CDPS stands for the Chronic Illness and Disability Payment System, a statistical model used to predict how much healthcare a Medicaid enrollee will need based on their diagnoses. It was originally developed at UC San Diego for state Medicaid programs that pay managed care plans a fixed monthly amount per patient. The core problem it solves is straightforward: a flat payment per person underpays plans that enroll sicker patients and overpays plans with healthier ones. CDPS adjusts those payments so they reflect how sick each person actually is.
How CDPS Calculates a Risk Score
A CDPS risk score starts with a person’s medical claims. For adults, the system looks at the last 15 months of paid claims and health plan encounters. For children under 18, it pulls from the last 24 months. Every diagnosis code on those claims gets mapped to one of 19 major disease categories: cardiovascular, psychiatric, skeletal, central nervous system, pulmonary, gastrointestinal, diabetes, skin, renal, substance abuse, cancer, developmental disability, genital, metabolic, pregnancy, eye, cerebrovascular, infectious disease, and hematological.
Within each of those 19 major categories, diagnoses are sorted into subcategories that reflect severity. The current version of the model (7.2, released in 2024) has 56 subcategories total. A key rule called the “hierarchy” applies here: if a person has multiple diagnoses in the same major category, only the most severe one counts. This prevents double-counting and makes it harder to inflate scores by piling on related diagnoses.
Once the system has identified a person’s highest-severity subcategory in each relevant disease group, it assigns a numerical weight to each one. It also assigns a weight based on the person’s age and sex, since healthcare costs vary significantly across demographic groups (the model uses 11 age/sex categories). The final risk score is simply the sum of all those weights. A higher score means the model predicts that person will cost more to care for, and the health plan receives a larger monthly payment accordingly.
The Pharmacy Add-On: CDPS+Rx
Most states now use an enhanced version called CDPS+Rx, which layers prescription drug data on top of diagnostic data. In addition to the 56 diagnosis-based categories, this version maps pharmacy claims (using national drug codes) into 45 separate medication-based risk categories. The medication weights are then added into the same risk score calculation. This matters because prescriptions often reveal conditions that don’t appear consistently in diagnosis codes, giving the model a more complete picture of a person’s health.
Where CDPS Is Used
CDPS is the dominant risk adjustment tool in Medicaid managed care. More than 30 state Medicaid programs currently use it to set payments to health plans, and nearly all of them have adopted the CDPS+Rx version. Early adopters like Colorado (1997), Oregon (1998), and Utah (1998) have been using some form of the system for over 25 years. Larger states including Texas, Florida, Ohio, Pennsylvania, and Michigan all rely on it. More recent adopters include New Mexico and North Carolina, both of which implemented CDPS+Rx in 2021. Puerto Rico adopted it in 2018.
The system was originally built for Medicaid, but a modified version called CDPS-Medicare was later developed for use in adjusting capitated Medicare payments to health plans as well.
How CDPS Differs From HCC Models
If you’ve encountered risk adjustment in Medicare Advantage, you’re probably familiar with the Hierarchical Condition Category (HCC) model. Both CDPS and HCC use diagnosis codes to predict costs and apply hierarchies within disease groups, but they were built for different populations. HCC was designed around the Medicare population, which skews older and has different cost patterns. CDPS was calibrated specifically for the Medicaid population, which includes younger adults, children, pregnant women, and people with disabilities. The disease categories, the severity tiers within them, and the dollar weights all reflect Medicaid spending patterns rather than Medicare ones.
CDPS also has a tighter structure by design. Its hierarchy system and the practice of collapsing fine-grained diagnosis groups into broader categories are explicitly intended to reduce “upcoding,” where providers or plans submit diagnosis codes in ways that inflate risk scores and payments.
How the Model Stays Current
CDPS is maintained by researchers at UC San Diego and undergoes periodic revisions. The most recent major update, version 7.2, was released in 2024 after a clinical review process in which medical experts evaluated every diagnosis assignment across all 19 major categories. Experts flagged diagnoses they felt were missing, misplaced, or incorrectly weighted, and the research team ran new statistical analyses to incorporate those recommendations.
The 2024 revision added four new subcategories and reassigned a significant number of diagnosis codes. Out of roughly 21,200 diagnoses in the model, about 92% stayed in the same category, but notable shifts occurred in cancer (503 codes reassigned), cerebrovascular disease (240), eye conditions (225), and central nervous system disorders (207). Hundreds of new diagnoses were also added, particularly in cardiovascular, genital, and pregnancy-related categories. These updates keep the model aligned with current medical coding practices and evolving treatment patterns.
Why It Matters for Patients
You’ll never see your own CDPS score on a medical bill or insurance statement, but it shapes the financial incentives behind your care. When risk adjustment works well, health plans have less reason to avoid enrolling people with serious chronic conditions, because the payments they receive reflect the actual cost of caring for those members. Without a system like CDPS, plans would profit most by attracting the healthiest enrollees and discouraging sicker people from joining. Risk-adjusted payments don’t guarantee better care, but they remove one of the strongest financial incentives to avoid providing it.