Off-Label Antipsychotic Withdrawal in People With Intellectual Disabilities: Development and Internal Validation of a...

Off-label antipsychotic use in people with intellectual disabilities and challenging behaviour is high. Antipsychotic withdrawal is recommended, but attempts are often unsuccessful. This study aimed to develop and internally validate a prediction model that provides insight into predicting factors for unsuccessful (i.e. incomplete) off-label antipsychotic withdrawal attempts in people with intellectual disabilities.

Data collected in two previous studies examining the withdrawal of off-label antipsychotics in people with intellectual disabilities and challenging behaviour living mostly in 24/7 care settings (98.6%) in the Netherlands were analysed. The dataset included 141 participants (64.5% male, median age 52). We selected candidate predictors (age, level of intellectual disability, defined daily dose, autism spectrum disorder and three subscales of the Aberrant Behavior Checklist [ABC], namely stereotypy, hyperactivity and lethargy) based on previous research and clinical relevance. A multivariable logistic regression analysis with backward selection procedures was conducted to identify significant predictors. The model was internally validated using bootstrapping procedures.

The analysis revealed the level of intellectual disability (p = 0.030, OR = 2.374), defined daily dose (p = 0.063, OR = 2.833), and ABC stereotypy (p = 0.007, OR = 1.106) as key predictors for unsuccessful withdrawals. The variables explained 20% of the variance (Nagelkerke's R-square, R² = 0.200). The model calibrated well as the Hosmer and Lemeshow test was not significant. The discrimination of the model was fair to good; the Area Under the Curve (AUC) was 0.728. Internal validation procedures showed an optimism-corrected AUC of 0.706; the optimism-corrected Nagelkerke's R² was 0.157.

The odds of unsuccessful withdrawal increase with a more severe level of intellectual disability, a higher antipsychotic defined daily dose and higher stereotypy scores. The results inform healthcare providers about the predictive factors enabling them to better anticipate and support future withdrawal attempts.