Predictors for relapse in a double-blind randomized placebo-controlled discontinuation study of remitted first-episode psychosis patients

Abstract

Background: Clinical decisions about relapse prevention and maintenance therapy following a first/single psychotic episode can be enhanced if features that predict higher relapse risk can be identified. While naturalistic studies suggest that maintenance medication reduces relapse risks, it is important that other risk factors controlling for medication status can also be identified.

Methods: Potential demographic, clinical and cognitive predictors for relapse were measured at the beginning of a double-blind randomized placebo-controlled study. Following a first/single episode with DSM-IV schizophrenia and related psychoses, remitted patients who had remained well on maintenance medication for at least 1 year were randomized to receive either maintenance therapy (with quetiapine 400 mg/day), or placebo for 12 months.

Results: 178 patients were randomized. Relapse rates were 35% (31/89) in maintenance group and 66% (59/89) in placebo group. Potential predictors were initially identified in univariate Cox regression models (p < 0.1) and were subsequently entered into a multivariate Cox regression model for measuring the relapse risk. Significant predictors included patients on placebo (hazard ratio = 0.388, CI = 0.235–0.641; p < 0.001); having more pre-morbid schizotypal traits (hazard ratio = 2.37, CI = 1.364–4.116; p = 0.002); scoring lower in the logical memory test (hazard ratio = 0.944, CI = 0.897–0.994; p = 0.028); and having more soft neurological signs (disinhibition) (hazard ratio = 1.305, CI = 0.996–1.709; p = 0.053).

Conclusions: Relapse predictors identified in this study may help to inform clinical decisions about discontinuation of maintenance therapy specifically for patients with a first/single episode psychosis following at least 1 year of maintenance therapy.

Acknowledgement: We are grateful to Dr TJ Yao at the Clinical Trials Center, University of Hong Kong, for statistical advice. The study was supported by investigator initiated trial award from AstraZeneca and the Research Grants Council Hong Kong (Project number: 765505).