Examining the factorial validity of the Quality of Life Scale

Abstract

Background: Quality of life (QoL) is important to assess in patient care. Researchers have previously claimed validity of the Quality of Life Scale (QOLS) across multiple samples of individuals, but close inspection of results suggest further psychometric investigation of the instrument is warranted. Therefore, the purposes of this study were to: 1) evaluate the proposed five-factor, 15-item and three-factor, 16-item QOLS; 2) if the factor structure could not be confirmed, re-assess the QOLS using exploratory factor analysis (EFA) and covariance modeling to identify a parsimonious refinement of the QOLS structure for future investigation. Methods: Participants varying in age, physical activity level, and identified medical condition(s) were recruited from clinical sites and ResearchMatch. Confirmatory factor analyses (CFA) were performed on the full sample (n = 1036) based on proposed 15- and 16-item QOLS versions. Subsequent EFA and covariance modeling was performed on a random subset of the data (n1 = 518) to identify a more parsimonious version of the QOLS. The psychometric properties of the newly proposed model were confirmed in the remaining half of participants (n2 = 518). Further examination of the scale psychometric properties was completed using invariance testing procedures across sex and health status sub-categories. Results: Neither the 15- nor 16-item QOLS CFA met model fit recommendations. Subsequent EFA and covariance modeling analyses revealed a one-factor, five-item scale that satisfied contemporary statistical and model fit standards. Follow-up CFA confirmed the revised model structure; however, invariance testing requirements across sex and injury status subgroups were not met. Conclusions: Neither the 15- nor 16-item QOLS exhibited psychometric attributes that support construct validity. Our analyses indicate a new, short-form model, might offer a more appropriate and parsimonious scale from some of the original QOLS items; however, invariance testing across sex and injury status suggested the psychometric properties still vary between sub-groups. Given the scale design concerns and the results of this study, developing a new instrument, or identifying a different, better validated instrument to assess QoL in research and practice is recommended.