Capturing and reporting topical treatment use in childhood eczema: lessons for data collection in eczema trials.

Abstract

Emollients and topical corticosteroids (TCS) prevent and treat flares in eczema. However, topical treatment use is poorly recorded and reported in clinical trials. There is no clear consensus of how best to capture and summarise topical treatment use. To explore different ways of capturing and reporting topical treatment use in childhood eczema. Secondary data analysis using 450 participants from the Best Emollients for Eczema (BEE) trial. Participants were allocated to use one type of emollient (lotion, cream, gel, or ointment) 'twice daily and when required' for 16 weeks. Otherwise, clinical management remained unchanged. Parents completed weekly questions about topical therapy use and eczema symptoms. Two versions of topical treatment use questionnaires were used. The first (n=202, 44.9%) asked parents to report treatment use on days 1-7, starting completion on the day they were randomised. The second (n=248, 55.1%) reported use by day of the week (Monday to Sunday), starting completion the first Monday after randomisation. Both underwent Patient and Public Involvement (PPI) review, but the second version was tested more thoroughly using cognitive interviewing techniques, following parent feedback that questions on the first version were confusing. Descriptive statistics compared questionnaire completion and differences in emollient and TCS use. Overall, questionnaire completion for both emollient and TCS use decreased with time: but at weeks 1 and 16 were 84.7% (381/450) and 58.9% (265/450) for emollient use, and 94.2% (424/450) and 80.4% (362/450) for TCS use, respectively. Fewer emollient use questionnaires were completed with first (33.5%) than the second (87.9%) version (p<0.001). TCS use questionnaire completion were similar for both (84.9% and 87.4%, p=0.002). We present different ways of summarising topical treatment use. While questionnaire completion was similar for TCS use, emollient use data completeness was higher in the second version. When designing questionnaires, balancing the detail and complexity of questions is important, especially if being collected as a secondary outcome measure. Numerous ways of summarising the same data can provide different information. Future collection and reporting of treatment use should reflect specific trial aims.