What matters: Factors impacting the recovery process among outpatient mental health service users

Abstract

Objective:

Understanding factors that promote and hinder the recovery process for people living with serious mental illness remains of critical importance. We examine factors, including limited literacy, associated with mental health recovery among public mental health service users.

Methods:

This study uses data from a mixed-methods, service-user informed project focused on the impact of limited literacy in the lives of people with serious mental illness. Data from structured interviews evaluate perceptions of recovery as assessed with the Recovery Assessment Scale (RAS). Regression models examine factors related to recovery controlling for sociodemographic factors, literacy, neurocognition, mental health status, and perceived social support.

Results:

Despite bivariate relationships between RAS and limited literacy, the full models suggest that other factors account for this relationship. These include mental health status, higher social support, higher self-reported community status, and higher stigma consciousness, as well as race for some models.

Conclusions and Implications for Practice:

Our findings that social support and perceptions of community status are associated with higher scores on the RAS echo prior work demonstrating the importance of social connection and context in mental health recovery. Though literacy was not a predictor of recovery, further research should examine the relationship between literacy and recovery given the deep literature on literacy on health outcomes. In order to better support people in the recovery process it is important that more research is done to examine the complex relationship between stigma consciousness and recovery as well as understand the racial disparities that exist within the recovery subscales.

Recovery-oriented care is an important part of behavioral health services, including those delivered in an integrated setting. With increased attention to the need for recovery oriented mental health services understanding the recovery processes of people with serious mental illness remains a critical challenge (Substance Abuse and Mental Health Services Administration, 2019). Many of the first voices to call for understandings of recovery were those of peers and service users. Although recovery is often described as a subjective or self-defined experience and thus not easily operationalized and measured, a rich body of research has emerged to further our understanding of recovery and the factors that may help or hinder the recovery processes. While this evolving knowledge base is often used to inform developing mental health services and delivery models (Agency for Health Care Research and Quality, n.d.), understanding factors associated with the recovery process has proven challenging.

Understanding the Recovery Process

The Substance Abuse and Mental Health Services Administration (SAMHSA) offers a framework for understanding recovery as “a process of change through which individuals improve their health and wellness, live self-directed lives, and strive to reach their full potential.” (SAMHSA, 2019). Measures of recovery typically ask respondents to assess the degree to which they see themselves as meaningfully engaged in different aspects of their lives consistent with SAMHSA’s framework. High scores reflect the person’s perception that that they are doing well reflecting ‘better’ recovery. Most studies find little or no relationship between recovery and gender, age, or race (Girard, Tinland, Boyer, Auguier, & French Housing First Study Group, 2015; Jasskelainen et al., 2012; Lloyd, King, & Moore, 2010); however, there are exceptions to this pattern. Leung and Psych (2000) found that women with schizophrenia had higher perceptions of one’s recovery process than men. In addition, Corrigan (2006) found that non-white participants reported higher perceptions of where one is in the recovery process than whites; and that age was associated with only the Goal and Success Orientation RAS subscale.

The relationship between health status and recovery is also equivocal. Several studies demonstrate that poor physical health is associated with lower perceptions of one’s recovery (Chiba & Miyamota, & Kawakami, 2010; Corrigan, Giffort, Rashid, Leary, & Okeke, 1999; Leung & Psych, 2000; McNaught, Caputi, Oades, & Deane, 2007; Norman, Windell, Lynch, & Manchanda, 2013; Roe, Maschiach-Eizenberg, & Lysaker, 2011; Salyers, Godfrey, Mueser, & Labriola, 2007). Research on recovery and diagnosis is less clear (Salzer and Brusilovskiy, 2014). Fuller (2010) reported that participants diagnosed with substance dependence, without other psychiatric diagnoses, reported higher recovery than participants with serious mental illness or those with co-occurring mental and substance use disorders. While, Lloyd (2010) found that people diagnosed with bipolar disorder had reported higher recovery than those with depression or schizophrenia, more recently Girard et al. (2015) found no difference between perceptions of recovery with participants diagnosed with schizophrenia and bi-polar disorder. When mental health status is operationalized as perceptions of one’s mental health (SF-12), a stronger relationship emerges. Gonzales, Hernandez, Douglas, and Chu (2015) found a significant association between lower perceived mental health impairment (SF-12) and higher perceived recovery scores.

Finally, higher neurocognitive functioning has emerged as one of the most consistent factors associated with recovery (Cavelti et al., 2012; Kopelowicz, Liberman, Bentura, Zarate, & Mintz, 2005; Roe et al., 2011; Valiente, Provencio, Espinosa, Chaves, & Fuentenebro, 2011). In addition, Giusti et al. (2015) found that higher neurocognition, and lower clinical insight and social functioning (assessed as involvement in socially useful activities); personal and social relationships; self-care and disturbing and aggressive behaviors were significant predictors of personal recovery among people with schizophrenia.

As social functioning is an integral aspect of recovery, researchers have investigated the relationship between social networks and support and a person’s recovery. Salzer and Brusilovskiy (2014) found that social network size, and greater social support were associated with higher recovery (Corrigan & Phelan, 2004; Corrigan et al., 1999; Muñoz, Sanchez, & Perez-Santos, 2011; Pernice-Duca & Onaga, 2009; Webb, Charbonneau, McCann, & Gayle, 2011). Others have focused on understanding the nature of support provided and demonstrated that the presence in one’s life of a person who supports incremental progress and provides realistic expectations, promotes higher perception of recovery (Kopelowicz et al., 2005; Liberman, 2002; Torgalsboen, 1999). Employment provides opportunities for additional social connection and meaning along with challenges and employment related stressors and results of studies examining the associations among employment status and recovery are mixed. For instance, Lloyd et al. (2010) study showed that people who were engaged in some form of paid employment had higher recovery than those that were receiving money from social security benefits. In contrast, Connell, King, and Crowe (2011) found that there was not a significant difference between workers and non-workers’ and level of recovery. Importantly, evidence suggests that just having a job might not matter as much as having a meaningful job. Hancock, Honey, & Bundy’s (2015) found that higher recovery was associated with being in a job that was defined as the type of job that was most meaningful to participants. Perception of higher social/community status and community engagement also promote higher perceptions of recovery (Lloyd et al, 2010; Townley, Kloss, & Wright, 2009).

Finally, stigma has been frequently examined in efforts to understand barriers to the recovery process (Salzer & Brusilovskiy, 2014; Muñoz et al., 2011; Ritsher, Otilingam & Grajales, 2003). These studies revealed that both perceived external stigma and internal stigma were negatively correlated with level of recovery.

Purpose

The current study aims to further examine and identify predictors of how one perceives one’s recovery process as measured by the Recovery Assessment Scale (RAS). Our original hypothesis focused on the relationship between reading literacy and recovery. Although there is growing evidence of the negative impact limited literacy has on people’s health and how they manage their health, there is very little literature on the impact limited literacy has on people living with serious mental illness. This gap in the literature related to limited literacy and mental health extends to the impact limited literacy has on the mental health recovery process. Limited literacy serves as a marker of distinct instrumental skills, reflects access to quality education and social position, and is itself highly stigmatized (Lincoln et al., 2017). Limited literacy thus merits further attention in the recovery field. Here we test whether limited literacy is associated with perceptions of recovery, whether neurocognition helps us to understand any associations among literacy and RAS scores as well as how societal level factors such as social support, social status, and stigma, relate to RAS scores. This study adds to the existing literature on correlates of recovery and helps us to understand the multi-level factors that shape the recovery process.

Methods

The data utilized were collected as part of a National Institute of Mental Health study examining the impact of limited literacy in the lives of people with serious mental illness. This larger study utilized a mixed-methods, service-user informed design including 294 structured interviews, 263 medical record reviews, and 45 qualitative interviews. These analyses draw upon the structured interviews and the medical record reviews. Multiple Institutional Review Boards approved the research study: Northeastern University, Beth Israel Deaconess Medical Center, The Massachusetts Department of Mental Health, Boston University Medical Center, and University of Massachusetts Dartmouth. Written informed consent was obtained after procedures of the study were explained. The research team worked with sites to develop a Consumer Consultant Group, (CCG) which consisted of five public mental health service users with varying levels of literacy. The CCG helped develop the structured interview and provided important guidance on recruitment strategies and interpretation of findings.

Participants

Study participants were recruited from two public urban mental health outpatient clinics between June 2013 and January 2016. Each site serves disadvantaged communities and patient populations that are diverse in race and diagnosis. Although the sites were chosen due to similarities, the two sites have important differences. Site 1 is a state funded, solely outpatient facility within a teaching hospital. Clients must meet Department of Mental Health eligibility criteria in order to receive services, including a diagnosis of a serious and persistent mental illness, e.g. psychotic disorder, major mood disorder, or anxiety spectrum disorder, with significant functional impairment. More than 1,500 people are seen each year at this outpatient clinic. The clinic is predominantly white, (61%) 14% African American, 6% Latino, and 4% Asian. Five percent do not identify a race or identify as “other” or “more than two races”.

Site 2 is an outpatient clinic that is part of a large, urban, safety-net hospital affiliated with an academic medical center. This facility provides services regardless of an individual’s resources and has no eligibility criteria for services. People with severe mental illness are well represented. Approximately 5% of patients have a diagnosis of schizophrenia or schizoaffective disorder, 20% have a diagnosis of bipolar disorder while the majority have a depressive disorder (45%) or anxiety spectrum disorders (35%). This clinic serves more than 4,000 people per year. Forty percent of patients are African-American, 30% white, 20% Latino, and the other 10% representing ‘more than one race’. The study sample reflects the diagnostic and racial/ethnic diversity found at our two sites (See Table 1 for the diagnostic composition of the study sample).

Table 1.

Summary Participant Characteristics and Mean RAS Score (N=244)

	N	%	Mean RAS Score
Sociodemographics
Gender
Male	103	41.37	3.72
Female	141	56.63	3.72
Race
White	94	37.75	3.64
Black	102	40.96	3.75
Other	48	19.28	3.80
Ethnicity
Yes	21	8.43	3.86
No	222	89.16	3.71
Missing	1	0.40	3.48
Age
18–29	31	12.45	3.79
30–39	35	14.06	3.72
40–49	76	30.52	3.66
50+	102	40.96	3.74
Employment
Employed	30	12.04	3.65
Student	4	2.0	4.13
Unemployed, Looking for work	70	28.11	3.75
Unemployed, not Looking for work	107	42.97	3.68
Other	34	13.64	3.77
Highest Level of Education
Did not finish High school	67	26.91	3.85
GED or equivalent	32	12.85	3.57
High school graduate o	50	20.08	3.68
Attended some vocational, trade, or business School	21	8.43	3.75
Some college without a degree	38	15.26	3.72
College, associate’s degree	7	2.81	3.50
College, bachelor’s degree	10	4.02	3.69
Some postdgraduate, without a degree	4	1.61	4.40
Postgraduate, master or doctoral degree	4	1.61	3.29
Reading Literacy
WCJ9
8th Grade and Below	173	69.48	3.77
9th Grade and Above	71	28.51	3.59
Health
Diagnosis
Depression	115	46.18	3.67
Missing	41	16.47	3.76
Bipolar	58	23.29	3.75
Missing	41	16.47	3.76
Schizophrenia	58	23.29	3.74
Missing	41	16.47	3.76
Anxiety	69	27.71	3.69
No Anxiety	137	55.02	3.72
Missing	41	16.47	3.76
PTSD	61	24.50	3.67
Missing	41	16.47	3.76
Substance use disorder	81	32.53	3.71
Missing	41	16.47	3.76

WC9 is subtest 9 – Passage Comprehension of the Woodcock Johnson III Test of Achievement

Recruitment

All patients receiving services at the outpatient mental health centers who conducted their appointments in English, aged 18 and over, were eligible for participation. Recruitment occurred on randomly selected days to avoid biases associated with temporal variation in clinical activities. Fliers were distributed and clinicians provided information to patients interested in participating and referred them to the project coordinator for further information and enrollment. Structured interviews were conducted with 294 participants and included collection and assessments of sociodemographic characteristics, reading literacy, cognitive functioning, health status, service utilization, civic participation, social support, attitudes towards recovery, and experiences of stigma and discrimination. A subset of these participants provided permission for review of their medical record (n=263) in order to gather diagnostic data.

Measures

Independent variables

Sociodemographic data include self-reported race, ethnicity, gender, and age. Due to small sample sizes of different race categories, race was characterized as an individual identifying as white, black, and other. Ethnicity was measured as an individual identifying with being latinx or not. Gender was characterized as female or male, and age was the self-report of the participant in years.

Reading comprehension was measured by the subtest 9 of the Woodcock Johnson III (WJ III) Tests of Achievement – Passage Comprehension. The WJ III is a standardized test normed against a representative U.S. population, ages 24 months to 90 year (Woodcock, McGrew, & Mather, 2001). Participants are shown a picture and given a passage describing the picture, which contains a missing word. Participants are asked to identify the missing word. As participants answer correctly, the difficulties of the questions advance. Grade equivalent scores are reported. This test has been widely used in psychology to measure cognition and appears to approximate literacy skills (Rosenfeld, Rudd, Emmons, Acevedo-Garcia, Martin, & Buka, 2011). Test 9 has a median test-retest correlation of .88 across all age groups (McGrew, Woodcock, & Schrank, 2007). Scores were dichotomized into two reading levels; reading at or above a ninth grade level; and reading at or below an eighth grade level. For purposes of this study reading at or below an eighth grade level is categorized as limited literacy.

Neurocognition was measured by a battery of 4 assessments: Trail Making Test, Part A, and B; Verbal Fluency – Animal Naming Test (ANT); and the Digit Symbol- Coding Test (DSCT). The Trail Making Test A and B provides information on visual search, scanning, speed of processing, mental flexibility, and executive functions (Tombaugh, 2004). Scores are normed by gender, race, age, and years of education (Heaton, Miller, Taylor, & Grant, 2004). The ANT task provides another indication of executive functioning including verbal retrieval and recall, self-monitoring aspects of cognition (the participant must keep track of responses already given), effortful self-initiation, and inhibition of responses, when appropriate (Rosser and Hodges, 1994; Da Silva, Petersson, Faisca, Igvar, & Reis, 2004). Scores are normed by age and education, ranging from 16–95 years old and 0–21 years of education (Tombaugh, Kozak, & Rees, 1999). The DSCT is one of fourteen subtests of Wechsler Adult Intelligence Scale-III (WAIS-III) and measures processing speed. It is assumed to be affected by visual perception, short-term memory, and motor coordination (Kaplan, Fein, Morris, & Delis, 1991). Scores are normed by age (Wechsler, 2008).

Participants’ mental health status was assessed utilizing the mental health component score of the SF-12, a standardized questionnaire used to assess patient’s perceptions of their health. The SF-12 is scored to produce 2 summery scores, the physical and mental health composite scores (PCS and MCS) (Ware, Kosinski, Dewey, & Gandek, 2000). The higher the score the better one’s health status.

Social Support was measured by the ENRICHD Social Support Inventory, (ESSI) which consists of 7 items that assess social support (Vaglio et al, 2004). Participants’ were asked how often in their daily life they received different types of support. For example, participants were asked to respond to statements such as, “Is there someone available that you count on to listen to you when you need to talk?” Responses were coded one for “none of the time” to five for “all of the time” with higher scores associated with more social support.

The MacArthur Scale of Subjective Social Status (the “MacArthur ladder”) was used to assess participants’ perception of their standing in the community (Adler & Stewart, 2007). Respondents were shown a picture of a ladder with 10 rungs. They were instructed to mark an “X” on the rung that best represented where they stand in the community.

Two types of stigma associated with mental illness were assessed: 1) mental health services concealment; and 2) mental illness stigma consciousness. The first measure is drawn from previous stigma research (Link, Cullen, Struening, Shrout, & Dohrenwend, 1989; Link & Phelan, 2014) and consists of five items that address secrecy about one’s mental health service use and fear of disclosing one’s mental health service utilization in a variety of interpersonal contexts. Stigma consciousness was measured by an adaption of Pinel’s (1999) 7-item Stigma Consciousness Questionnaire. Participants were asked five questions about the extent they are aware that they are perceived or treated differently because of having mental health diagnosis with higher score reflecting higher stigma consciousness.

Dependent variables

The Recovery Assessment Scale (RAS) (Giffort, Schmook, Woody, Vollendorf, & Gervain, 1995) is a 41-item scale developed to measure the recovery process. A factor analysis of the original RAS identified five factors comprising the recovery process and resulting in a 24-item scale (Corrigan, Salzer, Ralph, Sangster, & Keck, 2004). The RAS has consistently proven to be one of the best recovery measures based on several reviews utilizing specific evaluative criteria (e.g. psychometric properties, ease of administration, service-user involvement in creation of the measure, service-user satisfaction, use in academic research) (Burgess, Pirkis, Coombs, & Rosen, 2011; Cavelti, Kvrgic, Beck, Kossowsky, & Vauth, 2012; Law, Morrison, Bryn, & Hodson, 2012; Shanks, Williams, Leamy, Bird, Boutillier, & Slade, 2013). The RAS is a scale that measures a person’s perceptions about where they stand on five different components of the recovery process measured by 5 subscales: 1) personal confidence and hope; 2) willingness to ask for help; 3) goal and success orientation; 4) non-dominance by symptoms; and 5) reliance on others.

Data Analysis

An in-depth literature review of identified predictors of recovery measured by the Recovery Assessment Scale guided the determination of the independent variables that were used in univariate analysis with the dependent variable. Next, Pearson Correlations (see Table 2) were examined to determine significant correlations with independent variables and the RAS. All variables that had significant associations with RAS were utilized in multiple regression models. Although verbal fluency was the only neurocognitive measure with a significant association to RAS, the four neurocognitive measures were conceptualized as a battery and all four were included. Other multiple regression model variables were race, gender, and age to control for sociodemographics.

Table 2.

Recovery Assessment Scale (RAS) Correlation Table (N=244)

	RAS
Recovery Assessment Scale (RAS)	1.000
Reading Literacy (WCJ9)	−0.199**
Sociodemographics
Gender
Male	−0.024
Female	0.024
Other
Race
White	−0.108
Black	0.029
Other	0.096
Age
Age	0.017
Ethnicity
Latino	0.102
Education
Years of Education	−0.003
Employment
Employed	−0.059
Neurocognition
Trails A	−0.059
Trails B	−0.034
Verbal Fluency	−0.210***
Digit Symbol Coding	−0.086
Health
Mental Health Status
SF-12 MCS	0.477***
Diagnosis
Depression	−0.087
Bipolar	0.086
Anxiety	−0.023
Schizophrenia	0.029
Substance abuse	−0.009
PTSD	−0.028
Social Support
Enrichd	0.517***
Community Engagement
Community Status Ladder	0.387***
Stigma
Mental Health Concealment	0.383***
Mental Health Stigma Consciousness	−0.271***

^*p<.05

^**p<0.01

^***p<0.001

WC9 is subtest 9 – Passage Comprehension of the Woodcock Johnson III Test of Achievement

SF-12 MCS is the mental health composite score of the SF-12

Enrichd is the Enrichd Social Support Inventory

Community Status Ladder is the MacArthur Scale of Subjective Social Status

Three multiple regression models were identified, the first controlling for sociodemographics (race, gender, and age); the second model controlling for sociodemographics as well as limited literacy (Test 9 of Woodcock Johnson III) and neurocognition (Trail Making Test Part A and Part B, verbal Fluency, and Digit Symbol-Coding Test). The third and final model included variables in model two and also controlled for health status (SF-12 MCS) and social level factors, such as social support (Enrichd Social Support Inventory), community status (MacArthur Scale of Subjective Social Status), and stigma (stigma consciousness and stigma concealment) with a final model of 13 variables.

Analyses were conducted examining total RAS scores and then scores of the five subscales. Models were tested for assumptions of linear regression, including for linear relationships between predictors and dependent variables, influential observations, heteroskedastic errors, autocorrelation, and multicollinearity. Only one subscale model showed concerning results on the autocorrelation tests, and robust standard errors were therefore used for that model (the reliance on other models presented below). Listwise deletion was used and fifty-five participants were excluded from analysis due to missing data. All analyses were conducted using R (R Core Team, 2013). Models were fit using the base R ‘stats’ package (R Core Team, 2018), with assumptions checked using the ‘car’ (Fox & Weisberg, 2019) and ‘lmtest’ (Zeileis & Hothorn, 2002) packages and effect sizes obtained using ‘heplots’ (Fox &Friendly, 2018). For the “Reliance on Others” model, robust standard errors were calculated using the ‘sandwich’ package (Zeileis, 2004).

Results

Study participants reflected the diversity that exists in the two sites (Table 1). A little over half of the participants (56.63%) were female, and there was significant racial diversity with 38% of participants identifying as white, 41% black, and 19% of other racial backgrounds. The majority of the participants did not identify as being Latinx (89%). Participant age ranged from 19 to 77. More than two thirds of the participants were unemployed (71.08%), with 43% not looking for work and 28% seeking employment. Nearly two thirds had a high school diploma/GED (equivalency test to receiving a high school diploma in the U.S.) (33%) or less (27%). The remaining third had some post-high school education with 10% reporting some type of college degree and 3% with postgraduate experience. The prevalence of limited literacy in our study population was high. Despite the reported educational achievement, almost three quarters of our study participants read at or below the 8th grade level (69.48%). Participants had a range of mental health diagnoses, the majority being depression (46.18%) and substance abuse (32.53%).

Pearson correlations were examined to determine the relationship between RAS scores and the variables identified by the literature as being predictors of overall RAS scores (see Table 2). Our primary relationship of interest, reading literacy, had a weak, negative relationship with recovery (r = 0.199**). Mental health status (r = 0.477***), social support (r = 0.517***), community engagement (r = 0.387***), and mental health concealment (0.383***) were all significantly positively associated with RAS. Verbal fluency (r = −0.210***) and mental health services stigma consciousness (−0.271) were significantly negatively correlated with overall RAS scores. All diagnostic variables were not significantly correlated with RAS scores and therefore were not included in multiple regression models.

Next, multiple regression was used to examine the relationship between sociodemographics and recovery attitudes (see Table 3). For the total RAS score, sociodemographics alone, including gender, did not predict recovery scores. The second model included additional literacy and neurocognitive variables also did not significantly predict recovery scores. For both of these models, the F-test was not statistically significant, and the adjusted R-squares were essentially zero. Our third model, however, provided far more robust estimates of variation in recovery. This model contains additional individual level factors such as mental health status (β = 0.01***) and social factors such as social support (β = 0.03***), community status (β = 0.03*), and stigma consciousness (β = 0.18**). In the final models for total RAS score, social support, community status, and stigma concealment and consciousness are positively associated. Verbal Fluency was no longer significant (−0.08** in model two, −0.04 in model three). Adding mental health status and societal level factors increased model fit substantially, with decreases in AIC and BIC numbers. This final model explains 42% of the variation in overall RAS scores (adjusted R²=0.42, F=13.32, p < .001). Effect sizes in these models varied. The social support scale (η² = .132) has a moderate effect on RAS and the community status (η² = 0.019), and stigma consciousness (η² = 0.035), and SF-12 MCS (η² = 0.06) variables had small effects on total RAS score.

Table 3.

Multiple Linear Regression Models for Associates with the Total Score of the Recovery Assessment Scale

	Model 1 β (95% CI)	Model 2 β (95% CI)	Model 3 β (95% CI)
Race
White (ref.)
Black	0.11 (−0.05–0.26)	0.06 (−0.11–0.22)	0.02 (−0.11–0.16)
Other	0.16 (−0.03–0.35)	0.13 (−0.07–0.32)	0.12 (−0.04–0.28)
Age
Age	0.0000 (−0.01–0.01)	0.001 (−0.005–0.01)	−0.001 (−0.01–0.004)
Gender
Male (ref.)
Female	−0.001 (−0.14–0.14)	0.01 (−0.13–0.15)	−0.02 (−0.14–0.09)
Literacy
Above 8th grade (ref.)
8th grade and below		0.15 (−0.02–0.31)	−0.01 (−0.15–0.12)
Neurocognitive
Trails A		−0.001 (−0.01–0.01)	−0.0004 (−0.01–0.01)
Trails B		0.001 (−0.004–0.01)	−0.001 (−0.01–0.003)
Verbal Fluency		−0.08* (−0.14–−0.02)	−0.04 (−0.08–0.01)
Digit Symbol Coding		0.004 (−0.03–0.04)	−0.004 (−0.03–0.02)
Mental Health
SF-12 MCS			0.01*** (0.01−0.02)
Social Support
Community Status Ladder			0.03* (0.01−0.05)
ENRICHD			0.03*** (0.02−0.04)
Stigma
Stigma Consciousness			0.18** (0.07−0.28)
Stigma Concealment			−0.02 (−0.08–0.03)
Constant	3.64*** (3.32–3.96)	3.46*** (2.98–3.94)	2.23*** (1.71–2.74)
AIC	403.421	402.252	281.45
BIC	424.404	440.721	337.405
Adjusted R2	−0.003	0.02	0.42
F Statistic	0.83 (df = 4; 239)	1.59 (df = 9; 234)	13.32*** (df = 14; 229)

^*p<.05

^**p<0.01

^***p<0.001

WC9 is subtest 9 – Passage Comprehension of the Woodcock Johnson III Test of Achievement

SF-12 MCS is the mental health composite score of the SF-12

ENRICHD is the Enrichd Social Support Inventory

Community Status Ladder is the MacArthur Scale of Subjective Social Status

Examination of the RAS subscales demonstrated primarily consistent findings with that for the overall RAS model (see Table 4). Mental health status, perceived levels of social support and community status, and stigma consciousness were all consistently associated with RAS subscales. Exceptions from the overall models in Table 3 include that black race was associated with recovery on the Goal and Success Orientation subscale and black race was negatively associated with recovery on the Reliance on Other’s subscale. Additionally, mental health status was not related to the Willingness to Ask for Help and Reliance on Other subscales; perception of community status was not associated with the Goal and Success Orientation and Non-dominance by Symptoms subscales; stigma consciousness was not associated with the Willingness to Ask for Help, and Reliance on Others subscales; and stigma concealment was only significant in the Personal Confidence and Hope subscale.

Table 4.

Multiple Linear Regression Model 3 for Associates with the Recovery Assessment Subscales (n=244)

	Personal confidence and hope β (95% CI)	Willingness to ask for help β (95% CI)	Goal and success orientation β (95% CI)	Non-dominance by symptoms β (95% CI)	Reliance on others β (95% CI)
Race
White (ref.)
Black	0.09 (−0.06–0.25)	−0.02 (−0.25–0.21)	0.21* (0.03–0.39)	−0.03 (−0.27–0.22)	−0.27** (−0.46–−0.09)
Other	0.17 (−0.01–0.36)	−0.11 (−0.39–0.16)	0.28** (0.07–0.49)	0.22 (−0.07–0.51)	−0.10 (−0.31–0.12)
Age
Age	−0.001 (−0.01–0.004)	0.003 (−0.01–0.01)	−0.004 (−0.01–0.002)	0.001 (−0.01–0.01)	−0.0001 (−0.01–0.01)
Gender
Male (ref.)
Female	−0.04 (−0.17–0.09)	0.03 (−0.17–0.22)	0.01 (−0.14–0.15)	0.03 (−0.17–0.23)	−0.10 (−0.26–0.05)
Literacy
Above 8 th grade (ref.)
8th grade and below	0.02 (−0.14–0.17)	−0.03 (−0.27–0.20)	0.02 (−0.16–0.20)	−0.11 (−0.35–0.14)	−0.03 (−0.21–0.16)
Neurocognitive
Trails A	0.004 (−0.003–0.01)	−0.0002 (−0.01–0.01)	0.001 (−0.01–0.01)	−0.002 (−0.01–0.01)	−0.01* (−0.02–−0.001)
Trails B	−0.002 (−0.01–0.003)	−0.004 (−0.01–0.003)	0.001 (−0.004–0.01)	−0.005 (−0.01–0.003)	0.002 (−0.004–0.01)
Verbal Fluency	−0.04 (−0.09–0.02)	−0.05 (−0.13–0.03)	−0.04 (−0.10–0.02)	−0.03 (−0.11–0.06)	−0.04 (−0.10–0.02)
Digit Symbol Coding	0.001 (−0.03–0.03)	−0.05 (−0.10–0.001)	0.01 (−0.03–0.04)	0.0004 (−0.05–0.05)	−0.0003 (−0.04–0.04)
Mental Health
SF-12 MCS	0.02*** (0.01–0.03)	0.01 (−0.004–0.01)	0.01** (0.002–0.02)	0.02** (0.01–0.03)	0.0001 (−0.01–0.01)
Social Support
Community Status Ladder	0.03 (−0.0003–0.05)	0.04* (0.001–0.08)	0.02 (−0.01–0.05)	0.03 (−0.01–0.07)	0.04* (0.003–0.07)
ENRICHD	0.02** (0.01–0.03)	0.04*** (0.02–0.06)	0.02*** (0.01–0.04)	0.03*** (0.01–0.05)	0.06*** (0.05–0.07)
Stigma
Stigma Consciousness	0.26*** (0.13–0.38)	0.07 (−0.12–0.25)	0.18* (0.04–0.32)	0.20* (0.005–0.39)	0.09 (−0.06–0.23)
Stigma Concealment	−0.07* (−0.14–−0.005)	−0.01 (−0.11–0.09)	0.01 (−0.06–0.09)	−0.03 (−0.13–0.08)	0.02 (−0.06–0.10)
Constant	1.90*** (1.30–2.49)	2.79*** (1.90–3.68)	2.46*** (1.78–3.14)	1.61*** (0.67–2.55)	2.62*** (1.92–3.32)
Observations	244	244	244	244	244
Adjusted R2	0.41	0.19	0.21	0.18	0.34
F Statistic (df = 14; 229)	12.94***	5.19***	5.73***	4.91***	9.91***

^*p<.05

^**p<0.01

^***p<0.001

WC9 is subtest 9 – Passage Comprehension of the Woodcock Johnson III Test of Achievement

SF-12 MCS is the mental health composite score of the SF-12

ENRICHD is the Enrichd Social Support Inventory

Community Status Ladder is the MacArthur Scale of Subjective Social Status

Discussion

We did not find support for our initial hypothesis, that limited reading literacy would be related to perceptions of the recovery process despite a negative univariate association between limited literacy and perceptions of recovery. This negative association may be explained similarly to that proposed by Giusti et al. (2015) who found that people with schizophrenia who had higher insight predicted lower perceptions of recovery such that people who are more aware of their illness and the issues it brings for them, may perceive themselves lower in the recovery process. Similarly, people who have higher levels of literacy are able to gain more knowledge about their illness and understanding of the negative impacts it can have on their lives. Although literacy does not remain significant in the regression models, several important findings remain illustrating the importance of social context in the recovery process including social support, community engagement, and stigma consciousness which each play a significant role in predicting RAS score and help explain participants’ perceptions of being further in their recovery process.

Our findings parallel prior studies by suggesting that better mental health status, greater levels of social support, and more engagement in the community are all factors that are positively associated with recovery as measured by the RAS. This is similar to Gonzales et al. (2015) study of 80 youth diagnosed with substance abuse who were engaged in a pilot aftercare study. They found better mental health status, as measured by the SF-12, predicted higher perceptions of recovery. Measures of subjective well-being (such as the SF-12) share conceptual underpinnings with the operationalization of measurement of the recovery process of the RAS and thus these associations are not surprising.

There is rich evidence of the positive associations among types of social support and community engagement and mental health recovery (Corrigan et al., 1999; Corrigan & Phelan, 2004; Muñoz et al., 2011; Pernice-Duca & Onaga, 2009; Webb et al., 2011). People living and experiencing mental illness speak of the power brought by social support. Leete (1989) in her personal narrative about what it is like living with mental illness describes the importance of a support system, “An ongoing and reliable support system has been extremely important” (p. 198). She further talks about how important and invaluable it is to have at least one person you can trust. Other personal narratives also discuss the importance of being surrounded by supportive people, which have helped their recovery. Houghton (1982) describes the powerful impact having people around her who accept her weakness and strengths has on her recovery. “They reassure me that ‘I’m OK’ (p. 551). Community engagement has also been demonstrated to promote recovery (Lloyd et al., 2010; Townley et al., 2009). These positive associations may be understood as reflecting the idea that recovering from mental illness is social in nature; a process which involves people and the greater community. Being engaged in one’s community helps people feel like an active citizen (Ware, Hoper, Tugenberg, Dickey, & Fisher, 2007). Others have identified both social support and community engagement as reflections or components of increased social capital (Tew, Ramon, Slade, Bird, Melton, & Le Boutillier, 2011) that enables recovery.

Our findings on the relationships among stigma and recovery are more complex. Stigma has been identified as a barrier to a person’s mental health recovery process (Tew et al., 2011). Our finding that one’s stigma consciousness is positively associated with one’s perceptions of recovery seems to contradict this literature. When identifying the components that make up the recovery process, the literature identifies acceptance as being an essential component of the recovery process (Chamberlin, 1978; Deegan, 1988; Jacobson & Greenley, 2001; Roe & Chopra, 2003). Acceptance in the recovery process is defined by a person’s ability and willingness to accept that due to having a mental illness they now have limitations; they now have a new reality and acceptance of one’s self is a crucial factor in this process. Deegan (1988) talks about the paradox of recovery in that accepting what one cannot do is the start of a person being able to recognize what can actually be done. Many have written about the trauma of being a patient with mental illness, the illness itself and the impossibility of returning to the state before becoming ill, and the observation that it is important to accept one’s illness to help in the process of building a new life and meaning (Chamberlin, 1978; Jacobson & Greenley, 2001; Roe & Chopra, 2003). This positive association between stigma consciousness and perceptions of recovery may be a function of this idea of acceptance moving one further along in the recovery process such that being aware of the stigma that exists in society about having a mental illness may tie into the idea that being able to accept your reality helps one continue on the path of recovery. Our finding that one’s stigma consciousness is positively associated with one’s perceptions of recovery attitudes suggests that experiences of stigma are complex and may operate in differing ways throughout the recovery process.

Finally, our finding of racial differences in the experience of recovery provides additional evidence to remind us of the importance of social context in understanding stigma and recovery processes. Little research has examined the relationship between race/ethnicity and recovery attitudes, and it is important to identify how race is a factor in people’s perceptions of recovery. Similar to Corrigan (2006), we analyzed the 5 factors of the Recovery Assessment Scale to determine if there were predictors specific to a recovery factor. Being black is associated with higher recovery scores in the Goal and Success Orientation subscale and is associated with lower recovery scores in the Reliance of Others subscale. Blacks’ lower scores on the Reliance on Others RAS subscale is consistent with research indicating that people who identify as being black may ask for help less, especially when it comes to seeking care for their health (Wells, Klap, Koike, & Sherbourne, 2001; Harris, Edlund, & Larson, 2005; Wang, Lane, Olfson, Pincus, Wells, & Kessler, 2005). This racial difference is an important aspect of how someone who is Black may conceptualize their recovery process and needs to be considered by providers and programs aiming to support people of color in their recovery.

We note several study limitations. First, related to the outcome variable (RAS), during the creation of the interview guide, one item from the Personal Confidence and Hope subscale (“I have an idea of who I want to become”) was excluded through our community informed process, and thus twenty-three questions were included in this analysis instead of twenty-four. Additionally, due to missing data patterns in the RAS subscales both the subscales and the full RAS were scored using R’s psych package (Revelle, 2018), which includes functionality for calculating mean scores from only the available, valid responses. For the final, analytical sample, a total of six respondents (2.46% of the 244 respondents) had missing data on the RAS. Five of those respondents had one response to a RAS item that was missing, and one respondent has two items missing. Given the relatively small number of missing observations within the data set, this was deemed the optimum approach following Newman (2014).

In addition, data were not available on some important predictors of recovery related to employment experiences such as the meaningfulness of a job (Hancock, Honey, & Bundy, 2015). Although this study did capture employment status, we did not have any scales to assess employment engagement and/or satisfaction. We were also not able to capture the role that mental health professionals have in helping people towards recovery which has been identified by Giusti et al. (2019) as an important factor in recovery. Lastly, several of the effect sizes of social factors associated with recovery were small (stigma and community status ladder) and therefore the clinical meaningfulness of these effects should be further examined, along with the meaning of employment and the role of mental health professionals in the recovery process.

Conclusion

These analyses demonstrate the importance of multiple types of social connection, including social support, community engagement, perceived social status and stigma, as well one’s mental health status, in a person’s recovery from mental illness. Our hypothesis that literacy level would positively predict people’s perception of recovery was not supported. Further research needs to continue to examine the relationship between literacy and recovery as we know there is an impact of literacy on health outcomes (DeWalt, Berkman, Sheridan, Lohr, & Pignone, 2004; Nielsen-Bohlman, Panzer, & Kindig, 2004) and we need to learn more about its possible impacts on mental health, mental illness, and recovery. Longitudinal designs that follow participants over time would be a methodologically robust way to examine how recovery changes over the course of a client’s illness trajectory and examine if literacy impacts outcomes at different points in time.

Our findings of racial differences in the subscales reminds us that more work is needed to understand racial disparities in recovery and that these efforts are key to promoting health equity. Finally, further research is needed to untangle the complexity of the association between stigma consciousness and perceptions of recovery. This may produce a better understanding of not only how the stigma phenomena works, is managed or resisted, but how to best support people in their recovery process and to create strategies to minimize the barriers in helping people who live with serious mental illness “recover a new and valued sense of self and purpose” (Deegan, 1988).

Impact and Implications:

Limited literacy was not significantly associated with recovery attitudes however individual and social level factors, mental health status, social support, perception of community status, and stigma consciousness were found to positively impact one’s recovery attitudes. These factors need to be understood so they can be leveraged to help people move towards recovery.