Using Predictive Modeling to Evaluate Disease Management Financial Impact

1
Using Predictive Modeling to Evaluate Disease
Management Financial Impact
Terry Whitlock, MBA Kenton Johnston, MPH, MS
June 28, 2005
2
What is Disease Management at BCBST?

• CAD, CHF, COPD, Asthma and Diabetes
• Opt out program
• External 3rd party vendor relationship
• Self Insured Groups that participate
• Fully Insured Groups for CAD and CHF diseases
• FEP Members

3
What does BCBST look for in DM?

• Outcomes Improvement
• Improve health of population
• Enhance patient satisfaction care experience
• Enhance physician satisfaction delivery
• Improve work force productivity (Indirect Costs)
• Reduce total health care costs (Direct Costs)

4
Impact of Chronic Illness on costs

• 78 of Total Medical Costs are attributed to
  treatment of chronic diseases
• Half of Americans report having a chronic disease
  and these patients cost 3.5 times as much to
  those without chronic diseases

5
How do we measure DM impact?

• Measuring what would have happened vs. what
  actually happened is subjective to say the least
• Methods used to assess DM impact
• Randomized Control Group studies
• Population Based Pre-Post Methodology
  (Quasi-experimental design)
• Predictive Modeling

6
Why Predictive Modeling?

• Predictive Modeling refers to the process of
  finding rules (models) for predicting an event
  from prior patterns within a given time frame and
  applying these rules to current data in order to
  predict a future event
• Helps us account for Regression to the mean
• High cost cases in a base year becoming lower
  cost in subsequent years as well as normal cases
  in a base year becoming outlier cases in
  subsequent years

7
Population based Pre-Post Methodology
8
Predictive Modeling
9
Predictive Modeling w/Adjustments
10
Conclusions of DM Evaluations

• A statistically valid predictive model should be
  incorporated in lieu of randomized control group.
• Adjustments (inflation factors, inaccuracy of
  predictive models, etc.) should be made to model.
• Predictive scores and/or illness burdens should
  be incorporated into initial disease management
  patient identification and stratification process
  to improve impactibility.