Conjoint analysis applications in health-how are studies being designed and reported? An update on current practice in the published literature between 2005 and 2008

Deborah Marshall, John F.P. Bridges, Brett Hauber, Ruthanne Cameron, Lauren Donnalley, Ken Fyie, F. Reed Johnson

Research output: Contribution to journalReview articlepeer-review

Abstract

Despite the increased popularity of conjoint analysis in health outcomes research, little is known about what specific methods are being used for the design and reporting of these studies. This variation in method type and reporting quality sometimes makes it difficult to assess substantive findings. This review identifies and describes recent applications of conjoint analysis based on a systematic review of conjoint analysis in the health literature. We focus on significant unanswered questions for which there is neither compelling empirical evidence nor agreement among researchers.We searched multiple electronic databases to identify English-language articles of conjoint analysis applications in human health studies published since 2005 through to July 2008. Two independent reviewers completed the detailed data extraction, including descriptive information, methodological details on survey type, experimental design, survey format, attributes and levels, sample size, number of conjoint scenarios per respondent, and analysis methods. Review articles and methods studies were excluded. The detailed extraction form was piloted to identify key elements to be included in the database using a standardized taxonomy.We identified 79 conjoint analysis articles that met the inclusion criteria. The number of applied studies increased substantially over time in a broad range of clinical applications, cancer being the most frequent. Most used a discrete-choice survey format (71), with the number of attributes ranging from 3 to 16. Most surveys included 6 attributes, and 73 presented 715 scenarios to each respondent. Sample size varied substantially (minimum13, maximum1258), with most studies (38) including between 100 and 300 respondents. Cost was included as an attribute to estimate willingness to pay in approximately 40 of the articles across all years.Conjoint analysis in health has expanded to include a broad range of applications and methodological approaches. Although we found substantial variation in methods, terminology, and presentation of findings, our observations on sample size, the number of attributes, and number of scenarios presented to respondents should be helpful in guiding researchers when planning a new conjoint analysis study in health.

Original languageEnglish (US)
Pages (from-to)249-256
Number of pages8
JournalPatient
Volume3
Issue number4
DOIs
StatePublished - 2010

Keywords

  • Conjoint-analysis

ASJC Scopus subject areas

  • Nursing (miscellaneous)

Fingerprint Dive into the research topics of 'Conjoint analysis applications in health-how are studies being designed and reported? An update on current practice in the published literature between 2005 and 2008'. Together they form a unique fingerprint.

Cite this