In mental health, HCC stands for Hierarchical Condition Categories, a classification system used by the Centers for Medicare and Medicaid Services (CMS) to predict how much a patient’s care will cost. Each HCC groups related diagnoses into a single category, and when a patient has a qualifying mental health condition documented in their record, it raises their “risk score,” which directly affects how much an insurance plan gets paid to cover that person. The system matters because it shapes funding for mental health care, and many common psychiatric conditions fall outside it entirely.
How HCCs Work
The HCC system maps thousands of individual diagnostic codes into broader clusters. Rather than tracking every possible diagnosis separately, CMS groups conditions that carry similar cost implications into one category. These categories are “hierarchical” because when a patient has multiple related conditions, only the most severe one counts toward their risk score. A person diagnosed with both schizophrenia and major depression, for example, would have their score driven by the higher-weighted condition rather than stacking both.
Each HCC category assigned to a patient increases their Risk Adjustment Factor, or RAF score. A higher RAF score signals to insurers and government programs that this person is expected to need more care, which triggers higher payments to the plan covering them. The logic is straightforward: plans that enroll sicker patients should receive more money so they aren’t penalized for taking on complex cases. In practice, though, the system has significant gaps when it comes to mental health.
Which Mental Health Conditions Qualify
Only a narrow set of psychiatric diagnoses carry HCC weight in the current CMS model. The four mental health and substance use HCC categories are:
- HCC 54: Drug/Alcohol Psychosis
- HCC 55: Drug/Alcohol Dependence
- HCC 57: Schizophrenia
- HCC 58: Major Depressive, Bipolar, and Paranoid Disorders
CMS also groups these for purposes of calculating interaction effects: HCCs 57 and 58 are classified together as “Psychiatric,” while HCCs 54 and 55 fall under “Substance Abuse.” When a patient has qualifying diagnoses in both groups alongside other medical conditions, the interaction can further adjust their risk score.
The list of what doesn’t qualify is arguably more important. Generalized anxiety disorder, adjustment disorders, mild depression, and many other conditions that drive significant mental health spending are not recognized by the model. Nicotine dependence, mild alcohol use disorder, and mild cannabis use disorder are all explicitly excluded, partly because CMS considers them prone to overcoding and partly because their individual cost impact is lower. The recent shift from ICD-9 to ICD-10 coding also tightened the depression category: what now qualifies as HCC 88 in the marketplace model has been relabeled to cover only severe major depressive disorder and bipolar disorders, excluding milder presentations.
Why the Gaps Matter for Patients
The consequences of this narrow recognition are substantial. Research published in Health Affairs found that only 20 percent of individuals with a mental health or substance use diagnosis were recognized by the Marketplace risk adjustment model. The remaining 80 percent had qualifying psychiatric diagnoses in their medical records but conditions that didn’t map to any HCC category, meaning their plans received no extra payment to cover their care.
People with mental health and substance use diagnoses spent 2.6 times the sample average in 2013, making them a high-cost group. Yet the risk adjustment model undercompensated plans for these individuals by 16 percent overall. For the unrecognized group (those with mental health diagnoses that don’t trigger an HCC), the underpayment was 21 percent. Specific conditions fared even worse: adjustment disorders were undercompensated by 31 percent, anxiety disorders by 24 percent, and mood disorders by 22 percent.
This creates a financial incentive problem. When insurers lose money on patients with common psychiatric conditions, they have less motivation to build robust mental health networks or make those services easy to access. The underpayment doesn’t mean a patient gets denied care outright, but it can quietly shape which providers are in-network, how many therapy appointments are readily available, and how aggressively a plan invests in behavioral health programs.
Documentation Requirements
For the mental health conditions that do qualify, proper documentation is essential. A diagnosis only counts toward a patient’s HCC if it appears in the medical record with adequate clinical support each year. CMS does not carry forward diagnoses from previous years, so a patient with schizophrenia needs that condition documented at every annual visit cycle to maintain their risk score.
The standard framework for sufficient documentation is known by the acronym MEAT. Each qualifying diagnosis should show evidence that the provider did at least one of the following during the encounter: monitored the condition’s status, evaluated it with testing or clinical assessment, assessed it by reviewing symptoms or treatment response, or treated it with medication, therapy, or another intervention. Simply listing a diagnosis in the problem list without addressing it in the visit note is not enough. If an auditor reviews the chart and finds no MEAT-based support, the HCC can be removed and the plan’s payment clawed back.
This is particularly relevant in mental health because psychiatric conditions are often managed across multiple providers. A psychiatrist handling medication and a therapist providing counseling may each document pieces of the picture, but the HCC-qualifying diagnosis needs to be clearly supported in at least one clinical encounter per reporting period.
HCC Limitations for Mental Health Prediction
Researchers have found that the HCC system is a relatively blunt tool for predicting mental health spending. A study in PMC compared HCCs against a more granular classification system called Clinical Classifications Software (CCS) categories and found that CCS categories were more predictive of mental health and substance use spending than HCCs alone. This makes sense: collapsing the full spectrum of psychiatric diagnoses into just four categories inevitably loses information about which patients will need intensive services.
The same research explored whether machine learning approaches could improve mental health risk adjustment, and found promising results. The takeaway for the broader system is that HCCs were originally designed to predict total medical spending across all conditions, not to capture the nuances of behavioral health. Mental health spending patterns look different from those of chronic physical conditions. A person with poorly controlled diabetes, for instance, generates predictable recurring costs, while someone with a severe depressive episode may have a spike in spending followed by periods of lower utilization. The current model struggles with that variability.
CMS has acknowledged some of these shortcomings and periodically updates the HCC model, most recently with the V28 revision for Medicare Advantage. But the core architecture, where only the most severe psychiatric diagnoses receive risk adjustment weight, remains largely intact. For providers, coders, and administrators working in mental health, this means accurate capture of qualifying diagnoses is one of the few levers available to ensure appropriate reimbursement for the populations they serve.