An Exploratory Study on Using Principal-Component Analysis and Confirmatory Factor Analysis to Identify Bolt-On Dimensions: The EQ-5D Case Study

Aureliano Paolo Finch; John Edward Brazier; Clara Mukuria; Jakob Bue Bjorner

doi:10.1016/j.jval.2017.06.002

An Exploratory Study on Using Principal-Component Analysis and Confirmatory Factor Analysis to Identify Bolt-On Dimensions: The EQ-5D Case Study

Value Health. 2017 Dec;20(10):1362-1375. doi: 10.1016/j.jval.2017.06.002. Epub 2017 Jul 14.

Authors

Aureliano Paolo Finch¹, John Edward Brazier², Clara Mukuria², Jakob Bue Bjorner³

Affiliations

¹ Health Economics and Decision Science, School of Health and Related Research, University of Sheffield, Sheffield, UK. Electronic address: APFinch1@sheffield.ac.uk.
² Health Economics and Decision Science, School of Health and Related Research, University of Sheffield, Sheffield, UK.
³ Health Economics and Decision Science, School of Health and Related Research, University of Sheffield, Sheffield, UK; Optum Insight, Eden Prairie, Minnesota, USA; Faculty of Health and Medical Science, University of Copenhagen, Copenhagen, Denmark.

PMID: 29241896
DOI: 10.1016/j.jval.2017.06.002

Abstract

Objective: Generic preference-based measures such as the EuroQol five-dimensional questionnaire (EQ-5D) are used in economic evaluation, but may not be appropriate for all conditions. When this happens, a possible solution is adding bolt-ons to expand their descriptive systems. Using review-based methods, studies published to date claimed the relevance of bolt-ons in the presence of poor psychometric results. This approach does not identify the specific dimensions missing from the Generic preference-based measure core descriptive system, and is inappropriate for identifying dimensions that might improve the measure generically. This study explores the use of principal-component analysis (PCA) and confirmatory factor analysis (CFA) for bolt-on identification in the EQ-5D.

Methods: Data were drawn from the international Multi-Instrument Comparison study, which is an online survey on health and well-being measures in five countries. Analysis was based on a pool of 92 items from nine instruments. Initial content analysis provided a theoretical framework for PCA results interpretation and CFA model development. PCA was used to investigate the underlining dimensional structure and whether EQ-5D items were represented in the identified constructs. CFA was used to confirm the structure. CFA was cross-validated in random halves of the sample.

Results: PCA suggested a nine-component solution, which was confirmed by CFA. This included psychological symptoms, physical functioning, and pain, which were covered by the EQ-5D, and satisfaction, speech/cognition,relationships, hearing, vision, and energy/sleep which were not. These latter factors may represent relevant candidate bolt-ons.

Conclusions: PCA and CFA appear useful methods for identifying potential bolt-ons dimensions for an instrument such as the EQ-5D.

Keywords: EQ-5D; bolt-on; health- related quality of life; preference-based measures.

MeSH terms

Factor Analysis, Statistical
Health Status*
Humans
Patient Preference / psychology*
Principal Component Analysis
Psychometrics
Quality of Life*
Surveys and Questionnaires*