A decision-theoretic approach to identifying future high-cost patients

Kenneth Pietz, Margaret M. Byrne, Laura A. Petersen

Research output: Contribution to journalArticlepeer-review

4 Scopus citations


OBJECTIVE: The objective of this study was to develop and evaluate a method of allocating funding for very-high-cost (VHC) patients among hospitals. RESEARCH DESIGN: Diagnostic cost groups (DCGs) were used for risk adjustment. The patient population consisted of 253,013 veterans who used Department of Veterans Affairs (VA) medical care services in fiscal year (FY) 2003 (October 1, 2002-September 30, 2003) in a network of 8 VA hospitals. We defined VHC as greater than $75,000 (0.81%). The upper fifth percentile was also used for comparison. METHODS: A Bayesian decision rule for classifying patients as VHC/not VHC using DCGs was developed and evaluated. The method uses FY 2003 DCGs to allocate VHC funds for FY 2004. We also used FY 2002 DCGs to allocate VHC funds for FY 2003 for comparison. The resulting allocation was compared with using the allocation of VHC patients among the hospitals in the previous year. RESULTS: The decision rule identified DCG 17 as the optimal cutoff for identifying VHC patients for the next year. The previous year's allocation came closest to the actual distribution of VHC patients. CONCLUSIONS: The decision-theoretic approach may provide insight into the economic consequences of classifying a patient as VHC or not VHC. More research is needed into methods of identifying future VHC patients so that capitation plans can fairly reimburse healthcare systems for appropriately treating these patients.

Original languageEnglish (US)
Pages (from-to)842-849
Number of pages8
JournalMedical Care
Issue number9
StatePublished - Sep 1 2006


  • Cost
  • Decision theory
  • Diagnostic cost groups
  • Logistic regression
  • Risk adjustment

ASJC Scopus subject areas

  • Nursing(all)
  • Public Health, Environmental and Occupational Health
  • Health(social science)
  • Health Professions(all)


Dive into the research topics of 'A decision-theoretic approach to identifying future high-cost patients'. Together they form a unique fingerprint.

Cite this