Risk Stratification

Risk Stratification

Using risk stratification techniques is critical to identifying and ensuring that most needed and appropriate care is provided by the right provider, at the right time, and in the most appropriate setting.

As new members are enrolled in the program and in case of paucity of claims data, CHS has used an alternative approach for risk stratification, the Health Risk Screening (HRS). Responses to the HRS are used to identify and segment members into different risk groups, allowing subsequent care coordination interventions for the new enrollees. Furthermore, these responses are used to properly assign and prioritize care coordination activities.

Once an average of at least two years of historical claims data exists for the population, the stratification approach evolves to incorporate the Johns Hopkins ACG® Predictive Model. ACG is one of the most widely used, population-based, case-mix/risk adjustment methodologies.

Another approach that has been used by CHS for large Medicaid populations in the past is the identification of members with co-occurring conditions, based on empirical and experiential evidence that this population is the most unstable and costly subset of non-aged populations. We identified this population by a combination of factors analyzed internally:

  • Behavioral disease state
  • Chronic medical condition of:
    • Diabetes
    • Digestive
    • Renal
    • Central nervous system
    • Cardiovascular
  • Further stratification of this population may occur based on:
    • Atypical prescription drug utilization
    • Medication adherence ratio
    • Substance abuse disorder

ACG Risk Scores for the population are only one set of information. To achieve a comprehensive 360° view of the member, care coordinators also monitor these members over time by reviewing claims, immunization registry data, daily admission and discharge data from acute care facility EHRs, pharmacy imported data, EMRs, or hard copy patient records, and speaking with the member or caregiver and treatment team. All this data is accessible real-time in our proprietary software application (Consensus™) and is used to update the member’s risk stratification level. Aside from the ACG risk stratification, Consensus system triggering of indicators based upon utilization and/or diagnostic characteristics can be easily implemented to monitor claims experience for events of concern and to implement targeted quality campaigns. Example events include:

  • High cost members (top 10% of population in manageable conditions)
  • Emergency room utilization above a statistical threshold (mean utilization plus two standard deviations)
  • Re-admissions to the inpatient hospital setting within 15 and 30 days
  • Initial diagnosis of target conditions (e.g., metabolic syndrome, HIV/AIDS, renal failure, and behavioral health conditions)
  • Utilization of high cost or specialty medications (e.g., Remicade, Enbrel, Synagis, Olysio, Sovaldi)
  • Medication adherence ratio of less than 0.80 with a diagnosis of hypertension, diabetes, high cholesterol