Parsing the impact of early detection on duration of untreated psychosis (DUP) : Applying quantile regression to data from the Scandinavian TIPS study

Abstract

Background: Prolonged duration of untreated psychosis (DUP) is associated with poor outcomes. The TIPS study halved DUP with an early detection (ED) campaign; however, conventional statistical analyses, focused on mean estimates, failed to reveal the effects of ED across the full DUP distribution, restricting inferences about ED's effectiveness. Utilizing a novel quantile regression based analysis, we examined the differential impact of ED across DUP. Secondary analysis explored possible predictors of DUP, and moderators of the effect of the campaign. Methods: The TIPS ED campaign was conducted in two health care sectors in Norway, with two equivalent health care sectors serving as controls. Quantile regression analysis was performed to analyze ED campaign's effect. Results: 281 patients with first episode psychosis were recruited, including 141 from the ED area. ED had no effect on the first quartile (Q1) of DUP, whereas a significant reduction in Q2 (11 weeks), and Q3 (41 weeks) of DUP was observed. The effect of ED was significantly stronger on reducing Q3 than Q1 or Q2, suggesting that the campaign was more effective in longer DUP samples. Male gender and single status predicted longer DUP in Q3: by 38 and 27 weeks, respectively. Single status, but not gender, emerged as a significant moderator of ED campaign effect. Conclusions: Quantile regression provided in depth information about the non-uniformity, and moderators, of TIPS's ED effort across the full distribution of DUP, demonstrating the value of this analytic approach to re-examine prior, and plan analyses for future, early detection efforts.