¹Dongguk University Internatinoal Hospital, Psychiatry, Goyang, Korea, Republic of, ²Sunshine Childrens Home & Rehabilitation Center, Psychiatry, New York, United States

Background: Long-acting injectable (LAI) antipsychotics are widely used to improve treatment adherence in patients with schizophrenia and related disorders. Despite their clinical advantages, concerns about side effects often influence patient acceptance and continuation of LAI treatment. While side effects are known to change over time, the temporal pattern of patient-reported side effects during LAI initiation and maintenance—and particularly when meaningful improvement begins—remains insufficiently characterized.

Aims & Objectives: This study aimed to (1) examine longitudinal changes in patient-reported side effects measured by the Liverpool University Neuroleptic Side Effect Rating Scale (LUNSERS) following initiation of LAI antipsychotics, and (2) identify change-points indicating the onset or acceleration of improvement in side effect burden.

Method: We analyzed longitudinal LUNSERS data from patients treated with paliperidone palmitate (Invega Sustenna) or aripiprazole LAI (Abilify Maintena). LAI initiation was defined as each patient’s earliest recorded LUNSERS assessment (day 0). Follow-up assessments were collected irregularly during monthly LAI administration. Missing values were imputed using variable-wise global mean imputation. Longitudinal changes in LUNSERS total and subscale scores were evaluated using generalized estimating equations (GEE), comparing baseline with subsequent 30-day time bins. To detect improvement onset, piecewise regression models were fitted with candidate breakpoints at 14-day intervals, and the optimal breakpoint was selected using the minimum Akaike Information Criterion (AIC). Patient-level cluster-robust standard errors were applied to account for within-subject correlation. Breakpoint analyses were conducted for the total score, subscales, and all 51 individual LUNSERS items.

Results: The dataset comprised 468 patients with 1,953 LUNSERS observations. GEE analyses demonstrated significant reductions from baseline in the LUNSERS total score and selected subscale domains over time (all p < 0.05). Piecewise regression identified significant post-breakpoint slope changes for multiple outcomes, reflecting either a transition to decreasing side effect burden or an acceleration of improvement (p < 0.05). At the item level, 20 of 51 LUNSERS items showed statistically significant slope changes followed by decreasing trends after the breakpoint. Identified change-points most frequently occurred during the early post-initiation period, approximately several weeks to a few months after LAI initiation (roughly 0.5–3 months).

Discussion & Conclusions: Patient-reported side effects following LAI antipsychotic initiation do not appear to improve in a simple linear manner. Instead, many side effects demonstrate a clinically meaningful change-point after an early adaptation phase, beyond which improvement begins or accelerates. Recognizing these temporal patterns may help clinicians refine monitoring strategies, provide realistic expectations to patients during LAI initiation, and support treatment continuation. Future studies should incorporate advanced missing-data approaches and relevant clinical covariates within mixed-effects frameworks to further clarify determinants of side effect trajectories.