Depressive symptoms and use of HIV care and medication-assisted treatment among people living with HIV who inject drugs

Abstract

Objective:

Vietnam, Indonesia, and Ukraine have major burdens of injection drug use (IDU) and HIV. We estimated prevalence of depressive symptoms at baseline among people living with HIV (PLWH) who inject drugs, evaluated associations between depression at baseline and 12-month HIV care outcomes and medication-assisted treatment (MAT), and evaluated the study intervention effect by baseline depression subgroups.

Design:

HPTN 074 was a randomized study. The study intervention included psychosocial counseling, systems navigation, and antiretroviral treatment (ART) at any CD4 count.

Methods:

Moderate-to-severe depression was defined as a Patient Health Questionnaire (PHQ-9) score of 10 or above. ART and MAT were self-reported. Eligibility criteria were: 18–60 years of age, active IDU, and viral load ≥1000 cp/mL. Adjusted probability differences (aPD) were estimated using inverse-probability weighting.

Results:

502 participants enrolled from April 2015-June 2016. Median age was 35 years; 85% identified as men. Prevalence of baseline moderate-to-severe depression was 14% in Vietnam, 14% in Indonesia, and 56% in Ukraine. No evident associations were detected between baseline depression and ART, viral suppression, or MAT at 12-month follow-up. The study intervention improved the proportions of PWID achieving 12-month viral suppression in both the depressed (intervention 44%; standard of care (SOC) 24%; estimated aPD=25% [95%CI: 4.0%, 45%]) and non-depressed subgroups (intervention 38%; SOC 24%; aPD=13% [95%CI: 2.0%, 25%]).

Conclusions:

High levels of depressive symptoms were common among PLWH who inject drugs in Ukraine but were less common in Vietnam and Indonesia. The study intervention was effective among participants with or without baseline depression symptoms.

Introduction

Depression is the most commonly reported neuropsychiatric complication among people living with HIV (PLWH).^[1] The prevalence of depression symptoms in people who inject drugs (PWID) is even higher.^[2–7] Depression may negatively affect HIV treatment initiation^{[8, 9]} and adherence^{[10, 11]} and may increase mortality among PLWH who inject drugs.^[12]

Our report examines the relationships between pre-existing depressive symptoms and prospectively measured antiretroviral treatment (ART), viral suppression, medication-assisted treatment (MAT), mortality, and daily injection drug use (IDU). We also assess depression subgroup effects of the randomly assigned intervention from the parent HPTN 074 trial.^[13]

Methods

HPTN074 was a randomized, controlled, phase 3 feasibility and efficacy study among index PLWH who inject drugs and their HIV-negative injecting partners.^[13] The study was conducted in 3 sites: Kyiv, Ukraine; Thai Nguyen, Vietnam; and Jakarta, Indonesia. Depression symptoms were measured among the HIV-positive index participants who served as the cohort for this assessment. Respondents were surveyed by personnel certified to use non-judgmental interviewing technics. In pre-specified analyses, participants with a baseline PHQ-9 score of 10 or above were classified as having moderate or severe depression (D+), and participants scoring 0–9 were classified as having no or mild depression (D-). In a sensitivity analysis, participants with a PHQ-9 score of 5 or above were classified as having mild-to-severe depression. The prevalence of baseline D+ was estimated by site. Estimated probability differences (PDs) compared participants with baseline D+ to those who were D- (referent) using inverse probability weighted linear-binomial models. Adjusted probability ratios were estimated using an inverse probability weighted modified-Poisson model.^[14] Throughout, 95% confidence intervals (CIs) are constructed with no adjustment for multiplicity. At 12-month follow-up, we compared the proportions of participants who were on ART, virally suppressed, on MAT, and reporting daily IDU (see Text, Supplemental Digital Content 1 where we provide additional Methods information).

Results

Participant characteristics

Between April 2015 and June 2016, 502 eligible PLWH who inject drugs were enrolled.^{[13, 15]} 194 (39%) were enrolled in Vietnam, 187 (37%) in Ukraine, and 121 (24%) in Indonesia. The majority identified as men (427, 85%), and most participants who identified as women were enrolled in Ukraine (84% or 63/75) (see Table, Supplemental Digital Content 2 for participant baseline characteristics by site and depression status).

Depression at baseline

At study entry, the estimated prevalence of moderate/severe depression symptoms was heterogeneous across sites: 14% in Indonesia, 56% in Ukraine, and 14% in Vietnam. Median PHQ-9 depression scores were 4 (Q1, Q3: 3, 7) in Indonesia, 10 (7, 15) in Ukraine, and 5 (2, 8) in Vietnam. (see Table, Supplemental Digital Content 3 for details regarding the prevalence of depression and suicidal thoughts).

ART and viral suppression

At 12-month follow-up, in Indonesia, the percentage of participants on ART was 55% in the D+ group and 34% in the D- group (estimated adjusted probability difference, (aPD) 21% [95%CI: −13%, 56%], Table 1). In Ukraine the percentage on ART was 47% among D+ participants and 39% among D- participants with aPD 8% [95%CI: −8%, 23%]. In Vietnam, the percentage on ART was higher, 67% and 60% in the D+ and D- groups, respectively (aPD 7% [95%CI: −9%, 22%]).

Table 1.

Association between baseline depression and HIV and IDU outcomes at month 12 (n=502) ^a

Month 12 Outcomes	Estimated Probability of Outcome
	D+	D-	Probability Difference (95% CI)	Probability Ratio (95% CI)
Alive and using ART
Indonesia	55.2	34.2	0.21 (−0.13, 0.56)	1.61 (0.82, 3.15)
Ukraine	46.8	39.2	0.08 (−0.08, 0.23)	1.20 (0.83, 1.73)
Vietnam	67.1	60.5	0.07 (−0.09, 0.22)	1.11 (0.78, 1.58)
Overall	53.0	46.0	0.07 (−0.17, 0.30)	1.15 (0.89, 1.49)
Alive with viral suppression (<40 copies per mL)
Indonesia	12.4	12.6	0.00 (−0.20, 0.20)	0.92 (0.22, 3.78)
Ukraine	30.7	35.8	−0.05 (−0.20, 0.10)	0.86 (0.54, 1.35)
Vietnam	22.3	33.6	−0.11 (−0.30, 0.07)	0.66 (0.30, 1.47)
Overall	24.3	29.2	−0.05 (−0.16, 0.06)	0.83 (0.54, 1.27)
Alive and using MAT
Indonesia	15.0	18.2	−0.03 (−0.19, 0.13)	0.82 (0.29, 2.33)
Ukraine	18.4	25.3	−0.07 (−0.21, 0.07)	0.73 (0.39, 1.35)
Vietnam	48.4	35.8	0.13 (−0.13, 0.38)	1.35 (0.78, 2.33)
Overall	25.9	27.9	−0.02 (−0.13, 0.09)	0.93 (0.62, 1.39)
Daily IDU or deceased
Indonesia	49.3	32.7	0.17 (−0.20, 0.53)	1.49 (0.67, 3.28)
Ukraine	61.3	54.5	0.07 (−0.09, 0.22)	1.13 (0.86, 1.48)
Vietnam	16.3	29.3	−0.13 (−0.30, 0.04)	0.55 (0.21, 1.46)
Overall	42.4	40.1	0.02 (−0.11, 0.16)	1.06 (0.76, 1.46)

^aAdjusted estimates accounted for the following baseline covariates: study site, study arm, gender (Ukraine only), educational attainment (Ukraine only), employment status, ART status, MAT status, relationship status, age, CD4+ count, years since HIV diagnosis, and age of IDU initiation. Continuous variables were fit using a restricted cubic spline with 4 equally spaced knots. Multiple imputation was used throughout, and probabilities of each outcome were estimated from adjusted models.

The proportions of participants achieving viral suppression (<40 copies/mL) were similar between D+ (12%) and D- (13%) in Indonesia (aPD 0% [95%CI: −20%, 20%]), and in Ukraine (31% vs. 36%; aPD −5% [95%CI: −20%, 10%], Table 1). Though imprecise, in Vietnam, the proportion of people with viral suppression was lower for D+ (22%) compared to the D- (34%) group (aPD −11% [95%CI: −30%, 7%]).

MAT and daily IDU

At 12-month follow-up, the percentage of participants on MAT in Indonesia was 15% for D+ vs 18% for D- (aPD −3% [95%CI: −19%, 13%]) (Table 1). In Ukraine, 18% of D+ and 25% of D- participants were on MAT (aPD −7% [95%CI: −21%, 7%]). More frequent MAT use was observed in Vietnam with D+ 48% vs D- 36% (aPD 13% [ 95%CI: −13%, 38%]).

Daily IDU was more prevalent among D+ participants in Ukraine and Indonesia at 12 months compared to D- participants, with the largest estimated difference in Indonesia: D+ 49% vs D- 33% (aPD 17% [95%CI: −20%, 53%]) (Table 1). In Vietnam, daily IDU was more frequent in D- (aPD −13%, [95%CI: −30%, 4%]).

Mortality

The estimated 12-month, unadjusted mortality rate was 11% [95%CI: 7%, 18%] among D+ participants and 11% [95%CI: 8%, 14%] among D- participants, overall, with an estimated difference of 0.6% [95%CI: −6%, 7%]. Unadjusted mortality confidence intervals for D+ and D- participants overlapped, overall and within site (see Figure, Supplemental Digital Content 4 where we display crude mortality estimates).

Effect of intervention by baseline depression subgroup

The intervention was effective at increasing self-reported ART use at 12-month follow-up in both the D+ (78% of intervention vs 42% of SOC; estimated aPD 33% [95%CI: 13%, 53%]) and D- subgroups (68% intervention vs 41% SOC; aPD 25% [95%CI: 12%, 37%]) (Table 2). The study intervention improved 12-month viral suppression in both the D+ (44% intervention vs 24% SOC; estimated aPD 25% [95%CI: 4%, 45%]) and D- subgroups (38% intervention vs 24% SOC; aPD 13% [95%CI: 2%, 25%]).

Table 2.

Effect of randomized intervention by baseline depression subgroups at month 12

Month 12 Outcomes	Intervention	Standard of care	Adjusted Probability Difference (95% CI)a	Interaction p-value b
ART				0.50
Depressed	25/32 (78.1)	50/118 (42.4)	0.33 (0.13, 0.53)
Not depressed	64/94 (68.1)	105/258 (40.7)	0.25 (0.12, 0.37)
Viral suppression (<40 copies per mL)				0.37
Depressed	14/32 (43.8)	28/118 (23.7)	0.25 (0.04, 0.45)
Not depressed	36/94 (38.3)	61/258 (23.6)	0.13 (0.02, 0.25)
MAT				0.29
Depressed	10/32 (31.3)	27/118 (22.9)	0.03 (−0.14, 0.21)
Not depressed	37/94 (39.4)	58/258 (22.5)	0.15 (0.03, 0.26)
Daily IDU				0.08
Depressed	16/32 (50.0)	61/118 (51.7)	0.04 (−0.17, 0.26)
Not depressed	23/94 (24.5)	102/258 (39.5)	−0.17 (−0.28, −0.06)

Depressed is defined as moderate/severe depression on PHQ-9 at baseline. Deceased participants are counted as failures for each outcome. Unadjusted event/n (%) results are shown, and multiple imputation was used throughout. Standard of care arm is the referent.

^aAdjusted for the following baseline covariates: study site, depression group, ART status, MAT status, relationship status, age, CD4+ count, years since HIV diagnosis, and age of IDU initiation.

^bStatistical interaction test between study arm and baseline depression group.

MAT use was more frequent in the intervention arm with a more pronounced effect among D- participants (aPD 15% [95%CI: 3%, 26%]) compared to D+ participants (aPD 3%, [95%CI: −14%, 21%]). The effect of the intervention on 12-month daily IDU varied by baseline depression status. Among D- participants, the percentage of people having daily IDU was substantially lower in the intervention arm (24%) compared to SOC arm (40%) (aPD −17% [95%CI: −28%, −6%]). Yet among D+ participants, daily IDU was similar in the intervention and SOC arms (aPD 4% [95%CI: −17%, 26%]).

In a sensitivity analysis, a PHQ-9 score of 5 or above was classified as depression (mild, moderate, or severe) and depressed participants were compared with participants without depressive symptoms. Results were similar to the primary approach (see Table, Supplemental Digital Content 5 where we discuss results of the sensitivity analysis).

Discussion

Prevalence of depressive symptoms measured by PHQ-9 was strikingly higher in Ukraine as compared to Vietnam and Indonesia. But we saw little relationship between baseline depressive symptoms among PLWH who inject drugs on ART use, MAT uptake, viral suppression, daily IDU and mortality. Importantly, the study intervention was effective in improving ART use and viral suppression among participants with or without baseline depression.

In recent studies from Ukraine, the prevalence of self-reported depression among PWID as measured by CES-D and Hospital Anxiety and Depression Scale has been high, up to 60%, which is similar to our findings.^{[16, 17]} Among PLWH in Ukraine, the prevalence of depression was 25%^[18] as compared to 12% in the general population.^[19] The high depression prevalence is exacerbated by the military conflict with Russia in Eastern Ukraine, which has displaced approximately 1.5 million people since 2014. The prevalence of depression in these displaced people is estimated to be 25%.^[20]

Our participants from Indonessia and Vietnam were primarily men, which may have led to a lower estimate of depression prevalance given that gender is a recognized risk factor for depression, with women at increased risk.^{[1, 21–24]} In Vietnam, we detected lower depression prevalence (14%) than was recently described in a study among men living with HIV who inject drugs where 40% reported severe depressive symptoms on the CES-D scale.^[4] Similarly, in Indonesia, PLWH who inject drugs had higher depression prevalence measured by CES-D (33%) than in our sample.^[5]

Discrepancies in depression prevalence may be explained by differences in study-population characteristics, e.g. rural vs. urban, levels of social support, employment status, and similar factors. Alternatively, differences in depression measurement instruments, specifically the use of the PHQ-9 versus the CES-D, may explain the observed differences compared to previous studies in Vietnam and Indonesia.

In our study, self-reported ART was not markedly different between depressed and non-depressed participants. In a recent meta-analysis, depression was negatively associated with ART initiation.^[25] But this relationship is inconsistent. In some studies, PLWH, and especially PWID^[26–28], with depression were less likely to initiate ART^{[9, 26, 29–33]}; but in other studies, PLWH with depression had increases in ART use.^[34–37]

In HPTN 074, baseline depressive symptoms were not substantially associated with 12-month viral suppression, which was observed in several other studies.^[38–42] However, again the results are inconsistent, depression was found to decrease viral suppression in PLWH in some settings.^{[43, 44]} Lower adherence to ART appears to be one mechanism for these effects^{[10, 45–47]}, although adherence does not explain all of the observed differences.^[48]

In our study, overall, MAT use was similar among depressed and non-depressed participants. In previous studies, PWID with depressive symptoms were more likely to enroll in MAT.^[49–51] Depression was also associated with willingness to begin MAT in Ukraine.^[17]

PWID with depressive symptoms may use drugs more frequently than those without depressive symptoms.^[52] However, we did not observe this relationship, which also has not been consistent in other studies. In Vietnam, daily injections were more common among respondents with depression in a previous study,^[4] but among HIV outpatients in Vietnam drug use was not associated with depressive symptoms.^[53] Similarly, in Ukraine, depressive symptoms were not associated with frequency of injection.^{[16, 54]} In contrast, in Indonesia, depressive symptoms have been associated with recent substance use.^[5]

The HPTN 074 study intervention was effective in improving 12-month ART use and viral suppression in both the depressed and non-depressed subgroups. However, the effectiveness of the intervention on MAT use in PWID with depression was not conclusive. In participants with no/mild depressive symptoms at baseline, the intervention substantially reduced daily IDU use compared to our standard of care. However, there was no clear evidence of an intervention impact upon daily IDU among PWID with moderate/severe depressive symptoms at baseline. Importantly, the intervention included modules on depression and injection risk reduction and it is possible that these modules affected self-reported ART and daily IDU in the intervention group.

HPTN 074 was a rigorously conducted study in three diverse sites. Depression in PLWH who inject drugs has been understudied, and this analysis serves to help fill an important research gap. But our analyses of depression were limited by small numbers of people reporting depressive symptoms, especially in Indonesia and Vietnam, which limits precision. In addition, our results were heterogeneous across the three sites and therefore the overall combined results should be interpreted with caution.

In HPTN 074, we did not identify associations between moderate/severe depression at baseline and 12-month HIV or MAT outcomes. In addition, the study intervention effect on HIV outcomes was not substantially modified by baseline depression. Our results indicate that the intervention was effective in terms of viral suppression and self-reported ART in both people with and without depressive symptoms. Unlike previous studies, we did not find a clear link between prior depression and HIV outcomes.

Supplementary Material

Supplemental Data File (.doc, .tif, pdf, etc.)_1

Supplemental Digital Content 1. Text containing additional methods information. pdf

Supplemental Data File (.doc, .tif, pdf, etc.)_2

Supplemental Digital Content 2. Table of detailed participant baseline characteristics by site and depression status. pdf

Supplemental Data File (.doc, .tif, pdf, etc.)_3

Supplemental Digital Content 3. Table showing the prevalence of depression and suicidal thoughts. pdf

Supplemental Data File (.doc, .tif, pdf, etc.)_4

Supplemental Digital Content 4. Figure displaying crude mortality estimates. pdf

Supplemental Data File (.doc, .tif, pdf, etc.)_5

Supplemental Digital Content 5. Table showing the results of the sensitivity analysis comparing mild/moderate/severe depression to no depression symptoms. pdf