Sample and procedures

The data for this study were obtained from National Guard soldiers from two brigades anonymously surveyed between 2005 and 2007 at approximately 3 months and 12 months after their deployments to Iraq. The survey procedures have been previously described. ^{Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2,Reference Hoge, Castro, Messer, McGurk, Cotting and Koffman8,Reference Kim, Thomas, Wilk, Hoge and Castro13} Recruitment briefings occurred in garrison at times convenient to the brigades. In total 94–99% of the soldiers who were present for the recruitment briefings gave consent. The soldiers that did not attend the briefings were absent for reasons such as medical or personal leave, training and other duties. Data were collected under a Walter Reed Army Institute of Research Institutional Review Board approved protocol. At 3 months 3127 National Guard soldiers gave content and at 12 months 2449 National Guard soldiers gave consent (5576 in total). Unit personnel indicated 8848 National Guard soldiers across the two brigades and times (approximately 5378 respondents at 3 months post-deployment and approximately 3470 respondents at 12 months post-deployment were available for the data collections). Therefore, the response rate at 3 and 12 months post-deployment is 58% (3127/5378) and 71% (2449/3470) respectively. Overall, the response rate is 63% (5576/8848), which is similar to the rates reported elsewhere. ^{Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2,Reference Kim, Thomas, Wilk, Hoge and Castro13,Reference Wilk, Thomas, McGurk, Riviere, Castro and Hoge14} We restricted our sample to 4034 National Guard soldiers who reported that they had deployed only one time to Iraq (2539 at the first and 1495 at the second time point). This sample is a subset of the National Guard sample used by Thomas et al ^{Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2} and, as reported by these authors, the demographics of National Guard soldiers in this sample are comparable with those of the population of National Guard soldiers that were screened after Iraq deployments.

Measures

Statistical analysis

All analyses were performed with SPSS software (Version 13.0) for Windows. We used chi-squared analyses to assess the association between depression and PTSD rates at the two time points across the categories of the demographic variables and the four National Guard specific variables. For the three National Guard specific variables with Likert response scales the responses were collapsed to three levels (e.g. strongly agree and agree were combined). Separate logistic regression models are estimated per time point to assess the effect of the National Guard specific variables on depression and PTSD controlling for demographic variables and combat exposure. We also assessed whether the National Guard specific variables, the demographic variables and combat exposure varied in their associations with depression and PTSD at 12 months compared with 3 months post-deployment with predictor/time interactions. Preliminary analyses revealed that marital and parental status were unrelated to depression or PTSD so they were not included in the final logistic regression models. No substitutions were made for missing data; for both the chi-squared tests and logistic regression analyses missing data were deleted case-wise.

Main findings

A previous study ^{Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2} showed that the rates of depression and PTSD among active component soldiers remained either fairly stable or increased modestly between 3 and 12 months post-deployment. However, the rates of these two disorders increased substantially more in National Guard soldiers. That study and a previous study ^{Reference Milliken, Auchterlonie and Hoge1} offered explanations for the discrepancies in the rate increases in National Guard soldiers. We believe that our study's findings offer empirical evidence that the National Guard specific variables examined here are risk factors for developing post-deployment mental health problems after adjusting for combat exposure and demographic variables. They may explain why National Guard soldiers fare so much worse than active component soldiers at 12 months.

Consistent with both the civilian and military literature, both unemployment ^{Reference McKee-Ryan, Song, Wanberg and Kinicki18–Reference Zatzick, Marmar, Weiss, Browner, Metzler and Golding22} and subjective financial hardship ^{Reference Kessler, Turner and House21,Reference Butterworth, Rodgers and Windsor23} are associated with poor psychological health. Financial hardship has sometimes been conceptualised as a mediator in the relationship between unemployment and poor mental health (not PTSD). ^{Reference Kessler, Turner and House21} That is, unemployment only leads to poor mental health when it results in financial hardship. Other studies have found financial hardship to be independently associated with depression. ^{Reference Butterworth, Rodgers and Windsor23} Likewise, this study shows that at 12 months postdeployment, both financial hardship and unemployment are independently associated with both depression and PTSD. Financial hardship also increases the likelihood, at 3 months, of screening positive for depression and PTSD. Unemployment, however, did not increase the likelihood of screening positive for PTSD at 3 months. A single item was used in this study to measure financial hardship, unlike the three- or four-item indices used in previous studies. Notwithstanding, financial hardship appears to be an important variable in the development of both depression and PTSD. We do not know whether the National Guard soldiers who lost their jobs were unemployed at the time of the survey or had experienced unemployment prior to it. It may be that, in line with Kessler et al, ^{Reference Kessler, House and Turner19} current as well as prior unemployment negatively affected mental health.

It has been established that social support has salutary effects on the mental health of both military service personnel and civilians. ^{Reference Vinokur, Caplan and Williams20,Reference Pietrzak, Johnson, Goldstein, Malley and Southwick24–Reference Laffaye, Cavella, Drescher and Rosen26} Consistent with that literature, we found that National Guard soldiers who indicated that their employers did not support their military affiliation were more likely to meet criteria for depression (at both time points) and PTSD (at 3 months). An interesting finding was that National Guard soldiers who had a neutral response as to whether their employer supports their military affiliation were also more likely, at 3 months, to meet criteria for depression and PTSD. Perhaps the neutral response reflects ambivalence as to whether employers were supportive, and this ambivalence may be just as harmful as the perception that employers are unsupportive. Even more surprising is that, at 3 months, the likelihood of screening positive for PTSD was marginally greater for National Guard soldiers who gave a neutral response than for those who indicated that their employer did not support their military affiliation. It is unclear why this occurred, unless in this case ambivalence is actually more harmful.

What is it about employer support that is associated with depression or PTSD? Some authors have posited that negative social support contributes to the perception that the world is less safe, ^{Reference Charuvastra and Cloitre27} which may be compounded in soldiers with combat-related PTSD who have a heightened threat perception. For National Guard soldiers, feeling unsupported in the work environment, in which they likely spend a large part of their day, may contribute to less adaptive coping with post-deployment stress and higher rates of mental health problems. Another way of looking at this finding is that National Guard soldiers with mental health problems might have perceived the level of support from employers less favourably. A longitudinal study of the relationship between social support and PTSD in active component soldiers ²⁸ found that PTSD negatively affected the quality and quantity of social support; social support at time-point one did not reduce later PTSD symptom severity, but PTSD symptom severity at time-point one eroded social support at time-point two. The perception of lack of employer support may also be reflective of poor working relationships and other adverse working conditions such as employment discrimination, which National Guard soldiers have reported. ^{Reference King, King, Taft, Hammond and Stone29} However, we do not have the measures to assess this.

Also unhelpful for coping with PTSD appears to be the perception of National Guard soldiers that their deployment absence negatively affected their co-workers. We do not know whether National Guard soldiers felt concerned about this at the time of trauma exposures while they were deployed, nor to what degree there was an unpleasant work environment as a direct result of their absence and subsequent return to the workplace; an unpleasant work environment may also contribute to the perception that the world is hostile. ^{Reference Charuvastra and Cloitre27}

Limitations

The findings of this study should be interpreted in light of some limitations. First, we do not have pre-deployment data on the mental health status of these National Guard soldiers. Without such longitudinal data, we are unable to assess whether depression or PTSD preceded or were a consequence of the National Guard specific stressors. For example, we were unable to assess whether having a mental health problem preceded job loss or vice versa, but we would expect that having a mental health problem would impede continuous employment. Although we have two data time points, the samples at 3 and at 12 months are cross-sectional not longitudinal ones. This limits examination of the effects over time of the National Guard specific variables.

A second limitation is that our measures of financial hardship and employer support were both single items, which may not provide as robust a measure of these constructs as multiple items would; however, some studies have shown that single items can be valid. ^{Reference Elo and Leppanen30,Reference Wanous, Reichers and Hudy31} Nonetheless, the consistent and significant associations of these variables with both depression and PTSD in this study suggest that their influence on the mental health of National Guard soldiers needs further examination.

Third, the samples for this study were drawn from surveys of two National Guard brigades. National Guard brigades, unlike active component brigades, likely reflect the demographic characteristics of the state in which they are located. Although this study's findings point to the importance of examining the mental health effects of the stressors that National Guard soldiers experience, generalising these findings to all National Guard soldiers who had deployed to Iraq should be done cautiously. However, the comparability of our samples to the post-deployment population of National Guard soldiers referenced in Thomas et al ^{Reference Thomas, Wilk, Riviere, McGurk, Castro and Hoge2} gives us some confidence that the results in a more representative sample of National Guard soldiers would be similar.

Lastly, there are several risk factors for both major depression and PTSD some of which could not be assessed in this manuscript. Major depression in particular is a complex disorder that is the consequence of many factors. Consequently, we do not know whether the four factors would be associated differently or not at all with depression or PTSD if other risk factors were included in our analyses. However, one study that presented a comprehensive model of major depression in men showed that both low social support and stressful life events were among the risk factors for this disorder. ^{Reference Kendler, Gardner and Prescott32} Likewise, lack of social support was also found in a meta-analysis to be a risk factor for PTSD in both civilian and military populations, whereas ‘life stress’ was associated with PTSD only among civilians. ^{Reference Brewin, Andrews and Valentin33} Given that the respondents in this study are ‘citizen soldiers’, it is likely that findings about life stress would also be applicable.

Implications

The results of this study provide substantial evidence that National Guard soldiers have unique concerns compared with active component soldiers, which affect their well-being. The findings also indicate that although combat exposure contributes to the development of mental health problems, other experiences are also important. Studies that examine mental health among service personnel who experienced combat also need to consider how social and material conditions (daily stressors) affect the mental health of these individuals. ^{Reference Miller and Rasmussen12} Although it is understood that National Guard soldiers are different from active component soldiers, little research has examined how the post-deployment work-related experiences of National Guard soldiers affect their well-being. This study shows that future research within this group of soldiers is needed to address this important gap. A better understanding of the factors that influence the mental health of these soldiers, especially in the period before their mental health disease rates exceed those of active component soldiers, will be instrumental in the prevention of mental health problems.