Mental Health Treatment Delay: A Comparison Among Civilians and Veterans of Different Service Eras

Since October 2001, approximately 4 million veterans have been involved in active conflicts in and around Afghanistan and Iraq (1). Post-9/11 veterans have high rates of psychiatric conditions (36.9% to 43%) (2, 3). Three of the most common psychiatric conditions among recent veterans are posttraumatic stress disorder (PTSD), major depressive disorder, and alcohol use disorder (2–7).

Considerable attention has been paid to the unmet treatment needs of post-9/11 veterans (1, 3, 8, 9), aimed at decreasing delays in treatment. In response to high rates of psychiatric diagnoses among post-9/11 veterans and associated difficulties with reintegration (10), the Veterans Health Administration (VHA), the Department of Defense (DoD), veterans service organizations (11), and nongovernmental agencies have made concerted efforts in the past decade to increase veterans’ engagement in mental health treatment (1, 12). Recent initiatives include providing mental health treatment outside specialty mental health settings (e.g., primary care–mental health integration [PCMHI]) (13, 14), public awareness campaigns (e.g., the Veterans Crisis Line and the “Make the Connection” Web site) (15), postmilitary discharge outreach (16, 17), increased peer support programs (18), and enhanced eligibility among post-9/11 veterans for VHA benefits for 5 years following discharge (19).

Nonetheless, recent evaluations indicate that there are substantial remaining unmet mental health needs among post-9/11 veterans. For example, 31% of post-9/11 veterans with alcohol use disorder received mental health treatment in the past year (8), and 48% of soldiers screening positive for PTSD after deployment received mental health treatment in the past 6 months (9). Failure to engage in and delay in mental health treatment are common within the general U.S. population. Data collected between 2001 and 2003 in the National Comorbidity Survey Replication indicate that, on average, individuals with PTSD, depression, and alcohol use disorder delay treatment after symptom onset for 12, 8, and 9 years, respectively (20). Treatment delay is associated with persistent symptoms, adverse functional outcomes, and social costs (21, 22). Given substantial changes in policy and initiatives to increase veteran engagement in mental health care, it is important to use recent epidemiological data to understand whether treatment delay among post-9/11 veterans is comparable with delay among veterans of prior service eras and among civilians.

This study compared treatment delay for PTSD, depression, and alcohol use disorder among post-9/11 veterans and veterans of prior service eras and civilians. On the basis of recent initiatives and policy changes aimed at engaging recent veterans in mental health care, we expected shorter treatment delays among post-9/11 veterans compared with pre-9/11 veterans and civilians. We employed data from the National Epidemiologic Survey on Alcohol and Related Conditions–III (NESARC-III), a nationally representative survey of the general U.S. population (23).

Methods

All procedures for NESARC-III data collection were approved by the institutional review boards of the National Institutes of Health and Westat, Inc. This study was approved by the Institutional Review Board at the U.S. Department of Veterans Affairs Puget Sound Health Care System–Seattle Division.

Sample

The NESARC-III used probability sampling to select a representative sample of noninstitutionalized U.S. residents ages 18 and older who were not on active duty in the military during 2012–2013 (23). Individual counties or groups of contiguous counties served as primary sampling units; census blocks served as secondary sampling units; and households within secondary sampling units served as tertiary sampling units, with eligible adults randomly selected within households (23). Hispanic, black, and Asian respondents were oversampled. The full NESARC-III sample was 36,309 respondents, with a response rate of 60.1%, comparable to other U.S. national surveys (24).

Measures

Psychiatric diagnoses and symptom severity.

The Alcohol Use Disorder and Associated Disabilities Interview Schedule–5 (AUDADIS-5) is a validated, in-person diagnostic interview used to measure DSM-5 criteria for PTSD, major depressive disorder, and alcohol use disorder (25, 26). To receive a lifetime diagnosis of PTSD in NESARC-III, an individual must report witnessing, learning about, or experiencing at least one of 19 potentially traumatic events (criterion A) along with experiencing one or more intrusion symptoms (criterion B), one or more avoidance symptoms (criterion C), three or more negative mood or cognitive changes (criterion D), and three or more increased arousal symptoms (criterion E). This definition requires that more symptoms be present for criteria D and E than does DSM-5, which requires only two; however, it has been used in other studies based on NESARC-III (27, 28). Lifetime major depressive disorder and alcohol use disorder diagnoses were assessed with the AUDADIS-5, consistent with DSM-5 criteria (23, 29). As metrics of severity, counts of lifetime symptoms of each disorder were calculated from a list of potential symptoms (PTSD, N=20; depression, N=9; and alcohol use disorder, N=11).

Treatment delay.

Among those meeting lifetime diagnostic criteria for PTSD, major depressive disorder, or alcohol use disorder, a single item assessed age at onset of diagnosis and a separate item assesses age when help was first received for that disorder, if ever. Additional items assessed various types of help received (e.g., inpatient treatment and self-help group). We categorized treatment types into formal (e.g., inpatient treatment) or informal (e.g., self-help group) (see Supplemental Table 1 in the online supplement).

For survival analyses, the dependent variable was the length of time in years between the onset of diagnosis and first receipt of care (formal or informal). Those who did not report ever receiving treatment at the time of the survey were censored. A portion of respondents reported receiving treatment before onset of diagnosis, yielding negative values for estimates of time to treatment (PTSD: N=95, 4.1%; depression: N=552, 7.5%; alcohol use disorder: N=455, 4.6%). To avoid introducing bias through imputation of our dependent variable, negative values were coded as missing. Veteran status (active duty or not) was not associated with likelihood of reporting a negative time to treatment (see Supplemental Table 3 in the online supplement).

Explanatory variables.

The primary predictor of interest was veteran status, which was assessed by the question, “Have you ever served on active duty in the U.S. Armed Forces, Military Reserves, or National Guard?” Respondents answering “Yes, on active duty in past, but not now” were classified as veterans. Those indicating having trained in the National Guard and Reserve (N=200) but no other service or those with unknown veteran status (N=8) were excluded from analyses because of ambiguity regarding whether they should be categorized as veterans or civilians. Subsequent items assessed time periods of active duty. Individuals who endorsed any service post-9/11 were categorized as post-9/11 veterans, and those who did not were categorized as pre-9/11 veterans.

Given that efforts by the VHA and DoD may have contributed to increased probability of treatment for post-9/11 veterans, in exploratory analyses we included a survey item that asked respondents whether they were enrolled in VHA, CHAMPVA, or CHAMPUS/TRICARE in the past 12 months. This variable, referred to as VHA/TRICARE, was set to zero for civilian respondents. All post-9/11 veterans are eligible for VHA for 5 years after discharge. CHAMPVA is available only to dependents of veterans and thus is not a relevant response option for veterans. TRICARE is available only to retired military personnel. Because almost none of the post-9/11 veterans were over age 65 and relatively few were between 45 and 64, the VHA/TRICARE variable mostly represents enrollment in VHA. For the pre-9/11 veteran groups, this variable likely represents a combination of VHA enrollment and TRICARE coverage. The VHA/TRICARE variable does not indicate whether veterans actually use VHA/TRICARE services. Thus analyses including VHA/TRICARE coverage were considered exploratory. We created a five-level variable reflecting VHA/TRICARE coverage, veteran status, and era of service (civilians, pre-9/11 veterans with or without past year VHA/TRICARE coverage, and post-9/11 veterans with or without past year VHA/TRICARE coverage).

Demographic covariates.

Demographic characteristics that could not be influenced by military service (age, gender, and race-ethnicity) and that have been previously linked with treatment seeking (30, 31) were included as covariates. Sensitivity analyses included other demographic characteristics that could be influenced by military service (marital status and income).

Analysis Procedures

All analyses were adjusted for clustering, oversampling, and nonresponse and were weighted to represent the U.S. noninstitutionalized population, accounting for the complex survey design of NESARC-III. Survival analysis was used to model delay in treatment, while accounting for individuals who had not received treatment at the time of the survey (right-censoring) (32). Cox proportional hazard models were constructed by using the “svycoxph” function in the “survey” package in R (33, 34). A lower cumulative hazard represents a longer delay in treatment. Models were run separately for each diagnostic category (PTSD, depression, and alcohol use disorder) by using the subsample of individuals who met criteria for a lifetime diagnosis of the given disorder. An individual could be present in more than one subsample.

Subsequent sensitivity analyses assessed several potential sources of bias by controlling for additional demographic covariates that could be influenced by military service (marital status and income); restricting the sample to male respondents to account for the possibility that the increasing representation of women in the military influenced patterns of treatment among recent veterans (31); restricting treatment to formal care only, to assess exposure to more effective treatments; controlling for the co-occurrence of the other psychiatric disorders of interest (PTSD, depression, and alcohol use disorder), given that comorbidity has been associated with reduced treatment delay (31); and controlling for severity of PTSD, depression, and alcohol use disorder to rule out the possibility that differences in treatment delay were driven by differences in severity (22).

A final set of models examined the relationship between past-year VHA/TRICARE coverage and treatment delay to explore the possibility that treatment delay was influenced by efforts within VHA- or TRICARE-affiliated clinics to engage recent veterans in mental health treatment.

Results

The sample included 14,916 unique respondents, including those with a lifetime diagnosis of PTSD (N=2,335), major depressive disorder (N=7,406), and alcohol use disorder (N=9,943). A majority of the sample were civilians (N=13,528), with 1,130 pre-9/11 veterans and 258 post-9/11 veterans (Table 1). Across all three groups, most respondents with lifetime PTSD received treatment for PTSD (58.3%–79.8%, in weighted percentages), and the majority with lifetime major depressive disorder received treatment for depression (69.3%–76.3%). Consistent with prior reports, a considerably smaller percentage of respondents with alcohol use disorder received treatment for alcohol use disorder (17.5%–26.3%) (23). (See Supplemental Table 2 in the online supplement for comparisons between pre-9/11 veterans, post-9/11 veterans, and civilians on the probability of ever receiving treatment.)

On the basis of unadjusted Kaplan-Meier models, median survival time (time at which 50% had received treatment) was shortest among post-9/11 veterans for PTSD (median=2.5 years versus 16.0 years for pre-9/11 veterans and 15.0 years for civilians) and for depression (1.0 years versus 7.0 years and 5.0 years, respectively). Median survival times could not be computed for alcohol use disorder, given that fewer than 50% of members in each group ever received treatment .

Cox proportional hazard models included 14,088 respondents with complete data (828 respondents were excluded, largely because of missing treatment data). Several demographic variables were associated with treatment delay (Table 2). In particular, men showed longer delays for depression treatment and shorter delays for alcohol use disorder treatment than women. Older individuals showed longer delays for PTSD and depression treatment than younger individuals. Black and Hispanic individuals showed longer delays for PTSD and depression treatment than non-Hispanic white individuals.

When post-9/11 veterans were specified as the reference group (specification 1), post-9/11 veterans with PTSD were less likely to delay PTSD treatment than both pre-9/11 veterans (adjusted hazard ratio [AHR]= 0.69, 95% CI=0.49–0.96) and civilians (AHR=0.60, 95% CI=0.47–0.76) (Table 2, Figure 1), after the analyses were adjusted for age, gender, and race-ethnicity. Similarly, post-9/11 veterans with depression were less likely to delay depression treatment than both pre-9/11 veterans (AHR=0.74, 95% CI=0.56–0.98) and civilians (AHR=0.67, 95% CI=0.53–0.85). No differences in treatment delay were observed between post-9/11 veterans and pre-9/11 veterans or civilians for alcohol use disorder. In models with civilians as the reference group (specification 2), pre-9/11 veterans and civilians did not differ in treatment delay for any of the three disorders.

In a series of five sensitivity analyses, post-9/11 veterans were less likely to delay PTSD and depression treatment than both pre-9/11 veterans and civilians, with one exception (Table 3). There was no difference between post-9/11 veterans and pre-9/11 veterans in delay of formal— as opposed to formal or informal—depression treatment.

In exploratory analyses, past-year VHA/TRICARE coverage was not associated with PTSD treatment delay (see Supplemental Table 4 in the online supplement). In particular, no differences were found in PTSD treatment delay between post-9/11 veterans with and without past-year VHA/TRICARE coverage (specification 1). Post-9/11 veterans with and without past-year VHA/TRICARE delayed PTSD treatment less than civilians did (AHR=1.70, 95% CI=1.24–2.34, and AHR=1.62, 95% CI=1.16–2.27, respectively) (specification 2).

Past-year VHA/TRICARE coverage was associated with depression treatment delay, however. Specifically, post-9/11 veterans without past year VHA/TRICARE coverage showed longer delays in treatment for depression compared with post-9/11 veterans with past-year VHA/TRICARE coverage (AHR=0.56, 95% CI=0.35–0.90 (specification 1). In fact, only post-9/11 veterans with past-year VHA/TRICARE coverage delayed depression treatment less than civilians (AHR=1.89, 95% CI=1.43–2.49) (specification 2). No differences were observed in alcohol use disorder treatment delay between veterans with and without past-year VHA/TRICARE coverage.

Discussion

To better understand patterns of mental health treatment among post-9/11 veterans, this study examined data from NESARC-III, a nationally representative sample of the U.S. population. We compared the degree to which post-9/11 veterans delay treatment for common psychiatric conditions versus treatment delay among pre-9/11 veterans and civilians. Results indicated that post-9/11 veterans are less likely than both pre-9/11 veterans and civilians to delay mental health treatment for PTSD and depression.

The differences were clinically meaningful. Median time to PTSD treatment was 2.5 years for post-9/11 veterans, compared with 16.0 years and 15.0 years, respectively, for pre-9/11 veterans and civilians. Median time to depression treatment was 1.0 years for post-9/11 veterans, compared with 7.0 years and 5.0 years, respectively, for pre-9/11 veterans and civilians. This pattern was not present for alcohol use disorder, for which all three groups reported low rates of treatment engagement.

Results for PTSD and depression were largely consistent across sensitivity analyses, suggesting that shorter treatment delay for post-9/11 veterans is not attributable to differences in demographic factors, symptom severity, or psychiatric comorbidity. However, there was no significant difference between pre- and post-9/11 veterans in delay of formal depression treatment. It may be that efforts to engage recent veterans in mental health treatment have placed greater emphasis on the need for formal treatment of PTSD than for depression.

The increased engagement in PTSD and depression treatment among post-9/11 versus pre-9/11 veterans could be attributable to a host of recent historic, cultural, and policy changes. A new policy requiring universal mental health screening postdeployment may have contributed to reduced treatment delay for post-9/11 veterans (35). Screening efforts may have improved detection of postdeployment mental health concerns and increased engagement in treatment (9, 17).

Other potential contributors are efforts by the VHA and other organizations to engage post-9/11 veterans in treatment (1, 12, 35, 36). These include educational public health initiatives, dissemination of educational resources for community providers, creation of programs targeting reintegration and mental health among returning veterans, suicide hotlines, and development of veteran-specific mobile health resources (16, 36–39). Although not designed specifically for post-9/11 veterans, efforts to improve detection of mental health concerns in VHA primary care settings (40) and integration of mental health services in these settings (PCMHI) may also have increased engagement (41).

Findings from exploratory analyses partially supported the possibility that past-year VHA/TRICARE coverage may be associated with decreased treatment delay for post-9/11 veterans. Post-9/11 veterans reporting no past-year VHA/TRICARE coverage delayed depression treatment longer than post-9/11 veterans reporting past-year VHA/TRICARE coverage. Past-year VHA/TRICARE coverage was not, however, associated with treatment delay for PTSD or alcohol use disorder. These analyses should be interpreted cautiously, given the small number of post-9/11 veterans with VHA/TRICARE coverage.

Although post-9/11 veterans experienced shorter treatment delays for PTSD and depression compared with civilians and pre-9/11 veterans and high rates of treatment receipt (74.8%−79.8%), engagement with alcohol use disorder treatment for post-9/11 veterans was low (17.5%). This finding, which is consistent with previous reports (17, 23), is concerning, given the morbidity and mortality associated with alcohol use disorder (42, 43). Several factors may help explain low rates of treatment engagement among both veterans and civilians with alcohol use disorder. A study of over 22 million Americans with current substance use disorders who had not sought care in the past year found the most commonly stated reasons were feeling unready to stop drinking and a lack of health insurance or means to pay for treatment (44). Lack of problem recognition and stigma are also common reasons for not seeking substance abuse treatment (8, 45, 46). A review of 17 population studies indicated that alcohol misuse is less likely to be perceived as a mental illness and is more stigmatized than other psychiatric disorders (47), which may account for the lower rate of treatment engagement across our sample for alcohol use disorder compared with PTSD and depression. Limited alcohol use disorder treatment across multiple studies suggests that future research and clinical efforts to engage individuals in treatment are critically important. Future research could also explore why recent veterans are not seeking treatment faster for alcohol use disorder compared with veterans of earlier eras, unlike the pattern for PTSD and depression.

Although racial-ethnic differences were not the focus of this study, longer delays for PTSD and depression treatment were found for individuals from racial-ethnic minority groups compared with non-Hispanic whites. This finding is consistent with broader literature on mental health disparities indicating that individuals from racial-ethnic minority groups are less likely to engage in and more likely to delay mental health treatment (20). Future studies could examine risk and protective factors associated with treatment delay among members of racial-ethnic minority groups and interventions to increase engagement.

Limitations of this study included a modest number of veterans, particularly post-9/11 veterans, affecting our ability to detect small effects and the accuracy of observed effects. We also lacked information regarding where treatment was received, limiting our ability to draw conclusions regarding which engagement efforts were responsible for reduced treatment delay among post-9/11 veterans. Likewise, the VHA/TRICARE variable likely reflected different levels of access for pre-9/11 and post-9/11 veterans, and service connection, another aspect of access, was not assessed. The cross-sectional nature of the study increased risk of recall biases (e.g., inaccuracies in recalling age at symptom onset and treatment initiation for more temporally distant events), which may be larger for pre-9/11 veterans. Despite these limitations, the study had several strengths, including a large and nationally representative sample of the U.S. population, assessment of DSM-5 diagnoses of psychiatric conditions prevalent among veterans and for which evidence-based treatments are available, and inclusion of a number of sensitivity analyses to increase confidence in the pattern of findings.

Conclusions

Post-9/11 veterans were less likely than pre-9/11 veterans and civilians to delay mental health treatment for PTSD and depression. Efforts in the past decade by the VHA, DoD, and nongovernmental organizations may have been influential in promoting mental health treatment engagement for this recent generation of military service members.