Feasibility and Preliminary Effects of Implementing Acceptance and Commitment Therapy for Inpatients with Psychotic-Spectrum Disorders in a Clinical Psychiatric Intensive Care Setting

Abstract

The aim of this study was to adapt Acceptance and Commitment Therapy for Inpatients (ACT-IN) for implementation in a typical hospital setting to prepare for a larger clinical trial. The sample consisted of 26 inpatients diagnosed with psychotic-spectrum disorders. Using an open trial design, patients received individual and group ACT-IN sessions during their stay. We assessed the feasibility/acceptability of ACT-IN and preliminary changes on patient outcomes at baseline, discharge, and 4 month follow-up. We successfully recruited and retained participants as planned. Patients reported satisfaction with treatment, and routine hospital staff showed adoption and fidelity to the intervention. Relative to baseline, patients demonstrated significant improvements in symptoms and functioning across the 4 month follow-up. The current study shows that ACT-IN may be feasible and acceptable for inpatients with psychotic disorders in a psychiatric intensive care setting and should be tested in a future effectiveness-implementation trial.

INTRODUCTION

Psychotic disorders (PDs), including schizophrenia and psychotic mood disorders, are disabling illnesses present in approximately 3.5% of the general population (Perala et al., 2007). PDs are considered some of the most disabling disorders and accrue some of the highest healthcare costs across psychiatric diagnoses, with total costs of 155.7 billion USD annually (Cloutier et al., 2016). The high economic costs associated with PDs are partially a result of recurrent inpatient admissions (Knapp, 2005). Among patients hospitalized for psychiatric reasons, PDs represent up to 38% of cases (Wheeler, Robinson, & Robinson, 2005). In addition, over 50% of PD patients are rehospitalized within one year of a hospital discharge (Schennach et al. 2012).

Hospitalization can serve an important purpose in providing management and stabilization during acute phase of illness. Nevertheless, it also can cause difficulties for patients and their families due to the disruption to their living situation and daily functioning stemming from removal from the community (Sharfstein, 2009). Furthermore, the post-discharge period is characterized by increased risk for an array of negative outcomes, including suicide (Jencks, Williams, & Coleman, 2009; Popovic et al., 2014). In addition, the recovery rate of discharged patients with PDs remains poor, with only 22% of patients reaching functional and symptomatic recovery by one year post-discharge (Ventura et al., 2011).

Acute hospitalization represents a crucial time to teach patients skills that they can use to better manage symptoms and improve their functioning following discharge to prevent negative outcomes and to facilitate sustainable recovery. However, a dearth of systematic research exists on the effectiveness of inpatient treatment and strategies for reducing rehospitalizations, particularly in the US (Jacobsen, Hodkinson, & Peters, 2018; Loch, 2014), highlighting the need for easily-implemented, feasible, and empirically-based interventions that can be effectively delivered on inpatient units to improve outcomes. Current interventions for PDs focus almost exclusively on the post-hospital period as a means of preventing rehospitalization (Pitschel-Walz, Leucht, Bäuml, Kissling, & Engel, 2001). Given the escalating costs to the healthcare system resulting from potentially preventable recurrent rehospitalizations (Heslin & Weiss, 2015), there exists an urgent need for psychosocial treatments during hospitalization that address functional outcomes, teach relapse skills, treat associated mood symptoms, and promote post-discharge treatment adherence.

ACT for PDs

Acceptance and commitment therapy (ACT; Hayes, Strosahl, & Wilson, 1999) is part of the family of cognitive-behavioral interventions. ACT emphasizes psychological acceptance and mindfulness of distressing thoughts and feelings as an alternative to avoidance strategies that impede functioning and goal attainment. ACT has been applied to a variety of clinical populations, and randomized controlled trials have demonstrated its efficacy for treating common conditions such as depression, anxiety, substance abuse, and behavioral health problems. For example, A-Tjak et al. (2015) recently published a meta-analysis of 39 controlled trials of ACT for psychological and physical health problems and reported that the intervention was superior to control conditions at post-treatment and follow-up (Hedges’ g = 0.57) and similar in efficacy to other cognitive-behavioral approaches. In addition, Khoury, Lecomte, Gaudiano, & Paquin (2015) conducted a meta-analysis of mindfulness, acceptance, and compassion-based approaches for PDs (including ACT) and found an effect size improvement of Hedges’ g = 0.41 (n = 7) for primary outcomes at post-treatment compared with control conditions.

Cognitive-behavioral therapies (CBT) have been found to be efficacious for treating PDs, but less research has focused on short-term CBT for this population, such as when treating patients with PDs during acute hospitalizations typically lasting only one week (Hazell, Hayward, Cavanagh, & Strauss, 2016). Initial research indicates that ACT is one promising CBT approach for this population. Bach and Hayes (2002) conducted the first randomized trial of ACT for inpatient with PDs (n = 80) and demonstrated that individuals who received ACT during their hospital stay were significantly less likely to be hospitalized at 4-months post-discharge compared to those receiving hospital treatment as usual (TAU). Gaudiano and Herbert (2006) replicated and extended the findings of Bach and Hayes by showing that ACT provided to inpatients with psychotic-spectrum disorders (n = 40) produced clinically significant improvements in symptoms and functioning by hospital discharge, and resulted in a longer time to rehospitalization over 4 months compared to those receiving TAU (as reported in Bach, Gaudiano, Hayes, & Herbert, 2012). Most recently, Tyrberg, Carlbring, & Lundgren (2016) conducted a small randomized trial (n = 22) in Sweden showing significant effects of ACT relative to TAU on reducing risk of rehospitalization in inpatients with PDs. Additional evidence of ACT’s efficacy for treating psychosis has been reported from trials conducted in outpatients with PDs (Gumley et al., 2017; Johns et al., 2015; Shawyer et al., 2012, 2017; White et al., 2011). Limitations of previous research on ACT for PDs included the lack of focus on implementation and sustainability of the intervention. Therefore, a research-practice gap exists, such that ACT for PDs is still not widely used in inpatient facilities, due to lack of staff training in the approach and the absence of guidelines for how to integrate it with typical inpatient milieu programming.

Rationale for the Current Project

Whereas efficacy research focuses on methodology that enhances the internal validity of findings under tightly controlled conditions, effectiveness studies favor designs that maximize external validity to better inform real-world practices. The previous studies by Bach and Hayes (2002) and Gaudiano and Herbert (2006) were efficacy studies, in which ACT was provided in an individual format alone, delivered by research interventionists, and provided in addition to routine inpatient care that ultimately was not designed to be sustainable. In contrast, the aim of the current treatment development study was to implement and examine the effectiveness of ACT for PDs as part of a model of care in a real-world acute care hospital setting. Thus, we modified ACT in the current project by: 1) incorporating a multi-modal format that better mirrors typical inpatient programming (individual and group sessions), 2) training inpatient clinicians who typically work in these settings (e.g., social workers, nurses, occupational therapists, activities therapists) to deliver the treatment, and 3) integrating the treatment into other inpatient milieu services. Our longer-term goal was to test the real-world effectiveness of ACT for PDs compared with routine hospital care in order to assess the potential for a future adequately-powered randomized controlled trial.

The first step in this process was to refine and test the ACT for inpatients (ACT-IN) protocol and to develop appropriate procedures and methods to facilitate the conduct of a larger effectiveness-implementation trial. We hypothesized that ACT-IN would be deemed feasible/acceptable based on recruitment/retention goals and satisfaction ratings by patients and hospital staff. Compared to baseline, we also hypothesized that inpatients with PDs would show significant improvements (compared to baseline) through a 4-month follow-up on indices of symptoms, functioning, quality of life, and processes targeted by the intervention (acceptance and mindfulness).

METHOD

Participants

The current study was part of the Research the Effectiveness of Acceptance-based Coping (REACH) project and methods are reported in more detail elsewhere (Gaudiano et al., 2017). Participants were recruited during an index hospitalization at a private psychiatric hospital in the Northeastern U.S. Inclusion criteria were: 1) currently hospitalized (voluntary status); 2) DSM-5 diagnosis of a psychotic-spectrum disorder, including schizophrenia, schizoaffective disorder, schizophreniform, delusional disorder, psychotic disorder unspecified, or primary major depressive/bipolar disorder with psychotic features, based on the Structured Clinical Interview for DSM-IV modified for DSM-5 criteria (SCID; First, Spitzer, Gibbon, & Williams, 2002); 3) 18 years or older; and 4) ability to read and speak English. Exclusion criteria were: 1) severe psychosis/safety risk such that patients were unable to participate in unit programming as determined by their treatment team, 2) psychosis due to a general medical condition or substance-induced psychotic disorder, or 3) severe cognitive impairment based on the Mini-Mental State Exam (MMSE < 15; Folstein, Folstein, & McHugh, 1975).

Procedure

This study was approved by the Butler Hospital Institutional Review Board (IRB). Written informed consent was obtained by research staff after permission to approach the patient was obtained from the attending physician. For those who consented, baseline assessments were administered shortly thereafter (usually within the first couple of days after admission), immediately followed by provision of ACT-IN group and individual therapy. Individual therapy was delivered in a private room on the inpatient unit by a master’s level therapist. Patients also attended ACT-IN groups over the course of their inpatient stay, which were delivered multiple times weekly on the unit, provided by various inpatient staff, including activities therapists, nurses, and occupational therapists. The number of ACT-IN sessions completed depended on the duration of inpatient stay. ACT-IN was offered in addition to hospital TAU.

All interviews were audio recorded for supervision purposes. SCID results for new participants were reviewed with the first author (B.G.) to determine the diagnosis and to ensure interrater reliability. Research staff implemented a discharge assessment shortly before patients left the hospital. If the assessment was unable to be completed prior to discharge, it was scheduled shortly thereafter. Patients also were asked to return to the hospital to complete a four-month follow-up. In the interim period between discharge and follow-up, research staff monitored inpatient rosters to identify potential rehospitalizations or other adverse events. Participants were compensated monetarily upon completion of each stage of the study.

Measures

At baseline, patients were administered the MMSE to screen for cognitive functioning, and the SCID, a semi-structured clinical interview, to determine DSM-5 psychiatric diagnosis. At hospital baseline, discharge, and 4 months post-discharge, we assessed multiple domains of interest. Symptom severity and overall clinical improvement were assessed using the interviewer-rated 18-item Brief Psychiatric Rating Scale (BPRS; Overall & Gorham, 1962) as our primary outcome (total score). We also examined the BPRS subscales (Positive, Negative, Disorganized, and Mood) as secondary outcomes (Long & Brekke, 1999). Assessments were conducted by trained raters whom obtained initial interrater reliability >.80 on interview measures (e.g., SCID, BPRS) prior to assessing study participants.

Additionally, the 34-item, self-report Clinical Outcomes in Routine Evaluation (CORE; Barkham et al., 1998) was used as a secondary measure of patient-centered improvement (symptoms, well-being, functioning, risk). Additional outcomes also included quality of life, as assessed using an interviewer-rated and abbreviated version of the Quality of Life Scale (QLS; Bilker et al., 2003), along with the self-report Schizophrenia Quality of Life Scale (S-QoL; Boyer et al., 2010). In addition, psychosocial functioning was assessed using the 12-item self-report version of the World Health Organization Disability Assessment Schedule 2.0, which assesses cognition, mobility, self-care, social interactions, life activities, and community participation (WHODAS 2.0; Andrews, Kemp, Sunderland, Von Korff, & Ustn, 2009). Mechanisms targeted by the intervention were assessed using the following self-report measures: Acceptance and Action Questionnaire-II (AAQ-II; Bond et al., 2011) as a measure of psychological flexibility, Cognitive and Affective Mindfulness Scale-Revised (CAMS-R; Feldman, Hayes, Kumar, Greeson, & Laurenceau, 2007) as a measure of mindfulness, and Valuing Questionnaire (VQ: Smout, Davies, Burns, & Christie, 2014) as a measure of values-action consistency. Treatment adherence was determined using the interviewer-rated Brief Adherence Rating Scale (BARS; Byerly, Nakonezny, & Rush, 2008) which assesses the percentage of antipsychotic medication doses prescribed versus taken over the preceding month. At hospital discharge, patients were assessed for level of satisfaction with ACT-IN using the self-report Client Satisfaction Questionnaire-8 (CSQ-8; Larsen, Attkisson, Hargreaves, & Nguyen, 1979).

Following an ACT-IN training workshop, hospital staff were administered a 12-item survey assessing their attitudes (1 = completely disagree to 5 = completely agree) toward ACT (e.g., appropriateness, helpfulness, ease of use, willingness to use, confidence) adapted from previous research (Scott, Montesinos Marin, Gaudiano, & McCracken, 2017). Also, an ACT Knowledge Questionnaire was administered to therapists after ACT training. This questionnaire was adapted from the ACT Core Competency Self-Rating Form (Luoma, Hayes, & Walser, 2007) and consisted of 10 true/false questions (e.g., “The therapist helps the patient to commit to the outcomes to be achieved while deemphasizing the process used to achieve the outcome.” Correct answer: False). The ACT-IN fidelity scale was designed for the study based on the treatment manual. Items included topics that were to be covered in each session as well as topics that were not consistent with the intervention (e.g., direct confrontation, cognitive restructuring). For each item, adherence was rated as yes/no and competence was rated on a Likert scale from 0 = Poor to 6 = Excellent.

Acceptance and Commitment Therapy for Inpatients (ACT-IN)

The ACT-IN treatment manual was adapted from the protocols used in the Gaudiano and Herbert (2006) and Bach and Hayes (2002) studies. The following changes were made to the protocol in the current study compared to previous versions: a) adding in corresponding group sessions to the original individual only format, b) adapting existing inpatient therapy techniques being used by hospital staff to illustrate and teach ACT concepts, and c) enhancing treatment manual detail and providing sample therapist scripts to offer more guidance for master’s level compared with the original doctoral-level providers. The intervention used an open-enrollment format for therapy groups so that patients could begin and end sessions as their length of stay dictated, with a target of three individual and three group sessions for the typical 1-week stay. As in the original protocols, the number of ACT-IN sessions varied based on the length of hospital stay as done in our previous studies.

Individual Therapy.

Session #1 was designed to introduce the patient to ACT. The therapist reviewed the patient’s history, built rapport, assessed avoidance behaviors and associated impairment to highlight motivation for change, clarified the individual’s deeply held values, and set initial treatment goals consistent with those values. Each subsequent session ended with a further discussion of values and review of progress toward goals. Furthermore, a discussion of post-discharge planning and relapse prevention was also incorporated across sessions. Next, Session #2 focused on the theme of mindfulness. The therapist introduced and practiced various mindfulness exercises (e.g., mindful eating and walking activities) with patients that were specifically adapted and tailored for those experiencing acute psychotic symptoms. Session #3 emphasized building acceptance and willingness to experience unavoidable distress in the service of values. Various acceptance-based metaphors (e.g., Tug of War with a Monster, in which the ultimate futility and cost of struggling with internal experiences is illustrated) and experiential exercises (e.g., finger trap device, which highlights how avoidance leads to more and not less distress over time) were used to reinforce concepts. Session #4 introduced the concept of defusion, which focuses on decreasing the dominance of language and cognition over behavior by labeling thoughts as thoughts and feelings as feelings. In addition, this session focused on building a stable sense of self (self-as-context vs content), which involves shifting perspective to one’s stable sense of self that is separate from one’s momentary thoughts and feelings. Using illustrative methods (e.g., “I’m noticing that I’m having the thought that…” exercise), defusion and self-as-context highlighted the transient and fluctuating nature of mental events. If psychotic symptoms were present during the session, the therapist encouraged the patient to practice mindful awareness of these experiences, accept any accompanying distress that the person might be feeling in the moment, and continue session participation in the service of values.

Group Therapy.

Each group format began with a brief explanation of ACT, and then focused on practicing one core ACT process using a variety of exercises. Group strategies often were adapted from intervention strategies already being used by staff in other inpatient programming, and then specifically modified to be consistent with ACT. Mindfulness exercises practiced in the groups included those being used in the individual sessions (e.g., mindful eating), but also were expanded to include additional strategies. For example, patients practiced mindfulness of smell using scented lotions or sound using a sound machine, adapted from other sensory de-escalation approaches being used on the unit. Other groups emphasized themes of acceptance, defusion, values, and self-as-context. For example, patients constructed personal collages from various materials (e.g., cut out words and pictures from magazine articles) to foster defusion from their negative thoughts and feelings and to explore self-as-context vs content. In addition, patients participated in values-building activities that involved explaining their personal values to the group and setting consistent goals (e.g., 80^th birthday exercise). One values exercise involved patients passing around a beach ball to each other with various values written on it (e.g., family, friends, recreation, work, learning) and describing consistent actions that were relevant to them.

ACT-IN Therapist Training and Supervision

ACT-IN therapists included master’s level counselors, social workers, nurses, activities therapists, and occupational therapists who worked at the hospital. Training was first provided in a didactic training workshop, in which investigators presented training materials, roleplayed various strategies, and answered questions. Supervision was held with therapists on the unit weekly by the investigators to provide ongoing training and to prevent drift from the protocol. Treatment fidelity scales developed for ACT-IN were used to assess therapist adherence to the protocol. Individual treatment sessions were audio-recorded for supervision purposes. We were unable to record ACT group sessions due to confidentiality reasons because they also contained non-study patients. Instead, novice therapists co-led groups with a more experienced supervisor, and some groups were directly observed by investigators for adherence monitoring and supervision purposes.

Inpatient Treatment as Usual (TAU)

All patients also received inpatient TAU, which included medication management sessions with a psychiatrist, group therapy, case management, occupational therapy, open air walks, discharge planning, and medical treatment as indicated. Discharge plans included an appointment with a medication provider scheduled shortly after leaving the hospital. Patients were provided with a follow-up medication management appointment within the first month post-discharge and referrals for other follow-up treatment as needed (e.g., individual therapy, case management). The average length of hospital stay was 5 days but varied (a few days to several weeks) based on individual needs.

Statistical Analyses

Multilevel modeling (MLM) in IBM SPSS 22.0 (SPSS, 2010) was used to examine trajectories of change along three time points (baseline, discharge, and 4-month follow-up). MLM is ideal for use with small sample sizes and accommodates for missing data, instead of using listwise deletion or other method of imputation. MLM takes into account within person variability, modeling the effects more accurately. A −2 log likelihood (−2LL) test justified the use of the current model. Using a restricted estimated maximum likelihood (REML) estimator approach, random intercept, and heterogeneous first-order autoregressive covariance structure for repeated measures, we modeled the fixed effects for the trajectory of change in all indices including symptomatology, functioning, and quality of life using intent-to-treat participants. A linear trajectory was tested for all analyses to examine the hypothesis that outcomes improved (relative to baseline) through 4 month follow-up. Log10 transformations were used to correct for skewness for continuous variables when necessary. Alpha level was set at p < .05. The following regression equation models the prediction of treatment outcomes as a function of time at level 1:

$$\text{Yij}={\pi }_{0\text{j}}+{\pi }_{1\text{j}}\left({\text{Time}}_{\text{ij}}\right)+{\text{e}}_{\text{ij}}$$

Yij represents the participant j’s outcome for assessment i; π_0j represents individual j’s intercept; π_1j represents the slope of j at the time of assessment i; and e_ij represents the error term.

RESULTS

Sample Characteristics

See Figure 1 for subject flow diagram. In total, we screened 285 hospital charts, approached and assessed 51 patients with chart diagnoses of PDs for potential study eligibility. A total of 36 of 51 (71%) who were approached for the study agreed to participate, suggesting a high degree of interest in the intervention. After excluding 10 patients who did not meet study criteria, we enrolled 26 patients in the study. We were able to recruit 3–4 patients per month into the study, which allowed us to exceed our initial recruitment goal of 20 patients total.

Figure 1.

Subject Flow Diagram

See Table 1 for a detailed description of participant demographics. All patients had been psychiatrically hospitalized two or more times previously, suggesting high levels of chronicity and impairment. We achieved acceptable levels of diversity in the sample, including an average age of 38 (SD = 12.5), 62% female, 15% Latino/Hispanic ethnicity, 69% White/Caucasian, 19% Multiple Races, and 12% other races (Asian, Native American, African-American/Black). Fifty percent of patients reported receiving physical or psychiatric disability compensation and only 23% were working part- or full-time. Income ranged from $0 - $59,999 US, with the majority (69%) earning under $20,000. The majority (58%) reported being single or never married. The sample completed an average of 12.9 years of education (slightly more than high school level). According to the SCID, 12% had schizophrenia, 35% had schizoaffective disorder, 15% had major depressive disorder with psychotic features, and 38% had unspecified psychotic disorder.

Table 1.

Demographics and Descriptive Statistics

Variables	n (%) or M (SD)
Age (years)	37.9 (12.5)
Gender (female)	61.5 (16)
Education Level (years completed)	12.9 (2.3)
Employment
Disabled	50 (13)
Employed (part/full-time)	23.1 (6)
Unemployed/Student	11.5 (3)
Other (retired, homemaker, student)	15.4 (4)
Latino/Hispanic	15.4 (4)
Race
Caucasian/White	69.2 (18)
African American/Black	3.8 (1)
Native American	3.8 (1)
Asian	3.8 (1)
Multiple Races	19.2 (5)
Annual Income ($ US)
0–19,999	69.2 (18)
20,000–39,999	19.2 (5)
40,000–59,000	7.7 (2)
Did not respond	3.8 (1)
Marital Status
Single/never married	57.7 (15)
Married	11.5 (3)
Divorced/separated	30.8 (8)
Diagnosis
Schizophrenia	11.5 (3)
Schizoaffective Disorder, Depressive	19.2 (5)
Schizoaffective Disorder, Manic	15.4 (4)
Psychotic Disorder Unspecified	38.5 (10)
Major Depression with psychotic features	15.4 (4)
No. Previous Psychiatric Hospitalizations	18.6 (24.6)
Current Electroconvulsive Therapy (Yes)	7.7 (2)

Note. n = 26.

Treatment Fidelity

Treatment fidelity ratings were conducted by an independent, master’s level clinician who had previous training in ACT and experience treating patients with psychotic disorders. First, the fidelity rater and a study investigator co-rated sample session recordings until reaching over 90% agreement. The fidelity rater then completed adherence ratings for 10 randomly selected sessions (approximately 15% of all sessions). Results showed 100% adherence to topics in the ACT-IN protocol across sessions, 0% inclusion of non-ACT consistent content, and an average session competency rating (0–6) of 4.7 (SD = 0.67), suggesting high levels of fidelity.

Feasibility/Acceptability

In terms of the ACT-IN intervention, the mean number of individual sessions received was 2.9 (SD = 0.9) and the mean number of group sessions received was 2.7 (SD = 2.4), which was close to our goal of delivering 3 individual and 3 group sessions each. The mean CSQ-8 score at the discharge assessment was 29.5 (SD = 3.8) out of a possible score of 32, indicating high treatment satisfaction. In addition, 8 hospital therapists completed a satisfaction survey (1–5 scale) following training in ACT-IN. Again, results showed that staff found ACT-IN to be acceptable and the training helpful in preparing them for the intervention with a mean of 4.55 (SD = 0.15) out of 5. The ACT Knowledge Test was also administered after initial training and therapists received an average score of 76.3% out of 100%, suggesting adequate knowledge of ACT concepts.

Clinical Outcomes

Although the study was primarily designed to assess the feasibility and acceptability of ACT-IN, preliminary treatment effects are listed in Table 2 to help identify the potential clinical benefits of the intervention and targets for further study. Given the lack of a control group, we were mainly concerned with the potential for patients to remain significantly improved through 4 months post-discharge relative to hospital baseline, on various clinical indices to better understand any enduring effects of treatment. Significant time effects from baseline through 4-month follow-up were seen on our primary outcome (BPRS-Total), as well as for each BPRS subscale (positive symptoms, negative symptoms, mood symptoms, disorganization). In addition, significant time effects were observed for our secondary outcomes, including self-reported symptoms (CORE), interviewer-rated and self-reported quality of life (QLS and SQoL, respectively) and psychosocial functioning (WHODAS). Finally, significant time effects were observed on our hypothesized mechanisms of action, including psychological flexibility (AAQ), mindfulness (CAMS-R), and values consistent living (VQ). Effect sizes demonstrated generally large within-subject improvements from baseline to hospital discharge. However, we noted some degree of attenuation of effects on certain measures from baseline through 4 month follow-up, as would be anticipated for a brief, inpatient only intervention.

Table 2.

Time Effects on Outcomes using Multilevel Modeling from Baseline through 4 Months Post-Hospital Discharge

Measures	Baseline M (SD)	Discharge M (SD)	Follow-up M (SD)	Time Effect F-value (dfs)	Cohen’s d (95%CI): BL to DC	Cohen’s d (95%CI): BL to 4 mos.
Primary Outcome
BPRS-Total	51.0 (11.0)	35.3 (11.3)	40.6 (11.8)	25.84*** (2,44.6)	1.41 (.78,1.99)	.91 (.33,1.47)
Secondary Outcomes
BPRS-Positive	14.3 (5.9)	8.9 (5.0)	11.2 (6.6)	19.28*** (2,22.8)	1.08 (.48,1.60)	.53 (−.03,1.07)
BPRS-Negative	2.8 (.64)	2.6 (.48)	2.5 (.43)	3.48* (2,29.3)	.35 (−.20,.90)	.54 (−.01,1.10)
BPRS-Mood	20.5 (5.5)	12.3 (5.7)	15.4 (6.6)	22.88*** (2,25.9)	1.46 (.83,2.05)	.84 (.26,1.39)
BPRS-Disorganized	.2 .08)	.3 (.08)	.3 (.08)	7.01** (2,23.8)	−.63 (−1.17,−.06)	−.63 (−1.17,−.06)
CORE	63.7 (26.6)	32.7 (25.1)	45.5 (28.5)	18.54*** (2,23.8)	1.20 (.59,1.77)	.66 (.09,1.21)
QLS†	3.5 (.94)	4.5 (.96)	3.9 (1.3)	10.60*** (2,25.9)	−1.05 (−1.62,−.46)	−.35 (−.89,.20)
SQoL†	80.4 (23.9)	92.3 (22.3)	86.1 (27.8)	5.35* (2,24.5)	−.51 (−1.06,.05)	−.22 (−.76,.33)
WHODAS 2.0	18.1 (10.0)	12.4 (9.2)	13.8 (10.6)	5.32* (2,23.7)	.56 (.00,1.11)	.41 (−.14,.96)
Potential Mechanisms
AAQ-II†	24.9 (11.9)	33.1 (10.8)	28.1 (11.1)	6.60** (2,23.3)	−.72 (−1.27 −.15)	−.28 (−.82,.27)
CAMS-R†	28.0 (7.5)	35.3 (8.0)	33.6 (9.0)	12.50*** (2,25.8)	−.94 (−1.50,−.35)	−.68 (−1.22,−.11)
VQ†	2.8 (1.2)	3.9 (1.3)	3.5 (1.5)	11.12*** (2,24.5)	−.88 (−1,43,−.30)	−.51 (−1.06,.04)

Notes: n = 26,

^†increased scores indicate improvement, BL = Baseline (Admission), DC = Discharge, BPRS = Brief Psychiatric Rating Scale, CORE = Clinical Outcomes in Routine Evaluation, QLS = Quality of Life Scale, SQoL = Schizophrenia Quality of Life Quality of Life, WHODAS 2.0 = WHO Disability Assessment Scale 2.0, AAQ-II = Acceptance and Action Questionnaire-II, CAMS-R = Cognitive and Affective Mindfulness Scale Revised, VQ = Valuing Questionnaire. Estimated marginal means and standard deviations from mixed effects analysis depicted and used for calculating effect size estimates.

Additionally, we examined rates of treatment response based on at least a 30% reduction of symptoms from baseline on the BPRS total score, which corresponds to an assessment of being at least “minimally improved” when linked to Clinical Global Impressions Ratings (Leucht et al., 2005). In addition, a more conservative assumption was made to count those with missing data as not meeting criteria for treatment response. Based on these criteria, a total of 50% (n = 13) of the sample showed a treatment response on the BPRS at hospital discharge and 35% (n = 9) at follow-up.

Follow-up

No patient withdrew from the study during hospitalization and 84.6% (n = 22) were able to be reached for 4 month follow-up, which exceeded our a priori goal of 80% retention. Based on the BARS at 4-month follow-up, average antipsychotic medication adherence (taken/prescribed) over the 4-month follow-up period was 80.3% (n = 25.2), suggesting that patients were generally adherent to their pharmacotherapy (although this was not objectively verified). Also at 4 month follow-up, 40.9% (n = 9) reported receiving adjunctive, community psychosocial treatment (e.g., group or individual therapy).

The only adverse events observed over the trial included psychiatric rehospitalizations, which were expected given the nature of the sample. No patients attempted suicide during follow-up. Based on treatment history interview and review of hospital records, 54% (n = 14) of the sample was rehospitalized over 4 month follow-up, which was consistent with the sample’s high frequency of previous hospitalizations. Rehospitalizations occurring within 30 days of discharge are considered more likely to be tied to the previous admission, and thus may be a better indicator of poor outcomes such as relapse or lack of treatment follow-up (Vigod et al., 2013). For comparison purposes, 30-day rehospitalization rates in the U.S. for PDs are 22.4% (Heslin & Weiss, 2015), whereas in the current study only 15.4% (n = 4) of the sample was rehospitalized during this time period. This suggests that ACT may have had some protective effect for rehospitalization like in previous studies; however, a clear cause-and-effect relationship cannot be ascertained due to lack of a control group in the current study.

DISCUSSION

The aim of this treatment development study was to adapt ACT-IN for implementation on a typical hospital unit and to assess its feasibility/acceptability and preliminary effects on clinical outcomes among inpatients with PDs. Results supported the feasibility and acceptability of ACT-IN as previously demonstrated in efficacy studies. We were able to recruit and retain sufficient numbers of patients in the study as planned. We also were able to train hospital staff to implement ACT-IN and to integrate this approach into the existing inpatient milieu. Treatment fidelity checks of individual sessions demonstrated that staff could reliably and effectively implement ACT-IN when working with inpatients with PDs. Hospital therapists and inpatients reported high degrees of satisfaction with ACT-IN and overall their feedback suggested that it addressed their needs and provided a useful approach for promoting recovery. Exploratory analyses suggested improvements in symptoms, functioning, quality of life, and potential target mechanisms that were observed from hospital baseline to 4 months post-discharge (although there was some loss of effects over longer-term follow-up as anticipated). These findings support the viability of this research program and the aim to test ACT-IN in a pilot randomized-controlled effectiveness trial as the next step. Feedback indicated that this project helped to improve staff skills in working with some of the most challenging patients in the hospital and that satisfaction with the model was essential in ensuring that quality treatment would continue after the project is finished. The project also served as a useful method to increase staff engagement, facilitate provider creativity, and promote interprofessional engagement.

Consistent with current findings, Gaudiano and Herbert (2006) showed significant improvements from hospital admission to discharge on measures of symptoms (including the BPRS), impairment, and ACT-related processes (e.g., believability of psychotic symptoms). The current study extended these findings by documenting sustained improvement through 4 months on most outcomes and process measures, across both self-reports and interviewer-ratings. In addition, patients showed improvement on a broad array of patient-centered and process variables, including quality of life, psychosocial functioning, valued living, and mindfulness. Within group effect sizes were generally in the medium to large range, suggesting that patients achieved clinically meaningful levels of change on these outcomes. However, measures showed some attenuation of effects by follow-up, as it would be expected that a brief inpatient treatment would show greatest impact at discharge with effects slowly decreasing over the proceeding months. For example, results on the measure of psychological flexibility (AAQ-II), which is considered a key mechanism of ACT, showed significant change at discharge, but most of the effect was lost by follow-up, suggesting that further intervention might be needed to maintain effects on this construct. Nonetheless, we would emphasize that it is difficult to disentangle the effects of ACT-IN from those of medications, other psychosocial interventions that patients may have been receiving, and the naturally fluctuating course of psychotic disorders that result in people being hospitalized at times of greatest severity with subsequent regression to the mean. As stated, the primary aim of the current project was to assess feasibility and acceptability of ACT-IN delivered by routine hospital staff to plan for a more definitive effectiveness trial later.

ACT-IN might be improved in the future by adding an aftercare element to the program to solidify short-term gains. For example, patients often experience a discontinuity of care following inpatient treatment and require increased support and contact to “bridge this gap” to prevent relapse and rehospitalization. Rehospitalization rates over follow-up were high (consistent with the sample’s previous hospitalization rate), but were lower than national samples in term of rapid rehospitalizations (i.e., within 30 days of discharge). Thus, future iterations of ACT-IN may benefit from the inclusion of brief (e.g., 1 month), phone follow-up contacts to help patients transition successfully outside of the hospital and maintain gains achieved at discharge. Increasingly, such telephone aftercare interventions are being used in other high risk psychiatric populations following hospital discharge with promising results (Miller, Gaudiano, & Weinstock, 2016). In addition to considering a post-discharge transitions component to the program, future iterations of ACT-IN may also benefit from a greater emphasis on developing a specific relapse prevention plan during the hospital stay.

As discussed in the implementation and dissemination literature (Forman, Olin, Goagwood, Crowe, & Saka, 2009), we found that having local clinician “champions” of the approach helped to foster increased buy-in and enthusiasm by other staff in terms of their interest in learning and delivering ACT-IN. Another lesson learned from our implementation efforts was that it worked best to train hospital therapists in ACT principles and then modify the existing inpatient groups in ways that were consistent with this approach. For example, many of the inpatient groups were being run by occupational therapists who had their own unique set of professional skills and background. Instead of training these clinicians to become traditional ACT “psychotherapists,” we modified existing group content that it also could be used to illustrate ACT concepts. For example, a previously used collage group was revised to target cognitive defusion and self-as-context. Given the overall flexibility and transdiagnostic nature of the ACT approach, it was relatively easy to identify ways of re-engineering milieu group content to fit specific ACT principles. This allowed us to capitalize on the existing strengths and experience of the hospital staff who have their own unique skillsets, yet to expand their psychotherapeutic repertoire to enable effective ACT delivery. Staff feedback suggested that ACT also provided a consistent case conceptualization and professional language to facilitate cross-disciplinary collaboration, which was especially valuable when working with patients who had significant clinical needs.

Additional challenges were faced implementing ACT-IN on a psychiatric intensive care hospital unit, but these were not unique to this project and typical of any attempt to disseminate empirically-supported therapies in real-world practice settings (Proctor et al, 2008). Staff variability and turnover created difficulties ensuring consistent ACT training and fidelity, and required frequent adjustments over time and re-trainings. Finding time for weekly staff supervision also was limited by hospital schedules and staff workloads. Patient-related barriers to implementation included those typically noted when working with inpatients, including their overall level of severity, chronicity, presence of cognitive deficits, effects of changing medication regimens (e.g., excessive drowsiness), and concerns related to safety (self and other harm) that impacted treatment delivery and study participation. We found it necessary to tailor ACT-IN to the unique needs of these patients (via the ACT values-based approach) and found that it was important to repeat and reinforce ACT concepts across multiple individual and group sessions to maximize impact.

Based on the open trial nature of this investigation, it is impossible to know if the changes observed were the direct result of ACT-IN or the general effects of competent treatment delivery. Only a subsequent randomized trial can determine this. Furthermore, clinical raters were not blind to treatment provision, and this could have biased their judgments and magnified effects. It should be noted, however, that there was general consistency of effects between interviewer ratings and self-report scales. In addition, patients received additional treatment modalities (e.g., pharmacotherapy, non-ACT therapy) as part of their inpatient and outpatient care that also could have influenced results (for better or worse depending on the quality of care received, much of which was outside the control of the study). Also, we believe that the individual therapy component of ACT-IN was essential to its effectiveness because it allowed patients to receive more personalized and tailored treatment. However, many inpatient units do not have the staff or resources to provide individual therapy to inpatients at the moment, and we do not know if the ACT-IN groups alone would have been as effective. Furthermore, not all inpatients with PDs qualified for the study, due to our exclusion criteria (e.g., severe cognitive impairment, involuntary admission status). Thus, caution should be made when attempting to generalize the current results to all types of inpatients with PDs. Finally, patients received a variable amount of ACT treatment based on their variable length of study. Sample size was too small to statistically examine the effect of treatment dose, but this should be an avenue for future larger-scale research. Also, since the mean number of individual sessions completed was 2.9 and our target was at least 3, we could consider reducing our ACT-IN individual session content to 3 sessions total, instead of 4 as currently written in the protocol.

The next step will be to attempt to further bolster the feasibility, acceptability, and effectiveness of ACT-IN by conducting a smaller pilot randomized controlled trial to compare its effects relative to a control condition and test the feasibility of randomization. The ultimate aim of this program of research will be to conduct a fully-powered parallel group randomized controlled design with an effectiveness-implementation focus (Curran, Bauer, Mittman, Pyne, & Stetler, 2012). A time and contact matched control condition that provides general support but does not include ACT principles could be useful for assessing whether the effects observed are specific to ACT-IN. Also, brief phone follow-up for 1 month post-discharge will be explored to help consolidate and maintain treatment gains from the inpatient stay. Furthermore, the intervention may benefit from a more specific emphasis on teaching relapse prevention skills during the inpatient stay. Finally, due to the loss of some treatment effects by 4 months on certain outcomes, we plan to conduct more frequent assessments during the post-hospital period to better map change trajectories for patients to understand the longer-term impact of treatment.