Accountability for Outcomes in Payment Programs: Going Beyond Better Measures

Numerous federal payment reforms have emerged as part of the ongoing nationwide shift toward value-based payment and care. Although individual programs possess different features and rules, they share the goal of encouraging clinicians and provider organizations to assume accountability for patient outcomes. For example, accountable care organizations are designed to hold participating clinicians accountable for the total costs of care and quality outcomes for a defined population of patients. Episode-based payment models, such as bundled payments, hold participants accountable for episode-specific quality and costs.

As noted in an Annals Ideas and Opinions article (1), achieving this overarching goal requires use of strong outcome measures. In particular, the authors suggest 4 key criteria in evaluating the appropriateness of measures used in payment and other programs:

1. Clear relationship between clinical care and changes in the outcome measure
2. High precision in measuring outcome
3. Low risk for the outcome measure to create unintended consequences
4. Accurate methods for risk adjusting the outcome measure

I agree with the rationale underlying these criteria. However, from my perspective as a clinician, health system administrator, and policy researcher, few programs would likely pass muster if fully held to the outcome measure standards embodied in these 4 criteria. Consistent with this concern, only 3 of 10 evaluated measures in the article met all criteria (it is also notable that none of the 3 were based on claims data that are most widely used in federal programs). Perhaps more important, the issues raised through the latter 2 criteria—avoidance of unintended consequences and risk adjustment—apply not just outcome measures, but to programs more broadly.

Setting aside outcome measure issues, payment initiatives have long been plagued by the risk for unintended consequences and need for risk adjustment. For example, the Annals authors use 30-day stroke mortality as an example of unintended consequences, noting how clinicians might attempt to keep patients alive beyond the 30-day window rather than implement more appropriate measures that would lead to death within 30 days (e.g., palliative care). While that dynamic is certainly a risk, there are also larger concerns: Providers might respond alternatively by avoiding program participation altogether or avoiding certain types of patient after participating. Such “cherry-picking” has been well documented in prior programs.

Across a variety of settings, risk adjustment has also been inadequate due to limitations in underlying data (not capturing all factors associated with outcomes of interest) and/or implementation (not utilizing captured factors in program design). Other factors, such as changes in coding practices, not only complicate efforts to risk adjust outcome measures, as the Annals authors point out, but also efforts to risk adjust other (e.g., cost) measures and understand the impact of payment programs on unintended consequences.

Collectively, these issues highlight that while outcome measure selection plays a role in promoting accountability for patient outcomes, policymakers may be more able to address the problem by contextualizing it alongside other considerations in payment program design. Given limited resources and competing priorities, policymakers are unlikely to have the ability to focus on outcome measure issues separate from other program considerations. Instead, it could be beneficial to identify a challenging topic and address all of its potential manifestations together. For example, focusing on the topic of risk assessment could allow policymakers to weigh and address its multiple implications—inclusion in or exclusion from a program, protections for high-risk individuals within programs, risk adjustment for outcome and cost measures—together, rather than address each of these separately.

Ultimately, there are no quick or easy answers for how to address these complexities while pursuing aspirational goals in outcome measure and program design. However, it may be fruitful to wrap considerations about outcome measures together with other related issues to help ensure that payment programs hold providers accountable and achieve the goal of improving patient outcomes.

Reference

1. Baker DW, Chassin MR. Holding providers accountable for health care outcomes. Ann Intern Med. 2017;167:418-423. [PMID: 28806793] doi:10.7326/M17-0691