Early versus late administration of long-acting injectable antipsychotic agents among patients with newly diagnosed schizophrenia: an analysis of a commercial claims database

This study was designed to assess healthcare resource utilization (HCRU) and costs in patients with newly diagnosed schizophrenia based on timing and context of long-acting injectable antipsychotic agent (LAI) initiation. Using claims data, patients (aged 18–40 years) with first schizophrenia diagnosis January 2013–September 2019 (index date), no LAI or oral antipsychotic agent claims during 12-month preindex period, and continuous benefit enrollment from 12 months before index date to 12 months after first LAI administration were identified. Patients were grouped based on timing [early (≤1 year after index date) vs. late] and circumstances [reactive (after schizophrenia-related event) vs. proactive] of LAI initiation. Of 1290 patients with at least one LAI claim, 306 met criteria for early (n = 204; reactive, n = 107; proactive, n = 97) and late (n = 102; n = 75; n = 27) initiation. HCRU and costs were numerically lower in early versus late groups, and significantly lower for proactive initiation in both groups. Comparing worst-case (late-reactive) and best-case (early-proactive) scenarios, the average annual cost difference was $7195.13 (P = 0.0233), with major drivers being emergency department ($171.28; P < 0.05) and other outpatient ($2845.73; P < 0.00001) visits. In addition to the clinical advantages previously described in the literature, the proactive use of LAIs in early-phase schizophrenia is associated with lower healthcare costs.


Introduction
Schizophrenia is a complex, chronic psychiatric disorder characterized by impaired perception, cognitive ability, emotional expression, and social interactions (Patel et al., 2014;Kahn et al., 2015). Common symptoms of schizophrenia include delusions, hallucinations, and impaired communication and executive function (Patel et al., 2014;Kahn et al., 2015). Schizophrenia affects 0.6-1.9% of the USA population is associated with substantial burden for patients and their families, and is ranked among the top 20 illnesses that contribute to the total global burden of disease (Patel et al., 2014;GBD 2017 Disease andInjury Incidence andPrevalence Collaborators, 2018).
Antipsychotic treatment, such as oral antipsychotic agents (OAs) and newer long-acting injectable antipsychotic agents (LAIs), is the mainstay of therapy (Patel et al., 2014). Regarding relapse prevention and control of psychotic symptoms, LAIs confer several advantages over OAs (Subotnik et al., 2015;Kishimoto et al., 2021). Relapse prevention in schizophrenia is of the utmost importance because attenuated responses to schizophrenia treatment and treatment resistance have been observed following multiple relapses (Takeuchi et al., 2019;Taipale et al., 2022). In addition, LAIs have demonstrated improved patient outcomes, including reduced medication nonadherence, and reduced healthcare utilizations and related costs (Kishimoto et al., 2013(Kishimoto et al., , 2021Munday et al., 2019). Treatment nonadherence is most common early in the course of illness (Tiihonen et al., 2011;Rubio et al., 2021). It is during this phase of the illness when preventing relapses is most crucial, to avoid potential accumulation of morbidity later. Furthermore, LAI utilization in early-phase schizophrenia provides a clinically meaningful reduction in time to first hospitalization, an economically important outcome (Tiihonen et al., 2011;Kane et al., 2020). Given the clear advantage of LAIs in preventing relapses, and the high risk and impact of relapse during the early phase of the illness, LAIs are increasingly seen as an appropriate treatment option for early-phase schizophrenia, instead of only in patients who have poor adherence (Schreiner et al., 2015;Sajatovic et al., 2018;Florida Medicaid Drug Therapy Management Program, 2020;Kane et al., 2020). Furthermore, early use of LAI therapy is associated with neuroprotection that can further improve patient outcomes (Stevens et al., 2016).
Although LAIs have demonstrated improved clinical effectiveness and adherence versus OAs, they are consistently underutilized for the treatment of patients with schizophrenia (Pilon et al., 2017;Greene et al., 2018). It is of particular interest to further investigate the impact of early versus later LAI initiation in the disease course on healthcare utilization and costs among patients with newly diagnosed schizophrenia (Stahl, 2014), as well as in a commercially insured population (Fitch et al., 2014).
Given the possible advantages of initiating LAIs in the early phase of the illness, this US-based, retrospective claims database study was designed to assess whether earlier use of LAIs is associated with advantages in healthcare resource utilization (HCRU), as well as healthcare costs. In addition, this study sought to determine whether LAIs used proactively as relapse prevention tools, versus reactively following a relapse, corresponded to lower schizophrenia-related HCRU and costs among patients. We hypothesized that earlier implementation, as well as proactive initiation of an LAI, would provide the greatest HCRU-and cost-benefits to the patient population.

Study design
Patients with newly diagnosed schizophrenia [International Classification of Diseases (ICD)-9 and ICD-10] were followed using claims data from the US-based IBM MarketScan Commercial and Medicare Supplemental databases (IBM Watson Health, 2020) from 1 January 2012 to 30 September 2020 ( Supplementary  Fig. 1, Supplemental Digital Content 1, http://links.lww. com/ICP/A112). The IBM MarketScan databases use deidentified patient-level health data to capture information across the continuum of care (IBM Watson Health, 2020). Its Commercial Database consists of medical and pharmacy claims data for individuals enrolled in a variety of fee-for-service, fully capitated, and partially capitated health plans, including preferred provider organizations, exclusive provider organizations, and point-of-service plans (IBM Watson Health, 2020). The IBM MarketScan Medicare Supplemental Database covers medical and pharmacy claims for individuals enrolled in Medicare Supplemental health insurance plans (IBM Watson Health, 2020).
Key inclusion criteria were the first diagnosis of schizophrenia from 1 January 2013 to 30 September 2019 (index date), ≥1 LAI claim after index date, 18-40 years old, no schizophrenia diagnosis and no LAI or OA claims during 12-month preindex period (i.e. newly diagnosed schizophrenia), and continuous enrollment in a medical and pharmacy insurance plan from 12 months before index date through 12 months after first LAI administration (<30-day gaps allowed). The rationale for these criteria was to ensure that included patients were not enrolled in another insurance plan in which they could have received an earlier schizophrenia diagnosis or treatment as well as to ensure that patients remained enrolled after their first LAI administration in order to capture any HCRU and costs accrued after their first LAI administration. Patients were grouped based on the timing of their first LAI administration (early, ≤1 year after first schizophrenia diagnosis/index date; late, >1 year after) and context of LAI initiation [reactive after schizophrenia-related hospitalization (medical claim coded as nonemergency inpatient hospital visit) or emergency department (ED; medical claim coded as emergency hospital visit) visit; proactive, before hospitalization or ED visit]. In addition to schizophrenia-related hospitalizations and ED visits, medical claims were also categorized as office visits (coded as office visit), or other outpatient visits (outpatient medical claims without office visit, hospitalization, or ED coded).

Outcomes
Outcomes of interest included proportion of patients with newly diagnosed schizophrenia who initiated an LAI early or late; proportion of patients within early and late LAI groups that initiated an LAI reactively or proactively; proportion of patients with successful LAI implementation (≥90 days of continuous treatment with ≤7-day gaps); time to successful LAI implementation; and schizophrenia-related HCRU and costs within 12 months after first LAI administration.

Statistical analysis
All outcomes were examined descriptively. Mann-Whitney U and Chi-square/Fisher's exact tests were used to test statistical significance for continuous and categorical values, respectively. All outcomes and costs were annualized and reported as per patient per year, and costs were standardized to 2020 $USD.
Patients in the early and late LAI groups were then divided based on whether LAIs were initiated reactively (i.e. after ≥1 schizophrenia-related hospitalization or ED visit) or proactively (i.e. not after schizophrenia-related hospitalization or ED visit). In the early and late LAI groups, 107 patients and 75 patients started LAIs reactively, and 97 patients in the early LAI group and 27 patients in the late LAI group started LAIs proactively.
Proportions of patients with successful LAI implementation, defined as at least 90 days of continuous use with an allowance of 7 days or less for refills, were 53.9 and 48.0% in the early and late groups, respectively. Median [interquartile range (IQR)] times to successful LAI implementation were 177 (153) and 184 (237) days. Higher proportion of patients in the late-reactive subgroup had ≥1 and ≥2 schizophrenia-related hospitalizations or ED visits before LAI initiation versus those in the early-reactive subgroup; conversely, lower proportions in the late-proactive subgroup compared with the early-proactive subgroup had ≥1 and ≥2 schizophrenia-related hospitalizations or ED visits before LAI initiation ( Table 2). Implementation of LAI therapy was successful in approximately half of the schizophrenia cohort (52.0%; n = 159); Selection of study population. ED, emergency department; LAI, long-acting injectable antipsychotic agent; OA, oral antipsychotic agent. a The first diagnosis claim is the index date. b Enrolled 12 months before the index date and in the 12 months following initiation of first LAI (≤30-day gaps allowed). c Patients with (reactive) or without (proactive) ≥1 schizophrenia-related hospitalization or ED visit before LAI initiation. median (IQR) time to successful implementation was 180 (171) days from diagnosis. Among patients who initiated LAIs reactively, it took a median (IQR) of 45.0 (96.0) days for the early group compared with 525.0 (414.0) days for the late group (Table 2). Patients who initiated an LAI late and after at least one schizophrenia-related hospitalization or ED visit (i.e. late-reactive) took approximately 10 times longer to initiate LAI therapy after the index date [median (IQR), 608.0 (502.0) days] than those who initiated an LAI early [59.0 (148.0) days] ( A lower proportion of the early group (n = 107; 52.5%) initiated an LAI reactively compared with the late group (n = 75; 73.5%; P = 0.0004) ( Table 2). In the early and late groups, proportions of patients with at least one schizophrenia-related hospitalization after-LAI initiation were higher for patients who initiated an LAI reactively than for those who initiated it proactively (early, 20.6% vs. 10.3%, P = 0.0444; late, 20.0% vs. 7.4%, P = 0.2266) ( Table 3). Proportions of patients with other outpatient visits were significantly different between reactive and proactive subgroups in early (46.7% vs. 20.6%; P = 0.0001) and late groups (50.7% vs. 25.9%; P = 0.0264) ( Table 3). In early and late LAI groups, overall healthcare utilizations were lower in the proactive subgroups than those in the reactive subgroups. This translated into average total costs of schizophrenia-related healthcare differences of $5751.87 (P = 0.0086) and $7091.60 (P = 0.4851) between the reactive and the proactive subgroups for early and late groups, respectively (  (Table 3 and Fig. 2). The reactive subgroups had lower annual antipsychotic medication costs than the proactive subgroups, with the difference in the early group being lower than in the late group ($317.58 vs. $1075.16, respectively). Differences in antipsychotic medication use and cost between reactive and proactive subgroups in the late LAI group were significant where the proactive subgroup had a greater proportion of patients with ≥1 prescription (P = 0.0200) and higher costs (P = 0.0201).
Main drivers for cost savings between reactive and proactive subgroups were other outpatient visits and hospitalizations costs: the reactive subgroup had $1481.94 (P < 0.00001) and $3156.54 (P = 0.0063) higher annual other outpatient costs in the early and late groups, respectively (Table 3). Comparing late-reactive to early-proactive subgroups, the average cost difference was $2845.73 (P < 0.00001) ( Table 3 and Fig. 2). The largest significant differences in other outpatient and hospitalizations costs between the reactive and proactive subgroups were seen in the late LAI group where the reactive group had $3156.54 (P = 0.0063) and $4413.63 (P = 0.1347) higher costs for outpatient visits and hospitalization, respectively (Table 3). The early LAI group had significant differences in hospitalization costs where the reactive subgroup had $3356.45 (P = 0.0314) higher annual hospitalizations costs than the proactive subgroup (Table 3).

Discussion
Using a US-based commercial claims database, this study evaluated treatment patterns, HCRU, and costs among patients with newly diagnosed schizophrenia who began treatment with LAIs early (≤1 year) versus late (>1 year) following diagnosis. These data reveal that early LAI initiation can lead to more patients with successful implementation (≥90 days of continuous adherence) of LAI therapy. This aligns with several recent studies demonstrating that LAIs effectively improved daily adherence, and even more so when the LAI was initiated at illness onset or early in the diagnosis (Subotnik et al., 2015;Stevens et al., 2016;Greene et al., 2018).
The proportion of healthcare resource use, such as hospital admissions, readmissions, ED visits, and other    outpatient visits, was lower in the early LAI group than in the late LAI group. In addition, the early LAI group had lower schizophrenia-related costs per person than the late LAI group. This cost difference was primarily related to differences in hospitalization and other outpatient visit costs. These results align with a previous study that found a reduction in healthcare costs in a US-based commercially insured schizophrenia population after the initial diagnosis was largely driven by a reduction in hospitalizations and ED visits (Fitch et al., 2014). Furthermore, LAI use in early-phase schizophrenia significantly reduced the incidence rate of hospitalization after their implementation, which is an important economic outcome (Kane et al., 2020). Our assessment of total cost of care for schizophrenia showed an unadjusted average of $2301.03 in cost savings among the early group compared with the late LAI group. However, neither differences in utilizations nor costs between early LAI and late LAI groups were found to be statistically significant.
The results from this study suggest that patients with schizophrenia are not offered LAI therapy in a timely manner as demonstrated by the substantially lengthier time (10 times longer) for patients in the late LAI group to initiate an LAI after at least 1 schizophrenia-related high-risk event (median, 608 days); patients in the early LAI group that initiated an LAI reactively did so considerably more quickly (median, 59 days). There are several reasons why a clinician may be reluctant to initiate an LAI in a patient with newly diagnosed schizophrenia, including efficacy and tolerability concerns, reluctance to deviate from earlier treatment guidelines, expert opinions that advised OAs as first-line treatment, and limited availability of atypical LAIs on the market (Kirschner et al., 2013). In addition, a recent real-world survey found that clinicians tended to underestimate disease severity and overestimate medication adherence, leading to suboptimal treatment and poor patient outcomes (Keenan et al., 2022). However, in a cluster-randomized LAI clinical trial, in which participating investigators were required to undergo training about the study protocol and procedures, as well as training on how to discuss frequently asked questions with patients being offered LAI therapy, 91.0% of patients with first-episode or early-phase schizophrenia were willing to try LAI therapy when the option to do so was presented in a supportive manner by trained staff (Kane et al., 2019).
For early and late LAI groups, HCRU was lower among the proactive subgroup than the reactive subgroup. Antipsychotic medication costs did not differ greatly between early and late groups, which can counteract arguments against LAI use because of high medication costs. Although medication costs were lower in the reactive subgroups than in the proactive subgroups, the early-proactive subgroup had the lowest cost difference. Proactive Late-reactive versus early-proactive schizophrenia-related healthcare costs per patient. Early LAI, first LAI claim ≤1 year after index date. Late LAI, first LAI claim >1 year after index date. Proactive, no schizophrenia-related hospitalization or ED visit before LAI initiation. Reactive, schizophrenia-related hospitalization or ED visit prior to LAI initiation. *P < 0.05 **P < 0.00001. AP, antipsychotic agent; ED, emergency department; LAI, long-acting injectable antipsychotic agent.
LAI utilization, regardless of early or late initiation, was associated with lower HCRU and costs than reactive use; this was largely driven by ED visits and other outpatient visits. The largest significant differences in other outpatient visits and hospitalization costs between reactive and proactive subgroups were seen in the late LAI group, suggesting that proactive use, even if initiated late, would still provide cost savings. Overall, the early-proactive LAI subgroup had the lowest annual average schizophrenia-related costs, $4072.00 (8478.38), whereas the late-reactive LAI subgroup had the highest annual schizophrenia-related costs, $11 267.13 (23 741.67).
These data show that LAI implementation is generally associated with lower HCRU and costs among patients who received an LAI within 1 year of a schizophrenia diagnosis versus with those who initiated treatment with an LAI more than 1 year after diagnosis. Treatment guidelines for schizophrenia management recommend that LAIs be considered for initial therapy instead of reserving them strictly for patients who have demonstrated nonadherence or have had multiple relapses (Florida Medicaid Drug Therapy Management Program, 2020; Keepers et al., 2020). LAIs are a highly effective treatment option that offer a greater likelihood of stability for many patients with schizophrenia (Kishimoto et al., 2013(Kishimoto et al., , 2021. In real-world studies, LAIs have proved superior to OAs, lowering the risk of relapse, hospitalization, and all-cause discontinuation risk, despite greater illness severity in patients who initiate LAIs versus OAs (Kishimoto et al., 2013(Kishimoto et al., , 2021. Initiation of LAIs should be considered based on a combination of patient preference, patient education, and outcomes evidence (Florida Medicaid Drug Therapy Management Program, 2020; Keepers et al., 2020). These data support the early use of LAIs in newly diagnosed schizophrenia and demonstrate early use of LAIs corresponding to lower HCRU and schizophrenia-related healthcare costs.
There were several limitations in this study. Although young adults are more likely to have commercial insurance than Medicaid or Medicare, which are often utilized by older patients, claims data presented here are more reflective of younger patients with earlier-stage disease than earlier studies based on public insurance claims (Cohen et al., 2020). However, commercial claims data are still not entirely representative of all patients with early-phase schizophrenia. The 1-year cutoff timeframe to distinguish between early and late LAI initiation may be arbitrary since early use of an LAI has been variably defined as 1-5 years after initial schizophrenia diagnosis (Sliwa et al., 2012;Munday et al., 2019). However, the 1-year cutoff used in this study was based on a prior study evaluating LAI utilization among patients with schizophrenia that found that the median time between diagnosis and LAI initiation was approximately 290 days . In addition, the 1-year cutoff point was used in a previous study also looking at cost outcomes among patients with schizophrenia (Munday et al., 2019). The use of an LAI was regarded as reactive if initiated after at least one schizophrenia-related hospitalization or ED visit. However, it was possible that an LAI was used 'reactively' if a patient required a change in treatment without the need for hospitalization or an ED visit. The results of this analysis are descriptive in nature and are not adjusted for covariates such as patient demographics and schizophrenia-related factors. The inability to randomize the various conditions in the study (i.e. early vs. late initiation, reactive vs. proactive), prevents making causal interpretations. However, these results align with those of the PRELAPSE study, in which individuals were cluster-randomized by clinic to LAI prescriber training versus usual care, and it was found that the use of LAIs early in the course of illness was associated with greater clinical stability (Kane et al., 2019).
Overall, the study provides cost-based evidence in favor of early use of LAIs for patients with newly diagnosed schizophrenia with potential additive benefits if patients start LAIs proactively. Early intervention with LAI treatment could be an effective method for reducing schizophrenia-related hospital admission or readmission, as well as ED admission and other outpatient visits, ultimately resulting in the potential for substantial cost-savings, although these associations need to be tested in rigorous, randomized prospective trials.
Mental Health (NIMH); has received royalties/licensing fees from UpToDate; and owns stock/stock options in Doximity.