Factors Associated With Civilian Employment, Work Satisfaction, and Performance Among National Guard Members

In 2013, the U.S. Bureau of Labor Statistics (BLS) reported that there were over 21 million veterans in the U.S. population, with an overall jobless rate of 7.0%, versus 7.9% for nonveterans (1). However, veterans ages 18–24 and 25–34 had higher unemployment rates than did similarly aged civilian counterparts (20.4% versus 9.9% and 15% versus 8.2%, respectively) (1). This discrepancy is concerning, considering younger veterans’ longer-term work trajectories and elevated risk of developing a mental or substance use disorder as a result of their deployments, which might negatively affect current work status and longer-term employment outcomes (2–4). Military deployments have been found to be an independent risk factor contributing to the development of mental health and substance use conditions that impair social functioning and community reintegration of service members (4–10).

Psychiatric symptom severity and substance use are commonly associated with poor work outcomes (11–13). Users of U.S. Department of Veterans Affairs (VA) services who were diagnosed as having major depressive disorder, posttraumatic stress disorder (PTSD), and generalized anxiety or panic disorder reported diminished work role functioning (14). Similarly, lower rates of employment have been found among VA service users with diagnosed major depressive disorder (15), PTSD (15,16), or substance use disorders. Furthermore, unemployed VA service users were twice as likely as employed veterans to have a substance use disorder (15). There have been fewer studies regarding these associations in nonclinical samples of National Guard (NG) service members reentering the civilian workforce, although Erbes and colleagues (2) found in a general sample of NG members that those meeting criteria for PTSD, depression, or an alcohol use disorder had diminished work role function. In their combined sample, only the Beck Depression Inventory–Second Edition was predictive of work role functioning, whereas groups with and without psychiatric disorders differed on employment status.

Several sociodemographic factors have been linked to veteran employment, such as age (17,18), rank, education, and socioeconomic status (17). One study found lower employment among veterans discharged before the end of their service contract, followed by those who completed their service contacts; military retirees were found to have the highest rates of employment (18).

Most, although not all (2), employment research involving veterans and military service members has primarily focused on employment status rather than on work functioning; however, current employment status provides only a cursory understanding of the employment experiences within these groups. Diminished work performance, low job satisfaction, and elevated psychiatric symptoms are likely to negatively influence job retention, promotion, or future job seeking. Considerable literature supports the relationship between work outcomes (including absenteeism, presenteeism, productivity, and mental and interpersonal demands) and mental health problems (2,11–13). This is the only study that has assessed self-rated work performance and satisfaction in a general NG sample returning from an overseas military deployment.

Unlike active-duty service members, NG service members are commonly faced with employment issues immediately after deployments, a scenario that has been shown to increase stress and risk of readjustment issues (16,19). Our study examined work outcomes in a large sample of Midwestern NG members approximately six months after their overseas deployments. We examined predictors of employment status and self-rated work performance and work satisfaction. On the basis of results from prior studies, we hypothesized that age, rank, and education would predict current employment and that combat experience, mental health symptoms, and hazardous alcohol use would predict lower self-rated work performance and satisfaction.

Methods

The VA Ann Arbor Institutional Review Board approved this study with survey data collected under a waiver of written informed consent.

Data

Data were collected from service members of a Midwestern NG organization between August 2010 and July 2013, approximately six months after their return from overseas deployments (Operation Enduring Freedom, Operation Iraqi Freedom, or Operation New Dawn). Service members were recruited in person during monthly drill, by mail, or by both methods. For in-person recruitment, a study team member visited the unit during a scheduled drill weekend and briefed service members on the details of the study, emphasizing the voluntary and confidential nature of the survey. Team members distributed the survey packets and remained on site to collect completed surveys, allowing them to be completed during nonscheduled events. Surveys were mailed to soldiers who did not complete surveys during drill weekends. We used the modified Dillman (20) method for survey research, which entailed sending presurvey notification letters, followed by the survey packets. Follow-up surveys were sent two additional times to nonrespondents, followed by thank you/reminder letters, another round of survey packets mailed three weeks after, and one more round mailed seven weeks after the initial survey mailings. A total of 1,475 surveys were returned (55% response rate), 1,020 in person and 455 by mail.

Measures

The outcomes of interest were employment status, work performance, and work satisfaction; the latter two items were self-reported. Employment status was determined from survey items that queried about participants’ current work situation with the following response options: homemaker, student, maternity or paternity leave, sick leave, disability, working full-time, working part-time, unemployed—looking for work, unemployed—not looking for work, retired, or other. Categories of employed versus unemployed were based on BLS definitions for identifying employed versus unemployed persons. The BLS considers people who report having jobs as being employed, people who report being jobless, seeking jobs, and available for work to be unemployed, and people who report being neither employed nor unemployed as not in the labor force (1). Because of the heterogeneity (for example, homemakers, retirees, and people with disabilities) and low percentage of participants in the sample who identified as not in the labor force (14%), we decided to remove this category from analyses. Of note, 216 (61%) of the 355 NG participants who identified as students also checked another option for employment, with 136 (38%) reporting full- or part-time work and 80 (23%) indicating that they were seeking work, whereas 139 (39%) did not mark additional work categories and were considered out of the workforce.

Work performance was assessed with a question from the Health and Performance Questionnaire that asked respondents to rate their job or school performance over the past four weeks (21). Work satisfaction was assessed with the following item developed for the study: “How would you rate your job/school satisfaction in the past 4 weeks?” Both questions were rated on an 11-point scale, with lower ratings reflecting diminished performance and satisfaction.

Independent variables included age, gender, education, income, rank, marital status, and whether individuals had a child at home. Additional measures include barriers to care, combat exposure, psychiatric symptoms, alcohol use, general medical health, and perceived social resources.

The Hoge adaptation of the Perceived Stigma and Barriers to Care for Psychological Problems was used to assess barriers to care, because not receiving care when needed may negatively affect employment outcomes (22–24). The scale consists of 16 items measured on a 5-point Likert scale (1, strongly disagree; 5, strongly agree) that address concerns about seeking mental health care, with Cronbach’s alpha ranging between .75 and .95 (25).

Respondents’ perception of social resources was measured with the 12-Item Interpersonal Support Evaluation List (ISEL-12) (26), which assesses perceptions about the availability of social resources. It contains three subscales measuring perceived availability of material aid, emotional support, and socially shared activities, as well as a total score that provides an estimate of overall support. Items are rated on a 4-point scale ranging from “definitely true” (3) to “definitely false” (0). The ISEL-12 has excellent internal consistency (α=.88–.90), as well as convergent and predictive validity (26).

Items taken from the Postdeployment Health Assessment (PDHA) were used to assess combat exposure (27). The PDHA is a two-part comprehensive health screen required for soldiers deployed in support of any contingency operation outside the continental United States for longer than 30 days. This screen was designed to proactively identify health concerns and provide referrals for follow-up treatment for deployment-related health issues. Combat exposure items inquire about injuries sustained during deployment, engagement in direct combat, discharge of weapons, and subjective feelings of danger. Endorsement of any of these items for “any or most recent” deployments were used to determine combat exposure for this study.

The PTSD Checklist–Military version (PCL-M) (28) was used to assess symptoms of PTSD. The PCL-M is a self-report measure that screens 17 DSM-IV symptoms of PTSD and asks respondents to rate, using a 5-point scale (1, not at all; 5, extremely), the degree to which they were “bothered” by specified military experiences. Scores above 50 represent a positive screen for PTSD in military samples. The PCL-M has shown excellent internal consistency in use with Vietnam and Persian Gulf veterans (r=.94–.97) (29).

The seven-item Generalized Anxiety Disorder scale (GAD-7) was used to measure symptoms of anxiety (30), with scores ranging from 0 to 21 and “cut scores” for mild, moderate, and severe anxiety. Scores above 10 were used as a positive screen on the GAD-7. The scale has demonstrated good psychometric properties, good operating characteristics for detection and severity ratings of panic and social anxiety disorders (31).

The nine-item Patient Health Questionnaire (PHQ-9) (32) was used to assess symptoms of depression. Total scores range from 0 to 27, with higher scores suggestive of increased psychopathology and lower functioning. Scores above 10 were used as a positive screen on the PHQ-9. A meta-analysis found pooled sensitivity of 80% and specificity of 92% (33), and the PHQ-9 has demonstrated good internal consistency and convergent validity with the Center for Epidemiologic Studies Depression Scale (34).

The Alcohol Use Disorders Identification Test (AUDIT) (35) indexes average alcohol consumption and binge drinking over the past three months, with higher scores reflecting increased problematic alcohol use. The Alcohol Use Disorders Identification Test–Consumption Questionnaire (AUDIT-C) uses the first three items from the AUDIT and was used in this study to assess risk of problematic drinking. Hazardous alcohol use was determined with score cutoffs of 3 for women and 4 for men. A recent review supports the psychometric properties of the AUDIT-C for identifying hazardous alcohol use at these cutoffs (36).

The physical health component subscale (PCS) of the 12-Item Short Form (SF-12) (37) was used to measure physical functioning. This six-item component of the SF-12 provides a summary of physical functioning over the past four weeks (37). Items assess participants’ opinion of their general medical health and the extent to which it limits their day-to-day activities. PCS scores are standardized with a mean±SD of 50±10, with higher scores indicative of better health. The reliability and validity of the SF-12 in general and clinical populations have been well documented (37).

Statistical Analysis

Descriptive statistics were used to summarize sample characteristics. Unadjusted predictors were assessed in bivariate analyses on the primary outcomes, with t tests used for continuous variables and chi square used for dichotomous variables. We included demographic characteristics and significant bivariate predictors into the adjusted multivariate models. Multivariable logistic regression was used to assess the effects of these variables on employment status by calculating odds ratios with 95% confidence intervals. For employed NG participants, predictors of self-rated work performance and satisfaction were determined with multiple linear regressions. Significant bivariate predictors were included in each adjusted model. Age, gender, service rank, and level of education were included as predictors in all adjusted analyses. Psychiatric symptom load was created to examine the impact of one, two, or three positive psychiatric symptom screens (on the PHQ-9, GAD-7, and PCL-M) on outcome variables. Psychiatric symptoms were used as categorical variables (positive or negative screen) for all analyses. All analyses were conducted with SAS statistical software.

Results

The study sample consisted of 1,151 NG members, of whom 839 (73%) reported being employed and 312 (27%) reported being unemployed. The sample composition was 84% (N=967) Caucasian, 6% (N=64) African American, 4% (N=45) Hispanic, 3% (N=31) multiethnic, and 2% (N=27) “other.” Most of our sample (74%, N=848) reported combat experience (Table 1).

Multivariable logistic regression was used to predict employment status among NG respondents. In bivariate analyses, factors associated with increased odds of being employed were higher age, education, rank, and income. Being male, being in a committed relationship, having a child at home, and reporting greater perceived social resources contributed to greater odds of being employed, whereas hazardous alcohol use and positive screens for depression and anxiety reduced odds of being employed (Table 2). The adjusted model revealed that higher rank and income increased odds of employment, whereas hazardous alcohol use reduced odds of being employed (Table 2), classifying 77% cases correctly. Likelihood of employment increased for noncommissioned officers (adjusted odds ratio [AOR]=1.51) and commissioned officers (AOR=1.91), relative to enlisted respondents, but decreased for respondents reporting hazardous alcohol use (AOR=.73).

For employed NG respondents (N=787), bivariate analyses revealed that self-rated work performance was inversely related to age, rank, combat experience, psychiatric symptom severity, hazardous alcohol use, and barriers to care and positively associated with physical functioning and perceived social resources. Adjusted analysis revealed an inverse relationship between self-rated work performance and psychiatric symptom load (two or more positive psychiatric symptom screens), and greater physical functioning and greater perceived social resources were associated with better work performance (R²=.19, F=16.19, df=11 and 775, p<.001) (Table 3).

Multiple linear regressions were used to investigate predictors of self-rated work satisfaction among employed NG respondents (N=786). Bivariate analyses found that ratings of work satisfaction were inversely related to age, rank, combat experience, psychiatric symptom severity, alcohol use, and barriers to care and were positively associated with physical functioning and perceived social resources. Adjusted analyses revealed inverse associations between self-rated work satisfaction and rank, hazardous alcohol use, barriers to care, and psychiatric symptom load. In addition, the variable for perceived social resources was associated with greater self-rated work satisfaction (R²=.16, F=11.34, df=13 and 772, p<.001) (Table 4).

Discussion

To our knowledge, this is the largest study to investigate predictors of employment-based outcomes in a general sample of NG members six months after returning from a deployment. Consistent with previous studies (17,18) and national reports (1), we found elevated rates of unemployment among junior-ranking enlisted service members, who made up approximately two-thirds (61%) of the unemployed respondents. General rates of employment in military and veteran samples are influenced by multiple factors (such as populations sampled, economic factors, and time frame), and we found lower rates than found in some studies (2,17,18) but not others (14,38).

Consistent with previous studies, we did not find an effect of psychiatric symptoms on employment status (2,18), although Horton and colleagues (18) found that veterans screening positive for depression or anxiety, who were routinely retired from the military, were more likely to be unemployed. These studies used nonclinical samples from active-duty, NG, and Reserve forces (2). NG members have been shown to experience significant increases in mental health symptoms between three and 12 months after a combat deployment (9). As such, NG and Reserve service members may not initially experience difficulty returning to civilian employment but may struggle to stay employed as a result of absenteeism and presenteeism as symptoms increase. This was not supported in a study by Erbes and colleagues (2) that examined employment status one year postdeployment.

Greater psychiatric symptom load was associated with lower self-ratings of work performance and satisfaction in this study, which is consistent with other research showing reduced work functioning among individuals with more severe psychiatric symptoms (2,13,14). Diminished perceived work performance and satisfaction coupled with elevated psychiatric symptoms likely exert a negative influence on job retention, promotion, or future employment seeking. This is concerning for NG members who do not receive VA services and for those who are eligible for but are not aware of, or are not able to access, VA vocational services. Veteran recognition of vocational need represents the largest delay in service entry, with many participating in mental health or substance abuse treatment programs to address their vocational needs (39). This finding highlights the need to screen for vocational needs and to increase access to vocational services for veterans prior to developing mental and substance-related issues.

Hazardous alcohol use was high in our sample (48%) and had a negative effect on employment status. Despite a clear link between substance use disorders and poor employment outcomes, alcohol use had not previously been associated with employment status (2,18) but has been shown to negatively affect work satisfaction. Diminished work satisfaction, coupled with reintegration stressors, may reduce job retention among employed NG members over time. This finding is consistent with studies reporting poor postdeployment readjustment after rapid return to work (16,19).

Our results also underscore the value of perceived social resources and physical functioning on work outcomes among employed respondents. These findings are consistent with research showing that diminished physical functioning is associated with work functioning (12,14,40,41). Burnett-Zeigler and colleagues (17) found that although physical functioning was not a predictor of employment status, it was predictive of whether someone was employed full-time. The link between physical functioning and work performance is intuitive, considering that soldiers returning home injured will likely experience difficulties performing their civilian jobs.

There were several limitations to this study. The response rate was only 55%, and we were unable to assess differences between study respondents and nonrespondents. However, this response rate is typical of surveys of NG personnel (9,17), and our sample was fairly consistent with the NG demographic characteristics overall (42). Another limitation involved our two self-rated outcome variables (work performance and work satisfaction). Subjective indices are influenced by a variety of transient factors, such as mood, personality, and response set. Stability of these indices and their effects has yet to be determined and requires replication to support our findings. In addition, the psychiatric symptom measures that made up our composite variable of psychiatric symptom load are likely influenced by worry and rumination, which may be elevated among service members reintegrating with civilian roles. Future studies are needed to elucidate the effects of psychiatric symptom load on work outcomes in this population.

Conclusions

Our findings underscore the importance of assessing alcohol use among unemployed NG members and of aligning treatment options that can facilitate awareness of vocational needs while addressing hazardous alcohol use. For employed service members reporting work-related difficulties or low satisfaction with their current employment, VA service providers should also consider the assessment of and interventions targeting physical functioning and psychiatric symptom load. Within the VA, many of the symptom screens used in this study to determine psychiatric symptom load are already in use and could be used to prompt a screen to assess for vocational need. Furthermore, early engagement in mental health care may play an important role in increasing work performance and satisfaction, and subsequently in improving long-term employment outcomes. Future studies are needed to clarify the causal links between these predictors and employment outcomes.