The effects of psychological functioning on substance use and treatment outcomes among people with HIV

Abstract

The interrelatedness of mental health status and HIV-related outcomes is well-documented. However, the long-term relationship between psychological distress and health outcomes among persons with HIV, co-diagnosed with substance use disorders (SUD), is understudied. We measured psychological distress among men and women with HIV who use drugs, using a low-burden instrument, and tested its effect, longitudinally, on HIV and substance use-related outcomes. Recently hospitalized, adult men and women co-diagnosed with HIV and SUD were surveyed for psychological distress, using the 18-item Brief Symptom Inventory (BSI-18). We tested the short-term (6 months) and long-term (12 months) effect of psychological distress on HIV-related and substance use-related outcomes. Psychological distress predicted higher engagement with SUD treatment and higher substance use, which decreased rapidly, and significantly, over time. No significant relationship was found between psychological distress and HIV viral load suppression. Using brief and easy to administer measures, early detection of psychological distress among persons with HIV and SUD, could avert negative, long-term health consequences—warranting further investigation of interventions that address mental health challenges faced by this population.

Introduction

Background

Despite progress towards the 90-90-90 goals, in the modern antiretroviral therapy (ART) era, gaps along the HIV care cascade persist for people living with HIV (PLWH)¹. In the United States (US), men having sex with men (MSM), racial and ethnic minorities and youth are disproportionally affected by the HIV epidemic². Poor health outcomes persist as a public health challenge among PLWH who use substances³, as they share a disproportionate burden of negative health outcomes, such as death by overdose and suicide⁴. PLWH who use substances are less likely to engage in medical care and less likely to achieve HIV (viral) suppression–consequently have more frequent, yet preventable, hospitalizations and higher rates of viral transmission and greater morbidity and mortality⁵.

The interrelatedness of mental health status, psychiatric conditions and HIV-related outcomes is well-documented in the literature. PLWH are more likely to be challenged with psychological problems and psychiatric co-morbidities, which present a threat to their life and well-being⁶. These conditions may go undiagnosed, particularly in clinical settings where patients may be managed by infectious disease physicians⁶. PLWH have four times higher prevalence of depression and depressive symptoms⁷, which frequently co-occur with substance use⁸. In fact, a recent meta-analysis found that the worldwide prevalence of depression among PLWH was approximately 31% (95% CI), ranging from 22% in Europe to 44% in South America, and occurred more commonly among men⁹. Mental health conditions and psychiatric diagnoses have been associated with poor health behaviors such as decreased likelihood of initiation of—and retention in—HIV care^10,11, poor antiretroviral treatment adherence¹², lower likelihood of virologic suppression¹³, faster HIV disease progression¹⁴, and overall higher mortality rates^6,15.

An instrument designed to assess mental health status in clinical settings, while screening for psychiatric conditions, is the Brief Symptom Inventory (BSI)¹⁶. The BSI screens for psychiatric conditions among PLWH¹⁷ and among persons who use drugs¹⁸. The shortest version of the BSI is less burdensome for patients, as it takes less time to complete¹⁹, and is appropriate in clinical or hospital-based settings, where there is limited access to licensed mental health professionals⁶. Cross-sectionally, studies have measured psychological distress within the context of drug use or mental health conditions and HIV—mostly among HIV population subgroups, like sexual/gender minorities. Cost-effective and early detection of mental health conditions and psychiatric disorders is important to improve clinical outcomes for PLWH who use substances²⁰. Only a small number of studies assessed psychological distress among persons with HIV, who also use substances, and examined its effects on HIV and substance use outcomes. The objective of this study, therefore, was to measure psychological distress among adult men and women with HIV who use drugs, with a brief, low-burden instrument, and test its effect, both short-term (6 months) and long-term (12 months), on HIV and substance use-related outcomes.

HIV-related factors

For persons with HIV living in the US, specific age, racial and ethnic background, gender²¹ and housing status are linked to negative HIV and general health outcomes²²—including lower and suboptimal adherence to HIV treatment^23,24 and viral non-suppression²⁵. In prior studies, social support^26,27 was positively associated with HIV treatment adherence—among racial and ethnic minorities; specifically, lower adherence to HIV treatment has been reported²⁸.

Regarding food/nutrition/energy-related factors, food insecurity is associated with suboptimal medication adherence and viral non-suppression^29–32, postponing medications, higher emergency department use and increased hospitalizations³³. In CTN-0049, the parent study, participants with severe food insecurity at baseline were less likely to be linked to HIV care at 6 months³⁴. Another important consideration is that although the prevalence of wasting has drastically decreased with the advent of antiretroviral treatment, lower BMI, however, still occurs amongst PLWH³⁵.

Substance use-related factors

Demographic characteristics are linked, as well, to health outcomes,^36–38 and substance use-related outcomes for persons who use substances³⁸. Negative substance use-related outcomes have been demonstrated for younger PLWH³⁹. A higher prevalence of SUD has been reported among racial and ethnic minorities²⁸, who fare worst regarding substance use-related outcomes³⁶. It has been also been shown that factors related to biological sex contribute to differences in the use and abuse of substances, where women report greater impairment, compared to men; women, however, are more likely to receive SUD services, have greater reductions in substance use, post-treatment⁴⁰ and respond equally well to treatment, compared to men³⁷.

Regarding housing status, a bi-directional relationship between housing instability and negative substance use outcomes has been reported, where substance use increases the risk of first-time homelessness, and homelessness is linked to increase in subsequent substance and relapse⁴¹. Similarly, wasting is a factor amongst persons with substance use disorders⁴²; and, food insecurity, longitudinally is related to substance use⁴³.

Assessing psychological distress, using a brief measure

The 18-item, 3-factor BSI (BSI-18)⁴⁴ was used in the current study. Historically, the BSI has been selected due to its widely validated psychometric properties⁴⁵. The 53-item, 9-factor BSI⁴⁶, is a self-report instrument used to measure psychological distress among adolescents and adults. The BSI-18, which measures global psychological distress, assesses the three most common psychiatric disorders of depression, anxiety, and somatization (i.e., physical symptoms). The BSI-18, which captures the sum scores of these three subscales, is also known as the Global Severity Index,. The BSI-18 is particularly useful to screen for symptoms in clinical environments where patients present with other co-morbid conditions. The BSI-18 has been used in many different contexts to measure psychological distress, cross-sectionally. Studies have included adult and adolescent sexual gender minorities, including LGBT youth⁴⁷; trans-female youth⁴⁸, adult and older MSM⁴⁹ and young, Black MSM. Study populations have focused also on racial and ethnic minority young MSM⁵⁰, adult and adolescent racial and ethnic minorities with HIV who have experienced trauma and violence^51–53, adult and adolescent substance-using populations^54–59, and general populations of persons living with HIV^60,61. Among persons with HIV and co-occurring substance use, the BSI depression scale, only, was used⁶². Another longitudinal study⁶³ examined change in psychological distress over a 12-month period among youth newly diagnosed with HIV. To our knowledge, this is the first study to examine psychological distress longitudinally, as assessed by the BSI-18, quantifying depression, anxiety, and somatization—among PLWH who use drugs.

Given the interrelatedness between depression that co-occurs with substance use⁸, the importance of early detection of psychological dysfunction for PLWH, the importance of sustained engagement in treatment for HIV and substance use for PLWH who use drugs, and, finally, the importance of sustained viral suppression to preserve health quality, the current study evaluated psychological distress as a predictor of substance use and HIV outcomes, overtime, and the relationship between psychological distress and the outcomes of engagement and retention in substance abuse treatment. The current study included a cohort of patients with confirmed HIV diagnosis, hospitalized for an HIV- or AIDS-defining illness, who were participants of Project Hospital visit as an Opportunity for Prevention and Engagement (HOPE). The hypothesis was that higher psychological distress would predict, longitudinally, higher drug use, lower HIV and SUD engagement, and lower viral suppression.

Methods

Human ethics and consent to participate

Work with human subjects were in accordance with the ethical standards of the Helsinki Declaration (1964, amended most recently in 2008) of the World Medical Association. Patients’ written informed consent was obtained. The design of the work was approved by local ethical committees for the 11 hospitals (see https://pmc.ncbi.nlm.nih.gov/articles/instance/5339876/bin/NIHMS809575-supplement-Protocol.pdf for the full protocol).

Study population and participant recruitment

Project HOPE was a multisite, randomized controlled trial conducted within the Clinical Trials Network (CTN-0049) supported by the National Institute on Drug Abuse (NIDA). The trial tested the efficacy of patient navigation with or without financial incentives for achieving durable viral suppression among PLWH who use substances over 12 months⁶⁴. In total, 801 inpatients were enrolled from 11 hospitals in the US, from July 2012 through September 2014. The 11 enrolling hospital sites had high HIV inpatient censuses (≥ 200/year) and high prevalence of substance use among patients with HIV infection. Hospitals were located in the following 11 cities: Atlanta, Georgia; Baltimore, Maryland; Boston, Massachusetts; Birmingham, Alabama; Chicago, Illinois; Dallas, Texas; Los Angeles, California; Miami, Florida; New York, New York; and Philadelphia and Pittsburgh, Pennsylvania.

Eligible participants were: (1) HIV-positive inpatients; (2) at least 18 years old; (3) able to communicate with project staff in English; (4) willing and able to sign a medical record release; (5) willing and able to provide locator information; (6) lived near study sites and were able to return for follow-up visits; (7) had sufficient functional status as measured by a Karnofsky Performance Scale Index score of ≥ 60⁶⁵; (8) completed a baseline assessment and blood draw; (9) reported or had medical records documenting any opioid, stimulant, or heavy alcohol use within the past 12 months as determined by the Alcohol Use Disorders Identification Test (AUDIT)-10⁶⁶; and, (10) met one of the following requirements: had an AIDS-defining illness; had a CD4 + T-cell count less than 350 cells/mm3 at their most recent screening and a viral load of more than 200 copies/mL within 6 months; or had a CD4 + T-cell count within 12 months that was 500 cells/mm3 or less and their viral load was more than 200 copies/mL (or their viral load was unknown with clinical indicators that the patient was likely to have a detectable viral load). Those who met these eligibility criteria were randomly assigned to 3 study arms: (1) patient navigation only (PN); 2) patient navigation with financial incentives (PN + CM); or (3) treatment as usual (TAU). More detailed information on the protocol and intent-to-treat outcomes for this multi-site randomized controlled trial are provided in the trial’s primary outcomes paper⁶⁴.

Variables and instrumentation

Socio-demographic characteristics and health status

Given the disparities in HIV-related health outcomes by demographic characteristics⁶⁷, age, race, ethnicity, gender, education, relationship status, time since HIV diagnosis, housing, food insecurity, social support and education were collected at baseline for consideration to be entered into the statistical models (see Table 1).

Table 1.

Participant characteristics across the study and by treatment arm.

	Overall (N = 801)		PN (N = 266)		PN + CM (N = 271)		TAU (N = 264)
	N/mean	% / Std among total	N/mean	% / Std among PN	N/mean	% / Std among PN + CM	N/mean	% / Std among TAU
Gender
Male	540	67.4	179	67.3	177	65.3	184	69.7
Female	261	32.6	87	32.7	94	34.7	80	30.3
Age*	44.6	10.0	44.8	9.9	44.7	10.0	44.5	10.1
Race
Hispanic	84	10.5	26	9.9	25	9.3	33	12.5
NH black	579	72.6	195	73.9	194	71.9	190	72.0
NH white	97	12.2	27	10.2	41	15.2	29	11.0
NH other	38	4.8	16	6.1	10	3.7	12	4.6
Education
Less than high school	319	39.8	117	40.0	105	38.8	97	36.7
High school	271	33.8	81	30.5	94	34.8	96	36.4
More than high school	211	26.3	68	25.6	72	26.6	71	26.9
Marital status
Married or living with partner	90	11.2	32	12.0	33	12.2	25	9.5
Widowed, divorced, or separated	181	22.6	62	23.3	58	21.4	61	23.1
Never married	530	66.2	172	64.7	180	66.4	178	67.4
BMI*	23.5	5.4	24.0	5.5	23.5	5.2	23.1	5.6
Prescribed medication for HIV
Yes	404	50.5	130	49.1	141	52.0	133	50.4
No	396	49.5	135	50.9	130	48.0	131	49.6
CD4 + T-cell count < 200
Yes	530	66.2	173	65.0	177	65.3	180	68.2
No	271	33.8	93	35.9	94	34.7	84	31.8
Housing insecurity
Yes	298	37.9	106	40.8	101	37.8	91	35
No	489	62.1	154	59.2	166	62.2	169	65
Food security*	6.2	7.9	6.4	8.1	5.8	7.3	6.5	8.2
Social Support*	14.7	6.5	14.6	6.3	14.7	6.7	14.7	6.8
AUDIT-10*	9.1	9.6	9.0	9.6	8.9	9.5	9.4	9.5
DAST-10*	4.7	2.9	4.6	3.0	4.8	2.9	4.6	2.9
Engagement in SUD treatment
Yes	125	15.6%	44	16.5%	35	12.9%	46	17.5%
No	675	84.4%	222	83.5%	236	87.1%	217	82.5%
Viral suppression
Yes	87	10.9	30	11.3	28	10.3	29	11.0
No	714	89.1	236	88.7	243	89.7	235	89.0
Length of time since HIV diagnosis in months*	11.9	8.6	12.1	8.6	11.3	8.4	12.4	8.9
BSI*	22.5	16.1	23.0	16.4	22.4	15.8	22.2	16.1

* Mean and standard deviation of the non-standardized continuous variables and scales are presented.

Data source, measurement and instruments

Participants’ age, racial and ethnic background, and gender were collected by self-report. BMI, mental health and substance use history and ART regimen were extracted from the medical record. Housing stability, based on where participants reported living in the last six months, was assessed. If there were any days or nights where the participant stayed in a shelter, group room, drug treatment facility, transitional or permanent single-room occupancy in hotel, or other residential facility or institution, housing status was classified as unstable (coded as 0). Participants who reported living in their own home, renting, or staying with family or friends, were classified as having stable housing (coded as 1). The Household Food Security Access Scale was used to measure food insecurity. This is a 9-item measure with total scores ranging from 0 to 27 and higher scores representing greater food insecurity⁶⁸. The Short Social Support Scale and Conflictual Social Interaction scales were included to assess social support. The Short Social Support Scale is a 6 item measure in which participants report the number of available individuals they feel they can turn to in times of need as well as how satisfied they are with the perceived support via a 6-point Likert scale^69,70. The Conflictual Social Interaction scale is an 11 item measure assessed on a 5-point Likert scale which assesses frequency of conflict with others⁶⁹. In the parent study, CTN-0049, participants were asked to respond to 3 questions from this measure.

Study exposure: psychological distress

BSI-18 the Global Severity Index. The BSI-18 provides a quick assessment of psychological distress. The BSI-18, a reduced version of the original 53-question BSI, measures three dimensions of psychological distress—somatization, depression and anxiety. The total score, known as the Global Severity Index (GSI), is calculated by summing the scores on each of the 18 items with a max score of 72. In addition, subscales can be calculated for each dimension by calculating the scores for each item pertaining to the subscale⁴⁴. The BSI-18 was validated in US general populations (GSI: Cronbach’s alpha = 0.98; Somatization = 0.82; Depression = 0.87; Anxiety = 0.84)⁴⁴. For US populations, the BSI-18 has been demonstrated to consistently measure depression (e.g., “feeling hopeless about the future”), anxiety (e.g., “spells of terror or panic”), somatization and psychological distress, specifically, among diverse illicit drug use populations in the US⁵⁴, ethnically-diverse persons with serious mental illness⁷¹. A BSI-18 total score (GSI) of < 50 is considered not elevated, 50–56 would be scored as minor elevation, 57–62 is considered moderately elevated, and 63 and above is markedly elevated⁷².

Study outcomes: alcohol and substance use, substance use treatment engagement, viral suppression

Self-report measures of alcohol use severity, substance use severity and substance use treatment engagementwere administered at baseline, 6 months, and 12 months. The 10-item Alcohol Use Disorders Identification Test (AUDIT-10) and 10-item version of the Drug Abuse Screening Test (DAST-10) indexed the severity of alcohol and substance use, respectively^66,73. The AUDIT-10 yields scores ranging from 0 to 40, with scores of 8 or higher generally indicating hazardous or harmful alcohol use. The AUDIT-10 has demonstrated good internal consistency in diverse populations (Cronbach’s α ~ 0.80–0.94). The DAST-10 scores range from 0 to 10, with higher scores indicating more severe drug use problems. Scores of 3 or more typically suggest moderate to severe drug-related issues, warranting further assessment. The DAST-10 has also shown strong internal consistency (Cronbach’s α ~ 0.86–0.94) across various settings. Engagement in substance use disorder treatment was determined by one item on the Service Utilization Detail questionnaire: “During the past 6 months, did you see any professional for the primary purpose of getting alcohol or drug treatment, including methadone maintenance, or getting help for an alcohol or drug problem?”. Responses indicating not engaged or engaged were coded as binary 0 or 1, respectively. Finally, viral suppression was defined as having an HIV viral load ≤ 200 copies/mL (coded as 1) versus > 200 copies/mL (coded as 0). HIV viral load laboratory values were extracted from the medical record. Consistent with prior analyses from this randomized controlled trial⁶⁴, participants who died during the 12-month follow-up were coded as unsuppressed (> 200 copies/mL).

Statistical analyses

Descriptive statistics of all variables, including means and standard deviations for continuous variables and frequencies and percentages for categorical variables, were calculated per each time point as applicable. Generalized linear mixed models (GLMM) were used to examine psychological distress (BSI-18) as a baseline predictor of the following time-varying outcomes at three visits—baseline, 6 and 12 months: (a) AUDIT-10; (b) DAST-10; (c) binary engagement in substance use disorder treatment (Service Utilization Detail questionnaire); and, (4) binary viral suppression (HIV viral load). Post hoc comparisons were conducted in instances of significant interactions between baseline psychological distress and time, using the least squares means (LSmeans) procedure, to examine changes in outcomes over time. Other baseline sociodemographic and medical variables were included in the model as covariates. Adjusted odds ratios (AORs) or standardized beta coefficients with 95% confidence intervals (CI) are reported. The continuous exposure of BSI-18 was rescaled to have a mean of zero and a standard deviation of one to facilitate interpretation of the AORs. Similarly, the AUDIT-10, DAST-10, social support and food insecurity scales were also standardized. All analyses were conducted with SAS 9.4 (SAS Institute, Cary, NC) statistical software, using a significance level of 0.05 (two-tailed).

Results

Baseline characteristics

The sample of 801 randomized participants (3 study arms: PN, PN + CM, TAU) was predominantly Black/African American (73%), middle-aged (Mean = 44.6; SD = 10.0), never married (66%), and men (67%). At baseline, approximately two-thirds (66%) of participants had CD4 + T-cell count less than 200 cells/mm3 and approximately 11% of participants were virally suppressed (≤ 200 copies/mL). The baseline average BSI-18 score, quantifying depression, anxiety, and somatization, was 22.5 (SD = 16.1) (see Table 1). Overall, 8.5% of participants scored above the cutoff considered elevated.

Primary study outcomes

Psychological distress and alcohol use, drug use

GLMM results showed that higher baseline BSI-18 predicted higher AUDIT-10 at baseline (Beta = 0.18, 95% CI = [0.10, 0.25]) and DAST-10 scores at baseline (Beta = 0.01, 95% CI = [0.19, 0.33]) (See Table 2). The impact of BSI-18 on both alcohol and substance use changed over time; there was a significant interaction of BSI-18 and visit (over time) for alcohol use at both 6 and 12 months, and for substance use at 12 months. Median (Q50), and Interquartile Ranges (Q25, Q75) of baseline standardized BSI-18 scores were chosen to represent three different levels of BSI-18 —low, median and high. As shown in Table 3, at the level of BSI-18 = -0.84 (low), the 12-month, 6 month and baseline standardized AUDIT-10 were − 0.25, − 0.20, and 0.05, respectively. For BSI-18 = − 0.22 (median), the 12-month, 6-month and baseline standardized AUDIT-10 were − 0.22, − 0.17, and 0.16, respectively. For BSI-18 = 0.65 (high), the 12-month, 6-month and baseline standardized AUDIT-10 were − 0.19, − 0.13, and 0.31, respectively. Therefore, higher BSI-18 scores at baseline were associated with a larger drop of AUDIT-10 scores over time. A similar relationship was found with DAST-10 scores, comparing baseline to 6 and 12 months, where DAST-10 scores decreased more rapidly over time, among those with higher BSI-18 scores at baseline (see Tables 3 and 1).

Table 2.

Generalized linear mixed models (GLMM) results: Psychological distress (BSI-18) as a predictor of drinking, substance use, SUD treatment engagement, and viral suppression.

Variable (Ref)		AUDIT-10		DAST-10		SUD Treatment		Viral suppression
		Adj. co-efficient (95% CI)	P Value	Adj. co-efficient (95% CI)	P Value	Adj. OR (95% CI)	P Value	Adj. OR (95% CI)	P Value
BSI		0.18 (0.10, 0.25)	< .001	0.26 (0.19, 0.33)	< .001	1.40 (1.12, 1.75)	0.003	0.97 (0.81, 1.17)	0.75
Age		0.01 (0.00, 0.02)	0.03	0.01 (0.01, 0.02)	< .001	1.05 (1.02, 1.07)	< .001	1.04 (1.02, 1.06)	< .001
Gender (Ref = Male)	Female	− 0.14 (− 0.28, − 0.01)	0.03	0.26 (0.14, 0.39)	< .001	2.47 (1.58, 3.85)	< .001	0.93 (0.64, 1.35)	0.69
Race/ ethnicity (Ref = White)	Black	0.00 (− 0.18, 0.19)	0.97	− 0.31 (− 0.49, − 0.13)	< .001	0.31 (0.17, 0.58)	< .001	0.47 (0.28, 0.78)	0.00
	Hispanic	− 0.16 (− 0.41, 0.09)	0.21	0.04 (− 0.20, 0.28)	0.72	0.67 (0.29, 1.53)	0.34	0.69 (0.35, 1.38)	0.30
	Other	0.00 (–0.32, 0.32)	1.00	− 0.08 (− 0.39, 0.23)	0.61	0.36 (0.12, 1.06)	0.06	0.47 (0.19, 1.16)	0.10
Education (Ref = Less than high school)	High school	0.01 (− 13, 0.15)	0.93	0.10 (− 0.04, 0.23)	0.15	1.25 (0.77, 2.01)	0.37	1.41 (0.94, 2.11)	0.10
	More than high school	− 0.16 (–0.31, − 0.01)	0.04	− 0.08 (− 0.23, 0.06)	0.26	1.06 (0.63, 1.81)	0.82	1.6 (1.10, 2.61)	0.02
Marital status (Ref = Never Married)	Married/ Living together	0.05 (− 0.15, 0.24)	0.63	0.06 (− 0.13, 0.25)	0.54	0.58 (0.30, 1.15)	0.12	0.76 (0.43, 1.32)	0.32
	Widowed/ Divorced/ Separated	− 0.02 (− 0.18, 0.13)	0.75	0.04 (− 0.10, 0.19)	0.56	1.05 (0.64, 1.73)	0.85	0.78 (0.51, 1.20)	0.25
BMI		0.01 (− 0.00, 0.02)	0.07	0.00 (− 0.01, 0.01)	0.48	1. (0.97, 1.05)	0.67	1.06 (1.03, 1.10)	< .001
ART Regimen (Ref = No)	Yes	0.06 (− 0.06, 0.18)	0.32	0.05 (− 0.07, 0.17)	0.42	2.20 (1.43, 3.37)	< .001	1.48 (1.04, 2.10)	0.03
CD4 + T-cell < 200 cells/mm3 (Ref = No)	Yes	0.00 (− 0.12, 0.13)	0.94	− 0.15 (− 0.28, − 0.03)	0.01	0.74 (0.48, 1.14)	0.17	0.62 (0.43, 0.89)	0.01
Housing instability (Ref = No)	Yes	0.15 (0.02, 0.28)	0.03	0.26 (0.14, 0.39)	< .001	1.90 (1.21, 2.97)	0.01	0.74 (0.51, 1.09)	0.12
Food security		0.00 (− 0.07, 0.07)	0.93	0.13 (0.07, 0.20)	< .001	1.26 (1.0, 1.59)	0.05	1.01 (0.83, 1.23)	0.94
Social support		− 0.01 (− 0.08, 0.05)	0.61	0.02 (− 0.04, 0.09)	0.41	0.97 (0.78, 1.21)	0.80	1.03 (0.86, 1.24)	0.72
Length of HIV		− 0.00 (− 0.01, 0.00)	0.29	0.01 (0.00, 0.02)	0.01	1.03 (1.01, 1.06)	0.02	0.98 (0.96, 1.00)	0.03
Treatment Arm (Ref = TAU)	PN	− 0.07 (− 0.22, 0.07)	0.32	− 0.07 (− 0.21, 0.07)	0.32	0.02 (0.72, 1.97)	0.50	1.12 (0.74, 1.70)	0.58
	PN + CM	− 0.08 (− 0.23, 0.06)	0.26	0.01 (− 0.13, 0.14)	0.93	1.31 (0.80, 2.15)	0.28	1.29 (0.86, 1.93)	0.22
Time (Ref = Baseline)	12 mo	− 0.41 (− 0.48, − 0.34)	< .001	− 0.21 (− 0.27, − 0.15)	< .001	1.71 (1.21, 2.45)	0.00	10.70 (7.40, 15.48)	< .001
	6 mo	− 0.36 (− 0.42, − 0.28)	< .001	− 0.20 (− 0.27, − 0.14)	< .001	2.50 (1.77, 3.53)	< .001	12.88 (8.85, 18.72)	< .001

Significant values are in bold.

Table 3.

 Least-Squares Estimates of Alcohol and Substance Use Over Time Stratified by Baseline BSI

Error	Total BSI	Visit	LS mean	Standard	t value	p value
AUDIT-10	Low (Q25: − 0.84)	12 Months	− 0.25	0.07	− 3.49	< .001
		6 Months	− 0.20	0.07	− 2.81	0.01
		Baseline	0.05	0.07	0.67	0.50
	Med (Q50: − 0.22)	12 Months	− 0.22	0.06	− 3.61	< .001
		6 Months	− 0.17	0.06	− 2.74	0.01
		Baseline	0.16	0.06	2.60	0.01
	High (Q75: 0.65)	12 Months	− 0.19	0.07	− 2.90	0.00
		6 Months	− 0.13	0.07	− 1.98	0.05
		Baseline	0.31	0.06	4.89	< .001
DAST-10	Low (Q25: − 0.84)	12 Months	0.05	0.07	0.82	0.42
		6 Months	0.01	0.07	0.10	0.92
		Baseline	0.18	0.06	2.76	0.01
	Med (Q50: − 0.22)	12 Months	0.15	0.06	2.57	0.01
		6 Months	0.14	0.06	2.46	0.01
		Baseline	0.34	0.06	5.92	< .001
	High (Q75: 0.65)	12 Months	0.28	0.06	4.61	< .001
		6 Months	0.33	0.06	5.43	< .001
		Baseline	0.56	0.06	9.31	< .001

Psychological distress and substance use treatment engagement

Greater baseline BSI-18 scores also predicted greater odds of engagement in SUD treatment (AOR = 1.40, 95% CI = [1.12, 1.75]). There was no significant interaction of BSI and visit, i.e. time 6 or 12 months, for SUD treatment (see Table 2; see Fig. 1).

Fig. 1.

Changes in standardized AUDIT-10 and DAST-10 scores over time based on BSI quartile.

Psychological distress and viral suppression

Baseline BSI-18 scores were not associated with viral suppression, nor was there a significant interaction of BSI-18 and visit, i.e. time, for viral suppression (see Table 2).

Other significant findings

In addition to the main findings, there were significant relationships between the outcomes of interest (AUDIT-10, DAST-10, SUD treatment engagement, viral suppression) and baseline demographic, clinical, and social covariates. Older age was associated with higher AUDIT-10, DAST-10, and higher likelihood of engagement in SUD treatment and viral suppression. Housing instability was associated with higher AUDIT-10, DAST-10, and higher likelihood of engagement in SUD treatment. Being female (compared to male) was associated with lower AUDIT-10, higher DAST-10, and higher likelihood of engagement in SUD treatment. High school education and higher (compared to less than high school) was associated with lower AUDIT-10 scores and higher likelihood of viral suppression. Longer HIV history was associated with higher DAST-10, higher likelihood of engagement in SUD treatment, and lower likelihood of viral suppression. Higher food insecurity was associated with lower DAST-10 scores. Black race, compared to White race, was associated with lower DAST-10, and lower likelihood to be engaged in SUD treatment and be virally suppressed. Having a CD4 count below 200 was associated with lower DAST-10. ART regimen was associated with higher likelihood of engagement in SUD treatment and viral suppression. Higher BMI was associated with higher likelihood of viral suppression.

Discussion

Identifying psychological dysfunction, which may reveal underlying psychiatric diagnoses, may avert mortality, reduce morbidity and improve health outcomes for persons with HIV who use drugs. Cost-effective, early detection of psychiatric disorders is an important step to improve clinical outcomes for persons who may be co-diagnosed with psychiatric, and substance use disorders²⁰. People with comorbid HIV and mental health/ substance use disorders experience negative HIV-related outcomes and higher mortality⁷⁴. This study utilized the BSI-18—the shortest version of the psychological distress measure. The BSI-18, which includes three subscales of depression, anxiety and somatization, was chosen because it is less burdensome for patients, as it takes a shorter time to complete¹⁹, and is appropriate in settings where there is limited access to licensed, mental health professionals⁶. The BSI-18 was also chosen because of its good psychometric properties^75–79 and has been demonstrated to measures psychological distress consistently, overall, across diverse populations of drug users⁵⁴.

Our hypothesis was that higher psychological distress would predict, longitudinally, (a) higher drug use; (b) lower HIV engagement; (c) lower SUD treatment engagement; and, (d) lower viral suppression. In fact, contrary to what was hypothesized, we found that participants with higher levels of psychological distress were more likely to be engaged with SUD treatment. However, we confirmed the relationship between psychological distress with both alcohol and substance use severity, finding that higher levels of psychological distress predicted higher alcohol use severity. For this analysis, longitudinally, alcohol use severity decreased faster, both short-term (6 months) and long-term (12 months), among those with higher BSI scores at baseline. We also found that the higher psychological distress was associated with decreased drug use, which decreased more rapidly at 12 months (not 6 months). It is plausible that since there was a higher likelihood of being engaged into SUD treatment among participants with higher baseline BSI-18, being engaged in SUD treatment resulted in decreased use at 6 months and even more rapidly at 12 months.

Our hypothesis regarding viral suppression was not upheld—specifically, psychological distress did not predict, longitudinally, lower viral suppression. The current study did find that demographic characteristics (age, race, education) and clinical factors (higher BMI, ART regimen, CD4 count below 200, length of HIV) were associated with viral suppression, which was consistent with other studies—i.e., older age, Black race, being female²¹ and endorsing housing insecurity were linked to negative HIV outcomes²²—including viral non-suppression²⁵.

The current study also found that psychological distress did predict better SUD-related outcomes (engagement with SUD treatment, faster decrease in alcohol and drug use), independent of study group. As we did not test for moderation in the statistical models, it is plausible that study co-variates may have moderated the relationship between psychological distress and viral suppression, and that the observed improvement in SUD-related factors, which persisted over time, may have been a factor in the relationship between psychological distress and viral suppression. Stahler, Mennis³⁸ reported a moderation effect of demographic characteristics and SUD-related outcomes.

To date, only two published studies examined psychological distress, as measured by the BSI-18, among general, US-based populations of persons with HIV. Kecojevic, Wong’s ⁵⁷ study included comparisons with the individual scales (depression, anxiety, somatization), and did not use the general, psychological distress, scale; Valverde, Purcell’s ⁶² longitudinal study used the depression scale only. Hosek, Harper’s ⁶³ longitudinal intervention, conducted with newly HIV diagnosed youth, examined BSI as an outcome. Hosek, Harper⁶³ reported a significant main effect of time on psychological distress, as measured by the BSI-18, which declined across both groups over time (− 5.07 [1.81], p = 0.01), indicating decreases in psychological distress; however there were no study condition by time effects associated with BSI, indicating that the intervention/study condition did not contribute to a decrease in overall psychological distress. Marino, Gukasyan⁵⁸ also reported longitudinal improvement in treatment that was independent of study condition.

Regarding the social covariates, social support has been linked positively to better HIV-related outcomes^26,27 and to mental health-related outcomes⁶². However, we did not find a significant relationship between baseline social support, as it relates to psychological distress, as a predictor of the current study’s outcomes—alcohol and drug use, SUD treatment engagement, and viral suppression. Valverde, Purcell’s ⁶² study, which examined depression only, found that among men and women, independently, lower social support and lower empowerment was related to psychological distress, measured as depression only. Of note, participants of the current study received some form of assistance (standard of care, or two study conditions), upon discharge from the hospital, which is a type of social support. For PLWH, unstable housing correlates to poor HIV medication adherence^23,24 and HIV virologic suppression²⁵. In our study, examining psychological distress as a predictor of alcohol use, drug use and engagement into SUD treatment, we found housing instability at baseline was related to more alcohol and substance use. Similarly, Zivanovic, Milloy⁴¹ found that housing instability and homelessness correlated to subsequent substance use and substance use relapse, highlighting the need to address socio-structural factors among this population. Interestingly, our model also demonstrated that housing instability predicted engagement into SUD treatment. Given that patient navigation support was provided to the two of the three study arms, it is plausible that individuals with higher psychological distress were supported to engage in treatment regardless of housing status, which resulted in decreasing substance use over time. Specifically, among PLWH, Valverde, Purcell⁶² reported on the association between housing instability (homelessness) and psychological distress (assessed as depression, only). In the current study, housing instability may have been additional motivator for participants to engage into SUD treatment, as some form of assistance was provided (standard of care, or two study conditions), upon discharge from the hospital, post-enrollment.

Prior studies reported an association between food insecurity with substance use⁴³, suboptimal ART adherence, and viral non-suppression³². In CTN0049, the parent study, participants with severe food insecurity at baseline were less likely to be linked to HIV care at 6 months³⁴. In the current study, we found a correlation between baseline food insecurity and higher drug use (DAST-10).

We recognize several aspects of the study that may limit the interpretation of our findings. First, the findings are a result of secondary analyses, which were not the primary purpose of the study. Second, the BSI-18 only provides an indication of psychological stress and not psychopathology. Third, while participants were recruited from multiple sites around the US (in-patient hospitalization), results are applicable only to patients who were enrolled in the parent study. Results, therefore, may not be generalizable to all populations of men and women with HIV, in the United States. Fourth, we were not able to account for other mental health factors such as HIV-related stigma, a factor related (positively) to psychological distress and to poor HIV-related outcomes^50,80; and resilience, a factor inversely related to psychological distress and to positive HIV-related health outcomes⁴⁹. Finally, an important limitation of this study is that data were collected between 2012 and 2014. While the findings offer valuable insights, changes in health policy, service delivery models, and patterns of substance use and mental health care engagement since that time may affect the generalizability of the results to current contexts.

Fifth, although we did adjust for alcohol (AUDIT-10) and substance use severity (DAST-10), our statistical models did not account for types of substances used (i.e. opioids versus marijuana), nor did we account for polysubstance use. This is important as there could be distinct implications based on the type(s) of substance(s) used. It has been reported, for example, that methadone-maintained patients have very specific experiences of pain, which affects psychological distress differently, and that polysubstance affects psychological state⁵⁵. Conversely, the current study did not find a relationship between psychological distress and HIV viral load suppression. Of note, the current study did not test for types of substances used. Starks et al.⁵⁵ did not find that types of substances (illicit, cannabis, poly) affected HIV viral load suppression. We also did not account for types of substance use treatment participants engaged in. A prior study found that opioid misusers are less likely to engage in substance use care, long term⁵⁹; yet, another study found that opioid users benefited to a greater extent from residential treatment³⁸, compared to non-residential types of treatment for substance use.

Despite the study limitations identified, to our knowledge, this is one of the few studies examining psychological distress, amongst a general population of adult men and women in the US, with HIV, who use drugs, and to examine this effect, longitudinally, on both short-term and long-term HIV and substance use-related outcomes. Our findings regarding the relationship between psychological distress and SUD outcomes, and the relationship between specific demographic and social covariates and SUD outcomes, underscore the importance of measuring psychological distress and the personal context of individuals living with HIV, who use drugs. Specifically, understanding a person’s social and demographic context (social determinants) may improve the implementation of effective programs that integrate treatment for mental health disorders, SUD, and HIV, which are essential to ensure greater chance for positive outcomes. Culturally sensitive programs may also improve the outcomes of persons with HIV and comorbid mental illness⁷⁴. Interventions tailored for specific subpopulations of PLWH who use substances have proved successful. In fact, it has been demonstrated among substance using MSMs with HIV, that building skills which improve positive affect reduces stimulant use and optimizes durable HIV viral suppression⁸¹.

Conclusion

Improving health outcomes for persons with HIV is a public health goal, both domestically and internationally. Screening for co-morbid mental health conditions is especially important for PLWH who use drugs, as HIV- and substance use-related outcomes may be hindered by underlying psychiatric diagnoses, such as depression and anxiety, which may go undetected. For the current study, we tested to what extent psychological distress, measured using a brief instrument, predicted, longitudinally, HIV- and substance use-related outcomes. Instruments which are less burdensome for patients to complete and can be administered by trained, healthcare staff, and subsequently evaluated by licensed mental health professionals are desirable in busy and under-resourced healthcare systems and may mitigate gaps in treatment for persons triple co-diagnosed with HIV, substance use disorders and psychiatric diagnoses.

In the current study we measured psychological distress among adult men and women with HIV who use drugs, using a brief, low-burden instrument, and tested its effect, both short-term (6 months) and long-term (12 months), on HIV-related and substance use-related outcomes. In general, we found a significant relationship between higher levels of psychological distress and better SUD-related outcomes—specifically, engagement into SUD treatment, rapid decrease in alcohol use , both short-term (6 months) and long-term (12 months), as well as a rapid decrease substance use (long-term). Contrary to what was hypothesized, we did not find an association between psychological distress and viral suppression. We also found significant relationships of psychological distress by demographic characteristics (age, gender, race, education level, BMI), HIV-related outcomes (ART, CD4) and other social factors (housing insecurity and food insecurity), which predicted one or more of the main outcomes—i.e. alcohol and substance use, SUD treatment engagement, and HIV viral suppression. In summary, it is important to consider the differing demographic, social and clinical characteristics of an individual, when deciding on the interventions for men/women with HIV, who use drugs, to improve the HIV-related and substance use-related outcomes. Overall, our findings speak to the benefit of mental health interventions for people who use substances and a clear need for integration of mental health and substance use disorder treatment approaches in people with HIV who use substances.

Abbreviations

ART

Antiretroviral therapy

PLWH

People living with HIV

MSM

Men having sex with men

BSI

Brief symptom inventory

BSI-18

Brief symptom inventory-18

LGBT

Lesbian, gay, bisexual, transgender

HOPE

Project hospital visit as an opportunity for prevention and engagement

NIDA

National institute on drug abuse

AUDIT

10-Alcohol use disorders identification test

GSI

Global severity index

DAST

10-Drug abuse screening test

GLMM

Generalized linear mixed models

AORs

Adjusted odds ratio

PN

Patient navigator

CM

Contingency management

TAU

Treatment as usual

Author contributions

L.D., R.D.S. and A.C. wrote the main manuscript text. R.D. and D.F. conducted the statistical analyses. R.D.S., S.T.G. and R.D. created manuscript tables and figures. V.H. provided scientific oversight. All authors reviewed the manuscript.

Funding

This work was supported by the National Drug Abuse Treatment Clinical Trials Network (Grant Numbers UG1DA013720, UG1DA015815, UG1DA013035, HHSN271201400028C, HHSN271201500065C, UG1DA013034, UG1DA015831, UG1DA020024, UG1DA013732, and UG1DA013727); the University of Miami (Grant Number P30A1073961); the New York University Center for Drug Use and HIV/HCV Research (Grant Number P30DA011041); and the University of Miami Center for HIV and Research in Mental Health (Grant Number P30MH116867).

Data availability

The data are publicly available via the CTN DataShare and can be found here: https://datashare.nida.nih.gov/study/nida-ctn-0049.

Declarations

Competing interests

The authors declare no competing interests.