Lifetime and 12-Month Use of Psychiatric Services Among U.S. Army National Guard Soldiers in Ohio

The U.S. military force includes two components. The full-time, or active duty, component includes more than 1.4 million active duty service members, and the part-time, or reserve, component includes more than 1.2 million reservists. Although reservists are part-time soldiers—training one weekend a month and 15 days annually—when activated for deployment, they experience traumatic combat exposures at levels comparable to those of active duty personnel (1). Furthermore, reservists encounter a broad range of civilian challenges that are largely not germane to active duty personnel (civilian employment, for example) (2,3). There is abundant evidence that both the military (4) and the social stressors (2,3) experienced by reservists are associated with poor psychiatric health. In turn, the consequences of psychiatric disorders are profound. At the individual level, these stressors reduce quality of life and increase physical symptom severity (5) and aggressive actions (6,7); at the population level, the estimated economic burden of psychiatric illness in the United States is $300 billion (8).

The availability of effective treatments for psychiatric disorders provides an opportunity to minimize the burden of population-level psychiatric illness by ensuring that service members with psychiatric disorders receive care (9). Unfortunately, most reservists who screen positive for psychiatric disorders report not receiving care (10). There is evidence in the nonmilitary research literature that treatment-seeking behaviors vary greatly among anxiety, mood, and substance use disorders (11–13). The best available evidence suggests that persons with substance use disorders delay treatment the longest after disorder onset, whereas persons with mood disorders more readily perceive a need for professional help and seek both general medical treatment and specialized psychiatric treatment at greater rates than those with anxiety or substance use disorders (11,13).

Although there is reliable evidence that psychiatric disorder categories affect treatment seeking among civilians, military personnel are provided robust, equally distributed, psychiatric service access that on face value should reduce sociodemographic treatment disparities often documented in civilian populations (14). However, no study to date has longitudinally identified treatment-seeking behaviors across various psychiatric disorder categories in a representative sample of military personnel that was not limited to recently deployed service members. Therefore, we aimed to assess whether certain psychiatric disorder categories are associated with likelihood of psychiatric service use; whether individual characteristics affect the likelihood of psychiatric service use; and which individual characteristics, lifetime or current psychiatric disorder, or previous lifetime psychiatric service use has the greatest effect on treatment seeking by military personnel.

Methods

Study Sample and Procedure

The Ohio Army National Guard (OHARNG) Mental Health Initiative was a longitudinal cohort study of a representative sample of OHARNG service members between 2008 and 2012. Sample selection occurred in three distinct phases. First, a telephone survey randomly sampled 6,084 eligible soldiers and obtained a baseline sample of 2,616 (43% participation rate) (15); next, we randomly invited 40% of individuals completing the telephone survey to participate in the in-depth clinical cohort (91% agreed, N=952); and we scheduled a meeting with respondents through either University Hospital Case Medical Center or University of Toledo, depending on the participant’s proximity to each location. Finally, the target 500 OHARNG service members provided informed consent and were enrolled in the study. At year three, 105 new participants were added to the sample by using the identical protocols explained above for the baseline sample (N=605). We excluded 77 respondents who were assessed only at baseline, creating the final analytical sample of 528 respondents who contributed a total of 1,210 person-years (2.4±.9 years per person on average). The institutional review boards at both institutions approved all study methods.

Five study-trained clinicians completed in-person clinical interviews from June 2008 to February 2012 in a neutral private location (such as a private library room). The interview assessed respondents’ military experiences, psychiatric health history, treatment history, and sociodemographic characteristics. Recorded audio copies of each study clinician’s Structured Clinical Interview for DSM-IV (SCID [16]) and Clinician-Administered PTSD Scale (CAPS [17]) interviews were reviewed by other study-trained clinicians to ensure that the interviewers used standardized diagnostic assessment methods and interviewing techniques.

Measures

Dependent variables.

Recent psychiatric service use was determined by asking participants whether they had accessed since the last interview either psychotherapy (“Are you currently receiving or have you received psychotherapy, counseling, group or family therapy, or any family program since we last spoke?”) or pharmacotherapy (“Have you taken any medication [including over-the-counter (OTC) and herbal remedies] for emotional or sleep problems since we last spoke?”). Interviews were completed about 12 months apart on average.

Independent variables.

Psychiatric diagnoses were determined from a diagnostic interview that included the CAPS (17) and portions of the SCID (16). To assess posttraumatic stress disorder (PTSD) symptoms, the CAPS was administered twice, once based on the respondent’s self-selected worst traumatic event outside the most recent deployment and once with respect to the worst trauma during the respondent’s most recent deployment (18). PTSD diagnosis was based on DSM-IV criteria and the frequency ≥1 and intensity ≥2 (19) method, which is the original and most sensitive scoring method for the CAPS (17). Respondents meeting diagnostic criteria on either CAPS completed were classified as having PTSD.

The SCID was administered to assess axis I disorders (20), which were categorized into three groups, including anxiety disorders (panic disorder with or without agoraphobia, specific phobia, social phobia, obsessive-compulsive disorder, PTSD, and anxiety disorder not otherwise specified), mood disorders (major depressive disorder, bipolar disorder type I or type II, mania, dysthymia, and depression not otherwise specified), and substance use disorders (alcohol or drug abuse or dependence or combination). All diagnoses were based on DSM-IV criteria and a decision-tree approach to record the presence or absence of each disorder for current and lifetime occurrences. We did not include subthreshold diagnoses.

Respondents were classified as having a lifetime psychiatric disorder if they reported symptoms meeting diagnostic criteria that occurred prior to their previous interview. Conversely, respondents meeting diagnostic criteria in the interval since their last interview were determined to have a current psychiatric disorder. All disorder diagnoses were given equal weight (that is, we did not differentiate between primary and secondary diagnoses). Although distinct comorbidity patterns were not considered in this analysis, we created an ordinal variable to account for number of co-occurring disorders (no current disorder, one disorder, or two or more disorders).

Sociodemographic (age, gender, race-ethnicity, education, marital status, income, and insurance) and military (rank and lifetime deployment) characteristics were self-reported as part of the interview.

Analysis

First, period prevalence of psychiatric service use was calculated by dividing the total number of respondents reporting past-year psychiatric service use by the person-years contributed by respondents in each respective disorder category of analysis. Person-years in disorder were calculated as the sum of all survey years in which respondents met criteria for the examined disorder category. Respondents with past-year co-occurring disorders were included in the denominator of each diagnosed disorder.

Second, we examined the relationship between each dependent variable and the predictor variables through a series of unadjusted bivariate logistic regression models reporting odds ratio (OR) and 95% confidence interval (CI) for each predictor. Third, a generalized estimating equation (GEE) assuming a binomial logit-link function and autoregressive 1 correlation structure was used to estimate adjusted odds ratios (AORs) and CIs between 12-month psychiatric service use and current and lifetime psychiatric diagnosis and lifetime psychiatric service use, with adjustment for predetermined demographic, military, and clinical variables by the proc genmod procedure in SAS version 9.3. To determine best model fit, the model was examined to identify whether the removal of each demographic and military characteristic resulted in at least a 10% change in the regression coefficient estimates for current and lifetime psychiatric diagnosis or lifetime psychiatric service use. Variables that did not reach statistical significance at α<.05 were dropped from subsequent models to preserve degrees of freedom and improve precision.

Results

Table 1 shows the baseline demographic and military characteristics of the 528 participants. The sample was predominantly male (88%), non-Hispanic white (89%), and enlisted (89%); had served at least one deployment (61%); and was a mean age of 30. Further, 64% had a lifetime psychiatric disorder, 24% had a current psychiatric diagnosis, and 51% had received psychiatric services.

As shown in Table 2, period prevalence of recent treatment stratified by psychiatric diagnosis category ranged from 468.2 to 726.6 persons per 1,000 person-years for substance use and mood disorders, respectively. The period prevalence of psychiatric service use for respondents with no disorder, one disorder, and two or more disorders was 209.3 persons, 531.1 persons, and 634.2 persons per 1,000 person-years, respectively (21%, 53%, and 63% of persons per year, respectively).

Data on lifetime and current prevalence of psychiatric diagnosis and co-occurring disorders stratified by lifetime and recent treatment are presented in Figure 1 in the form of period prevalence per 100 person-years. Each of the figure’s four quadrants presents a subfigure of one of four potential lifetime and current psychiatric diagnosis categories; disorder categories are not mutually exclusive (in other words, a respondent with co-occurring disorders is represented in both lifetime prevalent diagnosis quadrants). Mood disorders consistently had the highest period prevalence of recent and lifetime treatment among current diagnoses (specifically, lifetime prevalent diagnosis with current diagnosis, no lifetime prevalent diagnosis with current diagnosis), whereas substance use disorders consistently had the highest period prevalence of reporting neither lifetime nor recent psychiatric treatment.

In bivariate analysis (Table 3), women (OR=1.9) and separated, divorced, or widowed respondents (OR=1.6) were significantly more likely than comparison groups to have used recent psychiatric services. Respondents with any current disorder were 4.4 times more likely than others to report recent treatment, with mood disorders most highly associated (OR=9.1) and substance use disorders (OR=2.0) least associated with treatment. Respondents with any lifetime psychiatric disorders and those reporting previous lifetime psychiatric service use were 3.2 times and 4.4 times more likely than others, respectively, to report recent treatment.

In the multivariable GEE (Table 4), current mood disorder was the strongest predictor of recent treatment (AOR=5.0), whereas current substance use disorder was not associated with psychiatric service use. Lifetime pharmacotherapy (AOR=4.1) and psychotherapy (AOR=1.4) significantly predicted recent treatment. No sociodemographic or military characteristics met the a priori model specification criteria described in the methods and were consequently removed from the final model.

Discussion

Using data from a representative sample assessing reservists’ treatment-seeking behaviors, we found, first, an annualized rate of 31% of persons per year accessed psychiatric services between 2010 and 2012. This annualized rate of psychiatric service use is higher than previous reports in representative military (21,22) and nonmilitary populations (23). In studies examining both the general U.S. population and active duty U.S. Army soldiers, approximately 20% of all persons reported past-year use of psychiatric services (22,23). Conversely, the estimated 31% of persons per year accessing psychiatric services in our study is closer to utilization rates among Operation Iraqi Freedom veterans within their first postdeployment year (35%) (24). This discrepancy in utilization may be explained by previous studies that documented that reservists report greater psychiatric service need than their active duty counterparts (24,25). Also, our data collected between 2008 and 2012 may have been affected by the implementation of recent Department of Defense initiatives aimed at increasing psychiatric service use among soldiers (26,27), which would not have affected previous military service rates based on 2008 data (22).

Second, our finding that persons with substance use disorders had the lowest annualized rate of psychiatric service use and that this was the only class of psychiatric disorder not predictive of current service use is supported by existing literature (12). Persons with substance use disorders often fail to perceive a need (13,28,29) and do not seek treatment until their disorder creates difficulties in daily living (12,30) or begins to affect several areas of their lives (31). In two different national surveys, about 10% of persons with alcohol use disorders perceived a need for treatment and did not seek treatment, whereas a majority of those who perceived a need received treatment (28). Conversely, that study showed that persons with mood disorders tended to experience several factors that increased their perceived need for services.

Although drug use disorders are uncommon in the military, untreated alcohol use disorder has been of considerable concern since the publication of several reports documenting that military personnel report heavy drinking at consistently higher levels than similar civilian samples (32,33). Further, reservists who face combat exposure during deployment are at increased risk of new-onset alcohol misuse and abuse compared with active duty personnel (34). Although over one-third of National Guard members have been documented to meet criteria for alcohol misuse (35), less than 1% of service members are referred to substance abuse treatment (4). In addition, a recent report documented that about a third of Operation Enduring Freedom/Operation Iraqi Freedom service members receiving health care from the U.S. Department of Veterans Affairs who met criteria for risky drinking were advised by a provider to drink less or to stop drinking (36). Although the negative repercussions of alcohol misuse in the military have been well documented, these studies suggest that recognition of and referral for alcohol use disorders remain insufficient.

Third, our observation that current mood disorder diagnosis was the most significant predictor of recent psychiatric service use is consistent with both military and civilian literature (11,13) and may be due to the effect of these disorders on disability, productivity, and health-related quality of life (3,37). Further, an increasing percentage of Americans have been seeking services for depression over the past two decades, which may be attributed both to increased population awareness of the signs and symptoms of mood disorders and of pharmacotherapy options for treating them and to reduced negative public perceptions about persons who experience depression (38,39).

Fourth, we found that psychiatric service use was not affected by sociodemographic or military characteristics when we adjusted for current and lifetime psychiatric disorder and treatment history. This finding is contrary to a recent U.S. Army report documenting that female, married, and enlisted personnel were more likely than other soldiers to access psychiatric services (22). Although McKibben and associates (22) documented an association between demographic variables and psychiatric service use, self-reported impaired functioning was the strongest driver of service use. Their finding that impaired functioning was the primary driver of treatment, in conjunction with our findings, suggests that the military environment may serve as a foil to racial (12,40), gender (12,40), age-related (13,40), and marital status (12,40) disparities documented in civilian populations, creating a system in the military where disorder status predominates over sociodemographic factors as a central driver of care seeking.

The results should be interpreted within the context of four limitations. First, self-report data may result in the underreporting of psychiatric symptoms. Even though military personnel are likely to provide truthful answers if they believe individual answers will remain confidential and the findings will be used for legitimate purposes (41), the presence or perception of stigma toward psychiatric illness (42) and a bias against reporting embarrassing behaviors (43) remain prevalent in the military. We compensated for this concern prior to participants’ volunteering for the study by assuring them, both verbally and in writing, that answers would remain confidential and by having all assessments conducted in neutral locations by civilian clinicians without the presence of military personnel.

Second, self-report of service utilization is unreliable (44). Although administrative records would have provided more reliable treatment information, we assessed participants annually to reduce the time between interviews and to limit recall concerns (45). Further, interviewers were trained to probe about the purpose, dates, and reasons for psychiatric service use to improve details.

Third, we assessed any psychotherapy or pharmacotherapy contact for self-perceived psychiatric health concerns, which included OTC medication, but we did not assess effectiveness or validity of treatment modalities. Although OTC medication use was rare (N=7), several studies have documented that people seeking psychiatric services do not receive minimally adequate treatment, often having only one visit with providers in the general medical sector (12). Although this topic is of interest, it was outside the scope of this report and will require further investigation with future studies. In addition, respondents may have sought care for reasons other than the diagnosed disorder. Although this is an area for future investigation, the analysis we performed prevents us from suggesting whether a causal mechanism exists between disorder diagnosis and subsequent treatment.

Fourth, these findings may not be generalizable to other reservists (for example, Navy Reserves) from other states. Although the OHARNG is similar in several key demographic and social factors to the U.S. Army (including percentage of high school graduates and per capita income) and National Guard (including age, gender, and rank) populations, replication of findings in other states and components would improve confidence in our findings.

Conclusions

In summary, we documented that psychiatric disorder categories differentially predicted psychiatric service use by military personnel, with mood disorders most predictive of treatment and substance use disorders not significantly associated with treatment seeking. Furthermore, sociodemographic and military characteristics were not significant predictors of service use when we accounted for current psychiatric disorders and lifetime treatment. These findings suggest that, even though reliable treatment methods exist, about half of soldiers with a psychiatric disorder are currently going untreated and that soldiers with substance use disorders have the lowest levels of treatment. These observations suggest that additional prevention and education are warranted among military personnel to increase the recognition and referral of substance use disorders to care. Explanation of treatment-seeking behaviors and evaluation of policy aimed at reducing substance use disorders in the military are important areas for future research to examine.