On the impact of predictive analytics-driven disease management interventions

Area of Science:

• Health Services Research
• Health Informatics
• Predictive Analytics in Healthcare

Background:

• Effective disease management (DM) is crucial for chronic conditions like chronic heart failure (CHF).
• Identifying high-risk patients for targeted interventions can optimize healthcare resource allocation.
• Traditional DM programs may not efficiently identify individuals most likely to benefit from outreach.

Purpose of the Study:

• To compare the effectiveness of a predictive algorithm-driven DM outreach program versus a standard DM program.
• To evaluate the impact on healthcare spending and utilization for Medicare Advantage members with CHF.
• To assess the predictive algorithm's ability to identify patients at high likelihood of hospitalization (LOH).

Main Methods:

• Propensity score matching was used for Medicare Advantage members with CHF.
• Claims data from 2013-2018 from a large commercial health insurer were analyzed.
• A predictive algorithm identified high-LOH patients for targeted DM outreach, compared to a standard outreach group.

Main Results:

• The predictive algorithm-driven (high-LOH) group showed a lower probability of hospitalization and emergency department (ED) visits.
• High-LOH intervention members experienced significantly lower total outpatient spending ($1517; P < .001).
• Predictive outreach was associated with reduced inpatient and ED utilization compared to traditional care.

Conclusions:

• A prediction-driven DM outreach program is effective for CHF patients.
• This approach significantly reduced medical spending in the year following the intervention.
• Predictive analytics enhance DM outreach efficiency and cost-effectiveness.