Frequently Asked Questions

After second-generation long-acting injectable antipsychotic initiation, psychiatric care shifted away from emergency and inpatient settings toward outpatient follow-up. In the 12 months after initiation, both stable users and non-stable users had more psychiatric outpatient visits and fewer psychiatric emergency department visits, admissions, and inpatient days than in the 12 months before initiation.

Among stable users, outpatient visits increased from 10.43 to 15.90, ED visits fell from 0.28 to 0.08, admissions fell from 2.96 to 1.92, and length of stay fell from 86.21 to 60.20 days. Among non-stable users, outpatient visits increased from 8.98 to 13.49, ED visits fell from 0.29 to 0.12, admissions fell from 2.45 to 2.01, and length of stay fell from 67.04 to 51.64 days.

Yes, psychiatric emergency department use decreased after second-generation long-acting injectable antipsychotic initiation in both groups. Among stable users, mean psychiatric ED visits decreased from 0.28 to 0.08 in the 12 months before versus after initiation (P < .001), and ED medical expenditures decreased from US$35.89 to US$9.16 (P < .001).

Among non-stable users, mean psychiatric ED visits decreased from 0.29 to 0.12 (P < .001), and ED medical expenditures decreased from US$40.87 to US$13.30 (P < .001). The study interpreted these findings as an association with reduced acute-care use, not definitive proof of a causal medication effect.

Hospital-based psychiatric use was lower after initiation in the primary mirror-image analysis, but the study advises cautious interpretation. Among stable users, psychiatric admissions decreased from 2.96 to 1.92 (P < .001) and length of stay decreased from 86.21 to 60.20 days (P = .0048). Among non-stable users, admissions decreased from 2.45 to 2.01 (P = .006) and length of stay decreased from 67.04 to 51.64 days (P < .001).

However, when the ±30-day index window was excluded in a sensitivity analysis, the apparent reductions in inpatient utilization among non-stable users were not robust; after exclusion, admissions, length of stay, and inpatient expenditures increased during the postindex period. The authors therefore frame the inpatient findings as associations that may partly reflect index-event bias, regression to the mean, or intensified monitoring after initiation.

Outpatient psychiatric engagement increased substantially after initiation. Among stable users, mean psychiatric outpatient visits increased from 10.43 to 15.90 (P < .001), outpatient drug costs increased from US$356.16 to US$2,860.08 (P < .001), and outpatient medical expenditures increased from US$670.95 to US$3,618.45 (P < .001).

Among non-stable users, outpatient visits increased from 8.98 to 13.49 (P < .001), outpatient drug costs increased from US$282.82 to US$1,213.78 (P < .001), and outpatient medical expenditures increased from US$537.98 to US$1,704.12 (P < .001). The authors suggest that regularly scheduled injection visits may create recurring points of clinical contact that support monitoring and continuity of care.

Not clearly. Total drug costs increased after second-generation long-acting injectable antipsychotic initiation in both groups, and reduced acute-care use did not consistently translate into short-term total medical cost savings.

In stable users, total drug costs increased from US$1,405.48 to US$3,631.88 (P < .001), while total medical costs decreased from US$8,112.19 to US$7,394.73 but this was not statistically significant (P = .188). In non-stable users, total drug costs increased from US$935.82 to US$1,664.40 (P < .001), and total medical costs decreased from US$6,593.09 to US$5,553.65 (P < .001). The between-group difference in total medical cost change was not significant (P = .584), whereas the between-group difference in total drug cost change was significant (P < .001).

Stable use meant sustained, label-frequency-concordant use during the 12-month follow-up period. Patients were classified as stable users if they met or exceeded the minimum recommended injection frequency for the specific long-acting injectable they started.

• Abilify Maintena or Invega Sustenna: at least 12 injections in the first postindex year
• Risperdal Consta: at least 24 injections
• Invega Trinza: at least 4 injections

The authors specifically state that this was an implementation-oriented indicator of treatment persistence and continued participation in scheduled injection-based care, not a surrogate for superior intrinsic pharmacologic efficacy.

Only 500 of 2,919 patients in the final analytic sample met the study's definition of stable use. The study identified 2,937 first-time second-generation long-acting injectable antipsychotic users with bipolar disorder, then excluded 18 patients with missing health care expenditure data, leaving 500 stable users and 2,419 non-stable users for the utilization and expenditure analyses.

The authors describe this as an important implementation finding, because persistence in this study reflects whether long-acting injectable-based care could be sustained in routine practice.

The study does not support that interpretation. The authors explicitly state that stable use should not be interpreted as evidence of superior intrinsic pharmacologic efficacy. Instead, stable use was treated as an implementation-related marker of sustained engagement with an LAI-based care pathway.

Patients who remained on scheduled injections for 12 months may have differed from those who discontinued earlier in baseline illness stability, treatment acceptability, family or service support, logistical feasibility, and willingness to stay engaged in care. For that reason, larger reductions in acute-care utilization among stable users are better understood as reflecting a more durable continuity-of-care pathway rather than a head-to-head efficacy advantage.

This was a nationwide retrospective mirror-image study using Taiwan's National Health Insurance data from 2015 to 2024. The analysis compared each patient's psychiatric outpatient, emergency, and inpatient service use during the 12 months before second-generation long-acting injectable antipsychotic initiation with the 12 months after initiation.

This design is useful for within-person comparison, but it cannot fully separate the effects of the long-acting injectable from temporal factors such as spontaneous improvement after a crisis, regression to the mean, or intensified follow-up after treatment initiation. The authors therefore interpret the observed reductions in acute-care utilization as associations rather than definitive causal effects.

The main limitations are the mirror-image design and the constraints of claims data. Because long-acting injectable antipsychotics are often started during or soon after deterioration, hospitalization, relapse, or poor adherence, the preindex period may be unusually high-use, which can make postinitiation reductions look larger because of index-event bias or regression to the mean.

The National Health Insurance claims data also did not include symptom severity, symptom trajectories, functional outcomes, patient preferences, family support, psychosocial interventions, or whether individual injection visits included psychotherapy, social work, or family-based care. As a result, changes in utilization should not be interpreted as direct evidence of improved clinical outcomes, cost-effectiveness, long-term functioning, or mortality benefit.

The study suggests that second-generation long-acting injectable antipsychotics may be most useful as a continuity-of-care strategy rather than a short-term cost-saving intervention. In this nationwide Taiwan cohort, initiation was associated with more planned outpatient follow-up and less emergency and inpatient psychiatric use, which is consistent with a structured care pathway for patients with recurrent relapse or unreliable oral adherence.

At the same time, clinicians should not assume that reduced acute-care use proves a causal medication effect or guarantees lower total medical costs in the first year. The authors also emphasize that stable use reflects sustained engagement with injection-based care, not proof that one group or one formulation had superior pharmacologic efficacy.