A clustering approach to identify severe bronchiolitis profiles in children

Area of Science:

• Pediatric respiratory diseases
• Clinical phenotyping
• Epidemiology of viral infections

Background:

• Bronchiolitis is typically viewed as a singular condition, but emerging evidence suggests significant underlying heterogeneity.
• Understanding these variations is crucial for accurate diagnosis and effective management of pediatric respiratory illness.

Purpose of the Study:

• To identify distinct clinical profiles (phenotypes) of severe bronchiolitis in hospitalized children.
• To utilize a clustering approach to reveal heterogeneity within severe bronchiolitis cases.

Main Methods:

• Latent class analysis was applied to clinical and etiological data from two large, multicenter cohorts of hospitalized infants (<2 years).
• Data included clinical presentation, medical history (wheezing, eczema), and identified viral pathogens.
• Cohorts comprised children from the USA (n=2207) and Finland (n=408).

Main Results:

• Four distinct profiles were identified in the US cohort: Profile A (wheezing history, eczema, rhinovirus), Profile B (wheezing, higher RSV probability), Profile C (most severe, longer hospitalization), and Profile D (least severe, non-wheezing).
• Profiles A and D were replicated in the Finnish cohort; a combined 'BC' profile was observed in Finnish children.
• Significant variations in clinical severity, hospitalization duration, and viral etiology were noted across profiles.

Conclusions:

• A clustering approach successfully identified distinct clinical phenotypes of severe bronchiolitis in children.
• The identified heterogeneity is critical for future research into bronchiolitis etiology, treatment strategies, and long-term outcomes, including childhood asthma risk.