Inconsistency in the relationship between duration of untreated psychosis (DUP) and negative symptoms: sorting out the problem of heterogeneity.

摘要

BACKGROUND: Several studies have found an association between duration of untreated psychosis (DUP) and clinical outcomes. However, there is inconsistency concerning the association between outcome on negative symptoms and DUP with some studies having found a correlation between DUP and negative symptoms, while other studies did not find such an association. OBJECTIVE: The aim of the present study was to investigate the role of heterogeneity associated with the relationship between DUP and negative symptoms in a sample of first episode psychosis (FEP) patients from a multicentre treatment study and a replication sample of subjects from a specialized service in a different jurisdiction. METHOD: FEP patients (n=116) treated in specialized programs in two medium sized and one large urban centre were evaluated. Latent class regression was employed to simultaneously classify respondents and estimate the effect of DUP on negative symptoms after one year. The process was repeated on 59 consecutive FEP patients in a specialized service in Montreal. RESULTS: The final model reflected three distinct sub-groups with different associations between DUP and negative symptoms: (a) for one fourth of the subjects there was a positive association between DUP and negative symptoms, indicating that long DUP was associated with poor negative symptoms outcome; (b) an opposite effect was observed for another sub-group of patients: patients with short DUP scored high on the negative symptoms scale and patients with long DUP reported only a few negative symptoms; (c) there was no association between DUP and negative symptoms outcome for nearly half of the patients. These models were replicated in the Montreal sample. CONCLUSIONS: The association between DUP and negative symptoms outcome might differ among sub-groups of first episode patients. Latent class regression analysis offers a flexible way to include unmeasured heterogeneity in outcome analyses.