Sequential decision making with continuous disease states and measurements: II. application to diastolic blood pressure

Research output: Contribution to journalArticle

7 Scopus citations

Abstract

The model and strategy for sequential decision making using normally distributed measure ments proposed in a companion paper are applied to the problem of diagnosing diastolic hypertension. The assumptions of the model are discussed and justified clinically. Methods for assigning values to the model’s parameters are explained and illustrated in the context of a hypothetical “generic” patient. Although current national recommendations and the sequential strategy both lead to an average of 1.89 measurements per patient prior to diagnosis, the sequential strategy applies a sequence of four or more measurements to 12% of patients. Fewer than 1% of patients would require ten or more measurements under this strategy. The sequential strategy leads to fewer patients’ receiving unnecessary treatment and substantially higher expected utility for the patient. The role of multiple blood pressure determinations per visit is explored in the absence of appropriate estimates. Even under “best-case” assumptions, however, it is shown that obtaining more than one observation per visit is called for only in about 15% of visits. While the exact role of multiple determinations cannot be specified from existing data, it is likely to be much more limited than current recommendations suggest.

Original languageEnglish (US)
Pages (from-to)256-265
Number of pages10
JournalMedical Decision Making
Volume10
Issue number4
DOIs
StatePublished - Dec 1990
Externally publishedYes

Keywords

  • Key words: Bayes’ theorem
  • conjugate-normal- linear model
  • decision analysis. (Med Decis Making 1990;10:256-265)
  • hypertension

ASJC Scopus subject areas

  • Health Policy

Fingerprint Dive into the research topics of 'Sequential decision making with continuous disease states and measurements: II. application to diastolic blood pressure'. Together they form a unique fingerprint.

  • Cite this