CMS defined a path for medical frailty. Now comes the hard part.

CMS’s new guidance on medical frailty gives states an important starting point for implementing Medicaid community engagement requirements. Use reliable clinical and administrative information already available before asking patients or clinicians to produce more documentation to prove a person meets the exemption criteria based on whether their physical, mental, or behavioral health conditions significantly impair their ability to comply.

The guidance also makes clear where the harder work begins. States have to turn that information into medical frailty determinations that are clinically credible, consistent, and workable across large Medicaid populations.

That challenge is bigger than medical frailty. Healthcare routinely has to translate complex clinical realities into classifications that can be applied consistently across populations. And the history behind that challenge is a learning opportunity.

The new framework narrows the questions states must answer, but it does not eliminate them. How should multiple diagnoses and their severity be weighed together? How should functional limitations affect a determination? How can states interpret the same kinds of information consistently? And when is the existing record truly insufficient, making additional patient or clinician documentation necessary?

CMS has directed states, where possible, to take a data-first approach and begin with the reliable information they already have, including claims, encounter data, and other available health information. The tiered framework also illustrates why implementation will not be equally straightforward in every case. For Tier 1, a qualifying diagnosis may be enough to establish medical frailty. Tier 2 is more complex. A diagnosis alone may not capture the severity of a person’s condition, the interaction of multiple comorbidities, or the extent of functional impairment.

Tier 2 is where the implementation challenge becomes clear. A diagnosis alone can’t necessarily show how severe a condition is, how it interacts with other illnesses, or whether it actually limits an individual’s ability to meet community engagement requirements.

Having the right data is part of the solution, but states also need a clinically sound way to interpret it.

For Tier 2, algorithms can help distinguish between the presence of a diagnosis and the degree to which a person’s condition actually limits their ability to comply with community engagement requirements.

Translating clinical information to a consistent, population-level determination is not new. Healthcare has been solving versions of this problem for decades. The challenge is determining how to translate that information into classifications that are reliable and practical to implement at scale. Healthcare organizations already collect information needed to identify people with significant medical needs, such as information about diagnoses, procedures, and patterns of care over time. Diagnoses submitted through claims exist within established coding, documentation, and compliance processes that make that information accountable and traceable. Combined with appropriate information about functional limitations, these existing sources can provide a fuller view of a person's health without unnecessarily duplicating relevant information already on record.

That fuller view is especially important when a single diagnosis does not tell the whole story. A person may be managing several chronic illnesses whose combined severity tells a very different story than any individual diagnosis viewed alone.

Functional limitations add another dimension. And because health changes over time, a point-in-time assessment may not provide the complete picture.

CMS’ data-first approach recognizes that additional physician or patient declarations should not automatically become another step in the process. When existing information is sufficient to support a determination, states should be able to use it. Additional documentation should be reserved for circumstances where the available information doesn’t provide enough clinical context.

CMS has provided direction on the inputs. States now need a trustworthy way to interpret them.

Identifying medical fraility accurately

While no single, existing methodology should simply be repurposed for medical frailty, we clearly already know what makes a clinical classification useful and trustworthy experience with population-based methodologies. Standardized algorithms offer important lessons about what works and what creates unintended complexity.

First, the methodology should be clinically grounded. It should account for the interaction and severity of conditions, not just a diagnosis on a list.

Second, it should be transparent and explainable. Data-informed methodologies used to support consequential decisions cannot become black boxes. State officials and healthcare organizations should be able to understand the information considered and how thresholds are applied.

Third, it should produce consistent, auditable results. Similar clinical circumstances should not produce dramatically different determinations because of where or how an assessment occurs.

Finally, it should account for change rather than locking someone into a determination based on a single moment in time. Chronic illnesses progress, new conditions emerge, and patients' circumstances evolve. The algorithm should also change as new diagnoses and codes are updated.

Classification is more than an administrative exercise. In this case, it can affect whether individuals with significant medical needs are appropriately identified for an exemption. It can also determine whether states create an efficient, scalable process or another layer of administrative work.

CMS has given states a useful starting point. The test now is whether they can translate that framework into a process that identifies medical frailty accurately, consistently, and transparently while reserving additional documentation for cases where existing information truly is not enough.

Getting that right would offer a model for a challenge healthcare will face again: translating complex clinical realities into population-level policy without losing the nuance that makes those distinctions meaningful.

Garri Garrison is president of Solventum Health Information Systems.