Development and validation of combined symptom-medication scores for allergic rhinitis

Background Validated combined symptom-medication scores (CSMSs) are needed to investigate the effects of allergic rhinitis treatments. This study aimed to use real-life data from the MASK-air® app to generate and validate hypothesis- and data-driven CSMSs. Methods We used MASK-air® data to assess the concurrent validity, test-retest reliability and responsiveness of one hypothesis-driven CSMS (modified CSMS: mCSMS), one mixed hypothesis- and data-driven score (mixed score), and several data-driven CSMSs. The latter were generated with MASK-air® data following cluster analysis and regression models or factor analysis. These CSMSs were compared with scales measuring (i) the impact of rhinitis on work productivity (visual analogue scale [VAS] of work of MASK-air® , and Work Productivity and Activity Impairment: Allergy Specific [WPAI-AS]), (ii) quality-of-life (EQ-5D VAS) and (iii) control of allergic diseases (Control of Allergic Rhinitis and Asthma Test [CARAT]). Results We assessed 317,176 days of MASK-air® use from 17,780 users aged 16-90 years, in 25 countries. The mCSMS and the factor analyses-based CSMSs displayed poorer validity and responsiveness compared to the remaining CSMSs. The latter displayed moderate-to-strong correlations with the tested comparators, high test-retest reliability and moderate-to-large responsiveness. Among data-driven CSMSs, a better performance was observed for cluster analyses-based CSMSs. High accuracy (capacity of discriminating different levels of rhinitis control) was observed for the latter (AUC-ROC = 0.904) and for the mixed CSMS (AUC-ROC = 0.820). Conclusion The mixed CSMS and the cluster-based CSMSs presented medium-high validity, reliability and accuracy, rendering them as candidates for primary endpoints in future rhinitis trials.