Consistency Between Three Different Ways of Administering the Short Form 6 Dimension Version 2

• Published in: Value in health Vol. 22; no. 7; pp. 837 - 842
• Main Authors: Poder, Thomas G., Fauteux, Vickie, He, Jie, Brazier, John E.
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.07.2019; Elsevier Science Ltd
• Subjects: Activities of Daily Living; Adult; Bias; consistency; Expected utility; Female; Health Status; Health Status Indicators; Health surveys; Humans; Internal Medicine; Male; Mental Health; Middle Aged; Public Health; Quality of Life; Quality-Adjusted Life Years; Quebec; Questionnaires; Reproducibility of Results; SF-6Dv2; Surveys and Questionnaires; Values; Variants; Young Adult; Quebec; SF-6Dv2; consistency
• ISSN: 1098-3015; 1524-4733; 1524-4733
• DOI: 10.1016/j.jval.2018.12.012

The Short Form 6 Dimension (SF-6D) is a multi-attribute utility instrument derived from the Short-Form 36 Health Survey Version 2 (SF-36v2) quality of life questionnaire and is used to calculate quality-adjusted life years (QALYs) on a scale 0 to 1. The SF-6Dv2 is a new version of the SF-6D. The aim of this study was to assess the consistency of respondents’ answers to 3 different methods to administer this new version. SF-6Dv2 utility values were generated from the SF-36v2 using the following: (1) full questionnaire with 36 items (SF-6Dv2SF-36); (2) subset questionnaire with 10 items (SF-6Dv2ind-10); (3) SF-6Dv2 administered as an independent instrument (rephrased questionnaire with only 6 items [SF-6Dv2ind-6]). The order of the 3 instruments was randomly allocated between respondents. A total of 782 respondents from Quebec, Canada, were interviewed, out of whom 697 fully completed the survey. Very few deviations in respondents’ answers were observed between the 3 instruments, with mean weighted kappa of 0.79 (range 0.61-0.91) and mean global consistency index of 70% (range 54-83). Maximal difference in utility values generated was found between SF-6Dv2ind-10 and SF-6Dv2ind-6 (mean difference 0.016, P < .01), whereas minimal difference was found between SF-6Dv2SF-36 and SF-6Dv2ind-10 (0.002, P = .38). No ceiling effect was observed. The SF-6Dv2 was designed to derive utilities from the SF-36v2, and our results indicate that it is still preferable to use the full questionnaire, although the difference with other variants of the questionnaire is very small. To use the SF-6Dv2 as an independent instrument will thus introduce minimal bias in utility values generated.