BACKGROUND: The age of onset (AOO) is a key factor for heterogeneity in major depressive disorder (MDD). Looking at the effect of AOO on symptomatology may improve clinical outcomes. This study aims to examine whether and how AOO affects symptomatology using a machine learning approach and latent profile analysis (LPA).
METHODS: The study enrolled 915 participants diagnosed with MDD from eight hospitals across China. Depressive symptoms were assessed using the 17-item Hamilton Depression Rating Scale. The relationship between symptom profiles and AOO was explored using Random Forest. The effect of AOO on symptom clusters and subtypes was investigated using multiple linear regression and LPA. A continuous AOO indicator was used to conduct the analyses.
RESULTS: Based on the Random Forest, symptom profiles were closely associated with AOO. The regression model showed that the severity of neurovegetative symptoms was positively associated with AOO (β = 0.18, p < 0.001), and the severity of cognitive-behavioral symptoms was negatively associated with AOO (β = -0.12, p < 0.001). LPA demonstrated that the subgroups characterized by suicide and guilt had earlier onset of depression. The subgroup with the lowest global severity of depression had the latest onset.
LIMITATIONS: AOO was recalled retrospectively. The relative scarcity of participants with childhood and adolescence onset depression.
CONCLUSIONS: AOO has an important impact on symptomatology. The findings may enhance clinical evaluations for MDD and assist clinicians in promoting earlier detection and individualized care in vulnerable individuals.
Language: en
Keywords
Machine learning; Major depressive disorder; Age of onset; Latent profile analysis; Random forest; Symptomatology