Priority setting in medical technology and medical practice assessment

Charles E. Phelps, Stephen T. Parente

Research output: Contribution to journalArticlepeer-review

77 Scopus citations


This study seeks to assist in setting priorities for assessing medical practices and technologies when assessment resources are scarce. It develops an objective index of expected gain from technology assessment, using modified DRG- level data on hospitalizations in NY State. The index uses standard economic concepts to combine measures of resource use, the coefficient of variation in use rates across regions, and the rate at which the incremental value of a medical intervention changes as its rate of use changes, providing a dollar-val- ued welfare loss from variations. For the entire US in 1987, the highest index occurred for coronary artery bypass graft ($0.95 billion per year), but most of the high-index interventions were nonsurgical, including hospitalizations for psychosis ($0.74 billion per year), cardiac catheterization ($0.62 billion per year), chronic obstructive lung disease ($0.55 billion per year), angina pectoris ($0.46 billion per year), adult gastroenteritis ($0.38 billion per year), adult pneumonia ($0.32 billion per year) and medical back problems ($0.28 billion per year). The top 25 interventions create an annual welfare loss of exceeding $7 billion. The present value of convincingly assessing the correct way to use these interventions sums many years of annual gains from eliminating these welfare losses. The gains from eliminating unexplained variation in medical practices appear greatly larger than costs of necessary studies.

Original languageEnglish (US)
Pages (from-to)703-723
Number of pages21
JournalMedical care
Issue number8
StatePublished - Aug 1990


  • Health economics
  • Health services research
  • Information
  • Technology assessment
  • Variations


Dive into the research topics of 'Priority setting in medical technology and medical practice assessment'. Together they form a unique fingerprint.

Cite this