Background/Purpose:

The aim of this study was to assess the validity of a short version of the electronic OAKHQOL questionnaire in a combined analysis with classical test theory (CTT) and a Rasch measurement model for item response theory.

Methods:

Patients with knee osteoarthritis (ACR criteria) were consecutively recruited by their general practitioners. They responded to the electronic version of the OAKHQOL questionnaire, called e-OAKHQOL ¹. It contains 43 items and describes health-related quality of life (HRQoL) in 5 dimensions: physical activity, mental health, pain, social functioning and social support. Each total score is normalized to a score from 0 (worst HRQoL) to 100 (best HRQoL). A shortened version of the OAKHQOL has been developed, the mini-OAKHQOL. It retains 20 items of the original instrument and maintains the same structure with five dimensions. The scoring method was kept as the original, for items and dimensions. For the present study, only items of the e-OAKHQOL retained in the mini-OAKHOQL were included in the statistical analysis. Metric properties were assessed by the CTT approach and a Rasch measurement model (partial credit model).

Results:

The confirmatory factor analysis showed that the five-factor model best fitted the data (RMSEA= 0.067 (90% confidence interval: 0.059-0.075), CFI=0.94, TLI=0.93). Cronbach’s alpha coefficients were good to excellent for four of the five subscales. As expected, discrimination was good for the physical activity and the pain dimensions. In Rasch analysis, one item (“Staying for a long time in the same position”) showed underfit (item not discriminative enough). A lack of fit to the model was observed for the social support and the social functioning dimensions (significant item-trait interaction). Internal consistency was excellent for the physical activity (Person Separation Index (PSI) = 0.92) and pain (PSI = 0.84) dimensions but were less than 0.80 for the other dimensions. Examination of person-item threshold distribution map demonstrated good targeting between persons and items in the physical activity and the pain dimensions while it was fairly good in the three other dimensions. No statistical differential item functioning was detected for all 5 dimensions by gender, age, educational level, low-back pain or body mass index.

Conclusion:

Analysis with CTT and Rasch measurement of the mini e-OAKHQOL questionnaire revealed the good measurement properties of the 5 dimensions of the questionnaire. The shortening of the e-OAKHQOL has considerably improved its structural properties. However, some shortcomings remain such as item-trait interaction. Further research is needed to confirm or refute these findings in an independent sample.

Reference: ¹ Wieczorek M, Rotonda C, Guillemin F, Rat AC. The Electronic Osteoarthritis Knee and Hip Quality of Life (OAKHQOL) Questionnaire: A Useful and Valid Alternative to Measure Health Related Quality of Life in Knee Osteoarthritis [abstract]. Arthritis Rheumatol. 2016; 68 (suppl 10)

« Back to 2018 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/could-the-shortening-of-the-electronic-osteoarthritis-knee-and-hip-quality-of-life-oakhqol-questionnaire-improve-its-metric-properties/