Poor functional recovery is better predicted than conversion in studies of outcomes of clinical high risk of psychosis: insight from SHARP

TianHong Zhang,ShuWen Yang,LiHua Xu,XiaoChen Tang,YanYan Wei,HuiRu Cui,HuiJun Li,YingYing Tang,Li Hui,ChunBo Li,XingShi Chen,JiJun Wang

DOI: https://doi.org/10.1017/s0033291719002174

IF: 10.592

2020-07-28

Psychological Medicine

Abstract:Background Few of the previous studies of clinical high risk of psychosis (CHR) have explored whether outcomes other than conversion, such as poor functioning or treatment responses, are better predicted when using risk calculators. To answer this question, we compared the predictive accuracy between the outcome of conversion and poor functioning by using the NAPLS-2 risk calculator. Methods Three hundred CHR individuals were identified using the Chinese version of the Structured Interview for Prodromal Symptoms. Of these, 228 (76.0%) completed neurocognitive assessments at baseline and 199 (66.3%) had at least a 1-year follow-up assessment. The latter group was used in the NAPLS-2 risk calculator. Results We divided the sample into two broad categories based on different outcome definitions, conversion ( n = 46) v. non-conversion ( n = 153) or recovery ( n = 138) v. poor functioning ( n = 61). Interestingly, the NAPLS-2 risk calculator showed moderate discrimination of subsequent conversion to psychosis in this sample with an area under the receiver operating characteristic curve (AUC) of 0.631 ( p = 0.007). However, for discriminating poor functioning, the AUC of the model increased to 0.754 ( p < 0.001). Conclusions Our results suggest that the current risk calculator was a better fit for predicting a poor functional outcome and treatment response than it was in the prediction of conversion to psychosis.

psychiatry,psychology, clinical