The Relationship Between Resilience,Self-Efficacy,and Employment in People With Physical Disabilities

Abstract

The aim of this cross-sectional study was to determine the importance of resilience and self-efficacy in explaining employment status for people with long-term physical disabilities when other sociodemographic and symptom variables were also examined. A multiple logistic regression with sequential predictor entry was used to predict employment status. Participants were individuals, 18 to 65 years of age, with a diagnosis of multiple sclerosis, muscular dystrophy, post-polio syndrome, or spinal cord injury (N = 882). Results indicated resilience but not self-efficacy was uniquely predictive of employment status. The combined effects of resilience and self-efficacy, however, did not significantly contribute to the variance in employment status above and beyond the sociodemographic and symptom variables. Other variables uniquely predictive of employment included education, age, marital status, disability benefits, and anxiety. We found in this study preliminary evidence to suggest that rehabilitation counseling practitioners should consider a client’s resilience with respect to employment. Knowledge of sociodemographic and symptom factors in conjunction with psychometrically sound measures of resilience and self-efficacy may be used to identify individuals with long-term physical disabilities whose beliefs and behaviors may limit the extent to which they prepare for, obtain, or maintain employment.

Keywords

positive psychology chronic illness stress and coping

Employment plays a key role in providing individuals with economic security, social participation, and access to health insurance and health care. Employment also contributes to personal identity and promotes physical and psychosocial health. For individuals with physical disabilities, these benefits are often elusive as many continue to be excluded from employment. Adults with physical disabilities experience higher rates of unemployment and underemployment than the general population. In 2016, the U.S. employment rate for adults with an ambulatory disability was about 25% where as adults without disabilities were employed at a rate of about 77% (Erickson, Lee, & von Schrader, 2017). Employment rates for adults with long-term physical disabilities which include multiple sclerosis (MS), muscular dystrophy (MD), post-polio syndrome (PPS), and spinal cord injury (SCI) can vary widely depending on the definition of employment used (i.e., full- or part-time status) and the time at which employment was measured (i.e., following injury or onset of the condition). For example, in two separate studies on employment outcomes for individuals with MS, the employment rate for individuals who were working 20 hours or more per week was about 40% (Johnson, Bamer, & Fraser, 2013; Julian, Vella, Vollmer, Hadjimichael, & Mohr, 2008). In a systematic review on return to work and employment in people with SCI, employment rates postinjury ranged from 12% to 74% (Lidal, Huynh, & Biering-Sørensen, 2007).

Several individual and societal level factors believed to be associated with employment for adults with long-term physical disabilities have been examined in previous studies. Sociodemographic factors frequently associated with unemployment for this population include older age (Hirsh, Molton, Johnson, Bombardier, & Jensen, 2009; I. Krause, Kern, Horntrich, & Ziemssen, 2013), female gender (Minis et al., 2010; Sweetland, Howse, & Playford, 2012), non-White race (J. Krause, Saunders, & Staten, 2010), and receipt of disability cash assistance (Chiu, Chan, Bishop, da Silva Cardoso, & O’Neill, 2013; Ottomanelli, Sippel, Cipher, & Goetz, 2011), whereas higher levels of education pre- and postinjury/disease have been associated with employment (J. Krause & Reed, 2009; Sweetland et al., 2012). The presence of specific symptoms such as pain (Shahrbanian, Auais, Duquette, Anderson, & Mayo, 2013), fatigue (Johnson et al., 2004), depression (Burns, Boyd, Hill, Hough, & Elliott, 2010; Honarmand, Akbar, Kou, & Feinstein, 2011), anxiety (Tan-Kristanto & Kiropoulos, 2015), perceived cognitive dysfunction (I. Krause et al., 2013), and physical impairment (Moore et al., 2013) also have been found to be associated with unemployment. Whether someone is employed or not also has been related to social and environmental factors such as perceived discrimination or support from employers and the community (Neath, Roessler, McMahon, & Rumrill, 2007; Nevala, Pehkonen, Koskela, Ruusuvuori, & Anttila, 2015).

Beyond these sociodemographic and symptom factors, there has been recent interest in the relationship between psychological resources and employment outcomes for individuals with physical disabilities because unlike many sociodemographic and some symptom factors, psychological resources are potentially modifiable. Peter, Müller, Cieza, and Geyh (2011) defined psychological resources as “inner, health protecting and health promoting potentials of a person, which represent a source or means to deal with difficult situations or obtain valued ends” (p. 188). Psychological resources such as self-reliance (Burns et al., 2010), degree of motivation (Peter et al., 2011), and internal locus of control (J. Krause & Broderick, 2006) have been positively correlated with employment for individuals with physical disabilities. There has been less research, however, on the relationship between the psychological resources of resilience and self-efficacy and employment for individuals with physical disabilities despite evidence to support strong relationships between these factors and improved psychosocial and physical functioning (White, Driver, & Warren, 2008).

Research on resilience and self-efficacy and the rehabilitation of people with physical disabilities has been conducted largely to identify characteristics associated with people who maintain functionality in the face of adversity. For example, Silverman, Molton, Alschuler, Ehde, and Jensen (2015) found that resilience is associated with improved functional outcomes such as depressive symptoms, social participation, and physical functioning in individuals aging with long-term physical disabilities. In a study on psychological factors and adjustment outcomes in those newly diagnosed with MS, the authors reported that lower levels of personal competence (measured on the Connor–Davidson Resilience Scale) significantly predicted both depressive and anxiety symptoms (Tan-Kristanto & Kiropoulos, 2015). An investigation of Kumpfer’s resilience model and quality of life outcomes for individuals with SCI found, among other things, that participation self-efficacy was a stronger predictor of quality of life than the severity of the SCI (Tansey, Bezyak, Kaya, Ditchman, & Catalano, 2017). Peter and colleagues (2011) found evidence for the relationship of self-efficacy with better mental health outcomes and higher subjective well-being for individuals with SCI. While strengthening psychological resources such as resilience and self-efficacy to support successful adjustment to disability is an aim of rehabilitation programs, these authors identified only a few intervention studies of acceptable quality for enhancing these resources.

Although there is evidence demonstrating that resilience and self-efficacy are strong and reliable predictors of adjustment to disability, to date, there are no published studies examining whether both resilience and self-efficacy are strong predictors of employment for people with physical disabilities. Vocational rehabilitation practitioners assist clients in adjusting to and coping with the often deleterious effects of job loss and periods of unemployment as well as the process of entering and reentering the labor force. Therefore, greater knowledge of the effects of resilience and self-efficacy on employment for individuals with physical disabilities has important implications for developing evidence-based psychological assessments and interventions for this group to enhance their personal resources in the face of these adversities (O’Sullivan & Strauser, 2009).

Using data from a survey project exploring factors associated with participation for individuals with long-term physical disabilities living in the United States, the purpose of the present study is to determine the importance of resilience and self-efficacy in explaining employment status when other sociodemographic and symptom variables are also examined. In this analysis, we seek to answer the following research question: What are the unique effects of resilience and self-efficacy on employment status for adults with long-term physical disabilities after controlling for covariates?

Based on the current literature on the relationship between the factors of resilience and self-efficacy and quality of life and other outcomes for individuals with physical disabilities, we hypothesized that individuals who report higher levels of resilience and self-efficacy were more likely to be employed than those who reported lower levels of resilience and self-efficacy. This study is an initial step in identifying individuals with long-term physical disabilities who may benefit from interventions to strengthen their resilience and self-efficacy with regard to employment.

Method

Participants and Procedure

The present study draws from extant data collected as part of the Rehabilitation Research and Training Center (RRTC) on Healthy Aging and Physical Disability (Healthy Aging RRTC) at the University of Washington (UW). The general purpose of this RRTC is to advance knowledge on healthy aging for individuals with medical conditions acquired or with onset before age 40 years (such as MS, MD, PPS, and SCI) that are typically associated with impairment in at least one activity of daily living. This population was selected by researchers at the Healthy Aging RRTC because these conditions are associated with different disease trajectories and prevalence of comorbidity or multimorbidity over time. Individuals with these conditions typically require intensive rehabilitation therapy services across the lifespan. The UW Healthy Aging RRTC has been conducting quality of life research, intervention research, and health policy research in individuals aging with long-term physical disabilities since 2008.

Eligible individuals were 18 years of age or older, were able to read and understand English, and self-reported a physician’s diagnosis of MS, MD, PPS, or SCI. Project year five data, the most current data set at the time of this study, were analyzed. Between October 2014 and May 2015, individuals with MS, MD, PPS, and SCI were recruited through invitation letters sent to disability-specific research registries (e.g., the UW Participant Pool and the University of Rochester MD Registry), invitation letters to former research participants at the UW, and web and print advertisements. Individuals who were eligible and interested in participating were either mailed a self-report paper questionnaire or directed to an online version of the questionnaire. Questionnaires were sent to 1,949 individuals, of which 1,581 were completed and returned to the researchers. Seven questionnaires were determined to be invalid because one individual incorrectly reported a diagnosis of MS, and six individuals returned questionnaires after the data set had been finalized (N = 1,574).

Individuals who were older than the full retirement age of 65 years were excluded from this analysis (n = 654). We did not have sufficient information to determine whether individuals age 65 years and older in this sample were out of the labor force voluntarily (and chose to retire) or involuntarily (and were actively looking for work). This rationale is consistent with previous studies on employment in people with physical disabilities (Botticello, Chen, & Tulsky, 2012; Chiu et al., 2013). In addition, 38 individuals were excluded from the analysis due to missing data. Therefore, complete data from N = 882 individuals were used for analyses.

Measures

Survey data were collected on sociodemographic and symptom characteristics. Variables included in this analysis were chronological age, gender, marital status, race, education level, disability benefits, and employment status. Individuals who indicated they were working part- or full-time hours were considered employed. Participants also completed a battery of self-report measures. The following measurement instruments were selected for inclusion in this analysis: the Connor–Davidson Resilience Scale (CD-RISC-10); the UW-CORR Self-Efficacy Scale for Disability Management (UW-SES); the Patient-Reported Outcomes Measurement Information System (PROMIS^®; www.nihpromis.org) for pain interference, fatigue, depression, anxiety, sleep disturbance, and physical function; and the Quality of Life in Neurological Disorders (Neuro-QoL) for cognitive function. Each measurement system is briefly described.

Resilience

Resilience was measured using the short-form (10-item) version of the CD-RISC-10 (Campbell-Sills & Stein, 2007; Connor & Davidson, 2003). This questionnaire assesses the extent to which the respondent can adapt and deal with change and stress on a 5-point scale ranging from 0 (not true at all) to 4 (true nearly all the time). An example item includes, “I am able to adapt when changes occur.” Summary scores range from 0 to 40 points, and a high score indicates better resilience. The CD-RISC-10 demonstrates adequate reliability and validity in a variety of samples including individuals with physical disabilities (Jeste et al., 2013; Kilic, Dorstyn, & Guiver, 2013). Cronbach’s alpha for the 10 items was .92 indicating excellent internal consistency.

Self-efficacy

Self-efficacy was measured using the short form (six-item) version of the UW-SES (Amtmann et al., 2012). This questionnaire assesses the extent to which a respondent is confident in managing challenges related to physical disability on a 5-point scale ranging from 1 (not at all) to 5 (completely). An example item includes, “How confident are you that you can keep your health condition or disability from interfering with your ability to deal with unexpected events?”. The final score is represented by a t score, a norm-referenced standardized score, with a mean of 50 and a standard deviation (SD) of 10. A high score indicates better self-reported self-efficacy. There is evidence of the validity and reliability of the UW-SES in older adults with physical disabilities (Amtmann et al., 2012). Cronbach’s alpha for the six items was .92 indicating excellent internal consistency.

Pain interference

Pain interference was measured using the PROMIS Pain Interference Short Form 4a (Amtmann et al., 2010; Revicki et al., 2009). This questionnaire assesses self-reported pain impact in key areas of functioning, including daily and social activities. Pain interference is measured on a 5-point scale ranging from 1 (not at all) to 5 (very much). An example item includes, “How much did pain interfere with your day to day activities?”. All the PROMIS instruments selected for this study use a 7-day recall period. The final score for all PROMIS measures is represented by the t score with a mean of 50 and an SD of 10. A high PROMIS pain score indicates worse self-reported pain interference. There is evidence of the validity and reliability of the PROMIS pain interference item bank in individuals with MS (Askew et al., 2013). Cronbach’s alpha for the four items was .97 indicating excellent internal consistency.

Fatigue

Fatigue was evaluated by questions from the PROMIS Fatigue Short Form 4a (Amtmann et al., 2011). Respondents are asked to rate their experience of fatigue and the impact of fatigue on physical, mental, and social activities. Fatigue is measured on a 5-point scale ranging from 1 (not at all) to 5 (very much). An example item includes, “I feel fatigued.” A high PROMIS fatigue score indicates worse self-reported fatigue. There is evidence of the validity and reliability of the PROMIS fatigue short form in individuals with MS (Bamer, Cook, Roddey, & Amtmann, 2009; Cook et al., 2012). Cronbach’s alpha for the four items was .96 indicating excellent internal consistency.

Depression

Depression was measured using the PROMIS Emotional Distress Depression Short Form 4a which assesses self-reported negative mood and negative views of self (Pilkonis et al., 2011). Depression is measured on a 5-point scale ranging from 1 (never) to 5 (always). An example item includes, “I felt worthless.” A high PROMIS depression score indicates worse self-reported depression. There is evidence of the validity and reliability of the PROMIS depression item bank in individuals with MS (Amtmann, Kim, et al., 2014). Cronbach’s alpha for the four items was .91 indicating excellent internal consistency.

Anxiety

Anxiety was measured using the PROMIS Emotional Distress Anxiety Short Form 4a which assesses self-reported fear, anxious misery, hyperarousal, and somatic symptoms related to arousal (Pilkonis et al., 2011). Anxiety is measured on a 5-point scale ranging from 1 (never) to 5 (always). An example item includes, “My worries overwhelmed me.” A high PROMIS anxiety score indicates worse self-reported anxiety. Cronbach’s alpha for the four items was .89 indicating good internal consistency.

Sleep disturbance

Sleep disturbance was measured using the PROMIS Sleep Disturbance Short Form 4a which assesses both self-reported sleep quality and sleep intensity (Buysse et al., 2010; Yu et al., 2012). An example item includes, “My sleep quality was . . .,” which is measured on a reverse-coded 5-point scale ranging from 5 (very poor) to 1 (very good). A high PROMIS sleep score indicates worse self-reported sleep quality. There is evidence of the validity and reliability of the PROMIS sleep disturbance short form in individuals with MS and SCI (Fogelberg, Vitiello, Hoffman, Bamer, & Amtmann, 2015). Cronbach’s alpha for the four items was .86 indicating good internal consistency.

Physical function

Physical function was measured using the PROMIS Short Form version 1.0 Physical Function Samples with Mobility Aid Users 11a (Rose, Bjorner, Becker, Fries, & Ware, 2008). Physical function is a measure of self-reported capability of one’s upper and lower extremities and of activities of daily living. Physical function is measured on a 5-point scale ranging from 5 (without any difficulty) to 1 (unable to do). An example item includes, “Are you able to walk a block on flat ground?”. A high PROMIS physical function score indicates better self-reported physical function. There is evidence of the validity and reliability of the PROMIS physical function item back in individuals with spinal disorders (Hung et al., 2014) and individuals with MS (Cook et al., 2009). Cronbach’s alpha for the 11 items was .92 indicating excellent internal consistency.

Cognitive function

Cognitive function was assessed using the Neuro-QoL Measures Version 2.0 Applied Cognition Function Short Form (Gershon et al., 2012). The Neuro-QoL measure of self-reported cognitive function includes eight items measuring both executive function and general concerns for individuals with neurological disorders. Some items in the cognitive function subdomains use “In the past 7 days,” measured on a 5-point scale ranging from 5 (never) to 1 (very often), as context, whereas others use the lead-in phrase “How much difficulty do you currently have,” measured on a 5-point scale ranging from 5 (none) to 1 (cannot do). An example item includes, “How much difficulty do you currently have learning new tasks or instructions?”. The final score is represented by a t score with a mean of 50 and an SD of 10. A high Neuro-QoL cognitive function score indicates better self-reported cognitive function. There is evidence of the validity and reliability of the Neuro-QoL short forms in individuals with MS (Miller et al., 2015). Cronbach’s alpha for the eight items was .94 indicating excellent internal consistency.

Data Analysis

Analyses were carried out using the Statistical Package for the Social Sciences (SPSS) Statistics^® version 23.0. First, data were screened for violation of statistical assumptions. Second, data were analyzed using multiple logistic regression with sequential predictor entry. Sequential predictor entry specifically allows for testing incremental improvement in model fit as predictors are added to the model. For ease of interpretation, gender (1 = male, 0 = female), marital status (1 = married/cohabitating, 0 = not married/cohabitating), race (1 = White, 0 = not White), education level (1 = college, 0 = no college), and disability benefits (1 = yes, 0 = no) were dummy coded, and age and resilience were standardized. Chronological age, gender, marital status, race, education level, and disability benefits were entered in Block 1; pain interference, fatigue, depression, anxiety, sleep disturbance, physical function, and cognitive function in Block 2; and resilience and self-efficacy in Block 3 with employment status (1 = employed, 0 = not employed) as the outcome.

Results

Univariate Analysis

The sociodemographic and symptom characteristics for the entire sample and stratified by employment status are summarized in Table 1. The average age of participants was about 54 years (SD = 9.02 years). About 63% of the participants were women. For statistical purposes, racial categories were collapsed into White (86%) and not White (14%). Individuals reported a diagnosis of MS (36%) or SCI (39%). Approximately 64% were married or living with a partner. In terms of education level, 78% had a college degree or had attended college. Most participants (61%) were receiving some type of disability cash assistance. Finally, about 33% of participants indicated that they were employed full or part time, whereas 67% reported they were not employed at the time of the survey.

Table 1.

Sociodemographic and Symptom Characteristics by Employment Status.

Measure	Total	Employed	Not employed
Sample size	882	291	591
Gender
Male	328 (37)	105 (36)	223 (38)
Female	554 (63)	186 (64)	368 (62)
Race
White	762 (86)	264 (91)	498 (84)
Not White	120 (14)	27 (9)	93 (16)
Diagnosis
MS	320 (36)	112 (39)	208 (35)
SCI	274 (31)	88 (30)	186 (32)
MD	211 (24)	74 (25)	137 (23)
PPS	77 (9)	17 (6)	60 (10)
Marital status
Married/Cohabitating	565 (64)	205 (70)	360 (61)
Not married/Cohabitating	317 (36)	86 (30)	231 (39)
Education level
College	689 (78)	259 (89)	430 (73)
No college	193 (22)	32 (11)	161 (27)
Disability benefits
Yes	537 (61)	58 (20)	470 (81)
No	345 (39)	233 (80)	112 (19)
Age	53.52 ± 9.02	50.66 ± 9.85	54.92 ± 8.23
Pain	56.15 ± 9.98	52.66 ± 8.93	57.87 ± 10.02
Fatigue	55.74 ± 10.77	53.49 ± 10.05	56.84 ± 10.95
Depression	52.57 ± 8.97	50.50 ± 8.32	53.58 ± 9.12
Anxiety	51.50 ± 9.43	50.66 ± 8.78	51.91 ± 9.71
Sleep	52.12 ± 8.45	50.57 ± 7.70	52.88 ± 8.70
Physical function	37.13 ± 11.10	41.73 ± 11.38	34.87 ± 10.24
Cognition	48.78 ± 9.50	51.33 ± 8.69	47.52 ± 9.59
Resilience	−0.06 ± 1.01	0.16 ± 0.90	−0.17 ± 1.05
Self-efficacy	47.88 ± 9.66	50.23 ± 9.14	46.72 ± 9.71

Note. Values are M ± SD or as n (%). MS = multiple sclerosis; SCI = spinal cord injury; MD = muscular dystrophy; PPS = post-polio syndrome.

Pearson chi-square values and point-biserial correlations among all variables are given in Table 2. Preliminary analyses showed no violations to the absence of multicollinearity and singularity (r ≥ .90; Tabachnick & Fidell, 2000). In addition, the variance inflation factor (VIF) for each variable was in the range of 1.0 to 3.0. Independence was assumed for this sample. Therefore, the assumptions for logistic regression were found tenable.

Table 2.

Pearson Chi-Square Values and Point-Biserial Correlations (N = 882).

Measure	M	(SD)	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.	14.	15.	16.
Outcomes
1. Employment status	0.33	(0.47)	—
Block 1 predictors
2. Sex	0.37	(0.48)	0.23	—
3. Marital status	0.64	(0.48)	7.70**	4.82*	—
4. Race	0.86	(0.34)	6.92**	0.47	11.92***	—
5. Education level	0.78	(0.41)	30.11***	8.43**	4.60*	1.86	—
6. Disability benefits	0.61	(0.49)	305.84***	7.59**	58.08***	13.03***	19.55***	—
Block 2 predictors
7. Age	53.52	(9.02)	−0.22***	−0.04	0.10**	0.05	−0.03	0.14***	—
8. Pain	56.15	(9.98)	−0.25***	0.00	−0.06	−0.11***	−0.06	0.29***	0.13***	—
9. Fatigue	55.70	(10.77)	−0.15***	−0.18***	0.01	−0.03	0.02	0.15***	0.05	0.52***	—
10. Depression	52.57	(8.97)	−0.16***	−0.03	−0.16***	−0.02	−0.03	0.21***	−0.02	0.40***	0.47***	—
11. Anxiety	51.50	(9.43)	−0.06	−0.10**	−0.11***	−0.03	0.02	0.10**	−0.02	0.37***	0.46***	0.74***	—
12. Sleep	52.12	(8.45)	−0.13***	−0.09**	−0.10**	−0.07*	−0.04	0.16***	0.01	0.42***	0.47***	0.36***	0.33***	—
13. Physical function	37.13	(11.10)	0.29***	−0.24***	0.23***	0.08**	0.13***	−0.41***	−0.07*	−0.30***	−0.07*	−0.26***	−0.14***	−0.15***	—
14. Cognition	48.78	(9.50)	0.19***	0.17***	0.05	0.02	0.05	−0.18***	−0.10**	−0.31***	−0.52***	−0.43***	−0.46***	−0.29***	0.03	—
Block 3 predictors
15. Resilience	−0.06	(1.01)	0.15***	0.06	0.09**	0.02	0.10**	−0.14***	0.05	−0.25***	−0.36***	−0.58***	−0.50***	−0.28***	0.15***	0.45***	—
16. Self-efficacy	47.88	(9.66)	0.17***	0.01	0.09**	−0.14	0.05	−0.20***	−0.06	−0.39***	−0.45***	−0.56***	−0.48***	−0.32***	0.30***	0.44***	0.69***	—

Note. Employment dummy coded with 1 = employed, 0 = not employed; sex dummy coded with 1 = male, 0 = female; marital status dummy coded with 1 = married/cohabitating, 0 = not married/cohabitating; race dummy coded with 1 = White, 0 = not White; education level coded with 1 = college, 0 = no college; disability benefits dummy coded with 1 = yes, 0 = no.

p < .05. **p < .01. ***p < .001.

Of the demographic variables, only gender was not significantly correlated with employment status. There was a very weak association between marital status and race with the outcome, χ²(1) = 7.70, p < .01 and χ²(1) = 6.92, p < .01, Cramer’s V = 0.09. There was a weak association between education and employment, χ²(1) = 30.11, p < .001, Cramer’s V = 0.19, and a moderate association between age and employment, r_pb = −0.22, p < .001. The association between disability benefits and employment was very strong, χ²(1) = 305.84, p < .001, Cramer’s V = 0.59, which could mean that the variables are measuring the same concept (Tabachnick & Fidell, 2000).

The associations between all of the symptom characteristics and the outcome were statistically significant except for anxiety. There was an inverse relationship between pain and employment, r_pb = −.25, p < .001, and a positive relationship between physical function and employment, r_pb = .29, p < .001, both of moderate strength. There were weak relationships between fatigue, depression, sleep, cognition, and employment, r_pb = −.15, p < .001; r_pb = −0.16, p < .001; r_pb = −.13, p < .001; and r_pb = .19, p < .001, respectively.

Finally, the associations between the psychological factors resilience and self-efficacy with employment were also statistically significant yet weak, r_pb = .15, p < .001 and r_pb = .17, p < .001.

Multivariate Analysis

A multiple logistic regression with sequential predictor entry was used to predict employment status using N = 882 individuals with long-term physical disabilities (see Table 3). A test of the model with chronological age, gender, marital status, race, education level, and disability benefits against the null model with no predictors was significant, χ²(6) = 363.75, p < .001, Nagelkerke pseudo R² = 0.47 (correct classification hit rate of 81%, which was better than the null model’s hit rate of 67%).

Table 3.

Model Fit Results for Employment Status (N = 882).

Model	χ² change	χ² total	Pseudo R²	HR
Block 1	—	363.75***	0.47	0.81
Block 2	15.98*	381.07***	0.49	0.82
Block 3	4.03	385.14***	0.49	0.83

Note. Block 1 chi-square change test df = 6, Block 2 df = 13, Block 3 df = 15. HR = hit rate.

p < .05. **p < .01. ***p < .001.

A test of Block 2 with pain interference, fatigue, depression, anxiety, sleep disturbance, physical function, and cognitive function added to the model against the previous model with only the sociodemographic characteristics was significant, χ²(7) = 15.98, p < .05, indicating these predictors uniquely distinguish between individuals who are employed from those who are not employed above and beyond the sociodemographic characteristics. A test of the model fit with these 13 predictors was significantly better than the null model, χ²(13) = 379.73, p < .001, Nagelkerke pseudo R² = 0.49 (correct classification hit rate increased to 82%).

Finally, a test of Block 3 with resilience and self-efficacy added to the model against the previous model with the sociodemographic and symptom characteristics was not significant, χ²(2) = 4.03, p = .13, indicating these predictors do not uniquely distinguish between individuals who are employed from those who are not employed above and beyond the previous factors. However, a test of the full model fit with all 15 predictors was significantly better than the null model, χ²(15) = 383.76, p < .001, Nagelkerke pseudo R² = 0.49 (correct classification hit rate increased slightly to 83%).

For brevity, only the coefficient estimates from the final model with all predictors entered are interpreted here (see Table 4). Model results showed that the intercept was not significantly different from zero (i.e., the mean predicted probability was not significantly different from 50%). The log-odds of being employed across the sample (holding all predictors constant) was b = −0.96 (SE = 1.73), Wald(1) = 0.31, p = .58, odds ratio (OR) = 0.38.

Table 4.

Log-Odds for the Associations Between Predictor Variables and Employment Status (N = 882).

	Block 1		Block 2		Block 3
	b	e ^b	b	e ^b	b	e ^b
Coefficients
Intercept	−0.18	0.84	−1.82	0.16	−0.96	0.38
Gender	0.30	1.35	0.32	1.37	0.30	1.35
Marital status	−0.40	0.68	−0.46*	0.63	−0.46*	0.63
Race	0.33	1.40	0.28	1.32	0.24	1.28
Education level	0.95***	2.60	0.87**	2.39	0.83**	2.30
Disability benefits	−2.93***	0.05	−2.69***	0.07	−2.71***	0.07
Age	−0.44***	0.65	−0.40***	0.67	−0.43***	0.65
Pain			−0.01	0.99	−0.02	0.99
Fatigue			−0.00	1.00	−0.00	1.00
Depression			−0.02	0.98	−0.01	0.99
Anxiety			0.03	1.03	0.03*	1.03
Sleep			0.00	1.00	0.00	1.00
Physical function			0.02	1.02	0.02	1.02
Cognition			0.03*	1.03	0.01	1.02
Resilience					0.29*	1.33
Self-efficacy					−0.02	0.98

Note. Employment dummy coded with 1 = employed, 0 = not employed; gender dummy coded with 1 = male, 0 = female; marital status dummy coded with 1 = married/cohabitating, 0 = not married/cohabitating; race dummy coded with 1 = White, 0 = not White; education level coded with 1 = college, 0 = no college; disability benefits dummy coded with 1 = yes, 0 = no.

p < .05. **p < .01. ***p < .001.

Chronological age was uniquely predictive of employment after controlling for the other predictors, b = −0.43 (SE = 0.10), Wald(1) = 19.86, p < .001, OR = 0.65. For every SD increase in age, we expect a .43 log-odd unit decrease in employment holding all other predictors constant. Two other ways to interpret this is using ORs and predicted probabilities. Individuals who were one SD above average on age were .65 times less likely to be employed. Computing the predicted probabilities based on the model estimates provides a clearer interpretation (see Peng, Lee, & Ingersoll, 2002, for a discussion of interpreting and reporting logistic regression results). Individuals who were at least one SD above average on age had a 20% predicted probability of being employed, and those who were at least one SD below average on age had a 37% predicted probability of being employed. Marital status also was uniquely predictive of employment, b = −0.46 (SE = 0.22), Wald(1) = 4.48, p < .05, OR = 0.63. Individuals who were married had a 19% predicted probability of employment. Education was uniquely predictive of employment, b = 0.83 (SE = 0.25), Wald(1) = 10.86, p < .01, OR = 2.30. Individuals who had attended college had a 47% predicted probability of employment. Disability benefits were uniquely predictive of employment, b = −2.70 (SE = 0.22), Wald(1) = 156.55, p < .001, OR = 0.07. Individuals who received disability benefits had a 3% predicted probability of being employed.

Of the symptom variables, only anxiety was uniquely predictive of employment, b = 0.03 (SE = 0.02), Wald(1) = 3.97, p < .05, OR = 1.03. Individuals who were at least one SD above the mean on anxiety had a 28% predicted probability of being employed, and those who were at least one SD below the mean on anxiety had a 27% predicted probability of being employed.

Finally, resilience was uniquely predictive of employment status, b = 0.29 (SE = 0.14), Wald(1) = 4.00, p < .05, OR = 1.33. Individuals who were at least one SD above the mean on resilience had a 34% predicted probability of being employed, and those who were one SD below the mean on resilience had a 22% predicted probability of being employed. Self-efficacy, however, was not significant, b = −0.02 (SE = 0.02), Wald(1) = 1.40, p = .24, OR = 0.98.

Discussion

The purpose of the study was to examine whether resilience and self-efficacy were significant predictors of employment in adults with long-term physical disabilities after controlling for the effects of sociodemographic and symptom covariates. Results of a multiple logistic regression with sequential predictor entry indicated that resilience was uniquely predictive of employment status after controlling for covariates. This finding, in part, supports our hypothesis that individuals higher on resilience are more likely to be employed. Self-efficacy, however, was not uniquely predictive of employment status. It is also important to note that the average contribution of both resilience and self-efficacy did not result in a significant improvement to the predictive power of the regression model.

One possible explanation for these findings is that although the shared variance between self-efficacy and the outcome was low, r_pb = 0.17, p < .001, there was overlap among self-efficacy and resilience, depression, anxiety, and fatigue (r = 0.69, p < .001; r = −0.56, p < .001; r = −0.48, p < .001; r = −0.45, p < .001, respectively) so that self-efficacy may not have contributed anything unique to employment status after accounting for the shared variance with these predictors. Another possible explanation is that disease management self-efficacy as operationalized in the present study may not be a strong predictor of employment status in this population.

These findings do highlight, however, the importance of using multivariate models to identify the range of factors that in combination may be associated with employment for people with disabilities. Even given our findings in the model on the contribution of resilience and self-efficacy to employment status, we hesitate to recommend that future studies examine the association among resilience and self-efficacy and employment status in separate models. In the literature, self-efficacy and resilience are frequently examined together. For example, in a meta-analysis of demographic and psychological variables related to resilience in nonclinical populations, of all the protective factors examined in this study, self-efficacy was found to have the strongest relationship with resilience (Lee et al., 2013). A review of resilience and physical illness found a significant relationship between self-efficacy and resilience across disease categories (Stewart & Yuen, 2011).

In examining the combined effects of these two constructs with regard to employment, we recommend that researchers use a domain-specific measure of self-efficacy. In the previous studies that have found evidence supporting self-efficacy as a predictor of employment for people with disabilities, self-efficacy has been measured in a domain related to work. For example, there has been evidence for the relationship between employment and career search efficacy (Regenold, Sherman, & Fenzel, 1999), the self-efficacy of job seeking skills (Hergenrather, Rhodes, Turner, & Barlow, 2008), and work behavior self-efficacy (O’Sullivan, Strauser, & Wong, 2012). As mentioned previously, the measure of self-efficacy employed in the current study may have had limited predictive value because all of the items were related to the domain of disease management.

This study yielded other significant findings. Education was uniquely predictive of employment in this study. Not surprisingly, individuals with higher levels of education were more likely to be employed, a finding consistent in both MS and SCI populations (J. Krause & Reed, 2009; Ottomanelli & Lind, 2009; Sweetland et al., 2012). Other sociodemographic factors uniquely predictive of employment in this study were age and marital status. Higher age has been associated with unemployment for people with physical disabilities. Although we found those who were married or cohabitating were less likely to be employed, findings from previous studies are mixed (Botticello et al., 2012; Saunders, Leahy, McGlynn, & Estrada-Hernández, 2006). Next, although receiving disability cash benefits (i.e., Supplemental Security Income or Social Security Disability Insurance) was uniquely predictive of employment status, this factor may be tautological with unemployment in that individuals who could not work because of a disability were receiving assistance from these or other federal programs. Finally, even though anxiety was statistically significant, the effects of symptoms with regard to employment were miniscule and unlikely to be meaningful.

Findings from this study should be interpreted with caution due to a number of limitations. This is a convenience sample of individuals with MS, MD, PPS, or SCI, and it likely underrepresents individuals with more severe functional impairment. The sample also comprised mostly individuals who are women, White, older in age, and highly educated, which limits the generalizability of these findings. In addition, this study is cross-sectional and, as such, precludes inferences about the causal and directional relationships among variables. Finally, given the debate in the scholarly literature on the definition and operationalization of resilience (see, for example, Kolar, 2011; Luthar, Cicchetti, & Becker, 2000), a single-point measurement of this factor may not be representative of the patterns of the disease-related resilience process for the study participants. Therefore, these findings based on cross-sectional data should be supported by further replication and, preferably, longitudinal studies.

To our knowledge, this is the only published study examining the association of resilience, self-efficacy, and employment status in individuals with long-term physical disabilities. We found in this study preliminary evidence to suggest that rehabilitation counseling practitioners should consider the importance of a client’s resilience with respect to employment. Knowledge of sociodemographic and symptom factors in conjunction with psychometrically sound measures of resilience and work-related measures of self-efficacy may be used by rehabilitation counseling practitioners early in the vocational rehabilitation process to identify individuals with physical disabilities whose beliefs and behaviors may limit the extent to which they prepare for, obtain, or maintain employment. However, more research is needed to confirm and extend these findings. Longitudinal research is warranted to examine changes in an individual’s resilience over time and to evaluate what, if any, effect this has on an individual’s employment status. Only with an accumulation of evidence can targeted approaches to promote resilience for individuals with long-term physical disabilities with respect to employment be developed and tested.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institute on Disability, Independent Living, and Rehabilitation Research at the Administration for Community Living (HB133B080024, 90RT5023-01-00, and 90AR5026-01-00).

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Review Board at the University of Washington and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Informed Consent

Informed consent was obtained from all individual participants included in the study.

Data Availability

The data sets analyzed during the current study are available from the corresponding author on reasonable request.

References

Amtmann

Bamer

A. M.

Cook

Askew

Noonan

Brockway

(2012). University of Washington self-efficacy scale: A new self-efficacy scale for people with disabilities. Archives of Physical Medicine and Rehabilitation, 93,1757–1765.

Amtmann

Bamer

A. M.

Yorkston

K. M.

Smith

A. E.

(2014). Predictors of self-efficacy in individuals aging with a disability. Quality of Life Research, 23, 35.

Amtmann

Cook

Bamer

Roddey

Kim

Kraft

(2011). Development of a PROMIS Fatigue Short Form for use in multiple sclerosis. Multiple Sclerosis Journal, 17, S233.

Amtmann

Cook

Jensen

Chen

Choi

Revicki

. . . Lai

(2010). Development of a PROMIS item bank to measure pain interference. Pain, 150, 173–182.

Amtmann

Kim

Chung

Bamer

Askew

. . . Johnson

K. L.

(2014). Comparing CESD-10, PHQ-9, and PROMIS depression instruments in individuals with multiple sclerosis. Rehabilitation Psychology, 59, 220–229.

Askew

R. L.

Kim

Chung

Cook

K. F.

Johnson

K. L.

Amtmann

(2013). Development of a crosswalk for pain interference measured by the BPI and PROMIS pain interference short form. Quality of Life Research, 22, 2769–2776.

Bamer

A. M.

Cook

K. F.

Roddey

T. S.

Amtmann

(2009). Item selection and content validity of a multiple sclerosis fatigue short form using the PROMIS item bank. International Journal of MS Care, 11(S2), 13–14.

Botticello

Chen

Tulsky

(2012). Geographic variation in participation for physically disabled adults: The contribution of area economic factors to employment after spinal cord injury. Social Science & Medicine, 75, 1505–1513.

Burns

Boyd

Hill

Hough

Elliott

T. R.

(2010). Psychosocial predictors of employment status among men living with spinal cord injury. Rehabilitation Psychology, 55, 81–90.

10.

Buysse

D. J.

L. E.

Germain

A. L.

Stover

A. A.

Dodds

Johnston

. . . Pilkonis

P. A.

(2010). Development and validation of patient-reported outcome measures for sleep disturbance and sleep-related impairments. Sleep, 33, 781–792.

11.

Campbell-Sills

Stein

(2007). Psychometric analysis and refinement of the Connor-Davidson Resilience Scale (CD-RISC): Validation of a 10-item measure of resilience. Journal of Traumatic Stress, 20, 1019–1028.

12.

Chiu

Chan

Bishop

da Silva Cardoso

O’Neill

(2013). State vocational rehabilitation services and employment in multiple sclerosis. Multiple Sclerosis Journal, 19, 1655–1664.

13.

Connor

Davidson

(2003). Development of a new resilience scale: The Connor-Davidson Resilience Scale. Depression and Anxiety, 18, 76–82.

14.

Cook

Bamer

Roddey

Kraft

Kim

Amtmann

(2012). A PROMIS fatigue short form for use by individuals who have multiple sclerosis. Quality of Life Research, 21, 1021–1030.

15.

Cook

Dudgeon

Roddey

Johnson

Bamer

Amtmann

(2009). Adapting the PROMIS physical function outcome bank for MS AT users. Multiple Sclerosis Journal, 15(9), S53.

16.

Erickson

Lee

von Schrader

(2017). Disability statistics from the American Community Survey (ACS). Ithaca, NY: Cornell University Yang-Tan Institute. Available from www.disabilitystatistics.org

17.

Fogelberg

Vitiello

Hoffman

Bamer

Amtmann

(2015). Comparison of self-report sleep measures for individuals with multiple sclerosis and spinal cord injury. Archives of Physical Medicine and Rehabilitation, 96, 478–483.

18.

Gershon

Lai

Bode

Choi

Moy

Bleck

. . . Cella

(2012). Neuro-QOL: Quality of life item banks for adults with neurological disorders: Item development and calibrations based upon clinical and general population testing. Quality of Life Research, 21, 475–486.

19.

Hergenrather

K. C.

Rhodes

S. D.

Turner

A. P.

Barlow

(2008). Persons with disabilities and employment: Application of the self-efficacy of job-seeking skills scale. Journal of Rehabilitation, 74, 334–344.

20.

Hirsh

Molton

Johnson

Bombardier

Jensen

(2009). The relationship of chronological age, age at injury, and duration of injury to employment status in individuals with spinal cord injury. Psychological Injury and Law, 2, 263–275.

21.

Honarmand

Akbar

Kou

Feinstein

(2011). Predicting employment status in multiple sclerosis patients: The utility of the MS functional composite. Journal of Neurology, 258, 244–249.

22.

Hung

Hon

Franklin

Kendall

Lawrence

Neese

. . . Brodke

(2014). Psychometric properties of the PROMIS physical function item bank in patients with spinal disorders. Spine, 39, 158–163.

23.

Jeste

Savla

Thompson

Vahia

Glorioso

Martin

. . . Depp

(2013). Association between older age and more successful aging: Critical role of resilience and depression. The American Journal of Psychiatry, 170, 188–196.

24.

Johnson

K. L.

Bamer

Fraser

(2013). Disease and demographic characteristics associated with unemployment among working-age adults with multiple sclerosis. International Journal of MS Care, 15(S1), 44–51.

25.

Johnson

K. L.

Yorkston

K. M.

Klasner

E. R.

Kuehn

C. M.

Johnson

Amtmann

(2004). The cost and benefits of employment: A qualitative study of experiences of persons with multiple sclerosis. Archives of Physical Medicine and Rehabilitation, 85, 201–209.

26.

Julian

Vella

Vollmer

Hadjimichael

Mohr

(2008). Employment in multiple sclerosis. Exiting and re-entering the work force. Journal of Neurology, 255,1354–1360.

27.

Kilic

S. A.

Dorstyn

D. S.

Guiver

N. G.

(2013). Examining factors that contribute to the process of resilience following spinal cord injury. Spinal Cord, 51, 553–557.

28.

Kolar

(2011). Resilience: Revisiting the concept and its utility for social research. International Journal of Mental Health and Addiction, 9, 421–433.

29.

Krause

Kern

Horntrich

Ziemssen

(2013). Employment status in multiple sclerosis: Impact of disease-specific and non-disease-specific factors. Multiple Sclerosis Journal, 19, 1792–1799.

30.

Krause

Broderick

(2006). Relationship of personality and locus of control with employment outcomes among participants with spinal cord injury. Rehabilitation Counseling Bulletin, 49, 111–114.

31.

Krause

Reed

(2009). Obtaining employment after spinal cord injury: Relationship with pre- and postinjury education. Rehabilitation Counseling Bulletin, 53, 27–33.

32.

Krause

Saunders

Staten

(2010). Race-ethnicity, education, and employment after spinal cord injury. Rehabilitation Counseling Bulletin, 53, 78–86.

33.

Lee

J. H.

Nam

S. K.

Kim

A.-R.

Kim

Lee

M. Y.

Lee

S. M.

(2013). Resilience: A meta-analytic approach. Journal of Counseling & Development, 91, 269–279.

34.

Lidal

Huynh

Biering-Sørensen

(2007). Return to work following spinal cord injury: A review. Disability and Rehabilitation, 29, 1341–1375.

35.

Luthar

Cicchetti

Becker

(2000). The construct of resilience: A critical evaluation and guidelines for future work. Child Development, 71, 543–562.

36.

Miller

D. M.

Bethoux

Victorson

Nowinski

C. J.

Buono

Lai

J.-S.

. . . Cella

(2015). Validating Neuro-QoL short forms and targeted scales with persons who have multiple sclerosis. Multiple Sclerosis Journal, 22, 830–841.

37.

Minis

M. A.

Kalkman

S. J.

Akkermans

R. P.

Engels

J. A.

Huijbregts

P. A.

Bleijenberg

. . . van Engelen

(2010). Employment status of patients with neuromuscular diseases in relation to personal factors, fatigue and health status: A secondary analysis. Journal of Rehabilitation Medicine, 42, 60–65.

38.

Moore

Harding

Clarkson

Pickersgill

Wardle

Robertson

(2013). Demographic and clinical factors associated with changes in employment in multiple sclerosis. Multiple Sclerosis Journal, 19, 1647–1654.

39.

Neath

J. T.

Roessler

R. D.

McMahon

Rumrill

(2007). Patterns in perceived employment discrimination for adults with multiple sclerosis. Work, 29, 255–274.

40.

Nevala

Pehkonen

Koskela

Ruusuvuori

Anttila

(2015). Workplace accommodation among persons with disabilities: A systematic review of its effectiveness and barriers or facilitators. Journal of Occupational Rehabilitation, 25, 432–448.

41.

O’Sullivan

Strauser

D. R.

(2009). Operationalizing self-efficacy, related social cognitive variables, and moderating effects: Implications for rehabilitation research and practice. Rehabilitation Counseling Bulletin, 52, 251–258.

42.

O’Sullivan

Strauser

D. R.

Wong

A. W. K.

(2012). Five-factor model of personality, work behavior self-efficacy, and length of prior employment for individuals with disabilities: An exploratory analysis. Rehabilitation Counseling Bulletin, 55, 156–165.

43.

Ottomanelli

L. L.

Lind

(2009). Review of critical factors related to employment after spinal cord injury: Implications for research and vocational services. The Journal of Spinal Cord Medicine, 32, 503–531.

44.

Ottomanelli

L. L.

Sippel

J. J.

Cipher

Goetz

(2011). Factors associated with employment among veterans with spinal cord injury. Journal of Vocational Rehabilitation, 34, 141–150.

45.

Peng

C.-Y. J.

Lee

K. L.

Ingersoll

G. M.

(2002). An introduction to logistic regression analysis and reporting. Journal of Educational Research, 96, 3–14.

46.

Peter

Müller

Cieza

Geyh

(2011). Psychological resources in spinal cord injury: A systematic literature review. Spinal Cord, 50, 188–201.

47.

Pilkonis

P. A.

Choi

S. W.

Reise

S. P.

Stover

A. M.

Riley

W. T.

Cella

(2011). Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS^®): Depression, anxiety, and anger. Assessment, 18, 263–283.

48.

Regenold

Sherman

M. F.

Fenzel

(1999). Getting back to work: Self-efficacy as a predictor of employment outcome. Psychiatric Rehabilitation Journal, 22, 361–367.

49.

Revicki

Chen

Harnam

Cook

Amtmann

Callahan

. . . Keefe

(2009). Development and psychometric analysis of the PROMIS pain behavior item bank. Pain, 146, 158–169.

50.

Rose

Bjorner

J. B.

Becker

Fries

J. F.

Ware

J. E.

(2008). Evaluation of a preliminary physical function item bank supported the expected advantages of the Patient-Reported Outcomes Measurement Information System (PROMIS). Journal of Clinical Epidemiology, 61, 17–33.

51.

Saunders

Leahy

McGlynn

Estrada-Hernández

(2006). Predictors of employment outcomes for persons with disabilities: An integrative review of potential evidenced-based factors. Journal of Applied Rehabilitation Counseling, 37(2), 3–20.

52.

Shahrbanian

Auais

Duquette

Anderson

Mayo

(2013). Does pain in individuals with multiple sclerosis affect employment? A systematic review and meta-analysis. Pain Research & Management, 18(5), e94–e100.

53.

Silverman

Molton

Alschuler

Ehde

Jensen

(2015). Resilience predicts functional outcomes in people aging with disability: A longitudinal investigation. Archives of Physical Medicine and Rehabilitation, 96, 1262–1268.

54.

Stewart

Yuen

(2011). A systematic review of resilience in the physically ill. Psychosomatics, 52, 199–209.

55.

Sweetland

Howse

Playford

(2012). A systematic review of research undertaken in vocational rehabilitation for people with multiple sclerosis. Disability and Rehabilitation, 34, 2031–2038.

56.

Tabachnick

Fidell

L. S.

(2000). Using multivariate statistics (4th ed.). Boston, MA: Allyn and Bacon.

57.

Tan-Kristanto

Kiropoulos

(2015). Resilience, self-efficacy, coping styles and depressive and anxiety symptoms in those newly diagnosed with multiple sclerosis. Psychology, Health & Medicine, 20, 635–645.

58.

Tansey

Bezyak

Kaya

Ditchman

Catalano

(2017). Resilience and quality of life: An investigation of Kumpfer’s resilience model with persons with spinal cord injuries. Rehabilitation Counseling Bulletin, 60, 163–174.

59.

White

Driver

Warren

A.-M.

(2008). Considering resilience in the rehabilitation of people with traumatic disabilities. Rehabilitation Psychology, 53, 9–17.

60.

Buysse

Germain

Moul

Stover

Dodds

. . . Pilkonis

(2012). Development of short forms from the PROMIS™ sleep disturbance and sleep-related impairment item banks. Behavioral Sleep Medicine, 10, 6–24.