Research Objective: This paper evaluates the ability of two methods to identify a small group of people who are expected to have highest future costs.
Study Design: We study 2.7 million people enrolled during 1997 and 1998 in various commercially insured, nationally-disbursed plans. We use the Diagnostic Cost Group (DCG) risk adjustment model and diagnoses from both physicians and hospitals in 1997 to characterize each individual°¦s health status and to predict the cost of their health care in 1998. We also use actual expenditures in 1997 (Prior Cost) to identify the same number of people with the highest expected costs. We compare the ability of these two models to distinguish high and low expected future cost individuals.
Results: The medical costs are highly skewed, with the most expensive 1% of the population consumes about 31% of costs in any given year. Individuals who are more likely to be expensive next year often have multiple serious coexisting conditions, and 90% of them have 5 or more distinct categories of medical problems. The DCG and prior cost models overlap in identifying high cost users. However, the DCG model is much better at identifying subsets of people with specific medical problems whose costs will be highest.
Conclusions: The DCG model better identifies people with very high future costs.
Learning Objectives: As a result of attending this session, participants will know how to use diagnostic claims information to identify individuals who are expected to have future highest medical costs so that patient care and costs can be more effectively managed
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: Medstat MarketScan database, DxCG, Inc. risk adjustment software
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.