The role of depression in secondary HIV transmission among people who inject drugs in Vietnam: A mathematical modeling analysis.

BACKGROUND: Among people who inject drugs (PWID), depression burden is high and may interfere with HIV prevention efforts. Although depression is known to affect injecting behaviors and HIV treatment, its overall impact on HIV transmission has not been quantified. Using mathematical modeling, we sought to estimate secondary HIV transmissions and identify differences by depression among PWID.
METHODS: We analyzed longitudinal data from 455 PWID living with HIV in Vietnam during 2009-2013. Using a Bernoulli process model with individual-level viral load and behavioral data from baseline and 6-month follow-up visits, we estimated secondary HIV transmission events from participants to their potentially susceptible injecting partners. To evaluate differences by depression, we compared modeled transmissions per 1,000 PWID across depressive symptom categories (severe, mild, or no symptoms) in the three months before each visit.
RESULTS: We estimated a median of 41.2 (2.5th, 97.5th percentiles: 33.2-49.2) secondary transmissions from all reported acts of sharing injection equipment with 833 injecting partners in the three months before baseline. Nearly half (41%) of modeled transmissions arose from fewer than 5% of participants in that period. Modeled transmissions per 1,000 PWID in that period were highest for severe depressive symptoms (100.4, 80.6-120.2) vs. mild (87.0, 68.2-109.4) or no symptoms (78.9, 63.4-94.1). Transmission estimates fell to near-zero at the 6-month visit.
CONCLUSIONS: Secondary transmissions were predicted to increase with depression severity, although most arose from a small number of participants. Our findings suggest that effective depression interventions could have the important added benefit of reducing HIV transmission among PWID.

Introduction

Sharing drug preparation and injection equipment is one of the most efficient means of HIV transmission, enabling bodily fluids from a person with HIV to directly enter an injecting partner’s bloodstream [1, 2]. Globally, the estimated prevalence of HIV among people who inject drugs (PWID) is 18% [3], and injection drug use accounts for nearly one-third of incident HIV infections outside of sub-Saharan Africa [4, 5]. Unfortunately, coverage of harm reduction services for most people who inject drugs (PWID) remains inadequate [6-8], and barriers to HIV care engagement and antiretroviral therapy (ART) use can inhibit viral suppression, resulting in onward HIV transmission to injecting partners [6, 7, 9]. Equipment sharing in the absence of viral suppression continues to sustain the HIV epidemic among PWID, particularly in Asia and eastern Europe [10, 11].

Depression may amplify HIV transmission among PWID [12, 13]. Depression prevalence estimates among PWID range from 29% to 75% [14-18], considerably higher than pooled prevalence estimates of 13%-24% in other populations with HIV [19] and global estimates that 4.4% of the world’s population experiences depression [20]. Depression has been linked to both increases in sharing behaviors [12, 21–24] and decreases in ART initiation and HIV viral suppression [13, 25–28]. Despite evidence supporting these links, the role of depression in HIV transmission among PWID has not been quantified. Prior research has focused on individual components of transmission risk (i.e., sharing behaviors or viral load) but has not investigated the overall contribution of depression to HIV transmission through the combination of these behavioral and biological pathways.

In this study, we use mathematical modeling to: 1) estimate secondary HIV transmission events from PWID living with HIV to their potentially susceptible injecting partners, and 2) examine differences in transmissions by depression status. We base input parameter values on behavioral and viral load data from PWID in Vietnam, where injection drug use drives the HIV epidemic [29-31]. Given continued HIV incidence among PWID [9, 10] and the plausible role of depression, it is critical to better understand how depression may drive onward transmission and the extent to which successful depression treatment could avert future infections.

Methods

Study population

We modeled parenteral HIV transmissions arising from 455 adult, male PWID with HIV enrolling in a randomized controlled trial of an HIV prevention intervention in Thai Nguyen Province, Vietnam between 2009 and 2013. Participants were recruited via snowball sampling from the 32 sub-districts of Thai Nguyen with the most PWID. HIV testing was performed, and participants diagnosed with HIV were referred to ART clinics and enrolled in the trial. The intervention included structural- and individual-level components. The structural intervention aimed to reduce community stigma toward HIV infection and injection drug use, and the individual intervention provided support to participants in coping with HIV and reducing condomless sex and sharing of injection drug use equipment. As previously reported [32], no differences in injecting or sexual behaviors by intervention arm were observed over 24 months in the trial.

Measures

Participants responded to a questionnaire in face-to-face interviews at baseline and follow-up visits at 6, 12, 18, and 24 months. At each interview, participants were asked to name up to ten partners with whom they had injected drugs (with or without shared equipment) in the prior three months, and for each partner, to report the frequency in those three months of sharing: 1) the same needle or syringe, 2) drug solution, and 3) the same ampoule of distilled water or novocain (a short-acting local anesthetic used to dissolve heroin). Participants were asked if each partner had HIV, with response options being “Yes”, “No”, or “Don’t know.” The questionnaire also assessed depressive symptoms in the past week with the Center for Epidemiologic Studies Depression Scale (CES-D), a measurement tool that has been widely used in populations with drug use [15, 33–35] and shown to have strong reliability and validity in a similar study population in Vietnam [36]. Although not a clinical diagnostic, the CES-D has good sensitivity and specificity for detecting depression in the general population and among patients with comorbid chronic illness, such as HIV [18, 36–38]. Consistent with this past work, we defined severe symptoms as CES-D scores ≥23, mild symptoms as scores 16–22, and no symptoms as scores <16.

In the trial, blood specimens were collected to confirm HIV infection and measure CD4 cell count. To obtain viral load data for this analysis, we performed HIV RNA testing on stored plasma specimens using COBAS® AmpliPrep/COBAS® TaqMan® HIV-1 Test v2.0 (Roche Diagnostics GmbH) with a lower limit of detection of 20 copies/mL. Due to insufficient volume in >50% of plasma samples from the 12-, 18-, and 24-month visits, we limited the scope of our analysis to data collected at the baseline and 6-month visits.

Descriptive analysis

We performed descriptive analyses on viral load and behavioral data for the overall study population and by baseline depressive symptoms, including a summary of HIV susceptibility and exposure among injecting partners. “Potentially susceptible” partners were defined as those for whom the participant reported HIV-negative or unknown HIV status. Potentially susceptible partners were classified as “minimally exposed” if ≥1 sharing act occurred with a participant who had suppressed viral load (<400 copies/ml), but no acts occurred with a participant who had unsuppressed viral load (≥400 copies/ml). Potentially susceptible partners with ≥1 sharing act with a participant who was virologically unsuppressed were classified as “more exposed,” given the much higher likelihood of transmission under conditions of unsuppressed vs. suppressed viral loads [39].

Model equations

We developed a Bernoulli process model [40-45] to estimate the number of secondary HIV transmission events from each participant in two periods: the three months before baseline and the three months before the 6-month follow-up visit. Transmission estimates were based on viral load measurements taken at the end of a given period, reported numbers of injecting partners in the period, sharing acts within those partnerships over the period, and reported partner HIV status. With this model, we estimated the probability P that participant i would transmit HIV to named partner j in a specified period, given the probability π that the partner was already HIV-infected at the start of that period, the probability β of HIV transmission in one sharing act with a susceptible partner, and the reported number of sharing acts n between participant and partner over the period:

We specified the per-act transmission probability β as a function of viral load v, an average viral load set-point v0, and the transmission probability β0 corresponding to that set-point. In this expression, β increases relative to β0 at a rate α per log10-unit increase in viral load v above v0 [46]:

We then estimated λ, the total modeled number of secondary transmissions for each participant i across all partners j in a given three-month period:

Finally, we summed the λ values across participants in each of the two periods for a total number of estimated transmissions in each window.

Model parameters

The probability (π) of partner HIV infection at the start of a given period, along with the number of sharing acts (n) with that partner over the period, were based on questionnaire reports at baseline and 6 months. We assigned π a value of 0 if the partner was reportedly HIV-negative and 1 if a partner was reportedly HIV-positive. For partners whose HIV status was unknown, we specified π as the estimated HIV prevalence (34%) among PWID in Thai Nguyen during the study period [30, 31, 47]. Participant viral load (v) was fully observed at baseline. For missing viral load due to insufficient samples at 6 months (31% of participants), we used multiple imputation by chained equations [48, 49] to impute viral suppression status and assign viral load in 50 imputed datasets. We used estimates in the scientific literature for the average viral load set-point (v0 = 4.5 log10 copies/ml) [50], transmission probability per sharing act at that set-point (β0 = 0.008) [51], and increases or decreases in infectiousness per log10 change in viral load relative to the set-point (α = 2.09) [46]. Further details about parameters are in S1 Table.

Prior modeling of HIV transmission via shared injecting materials [52-57] has relied on a single per-act probability value that was estimated without specification of (or differentiation by) type of sharing act [51]. Consistent with this work, we conducted one set of model analyses in which we included all types of sharing acts (needles or syringes, drug solutions, and ampoules) as potential transmission events, using the single value of β0 for all types. Due to uncertainty about the extent to which the standard probability estimate reflects all acts vs. needle- or syringe-sharing specifically, we also conducted a set of analyses in which we included only needle- or syringe-sharing acts in the model.

Modeling analysis

To incorporate uncertainty around model inputs into our model outputs, we ran the model 50 times for each of 50 imputed datasets (2,500 runs total). In each run, we drew values for π (for partners with unknown HIV status), α, and β0 from the distributions specified by published summary estimates and confidence limits (S1 Table). Participants with competing events between baseline and 6 months (12% died or incarcerated) or missed 6-month visits not due to a competing event (5%) were excluded from the analysis at 6 months.

To evaluate differences in modeled transmission events by depression, we stratified the population according to baseline depressive symptoms (severe, mild, or no symptoms), and in each time window, we compared the total numbers of transmissions, the number of transmissions per participant, and the probability of infection acquisition per partner with ≥1 sharing act. To produce estimates that could be interpreted beyond our study sample, we also calculated expected numbers of transmissions in each time window per 1,000 PWID living with HIV in a given depressive symptom category (severe, mild, no symptoms). In addition, because depressive symptoms were assessed at both baseline and 6 months, we conducted an analysis of 6-month transmission by 6-month depressive symptoms (instead of baseline symptoms).

To account for possible misreporting of partner HIV status, we performed a sensitivity analysis in which a random sample of 25% of partner HIV status reports in each model run were assumed to be inaccurate (i.e., status was reversed for 25% of partners reported to be HIV-positive or HIV-negative). We also performed a sensitivity analysis around the assumed relationship between viral load and the transmission probability, as the current estimate for α is based on heterosexual transmission and may not strictly hold in the context of injection drug use, where viral load increases above a certain minimum threshold may not affect transmission risk to the extent seen in heterosexual transmission. For this analysis, instead of deriving the probability β from viral load v, we used the standard probability estimate β0 [51] for all viral load values. All analyses were conducted using R Version 3.6.3 [58].

Ethics

The trial and this analysis were approved by the ethical review committees at the Thai Nguyen Center for Preventive Medicine, the Johns Hopkins Bloomberg School of Public Health, and the University of North Carolina at Chapel Hill Gillings School of Global Public Health. The trial was registered at ClinicalTrials.gov (NCT01689545). Written informed consent was obtained from participants.

Results

Among the 455 male participants with HIV at baseline, the median age was 35 years, almost half (47%) were married or cohabitating, and about two-thirds (69%) were employed full-time. Three-quarters (74%) became newly HIV diagnosed at enrollment, and 41% had a CD4 cell count <200 cells/μL. Most (73%) reported sharing injection drug use equipment with partners over the prior three months. Nearly half (44%) had severe depressive symptoms, 25% had mild symptoms, and 30% had no symptoms.

Descriptive analyses

At baseline, participants had a median viral load of 4.3 log10 copies/mL and reported a total of 833 injecting partners (mean 1.8 per participant) in the prior three months (Table 1). Participants reported ≥1 act of sharing injection equipment with 578 of 833 partners. Among those partners with ≥1 sharing act, 45% were reported to have unknown HIV status, 28% were reported to be HIV-negative, and 27% were reported to already have HIV. On average, participants reported 22.3 sharing acts per partner (primarily sharing ampoules/solutions, rather than needles/syringes). At the 6-month visit, median viral load decreased to 3.6 log10 copies/mL, and the average number of injecting partners over months 3–6 dropped to 0.7 per participant. Among the 106 partners with ≥1 sharing act (out of 251 partners total), a higher percentage was reported to be HIV-positive (61%) and fewer sharing behaviors took place (mean 10.2 acts per partner). At baseline, 333 of 833 partners (40%) were classified as “potentially susceptible” and “more exposed” to HIV, decreasing to only 20 of 251 partners (8%) at 6 months (Fig 1).

Table 1

Transmission-related characteristics of study population.

	Overall	Baseline depressive symptoms
		Severe	Mild	None
	N = 455	n = 202	n = 115	n = 138
Baseline characteristics
Median (IQR) participant viral load	4.3 (2.6, 4.9)	4.5 (3.1, 5.0)	4.1 (1.9, 4.7)	4.4 (2.2, 4.9)
Number of injecting partners in last 3 months	833	400	211	222
Number with ≥1 sharing acts (% all partners)	578 (69%)	307 (77%)	143 (68%)	128 (58%)
HIV status (% partners with sharing)
Reported to be HIV-positive	27%	32%	25%	19%
Reported to be HIV-negative	28%	24%	24%	38%
Reported to be unknown	45%	44%	50%	43%
Mean (range) acts per partner with sharing
Sharing ampoules	7.9 (0, 90)	7.0 (0, 90)	8.5 (0, 90)	9.7 (0, 90)
Sharing drug solutions	13.1 (0, 90)	11.8 (0, 90)	13.7 (0, 90)	15.4 (0, 90)
Sharing needles or syringes	1.3 (0, 30)	1.4 (0, 30)	1.8 (0, 30)	0.6 (0, 30)
All acts	22.3 (2, 188)	20.2 (2, 182)	23.9 (2, 180)	25.6 (2, 188)
6-month characteristics
Number (%) for follow-up
Attended visit	377 (83%)	158 (78%)	97 (84%)	122 (88%)
Missed visit: competing event	55 (12%)	39 (19%)	8 (7%)	8 (6%)
Missed visit: no competing event	23 (5%)	5 (2%)	10 (9%)	8 (6%)
Median (IQR) participant viral load*	3.6 (1.8, 4.5)	3.8 (2.0, 4.6)	3.4 (1.8, 4.3)	3.7 (1.6, 4.5)
Number of injecting partners in last 3 months	251	117	69	65
Number with ≥1 sharing acts (% all partners)	106 (42%)	53 (45%)	33 (48%)	20 (31%)
HIV status (% partners with sharing)
Reported to be HIV-positive	61%	62%	67%	50%
Reported to be HIV-negative	17%	21%	15%	10%
Reported to be unknown	22%	17%	18%	40%
Mean (range) acts per partner with sharing
Sharing ampoules	3.4 (0, 30)	3.0 (0, 30)	3.2 (0, 30)	5.0 (0, 30)
Sharing drug solutions	5.8 (0, 30)	4.6 (0, 30)	10.0 (0, 30)	2.2 (0, 30)
Sharing needles or syringes	0.9 (0, 12)	1.2 (0, 8)	0.4 (0, 8)	1.1 (0, 12)
All acts	10.2 (2, 72)	8.8 (2, 45)	13.6 (2, 60)	8.3 (2, 72)

* Among 377 participants who attended the 6-month visit, 139 had missing data on viral load due to insufficient sample volume.

Fig 1

Distribution of partner susceptibility and exposure to HIV, overall and by participant’s baseline depressive symptoms.

Percentages are based on the 833 partners reported by 455 participants at the baseline visit and the 251 partners reported by 377 participants at the 6-month visit. “Potentially susceptible” refers to partners with reported HIV-negative or unknown HIV status. Among potentially susceptible partners, “minimally exposed” had ≥1 sharing act with a participant who had suppressed viral load (<400 copies/ml). “More exposed” partners participated in ≥1 sharing act with a participant who had unsuppressed viral load (≥400 copies/ml).

At baseline and 6 months, there were few clear differences in viral load or sharing behaviors according to baseline depressive symptoms, except for numbers of injecting partners and partner HIV status (Table 1). Participants with severe symptoms reported approximately twice as many partners as participants with mild or no symptoms at both time points. At baseline, partners of participants with severe symptoms were more likely to have shared injection equipment (77%) compared to partners of participants with mild symptoms (68%) or no symptoms (58%); however, these partners were also more likely to be reported as already HIV-positive (32% for participants with severe symptoms vs. 25% and 19% for mild and no symptoms, respectively). Overall, 43% of partners of participants with severe symptoms were classified as potentially susceptible and more exposed to HIV (vs. 37% for mild or no symptoms) at baseline, compared to 6%, 12%, and 8% of partners reported at 6 months by participants with severe, mild, or symptoms, respectively (Fig 1).

Modeling analyses

Across the 2,500 model runs, we estimated a median of 41.2 (2.5th-97.5th percentiles: 33.2–49.2) secondary transmissions from all sharing acts with 833 injecting partners in the three months prior to baseline (Table 2). This estimate fell to 3.0 (2.2–3.8) transmissions in analyses where only needle-/syringe-sharing acts were modeled. For 251 partners reported at 6 months, we estimated a median of 1.2 (0.8–1.7) transmissions in the prior three months based on all acts and 0.1 (0.0–0.2) based on needle-/syringe-sharing only.

Table 2

Total modeled secondary transmission events in study population (N = 455) and expected number of transmission events per 1,000 PWID living with HIV, overall and in each depressive symptom category.

	Overall	Baseline depressive symptoms
		Severe	Mild	None
	N = 455	n = 202	n = 115	n = 138
3 months prior to baseline
All acts, total in study population	41.2 (33.2–49.2)	20.3 (16.3–24.3)	10.0 (7.8–12.6)	10.9 (8.8–13.0)
Needle-sharing only, total in population	3.0 (2.2–3.8)	2.2 (1.5–3.0)	0.5 (0.4–0.7)	0.3 (0.2–0.4)
All acts, number per 1,000 PWID	90.6 (72.9–108.1)	100.4 (80.6–120.2)	87.0 (68.2–109.4)	78.9 (63.4–94.1)
Needle-sharing only, per 1,000 PWID	6.6 (4.9–8.4)	10.7 (7.3–14.6)	4.4 (3.1–6.5)	2.2 (1.6–2.7)
Follow-up months 3–6
All acts, total in study population	1.2 (0.8–1.7)	0.8 (0.5–1.2)	0.3 (0.2–0.5)	0.1 (0.1–0.2)
Needle-sharing only, total in population	0.1 (0.0–0.2)	0.1 (0.1–0.2)	0.0 (0.0–0.0)	0.0 (0.0–0.0)
All acts, number per 1,000 PWID	2.6 (1.7–3.8)	4.9 (2.9–7.4)	3.3 (2.1–5.5)	0.6 (0.4–1.5)
Needle-sharing only, per 1,000 PWID	0.3 (0.1–0.4)	0.8 (0.3, 1.2)	0.0 (0.0–0.0)	0.0 (0.0–0.0)

Estimates are reported as the median (2.5th-97.5th percentiles) from 2,500 model runs and are estimated separately by type of sharing act (all sharing acts vs. only needle-/syringe-sharing).

The highest numbers of transmissions were predicted for participants with severe depressive symptoms. Of an estimated 41.2 transmissions from all acts in the three months prior to baseline, 20.3 (16.3–24.3) were predicted to arise from participants with severe symptoms, 10.0 (7.8–12.6) from participants with mild symptoms, and 10.9 (8.8–13.0) from participants with no symptoms (Table 2). Transmissions per 1,000 PWID living with HIV in the baseline period were highest for severe symptoms (100.4, 80.6–120.2), compared to mild symptoms (87.0, 68.2–109.4) and no symptoms (78.9, 63.4–94.1) (Table 2, Fig 2). When modeled transmissions were restricted to needle-/syringe-sharing, 2.2 (1.5, 3.0) of 3.0 transmissions at baseline were predicted to arise from participants with severe symptoms (vs. <1 for participants with mild or no symptoms). Similarly, transmissions per 1,000 PWID remained highest for severe symptoms (10.7, 7.3–14.6), compared to mild symptoms (4.4, 3.1–6.5) and no symptoms (2.2, 1.6–2.7).

Fig 2

Modeled secondary transmission events in the three months prior to baseline per 1,000 PWID living with HIV in each depressive symptom category.

Estimates from 2,500 model runs are plotted by type of sharing act (all sharing acts vs. only needle-/syringe-sharing). Colored regions show all 2,500 estimates, the black point is the median, and the vertical line extends from the 2.5th to 97.5th percentiles.

In follow-up months 3–6, the 1.2 transmissions modeled for all acts primarily arose from participants with severe symptoms (0.8, 0.5–1.2) (Table 2). Transmissions per 1,000 PWID in months 3–6 were lower than baseline estimates but showed a similar relationship with depressive symptoms: 4.9 (2.9–7.4) for severe symptoms, 3.3 (2.1–5.5) for mild symptoms, and 0.6 (0.4–1.5) for no symptoms (Table 2, Fig 3). Estimates further decreased when only needle-/syringe-sharing acts were modeled; non-zero transmissions were predicted only for participants with severe symptoms. Results were similar when the 6-month depressive symptom measure was used (S1 Fig).

Fig 3

Modeled secondary transmission events in follow-up months 3–6 per 1,000 PWID living with HIV in each depressive symptom category.

Estimates from 2,500 model runs are plotted by type of sharing act (all sharing acts vs. only needle-/syringe-sharing). Colored regions show all 2,500 estimates, the black point is the median, and the vertical line extends from the 2.5th to 97.5th percentiles.

When we accounted for possible misreporting of partner HIV status, there was greater uncertainty in transmission estimates but differences by depression were similar (S2A and S2B Fig). When we assumed a constant transmission probability across viral load (S3A and S3B Fig), more transmissions at baseline were predicted for participants with mild vs. severe symptoms. This divergence from the main results likely relates to slightly higher viral loads among participants with severe (vs. mild) symptoms, resulting in higher transmission probabilities for severe (vs. mild) symptoms in the main analysis but not this sensitivity analysis.

Nearly half of modeled transmissions in the three months prior to baseline (17.1 of 41.2, 41%) arose from just 20 of 455 participants; in months 3–6, only 25 participants accounted for all modeled transmissions. Given that most participants had zero predicted transmissions (due to no sharing or ≥1 acts with only HIV-positive partners), the mean probability of infection per partner and mean transmissions per participant were heavily weighted toward zero regardless of depressive symptoms (S4 and S5 Figs). When modeling transmission using all acts at baseline, the mean probability of infection per partner (given ≥1 act) was 0.07–0.09 across depressive symptom categories, falling closer to zero when only needle-/syringe-sharing acts were modeled and across sharing acts at 6 months. However, there were outliers with higher mean probabilities, particularly for partners of participants with severe symptoms, and to some extent, partners of participants with mild symptoms. The mean transmissions per participant ranged from 0–3 across time points, types of sharing acts, and depressive symptoms. Among the 20 participants with ≥0.5 transmissions at baseline (resulting in nearly half of transmissions), 55% had severe symptoms (vs. 44% overall prevalence of severe symptoms at baseline). Among the 25 participants accounting for all modeled transmissions at 6 months, 52% had severe symptoms (vs. 42% overall prevalence of severe symptoms at 6 months).

Discussion

In this study of PWID living with HIV in Vietnam, we used mathematical modeling to estimate secondary HIV transmission events from PWID to their potentially susceptible injecting partners in two short windows. A high number of secondary transmissions was predicted for the three months prior to baseline when transmission was modeled for all sharing behaviors (i.e., sharing needles or syringes, drug solutions, and ampoules). This estimate dramatically decreased when only needle-/syringe-sharing behaviors were included. These disparate findings demonstrate the sensitivity of model estimates to the assumed infectivity of different sharing behaviors, a phenomenon that has not been routinely considered in prior modeling studies of injection-related HIV transmission. Due to substantial declines in viremia and sharing behaviors after baseline (i.e., after HIV diagnosis and referral to ART clinics), we estimated that very little secondary transmission occurred during follow-up months 3–6, underscoring the importance of diagnosis and treatment for HIV prevention in PWID populations.

Prior research has linked depression to increases in sharing behaviors [12, 21–24] and decreases in ART initiation and viral suppression [13, 25–28], and in the current analysis, we predicted more secondary transmissions for PWID with depression. Across our main analyses, estimated transmissions in each time period were highest for participants with severe symptoms. However, there was substantial variability in the magnitude of differences across model runs, time points assessed, and specific behaviors modeled. The majority of modeled transmissions arose from a small number of participants, resulting in group averages of the probability of infection per partner and transmissions per participant being near-zero regardless of depressive symptoms. Given the multi-faceted nature of transmission, it is plausible that depression has a combination of synergistic and antagonistic effects, with the contribution of each varying across individuals.

Future mathematical models could investigate the extent to which successful depression treatment might reduce continued HIV incidence, assuming that decreasing the severity of symptoms to mild or none through treatment would translate to the lower transmissions predicted for those groups in our study. This work could incorporate our study’s estimates of viral load and sharing behaviors by depressive symptoms and the corresponding differences in transmissions to injecting partners. Depression treatment at varying coverage and efficacy levels could be modeled alongside other prevention interventions (ART, needle exchanges, methadone maintenance) to understand the optimal combination of these approaches in decreasing HIV incidence. Future models of injection-related HIV transmission should also examine uncertainties in the infectivity of different sharing behaviors and the relationship between viral load and transmission probability in the context of injection drug use. Assumptions along these dimensions substantially influenced our transmission estimates, underscoring the importance of understanding these understudied phenomena.

Our mathematical modeling approach enabled the integration of behavioral and biological mechanisms to estimate HIV transmissions according to depression status, but we did not conduct a formal causal analysis with consideration of potential confounders or effect measure modifiers. We also expect that our model predictions underestimate all HIV transmissions arising from this population during the study period. Social desirability bias may have led to underestimates of sensitive sharing behaviors, and missing data precluded estimates for participants who did not attend the 6-month visit. Further, we focused on injection-related HIV transmission to injecting partners and excluded sexual transmission, as condomless sex was rarely reported with injecting partners in our study population. We also note that our model assumes that viral load measures taken at baseline and 6 months applied to the prior three months, which might have been especially likely to underestimate transmission in the second interval, when viral loads may have been decreasing over time due to diagnosis and treatment. Additional underestimation of transmission could have resulted from our assumption that partners reported to be HIV-positive at the end of an interval were HIV-positive at the beginning of the interval. Misreporting of partner HIV status and sharing acts may have been differential by depression status, as those with low levels of reported depressive symptoms may be less likely to report socially undesirable behaviors [59]. As a further limitation, we note that relationships between depressive symptoms and transmission may differ in the current epidemic era and in populations outside our study setting.

Despite these limitations, our modeling analyses suggest a skewed distribution of transmission by depression status, with secondary transmissions predicted to increase with symptom severity and a disproportionate burden of severe depressive symptoms among those participants from whom transmissions were predicted to arise. Our findings reflect a complex relationship of depression with HIV transmission and suggest that depression interventions have the potential to decrease secondary HIV transmissions, in addition to providing crucial mental health benefits among PWID.

Bernoulli model input parameters.

(DOCX)

Click here for additional data file.

Modeled secondary transmission events in follow-up months 3–6 per 1,000 PWID living with HIV, based on depressive symptoms assessed at 6 months instead of baseline.

(DOCX)

Click here for additional data file.

A. Sensitivity analysis accounting for possible misreporting of partner HIV status (baseline transmission by baseline depression). B. Sensitivity analysis accounting for possible misreporting of partner HIV status (transmission in months 3–6 by baseline depression).

(ZIP)

Click here for additional data file.

A. Sensitivity analysis allowing for constant transmission probability (baseline transmission by baseline depression). B. Sensitivity analysis allowing for constant transmission probability (transmission in months 3–6 by baseline depression).

(ZIP)

Click here for additional data file.

Mean probability of infection acquisition in prior 3 months per injecting partner with ≥1 sharing act reported at baseline and 6 months, by participant depressive symptoms.

Each colored point shows one partner’s mean probability across 2,500 model runs; dashed lines show the mean of means (across all partners and model runs). Mean probabilities are reported separately by the types of sharing acts included in the model (all sharing acts vs. only needle-/syringe-sharing). A) Baseline probabilities by baseline depression. B) 6-month probabilities by baseline depression. C) 6-month probabilities by 6-month depression.

(DOCX)

Click here for additional data file.

Mean numbers of secondary transmission events in prior 3 months per participant, by timepoint and depressive symptoms.

Each colored point shows one participant’s mean transmissions across 2,500 model runs; dashed lines show the mean of means (across all participants and model runs). Mean transmissions are reported separately by the types of sharing acts included in the model (all sharing acts vs. only needle-/syringe-sharing). A) Baseline transmissions by baseline depression. B) 6-month transmission by baseline depression. C) 6-month transmission by 6-month depression.

(DOCX)

Click here for additional data file.

(CSV)

Click here for additional data file.

19 Jul 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Hamid Sharifi

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf  and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“This study and the parent trial were funded by NIDA (R36 DA045569, R01 DA022962). Doctoral training support for S. N. Levintow was also provided by NIAID (T32 AI070114) and ViiV Healthcare (pre-doctoral fellowship). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

We note that you have provided additional information within the Acknowledgements Section that is not currently declared in your Funding Statement. Please note that funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“This study and the parent trial were funded by the National Institute on Drug Abuse (R36 DA045569 awarded to S.N.L.; R01 DA022962 awarded to V.F.G.; https://nida.nih.gov/). Doctoral training support for S.N.L. was provided by the National Institute of Allergy and Infectious Diseases (T32 AI070114; https://www.niaid.nih.gov/) and ViiV Healthcare (pre-doctoral fellowship; https://viivhealthcare.com/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Additional Editor Comments:

Major Revision

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Date: June 18, 2022

Study title

The role of depression in HIV transmission among people who inject drugs in Vietnam: A mathematical modeling analysis

General comments

This is one of the relevant literatures that presented the mathematical modelling of Bernoulli process to describe and explain the depression and secondary HIV transmission. The manuscript is well structured, narrated and put in logical order, the flow of which is easy to go through to the end. The introduction section has clearly put the gray area or the problem statement. The result section is detailed, with relevant finding addressing the objective of the study. The authors have summarized the finding, showed the relevance of the finding to HIV prevention and control policy suggestions. The persisted gray area and the study limitations were also well explained and justified for the future investigation. The

Specific comments

The authors might consider the following items to further improve the manuscript

1. It seems that the mainstay of HIV transmission in Vietnam is among people who inject drugs. However, the main ways of HIV transmission could be presented in introduction/method section objectively, so that the reader could easily understand which section of population are at risk for HIV, and then the significant of the modelling could be justified.

2. The key modelling finding, and discussion was about depression and secondary HIV transmission which might be reflected in the title and objective of the study.

3. There seem old references (older than 2005). The authors are advised to look for recent references and replace with the recent relevant evidence.

Reviewer #2: This is an interesting article with very important implications. It is important to evaluate the psychosocial factors influencing behavior and this was very clear as to the conclusions and implications. I am not sophisticated to judge the modeling but will assume it was done properly.

Reviewer #3: This paper is on an important subject that correlates PWID, depression and HIV infection risk. The association of PWID and the risk for secondary transmission of HIV seem adequately addressed by modelling. However, the role of depression as a variable was not thoroughly investigated to answer the following questions:

1. What is the prevalence of depression in PWID compared with the other drug users and general population?

2. Were robust measurement parameters used to ascertain the depression status of the recruited participants?

3. Were confounders/effect modifiers considered in associating depression among PWID with the risk of secondary HIV transmission

Following these responses, bring out a clear take home message from the study, that address how the limitations of the model affect the conclusions

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Richard Elion

Reviewer #3: Yes: dumiwaboka@gmail.com

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

23 Aug 2022

Response to Reviewers

Reviewer 1

This is one of the relevant literatures that presented the mathematical modelling of Bernoulli process to describe and explain the depression and secondary HIV transmission. The manuscript is well structured, narrated and put in logical order, the flow of which is easy to go through to the end. The introduction section has clearly put the gray area or the problem statement. The result section is detailed, with relevant finding addressing the objective of the study. The authors have summarized the finding, showed the relevance of the finding to HIV prevention and control policy suggestions. The persisted gray area and the study limitations were also well explained and justified for the future investigation. The specific comments the authors might consider the following items to further improve the manuscript.

1. It seems that the mainstay of HIV transmission in Vietnam is among people who inject drugs. However, the main ways of HIV transmission could be presented in introduction/method section objectively, so that the reader could easily understand which section of population are at risk for HIV, and then the significant of the modelling could be justified.

Response: We appreciate this suggestion from the reviewer and have revised the first paragraph of the Introduction to provide greater context for the modes of HIV transmission. This text now states: “Sharing drug preparation and injection equipment is one of the most efficient means of HIV transmission, enabling bodily fluids from a person with HIV to directly enter an injecting partner’s bloodstream (1,2). Globally, the estimated prevalence of HIV among people who inject drugs (PWID) is 18% (3), and injection drug use accounts for nearly one-third of incident HIV infections outside of sub-Saharan Africa (4,5).”

2. The key modelling finding, and discussion was about depression and secondary HIV transmission which might be reflected in the title and objective of the study.

Response: We updated the title to specify “secondary” HIV transmission, and we revised the objective statement in the abstract: “Using mathematical modeling, we sought to estimate secondary HIV transmissions and identify differences by depression among PWID.” We also clarified this objective in the first sentence of the last paragraph of the Introduction: “In this study, we use mathematical modeling to: 1) estimate secondary HIV transmission events from PWID living with HIV to their potentially susceptible injecting partners, and 2) examine differences in transmissions by depression status.”

3. There seem old references (older than 2005). The authors are advised to look for recent references and replace with the recent relevant evidence.

Response: We closely reviewed the references but found that the majority represent recent evidence on the HIV epidemic among PWID from the last 5-10 years. Only 4 of 60 references are older than 2005, and all 4 correspond to foundational work for the development of the methods used in this study. Radloff 1977 evaluated the psychometric properties of the CES-D for depression screening, and we include citations to more recent studies validating and applying the CES-D in Vietnamese populations (Thai et al. 2016, Huynh et al. 2017). Pinkerton & Abramson 1993 and Rottingen et al. 2002 developed the Bernoulli modeling methodology, and we also cite more recent studies using these models (Pearson et al. 2007, Fox et al. 2011, Adams et al. 2013, Kroon et al. 2017). Rubin 1987 developed the multiple imputation methodology, and we include a citation to more recent work programming these methods into statistical software (van Buuren & Groothuis-Oudshoorn 2011).

Reviewer 2

This is an interesting article with very important implications. It is important to evaluate the psychosocial factors influencing behavior and this was very clear as to the conclusions and implications. I am not sophisticated to judge the modeling but will assume it was done properly.

Response: We thank the reviewer for positive feedback on our manuscript.

Reviewer 3

This paper is on an important subject that correlates PWID, depression and HIV infection risk. The association of PWID and the risk for secondary transmission of HIV seem adequately addressed by modelling. However, the role of depression as a variable was not thoroughly investigated to answer the following questions:

1. What is the prevalence of depression in PWID compared with the other drug users and general population?

Response: We added greater background to the second paragraph of the Introduction on the prevalence of depression among PWID and in other populations. This text reads: “Depression prevalence estimates among PWID range from 29% to 75% (14–18), considerably higher than pooled prevalence estimates of 13%-24% in other populations with HIV (19) and global estimates that 4.4% of the world’s population experiences depression (20).”

2. Were robust measurement parameters used to ascertain the depression status of the recruited participants?

Response: We expanded on the description of the psychometric properties of Center for Epidemiologic Studies Depression Scale (CES-D) in the Measures subsection of the Methods: “The questionnaire also assessed depressive symptoms in the past week with the Center for Epidemiologic Studies Depression Scale (CES-D), a measurement tool that has been widely used in populations with drug use (15,33–35) and shown to have strong reliability and validity in a similar study population in Vietnam (36). Although not a clinical diagnostic, the CES-D has good sensitivity and specificity for detecting depression in the general population and among patients with comorbid chronic illness, such as HIV (18,36–38).”

3. Were confounders/effect modifiers considered in associating depression among PWID with the risk of secondary HIV transmission?

Response: We appreciate the reviewer bringing up these methodological considerations, which we have now described as a limitation of our study in the Discussion: “Our mathematical modeling approach enabled the integration of behavioral and biological mechanisms to estimate HIV transmissions according to depression status, but we did not conduct a formal causal analysis with consideration of potential confounders or effect measure modifiers.”

Given this study limitation, we also toned down some of the language in the Discussion paragraph on future research directions. Modeled interventions on depression would be hypothetical and could potentially reduce expected transmissions under certain assumptions, now described in the text: “Future mathematical models could investigate the extent to which successful depression treatment might reduce continued HIV incidence, assuming that decreasing the severity of symptoms to mild or none through treatment would translate to the lower transmissions predicted for those groups in our study.”

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

27 Sep 2022

The role of depression in secondary HIV transmission among people who inject drugs in Vietnam: A mathematical modeling analysis

PONE-D-22-09164R1

Dear Dr. Levintow

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Hamid Sharifi

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: excellent work, i have no further comments. This work raises an excellent point of the syndemic of psychological state and risk management.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

4 Oct 2022

PONE-D-22-09164R1

The role of depression in secondary HIV transmission among people who inject drugs in Vietnam: A mathematical modeling analysis

Dear Dr. Levintow:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Hamid Sharifi

Academic Editor

PLOS ONE