• Sidebar: A Closer Look at Risk Stratification Models

    Laura Ramos Hegwer Jul 28, 2016

    A number of risk stratification models are available from public agencies and third-party vendors. Each model has a distinctive focus. Below are some of the most commonly used models.a

    Hierarchical Condition Categories. Implemented in 2004 by the Centers for Medicare & Medicaid Services for its Medicare Advantage plans, the HCC model contains 70 condition categories selected from ICD-10 diagnosis codes along with demographic data to calculate patient risk scores.

    Adjusted Clinical Groups. Developed at The Johns Hopkins University to predict morbidity, the ACG model projects the use of medical resources in inpatient and outpatient services over a specific period of time, using the presence or absence of specific diagnoses, along with age and gender, to classify patients into one of 93 categories.

    Chronic Comorbidity Counts. Based on publicly available information from the Agency for Healthcare Research and Quality’s Clinical Classification Software, the CCC model groups patients into six categories based on risk as measured by the total sum of selected comorbid conditions.

    Elder Risk Assessment. The ERA model is used to identify patients 60 years or older who are at risk for hospitalization and ED visits. The index uses age, gender, number of hospital days in the prior two years, and marital status, as well as the presence of selected medical conditions (diabetes, coronary artery disease, congestive heart failure, stroke, chronic obstructive pulmonary disease, and dementia) to assign an index score to each patient.

    Charlson Comorbidity Measure. Based on administrative data, the CCM model uses the presence or absence of 17 specific conditions to predict the risk of one-year mortality for patients with a range of comorbid illnesses.

    Minnesota Tiering. The MT model groups patients into one of five complexity tiers based on their number of major conditions. The five tiers are designated as Tier 0 (Low: 0 Conditions), Tier 1 (Basic: 1 to 3), Tier 2 (Intermediate: 4 to 6), Tier 3 (Extended: 7 to 9), and Tier 4 (Complex: 10+).

    A study published in 2013 by the American Journal of Managed Care examines these six risk stratification models. The authors found that the ACG model generally was superior at predicting hospitalizations, emergency department visits, 30-day readmissions, and high-cost users, but that the other models did a good job as well.b

    What these models have in common is the importance of age and comorbidity as risk factors, says Paul Takahashi, coauthor of the study and professor of medicine at Mayo Clinic College of Medicine, Rochester, Minn. The models differ in how easy they are to implement. As noted in the study, for example, the software used in some models requires licensing, while other software can be downloaded. Some models have algorithms that need to be programmed to be applied for local clinical use.

    Healthcare organizations can develop their own risk stratification models—but given that so many are available, using one that is ready-made can be just as effective, says Urvashi Patel, PhD, of Montefiore Health System. Her advice is to review a few models, but not to spend a lot of time doing so. She says HCC scores are a good place to start, but organizations should not stop there.

    “I don’t think that there’s any one set of tools,” Patel says. “We use both basic aggregation of scores as well as more advanced regression modeling. And we’ve found that they all work. Each one works for different programs very well. It just really depends on where you want to focus your effort.”


    Karen Wagner is a freelance healthcare writer based in Forest Lake, Ill., and a member of HFMA’s First Illinois Chapter.

    Interviewed for this article: Paul Takahashi, MD, professor of medicine, Mayo Clinic College of Medicine, Rochester, Minn.

    Urvashi Patel, PhD, senior director and chief data scientist Montefiore Health System’s care management organization, Montefiore Health System, New York City.

    Footnotes

    a. Just, E., “Understanding Risk Stratification, Comorbidities, and the Future of Healthcare,” Health Catalyst, 2016; and Haas, L.R., Takahashi, P.Y., Shah, N.D., et al., “Risk-Stratification Methods for Identifying Patients for Care Coordination,” American Journal of Managed Care, Sept. 17, 2013.

    b. American Journal of Managed Care, ibid.

    YOU MIGHT ALSO BE INTERESTED IN:

ADVERTISEMENTS