Sex Differences in Opioid Use Disorder Prevalence and Multimorbidity Nationally in the Veterans Health Administration

Abstract

Objective:

Opioid use disorder (OUD) is a significant problem among US veterans with increasing rates of OUD and overdose, and thus has substantial importance for service delivery within the Veterans Health Administration (VHA). Among individuals with OUD, several sex-specific differences have begun to emerge regarding co-occurring medical, psychiatric and pain-related diagnoses. The rates of such multimorbidities are likely to vary between men and women with OUD and may have important implications for treatment within the VHA but have not yet been studied.

Methods:

The present study utilized a data set that included all veterans receiving VHA health care during Fiscal Year (FY) 2012 (October 1, 2011 through September 30, 2012), who were diagnosed during the year with opioid dependence or abuse. VHA patients diagnosed with OUD nationwide in FY 2012 were compared by sex on proportions with OUD, and among those with OUD, on sociodemographic characteristics, medical, psychiatric and pain-related diagnoses, as well as on service use, and psychotropic and opioid agonist prescription fills.

Results:

During FY 2012, 48,408 veterans were diagnosed with OUD, 5.77% of whom were women. Among those veterans with OUD, few sociodemographic differences were observed between sexes. Female veterans had a higher rate of psychiatric diagnoses, notably mood, anxiety and personality disorders, as well as higher rates of pain-related diagnoses, such as headaches and fibromyalgia, while male veterans were more likely to have concurrent, severe medical co-morbidities, including hepatic disease, HIV, cancers, peripheral vascular disease, diabetes and related complications, and renal disease. There were few differences in health service utilization, with women reporting greater receipt of prescriptions for anxiolytic/sedative/hypnotics, stimulants and lithium. Men and women did not differ in receipt of opioid agonist medications or mental health/substance use treatments.

Conclusions:

There are substantial sex-specific differences in patterns of multimorbidity among veterans with OUD, spanning medical, psychiatric and pain-related diagnoses. These results illustrate the need to view OUD as a multimorbid condition and design interventions to target such multimorbidities. The present study highlights the potential benefits of sex-specific treatment and prevention efforts among female veterans with OUD and related co-occurring disorders.

Opioid use disorder (OUD) and overdose are critical and worsening public health problems throughout the United States (US), with 10.3 million individuals in the US reporting prescription opioid misuse and over 800,000 individuals endorsing heroin use in 2018; US Department of Health & Human Services, 2018). Reflecting this nationwide epidemic, the Veterans Health Administration (VHA) has observed a parallel rise in opioid-related diagnoses, overdoses, and deaths among the veterans it treats (Bohnert et al., 2011; Wyse et al., 2018).

Specifically in the US, 3.5% of males endorsed past year heroin use from 2007 to 2014, while 1.5% of women endorsed use heroin use during this same time period. Non-medical prescription opioid use prevalence was significantly higher, with 52.6% of males and 40.3% of women endorsing past year use, with rates escalating more rapidly among women (Marsh et al., 2018). While there has been a vigorous call to action to address the OUD and overdose crisis (Kuehn, 2017; Volkow & Collins, 2017), a recent commentary in The Lancet (Mazure & Fiellin, 2018), highlighted the potentially important lack of research on sex differences among patients with OUD, including sex-specific differences in risk, chronicity, and treatment of OUD. This gap, they point out, may have significant implications for treatment access, sustained intervention and prevention efforts for women with OUD (Mazure & Fiellin, 2018).

One of the important sex differences associated with OUD is that women have been found, in some studies, to have a high rate of co-morbid psychiatric disorders including major depressive and anxiety disorders, as well as posttraumatic stress disorder (PTSD; Grella et al., 2009; McHugh et al., 2013). Specifically, among those with comorbid PTSD, avoidance-related symptoms are associated with increased odds of OUD among women, while arousal/reactivity-related symptoms are associated with increased odds of OUD among men (Smith et al., 2016). Thus, symptoms related to psychiatric co-morbidities, such as negative affect and stress reactivity may be an especially frequent risk factor for OUD among women. Such sex issues may be particularly important in the case of female veterans who may have unique treatment needs compared to their male counterparts due, in part, to differences in their military experiences, with men experiencing more frequent combat trauma and women more frequent sexual trauma and rape (Back et al., 2015; Finlay et al., 2015; Kimerling et al., 2010; Teeters et al., 2017).

In addition to high rates of psychiatric co-morbidity, OUD is associated with significant medical comorbidity and consequences, including chronic pain (Bawor et al., 2015; McHugh et al., 2018). Overall, women are more sensitive to acute and chronic pain than men and are more likely to report both wide-spread and localized pain conditions, including fibromyalgia, migraine, and chronic headache (Bartley & Fillingim, 2013; Fillingim et al., 2009; Manubay et al., 2015; Serdarevic et al., 2017). Further, women are also more likely to have first used prescribed opioids which they obtain at a higher rate than men (Cicero et al., 2009; McHugh et al., 2013).

The increased rates of co-morbid psychiatric disorders and pain conditions among women suggest the importance of treating women with OUD from a multimorbidity perspective. Recent evidence suggests that “multimorbidity,” a term to describe the intersection of multiple psychiatric, addictive and medical disorders, as well as socio-environmental factors, has been associated with reduced quality of life and overall poor physical and mental health (Barnett et al., 2012; Bhalla & Rosenheck, 2018). Addressing illness from a “multimorbidity” perspective may enhance treatment effectiveness by addressing symptoms both from the perspective of the primary disorder as well as with consideration of co-occurring or underlying diagnoses (Bhalla et al., 2018; Guthrie et al., 2012). It has been postulated that the multimorbidity perspective may usefully inform treatment development, as it encourages the treatment of symptoms that span multiple-related disorders, as contrasted with a more narrow focus on treatment of a “primary” diagnosis (e.g., Bhalla et al., 2018; MacLean et al., 2018).

Given the complex nature of OUD and its common co-morbidities, the multimorbidity perspective may help elucidate underlying mechanisms that differ between women and men (Iheanacho et al., 2018; McHugh et al., 2018). This may be particularly true among veterans, who in some studies, report high rates of psychiatric and substance use disorders (SUD; Pemberton et al., 2016; Teeters et al., 2017; Vazan et al., 2013). Specifically, recent data highlights the high rates of polysubstance use among Veterans with OUD, which has been associated with increased medical diagnoses, overdose deaths, more severe symptom presentation, and poor treatment retention and outcomes (Lin et al., 2021; Manhapra et al., 2021).

In the present study, we used national data from the VHA to elucidate sex differences in multimorbidity associated with OUD and thus explore potential sex differences in processes underlying and impacting OUD. This information may facilitate the development of novel interventions and prevention efforts specific to women with OUD.

Methods

The present study was conducted in accordance with the Declaration of Helsinki; the Veterans Affairs Connecticut Healthcare System Human Subjects Subcommittee approved and monitored the study; a waiver of consent was obtained given the archival nature of the data used in the study.

Source of data

The present study includes all veterans receiving VHA health care during Fiscal Year (FY) 2012 (October 1, 2011 through September 30, 2012), who were diagnosed during the year with opioid dependence or abuse as indicated by International Classification of Diseases, 9th edition (ICD-9) codes 304.00, 304.01, 304.02, 304.70, 304.71, 304.72, 305.50, 305.51, or 305.52, considered here to be the rough equivalent of a DSM-V diagnosis of opioid use disorder (OUD), excluding those in remission. The dataset was constructed by the VHA Northeast Program Evaluation Center using clinical encounter data from the VHA Computerized Personal Record System (CPRS) and data regarding psychotropic and opioid agonist prescription fills from the VHA Drug Benefit Management database. All data was de-identified to protect veterans’ privacy.

Measures

Sociodemographics

Sample characteristics included sex, age, race, income, type of residence (urban, large rural, small rural, isolated rural; (Morrill et al., 1999), and service in Operation Enduring Freedom/Operation Iraqi Freedom (OEF/OIF) Middle East conflicts (based on dichotomous data provided to VHA by the Department of Defense). Additionally, service connected disability (determination regarding whether a Veteran was disabled due to medical condition as a result of his/her service) was assessed as being 50% or more, or less than 50%, based upon scale of 0%−100% disability rating. Receipt of VHA pension and identification as homeless (based upon documented use of homelessness resource encounter codes or ICD-9 code) were also represented dichotomously.

Psychiatric diagnoses

Psychiatric diagnoses were based upon ICD-9 codes for clinical encounters within FY2012. Diagnoses included: substance dependences (cannabis, alcohol, cocaine, barbiturate, amphetamine, hallucinogen, nicotine), schizophrenia, bipolar disorder, major depression, other depression (e.g., dysthymia), anxiety disorder, PTSD, adjustment disorder, personality disorder, dementia, and other psychiatric diagnoses. Additionally, these diagnoses were used to identify the following aggregated variables: any SUD, any psychiatric disorder, dual diagnosis, number of SUDs, number of psychiatric disorders.

Medical diagnoses

All medical diagnoses were also identified based upon ICD-9 codes in FY2012 clinical encounters. Diagnoses selected for inclusion in the present study were components of the Charlson Comorbidity Index (Charlson et al., 1987, 1994), a weighted index that allows for the classification of comorbid conditions and predicts cumulative mortality. Diagnoses composing the Charlson Comorbidity Index include: seizures, insomnia, nausea, pruritus, myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular accident, chronic obstructive airway, connective tissue disease, peptic ulcer disease, hepatic disease, diabetes mellitus and related complications, paraplegia, renal disease, human immunodeficiency virus (HIV) and cancer (Charlson et al., 1987, 1994).

Pain diagnoses

Pain diagnostic categories, also based upon the ICD-9 codes were created from the 27 ICD-9 pain diagnostic codes and included: headache, diabetic pain, musculoskeletal pain, fibromyalgia pain, skeletal muscle spasm pain, and herpetic pain, as well as other types of pain and a dichotomous variable representing any pain (Barry et al., 2015).

Psychotropic medications

Psychotropic medications prescribed were assessed dichotomously by six classes: antidepressants, antipsychotics, anxiolytics/sedatives/hypnotics, stimulants, anticonvulsants/mood stabilizers, and lithium. Prescriptions for buprenorphine/naloxone, injectable naltrexone and attendance at methadone maintenance clinics were also documented.

Service use

The present study included data on all VHA medical and psychiatric services used by this sample. Mental health treatment (including inpatient and outpatient treatment) was identified from inpatient bed section codes and outpatient clinic “stop codes, indicating a variety of treatments encounters in specialty clinics during FY2012 and specifically the number of emergency room visits, medical/surgical visits, SUD visits, general psychiatric visits, and all outpatient visits.

Data analysis

Bivariate comparisons between female and male veterans with OUD addressed sociodemographic characteristics, psychiatric diagnoses, general medical diagnoses, pain diagnoses, additional SUD diagnoses, psychotropic medications prescribed, and services used. Due to the large sample sizes, which would render virtually all comparisons statistically significant by usual p < .05 criteria, regardless of how small the difference, effect sizes were used for bivariate comparisons. Thus, risk ratios were utilized to compare sexes on categorical variables (e.g., psychiatric, SUD, medical, and pain diagnoses), while Cohen’s d was used for comparison of continuous variables (e.g., age, number outpatient visits). Risk ratios >1.5 or <0.67 and Cohen’s d values >0.2 or < −0.2 were considered to be indicative of substantial differences. The variables reaching this threshold were included in subsequent multivariate analyses. Previous studies involving such large VHA sample sizes have used similar methodology for variable selection (Arout et al., 2018; MacLean et al., 2018).

Generalized estimation equations (GEE) was then used to identify factors that were independently associated with sex among veterans diagnosed with OUD including a random effect for facility to adjust for the potential correlation of observations with in VA facility. Data were analyzed using Statistics Analysis software (SAS).

Results

Sample

The present study included 48,408 veterans who had an OUD diagnosis in FY2012. The vast majority were men (45,614; 94.23%) with 2,794 (5.77%) women. Among those receiving VHA care in FY 2012, 0.97% (45,614 of 4,693,458) men were diagnosed with OUD, compared to 0.85% (2,794 of 330,087) of women.

Female versus male veterans with OUD

Sociodemographic characteristics were similar between female and male veterans with OUD (Table 1). In fact, only substantial differences between the sexes were observed in age and VHA pension, with female veterans nearly 7 years younger when compared to male veterans with OUD (44.86 years old versus 51.01 years old, respectively) and less likely to be receiving a VHA pension (2.68% versus 9.37%, respectively).

Table 1.

Comparisons of Sociodemographic Correlates of OUD in Female and Male Veterans.

	Female	Male
	M (SD)	M (SD)	Cohen’s d
	N = 2,794	N = 45,614
Age	44.86 (11.40)	51.01 (12.99)	−0.48*
Income	17,144.47 (18,122.27)	16,721.57 (23,336.08)	0.02
	N (%)	N (%)	Risk ratio
Race
White	2,079 (78.96%)	31,246 (72.46%)	1.09
Black	454 (17.24%)	29,095 (23.80%)	0.72
Residence
Urban	2,155 (79.05%)	35,554 (79.73%)	0.99
Large rural	280 (10.27%)	4,029 (9.03%)	1.14
Small rural	175 (6.42%)	2,935(6.58%)	0.98
Isolated rural	116 (4.26%)	2,077 (4.66%)	0.91
OEF/OIF	341 (12.20%)	6,517 (14.29%)	0.85
Service connection
50% or more	972 (34.79%)	11,836 (25.95%)	1.34
Less than 50%	508 (18.18%)	7,078 (15.52%)	1.17
VHA pension	75 (2.68%)	4,275 (9.37%)	0.29*
Homelessness	832 (29.78%)	13,903 (30.48%)	0.98

Note. M = Mean; SD = Standard Deviation; OEF/OIF = Operation Enduring Freedom/Operation Iraqi Freedom; VHA = Veteran’s Health Administration.

^*Cohen’s d > 0.20 or < −0.20; Risk ratio >1.50 or <0.50.

Overall, female veterans with OUD had more comorbid psychiatric diagnoses than their male counterparts (2.80 diagnoses versus 2.03, Cohen’s d = .50; Table 2). Female veterans with OUD were over one and a half times as likely to have diagnoses of major depression and anxiety disorders. They were also more likely to have been diagnosed with bipolar disorder or a personality disorder. However, the risk ratio for PTSD did not reached the substantial threshold (Risk Ratio = 1.49), with 47.75% of women having a diagnosis of PTSD, compared to 33.51% of men. There were no other substantial differences in non-opioid SUD diagnoses between the sexes.

Table 2.

Comparisons of Psychiatric and Medical Diagnoses Among Female and Male Veterans With OUD.

	Female	Male
	N (%)	N (%)	Risk ratio
Psychiatric diagnoses
Schizophrenia	92 (3.29%)	2,385 (5.23%)	0.63
Bipolar disorder	615 (22.01%)	5,127 (11.24%)	1.96*
Major depression	960 (34.36%)	9,978 (21.87%)	1.57*
Other depression	1,698 (60.77%)	22,470 (49.26%)	1.23
PTSD	1,334 (47.75%)	15,284 (33.51%)	1.42
Anxiety disorder	1,269 (45.42%)	13,756 (30.16%)	1.51*
Adjustment disorder	341 (12.20%)	5,044 (11.06%)	1.10
Personality disorder	651 (23.30%)	4,094 (8.98%)	2.60*
Dementia	4 (0.14%)	142 (0.31%)	0.46*
Narcolepsy	4 (0.14%)	41 (0.09)	1.59*
Other psychiatric diagnosis	735 (26.31%)	12,115 (26.56%)	0.87
Any psychiatric disorder	2,789 (99.82%)	45,495 (99.74%)	1.00
Medical diagnoses
Seizures	63 (2.25%)	797 (1.58%)	1.29
Insomnia	319 (11.42%)	4,202 (9.21%)	1.24
Nausea	286 (10.24%)	3,030 (4.45%)	2.30*
Pruritus	60 (2.15%)	672 (1.47%)	1.46
Myocardial infarction	16 (0.57%)	493 (1.08%)	0.53
Congestive heart failure	1,627 (58.23%)	24,290 (53.25%)	1.09
Peripheral vascular disease	52 (1.86%)	1,851 (4.06%)	0.46*
Cerebrovascular accident	84 (3.01%)	1,646 (3.61%)	0.83
Chronic obstructive airway	571 (20.44%)	8,142 (17.85%)	1.14
Connective tissue disease	51 (1.83%)	357 (0.78%)	2.33*
Peptic ulcer disease	36 (1.29%)	563 (14.76%)	1.04
Hepatic disease	199 (7.12%)	6,733 (14.76%)	0.48*
Diabetes mellitus	267 (9.56%)	7,415 (16.26%)	0.59
Complications of diabetes	44 (1.57%)	1,703 (3.73%)	0.42*
Paraplegia	36 (1.29%)	489 (1.07%)	1.20
Renal disease	44 (1.57%)	1,535 (3.37%)	0.47*
HIV	18 (0.64%)	820 (1.80%)	0.36*
Cancer	108 (3.87%)	2,580 (5.66%)	0.68
Pain diagnoses
Headache	723 (25.88%)	4,618 (10.12%)	2.56*
Diabetic pain	42 (1.50%)	1,522 (3.34%)	0.45*
Musculoskeletal pain	1,381 (49.43%)	18,217 (39.94%)	1.24
Fibromyalgia pain	381 (13.64%)	1,321 (2.90%)	4.71*
Skeletal muscle spasm pain	121 (4.33%)	1,327 (2.91%)	1.49
Herpetic pain	44 (1.57%)	592 (1.30%)	1.21
Any pain	1,554 (55.62%)	20,003 (43.85%)	1.11
SUD diagnosis
Cannabis	499 (17.86%)	9,011 (19.75%)	0.90
Alcohol	1,048 (37.51%)	20,447 (44.83%)	0.84
Cocaine	655 (23.44%)	12,480 (27.36%)	0.86
Barbiturate	323 (11.56%)	4,176 (9.16%)	1.26
Amphetamine	158 (5.65%)	2,329 (5.11%)	1.11
Hallucinogen	19 (0.68%)	268 (0.59%)	1.16
Nicotine	1,392 (49.82%)	22,168 (48.60%)	1.03
Dual diagnoses (SUD/Psych)	2,681 (95.96%)	42,977 (94.22%)	1.02
	M (SD)	M (SD)	Cohen’s d
Number and comorbidity of diagnoses
Number of psychiatric diagnoses	2.80 (1.69)	2.03 (1.57)	0.49*
Number of SUD diagnoses	1.83 (1.19)	1.95 (1.17)	−0.10
Charleson Comorbidity Index	1.28 (1.50)	1.59 (1.92)	−0.16

Note. M = Mean; SD = Standard Deviation; PTSD = posttraumatic stress disorder; HIV = human immunodeficiency virus; SUD = substance use disorder; Psych = psychiatric.

^*Cohen’s d > 0.20 or < −0.20; Risk ratio >1.50 or < 0.50.

Male veterans with OUD were more likely to have accompanying medical complications, such as higher rates of hepatic disease, HIV, cancers, peripheral vascular disease, diabetes-related complications, and renal disease (Table 2). Women, in contrast, were more than two times more likely to have diagnoses of nausea and connective tissue disease. Additionally, differences in pain diagnoses also emerged as female veterans with OUD were two and a half times more likely to be diagnosed with headaches and over four times as likely to have fibromyalgia (Table 2). However, of note, male veterans with OUD were more likely to be diagnosed with diabetic pain.

There were no substantial differences in service use between the sexes, including inpatient medical and psychiatric hospitalizations, and outpatient medical and psychiatric visits, including, most notably, specialty substance abuse treatment visits (Table 3). However, over 90% of female veterans received psychotropic medications as compared to only 82.46% of men. Women were over two times as likely to have been prescribed stimulants or lithium as compared to their male counterparts. There were no substantial differences in the Medication Assisted Treatment (MAT) with buprenorphine, injectable naltrexone, or methadone.

Table 3.

Comparisons of Treatments and Services Used Among Female and Male Veterans With OUD.

	Female	Male
	N (%)	N (%)	Risk ratio
Psychotropic medications
Antidepressant	2,242 (80.24%)	30,637 (67.17%)	1.19
Antipsychotic	1,056 (37.80%)	13,052 (28.61%)	1.32
Anxiolytic/Sedative/Hypnotic	1,530 (54.76%)	17,738 (38.89%)	1.41
Stimulant	120 (4.29%)	979 (2.15%)	2.00*
Anticonvulsant/Mood stabilizer	1,282 (45.88%)	16,014 (35.11%)	1.31
Lithium	159 (5.69%)	1,220 (2.67%)	2.13*
Any psychotropic medications	2,520 (91.84%)	36,515 (82.46%)	1.03
Medication assisted treatment
Buprenorphine	495 (17.71%)	7,568 (16.59%)	1.07
Methadone maintenance	332 (11.88%)	7,887 (17.29%)	0.69
Injectable naltrexone	10 (0.65%)	143 (3.14%)	1.14
Overall services
Any MH inpatient treatment	708 (25.34%)	10,204 (22.37%)	1.13
Any MH treatment	2,557 (91.57%)	40,237 (88.21%)	1.04
SUD medications prescribed	759 (27.17%)	14,206 (31.11%)	0.87
	M (SD)	M (SD)	Cohen’s d
Services used
Number of opioids prescribed^	8.57 (11.09)	7.43 (10.82)	1.15
Emergency room visits	2.41 (4.70)	2.19 (4.13)	0.05
Medical surgical visits	14.03 (14.57)	11.90 (13.00)	0.16
All outpatient visits	52.33 (62.87)	47.56 (60.43)	0.08
SUD visits	19.82 (43.89)	21.61 (45.76)	−0.04
General psychiatric visits	18.48 (31.02)	14.05 (27.86)	0.16
MH/SUD outpatient visits	38.30 (59.18)	35.66 (57.79)	0.05

Note. M = Mean; SD = Standard Deviation;

^{^ =}the number of opiate prescriptions filled by each patient during the year; MH = mental health; SUD = substance use disorder.

^*Cohen’s d > 0.20 or <−0.20; Risk ratio >1.50 or < 0.50.

Multivariate analysis of female and male veterans with OUD

Multivariate analysis (GEE) identifying independent factors showed men were older, on average, and scored more severely on the Charlson Comorbidity Index (Table 4). Women, in contrast, were two to four times as likely to have diagnoses of headache pain, fibromyalgia pain, nausea, and connective tissue disease. Women also had an increased likelihood of being diagnosed with several psychiatric disorders as they were 1.29–2.17 times more likely to have diagnoses of bipolar disorder, major depressive disorder, anxiety disorders, or personality disorders.

Table 4.

Generalized Estimation Equation Analysis (With Random Effect for VA Facility) of Correlates of Being Female Among Veterans Diagnosed With OUD.

	N = 48,408
	Estimate	SE	OR	95% CI
Sociodemographics
Age	−0.03	0.01	0.97	0.97–0.98
Medical/pain diagnoses
Charleson Comorbidity Index	−0.06	0.01	0.95	0.92–0.97
Headache	0.74	0.05	2.10	1.90–2.31
Fibromyalgia pain	1.45	0.07	4.27	3.75–4.87
Nausea	0.64	0.07	1.90	1.64–2.20
Narcolepsy	0.40	0.63	1.49	0.44–5.10
Connective tissue disease	1.00	0.17	2.72	1.96–3.78
Psychiatric diagnoses
Bipolar disorder	0.55	0.05	1.74	1.57–1.92
Major depressive disorder	0.42	0.04	1.52	1.39–1.66
Anxiety disorders	0.26	0.04	1.29	1.19–1.40
Personality disorders	0.77	0.05	2.17	1.96–2.40

Note. SE = standard error; OR = odds ratio; CI = confidence interval; PTSD = posttraumatic stress disorder.

Discussion

The present study found that female veterans with OUD, tended to have higher rates of psychiatric disorders and several pain diagnoses, but less severe medical comorbidity than men, despite few differences in sociodemographic characteristics from their male counterparts. There were no sex differences in other SUD diagnoses. While there were no differences between men and women regarding overall health service use, especially substance use specialty services and MAT; women with OUD were more likely to have filled prescriptions for stimulants and lithium even after controlling for psychiatric co-morbidities. These analyses thus highlight several sex-specific findings that impact the assessment and perhaps the treatment of women veterans with OUD.

The results of this study extend previous research, showing that women with OUD in VHA have higher rates of co-morbid psychiatric diagnoses than their male counterparts, as they do in civilian populations, (Grella et al., 2009; McHugh et al., 2013, 2018). The results are also similar to those of other VHA studies which also showed higher rates of psychiatric co-morbidities in female as compared to male veterans (Finlay et al., 2015; Seal et al., 2009). It is notable that the co-morbid psychiatric conditions observed in females with OUD in the present study tended to be mood and anxiety disorders, including major depression, bipolar disorder, and anxiety disorder (ranging from 22.01% to 45.42%). Personality disorders were also present in approximately 23% of female veterans with OUD in this study. Given that personality disorders are generally defined by a range of affect intensity, lability and appropriateness, as well as impairments in cognitions, interpersonal functioning and impulsivity (American Psychiatric Association, 2013), these findings suggest that additional research is needed to explore females’ motives for opioid use. This is especially important as previous research suggests that females are more likely to be motivated to use non-prescription opioids to cope with emotional distress than men (Back et al., 2011; McHugh et al., 2013; Smith et al., 2016). The available data did not include information on veteran’s motivation for substance use and whether OUD diagnoses preceded other diagnoses. Thus, we are unable to confirm that affect has a causal role in OUD among women. However, it remains salient that mood, anxiety and personality disorders are more frequent among women and they have been reported to underlie OUD among women, as well as negatively impacting the treatment of both psychiatric and substance use diagnoses (for review see Greenfield, Brooks, et al., 2007).

The present study also demonstrated that women with OUD were approximately 7 years younger than their male counterparts. These results extend previous findings that while there are mixed results regarding initiation of opioid use among women, there is an acceleration in frequency of regular use and treatment entry among women (Back et al., 2011; Hernandez-Avila et al., 2004). Women report treatment entry, following fewer years of regular use of opioids when compared to men, illustrating a “telescoping pattern.” This pattern, in which onset of substance use is later in women than men, but they nevertheless enter substance use treatment earlier than men, has been well established in the substance use literature (Hernandez-Avila et al., 2004; McHugh et al., 2013). The observed difference in age also likely explains the differences observed in severe medical complications, in that men with OUD had overall more serious medical diagnoses (e.g., peripheral vascular disease, and diabetes, as well as hepatic and renal disease) as compared to women with OUD.

In addition to the observed sex differences in concomitant psychiatric disorders and medical diagnoses, women with OUD were more likely to be diagnosed with headache and fibromyalgia, as well as diagnosis of connective tissue disease and nausea. These findings support previous research showing that women are at increased risk for chronic pain (Bartley & Fillingim, 2013; Cicero et al., 2009). Female VHA patients have been shown in a previous VHA study to have higher rates of fibromyalgia when compared to males, highlighting the need to screen and treat females with OUD for chronic pain with non-opioid therapies (Arout et al., 2018).

The observed higher rates of psychiatric disorders and pain conditions among women, as well as the higher rates of concurrent medical diagnoses among men in the present analyses emphasize the importance of treating OUD from a multimorbidity perspective, especially in the VHA, as both men and women at VHA have been shown to have multiple, concurrent medical and psychiatric diagnoses (Bhalla & Rosenheck, 2018). It is likely that implementing broad interventions that address OUD, as well as the span the symptoms of the multiple co-occurring/underlying diagnoses would improve treatment response and overall health among individuals with OUD (Bhalla et al., 2018; Guthrie et al., 2012; Iheanacho et al., 2018; MacLean et al., 2018). Specifically, it has been proposed that pain and addiction should be conceptualized as a multimorbidity state and treated within an integrative approach, utilizing dependency, comorbidity, and chronic pain treatments along with an effort toward polypharmacy reduction (Manhapra & Becker, 2018). VHA patients, would likely benefit from this integrative, interdisciplinary treatment approach. However, additional research of OUD from a multimorbidity perspective is needed in order to design and implement appropriate treatments to address opioid use as well as its common co-morbidities. Additionally, researching OUD as a multimorbidity condition may elucidate important sex-specific differences that allow interventions to be further tailored to address the medical, psychiatric or sociodemographic needs of men and women with OUD.

The present results are an initial step in this line of research, showing similar levels of medical and psychiatric service use between men and women veterans with OUD. Notably, there were no significant differences in mental health and SUD visits, as well as in receipt of prescriptions for MAT (i.e., opioid agonist treatments). This is a promising finding, as women seeking treatment for SUD have been reported to face substantial barriers to such care (Greenfield, Brooks, et al., 2007). A recent systematic review suggests that women veterans are likely to benefit from sex-specific SUD services, with reports of increased treatment utilization and more regular attendance, as well as greater satisfaction with such treatment (Hoggatt et al., 2015). Sex-specific mental health services have been incorporated into women veteran’s primary care settings at VHA medical centers and may partially explain the evidence of reduced barriers to mental health and SUD treatment observed in the present study (US Department of Veterans Affairs, 2017). Women veterans receiving care from VHA facilities that offer services tailored for women have been shown to be more receptive to SUD treatment and are more likely to engage in overall treatment (Lewis et al., 2016; Oliva et al., 2012).

There has been an emerging discussion in recent years regarding the efficacy of sex-specific, or women-only, treatment modalities for psychiatric disorders and SUDs (Greenfield & Grella, 2009; Greenfield, Trucco, et al., 2007; Messina et al., 2012). These tailored treatments have been shown, in the few trials conducted thus far, to be comparable in efficacy to mixed-gender treatment groups regarding reduction in mean number of days of substance use during treatment and six months post-treatment, but women participants in such programs report improved treatment satisfaction, greater reduction in trauma-related symptoms and greater retention/improvement at follow-up (Greenfield, Brooks, et al., 2007; Greenfield et al., 2014; Greenfield, Trucco, et al., 2007; Messina et al., 2012). Furthermore, a meta-analysis demonstrated that women-only versus mixed-sex substance abuse treatment programs reduced psychiatric problems, and that creating “enhanced” treatment programs tailored to address the unique needs of women promoted healthy behaviors (Orwin et al., 2001).

While single-sex women’s treatment groups have resulted in comparable results to mixed-gendered groups regarding substance use outcomes, female specific treatments may be particularly important among veterans, as substance use groups at the VHA are often male-dominated (Greenfield, Brooks, et al., 2007; Greenfield et al., 2014; Teeters et al., 2017). Future research is needed to understand the factors underlying the superior engagement and satisfaction in sex-specific interventions as well as to develop sex-tailored treatments that address the unique needs of women. Treatments targeting affect regulation and stress-reactivity may prove to be especially useful for female veterans and warrant continued exploration.

Strengths and limitations

The present study highlights important sex differences among veterans diagnosed with OUD, which identify distinct treatment needs of women veterans with OUD and suggest directions for treatment development. However, several methodological limitations warrant mention. First, this study utilized data drawn from VHA administrative databases in FY2012. The VHA population disproportionately represents older men and women (National Center for Veterans Analysis & Statistics, 2016), thus generalizability to non-VHA populations of veterans and non-veterans may be limited. Additionally, all ICD-9 diagnoses were based on judgements of diverse clinicians rather that assessments using standardized measures. However, the present study does reflect real-world, clinical practice at the VHA. While these practices may be subject to potential clinician variability and biases, they offer valuable insight into the multimorbidity and treatment patterns as perceived by VHA clinicians. The present data also lacked important details regarding OUD history, type of opioid use (illicit vs. prescription) or dose, motivations for use, and route of administration (oral, smoked, nasal, or IV). Additionally, details related to service use and the availability of sex-specific services, including access to VHA women’s health centers, were not available and there were no details regarding the temporary nature or duration of a diagnosis. Future research on smaller samples using more extensive assessment tools would provide further information about sex differences in opioid use and VHA service delivery.

Conclusions

The present national study of VHA patients with OUD demonstrated substantial sex-specific differences in patterns of multimorbidity. These differences show that females have higher rates of psychiatric co-morbidities, especially mood disorders, and some pain disorders, with no evidence of reduced access to specialized substance use services. These findings highlight the potential for development and dissemination of sex-specific and/or sex-sensitive services that could improve outcomes for women veterans with OUD.