D3.198 - Challenges of using clinical scores as primary endpoints in allergology trials

Composite clinical scores are widely used to assess the efficacy of allergen immunotherapy (AIT) in allergic rhinoconjunctivitis. Among these, the Combined Symptom–Medication Score (CSMS) is recommended for use in clinical trials. However, despite its seemingly quantitative nature, the CSMS raises important methodological concerns related to measurement theory, aggregation of heterogeneous clinical states, and the validity of associated statistical analyses.

We conducted a critical appraisal of the CSMS from three complementary perspectives: (1) measurement theory, focusing on the implications of conflating ordinal and cardinal scales; (2) statistical consequences of analyzing the CSMS as a continuous variable; and (3) the assessment of alternative endpoints and inferential approaches, including responder-based definitions.

The CSMS combines a symptom score derived from ordinal item ratings with a medication score reflecting treatment use. Although the resulting total score spans a quasi-continuous range from 0 to 6, combinatorial analyses demonstrate that multiple heterogeneous combinations of symptoms and medication use can map onto identical CSMS values, resulting in a substantial loss of clinical information. This is illustrated by the bar chart showing the number of possible combinations of symptoms and medication use (y-axis) according to CSMS levels (x-axis). For example, approximately 700 distinct combinations correspond to a CSMS value of 2. Treating the CSMS as a continuous endpoint entails unjustified equal-interval assumptions of symptoms intensity, potentially biasing effect estimates and increasing required sample sizes. In contrast, responder-based endpoints—such as achieving a predefined, clinically meaningful improvement—offer enhanced interpretability and stronger methodological validity.

The CSMS exhibits structural and inferential limitations that challenge its suitability as a primary continuous endpoint in allergology trials. The adoption of other clinically meaningful outcomes such as responder-based endpoints, together with appropriate statistical frameworks, may substantially improve the robustness and interpretability of AIT clinical trial outcomes.