Model based economic evaluations of diagnostic point of care tests were rarely fit for purpose

Kathryn Breheny, Andrew Sutton, Jonathan Deeks

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)
284 Downloads (Pure)

Abstract

Objective: Linked evidence models are recommended to predict health benefits and cost-effectiveness of diagnostic tests. We considered how published models accounted for changes in patient pathways that occur with point of care tests (POCTs), and their impact of on patient health and costs.


Study Design and Setting: Model based evaluations of diagnostic POCTs published from 2004-2017 were identified from searching six databases. For each model we assessed the outcomes considered, and whether reduced time to diagnosis and increased access to testing affected patient health and costs.


Results: Seventy-four model based evaluations were included: 95% incorporated evidence on test accuracy, but 34% only assessed intermediate outcomes such as rates of correct diagnosis. Of 54 models where POCTs reduced testing time, 39% addressed the economic and 37% the health benefits of faster diagnosis. No model considered differences in access to tests.


Conclusions: Many models fail to capture the effects of POCTs in increasing access, advancing speed of diagnosis and treatment, reducing anxiety and the associated costs. Many only consider the impact of testing from changes in accuracy. Ensuring models incorporate changes in patient pathways from faster and more accessible testing will lead to economic evaluations that better reflect the impact of POCTs.
Original languageEnglish
JournalJournal of Clinical Epidemiology
Early online date10 Nov 2018
DOIs
Publication statusE-pub ahead of print - 10 Nov 2018

Keywords

  • Diagnostic test
  • post of care test
  • decision model
  • clinical pathway
  • health economic model

Fingerprint

Dive into the research topics of 'Model based economic evaluations of diagnostic point of care tests were rarely fit for purpose'. Together they form a unique fingerprint.

Cite this