Clinical Versus Financial Risk in a Value-Based Purchasing Environment: Part I

EDITOR’S NOTE: What follows is the first of a two-part series examining how health risk, severity, and complexity impact healthcare policy, payment, and quality assessment.  

There is little doubt that healthcare policy is moving away from a service-centric model towards a value-centric model. An article in the New England Journal of Medicine on standardizing outcome measures sums up this evolving direction well: “The arc of history is increasingly clear: healthcare is shifting focus from the volume of services delivered to the value created for patients, with ‘value’ defined as the outcomes achieved relative to the cost.”

The proposed implementation of recent legislation under the Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) has caused the healthcare industry to focus on Alternate Payment Models (APMs), the Merit-Based Incentive Payment System (MIPS), Accountable Care Organizations (ACOs), the Comprehensive Care for Joint Replacement Model (CCJR), Hierarchical Condition Categories (HCCs), and a host of other acronyms. We seem to be so driven by regulatory acronyms that we may be losing sight of the basic concepts of the underlying directional change. The models that will eventually accomplish this aim are still in a confusing state of flux. There is no clear direction that points to which model or models may eventually replace the purely fee-for-service model that has failed the test of achieving sustainable growth.

United States’ expenditures for healthcare is over 2.5 times higher than Japan’s per capita spending, Japan’s mortality rate is ranked No. 1 globally, and the U.S. is ranked 41st on the OECD listing of countries. The rate per thousand for infant mortality for the United States is nearly four times the rate in Japan. 

Despite the fact that the U.S. spends more per capita than other nations, we still have 28.5 million uninsured residents in this country. While these data points do not tell us why, they clearly suggest a healthcare model that cannot be sustained.

Given these financial realities, it is not surprising that the clear focus of healthcare policy is moving towards value. To achieve this transition, costs must be reduced and healthcare outcomes must be improved. The thousands of pages of regulatory documentation that propose how this might happen can be overwhelming to comprehend and are likely to continue to evolve as proposed strategies are field-tested. There are, however, key concepts that are likely to persist regardless of the specific methodologies used. These concepts should be the focus for how healthcare entities prepare for this evolution.

It is highly unlikely that the fee-for-service model (that is, the service-based payment model) will go away anytime soon. Fee-for-service payment is deeply ingrained in how individual providers have been historically financed. What’s more likely is that payment will be modified over time based on value metrics and population-based budgets for integrated “at-risk” entities. There would be a direct impact on the entity’s payment, as well as on the individual provider’s compensation, if budget targets or outcome expectations are not met. The impact on revenue could occur at any level under a number of these proposed or currently implemented alternative models.

Alternative Payment Models

The Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) represents one of the most dramatic legislative policy approaches in decades for changing the existing payment paradigm. MACRA proposes a number of different alternatives for evolving this value-based, data-driven change.

MIPS provides a mechanism for providers who are not “significantly participating” in some Advanced Alternative Payment Model (A-APM) under an Eligible Alternative Payment Entity (EAPE). Proposed methodologies for value-based purchasing include payment bundling, risk adjustment, episode definitions, and clinical and financial outcome metrics that are intended to modify payment based on value and resource use.  Most of these approaches are complicated and confusing. It is unlikely that any of these methodologies will see long-term adoption, at least as currently configured, given historical experience with similar efforts. What is more likely is that data-driven, risk-adjusted payment and data-driven, risk-adjusted value measurement will persist in some form.

The Role of Risk

 A full explanation of all of the core concepts regarding MACRA and the value-based purchasing approach is beyond the scope of this paper. Rather, I focus more specifically on risk and how data may be used to assess it. Before proceeding, however, it is necessary to define risk, which could mean “financial risk” or “clinical risk,” and to understand the relationship between these different concepts.

Financial Risk:

Financial risk addresses the cost of providing care to a population or some stratification of the population that may not be supported by the level of reimbursement associated with that care. In a pure fee-for-service model, the approach to mitigating risk is different at the provider level versus the payer level. For the provider, mitigating financial risk means performing more payable services to generate more revenue. For the payer, mitigating financial risk involves efforts to curb the number of payable services.

As financial risk is increasingly borne by providers, it is incumbent upon each provider to mitigate risk by controlling costs. Historically, the concept of cost control has not been as significant a component of the provider’s business focus as the concept of revenue enhancement.

Clinical Risk:

Clinical risk is about the risk of an undesired outcome for a patient relative to the baseline status of the patient’s health condition. “Sicker” patients are more likely to experience complications or undesirable outcomes and to require a greater amount of healthcare resources in an effort to maintain or improve their health status.

Mitigating clinical risk becomes more difficult, and more resource-intensive, based on the degree of severity and complexity of underlying health conditions. For any given health condition, the level of clinical risk can be impacted by the following:

• The specific definition of the patient condition or diagnosis
• The level of severity of the condition
• Comorbid condition(s) that may impact outcomes
• The availability of proven and effective treatment modalities for that condition
• The ability of the patient to comply with recommended care
• Resources available for care

The Overlap:

In most instances, clinical risk has a significant impact on both the cost and outcome components of value measurement. Conditions with significantly higher risk of unfavorable clinical outcomes usually incur higher costs for care. Outside of death or loss of a member from the population cohort, clinical outcomes and financial risks are tightly related.

Accessing Risk

Accurate assessment of risk from both the clinical and financial perspective requires data and data aggregation that is:

• Accurate – reflects the relevant facts of the patient health state
• Complete – includes all parameters relevant to clearly defining the condition, as well as the factors that impact risk, severity, and complexity
• Specific – includes sufficient detail and granularity
• Properly attributed –  accurately related to providers, patients, conditions, and episodes of care so as not to over- or under-attribute data relevant to the assessment of costs and outcomes
• Properly categorized – aggregated at a level that is clinically meaningful, reasonably homogenous, predictive of risk, of sufficient granularity, consistently applied, and that clearly defines what is included or excluded in the data categorization scheme
• Adequate sample size – avoids statistical anomalies
• Hierarchal assignment – avoids over-assigning risk by double-counting risk and severity factors that may not be additive to the overall patient health state
• Incentivizes data quality – provides incentives needed to assure that the data meets the above requirements and does not create perverse incentives for manipulating data to improve revenue
• Standardized condition coding – limits variations in diagnosis coding patterns for the same condition at the same level of severity and complexity by different observers

While it is unlikely that all of these requirements will typically be met, understanding the limitations of the data used to assess risk is critical to achieving appropriate payment and adjustment to quality metrics. Assessment of data quality is every bit as important as risk assessment methodologies.

The best methodologies will fail if applied to inconsistent, inaccurate, and incomplete data.

A special thank you to Bob Perna, director of healthcare economics at the Washington State Medical Association, for his support, advice, and input into this paper.

EDITOR’S NOTE: The second installment of this series will focus on existing and evolving models for risk assessment, as well as challenges that need to be addressed.