Improving Severity Scoring of Food-Induced Allergic Reactions: A Global "Best-Worst Scaling" Exercise

There is no current consensus on assigning severity to food-induced allergic reactions, for example, to assess the efficacy of allergen immunotherapy. Existing severity scores lack the capability to discriminate between non-anaphylaxis reactions of different severities. Attempts are ongoing to develop a more discriminatory score, which should ideally be data-driven and validated in multiple cohorts.

To undertake an exercise using best-worst scaling (BWS) to define a potential gold standard against which severity scoring of food-induced allergic reactions can be refined.

We undertook a global survey to better understand how health care professionals rate the severity of food-induced allergic reactions, using BWS methodology. Respondents were given a number of patient case vignettes describing real-world allergic reactions and asked to select the pair that, in their opinion, reflected the maximum difference in severity. Responses were then modeled and a preference score (representing severity) determined for each scenario. Scenarios were also scored using existing published scoring systems and the scores compared with the BWS score using Spearman r correlation and Cohen kappa. Given the differences in definitions of anaphylaxis globally, we also evaluated differences in BWS ranking depending on the geographical location of respondents.

We received 334 complete responses, 183 (55%) from Europe and 65 (20%) from North America. Perception of severity of some reactions appeared to be affected by geographical location. The comparison of BWS ranking with current grading systems identified significant issues that varied from one grading system to another, such as prominence to some symptoms (eg, vomiting) that skew grading when using scoring systems not designed for food allergy. In general, current scoring systems poorly discriminate against more mild symptoms and often overestimate their severity.

These data provide a methodology free of user scale bias to help define a potential, consensus-driven gold standard that can be used to guide and validate the development of improved grading systems to score food-induced allergic symptoms and highlight areas for education where there is the potential to miscategorize severity.