Using Latent Profile Analysis to Classify Longitudinal Symptom Burden in Children with Advanced Cancer (GP767)

Abstract

• Explain how groups of children with advanced cancer experience different levels of symptom burden over time.• Describe factors associated with symptom classes. Children with advanced cancer experience physical and psychological symptom burden. Emphasis on cross-sectional evaluation of symptoms in children with advanced cancer has led to gaps in understanding the dynamic nature of symptoms over time and factors associated with different trajectories of symptom burden. The presence of distinct symptom trajectories in children with advanced cancer is largely unknown. To describe symptom burden trajectories and factors associated with those trajectories using latent profile analysis of longitudinal symptom data from children with advanced cancer. Symptom data were collected every two weeks for a median of 21 weeks from 46 children with advanced cancer. Data were analyzed using latent profile analysis to classify trajectories in symptom prevalence (burden) over time. Age, gender, and clinical characteristics were compared between the observed classes. The best fitting model described two distinct classes of symptom burden based on the prevalence of pain, fatigue, nausea, sleeping difficulties, dyspnea, anxiety, and sadness over time. Participants in Class 2 experienced a trajectory of higher symptom burden over time compared to Class 1. Most children were represented in either Class 1 (n=23) or Class 2 (n=20). Classes did not differ by diagnosis or age, but differed by sex and disease status. Class 1 had a higher proportion of males, while Class 2 had a higher proportion of females, children with active disease, and recent disease progression (<30 days). A subgroup of children with advanced cancer experienced a trajectory of high symptom burden. Female sex, active disease, and recent disease progression were associated with higher symptom burden over time.