Endocrine Abstracts

Introduction: Patients with Addison’s disease (AD) self-report impairment in specific dimensions on well-being questionnaires. An AD disease-specific quality of life questionnaire (AddiQoL) was developed to aid evaluation of patients. We here aimed to validate translated questionnaires in terms of internal consistency and reliability and to improve the construct of the scale.

Methods: After translation by a multistep approach, the final versions were tested in AD patients from Norway (n=107), Sweden (n=101), Italy (n=165) and Germany (n=200). Internal consistency was examined by exploratory factor analysis (EFA) and Rasch analysis, aiming at unidimensionality and fit to the Rasch model. Cronbach’s α and person separation index (PSI) were calculated for reliability. Longitudinal reliability was tested by differential item functioning (DIF, or item bias for time) in stable patient subgroups in Norway, Sweden and Italy.

Results: EFA identified four sub-dimensions of AddiQoL; fatigue (8 items), emotions (8), symptoms (11) and miscellaneous (sleep, sexuality and impact of intercurrent disease, 6). Three items concerning nocturia, weight gain, and dry skin did not fit any dimension and were discarded. The sub-dimensions fatigue, emotion and miscellaneous showed non-significant χ² interactions, implying fit to the Rasch model, also supported by the overall fit residual statistics. The symptoms sub-dimension initially displayed misfit, hence three items concerning salt cravings, light headedness and sweating were removed. The 30 remaining items, rearranged in four subtests by sub-dimensions, fitted the Rasch model; supported by a non-significant χ² interaction of 0.65 and no significant pattern deviation among the fit residuals. Crohnbach’s α was 0.94, and PSI was 0.86, indicating good reliability. Longitudinal reliability was excellent, as no significant DIF between separate time points was detected.

Conclusion: The Norwegian, Swedish, Italian and German AddiQoL versions have high internal consistency and reliability, and can be fitted to the Rasch model for optimisation of psychometric properties.