The DISABKIDS generic and diabetes-specific modules are valid but not directly comparable between Denmark, Sweden, and Norway.

ABSTRACT

BACKGROUND/OBJECTIVES: Government guidance promote benchmarking comparing quality of care including both clinical values and patient reported outcome measures in young persons with type 1 diabetes. The aim was to test if the Nordic DISABKIDS health-related quality of life (HrQoL) modules were construct valid and measurement comparable within the three Nordic countries. METHODS: Data from three DISABKIDS validation studies in Sweden, Denmark, and Norway were compared using Rasch and the graphical log-linear Rasch modeling. Monte Carlo methods were used to estimate reliability coefficient and target was defined as the point with the lowest SE of the mean. Self-report data were available from 99 Danish (8-18 years), 103 Norwegian (7-19 years), and 131 Swedish (8-18 years) young people. RESULTS: For the DISABKIDS higher scores on most subscales were noted in the Norwegian population. The Swedish sample had a significantly higher score on the "Diabetes treatment" subscale and scores closer to optimal target than the other countries. For each country, construct validity and sensitivity were acceptable when accounting for differential item function (DIF) and local dependency (LD). Less LD and DIF were found if only Denmark and Norway were included. The combined model was reliable; however, some differences were noted in the scale translations relating to the stem and response alternatives, which could explain the discrepancies. CONCLUSION: The Nordic versions of the DISABKIDS questionnaires measures valid and reliable HrQoL both within and between countries when adjusted for DIF and LD. Adjusting the Likert scales to the same respond categories may improve comparability.