Inherent biases in program development and modeling can contribute to health disparities, but existing ASOP guidance can help identify and mitigate these disparities.
By Ian McCulla
As trusted advisors for organizations interested in managing risk, actuaries are routinely tasked with performing financial estimates and collaborating with other business units in developing, implementing, and monitoring programs that carry financial risk. Within the health care sector, and with respect to major medical coverage in particular, actuaries play a role in setting premium rates, developing performance metrics and corresponding incentives or penalties, and identifying members who would benefit from case management or disease management programs.
The risk assessment component of these tasks generally relies on a model, which includes:
- an information input component, which delivers data and assumptions to the model;
- a processing component, which transforms input into output; and
- a results component, which translates the output into useful business information.
The modeling process, and even the development of the model itself, combines art and science, relying on a mix of objective and subjective decision-making based on actuaries’ professional judgment and professional guidance.
Key among available resources for actuaries are the actuarial standards of practice (ASOPs), which help actuaries to weigh decisions such as appropriate data sources, the appropriate risk classifications to use, and credibility of the data; however, it is the responsibility of each actuary to ensure appropriate application of the ASOPs.
The American Academy of Actuaries’ Health Equity Committee wrote this article to raise awareness of health equity considerations that actuaries may want to incorporate into their day-to-day work. Through a quality metrics lens, the Committee seeks to highlight how inherent biases in program development and modeling processes can contribute to health disparities and point to how existing guidance in the ASOPs can be leveraged to help identify and mitigate those disparities.
Health Plan Quality Metrics Example
Quality metricsare used to evaluate different kinds of health plans on quality of care provided, patient outcomes, and improvement efforts, which may influence plan enrollment and revenue. For example, the National Committee for Quality Assurance, an accrediting body for providers and plans, issues an annual report card on ratings for commercial, Medicare, and Medicaid plans based on “the quality of care patients receive, how happy patients are with their care and health plans’ efforts to keep improving.”
Quality metrics are generally measured across the entirety of the enrolled population that qualifies for the metric. In an effort to improve quality measures across populations, health plans may use algorithms to help identify members who will benefit most from an intervention program, such as disease management, aimed at improving health status. Employing programs focused on overall efficiency and improving members’ outcomes will often result in improvement in associated quality metrics. If the algorithm is using data that include an inherent bias, however, the identification process may miss entire groups of members who would otherwise be eligible for and benefit from the program.
Similarly, an actuary may be tasked with utilizing historical intervention data to identify which members experienced the greatest improvement in health status and thus improved the plans’ quality metrics when included in the intervention program. Actuaries look for similarities in the members where the intervention was most effective. An actuary may find that members in a certain geographic area respond better to the intervention. However, it is possible that members in the given geographic area respond better because of better access to services or more stable communication methods.
Whether in support of algorithm development or in developing data sets, it is critical to ensure that the data tell the right story. It may, in fact, be necessary to modify an intervention’s approach in other geographic areas to better address member needs. In this example, intervention methods that target certain geographies (explicitly or implicitly via other model variables) may be the result of, and further contribute to, health inequities.
Discussion
The example illustrates how certain geographic areas and the members living in that area may be adversely affected by social determinants of health (provider access or communication stability). With that in mind, actuaries tasked with improving health quality metrics for health plans might consider expanding their consideration to factors influencing the risk classification implied by their models and whether they are in alignment with the intended purpose.
Actuaries informing the health quality metric definition may consider adjustments to the metric measurement that increase alignment of quality program incentives and program health equity goals. For example, stratifying the health quality metrics by certain member attributes such as race (if available) or geography may provide more equitable access to interventions aimed at improving the health of the members.
While social determinants of health and equity are not explicitly included, careful consideration of relevant ASOPs can guide actuaries as they contemplate the actions above. For example, ASOP No. 12, Risk Classification (for All Practice Areas), states that “The actuary should select a risk classification system that is appropriate for the intended use.” Are the attributes used in the actuarial model or financial metrics consistent with the intended use of the model? Are there latent attributes that are influencing the model in an unintended fashion?
ASOP No. 56, Modeling, states that an actuary should understand “limitations of the data or information, time constraints, or other practical considerations that could materially impact the model’s ability to meet its intended purpose.” Many public and private health care organizations have explicit references to health equity objectives. Could the models mentioned above be enhanced to further these objectives?
Conclusion
As professionals who understand both the detailed statistical nuances of model development and implications of financial policy, actuaries have a unique opportunity to be at the forefront of developing health care models, metrics, and financial policy that could potentially reduce health disparities. As noted, this article is meant only as a starting point to raise awareness of health equity considerations that actuaries may want to consider in their day-to-day work and how the ASOPs help guide the actuaries work in this area.
Ian McCulla, MAAA, FSA, is a principal and consulting actuary with Milliman. McCulla primarily consults with Medicaid state agencies on topics such as capitation rate development, provider payment, and alternative payment models. McCulla is an active member of the American Academy of Actuaries and Society of Actuaries work groups and committees, including those focused on Medicaid and health equity issues.
Definitions
The committee refers to the following definitions in its work:
- Health equity means that everyone has a fair and just opportunity to be as healthy as possible. This requires removing obstacles to health such as poverty, discrimination, and their consequences, including powerlessness and lack of access to good jobs with fair pay, quality education and housing, safe environments, and health care.
- Health disparities are differences in health or its key determinants that adversely affect marginalized or excluded groups. Disparities in health and in the key determinants of health are the metric for assessing progress toward health equity.
- Social determinants of health are nonmedical factors such as employment, income, housing, transportation, childcare, education, and the quality of the places where people live, work, learn, and play that influence health.
Source: Braveman P, Arkin E, Orleans T, Proctor D, and Plough A. “What Is Health Equity? And What Difference Does a Definition Make? ”Princeton, NJ: Robert Wood Johnson Foundation, 2017.