HIV Prevalence and HIV Screening History Among a Veterans Association Cohort of People with Opioid and Alcohol Use Disorders

Abstract

Background

Veterans face high risk for HIV and substance use, and thus could be disproportionately impacted by the HIV and substance use disorder (SUD) “syndemic.” HIV prevalence among veterans with SUD is unknown.

Objective

To project HIV prevalence and lifetime HIV screening history among US veterans with alcohol use disorder (AUD), opioid use disorder (OUD), or both.

Design

We conducted a retrospective cohort analysis using national Veterans Health Administration (VHA) data.

Participants

We selected three cohorts of veterans with SUD: (1) AUD, (2) OUD, and (3) AUD/OUD. Included veterans had ICD codes for AUD/OUD from 2016 to 2022 recorded in VHA electronic medical records, sourced from the VA Corporate Data Warehouse (CDW).

Main Measures

We estimated HIV prevalence by dividing the number of veterans who met two out of three criteria (codes for HIV diagnosis, antiretroviral therapy, or HIV screening/monitoring) by the total number of veterans in each cohort. We also estimated lifetime HIV screening history (as documented in VHA data) by cohort. We reported HIV prevalence and screening history by cohort and across demographic/clinical subgroups.

Key Results

Our sample included 669,595 veterans with AUD, 63,787 with OUD, and 57,015 with AUD/OUD. HIV prevalence was highest in the AUD/OUD cohort (3.9%), followed by the OUD (2.1%) and AUD (1.1%) cohorts. Veterans of Black race and Hispanic/Latinx ethnicity, with HCV diagnoses, and aged 50–64 had the highest HIV prevalence in all cohorts. Overall, 12.8%, 29.1%, and 33.1% of the AUD/OUD, OUD, and AUD cohorts did not have history of HIV screening, respectively.

Conclusions

HIV prevalence was high in all SUD cohorts, and was highest among veterans with AUD/OUD, with disparities by race/ethnicity and age. A substantial portion of veterans had not received HIV screening in the VHA. Findings highlight room for improvement in HIV prevention and screening services for veterans with SUD.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11606-023-08452-5.

INTRODUCTION

In the United States (US), 2.7 million, 29.7 million, and 1.2 million people have opioid use disorder (OUD), alcohol use disorder (AUD), and HIV, respectively.^1–3 People with HIV are more likely to have substance use disorders (SUD) than people without HIV, and people with SUD have higher HIV risk than those without SUD.^4–8 Additionally, substance use exacerbates HIV morbidity, mortality, and transmission.^4–6 In a large study of veterans, HIV prevalence was markedly higher than in the general US population.⁹ Other studies have found that veterans have higher SUD and problematic substance use rates, partly due to chronic pain and psychiatric co-morbidities.^10–13 Thus, US veterans may be at particularly high risk for having HIV and SUD.^{9, 10}

The intersection between SUD and HIV is a “syndemic” that compounds negative impacts for patients and can burden healthcare systems.^{14, 15} In a national survey, 24% of people with HIV reported needing alcohol/drug use treatment.¹⁶ The US Centers for Disease Control and Prevention (CDC) reports that people who inject drugs (PWID) are 22 times more likely to acquire HIV than people who do not, and recognizes opioid and alcohol use as risk factors for HIV acquisition.^{6, 17, 18} Evidence suggests that similar intersections exist among US veterans; Veterans Aging Cohort Study (VACS) analyses have found that veterans with HIV are more likely than those without HIV to have SUD.^{19, 20} Additionally, another study found that among veterans with AUD, those with HIV were more likely than those without HIV to have additional, non-alcohol SUD.²¹

While studies have described higher substance use among people with HIV and higher HIV risk among people with SUD, fewer studies describe HIV prevalence among people with SUD. Prior research suggests HIV prevalence may be elevated in people with SUD, but studies are limited to PWID or people with OUD.^{16, 22} Similarly, while evidence supports increased substance use among veterans with HIV, HIV prevalence among veterans with different SUDs remains unknown. Such research is critical given the high prevalence of adverse social determinants of health, co-occurring psychiatric disorders and medical complications, and mortality among veterans, as well as future downstream effects of the Iraq and Afghanistan conflicts (the longest war in US history).^23–28

The United States Preventive Services Task Force (USPSTF), CDC, and Veterans Health Administration (VHA) recommend at least one universal HIV screening for all adults aged 18–65, and annual screening for those at high risk for HIV.^29–32 The “high risk” category includes men who have sex with men, people with a sex partner with HIV, and PWID. While SUD may contribute to HIV risk, diagnosis with OUD and/or AUD alone is not an independent criterion for HIV screening. Consequently, physicians following USPSTF, CDC, or VHA guidelines may not annually screen veterans with SUD for HIV. The Substance Abuse and Mental Health Services Administration (SAMHSA) additionally suggests that providers offer ongoing HIV screening for people with SUD.⁴ Actual HIV screening rates in current VHA practices are uncertain; a 2009 study found that veterans with SUD have low HIV screening rates.³³

Understanding HIV prevalence and screening history among veterans with SUD could provide insight into unmet need for increasing and streamlining access to HIV and SUD treatment and prevention, and could have implications for VHA system improvement.

METHODS

Design, Population, and Data Collection

We conducted a retrospective cohort analysis using national VHA data. The VHA dataset includes diagnoses (International Classification of Disease [ICD] codes), patient encounters (Current Procedural Terminology [CPT] codes), medication fills (National Drug Codes [NDC]), and procedure codes (Healthcare Common Procedure Coding System [HCPC]) for all patients recorded in the VHA Corporate Data Warehouses (CDW), and the VA Informatics and Computing Infrastructure (VINCI). The cohort we created used data from 2016 to 2022 and included records predating this timeframe (e.g., prior diagnoses).³⁴ The study was deemed exempt from VA Boston IRB and executed with approval from the Research & Development Committee, and meets Strengthening the Reporting of Observational Studies in Epidemiology guidelines for cross-sectional studies.³⁵

We selected three SUD veteran cohorts for analysis: (1) AUD only (AUD cohort), (2) OUD only (OUD cohort), and (3) co-occurring AUD and OUD (AUD/OUD cohort). The cohorts included all veterans in the VHA with AUD and/or OUD, as indicated by having at least one inpatient or two outpatient encounters with an ICD-10 code for AUD or OUD in any healthcare encounters during the 2016–2022 study period (excluding remission codes, Supplemental Table 1).³⁴

Study Variables

Demographic and Clinical Characteristics

SUD cohort demographic and clinical characteristics included age category (18–29, 30–39, 40–49, 50–64, 65 +), sex documented in medical records (male, female), race (white, American Indian/Alaska Native [AIAN], Asian, Black, multiple, Native Hawaiian or Pacific Islander [NHPI], unknown), ethnicity (Hispanic/Latinx, non-Hispanic/Latinx), and presence of hepatitis C virus (HCV) diagnosis (yes, no). Demographic characteristics were reported at earliest encounter during the study period; categories were based on documented categories in the medical record. HCV diagnosis presence was reported as “yes” if an individual had an ICD-10 code for HCV during the study period (Supplemental Table 1).

HIV Prevalence

HIV prevalence, defined as the number of veterans with HIV within AUD, OUD, and AUD/OUD cohorts divided by the total number of veterans in each cohort, was our primary outcome. Veterans were considered to have HIV if they met at least two out of three criteria at any point in their VHA medical history: (1) one or more ICD-9/ICD-10 diagnostic code for HIV, (2) one or more non-prophylaxis antiretroviral therapy HIV treatment, identified using NDC codes and drug names, or (3) one or more positive HIV screening test (i.e., antibody, antigen/antibody, or nucleic acid test) or one or more HIV monitoring test (e.g., viral load or CD4 test), identified by laboratory test names (Supplemental Table 1).³⁶

HIV Screening History

We evaluated the proportion of veterans who ever had HIV screening (i.e., antibody, antigen/antibody, or nucleic acid test), defined as the number of veterans who had a record of receiving an HIV test at any point in their VHA medical history or who had HIV (as defined above), indicating a prior HIV screening, within the AUD, OUD, and AUD/OUD cohorts divided by the total number of veterans in each cohort.

Analyses

We described demographic and clinical characteristics within AUD, OUD, and AUD/OUD cohorts, and conducted chi-square tests to test for significant differences between cohorts. Overall HIV prevalence and screening history were calculated separately for the three SUD cohorts. Stratified subgroup analyses were similarly conducted to project HIV prevalence and screening history within each cohort by demographic and clinical characteristics. We tested for significant differences in HIV prevalence and screening history between the AUD, OUD, and AUD/OUD cohorts, deriving overall differences between SUD groups and within demographic and clinical characteristic subgroups, using linear probability models with robust standard errors. We applied Benjamini–Hochberg corrections to reduce false discovery rate across multiple subgroup analyses, and separately for pairwise post hoc comparisons.³⁷ As a sensitivity check, we performed the same approach for age-adjusted HIV prevalence results.

All analyses were descriptive. Use of linear modeling with a binary outcome expresses coefficients as proportions in percent values, with mean comparisons reported herein relating to differences in the percent of veterans with HIV, and screened for HIV (separately), by SUD group and demographic or clinical characteristics. All analyses were conducted in R, version 4.1.2.

RESULTS

Demographic and Clinical Characteristics

We included 790,397 VHA patients, with 669,595 (84.7%) in the AUD cohort, 63,787 (8.1%) in the OUD cohort, and 57,015 (7.2%) in the AUD/OUD cohort. All cohorts were majority male (AUD: 93.7%, OUD: 91.0%, AUD/OUD: 93.7%), white (AUD: 66.1%, OUD: 78.5%, AUD/OUD: 69.8%), and non-Hispanic (AUD: 91.5%, OUD: 94.3%, AUD/OUD: 93.3%). The second largest racial/ethnic category was Black (AUD: 24.3%, OUD: 14.4%, AUD/OUD: 23.5%). A plurality was between ages 50 and 64 (AUD: 34.4%, OUD: 35.9%, AUD/OUD: 42.4%). The AUD/OUD cohort had the highest HCV diagnosis prevalence (28.1%, compared with 16.6% and 6.5% of the OUD and AUD cohorts, respectively) (Table 1; Supplemental Table 2).

Table 1.

Demographic and Clinical Characteristics in Veteran AUD, OUD, and AUD/OUD Cohorts

Characteristic1	n, %2, by cohort			P-value3
	AUD n = 669,595 (84.7%)	OUD n = 63,787 (8.1%)	AUD/OUD n = 57,015 (7.2%)	OUD vs. AUD	AUD/OUD vs. AUD	AUD/OUD vs. OUD
Age category
18–29	45,867 (6.8)	2851 (4.5)	5580 (9.8)	< 0.001	< 0.001	< 0.001
30–39	106,737 (15.9)	9720 (15.2)	12,265 (21.5)
40–49	90,501 (13.5)	6937 (10.9)	7402 (13.0)
50–64	230,227 (34.4)	22,910 (35.9)	24,172 (42.4)
65 +	196,258 (29.3)	21,369 (33.5)	7596 (13.3)
Sex in medical record
Male	627,418 (93.7)	58,062 (91.0)	53,404 (93.7)	< 0.001	0.750	< 0.001
Female	42,174 (6.3)	5725 (9.0)	3611 (6.3)
Race
White	442,493 (66.1)	50,092 (78.5)	39,821 (69.8)	< 0.001	< 0.001	< 0.001
AIAN	7398 (1.1)	589 (0.9)	577 (1.0)
Asian	4863 (0.7)	207 (0.3)	179 (0.3)
Black	162,703 (24.3)	9169 (14.4)	13,360 (23.5)
Multiple	7066 (1.1)	682 (1.1)	626 (1.1)
NHPI	5960 (0.9)	425 (0.7)	336 (0.6)
Unknown	39,109 (5.8)	2623 (4.1)	2116 (3.7)
Ethnicity
Hispanic or Latinx	57,198 (8.5)	3633 (5.7)	3815 (6.7)	< 0.001	< 0.001	< 0.001
Non-Hispanic or Latinx	612,394 (91.5)	60,154 (94.3)	53,200 (93.3)
HCV in study period
No	625,876 (93.5)	53,209 (83.4)	41,002 (71.9)	< 0.001	< 0.001	< 0.001
Yes	43,716 (6.5)	10,578 (16.6)	16,013 (28.1)

HIV, human immunodeficiency virus; AUD, alcohol use disorder; OUD, opioid use disorder; AIAN, American Indian or Alaska Native; NHPI, Native Hawaiian or Pacific Islander; HCV, Hepatitis C Virus

¹Age, sex in medical record, race, and ethnicity represent all categories reported at earliest encounter in the study period (2016–2022); HCV diagnosis reported based on ICD-9 or ICD-10 code for HCV during the study period (codes in Supplemental Table 1)

²Values may not add up to 100% due to rounding

³χ² values are reported in Supplemental Table 2

HIV Prevalence

Overall, HIV prevalence was 1.1%, 2.1%, and 3.9% in the AUD, OUD, and AUD/OUD cohorts, respectively. HIV prevalence in the AUD/OUD cohort was significantly higher than in the other two cohorts, and HIV prevalence in the OUD cohort was significantly higher than in the AUD cohort (Table 2, Fig. 1; Supplemental Table 3). HIV prevalence was higher in the AUD/OUD cohort, followed by the OUD cohort and then the AUD cohort across all demographic and clinical characteristic subgroups. Differences between the three SUD cohorts were statistically significant in most subgroups (exceptions are AIAN, Asian, multiple race, and NHPI racial categories, likely due to small samples [Supplemental Table 3]).

Table 2.

Unadjusted HIV Prevalence in Veteran AUD, OUD, and AUD/OUD Cohorts Overall and by Demographic and Clinical Characteristics

	Unadjusted HIV prevalence1, n %2, by cohort
Characteristic	AUD3 n = 669,595	OUD3 n = 63,787	AUD/OUD3 n = 57,015
Overall	7219 (1.1)	1315 (2.1)	2232 (3.9)
Age category
18–29	373 (0.8)	41 (1.4)	143 (2.7)
30–39	989 (0.9)	117 (1.2)	359 (2.9)
40–49	1069 (1.1)	118 (1.7)	265 (3.6)
50–64	3643 (1.6)	677 (3.0)	1189 (4.9)
65 +	1145 (0.6)	362 (1.7)	276 (3.6)
Sex in medical record
Male	6857 (1.1)	1233 (2.1)	2106 (3.9)
Female	362 (0.9)	82 (1.4)	126 (3.5)
Race
White	2882 (0.7)	768 (1.5)	1145 (2.9)
AIAN	64 (0.9)	8 (1.4)	22 (3.8)
Asian	28 (0.6)	2 (1.0)	7 (3.9)
Black	3817 (2.3)	485 (5.3)	963 (7.2)
Multiple	99 (1.4)	10 (1.5)	29 (4.6)
NHPI	58 (1.0)	5 (1.2)	10 (2.9)
Unknown	271 (0.7)	37 (1.4)	56 (2.6)
Ethnicity
Hispanic or Latinx	665 (1.2)	121 (3.3)	194 (5.1)
Non-Hispanic or Latinx	6554 (1.1)	1194 (2.0)	2038 (3.8)
HCV in study period
No	5853 (0.9)	759 (1.4)	1128 (2.8)
Yes	1366 (3.1)	556 (5.3)	1104 (6.9)

HIV, human immunodeficiency virus; AUD, alcohol use disorder; OUD, opioid use disorder; AIAN, American Indian/Alaska Native; NHPI, Native Hawaiian or Pacific Islander; HCV, Hepatitis C Virus

¹HIV defined based on meeting two out of three criteria for HIV; see “METHODS” and Supplemental Table 1

²Values may not add up to 100% due to rounding

³HIV prevalence was significantly different between the AUD and OUD cohorts, the OUD and AUD/OUD cohorts, and the AUD and AUD/OUD cohorts for the overall sample. Supplemental Table 3 shows P-values (adjusted using robust standard errors followed by Benjamini–Hochberg correction for multiple comparisons) for differences between the three cohorts overall and for each subgroup category. In omnibus tests, unadjusted HIV prevalence was statistically significantly different at the P < 0.05 level across AUD, OUD, and AUD/OUD cohorts overall and in all subgroups

Figure 1.

HIV prevalence and HIV screening history in veteran AUD, OUD, and AUD/OUD cohorts. Figure 1 shows HIV prevalence on the primary (left) vertical axis and the proportion of veterans with HIV screening history on the secondary (right) vertical axis. From left to right, values are shown separately with bars for the AUD, OUD, and AUD/OUD cohorts. HIV prevalence is shown with blue bars. HIV screening history is shown with yellow bars. Error bars represent 95% confidence intervals. Abbreviations: HIV, human immunodeficiency virus; AUD, alcohol use disorder; OUD, opioid use disorder.

In descriptive subgroup analysis of demographic and clinical characteristics (Table 2), HIV prevalence was highest among veterans age 50–64 compared with the other ages. HIV prevalence was higher in males than in females, and in Black veterans compared with veterans of other races. Veterans of Hispanic/Latinx ethnicity had higher HIV prevalence than non-Hispanic/Latinx veterans. Last, veterans with an HCV diagnosis in the study period had higher HIV prevalence than those without an HCV diagnosis. Age-adjusted HIV prevalence values were similar to unadjusted values (Supplemental Table 4).

HIV Screening History

Veterans in the AUD/OUD cohort (87.2%) were significantly more likely to have a VHA history of HIV screening compared with those in the OUD (70.9%) and AUD cohorts (66.9%), and veterans in the OUD cohort were more likely to have HIV screening history than those in the AUD cohort (Table 3).

Table 3.

HIV Screening History in Veteran AUD, OUD, and AUD/OUD Cohorts Overall and by Demographic and Clinical Characteristics

	HIV screening history1, n %2, by cohort
Characteristic	AUD3 n = 669,595	OUD3 n = 63,787	AUD/OUD3 n = 57,015
Overall	447,863 (66.9)	45,245 (70.9)	49,719 (87.2)
Age category
18–29	30,806 (67.2)	2070 (72.6)	4875 (87.4)
30–39	74,287 (69.6)	7266 (74.8)	10,732 (87.5)
40–49	62,725 (69.3)	5089 (73.4)	6427 (86.8)
50–64	166,391 (72.3)	17,294 (75.5)	21,516 (89.0)
65 +	113,653 (57.9)	13,526 (63.3)	6169 (81.2)
Sex in medical record
Male	415,741 (66.3)	40,924 (70.5)	46,459 (87.0)
Female	32,112 (76.2)	4321 (75.5)	3260 (90.3)
Race
White	275,776 (62.3)	34,422 (68.7)	33,998 (85.4)
AIAN	4908 (66.3)	404 (68.6)	506 (87.7)
Asian	3345 (68.8)	158 (76.3)	161 (89.9)
Black	131,726 (81.0)	7810 (85.2)	12,489 (93.5)
Multiple	5056 (71.6)	500 (73.3)	574 (91.7)
NHPI	3829 (64.3)	282 (66.4)	284 (84.5)
Unknown	23,223 (59.4)	1669 (63.6)	1707 (80.7)
Ethnicity
Hispanic or Latinx	42,311 (74.0)	2865 (78.9)	3451 (90.5)
Non-Hisp. or Latinx	405,552 (66.2)	42,380 (70.5)	46,268 (87.0)
HCV in study period
No	408,620 (65.3)	35,643 (67.0)	34,301 (83.7)
Yes	39,243 (90.0)	9602 (90.8)	15,418 (96.3)

HIV, human immunodeficiency virus; AUD, alcohol use disorder; OUD, opioid use disorder; AIAN, American Indian/Alaska Native; NHPI, Native Hawaiian or Pacific Islander; HCV, Hepatitis C Virus

¹HIV screening history defined as record of at least one HIV diagnostic test in a patient’s lifetime VHA medical record, HIV diagnostic tests identified based on laboratory names (see Supplemental Table 1)

²Values may not add up to 100% due to rounding

³HIV screening history was significantly different between the AUD and OUD cohorts, the OUD and AUD/OUD cohorts, and the AUD and AUD/OUD cohorts for the overall sample. Supplemental Table 5 shows P-values (adjusted using robust standard errors followed by Benjamini–Hochberg correction for multiple comparisons) for differences between the three cohorts overall and for each subgroup category. In omnibus tests, unadjusted HIV screening history was statistically significantly different at the P < 0.05 level across AUD, OUD, and AUD/OUD cohorts overall and in all subgroups

In subgroup analyses, groups with the highest HIV prevalence were also more likely to have a history of HIV screening. For example, veterans aged 50–64 years, of Black race, of Hispanic/Latinx ethnicity, and with an HCV diagnosis in the study period were more likely to have HIV screening history than their counterparts in respective categories. An exception to this was the sex subgroup; women had lower HIV prevalence than men, but were more likely to have been screened. The subgroup most likely to have HIV screening history was veterans with an HCV diagnosis; over 90% of veterans in all three SUD cohorts who had an HCV diagnosis had been screened for HIV in their VHA history (Table 3; Supplemental Table 5).

DISCUSSION

We found HIV prevalence ranged from 1.1 to 3.9% in veterans with AUD, OUD, and both. In all SUD cohorts, HIV prevalence was higher than published estimates for the US population (∼0.3%) and VHA veterans (∼0.5%).^{2, 9} While HIV prevalence was high overall, there was a significant rank-order effect wherein the AUD/OUD cohort had the highest HIV prevalence, then the OUD cohort, and finally the AUD cohort. Additionally, we found that nearly a third of veterans with AUD or OUD did not have a HIV screening history in the VHA system data, while most (∼90%) of veterans with both AUD/OUD had HIV screening history in VHA care. While this unscreened prevalence is low compared with older estimates (e.g., a 2009 study reported 80% of veterans with SUD were not screened),³³ proportions are still high given that the VHA, CDC, and USPSTF recommend at least one HIV screening for all adults aged 18–65 and annual screening for high-risk populations.^{29, 30, 32} Our findings highlight room for improvement in HIV prevention and screening among veterans with SUD.

Consistent with prior research documenting disparities, HIV prevalence varied across subgroups. Notably, HIV prevalence was higher in veterans with an HCV diagnosis during the study period compared with those without HCV. While injection drug use is not consistently recorded in VHA medical records, HCV is commonly acquired through sharing injection drug equipment.⁹ These results are consistent with evidence of increased HIV risk for PWID.¹⁷ Similarly, HIV screening history varied between subgroups. Within all cohorts and subgroups (except sex), categories with the highest HIV prevalence also were most likely to have received an HIV screening. This may reflect higher actual or perceived differences in HIV prevalence or high-risk patient demographics. Alternatively, higher testing rates could be driving higher HIV prevalence estimates, for example, higher-risk populations (e.g., AUD/OUD cohort) may have more interaction with health systems and opportunities for testing. Sex category was an exception to this pattern; women had lower HIV prevalence but higher probability of HIV screening than men (possibly due to reproductive-age screening).

Despite variation within subgroups, HIV prevalence and screening trends between AUD, OUD, and AUD/OUD cohorts were consistent for all subgroups, with the AUD/OUD cohort having the highest HIV prevalence and screening, followed by the OUD and then AUD cohorts. Higher HIV prevalence in veterans with AUD/OUD could be partly driven by compounding HIV risk factors among veterans with more than one SUD²¹ (e.g., injection-drug-related and sexual risk behaviors).⁶ People with two SUDs may also be more likely to experience adverse social determinants of health associated with both SUD and HIV risk, such as trauma, racism, or homelessness.^{38, 39} Ongoing efforts have characterized syndemic conditions among marginalized groups such as women experiencing physical/sexual violence,^{40, 41} but more research is needed to understand how polysubstance use and HIV risk intersect among veterans.

Our finding that veterans with AUD, OUD, and AUD/OUD have higher HIV prevalence than the general population has implications for improving veteran health services. Prior work has documented high SUD prevalence among people with HIV,⁴ and we now document elevated HIV prevalence among veterans with SUD, emphasizing the extent of the SUD and HIV syndemic. An important health service factor to examine further is to what extent these co-occurring concerns are being addressed within and across clinics, as opposed to being addressed separately (or at all) across siloed clinics treating one at the expense (or under-acknowledgement) of the other. From the patient perspective, siloed/fragmented care can be inconvenient or costly and can increase disengagement from services.^{42, 43} From the health system perspective, it can result in inefficiencies and decrease service quality.⁴⁴ Thus, preventing care fragmentation through integrative care coordination between substance use, behavioral health, and HIV services could address gaps in HIV prevention and detection within VHA systems.

The 2022–2025 HIV/AIDS strategy emphasizes the need for integration of HIV and substance use services as a strategy to end the US HIV epidemic.⁴⁵ In a systematic review, Haldane et al. identified three models for integrated HIV/SUD systems: (1) SUD services integrated at HIV treatment facilities, (2) HIV services integrated at SUD treatment facilities, and (3) SUD and HIV services integrated simultaneously into alternative care settings (e.g., drop-in or street-based settings).⁴⁴ These models may streamline services and increase numbers of people diagnosed, treated, and retained in care for both HIV and SUD.⁴² For example, SUD services (or other services provided to people with SUD, such as mental health services) could be designed to implement universal HIV screenings, coordinate with infectious disease clinics, and connect patients to preventive healthcare such as pre-exposure prophylaxis.^{46, 47} Such models may be useful for veteran populations with SUD who have high HIV prevalence and/or low HIV screening rates.⁴⁸

Reviews and studies of VHA care have noted the importance of integrated SUD, HIV, and mental health care. In a resource for providers, the Department of Veterans Affairs describes a model of integrated care that could include co-located services, diverse teams of providers, stigma-reducing cultures, availability of comprehensive services, and effective communication strategies.⁴⁹ A RAND corporation report suggests that all treatment facilities serving post-9/11 veterans should screen for co-occurring disorders; however, the report does not describe current implementation of integrated HIV/SUD screening and care.⁴⁶ One study found that HIV testing can successfully be incorporated into some VHA clinics, but other sites faced barriers such as lack of laboratory support and nursing resistance.⁵⁰ Another study found that an integrated alcohol treatment intervention increased treatment uptake among veterans with HIV, but noted that such integrated approaches are rare.⁵¹ In the US generally, barriers to system integration include policy- and system-level factors such as SUD treatment waiver requirements, prior authorization rules, and gaps in workforce training/development.⁵² More research is needed to understand which models of integrated HIV/SUD systems are effective in VHA settings, and how optimal models can be adapted to local contexts and successfully scaled.^{50, 53} This work should include policy and implementation studies investigating facilitators and barriers to successful integrated VHA SUD/HIV services.

Strengths of our study include the large sample size from the VHA CDW, a database that includes medical record history for included patients, which is important for identifying HIV (a lifelong disease). Additionally, this is the first study to our knowledge to document HIV prevalence among veterans with different SUD types, including co-occurring AUD/OUD. Our study also has limitations. The VHA CDW includes medical encounter data which may contain inaccuracies in encounter codes. However, we used a multi-criteria method that was previously validated using VHA data to identify veterans with HIV.³⁶ Also, VHA CDW data does not capture all encounters outside of the VHA system, so while we restricted the analysis to veterans with at least one inpatient or two outpatient visits with an SUD diagnosis within the VHA system, we may have missed veterans who received screening or SUD diagnoses outside of VHA. Last, our study is descriptive and does not establish causal relationships between SUD and HIV.

In conclusion, we find that VHA patients with AUD, OUD, and both AUD and OUD have high HIV prevalence. Veterans with both AUD and OUD have the highest HIV prevalence, followed by veterans with OUD and then veterans with AUD. Additionally, a substantial proportion of veterans with SUD do not have a history of HIV screening in their lifetime VHA data, despite USPTF recommendations that everyone age 18–65 should have at least one HIV screening, and that people at high risk for HIV should have annual screening. Our findings call attention to gaps in HIV prevention and screening services for veterans, and may be useful to veterans’ health service program managers and implementation researchers interested in improving integrated care, including screening, prevention, and treatment options, for the substance use and HIV syndemic.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

This work was funded by the National Institute for Drug Abuse (McCann and Stein: 1T32DA041898-01A1) and the Patient Centered Outcomes Research Institute (Davenport, Mandavia, Stein, and Livingston: COVID2020C2-11081).

Data Availability

The United States Department of Veterans Affairs (VA) places legal restrictions on access to veteran’s health care data, which includes both identifying data and sensitive patient information. The analytic data sets used for this study are not permitted to leave the VA firewall without a Data Use Agreement. This limitation is consistent with other studies based on VA data. However, VA data are made freely available to researchers behind the VA firewall with an approved VA study protocol. For more information, please visit https://www.virec.research.va.gov or contact the VA Information Resource Center (VIReC) at vog.av@CeRIV.

Declarations

Conflict of Interest

The authors have no conflicts of interest to declare.