Using a patient-centered approach to identify symptom clusters among adolescents with cancer

Background Prior studies identifying symptom clusters used a symptom-centered approach to demonstrate the relationship among symptoms. Latent profile analysis (LPA) is a patient-centered approach that classifies individuals from a heterogeneous population into homogeneous subgroups, helping prioritize interventions to focus on clusters with the most severe symptom burden. Objective The aim of this study was to use LPA to determine the best-fit models and to identify phenotypes of severe symptom distress profiles for adolescents with cancer who are undergoing treatment and in survivorship. Methods We used estimated means generated by the LPA to predict the probability of an individual symptom occurring across on- and off-treatment groups for 200 adolescents with cancer. Results The 3-profile solution was considered the best fit to the data for both on- and off-treatment groups. Adolescents on treatment and classified into the severe profile were most likely to report distress in appetite, fatigue, appearance, nausea, and concentration. Adolescents off treatment and classified into the severe profile were most likely to report distress in fatigue, pain frequency, and concentration. Conclusions Latent profile analysis provided a cluster methodology that uncovered hidden profiles from observed symptoms. This made it possible to directly compare the phenotypes of severe profiles between different treatment statuses. Implications for Practice The co-occurring 13-item Symptom Distress Scale symptoms found in the severe symptom distress profiles could be used as items in a prespecified severe symptom distress cluster, helping evaluate a patient's risk of developing varying degrees of symptom distress.