Objective: Fatigue is one of the most prevalent and significant symptoms experienced by breast cancer patients. This study aimed to investigate potential population heterogeneity in fatigue symptoms of the patients using the innovative non-normal mixture modeling.
Methods: A sample of 197 breast cancer patients completed the brief fatigue inventory and other measures on cancer symptoms. Non-normal factor mixture models were analyzed and compared using the normal, t, skew-normal, and skew-t distributions. Selection of the number of latent classes was based on the Bayesian information criterion (BIC). The identified classes were validated by comparing their demographic profiles, clinical characteristics, and cancer symptoms using a stepwise distal outcome approach.
Results: The observed fatigue items displayed slight skewness but evident negative kurtosis. Factor mixture models using the normal distribution pointed to a 3-class solution. The t distribution mixture models showed the lowest BIC for the 2-class model. The restored class (52.5 %) exhibited moderate severity (item mean = 2.8-3.2) and low interference (item mean = 1.1-1.9). The exhausted class (47.5 %) displayed high levels of fatigue severity and interference (item mean = 5.8-6.6). Compared to the restored class, the exhausted class reported significantly higher perceived stress, anxiety, depression, pain, sleep disturbance, and lower quality of life.
Conclusions: The non-normal factor mixture models suggest two distinct subgroups of patients on their fatigue symptoms. The presence of the exhausted class with exacerbated symptoms calls for a proactive assessment of the symptoms and development of tailored interventions for this subgroup.