Quality of life in infants and children with atopic dermatitis: addressing issues of differential item functioning across countries in multinational clinical trials.

ABSTRACT

BACKGROUND: A previous study had identified 45 items assessing the impact of atopic dermatitis (AD) on the whole family. From these it was intended to develop two separate scales, one assessing impact on carers and the other determining the effect on the child. METHODS: The 45 items were included in three clinical trials designed to test the efficacy of a new topical treatment (pimecrolimus, Elidel cream 1%) in the treatment of AD in infants and children and in validation studies in the UK, US, Germany, France and the Netherlands. Rasch analyses were undertaken to determine whether an internationally valid, unidimensional scale could be developed that would inform on the direct impact of AD on the child. RESULTS: Rasch analyses applied to the data from the trials indicated that the draft measure consisted of two scales, one assessing the QoL of the carer and the other (consisting of 12 items) measuring the impact of AD on the child. Three of the 12 potential items failed to fit the measurement model in Europe and five in the US. In addition, four items exhibiting differential item functioning (DIF) by country were identified. After removing the misfitting items and controlling for DIF it was possible to derive a scale; The Childhood Impact of Atopic Dermatitis (CIAD) with good item fit for each trial analysis. Analysis of the validation data from each of the different countries confirmed that the CIAD had adequate internal consistency, reproducibility and construct validity. The CIAD demonstrated the benefits of treatment with Elidel over placebo in the European trial. A similar (non-significant) trend was found for the US trials. CONCLUSION: The study represents a novel method of dealing with the problem of DIF associated with different cultures. Such problems are likely to arise in any multinational study involving patient-reported outcome measures, as items in the scales are likely to be valued differently in different cultures. However, where all items in a scale fit both a single theoretical construct and the Rasch measurement model, it is feasible to conceive of outcome measures with a different set of items in each language.