Forecasting risk

May 1, 2011

Risk management will become increasingly
critical in accountable care models.

Clayton Ramsey

Predictive modeling in healthcare represents a cataclysmic shift from more traditional methods of retrospective data analysis.

Previously, case managers examined the records of patients who were hospitalized in the previous year for conditions such as diabetes. Predictive modeling provides the case manager with a window into the future. With predictive modeling, she can review vast volumes of client data to identify patterns in care and interventions to prevent negative outcomes such as re-hospitalizations. Predictive modeling provides her with the rules or algorithms to apply against data to predict outcomes.

We have seen some health systems take the lead using this technology. Novant Health,
a not-for-profit integrated group of 13 hospitals in North Carolina, Virginia, South Carolina and Georgia, was searching for a solution that would deliver real-time alerts on hospitalized patients. Doing so would prevent readmissions and avoid transfers to a limited number of ICU beds. Elsevier/ MEDai responded with Pinpoint Review, predictive modeling software that has allowed Novant to deliver predictions on individual patients along with factors that drive risk. If, for example, a patient is admitted to the hospital at low risk for readmission and becomes a high risk for readmission just three days later, the software can identify predictive risks along with immediate provider action steps.

Payers face a similar challenge. Long-term users of predictive modeling used to proactively identify high-risk, high-cost members — payers increasingly have grown interested in isolating individual members at high risk for hospitalization and emergency department visits. Using this information, they can initiate proactive disease management to avoid adverse outcomes.

That's what occurred at Sentara Health Plan, which used Elsevier / MEDai's Risk Navigator Clinical to identify 2,000 high-risk members for placement in a care management program and save the plan some $6 million. The per-member, per-year costs of members who participated in care management declined from $20,000 to $15,000 over three years, while the per-member, per-year costs of 2,000 others unable to take advantage of care management increased over three years. Using Pinpoint Quality, Sentara saved more than $7 million on common clinical conditions, including congestive heart failure, pneumonia, stroke and cholecystectomy.

In 2006, Geisinger Health Plan began using Risk Navigator Clinical to identify high-risk members for placement in a medical home. Among medical home participants, Geisinger saw across-the-board increases in quality metrics, including risk assessment (100 percent), plan of care (94 percent), follow-up encounters (84 percent) and coronary artery disease (45 percent).

Efficiency also improved, with inpatient per member per month decreasing 15 percent compared to 10 percent increases for patients who did not participate in the medical home. Other gains included total allowed per member per month, pre-Rx allowed per member per month and total admissions per 1,000 members.

The message is clear: Providers and payers can generate savings and improve care quality by zeroing in on individual patients and members and providing long-term interventions and support, whether in the form of disease and case management or participation in evolving models of care such as medical home.

The best is yet to come. While predictive modeling vendors previously focused on forecasting patient cost and outcomes, a growing number have begun to zero in on disease progression — how far a patient or member's disease has progressed and how providers and payers can work together to slow or halt the disease.

Predictive modeling will also become increasingly critical in accountable care models, where providers and payers assume the risk of managing patient population's care and share any savings incurred. To determine where risk resides, providers and payers will need to identify existing and future high-risk patients and members, along with the most efficient and effective caregivers and hospitals.

Providers and payers can then identify how well risk is being managed — whether, for example, patients are facing re-hospitalization or tend to stay out of the hospital. By sharing this information with other members of the healthcare team, providers and payers can come to a shared understanding of risk and create the most effective, efficient care path for the patient.

Clayton Ramsey is COO,
Elsevier / MEDai.
For more information on
Elsevier / MEDai solutions:

Sponsored Recommendations

Patient Engagement and ML/AI – Modern Interoperability as an enabler for value based care

Discover how modern interoperability empowers patient engagement and leverages ML/AI for better outcomes in value-based care. Join us on June 18th to learn how seamless data integration...

New Research: The State of Healthcare Cloud Security and Compliance Posture

Compliance & Security Debt Awareness Could Have Prevented Change Healthcare & Ascension Healthcare Breaches

Telehealth: Moving Forward Into the Future

Register now to explore two insightful sessions that delve into the transformative potential of telehealth and virtual care management solutions.

Telehealth: Moving Forward Into the Future

Register now to explore two insightful sessions that delve into the transformative potential of telehealth and virtual care management solutions.