Jennifer A Whitty1, Ana Sofia Oliveira Gonçalves2. 1. Health Economics Group, Norwich Medical School, University of East Anglia, Norwich Research Park, Norwich, Norfolk, NR4 7JT, UK. Jennifer.whitty@uea.ac.uk. 2. Health Economics Group, Norwich Medical School, University of East Anglia, Norwich Research Park, Norwich, Norfolk, NR4 7JT, UK.
Abstract
OBJECTIVE: The aim of this study was to compare the acceptability, validity and concordance of discrete choice experiment (DCE) and best-worst scaling (BWS) stated preference approaches in health. METHODS: A systematic search of EMBASE, Medline, AMED, PubMed, CINAHL, Cochrane Library and EconLit databases was undertaken in October to December 2016 without date restriction. Studies were included if they were published in English, presented empirical data related to the administration or findings of traditional format DCE and object-, profile- or multiprofile-case BWS, and were related to health. Study quality was assessed using the PREFS checklist. RESULTS: Fourteen articles describing 12 studies were included, comparing DCE with profile-case BWS (9 studies), DCE and multiprofile-case BWS (1 study), and profile- and multiprofile-case BWS (2 studies). Although limited and inconsistent, the balance of evidence suggests that preferences derived from DCE and profile-case BWS may not be concordant, regardless of the decision context. Preferences estimated from DCE and multiprofile-case BWS may be concordant (single study). Profile- and multiprofile-case BWS appear more statistically efficient than DCE, but no evidence is available to suggest they have a greater response efficiency. Little evidence suggests superior validity for one format over another. Participant acceptability may favour DCE, which had a lower self-reported task difficulty and was preferred over profile-case BWS in a priority setting but not necessarily in other decision contexts. CONCLUSION: DCE and profile-case BWS may be of equal validity but give different preference estimates regardless of the health context; thus, they may be measuring different constructs. Therefore, choice between methods is likely to be based on normative considerations related to coherence with theoretical frameworks and on pragmatic considerations related to ease of data collection.
OBJECTIVE: The aim of this study was to compare the acceptability, validity and concordance of discrete choice experiment (DCE) and best-worst scaling (BWS) stated preference approaches in health. METHODS: A systematic search of EMBASE, Medline, AMED, PubMed, CINAHL, Cochrane Library and EconLit databases was undertaken in October to December 2016 without date restriction. Studies were included if they were published in English, presented empirical data related to the administration or findings of traditional format DCE and object-, profile- or multiprofile-case BWS, and were related to health. Study quality was assessed using the PREFS checklist. RESULTS: Fourteen articles describing 12 studies were included, comparing DCE with profile-case BWS (9 studies), DCE and multiprofile-case BWS (1 study), and profile- and multiprofile-case BWS (2 studies). Although limited and inconsistent, the balance of evidence suggests that preferences derived from DCE and profile-case BWS may not be concordant, regardless of the decision context. Preferences estimated from DCE and multiprofile-case BWS may be concordant (single study). Profile- and multiprofile-case BWS appear more statistically efficient than DCE, but no evidence is available to suggest they have a greater response efficiency. Little evidence suggests superior validity for one format over another. Participant acceptability may favour DCE, which had a lower self-reported task difficulty and was preferred over profile-case BWS in a priority setting but not necessarily in other decision contexts. CONCLUSION:DCE and profile-case BWS may be of equal validity but give different preference estimates regardless of the health context; thus, they may be measuring different constructs. Therefore, choice between methods is likely to be based on normative considerations related to coherence with theoretical frameworks and on pragmatic considerations related to ease of data collection.
Authors: Joris D van Dijk; Catharina G M Groothuis-Oudshoorn; Deborah A Marshall; Maarten J IJzerman Journal: Value Health Date: 2016-03-23 Impact factor: 5.725
Authors: Jennifer A Whitty; Julie Ratcliffe; Elizabeth Kendall; Paul Burton; Andrew Wilson; Peter Littlejohns; Paul Harris; Rachael Krinks; Paul A Scuffham Journal: BMJ Open Date: 2015-10-15 Impact factor: 2.692
Authors: Kei Long Cheung; Ben F M Wijnen; Ilene L Hollin; Ellen M Janssen; John F Bridges; Silvia M A A Evers; Mickael Hiligsmann Journal: Pharmacoeconomics Date: 2016-12 Impact factor: 4.981
Authors: Victoria Fischman; Eve Wittenberg; Sungjin A Song; Molly N Huston; Ramon A Franco; Phillip C Song; Matthew R Naunheim Journal: OTO Open Date: 2021-03-11
Authors: Elizabeth A Asiago-Reddy; John McPeak; Riccardo Scarpa; Amy Braksmajer; Nicola Ruszkowski; James McMahon; Andrew S London Journal: PLoS One Date: 2022-01-26 Impact factor: 3.240
Authors: Tatenda T Yemeke; Elizabeth E Kiracho; Aloysius Mutebi; Rebecca R Apolot; Anthony Ssebagereka; Daniel R Evans; Sachiko Ozawa Journal: PLoS One Date: 2020-07-30 Impact factor: 3.240
Authors: John M Humphrey; Violet Naanyu; Katherine R MacDonald; Kara Wools-Kaloustian; Gregory D Zimet Journal: PLoS One Date: 2019-10-30 Impact factor: 3.752