ensure that patients receive clinically appropriate care, especially needed services that may be costly. The ability to measure quality is not the same across disease states. For conditions like diabetes, hypertension, or hypercholesterolemia, there are routinely used quality metrics (e.g., HbA1c, blood pressure control, cholesterol levels) and these metrics could indicate if patients are not receiving needed services.
However, quality metrics are often not routinely used in many high-cost disease states such as rheumatoid arthritis, multiple sclerosis, and cancer. For example, excluding screening-related metrics, the quality metrics for the 2016 Medicare Shared Savings Program and Pioneer Model do not include any performance metrics for these high-cost diseases. Where metrics cannot accurately assess the value of care, payment bundles may not be appropriate.
Metric performance thresholds should incentivize appropriate care for all patients regardless of health status. When performance metrics are designed around a single threshold, it encourages providers to focus on a subset of patients rather than the entire population. For example, a single diabetes performance threshold of HbA1c < 7 percent encourages providers to focus on the set of patients that he/she is able to move below the threshold.
A significant percentage of patients will not satisfy this performance metric for a number of reasons including comorbidities. Despite not meeting this threshold, significant clinical benefit can still be delivered to these patients. This example highlights the need for comprehensive performance metrics; otherwise physicians may be incentivized to focus only on a population subset. Comprehensive approaches to performance metrics include multiple thresholds for a single quality metric, or rewarding incremental improvement.
Financial incentives should be tied to quality metrics. In many instances, the amount of financial risk linked to quality metrics is small in comparison to the financial incentives associated with bundled payment. Physicians typically receive anywhere from 50 percent to 100 percent of the shared savings, while the payment/penalties for quality remain modest at 1 percent to 2 percent of total reimbursement.
It is safe to say that the impact of quality metrics is significantly lessened when this type of imbalance exists. Rigorous studies on the appropriate balance between savings and quality are lacking, but an approach that seeks to protect patients from underuse would seek to assign greater financial weight to quality metrics than what is currently in place.
Moving Forward With Bundled Payments
Bundled payments can achieve the triple aim of improving experience of care, improving population health, and reducing total costs of care. However, this requires careful thought and consideration of provider incentives and how they impact patient care. Past successes and failures demonstrate that it is unlikely that any bundled payment program is going to initially achieve 100 percent success; programs should be built on continuous learning.
Successful bundled payment programs will 1) begin with mostly homogeneous patient populations and well-defined quality metrics and expand to other areas as better risk adjustment and quality metrics become available; 2) put protections in place to safeguard the most vulnerable stakeholder, the patient; and, 3) place an emphasis on aligning physician incentives with the patients’ best interest.
Expansion of bundled payment requires that the significant gaps and barriers in accurately adjusting payment and measuring quality be addressed. From a policy standpoint, this requires renewed focus on the development of payment adjustment techniques that accurately predict treatment costs for the individual and the expansion of quality metrics into areas where they do not currently exist. Without these changes, bundled payment will need to remain limited in scope; otherwise patient care could be compromised.