Diagnostic Cost Group and RxGroups Models Predict Total Health Care Costs Better Than Any Type of Data Alone, DxCG Study Reports

Diagnostic Cost Group and RxGroups® Models Are Most Accurate for Managing Population Health

BOSTON — January 18, 2005 — DxCG®, Inc., the leading provider of predictive modeling software for health care organizations, announced today the results of a study finding the comparative power of diagnostic and drug data for predicting future health care costs. The study, "Predicting Pharmacy Costs and Other Medical Costs Using Diagnoses and Drug Claims," concludes that combined drug and diagnostic data predicts total health care costs better than either type of data alone. The study also proved that using recent data substantially boosts model performance measures. The study — led by Yang Zhao, Ph.D., formerly a senior research associate at DxCG; Arlene Ash, Ph.D., DxCG senior scientist; and Randall Ellis, Ph.D., DxCG senior scientist — is published in Medical Care.

The study applied the combined Rx plus Diagnostic Cost Group (DCG) model to 1997 and 1998 data to predict health care costs in 1998 and 1999. The researchers compared the performance and R2 value of the combined model against using only diagnoses and then only drug claims to predict various future health care costs. The combined drug and diagnostic data predicted total health care costs by 11 percent more than either data alone. The results also found more recent data improved model performance by 20 percent for total costs and non-pharmacy costs.

"Many studies have examined the predictive performance of newer models, but none has systematically distinguished improvements as the result of more refined predictive models versus newer data," said Dr. Zhao. "Our study used this comparison and found improvements in predictive performance came from newer data rather than from more clinically refined classifications."

The research further explored the extent to which the newer data or more refined predictive models contributed to the observed improvements in predictive power. This finding proves that credible comparisons of the performance of different models require evaluation of the same data. The Diagnostic Cost Group (DCG) and RxGroups® models were found to be powerful predictors of future cost. Each model captures population disease burden and can be used to accurately summarize clinical information that can be managed to achieve better outcomes and lower costs. The RxGroups models predicted future pharmacy costs much better than the DCG models, while the DCG models predicted total costs and non-pharmacy costs more effectively than the RxGroups models.

About DxCG, Inc.
DxCG, a unit of ISO, promotes fair and efficient health care by providing software solutions to more accurately plan, budget and evaluate health care management programs. DxCG has more than 175 clients, covering over 75 million lives in the U.S. and abroad. The company's Diagnostic Cost Group (DCG) and RxGroups® predictive models are used to negotiate health-based payments, identify opportunities for disease management, profile physicians and evaluate managed care programs. Recognized by leading independent researchers as the most proven models available, DxCG's methodologies are used by the federal government to set payment rates for Medicare. For more information, visit the company's website at www.dxcg.com.

Release: Immediate

Contact:
Stephanie Martinovich (DxCG, Inc.)
617-896-5919
stephanie.martinovich@dxcg.com

Archived Press