Ellie S. R. Elliott, PharmD, BCPP

Schizophrenia is a chronic mental illness, characterized by episodes of relapse and control of symptoms, affecting both social and occupational functioning. Up to 80% of patients diagnosed with schizophrenia or schizoaffective disorder are nonadherent to oral antipsychotic medication, contributing to high relapse rates.

 Nonadherence to prescribed antipsychotics contributes to increased clinic and emergency department visits and psychiatric hospitalizations, resulting in increased healthcare expenditures.

 Typical long-acting injectable antipsychotics (eg, haloperidol and fluphenazine) were developed in the 1960s to improve adherence rates and decrease relapse.

 Risperidone long-acting injection (RLAI) therapy is the first atypical antipsychotic marketed in the United States and approved for the treatment of schizophrenia (2003) and for bipolar I disorder (2009). The unique drug delivery system releases medication over an extended period, allowing for dosing intervals every 2 weeks. Several pharmacoeconomic analyses have concluded that patients receiving RLAI have fewer hospitalizations compared with those receiving oral antipsychotic therapy or during the time period before initiation of RLAI, resulting in a reduction in total costs.

 Although there is a definitive relationship between RLAI adherence and reduced economic costs in the treatment of schizophrenia, adherence data in a naturalistic setting with RLAI is limited.

Medication adherence is affected by a multitude of individual and environmental factors. In schizophrenia, studies have demonstrated that medication adherence is infl uenced by global functioning level, substance abuse, level of social support, and quality of therapeutic alliances with healthcare professionals.

 The impact of discharge planning that occurs in the inpatient hospital setting in relationship to RLAI adherence in the outpatient setting is currently not defined. The objectives of this study are to (1) describe the adherence to RLAI following discharge from inpatient psychiatric hospitalization and (2) examine the predictors for outpatient RLAI adherence, especially factors related to discharge planning.

This is a retrospective study of patients receiving RLAI transitioning from an inpatient psychiatric state hospital to an outpatient community mental health center. Research was conducted at Western Missouri Mental Health Center (WMMHC) in Kansas City, Missouri, after approval from the University of Missouri-Kansas City Investigational Review Board and all outpatient sites included in the study. There are 6 outpatient community mental health centers in the surrounding Kansas City area; however, the majority of patients receiving inpatient psychiatric services at WMMHC receive outpatient psychiatric services from 2 of these facilities, Swope Health Services and Truman Medical Center Behavioral Health. Only individuals receiving outpatient psychiatric services from these 2 facilities were included in the study. All adult patients who received 

1 dose of RLAI during their inpatient stay and were discharged on RLAI from WMMHC to one of the community mental health centers listed above prior to June 30, 2008, were included in the study.

The primary end point is outpatient medication adherence to RLAI during the initial 3 months following discharge. The dates of RLAI administration at the community mental health center post-discharge were reviewed. Each patient should receive 6 doses of RLAI in total during the 3-month time frame, and the number of doses the patient actually obtained was used to assess adherence. A patient was considered as fully adherent if he or she obtained 6 doses.

Along with the number of doses obtained, the following variables were collected for each patient: age, gender, ethnicity, outpatient facility (Truman or Swope), length of stay since index admission, admission status (voluntary or involuntary), Axis I diagnosis, history of substance abuse, history of medication nonadherence, when RLAI was initiated (while inpatient or prior to admission), RLAI dose (25 mg, 37.5 mg, or 50 mg), discharge disposition (family, group home, self, or shelter), number of previous admissions, presence of a case manager, presence of a guardian, and establishment of an aftercare appointment within the time frame to receive the next RLAI dose. The last 3 variables reflected the degree of discharge planning. The data were collected using several different strategies. For Medicaid patients, a review of the Missouri Medicaid database was conducted. A manual review of institutional pharmacy and medical records was completed for all patients. Additionally, healthcare providers at the community mental health centers were contacted if needed.

Bivariate correlations were conducted for all variables. A linear regression analysis was performed where the number of doses obtained was the dependent variable. The independent variables included 2 groups of variables: (1) gender, outpatient facility, length of stay since index admission, and when RLAI was initiated, because their correlation coefficients with the dependent variable were not <0.2; and (2) presence of a case manager, presence of a guardian, and establishment of an aftercare appointment to assess the impact of discharge planning on adherence. In addition to these analyses, the reasons for discontinuation of RLAI and rate of rehospitalization were also explored.

A total of 126 patients were identified as being on RLAI during inpatient admission since the medication was approved in 2003; 73 patients were excluded due to various reasons which included: not referred to one of the included outpatient mental health facilities (n = 47), not discharged on RLAI (n = 10), and unable to access medical records (n = 16). A total of 53 patients met inclusion criteria and were included in the study. Patient characteristics are shown in 

. The majority of patients receiving RLAI were male (n = 40, 75%), African American (n = 36, 68%), had an Axis I diagnosis of paranoid schizophrenia (n = 37, 70%), and were involuntarily admitted (n = 32, 60%) to the inpatient psychiatric hospital due to acute psychosis. It is important to note our study population is reflective of people with serious and persistent mental illness from an urban setting.

Overall, 24% (13/53) of patients were considered fully adherent (6/6), as they received all 6 injections during the 3-month time frame, and 34% (18/53) were considered 80% adherent (5/6), as they received 

5 injections during the 3-month time frame. Sixty percent (32/53) of patients were initiated on RLAI and received their first injection while inpatient; more than half of these patients (19/32, 59%) did not follow up to receive even 1 injection in the outpatient setting. Out of the total number of patients included in the study, 40% (21/53) failed to receive a single dose of RLAI in the outpatient setting following discharge from the hospital; 19 of the 21 patients (90%) who failed to follow up were started on RLAI inpatient in an attempt to improve medication adherence in the outpatient setting.

The results of the regression are shown in 

. The regression model explained 11% of the variance in doses obtained. Among the independent variables, when RLAI was initiated was the strongest predictor for doses obtained (standardized beta = 0.34) and was the only significant predictor (

 <.05). Compared with patients who initiated RLAI while inpatient, patients who initiated RLAI before admission obtained approximately 2 more doses of RLAI, controlling for the other independent variables. None of the variables related to discharge planning, independently or collectively, were a significant predictor.

Documentation regarding reason for RLAI discontinuation was provided for only 15% (8/53) of patients and included patient refusal (n = 4), lack of follow-up (n = 3), and lack of efficacy (n = 1). A total of 15% of patients (8/53) included in the study were readmitted within 3 months of discharge. There is a possibility that patients were admitted to a hospital outside of our tracking ability via hospital records and the Medicaid database.

In this study, the rate of adherence to RLAI in the outpatient setting following discharge from inpatient psychiatric hospitalization was 24% or 34%, depending on the threshold set for adherence. That is, 24% of patients were considered adherent when the threshold was set to be 100% adherent, and 34% of patients were considered adherent when the threshold was set to be 80% adherent. Accordingly, the rate of nonadherence was 66% to 76%. This is comparable to nonadherence rates for oral antipsychotics found in the literature, though estimates of medication nonadherence vary significantly depending on each particular study’s definition of nonadherence, the study population, and methods for assessing adherence.

The use of long-acting injectable antipsychotics does not eliminate medication nonadherence but does address some of the issues associated with nonadherence, such as eliminating the need for daily dosing and identifying covert noncompliance.

 Most of the evidence for depot antipsychotics (eg, haloperidol, fluphenazine) indicates that their use is associated with increased medication adherence and a reduction in relapse rates.

 Low rates of discontinuation were found in RLAI randomized controlled trials.

One study reported a discontinuation rate of approximately 8% over 12 weeks

and another study reported a discontinuation rate of 35% over 12 months.

 The adherence rates derived from randomized controlled trials with strict inclusion and exclusion criteria may not be applicable to our study population.

The focus of most of the outcomes literature involving RLAI is the evaluation of healthcare resource utilization.

 In order to be included in these prospective and retrospective studies, patients had to have received a minimum number of injections, usually 

4 injections, in the outpatient setting. Many of these studies have shown improved medication adherence and reduced healthcare costs associated with RLAI compared with oral antipsychotic therapy or the time period prior to RLAI. But once again, these studies may not be reflective of RLAI in clinical practice as they do not include all patients initiated on RLAI (ie, those receiving <4 injections). Additionally, these studies were conducted in either a Veterans Affairs patient population or outside the United States, which may not be reflective of RLAI use in community mental health centers in the United States due to differences in healthcare access and delivery. European countries tend to mimic the Veterans Affairs closed system of care, in which there is potential for better follow-up and management of care.

There are limited data evaluating adherence to RLAI in a naturalistic setting. There are no published studies evaluating RLAI adherence following hospital discharge. It is a common occurrence for a long-acting injectable antipsychotic to be initiated in the inpatient setting with the intention that it will improve medication adherence in the outpatient setting following discharge. Evaluation of the effectiveness of this practice is needed to guide clinical practice.

Our study found that 40% (21/53) of the study population failed to receive RLAI in the outpatient setting following hospital discharge; 19 of these patients (90%, 19/21) were initiated on RLAI in the inpatient setting. This is in comparison with only 2 patients who were receiving RLAI prior to admission and failed to receive RLAI following hospitalization. The regression analysis identified RLAI initiation (inpatient or prior to admission) as a significant predictor for medication adherence. These findings question the effectiveness of the practice of initiating RLAI in the inpatient setting in an attempt to increase outpatient medication adherence. There are multiple factors which likely impact the effectiveness of such practice, including issues surrounding formulary guidelines, expense, lack of continuity in care between inpatient and outpatient settings, and poor medication reconciliation.

There are no published studies evaluating RLAI adherence during a 3-month time frame, regardless of whether initiation was in the inpatient or outpatient setting. The 3-month time frame following hospital discharge is critical because, during this vulnerable period, the patient may still be undergoing stabilization and, in some cases, is still being transitioned from oral risperidone to RLAI. Using strict criteria to define nonadherence (ie, received 6 doses of RLAI in a 3-month time frame), we found that 76% of patients were nonadherent to RLAI following hospital discharge, which is high by any standard. Using less stringent criteria, defining adherence as receiving >80% of medication (eg, 

5 RLAI doses in a 3-month time frame), nonadherence to RLAI therapy was found to be 66%, which still reflects a significant degree of nonadherence. Three studies have reported RLAI nonadherence rates at 6 months to be between 41% and 52.8%.

 Our rate of nonadherence to RLAI therapy at 3 months exceeds what has been reported in the literature for 6 months. Our 3-month nonadherence rate of 66% is more comparable to a 12-month discontinuation rate of 67.6%, which was reported in 1 study.

 This 12-month study included a description of the attrition rate, which was reported to be constant, suggesting a significantly lower nonadherence rate at 3 months than that reflected at 12 months. A comparison of our data with these previous studies is difficult as the time frames and patient populations are different. Our study included inpatients initiated or continued on RLAI then discharged to the outpatient setting, whereas other studies have included both inpatients and outpatients initiated on RLAI or outpatients only.

The nonadherence rate to RLAI observed in this study is higher than expected and may be due to several reasons attributed to our study population. As stated above, we included only patients with a recent inpatient psychiatric hospitalization. Our patient population may be less psychiatrically stable than other RLAI study populations. Although we did not evaluate treatment resistance, the fact that we included only those patients recently hospitalized may select for a more difficult-to-treat population compared with other studies. More than half of our study population was involuntarily admitted to the hospital, which is reflective of their lack of insight and resistance to treatment and likely carries over into the outpatient setting following hospital discharge. Additionally, the majority of our population were either actively abusing substances or had a history of substance dependence with alcohol, marijuana, or cocaine, a known barrier to medication adherence.

To our knowledge, there is only 1 other study evaluating factors or predictors associated with medication adherence in patients receiving RLAI.

 This study reported that continuation of RLAI at 6 months was more likely to occur in patients aged 

55 years and in those receiving RLAI 37.5 or 50 mg. The lack of association between discharge planning factors and RLAI adherence in our study may again be a reflection of our study population or could be attributed to the small sample size.

The rehospitalization rate in this study was 15%. The short duration of this study is optimal for evaluation of medication nonadherence during the transition from an inpatient to outpatient setting; however, it is insufficient in evaluating secondary outcomes, such as rehospitalization due to the occurrence of delayed relapse, which can occur several months following discontinuation of antipsychotic therapy.

Due to the retrospective nature of the study, many limitations exist which could potentially affect the outcome of the study. Although several databases were used for data collection to improve the accuracy of the data, relying on documentation for evidence of drug administration and other information is not optimal. Limiting data collection to 2 outpatient community mental health centers may reflect adherence rates that are affected by policies associated with the use of RLAI and clinicians’ perceptions of who is an appropriate candidate for RLAI treatment. The policies and clinicians’ perceptions at the 2 centers may differ from other community mental health centers. The small sample size may also have limited our ability to detect an association between discharge planning factors and medication adherence.

Notwithstanding these limitations, our study has several strengths. This was a retrospective study conducted as an evaluation of normal clinical practice. Thus, bias due to clinician awareness of the presence of a prospective study did not affect treatment decisions. We included patients consecutively and did not exclude patients for any clinical reasons (eg, treatment resistance, history of medication nonadherence, comorbid medical or psychiatric conditions, concomitant medications, etc), thus reflecting normal clinical practice. It is the only study to evaluate RLAI adherence at the 3-month time point following hospital discharge and the second study to assess predictors associated with RLAI adherence.

Our study found that patients who initiated RLAI prior to admission obtained more doses than those who initiated while inpatient. In addition, adherence to RLAI in the outpatient setting following discharge from inpatient psychiatric hospitalization was less than optimal and was extremely poor for those patients initiated on RLAI in the inpatient setting. This study emphasizes the continued need to discuss patient acceptance of risperidone with regard to route of administration in a collaborative setting. It also highlights the importance of communication, including accurate medication reconciliation, between inpatient and outpatient healthcare providers.

Articles

Adherence to risperidone long-acting injection (RLAI) therapy in the outpatient setting was poor. Degree of discharge planning from inpatient hospitalization was not associated with RLAI adherence.