Twelve-Month Cumulative Incidence, Characteristics, and Correlates of Non-adherence in Youths Treated with Second-Generation Antipsychotics for Mental Disorders: Results from a Prospective Cohort Study

Antipsychotic non-adherence in youths with mental disorders is multifactorial, may remain clinically unnoticed, and has been underresearched. To characterize antipsychotic non-adherence in youths, including its correlates. Data were collected from the naturalistic Second-Generation Antipsychotic Treatment Indications, Effectiveness and Tolerability in Youth (SATIETY) cohort study, which included youths (ages 4–17 years) who initiated clinician’s choice second-generation antipsychotics (SGAs) for psychosis-, mood-, or disruptive behavior-spectrum disorders and had ≥ 1 outpatient day of follow-up. Non-adherence was assessed via interview and antipsychotic plasma levels, and cumulative rates were analyzed via Kaplan–Meyer survival analysis and observed incidences. Characteristics of non-adherence and non-adherent patients were examined. Three multivariable regression models were used to identify predictors of observed cumulative non-adherence over 12 months. Altogether, 521 youths (median age 14.9 years; male 55.1%; post-pubertal 72.7%; antipsychotic-naïve 65.5%; inpatients 74.7%; mood-spectrum disorder 56.5%, schizophrenia-spectrum disorder 22.9%, aggression-spectrum disorder 20.6%) started risperidone (37.2%), aripiprazole (26.9%), quetiapine (22.8%), or olanzapine (13.1%). The observed 12-month cumulative SGA non-adherence (occurring after a median follow-up of 75 days, interquartile range [IQR] 30–135 days) was 24.4% (overt 61.4%, covert/blood level-detected 38.6%; decision by: patient 75.6%, caregiver 18.1%, both 6.3%), increasing to 36.2% (95% confidence interval [CI] 30.4–41.9%) in the survival analysis. Non-mutually exclusive reasons for non-adherence, defined as SGA discontinuation against medical advice, included perceived lack of SGA treatment need (87.4%), intolerability (22.1%), and/or inefficacy (4.7%). In the first multivariable model in the full dataset, non-adherence was predicted by substance use disorder (odds ratio [OR] 3.38; 95% CI 1.85–6.17), non-white origin (OR 1.62; 95% CI 1.04–2.52), more last visit psychotropic medications (OR 1.60; 95% CI 1.27–2.00), and shorter pre-baseline lifetime antipsychotic treatment (OR 1.09; 95% CI 1.00–1.13 per month) ( n = 515, r 2 = 0.10, p < 0.0001). Including non-dual-biological parent household (available in n = 435) in the second, exploratory model, non-dual-biological parent household and the four predictors of the first model predicted non-adherence ( r 2 = 0.13, p < 0.0001). Finally, including last visit patient and caregiver Drug Attitude Inventory (DAI)-30/DAI-10 ratings (available in n = 273) to the third, exploratory model, substance use disorder, non-dual-biological parent household, more last visit psychiatric medications, worse last visit DAI-30 patient ratings, and risperidone treatment predicted non-adherence ( r 2 = 0.17, p < 0.0001). One in four youths starting SGAs (increasing in survival analyses accounting for early dropouts to more than one in three youth) discontinued SGAs against medical advice within 12 months, mainly owing to perceived lack of treatment need. As much as 39% of the non-adherence would have remained undetected without SGA plasma level assessments. Interventions should target patient, caregiver, and medication factors to reduce barriers and improve outcomes.