Perceived Autonomy Support in the NIMH RAISE Early Treatment Program

Abstract

Objective:

The present study examined perceived autonomy support, the extent to which individuals feel empowered and supported to make informed choices, among the participants from the Recovery After an Initial Schizophrenia Episode Early Treatment Program (RAISE ETP). The aims of the study were to evaluate whether NAVIGATE, the active treatment studied in RAISE ETP, was associated with greater improvements in perceived autonomy support over the 2-year intervention as compared to Community Care (CC), and to examine associations between perceived autonomy support and quality of life and symptoms over time and across treatment groups.

Methods:

The present study examined perceived autonomy support among the 404 first episode psychosis individuals who participated in the RAISE ETP trial. Three-level conditional linear growth modeling was utilized given the nested data structure.

Key Findings:

The results revealed that perceived autonomy support increased significantly over time for the NAVIGATE group but not for the CC group. Moreover, once treatment began, higher perceived autonomy support was related to higher quality of life at 6, 12, and 18 months in NAVIGATE and at 12, 18, and 24 months in CC. Finally, higher perceived autonomy support was related to lower total and excited symptoms regardless of treatment group and time.

Conclusions:

Overall, perceived autonomy support increased in NAVIGATE but not in CC and was related to quality of life and symptoms across both treatment groups. Future research should examine the impact of perceived autonomy support on a wider array of outcomes including engagement, medication adherence, and functioning.

Introduction

The Recovery After an Initial Schizophrenia Episode Early Treatment Program (RAISE ETP) study developed and tested the NAVIGATE program, a multi-component, team-based treatment for individuals with first episode psychosis (FEP), aimed at improving quality of life and psychosocial functioning in addition to reducing symptoms (1). The implementation of the NAVIGATE program was guided by a core set of principles including shared decision-making and supporting clients’ self-determination and personal autonomy (2). As a result, it is important to evaluate whether NAVIGATE was more effective at improving clients’ perceptions of autonomy support than usual care, and whether those perceptions were related to symptoms and quality of life over the course of treatment.

Self-determination theory (SDT) posits that autonomy, defined as the ability to act out of personal choice (rather than control), is a basic psychological need for all people (3). Autonomy support refers to the extent to which an individual feels empowered by his/her social environment to make decisions based on his/her own values and preferences (4). The support of another person’s autonomy requires perspective taking, providing a meaningful rationale for suggestions, and supporting an individual’s choices without attempting to exert control (4, 5). Though autonomy support has been identified as a valuable construct across a wide range of settings (e.g., parent-child relations, workplace environment), it is especially germane to psychological and psychiatric treatments (4, 5). Specifically, autonomy support has been associated with engagement in psychological treatment, health behavior change (e.g., smoking abstinence), quality of life, motivation, depression, and medication adherence (4–10). Autonomy support has also been shown to increase over the course of treatment when specifically integrated into an intervention for smoking abstinence (6). Moreover, when examined in the context of treatment for depression, autonomy support has been found to increase over time in Interpersonal Therapy, Cognitive Behavioral Therapy, and Pharmacotherapy + clinical management (10). Nonetheless, the lack of comparable studies in FEP treatment limits conclusions as to the malleability of autonomy support and its association with outcomes in this population.

The NAVIGATE program incorporated the support and bolstering of client self-determination and autonomy into its comprehensive treatment model to target quality of life, functional outcome, and well-being in FEP (2). The purpose of this paper was to examine perceived autonomy support among the individuals who participated in the RAISE ETP trial. The aims of these secondary analyses were to evaluate whether NAVIGATE was associated with greater increases in perceived autonomy support over the 2-year intervention as compared to Community Care (CC) and to examine the associations between perceived autonomy support and quality of life and symptoms over time and across treatment groups.

Methods

Participants and Study Design

The sample comprised 404 individuals with FEP who had only experienced one episode of non-affective psychosis (Table 1). See Kane and colleagues (11) for full demographic and clinical characteristics. RAISE ETP utilized a cluster randomization design such that thirty-four community mental health clinics were randomized to provide either the NAVIGATE program (N=17), a coordinated specialty care intervention for FEP, or usual treatment (CC; N=17). The RAISE ETP study received Institutional Review Board approval from the Coordinating Center and participating sites. All participants provided written informed consent or assent if under 18 years old.

Table 1.

Demographic, Clinical, and Baseline Characteristics

	Community Care (n=181)		NAVIGATE (n=223)
Demographic Characteristics	N	%	N	%
Male	120	66	173	78
Age (years; M±SD)	23.08±4.90		23.18±5.21
Race
Caucasian	80	44	138	62
African American	89	49	63	28
Other	12	7	22	10
Ethnicity
Hispanic	18	10	55	25
Education
Some college or higher	54	30	71	32
Completed high school	58	32	75	34
Some high school	58	32	67	30
Some or completed grade school	11	6	9	4
Current student	47	26	35	16
Clinical Characteristics
Diagnosis
Schizophrenia	101	56	113	51
Schizoaffective bipolar	13	7	11	5
Schizoaffective depressive	25	14	32	14
Schizophreniform	24	13	43	19
Brief psychotic disorder	1	1	1	1
Psychotic disorder NOS	17	9	23	10
DUP (weeks; M±SD)	211.43±277.49		178.91±248.73
Baseline Characteristics (M±SD)
QLS Total Score	54.77±18.99		50.89±18.44
PANSS Total Score	74.54±14.87		78.32±14.95
PANSS Positive	12.13±3.79		12.32±3.88
PANSS Negative	16.34±4.96		16.98±5.34
PANSS Disorganized/Concrete	7.34±2.63		8.18±2.83
PANSS Excited	6.38±2.30		7.05±3.06
PANSS Depressed	7.93±3.42		8.16±3.22
Autonomy Support Average	5.48±1.37		5.59±1.08

Note. NOS = Not otherwise specified; DUP = Duration of untreated psychosis; QLS = Quality of Life Scale, Possible scores range from 0 to 126 (Total Score) with higher scores indicating greater quality of life; PANSS = Positive and Negative Syndrome Scale; Possible scores range from 30 to 210 (Total Score), 4 to 28 (Positive), 6 to 42 (Negative), 3 to 21 (Disorganized/Concrete), 4 to 28 (Excited), and 3 to 21 (Depressed) with higher scores indicating greater severity of symptoms (Note: 20/30 items are included in the factor scores whereas all 30 items are included in the total score). Autonomy Support Average: Possible scores range from 1 to 7, with higher scores indicating greater perceived autonomy support.

Intervention

NAVIGATE is a multi-element treatment comprising individualized medication management, family psychoeducation, resilience-focused individual therapy, and supported employment and education (2, 12). In contrast to a symptom-focused treatment, NAVIGATE was designed to promote recovery through a focus on the client as a whole person with strengths and resilience (2). The program was centered on helping clients achieve personal goals through active efforts to support self-determination. Moreover, goal setting, shared decision-making, and supporting self-efficacy were incorporated into each of the manuals guiding the interventions within NAVIGATE (2). NAVIGATE sites received training in team-based FEP intervention prior to beginning the study and continued to receive consultation and fidelity monitoring. Sites randomized to CC, the standard care available for individuals with FEP, did not receive additional training or supervision.

Measures

Since the present analyses examined a subset of measures administered in the RAISE ETP study, only these measures are described in full detail. See Kane and colleagues (11) for a full description of the methods and procedure.

The Positive and Negative Syndrome Scale (PANSS) (13) is a standardized semi-structured interview for the assessment of symptoms in persons with schizophrenia spectrum disorders. Thirty items are rated on a 1–7 scale producing a total score and five factor scores: positive, negative, disorganized/concrete, excited, and depressed (14). The PANSS total score and 5 factor scores were used in the analyses. The Quality of Life Scale (QLS) (15) is a semi-structured interview consisting of 21 items rated on a 0–6 scale. A total score and four domain scores are produced: interpersonal relationships, instrumental role functioning, intrapsychic foundations, and common objects and activities. The QLS total score was used in analyses. Both the PANSS and the QLS were administered at baseline, 6, 12, 18, and 24 months by trained clinician-interviewers masked to study design and participant treatment using live, two-way video conferencing.

The six-item Autonomy Support Scale is the short-form of the longer 15-item self-report Health Care Climate Questionnaire (8). Items are rated on a 1–7 scale. The scale assesses an individual’s perceived autonomy support from his/her treatment team by rating statements such as “I feel that my clinicians have given me choices and options” and “My clinicians convey confidence in my ability to make changes.” This scale reflects a global rating of autonomy support from the entire treatment team (i.e., clients were not instructed to rate specific providers such as a therapist or prescriber but rather the entire team). Participants completed this measure along with several other self-report measures at baseline, 3, 6, 12, 18, and 24 months. The Autonomy Support Scale had good internal consistency in the present study (α = .90; measured at baseline). A mean score across all 6 items (Autonomy Support Average) was used. In the analyses examining autonomy support as a predictor of quality of life and symptoms, the 3-month assessment was excluded given that the outcome measures (PANSS and QLS) were not administered at 3 months.

Procedure

Enrollment occurred between July 2010 and July 2012 and all participants could receive treatment for at least two years. Study assessments were suspended during hospitalizations and incarcerations, but resumed after release or discharge. The final participant completed 2 years of treatment in July 2014.

Data Analysis

Multilevel modeling was utilized given the nested structure of these data (time nested within client nested within site). All analyses were conducted using SAS (version 9.3) and diagnostics were examined for all models. Time was linearized by a square root transformation given that the greatest improvement occurred within the first 6 months of treatment (11). In addition, three covariates (student status, gender, and baseline PANSS total scores) were included to adjust for baseline differences between the treatment groups (11). To compare changes over time in perceived autonomy support between the treatment groups, we fit a three-level conditional linear growth model with time (square root month in treatment) as a level 1 predictor and treatment group (NAVIGATE or CC) as a level 3 predictor. We included fixed effects for both predictors and the time by treatment group interaction term. A random intercept and slope for time was included at both the site and participant level. Finally, we probed the interaction term by calculating simple intercepts and slopes for both treatment groups and then graphically depicted these trajectories (16, 17).

To examine the associations between perceived autonomy support and quality of life and symptoms, we fit a three-level conditional linear growth model with perceived autonomy support (person mean-centered) and time (square root month in treatment) entered as predictors at level 1 and treatment group entered at level three. A random intercept and slope for time at the participant and site levels were included in all analyses; however, if any of the estimated covariance parameters were zero, the model was re-fit without the corresponding random effect(s). The within-level interaction of time by perceived autonomy support and the cross-level interactions of treatment group by perceived autonomy support and time by treatment group were also included in the models.

Finally, we examined the 3-way interaction of perceived autonomy support by time by treatment group to evaluate whether the relationship between perceived autonomy support and outcomes varied based on the time point and/or treatment group. For significant 3-way interactions, we probed the interaction using the “pick a point approach” originally developed by Rogosa (17). Specifically, we considered time as the focal predictor and calculated simple intercepts and slopes at the 5 time periods assessed in the present study: baseline, 6, 12, 18, and 24 months (16, 17). As a result, we were able to examine the association of perceived autonomy support and these outcomes in both treatment groups and at each of the five time points. Though baseline slopes were calculated and described within the text, our principal focus was on the relationship between autonomy support and outcomes at time points after treatment had commenced (i.e., at 6, 12, 18, and 24 months). Finally, model-implied simple regression lines were plotted to provide a graphical depiction of significant interaction effects.

Results

Model estimates for the fixed effects of interest are presented in Table 2, organized by outcome variable. Figures 1 and 2 depict significant 2-way and 3-way interactions of interest.

Table 2.

Model Estimates for Perceived Autonomy Support, Quality of Life, and Symptoms

Fixed Effects by Outcome	Fixed Effect Estimate	Standard Error
Autonomy Support Averagea,b,c,d
Time	−.022	.027
TreatGrp	.214	.130
Time*TreatGrp	.102**	.037
Quality of Life Total Scored,e,f
AutSup	−2.198	1.311
Time*TreatGrp*AutSup	−1.494*	.592
PANSS Total Scoree,f,g
AutSup	−3.429**	1.215
Time*TreatGrp*AutSup	−.884	.533
PANSS Positivea,e,f,g
AutSup	−.340	.268
Time*TreatGrp*AutSup	.079	.133
PANSS Negativea,e,f,g
AutSup	−.636	.372
Time*TreatGrp*AutSup	−.091	.158
PANSS Disorganized/Concretea,b,e,g
AutSup	−.472	.278
Time*TreatGrp*AutSup	−.169	.091
PANSS Exciteda,e,g,h
AutSup	−.435*	.191
Time*TreatGrp*AutSup	−.158	.105
PANSS Depresseda,e,f,g
AutSup	−.632*	.314
Time*TreatGrp*AutSup	−.236*	.102

Note. TreatGrp = Treatment Group (NAVIGATE vs. Community Care); AutSup = Autonomy Support Average; PANSS = Positive and Negative Syndrome Scale. Normally distributed outcomes were fit using Kenward-Roger Fixed Effects SE and Degrees of Freedom Method. Time refers to square root month in treatment; Autonomy Support was person mean-centered as a predictor; Student status, gender, and baseline PANSS were grand-mean-centered.

^aEmpirical Fixed Effects SE Method with Between-Within degrees of freedom was used.

^bIncludes random intercept and slope at participant and site levels.

^cFixed effect for Intercept was also included in the model.

^dIncludes student status, gender, and baseline PANSS Total score as covariates.

^eIncludes fixed effects for Intercept, Time, TreatGrp, Time*TreatGrp, TreatGrp*AutSup, and Time*AutSup.

^fIncludes random intercept and slope at participant level and random intercept at site level.

^gIncludes student status and gender as covariates.

^hIncludes random intercept and slope at participant level.

^*p<.05

^**p<.01.

Figure 1. Model-implied Trajectories of Perceived Autonomy Support.

Perceived autonomy support increased significantly over the course of 24 months in NAVIGATE (solid line) but not in Community Care (dotted line). Note: Time was back-transformed in this figure.

Figure 2. Model-implied Simple Regression Lines of Perceived Autonomy Support on Quality of Life as a function of Time and Treatment.

Higher perceived autonomy support was related to higher quality of life at baseline, 6, 12, and 18 months in NAVIGATE and at 12, 18, and 24 months in CC. The strongest effects once treatment commenced were observed at 6 months for NAVIGATE (line with circles) and at 24 months for CC (line with arrows).

Perceived Autonomy Support

The results revealed a significant time by treatment group interaction (t=2.75, df=1685, p=.006) for perceived autonomy support (Figure 1). Probing this interaction revealed a significant positive slope of .08 (t=3.17, df=1685, p=.002) for the NAVIGATE group indicating that perceived autonomy support increased over time. The slope of −.022 in the CC group was not significantly different from zero (t=−.82, df=1685, p=.412), indicating that perceived autonomy support did not significantly change over time in CC.

Quality of Life and Symptoms

Results of quality of life analyses indicated a significant 3-way interaction of time by treatment group by perceived autonomy support (t=−2.52, df=1075, p=.012). To probe this interaction, we examined the effect of time on the perceived autonomy support by treatment group interaction. The slopes at baseline, 6, 12, and 18 months were significantly different from zero in NAVIGATE, whereas the slopes at 12, 18, and 24 months were significantly different from zero in CC. Figure 2 depicts significant slopes once treatment had commenced (i.e., excluding baseline), as these time points were of primary interest. Further, the strongest effects (i.e., steepest slopes) once treatment commenced occurred at 6 months in NAVIGATE and at 24 months in CC. Perceived autonomy support was not significantly related to quality of life at 24 months in NAVIGATE or at baseline or 6 months in CC. Overall, perceived autonomy support was related to quality of life in both conditions across multiple time points with the strongest effects occurring early in treatment in NAVIGATE and at the end of treatment in CC.

The results of the PANSS analyses indicated main effects of perceived autonomy support for the Total Score (t=−2.82, df=825, p=.005) and Excited Factor Score (t=−2.28, df=920, p=.023), with higher perceived autonomy support scores related to lower PANSS scores across both treatment groups and time points. None of the two-way or three-way interactions for these two variables were statistically significant. For the PANSS depressed factor, there was a significant 3-way interaction of time by treatment group by perceived autonomy support (t=−2.31, df=1287, p=.021). Probing this interaction revealed a significant negative slope at baseline in CC (t=−2.01, df=1287, p=.045) indicating that higher perceived autonomy support was associated with lower PANSS depressed at baseline in CC. All other slopes were not significantly different from zero indicating that perceived autonomy support was not significantly related to PANSS depressed factor scores at 6, 12, 18, or 24 months in CC or at any time point in NAVIGATE. Perceived autonomy support was not significantly associated with PANSS positive, negative, or disorganized/concrete factor scores.

Discussion

Perceived autonomy support increased significantly over the course of the 24-month treatment period only for those who received NAVIGATE. This finding is consistent with the NAVIGATE treatment model that emphasized supporting client self-determination and autonomy (2). Prior research has not evaluated whether interventions targeting FEP are associated with higher levels of perceived autonomy support than standard treatment. Moreover, NAVIGATE is the first intervention that has specifically targeted and effectively changed autonomy support in FEP. This finding extends the current literature that had only examined the trajectory of perceived autonomy support in the context of treatment for depression (10) and smoking (6) to an FEP population. Furthermore, because individuals with FEP report lower levels of autonomy than healthy controls (18), supporting autonomy during treatment may be especially critical for this population.

In terms of associations with outcomes, higher perceived autonomy support was related to higher quality of life at baseline, 6, 12, and 18 months for individuals in NAVIGATE and at 12, 18, and 24 months for those in CC. These results suggest three possible interpretations, any or several of which could be correct. First, when people are less symptomatic and are functioning better, they may be more likely to perceive their treatment providers as supporting their autonomy compared to when they are experiencing more challenges, irrespective of the actual behavior of providers. Second, treatment providers may actually be more supportive of the personal autonomy of clients who are functioning better and less symptomatic, and may employ more persuasion or pressure to change in clients with more challenges. Third, higher levels of autonomy support might lead to better treatment outcomes. The strongest effects were observed towards the beginning of treatment (6 months) for NAVIGATE participants, but at the end of treatment (24 months) for participants in CC. These findings may reflect the fact that shared decision-making and support for client self-determination were integrated into NAVIGATE and played an especially critical role in quality of life early in treatment, when the greatest gains in functioning and symptoms occurred (11). Despite the fact that autonomy support was not specifically infused within CC, it still predicted quality of life (at 12, 18, and 24 months), providing evidence that when autonomy support is present, it relates to quality of life.

Consistent with findings that greater perceived autonomy support was related to greater quality of life, higher levels were related to lower PANSS total and excited symptoms (i.e., better symptoms) across both treatment groups and time points. Yet, perceived autonomy support was not significantly related to positive, negative, or disorganized symptoms. Therefore, perceived autonomy support appears to be associated with certain symptom clusters; however, there is limited evidence to support its associations with these symptoms differentially across time points or treatment groups. Higher perceived autonomy support was related to lower PANSS depressed symptoms only at baseline in CC. It did not predict PANSS depressed symptoms at 6, 12, 18, or 24 months in CC or at any time point in NAVIGATE. Because this finding was only present in CC prior to the commencement of treatment, this result should be interpreted with caution.

The current study provided the first examination of perceived autonomy support within FEP treatment and demonstrated that this construct is related to quality of life and symptoms. NAVIGATE was based upon recovery-oriented principles that focused on cultivating and reinforcing self-determination among clients through goal-setting as well as supporting their autonomy and capacity to make informed choices (2). Shared decision-making, a chief component of NAVIGATE, is at the core of client-centered care in which practitioners and clients work collaboratively to address concerns while discussing values and supporting the client’s autonomy (19, 20). Shared decision-making has recently gained more attention in mental health, especially with hard to engage populations, like those with serious mental illnesses (21). Not only does shared decision-making have the potential to significantly impact engagement, but individuals with psychosis and their providers also value this therapeutic technique (21, 22). Furthermore, given the high rates of treatment dropout among individuals with FEP (about 30%), recovery-oriented programs that support client autonomy have the potential to powerfully impact client engagement and retention (23).

The study was limited in its reliance on a brief self-report measure (the Autonomy Support Scale) and the use of global autonomy support ratings of the treatment team rather than of individual provider ratings. Because clients likely had varying amounts of contact with different providers (e.g., prescriber vs. individual therapist), their ratings of autonomy support could represent experiences with different treatment team members. As a result, differences in perceived autonomy support between NAVIGATE and CC clients could reflect differences in provider contact and contribute to the observed treatment differences. Although the global ratings used in this study do not capture these differences, they underscore the importance of all providers supporting client autonomy and offer an initial examination of this construct in FEP treatment. Additionally, though statistically significant, the observed increase in autonomy support for those in NAVIGATE is small in magnitude. Despite these limitations, the present study was the first to examine perceived autonomy support in the context of FEP treatment. Given the promise of these initial findings, future research may consider examining its impact on a wider array of outcomes including engagement, medication adherence, and functioning. Further, important next steps include examining the impact of changes in perceived autonomy support on changes in outcomes, formally testing the directionality of relationships observed in the present analyses, and exploring how engagement in various components of treatment (e.g., individual therapy, medication management, family psychoeducation, supported employment and education) affect perceived autonomy support and treatment outcomes.

Conclusions

The present study demonstrated that perceived autonomy support improved over the 2-year treatment only among those who received NAVIGATE, which is consistent with its recovery-oriented framework. Further, individual differences in perceived autonomy support were related to individual differences in quality of life and symptoms for those in both treatment conditions, suggesting that these associations may be relevant to FEP treatment, more broadly. Overall, these findings illustrate that perceived autonomy support is malleable in FEP and that it is associated with important outcome variables.