Incidence of Mental Health Diagnoses in Veterans of Operations Iraqi Freedom, Enduring Freedom, and New Dawn, 2001–2014

Abstract

Objectives. To evaluate gender, age, and race/ethnicity as predictors of incident mental health diagnoses among Operations Iraqi Freedom, Enduring Freedom, and New Dawn veterans.

Methods. We used US Veterans Health Administration (VHA) electronic health records from 2001 to 2014 to examine incidence rates and sociodemographic risk factors for mental health diagnoses among 888 142 veterans.

Results. Posttraumatic stress disorder (PTSD) was the most frequently diagnosed mental health condition across gender and age groups. Incidence rates for all mental health diagnoses were highest at ages 18 to 29 years and declined thereafter, with the exceptions of major depressive disorder (MDD) in both genders, and PTSD among women. Risk of incident bipolar disorder and MDD diagnoses were greater among women; risk of incident schizophrenia, and alcohol- and drug-use disorders diagnoses were greater in men. Compared with Whites, risk incident PTSD, MDD, and alcohol-use disorder diagnoses were lower at ages 18 to 29 years and higher at ages 45 to 64 years for both Hispanics and African Americans.

Conclusions. Differentiating high-risk demographic and gender groups can lead to improved diagnosis and treatment of mental health diagnoses among veterans and other high-risk groups.

Veterans who served in Operation Enduring Freedom, Operation Iraqi Freedom, or Operation New Dawn (OEF/OIF/OND) are unique from veterans of other eras in sociodemographic composition and in Veterans Health Administration (VHA) service use. Approximately 61% of all separated (not on active-duty military service) OEF/OIF/OND veterans have used VHA services since October 1, 2001, making the VHA the single largest health care provider for these veterans.¹ Women account for 12.2% of OEF/OIF/OND-era VHA users, and represent the fastest growing segment of new users.^1–3 Nearly one third of men and 45% of women in this cohort are non-White, compared with the 18.4% of men and 29.3% of women veterans from all other eras.⁴ According to recent estimates, 57.6% of OEF/OIF/OND veterans who have used VHA services have been diagnosed with a mental health condition.¹

Research has demonstrated an increase in the prevalence of mental health diagnoses in this cohort as the recent conflict has continued,^5–7 and identified sociodemographic differences, with women more often diagnosed with depression, and men more often diagnosed with posttraumatic stress disorder (PTSD) and alcohol and substance use disorders.^6,8,9 Female gender, White race, and age younger than 25 years have been associated with earlier initiation of both primary care and mental health care in this cohort.⁹

Although prevalence of, and treatment seeking for, mental health diagnoses have been studied in OEF/OIF/OND veterans who used VHA services shortly after separation,⁵ less is known about incident mental health diagnoses among VHA users who do not receive a mental health diagnosis at their first VHA visit. Studies of OEF/OIF/OND and previous veteran cohorts suggest that mental health conditions, and PTSD in particular, can emerge months and even years following deployment¹⁰ or exposure to disasters, combat, and other large-scale traumas.^11,12 Therefore, articulating a clear temporal relationship between exposure and new diagnosis is an important strength of studying incidence rates that cannot be captured by prevalence studies. Furthermore, the ability of incidence studies to pinpoint age- and cohort-specific risk factors for new mental health diagnoses can provide insights into clinicians’ diagnosing tendencies and patients’ characteristics, which can allow health care providers to mobilize health care resources more effectively. We therefore determined incidence rates of 6 common mental health diagnoses among OEF/OIF/OND veterans who used VHA services who did not have a mental health diagnosis at the time of their first VHA visit. We examined age, gender, and race/ethnicity as predictors for incident mental health diagnoses from 2001 to 2014.

METHODS

We drew data from the OEF/OIF/OND Roster and the VA National Patient Care Database (NPCD). The Roster is a national database that contains demographic and military service data on the roughly 56% of OEF/OIF/OND veterans who have separated from military service and have enrolled in the VHA.¹³ We linked data from veterans in the OEF/OIF/OND Roster with the NPCD, which includes administrative and clinical data for veterans who use VHA services.

Consistent with other studies of this cohort,^5–9 veterans included in this study had used VHA services between October 1, 2001 (beginning of the US conflict in Afghanistan), and November 19, 2014 (the most recent data collection quarter available at the time of these analyses); we did not include veterans from other eras. Because the aim of this study was to determine incidence rates of new mental health diagnoses, we excluded those with a psychiatric diagnosis based on International Classification of Diseases, Ninth Revision (ICD-9) codes (see “Measures” section for further details) at first VHA visit (n = 16 504; 1.5%).¹⁴ We also excluded veterans if they had only 1 VHA inpatient or outpatient visit during the follow-up period (a follow-up period of 0 days; n = 182 625; 16.8%). There were no differences between included and excluded OIF/OEF/OND VHA users on age, race/ethnicity, gender, or marital status. The final sample comprised 888 142 (approximately 82%) of the 1.1 million OEF/OIF/OND veterans enrolled in the VHA during the follow-up period.

Measures

Incident mental health diagnoses.

We used previously validated ICD-9 diagnostic code groupings to determine the incidence of 6 mental health conditions in the electronic health record: PTSD, major depressive disorder (MDD), alcohol use disorders (AUDs), drug use disorders (DUDs), bipolar disorder (BPD), and schizophrenia (SZ)¹⁴ (Table A, available as a supplement to the online version of this article at http://www.ajph.org). Although AUD and DUD are often evaluated together as “substance use disorders,” because this study is the first to report on incidence rates in this population, and because rates of DUD are typically much lower than rates of AUD, we presented these conditions separately. Veterans were classified with a mental health condition if they had 1 inpatient hospitalization or 2 outpatient encounters coded with the same ICD-9 code. This approach has been found to improve the validity of mental health diagnoses taken from administrative data in previous studies of VHA patients.^8,15

Sociodemographic and military service covariates.

We obtained race/ethnicity, marital status, date of birth, military component (officer or enlisted), branch of service, dates of first and last deployment, and separation dates from the military from the OEF/OIF/OND Roster. We calculated age at first VHA visit and time between end of last deployment and first VHA visit.

Statistical Analysis

We compared men and women VHA users on demographic and military service variables by using the χ² test for categorical variables and the Wilcoxon ranked sum test for continuous variables. In 6 separate age- and gender- stratified models, we calculated incidence rates for each of the 6 mental health diagnoses from the time of first VHA encounter to the date of diagnosis. Veterans who were not diagnosed with the mental health condition being modeled were censored at the date of their last VHA visit through November 2014. Past research has demonstrated distinct median age of onset for different mental health conditions.¹⁶ To account for these known patterns in age of onset, we stratified incident-rate analyses by 4 age ranges (18–29, 30–44, 45–64, and ≥ 65 years), as has been done in previous cohort studies of mental health conditions,^17–19 and calculated incidence rate ratios (IRRs) for comparisons across age strata. To determine independent contribution of gender for each mental health diagnosis, we used Poisson regression models to calculate IRRs, with adjustment for sociodemographic and military service variables. Incidence rates and IRRs were not reported for veterans aged 65 years or older because of small sample size (n = 370) and low incidence of mental health diagnoses. (See Table B, available as a supplement to the online version of this article at http://www.ajph.org, for information on multiple morbidity and incident mental health diagnosis co-occurrence rates.)

RESULTS

Of the 888 142 veterans included, 87.6% were men and 12.4% were women. Women were younger (median age = 28.8 years; interquartile range [IQR] = 25.0–37.8) than men (median age = 29.6 years; IQR = 25.2–40.1; P < .001), more likely to be African American and to have multiple deployments, and less likely to be married (all P < .001; Table 1).

TABLE 1—

Sociodemographic and Military Service Characteristics of Veterans Without a Prevalent Diagnosis of Mental Illness at Initiation of Veterans Health Administration Service Use in the Era of Operations Iraqi Freedom, Enduring Freedom, and New Dawn: United States, 2001–2014

Characteristic	Men (n = 777 722; 87.6%), % or Median (IQR)	Women (n = 110 420; 12.4%), % or Median (IQR)	Total (n = 888 142), % or Median (IQR)	P
Age at first VHA visit, y	29.6 (25.2–40.1)	28.8 (25.0–37.8)	29.5 (25.1–39.8)	< .001
End of last deployment to first VHA visit, y	1.3 (0.4–2.9)	1.4 (0.4–3.2)	1.3 (0.4–2.9)	< .001
Married	47.5	33.1	45.7	< .001
Race/ethnicity				< .001
White	60.2	45.8	58.4
African American	12.3	24.5	13.8
Hispanic	10.6	11.1	10.6
Other	5.2	6.8	5.4
Unknown	11.7	11.9	11.7
Military rank				< .001
Enlisted	92.0	90.4	91.8
Officer	7.0	9.0	7.2
Branch of service				< .001
Army	59.6	59.8	59.6
Coast Guard	0.14	0.09	0.14
Air Force	11.3	17.9	12.1
Marines	15.7	4.4	14.3
Navy	13.3	17.8	13.8
Multiple deployments	47.9	55.3	48.8	< .001

Note. IQR = interquartile range; VHA = Veterans Health Administration.

For both men and women across all age groups, PTSD was the most frequent incident diagnosis. The highest incidence rates for a diagnosis of PTSD among men was in the group aged 18 to 29 years (41.2 per 100 000 person-years; 95% confidence interval [CI] = 41.0, 41.4) and the highest incidence rates among women was in group aged 30 to 44 years (27.2 per 100 000 person-years; 95% CI = 27.2, 28.3; Table 2). With adjustment for marital status, branch of service, rank, multiple deployments, and time to first VHA encounter, risk of incident PTSD diagnosis varied as a function of age and gender, and age and race/ethnicity, with risk being greater among older women and racial minorities. Specifically, risk of incident PTSD diagnosis was lower in women versus men aged 18 to 29 years (IRR = 0.58; 95% CI = 0.57, 0.59) and 30 to 44 years (IRR = 0.83; 95% CI = 0.81, 0.84), though higher in women aged 45 to 64 years (IRR = 1.06; 95% CI = 1.02, 1.11; Table 3). Compared with Whites, risk of incident PTSD diagnosis was lower in Hispanics aged 18 to 29 years and in African Americans aged 18 to 29 years and 30 to 44 years, and higher in African Americans and Hispanics aged 45 to 64 years (all P < .001; Table C, available as a supplement to the online version of this article at http://www.ajph.org).

TABLE 2—

Age-Stratified, Unadjusted Incidence Rates and Incident Rate Ratios for Diagnosis of Mental Illness in Women and Men in the Veterans Health Administration: United States, 2001–2014

Mental Illness	100 000 Person-Years	No. of Diagnoses	Incidence Rate (95% CI)	IRR (95% CI)
Posttraumatic stress disorder
Women, age, y
18–29	6.43	15 613	24.27 (23.89, 24.66)
30–44	3.46	9 606	27.73 (27.18, 28.29)	1.14 (1.11, 1.17)
45–64	1.17	2 644	22.67 (21.82, 23.55)	0.93 (0.90, 0.97)
Men, age, y
18–29	0.34	138 653	41.23 (41.00, 41.44)
30–44	0.24	80 946	34.33 (34.10, 34.57)	0.83 (0.82, 0.84)
45–64	0.10	21 681	21.35 (21.07, 21.64)	0.52 (0.50, 0.53)
Major depressive disorder
Women, age, y
18–29	7.26	8 937	12.30 (12.05, 12.56)
30–44	3.90	5 974	15.30 (14.91, 15.69)	1.24 (1.20, 1.28)
45–64	1.30	1 645	12.70 (12.10, 13.33)	1.03 (0.98, 1.09)
Men, age, y
18–29	0.45	38 499	8.55 (8.46, 8.63)
30–44	0.30	26 627	8.91 (8.81, 9.02)	1.04 (1.03, 1.06)
45–64	0.12	8 015	6.76 (6.61, 6.91)	0.79 (0.77, 0.81)
Alcohol use disorder
Women, age, y
18–29	7.79	4 004	5.14 (4.98, 5.30)
30–44	4.30	1 972	4.59 (4.39, 4.79)	0.89 (0.85, 0.94)
45–64	1.41	446	3.15 (2.87, 3.46)	0.61 (0.56, 0.68)
Men, age, y
18–29	0.43	59 268	13.67 (13.56, 13.78)
30–44	0.30	26 779	8.93 (8.83, 9.04)	0.65 (0.64, 0.66)
45–64	0.12	6 747	5.60 (5.47, 5.74)	0.41 (0.40, 0.42)
Drug use disorder
Women age, y
18–29	7.96	2 421	3.04 (2.92, 3.17)
30–44	4.41	791	1.79 (1.67, 1.92)	0.59 (0.54, 0.64)
45–64	1.44	113	0.78 (0.65, 0.94)	0.26 (0.21, 0.31)
Men, age, y
18–29	0.46	33 669	7.34 (7.26, 7.42)
30–44	0.32	10 076	3.20 (3.14, 3.26)	0.44 (0.43, 0.45)
45–64	0.13	1 459	1.16 (1.11, 1.22)	0.16 (0.15, 0.17)
Bipolar disorder
Women, age, y
18–29	7.78	3 720	4.78 (4.63, 4.94)
30–44	4.30	1 952	4.54 (4.34, 4.74)	0.95 (0.90, 1.00)
45–64	1.42	352	2.47 (2.23, 2.75)	0.52 (0.46, 0.58)
Men, age, y
18–29	0.47	21 745	4.68 (4.62, 4.74)
30–44	0.31	10 902	3.49 (3.42, 3.55)	0.75 (0.73, 0.76)
45–64	0.13	2 340	1.88 (1.81, 1.96)	0.40 (0.38, 0.42)
Schizophrenia
Women, age, y
18–29	8.09	189	0.23 (0.20, 0.27)
30–44	4.45	98	0.22 (0.18, 0.27)	0.94 (0.74, 1.20)
45–64	1.45	7	0.05 (0.02, 0.10)	0.21 (0.10, 0.44)
Men, age, y
18–29	0.48	2 235	0.47 (0.45, 0.49)
30–44	0.32	651	0.20 (0.19, 0.22)	0.43 (0.40, 0.47)
45–64	0.13	93	0.07 (0.06, 0.09)	0.16 (0.13, 0.19)

Note. CI = confidence interval; IRR = incident rate ratio. Incidence rates and IRRs were not reported for veterans aged 65 years or older because of small sample size (n = 370) and low incidence of mental health diagnoses.

TABLE 3—

Age-Stratified Incident Rate Ratios for Diagnosis of Mental Illness in Women Compared With Men in the Veterans Health Administration: United States, 2001–2014

Mental Illness	Model 1,a IRR (95% CI)	Model 2,b IRR (95% CI)
Posttraumatic stress disorder, age, y
18–29	0.59 (0.58, 0.60)	0.58 (0.57, 0.59)
30–44	0.81 (0.79, 0.83)	0.83 (0.81, 0.84)
45–64	1.06 (1.02, 1.11)	1.06 (1.01, 1.11)
Major depressive disorder, age, y
18–29	1.44 (1.41, 1.47)	1.17 (1.14, 1.20)
30–44	1.72 (1.67, 1.76)	1.50 (1.46, 1.55)
45–64	1.88 (1.78, 1.98)	1.77 (1.67, 1.87)
Bipolar disorder, age, y
18–29	1.02 (0.99, 1.06)	0.87 (0.84, 0.90)
30–44	1.30 (1.24, 1.37)	1.19 (1.13, 1.25)
45–64	1.32 (1.18, 1.47)	1.24 (1.10, 1.40)
Schizophrenia, age, y
18–29	0.50 (0.43, 0.58)	0.35 (0.31, 0.41)
30–44	1.08 (0.88, 1.34)	0.69 (0.56, 0.86)
45–64	0.66 (0.30, 1.41)	0.38 (0.16, 0.87)
Alcohol use disorder, age, y
18–29	0.38 (0.36, 0.39)	0.34 (0.33, 0.35)
30–44	0.51 (0.49, 0.54)	0.45 (0.43, 0.47)
45–64	0.56 (0.51, 0.62)	0.49 (0.45, 0.54)
Drug use disorder, age, y
18–29	0.41 (0.40, 0.43)	0.34 (0.33, 0.36)
30–44	0.56 (0.52, 0.60)	0.45 (0.42, 0.49)
45–64	0.67 (0.56, 0.81)	0.48 (0.39, 0.58)

Note. CI = confidence interval; IRR = incident rate ratio. Incidence rates and IRRs were not reported for veterans aged 65 years or older because of small sample size (n = 370) and low incidence of mental health diagnoses.

^aUnadjusted.

^bAdjusted for age, race/ethnicity, marital status, branch of service, rank, multiple deployments, and time to first Veterans Health Administration encounter.

In adjusted models, risk of incident MDD diagnosis was greater among women versus men in all age strata (18–29 years: IRR = 1.17; 95% CI = 1.14, 1.20; 30–44 years: IRR = 1.50; 95% CI = 1.46, 1.55; 45–64 years: IRR = 1.77; 95% CI = 1.67, 1.87). Compared with Whites, risk of incident MDD diagnosis was lower in African Americans and Hispanics aged 18 to 29 years and African Americans aged 30 to 44 years, and higher in Hispanics aged 30 to 44 years and 45 to 64 years (all P < .001).

In adjusted models, risk of incident BPD diagnosis was lower in women versus men aged 18 to 29 years (IRR = 0.87; 95% CI = 0.84, 0.90), though higher in women versus men aged 30 to 44 years (IRR = 1.19; 95% CI = 1.13, 1.25) and 45 to 64 years (IRR = 1.24; 95% CI = 1.10, 1.40). Compared with Whites, African Americans in all age groups had a lower risk of incident BPD diagnosis (all P < .001), and among Hispanics, risk of incident BPD diagnosis was lower for ages 18 to 29 years and 30 to 44 years and higher for age 45 to 64 years (all P < .001).

In adjusted models, risk of incident SZ diagnosis was lower in women compared with men in all age groups (18–29 years: IRR = 0.35; 95% CI = 0.31, 0.41; 30–44 years: IRR = 0.69; 95% CI = 0.56, 0.86; 45–64 years: IRR = 0.38; 95% CI = 0.16, 0.87). Compared with Whites, African Americans in all age groups had a higher risk of incident SZ diagnosis, and, among Hispanics, risk of incident SZ diagnosis was higher for those aged 45 to 64 years (all P < .001).

In adjusted models, risk of incident AUD diagnosis was lower in women compared with men in all age groups (18–29 years: IRR = 0.34; 95% CI = 0.33, 0.35; 30–44 years: IRR = 0.45; 95% CI = 0.43, 0.47; 45–64 years: IRR = 0.49; 95% CI = 0.45, 0.54). Compared with Whites, risk of incident AUD diagnosis was lower in Hispanics aged 18 to 29 years and African Americans aged 18 to 29 years and 30 to 44 years, and higher in Hispanics and African Americans aged 45 to 64 years (all P < .001).

In adjusted models, risk of incident DUD diagnosis was lower in women compared with men in all age groups (18–29 years: IRR = 0.34; 95% CI = 0.33, 0.36; 30–44 years: IRR = 0.45; 95% CI = 0.42, 0.49; 45–64 years: IRR = 0.48; 95% CI = 0.39, 0.58). Risk of incident DUD diagnosis did not differ between African Americans and Whites aged 18 to 29 years; however, African Americans aged 30 to 44 years and Hispanic Americans aged 18 to 29 years and 30 to 44 years had a lower risk of incident DUD diagnosis, and African Americans aged 45 to 64 years had a greater risk (all P < .001).

DISCUSSION

As women and racial/ethnic minority veterans increasingly use VHA services,^1–4,20 any differences by gender or race/ethnicity in diagnosis, treatment, and disease course of mental health conditions will be an important focus. This study reports on the incidence rates for 6 common mental health diagnoses among OEF/OIF/OND veteran users of VHA services by gender, age, and race/ethnicity. This is the first such prospective report of incident mental health conditions in the OIF/OEF/OND cohort, and the implications of these findings extend beyond that of previous prevalence studies by establishing temporal patterns between military service and sociodemographic (i.e., age, race/ethnicity, and gender) characteristics and mental health diagnoses.^5,6,8,9 Moreover, this study evaluates risk within different age cohorts, allowing us to precisely identify age-specific risk factors. Identifying high-risk individuals can inform screening, outreach, and care delivery for veterans who use the VHA. Employing informed preventive and early intervention strategies can lead to earlier diagnosis and treatment, which can prevent isolated episodes from becoming debilitating and costly chronic conditions.

Demonstrating the increase in incidence in PTSD, MDD, BPD, and SZ among veterans is particularly important to VHA policymakers as they plan to allocate resources for mental health services. Although some mental illnesses tend to require long-term outpatient care, each illness requires particular services. Examining incidence rates in our study revealed several patterns beyond those ascertained from existing prevalence studies that warrant further discussion.

Mental Health Diagnoses

Posttraumatic stress disorder.

Overall, incidence of PTSD diagnosis was higher than that of any other mental health diagnosis across all gender and age strata. This was expected, given that the majority of this cohort was deployed to combat theaters where they might be more likely to be exposed to Diagnostic and Statistical Manual of Mental Disorder Criterion A traumatic events. Whereas a small, previous study that controlled for self-report of exposure to combat and other trauma reported no gender differences in PTSD risk,²¹ we found that risk of incident PTSD diagnosis was lower for women younger than 45 years and greater for women aged 45 to 64 years, compared with same-age men. This was also generally found for both African American and Hispanic members of the cohort.

Information regarding specific exposures such as combat, military sexual trauma, other types of trauma, or deployment history (e.g., Iraq, Afghanistan, or other locations) were not available for our study. Thus, we were unable to ascertain the relative contribution that different types of exposures or exposure frequencies for men versus women—and African Americans or Hispanics versus Whites—might have had for the current findings.^21–23 In the absence of this information, we adjusted for possible indicators of trauma exposure, including military rank, branch of service, and being on multiple deployments. A previous study reported higher prevalence of PTSD diagnosis among men versus women OIF/OEF/OND veterans, but did not examine incidence of new PTSD diagnoses or examine differences by age cohorts, as was done in our study.²⁴ Stratifying by age in our study allowed us to identify age-by-gender and age-by- race/ethnicity differences in rates of PTSD diagnosis and we found that, among veterans aged 45 to 64 years, women are more likely to receive a new PTSD diagnosis than men.

Given the impact on health and quality of life associated with PTSD, it is essential for future studies to ascertain age-, gender-, and race/ethnicity–stratified risk factors for PTSD, including aspects of trauma exposure, to determine why age appears to differentially contribute to the risk of a PTSD diagnosis among women and minority veterans. In addition, future studies with longitudinal data on PTSD symptoms should examine sociodemographic differences between persistent and remitted cases of PTSD. These efforts will facilitate health care system planning and testing of risk surveillance models to support targeted interventions.

Major depressive disorder.

We found that incident MDD diagnosis peaked in midlife (age 30–44 years) for both men and women. In addition, we found a greater risk of incident MDD diagnosis for women versus men, and this risk increased with increasing with age. Thus, with each older age group, women were increasingly likely to receive a MDD diagnosis compared with men of the same age group. This increasing risk with age warrants further study to determine whether the incidence rate differences are attributable to true differences in the incidence of MDD in the population, or whether they are attributable to other factors, such as gender-based differences in clinician screening and diagnosis rates or differences in specialty mental health service use. There were also differences for African American and Hispanic veterans at differing age ranges versus Whites that similarly warrant further study to discern true differences in incidence from race/ethnicity-based differences in patient and provider behavior.

Bipolar disorder.

Incident BPD diagnosis was lower among younger women and higher among women aged 30 years and older relative to same-age men. Possible explanations for the observed age-by-gender differences include earlier age of onset or earlier treatment seeking among men in this cohort. Another possibility is differences in diagnosing tendencies, with men being more likely to be diagnosed with SZ than with BPD, a pattern observed in another VHA cohort.²⁵ Alternatively, younger women veterans may initially be given an MDD diagnosis that is later revised to BPD when hypomanic or manic symptoms emerge.^26,27 Future studies will require prospectively collected data on mental health symptoms, mental health–specific VHA service use, and diagnoses to elucidate the mechanisms driving our observed age-by-gender differences in diagnosis of BPD.

Schizophrenia.

Men across all age strata were more likely to receive an incident SZ diagnosis than were women. This finding is consistent with a previous study of VHA users with serious mental illness, which found that men were more likely than women to be diagnosed with SZ even after controlling for illness severity.²⁵ In addition, recent findings of a link between traumatic brain injury—a highly prevalent condition among OEF/OIF/OND veterans²⁸—and symptoms of schizophrenia,^29–31 demonstrate the need for further study of cohort trends in SZ incidence. Exploring potential links between traumatic brain injury and the emergence of mental health conditions is especially relevant for OEF/OIF veterans given higher rates of combat exposure and nonfatal injuries (which are often accompanied by traumatic brain injury) in this veteran group.²⁸

Alcohol use and drug use disorders.

Incident AUD and DUD diagnoses were highest in the youngest OEF/OIF/OND-era VHA users and declined with age. Consistent with previous studies involving this veteran cohort,^5–7 we found a decreased risk of incident AUD and DUD diagnosis in women compared with men. Differences that we observed in incident AUD and DUD diagnoses as a function of race/ethnicity by age may represent utilization trends and warrant a greater focus.^32,33

Age and Race/Ethnicity Comparisons

In general, we observed a lower risk of incident mental health diagnoses among younger African Americans and Hispanics and a higher risk among older African Americans and Hispanics compared with Whites. Risk of incident MDD, PTSD, BPD, and AUD diagnosis was lower among Hispanics aged 18 to 29 years and higher among Hispanics aged 45 to 64 years. We also observed differences in risk with age for DUD and SZ. Among African Americans, risk of incident PTSD or AUD diagnosis followed the same pattern—lower risk in younger age strata and higher risk in older age strata. We also observed differences in risk by age for MDD and DUD. Later primary care and mental health care initiation with VHA has been observed in racial/ethnic minorities from the OEF/OIF/OND cohort.³² Therefore, our observed greater incident diagnoses among older racial/ethnic minority groups could reflect delayed treatment seeking. Unfortunately, we were not able to determine whether differences in risk by race/ethnicity and age are attributable to treatment-seeking or diagnosing patterns.

A final noteworthy finding was the higher risk of incident SZ diagnosis and a lower risk of incident BPD diagnosis in African Americans versus Whites across all age strata, a finding that has been reported in other veteran populations.²⁷ As the number of racial/ethnic minorities using VHA services continues to rise, identifying disparities and establishing equitable treatment is an important area for future study.⁴

Limitations

There are several limitations that should be taken into consideration when one is interpreting the results of this study. First, the sample included only OIF/OEF/OND veterans enrolled in VHA health services and, therefore, may not reflect incidence rates of mental health diagnoses in the 39% of OIF/OEF/OND who are not enrolled in the VHA. Ideally, we would like to compare sociodemographic and deployment-related risk factors between veterans in the VHA and veterans who are not VHA users to assess the generalizability of our findings. In the absence of available sociodemographic and military service data on veterans not registered with the VHA with which to make these comparisons, the implications of our findings may not generalize to veterans not enrolled in VHA health services or to civilian populations.

In addition, incidence rates in this sample are based on data from the clinical records in the electronic health record, and not from a prospectively collected survey of an at-risk population. Therefore, we do not know how long veterans were experiencing symptoms before diagnosis, nor were we able to ascertain the severity of symptoms, a factor associated with initiation of treatment among veterans with mental illness.^34–36 This is particularly important given our findings that new diagnoses for women and minorities are higher in older age groups. Discerning whether these age-by-gender and age-by-race/ethnicity differences are attributed to age-specific adverse events and stressors in women and minorities, to health service use or clinician’s diagnosing tendencies, or to a combination of these factors is an important area of future study.

Furthermore, many of the VHA users in our sample received more than 1 incident mental health diagnosis during the follow up period (Table B), and we were unable to separate the effect of individual incident diagnosis from the potential effect of overall disease burden resulting from multiple incident diagnoses. In addition, we are not able to draw conclusions regarding exposure to combat or trauma and incidence of mental illness.

Finally, we examined interactions between age and gender and age and race/ethnicity, but we did not consider the effect of these 3 main effects at the same time. To further elucidate the relationship among age, race/ethnicity, gender, and incident mental health diagnosis, examining 3-way interactions among these predictors and incident mental health diagnoses is an important area for future study.

Despite these limitations, to our knowledge, this is the first study to examine incidence rates and risk factors for diagnoses of common mental health conditions among OEF/OIF/OND-era VHA users, and does so in an up-to-date sample spanning 14 years following initiation of the conflicts in Iraq and Afghanistan. Furthermore, the large, diverse sample in this study allows us to estimate age-specific risk factors with great precision. Future studies should continue to capitalize on this strength by comparing incidence rates over shorter time periods, such as before and after integration of mental health services into primary care or recent initiatives to improve services for women and minority veterans. Future studies should also employ incidence rates to compare the number of individuals in each high-risk group to the proportion receiving VA mental health services, to identify subpopulations who may not be accessing specialty mental health care and may require further outreach.