Longitudinal symptom-cluster trajectories and associated factors in patients with peripheral facial paralysis: a prospective cohort study

Objective To characterize stage-specific symptom-cluster organization, identify heterogeneous longitudinal trajectories of fixed-core psychological-sleep and fatigue burden, and examine sociodemographic, clinical, psychosocial, and treatment-timing correlates of trajectory membership in patients with peripheral facial paralysis (PFP). Methods In this prospective longitudinal cohort, 283 adults with PFP were assessed during the acute phase and at 1, 3, and 6 months after onset. Principal component analysis described stage-specific symptom-cluster organization. To ensure longitudinal comparability, trajectory analyses used fixed-core symptom-burden composites measured identically at all four assessments: anxiety and disturbed sleep for psychological-sleep burden, and lack of energy, drowsiness, and difficulty concentrating for fatigue burden. Piecewise random-intercept growth-mixture models were estimated in 239 complete cases, with full-information maximum-likelihood (FIML) sensitivity analyses in all 283 participants. Baseline factors associated with the final two-class solutions were examined using multiplicity-aware screening and Firth penalized binary logistic regression. Results The number of common symptoms decreased from 20 at baseline to 5 at 6 months, and time-specific PCA components were reorganized across recovery. Fixed-core psychological-sleep and fatigue burden both declined overall (Friedman χ 2 = 217.825 and 274.458, respectively; both p < 0.001). Two numerically admissible trajectories were retained for each domain. Psychological-sleep burden comprised an early-decreasing trajectory ( n = 199, 83.3%) and a delayed-decreasing trajectory ( n = 40, 16.7%); fatigue burden comprised a low-rapid-decreasing trajectory ( n = 200, 83.7%) and a high-persistent trajectory ( n = 39, 16.3%). FIML analyses in all 283 participants closely reproduced both two-class structures. No robust baseline predictor of psychological-sleep trajectory membership remained after multiplicity correction. For fatigue, autumn/winter onset (OR = 7.049, 95% CI 2.525–23.364), non-acute acupuncture initiation (OR = 11.281, 95% CI 4.625–30.030), and higher MCMQ avoidance scores (OR = 0.646 per point, 95% CI 0.482–0.810) were associated with membership in the high-persistent trajectory. Conclusion Symptom organization changed across recovery, while fixed-core psychological-sleep and fatigue burden showed two heterogeneous longitudinal courses. The primary class structures were robust to FIML handling of missing repeated measures, although censored-normal sensitivity analyses indicated that latent-class allocation may be sensitive to floor effects, particularly for fatigue. Associations with baseline characteristics should be interpreted as exploratory and non-causal.