Subtyping negative symptoms in first-episode psychosis: Contrasting persistent negative symptoms with a data-driven approach

Persistent negative symptoms (PNS) are linked to poor functional outcomes and may be primary or caused by secondary factors. Although several studies have examined PNS in first-episode psychosis (FEP), a comparison with a data-driven approach is lacking. Here, we compared clinically defined PNS subgroups with class trajectories identified through latent growth modeling (LGM). Patients admitted to an early intervention service (N = 392) were classified as PNS (n = 105), secondary PNS (sPNS; n = 74), or non-PNS (n = 213) based on longitudinal data collected six to twelve months after admission. LGM was used to stratify patients based on similar negative symptom course over the same time period. Using multiple linear regression, we assessed the utility of both approaches in predicting Social and Occupational Functioning Assessment Scale (SOFAS) scores at two-year follow-up. Three negative symptom trajectories were identified: low and remitting (LR; n = 158), moderate and improving (MI; n = 163) and delayed partial response (DR; n = 71). Most non-PNS patients followed the LR trajectory, while patients with PNS or sPNS were generally divided between MI and DR. Both PNS classification and trajectory membership were significant predictors of two-year functional outcomes; the DR and MI trajectories predicted greater increases in SOFAS scores (DR: b = -19.14; MI: b = -11.54) than either sPNS (b = -9.19) or PNS (b = -6.46). These findings demonstrate that combining PNS and symptom-based stratification can predict functional outcomes more accurately than either taxonomy alone. Such a combined approach could yield significant advances in developing more targeted interventions for patients at risk for poor functional outcomes.