Notice of Retraction: Worthington MA, et al. Dynamic Prediction of Outcomes for Youth at Clinical High Risk for Psychosis: A Joint Modeling Approach. JAMA Psychiatry. 2023;80(10):1017-1025.

To the Editor In consulting with another group attempting to replicate our analyses, we have identified a coding error in the joint modeling analyses in our article, “Dynamic Prediction of Outcomes for Youth at Clinical High Risk for Psychosis: A Joint Modeling Approach,” published online on August 2, 2023, and in the October 2023 issue of JAMA Psychiatry. Specifically, the converter by time interaction term used in the feature selection phase of the longitu dinal mixed-effects analyses was mistakenly retained in the longitudinal mixed-effects component of the joint models, but only the time effect should have been used in this phase. As a result of this error, the short-term longitudinal features used to boost performance of the baseline prediction mod els contained information on the outcomes to be predicted, making them not about prediction per se but about describing differential change as a function of outcome. The results for the base models and the feature selection stage are correct, but the results for the joint models that combine the bas e models and the selected longitudinal features are subject to this error. When the joint model analyses were rerun with the correct coding, they no longer showed improved prediction accuracy over and above the performance of the Cox regression models incorporating baseline-only predictors. Because the primary significance and novelty of the article were based on improved prediction in joint models incorporating information on ...

Source: JAMA Psychiatry - November 2, 2023 Category: Psychiatry Source Type: research