On-site health service delivery models at syringe services programs in the United States: Results of a national cross-sectional survey

Abstract

Background

People who inject drugs (PWID) have many needs for health services, but frequently lack access to and/or do not utilize those services. Syringe services programs (SSPs) are low-stigma environments where health services can be provided, but are not well described.

Objective

We characterized types of health services delivery available at syringe services programs (SSPs) nationally and assess SSP characteristics associated with them.

Methods

Using a national survey of SSPs on services provided in 2019 (N = 153), we conducted latent class analysis to determine the best fit model for health service delivery types. We examined the association between health delivery types and SSP organizational characteristics using multinomial logistic regression.

Results

A 3-class model was best fit and included 1) a comprehensive care model that had a high probability of offering multiple health services (11 %), 2) a testing and wound care model that offers mostly HIV/HCV testing and wound care (57 %), and 3) a minimal or no health services model that predominantly do no offer medical services (32 %). Comprehensive care and HIV/HCV testing and wound care SSPs were significantly more likely to have ≥ 50 % of their funding from public sources (OR=13.7 and OR=18.0), be a larger program (4th quartile in syringe distribution, OR=25.2 and OR=4.7), and less likely to be a grassroots program (OR=0.1 for both) compared to minimal care SSPs.

Conclusions

With 11 % of SSPs providing comprehensive care and one-third providing minimal services, there is an opportunity to expand health care services at SSPs with further public funding investments.

Highlights

• •Most syringe services programs (SSPs) have on-site health care models (68 %).
• •Few SSPs offered comprehensive care but 57 % offered HIV/HCV testing and wound care.
• •Having ≥ 50 % of funding from public funds results in more robust healthcare offerings.
• •Larger SSPs are associated with more robust onsite healthcare service models.

1. Introduction

People who inject drugs (PWID) have an elevated risk for injection-related infections, chronic diseases, and poor health outcomes related to their drug use (Degenhardt et al., 2017). Despite their needs for health care services to prevent or address these conditions, many do not engage in non-emergency health care because of limited access to services, stigmatizing prior experiences with the health care system, and discriminatory policies within the health care system (e.g., requiring abstinence to obtain care) (Austin et al., 2022; Biancarelli et al., 2019; Chan Carusone et al., 2019; Miller‑Lloyd et al., 2020; Muncan et al., 2020). PWID avoid stigmatizing healthcare experiences by delaying healthcare, not disclosing their drug use, and downplaying symptoms, sometimes including pain, which can adversely affect both their health and their receipt of appropriate care (Biancarelli et al., 2019). Among the few studies that have examined healthcare access among PWID, one found that about 50 % of PWID did not have a primary care physician and about 70 % reported not getting healthcare when they needed it (Miller‑Lloyd et al., 2020). Earlier studies found that 50 % or less of PWID have seen a primary care physician in the past 6 months (Artenie et al., 2015; Heinzerling et al., 2006). In addition, only 22 % of people with an opioid use disorder have received medications for opioid use disorder (MOUD) (Jones et al., 2023) and 23 % of PWID with hepatitis C virus (HCV) have been evaluated by a provider for HCV treatment (Falade‑Nwulia et al., 2020). Poor primary care access and low utilization result in a higher rate of costly emergency department visits for PWID (Nambiar et al., 2018) and their emergency department experiences often reaffirm or worsen their stigmatizing experiences with the U.S. healthcare sector (Mayer et al., 2023).

Syringe services programs (SSPs) are places where harm reduction services are provided to prevent the spread of infectious diseases, such as HIV and HCV, and to prevent opioid overdose fatalities. Prevention services include providing sterile injection supplies, overdose education, and distributing naloxone for reversal of opioid overdoses. In addition to these services, SSPs can provide a wide range of additional medical services. In a U.S. national survey of services available in 2019, the majority of SSPs reported that they offered on-site HIV and HCV testing, a quarter offered on-site confirmatory HIV/HCV testing and pre-exposure prophylaxis (PrEP), and just over 10 % offered on-site HIV and HCV treatment (Behrends et al., 2022). Additional on-site medical services offered at some SSPs included medications for opioid use disorder (20 % of SSPs), wound care (50 %), other primary care services (excluding wound care) (29 %), and medications for mental health (5 %) (Behrends et al., 2022).

In addition to the expanded access to care that SSP-based medical services provide, PWID may prefer to receive medical services at SSPs because they experience less stigma receiving medical care in SSPs (Frost et al., 2021; Muncan et al., 2020). Given the unfulfilled healthcare needs of PWID and the potential for SSPs to address these needs, this study aims to 1) characterize different health care delivery models offered at SSPs nationally, and 2) assess organizational and geographic characteristics associated with more comprehensive models of care to better understand the resources and infrastructure needed to expand these models in the future.

2. Methods

2.1. Data

We used data from the 2020 Dave Purchase Memorial Survey, the longest running national survey of SSPs in the United States (Centers for Disease Control and Prevention (CDC), 1995). In the 2020 survey, SSP directors were asked about harm reduction and medical services provided in 2019. SSP directors were recruited from the North American Syringe Exchange Network (NASEN) directory, which historically has represented the majority of SSPs in the US because of NASEN’s Buyer’s Club that offers harm reduction supplies at low cost to member SSPs (Des Jarlais et al., 2020). Because the survey was completed in 2020 when SSP services might have been paused due to the COVID-19 pandemic, SSP eligibility for the analytic sample was based on having 1) offered harm reduction services for > = 3 months in 2019 and 2) purchased syringes in 2019 or 2020 from the NASEN Buyer’s Club to indicate that the SSP had not shut down during the pandemic. Further details on the eligibility criteria are described in a previous publication, including potential response bias (Behrends et al., 2022). All 396 of SSPs on the NASEN list were emailed, resulting in 164 programs that responded, and 153 eligible respondents that completed the survey out of 211 SSPs that met the eligibility criteria. Based on NASEN records, among the 232 non-respondents that were emailed, 51 % did not purchase supplies in 2019 or 2020, 18 % had only purchased supplies in 2020, indicating a newly opened program, and 6 % had only purchased in 2019, suggesting a potential closure in 2020. The remaining non-respondents self-reported that they were not operational for 3 months or longer in 2019. We excluded two SSPs that did not respond to questions about on-site medical services, which yielded a final sample of 151 respondents (72 % response rate among 211 eligible SSPs). It is possible that some SSPs did not obtain syringes from NASEN because they were fully supplied from other sources or that the eligibility criteria still resulted in misclassification of some SSPs as operational in 2020 even though they were not, potentially contributing to some selection bias. Surveys were completed between August 2020 and February 2021. This survey was considered non-human subjects research because we collected only program administrative data.

2.2. Measures

Medical models were identified based on the medical services offered on-site at the SSP. Respondents were asked whether or not the SSP provided on-site HIV testing, HCV testing, HIV treatment, HCV treatment, HIV pre-exposure prophylaxis, wound care, other primary care (not including wound care), medications for opioid use disorder, and mental health care (which includes medication and/or counseling).

Descriptive measures described the healthcare infrastructure at SSPs required for delivery of some medical services, including having a private space available for clinical or counseling services, having a clinical provider available at least part-time who can prescribe medications, and having a phlebotomist available at least part-time.

To examine the program characteristics associated with medical models, the following covariates were examined: receiving > =50 % of funding from public funding (vs. <50 % from public funding); program size defined as number of syringes dispensed between January 1 to December 31, 2019 by quartile; organization type; and SSP-reported rurality (non-rural vs. rural). Public funding was defined as receiving funds from local, county, state, or federal sources and the 50 % or more cutoff for the public funding variable was developed to represent programs with the majority of their budget from public funds vs. those with a majority from private foundations or donations. Organization types were categorized as stand-alone, non-profit organizations that had 501c3 tax exempt status, SSPs that were operated by another non-profit organization (e.g., HIV- related community-based programs), SSPs operated by health departments (i.e., city, county, or state health department), and stand-alone, grassroots community-based organizations (without 501c3 status).

2.3. Analysis

First, latent class analysis was conducted to identify medical model classifications based on combinations of the on-site medical services described above. This analysis was conducted using SAS version 9.4 and the PROC LCA command (Lanza et al., 2007). Latent class analysis is a methodology that detects unobserved heterogeneity in samples based on study participants’ responses to selected indicator variables (Weller et al., 2020). This methodology allows for the identification of unobserved subgroups, called latent classes, with the purpose of identifying classes where individuals are similar to each other but different from individuals in other classes. We selected a model that minimized the Bayesian Information Criteria (BIC) to determine the optimal number of classes, and we selected a model with entropy values that exceeded 0.8 to determine the best fit model (Weller et al., 2020). In addition, we ensured that each class membership sample size was more than 5 % of the total sample. Class membership probabilities were also reported for this analysis.

Due to sparse cell sizes for some of the covariate combinations, we examined bivariate associations using multinomial logistic regression methodology and did not conduct multivariable analysis. This multinomial logistic regression analysis used a recommended three step approach (Bolck, 2004; Dziak et al., 2020; Weller et al., 2020) that allowed for examination of differences among classes without affecting the class structure; this was done by taking into account the posterior probabilities of latent class membership of each respondent in the model. This method provides more robust estimates than a regular multinomial logistic regression model, which relies on assigning fixed latent class memberships to each respondent (Dziak et al., 2020). Using this three-step approach, “BCH” (Bolck, Croon, and Hagenaars) weights are constructed to adjust for class membership uncertainty using posterior probabilities of latent class membership. We used BCH-adjusted modal assignment for the three-step approach, which uses the resulting BCH weight for adjustment in the three-step multinomial logistic regression and assigns each program to its most likely class even if there is uncertainty about its exact membership. In the case of sparse data and small cells where the three-step approach fails to converge, we instead used an exact multinomial logistic regression model with fixed latent class membership. The three-step approach for analysis was executed using a SAS macro (Dziak et al., 2020), removing respondents with missing covariate data (n = 7, 5 %), and resulting in a final analytic sample size of 144 respondents. The multinomial bivariate model examined covariates that may influence offering different types of medical models: public funding, program size, organization type, and rurality. Other covariates that may influence our outcome that were also considered include region (Northeast, Midwest, South, West) and infrastructure covariates (on-site provider availability, having a brick-and-mortar storefront, having a private space available, having a clinical provider who can prescribe medications, and having a phlebotomist available), but these covariates were not included in the final results as they were not significant in exploratory analyses. Significance was assessed with two-sided 95 % confidence intervals (CI) of the odds ratios (OR).

3. Results

3.1. Syringe services program characteristics

Out of the 153 respondents, 151 SSPs had completed all the questions on medical services provision on-site. In 2019, 60 % of SSPs had dispensed greater than the median number of syringes dispensed in that year (Table 1). Over half of SSPs had 50 % or more of their funding from public funding sources whereas 37 % had no public funding at all. In terms of organizational type, 51 % of SSPs were 501c3 tax exempt stand-alone non-profit organizations, 26 % were operated by health departments, 15 % were stand-alone grassroots, community-based organizations (not 501c3), and 7.3 % were programs operated by non-profit organization (not a stand-alone program). The majority of SSPs (57 %) were located in urban areas, but 30 % were in rural areas and 13 % were in suburban areas. Regionally, one third of SSPs were in the West, about one quarter were in the South, about one quarter in the Midwest, and 15 % were in the Northeast.

Table 1.

Program characteristics from a 2019 sample of syringe service programs in the United States (N = 151).

	n	%
Program size (syringes dispensed in 2019)
1st quartile (<15,000)	37	24.5
2nd quartile (15,000−96,000)	39	25.8
3rd quartile (96,001−381,000)	37	24.5
4th quartile (>381,000)	38	25.2
Annual budget size
< $25,000	51	34.7
$25,000-$100,000	38	25.9
> $100,000	58	39.5
% of budget from public funds
None	55	37.2
Less than 50 %	17	11.5
50 % or more	76	51.4
Organizational type
Standalone non-profit organization (501c3, tax exempt status)	77	51.3
Operated by city, county, or state health department	39	26.0
Standalone grassroots community-based organization (no 501c3 status)	23	15.3
Operated by another non-profit organization	11	7.3
Location
Urban	85	57.1
Suburban	20	13.4
Rural	44	29.5
Region
Northeast	23	15.5
Midwest	40	27.0
South	36	24.3
West	49	33.1
Private space available for clinical or counseling services	99	66.0
Clinical provider available who can prescribe medications	44	29.1
Phlebotomist available	65	43.3

In terms of healthcare infrastructure, 66 % of SSPs had a private space available for clinical or counseling services, 29 % had a clinical provider available at least part time who could prescribe medications, and 43 % had a phlebotomist available at least part time. Fig. 1 shows the percentage of SSPs that offer each on-site medical service. Overall, 34 % either did not offer any medical services or offered only HIV and/or HCV testing, 41 % offered testing and 1–2 additional services on site, 12 % offered testing and 3 additional services, and 13 % offered testing and 4 or more additional services on site (Fig. 1).

Fig. 1.

Types and number of health services provided on-site (excluding HIV and HCV testing) from a 2019 sample of syringe service programs in the United States (N = 151).

3.2. Medical model classification

When determining the best fit model, a 3-class model was chosen because it minimized the BIC, had an entropy measure of 0.92 (close to 1), and had more than 5 % of the sample represented in each class (Appendix Table 1) (Weller et al., 2020). In the 3-class model, we identified and named the following medical model classes: 1) a Comprehensive Care Model that had a high probability of offering all or most medical services on site (11 % of sample), 2) a Testing and Wound Care Model that had a high probability of offering on-site HIV/HCV testing and wound care (57 %), and 3) a Minimal or No Medical Services Model with a low probability of offering any on-site medical services (32 %) (Fig. 2). When examining differences among these three classes across healthcare infrastructure variables, there were significant differences in having a private space (p < 0.001), a clinical provider available who can prescribe medications (p < 0.001), and a phlebotomist available (p < 0.001) with the comprehensive care model having the highest percentage of each compared to the other medical models (Appendix Fig. 1).

Fig. 2.

Probability of offering each on-site medical service for three syringe services health models identified through latent class analysis from a 2019 sample of syringe services programs in the United States (N = 151).

3.3. SSP characteristics associated with SSP medical models

For the multinomial bivariate model examining SSP characteristics associated with medical model classes (Table 2), the comprehensive care model SSPs were more likely to have 50 % or more total funding from public funds (odds ratios [OR]=13.7, 95 % confidence interval [CI]: 3.1, 60.6) than minimal or no medical services model SSPs. The testing and wound care model SSPs were also more likely to receive 50 % or more of funds from public sources (OR=18.0; CI: 5.7, 56.7), and less likely to have a budget size of less than $25,000 (OR=0.1, CI: 0.0, 0.7) than minimal or no medical services model SSPs (Table 2). By program size—calculated using number of syringes distributed in 2019—comprehensive care and testing and wound care SSPs had higher odds of belonging in the 3rd (OR=22.5, CI: 2.9, 256.2; OR=6.2, CI: 1.7, 21.9, respectively) and 4th quartile (OR=25.2, CI: 2.4, 267.8; OR=4.7, 1.4, 15.8, respectively) than SSPs with the minimal or no services model (Table 2). Grassroots programs have a lower odds of having a comprehensive care model (OR=0.1, CI: 0.0, 0.4) or testing and wound care model (OR=0.1, CI: 0.0, 0.3) than a minimal or no medical service model compared to standalone non-profit organizations (Table 2). Of note, no grassroots organizations had a comprehensive care model (Appendix Table 2). Rurality was not associated with significant differences in odds of having either a comprehensive care or testing and wound care model compared to a minimal or no medical services model.

Table 2.

Program characteristics associated with medical models identified through latent class analysis from a 2019 sample of syringe service programs in the United States (N = 144).

			Comprehensive Care Model vs. Minimal or No Services Model	Testing and Wound Care Model vs. Minimal or No Services Model
SSP characteristics			OR (95 % CI)	OR (95 % CI)
Public Funding
None			Ref	Ref
Less than 50 %			2.8 (0.4, 19.9)	3.1 (0.8, 11.7)
50 % or more			13.7 (3.1, 60.6)*	18.0 (5.7, 56.7)*
Budget size
$25,000-$100,000			Ref	Ref
< $25,000			0.3 (0.0, 5.4)	0.1 (0.0, 0.7)*
> $100,000			4.1 (0.6, 30.1)	1.9 (0.7, 5.3)
Program Size (syringes dispensed in 2019)
1st quartile (<15,000)			Ref	Ref
2nd quartile (15,000−96,000)			3.0 (0.2, 40.0)	2.2 (0.8, 6.3)
3rd quartile (96,001−381,000)			22.5 (2.0, 256.2)*	6.2 (1.7, 21.9)*
4th quartile (>381,000)			25.2 (2.4, 267.8)*	4.7 (1.4, 15.8)*
Organization Type†
Standalone non-profit organization (501c3, tax exempt status)		Ref		Ref
Operated by another non-profit organization			1.8 (0.2, 15.8)	0.7 (0.1, 5.1)
Operated by city, county, or state health department			1.1 (0.1, 6.9)	2.7 (0.9, 10.1)
Standalone grassroots community-based organization (no 501c3 status)			0.1 (0.0, 0.4)*	0.1 (0.0, 0.3)*
Rurality†
Not rural		Ref		Ref
Rural			0.2 (0.0, 1.1)	1.2 (0.5, 2.7)

95 % CI: 95 % confidence intervals.

Ref: reference category.

* statistically significant (p < 0.05).

† due to sparse cells, bivariate associations were examined using unadjusted exact logistic regression, not the 3-step approach.

4. Discussion

The majority of SSPs in this national survey provided HIV and HCV testing and one or more additional on-site medical services, indicating the large potential role of SSPs in providing non-emergency prevention and treatment services to an underserved population of PWID. SSPs have been shown to effectively provide on-site HCV treatment (Eckhardt et al., 2022; Muncan et al., 2021), MOUD (Jakubowski et al., 2023), and vaccinations (Heidari et al., 2023). SSPs provide services in a nonjudgmental, respectful and trusted environment for PWID and deliver these services using a low threshold model that is flexible about missed appointments, allows walk-in visits, and makes service access convenient via mobile units (Bartholomew et al., 2022; Heidari et al., 2023; Kapadia et al., 2023). PWID have expressed interest in a “one-stop shop” for their harm reduction and medical services needs in settings such as mobile units to improve access to all services (Bartholomew et al., 2022). These approaches to health care delivery are in stark contrast to the experiences of PWID who describe receiving poor and stigmatizing treatment in traditional and emergency-based healthcare settings and cite these experiences as major barriers to obtaining health care (Austin et al., 2022; Chan Carusone et al., 2019; Kapadia et al., 2023; Mayer et al., 2023; Miller‑Lloyd et al., 2020; Voon et al., 2021).

While SSPs have been shown to be important places for providing a diverse set of medical services, medical service models at SSPs have been understudied. We identified three common medical models at SSPs: a comprehensive care model, a testing and wound care model, and a minimal or no service model. These models align with different levels of healthcare infrastructure and staffing. A comprehensive model requires private space for healthcare visits, a prescribing healthcare provider to provide care such as MOUD and HIV/HCV treatment, PrEP, or mental health medications, and phlebotomy for some testing and for treatment monitoring. A testing and wound care model requires private space to conduct testing as well as training in basic wound care (cleaning and dressing a wound) (NASTAD, 2023; National Harm Reduction Coalition, n.d.) and testing, but these services can be provided by individuals without healthcare credentials. Transitioning between a minimal care model to a testing and wound care model may be achievable with some additional resources for staff training and supplies (e.g., testing kits, wound care supplies). Some programs, however, may not be positioned to provide these services, such as volunteer-based grassroots services on a shoe-string budget, programs with limited physical space, and unsanctioned programs with limited resources. Nonetheless, efforts to transition SSPs that have minimal or no medical services to the testing and wound care model may be cost-saving in the long run by reducing emergency department visits and hospitalizations that are common among PWID who delay or avoid non-emergency health care services (Islam et al., 2012). For example, skin and soft tissue infections cost the healthcare system $15 billion US dollars annually (Tun et al., 2018).

SSPs with 50 % or more of the budget supported by public funding and that distributed more syringes (i.e. larger SSPs) were the most important predictors of providing a comprehensive medical model or a testing and wound care model. Stand-alone 501c3 non-profits or health department-related programs may be better positioned to receive public funding than grassroots programs because they meet common government funding requirements that programs are sanctioned, non-profit organizations (Akiba et al., 2024). Even if they meet funding requirements, small program size and low staff capacity of some SSPs may still limit their ability to acquire public funds given the complicated grant funding mechanisms available and high burden reporting required of these grants (Akiba et al., 2024; Green et al., 2012) Unsurprisingly, none of the grassroots programs had a comprehensive care model which may reflect lack of funding stability, lack of access to a reliable space for clinical care, unsanctioned legal status of the program, and/or staffing needs that go beyond the volunteer structure of many grassroots organizations. There are multiple barriers to obtaining public funds that include organizational requirements (e.g., non-profit status), complex application processes, state-based or local policies that prevent legal establishment of SSPs or funding of SSPs and/or syringes, and grant reporting requirements that are difficult to meet for under-resourced programs (Akiba et al., 2024; Green et al., 2012; Jones, 2019). Attaining funding from these public sources can be improved with policy changes that support SSP operations and funding, technical assistance for SSPs, more accessible application processes, and the provision of more programmatic support from the public agencies that fund these grants (Akiba et al., 2024; Green et al., 2012).

Larger programs are more likely to have the healthcare infrastructure and funding resources to support on-site medical service delivery. However, among comprehensive care model SSPs, 43 % were small/medium sized organizations, indicating that there is capacity at smaller SSPs to offer comprehensive care service models. Some SSPs may be located in settings with more medical infrastructure where SSP harm reduction services are provided as one of many services. Further research is needed to understand the infrastructure, staffing, and support needed to feasibly expand medical services at SSPs. In addition, understanding the context in which comprehensive medical models are successful and how this affects harm reduction service delivery is also needed.

There are limitations to this study. The sample was limited to programs recruited from a NASEN directory, which may be less likely to include grassroots and unsanctioned SSPs (Lambdin et al., 2023). However, our aim was to examine characteristics associated with more comprehensive models of care, which grassroots SSPs are less likely to provide. This study may have been limited by sample size in being able to identify all of the potential organizational aspects that may be related to medical models. The survey was completed in 2020 during the COVID-19 pandemic, and therefore, the response rate was limited by potential closures of SSPs and may not be generalizable to all SSPs. In addition, while some SSPs may offer several services, uptake of these services may be low. For example, a previous study at one SSP reported that HIV/STD service engagement was low (9 %), but also demonstrated that more encounters and engagement with the SSP was associated with more medical service engagement (Beidelman et al., 2023). This study did not collect information on SSP participants’ use of medical services at the SSP versus elsewhere, and therefore cannot definitively address whether specific SSP healthcare models fill in health care access gaps for PWID. Collecting this information from individual PWID could help guide future efforts to scale up healthcare services for PWID at SSPs and elsewhere.

We identified three classes of medical models at SSPs nationally: a comprehensive care model, a testing and wound care model, and a minimal or no medical services model. Large SSPs and those with over 50 % public funding currently appear to be better positioned to provide comprehensive healthcare services but a minority are currently doing so. With only 11 % of SSPs providing comprehensive care and one-third providing minimal or no services, there is an opportunity to improve health care access for PWID through further public funding investments in SSPs to support expansion of the medical services that they provide. Given that PWID may avoid traditional healthcare settings due to prior stigmatizing experiences, SSPs have an important role in engaging PWID in healthcare and averting downstream costs from emergency department visits and hospitalizations resulting from delayed care. Further efforts and resources should be directed to encourage SSP adoption of, at minimum, a testing and wound care model which requires less healthcare infrastructure than comprehensive care. In settings where providing more comprehensive services is feasible, innovative low-threshold models of PWID healthcare pioneered by SSPs should be championed to support better access to healthcare for PWID. While providing medical services at SSPs hold great promise for overcoming low healthcare engagement among PWID, not all harm reduction spaces are optimal settings for medical services delivery, due to potential philosophical conflicts about healthcare service delivery and structural barriers to integrating services (Heller et al., 2004). To overcome potential barriers to integrating medical service at SSPs, implementation research is needed to understand the specific settings and supports required for successful integration and we need to more information about the type of medical services PWID prefer to receive in the SSP setting.

CRediT authorship contribution statement

Behrends Czarina N.: Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Writing – original draft. Des Jarlais Don C.: Project administration, Resources, Supervision, Writing – review & editing. Luhur Winston: Formal analysis, Writing – review & editing. Lu Xinlin: Formal analysis, Writing – review & editing. Corry Grace J.: Investigation, Writing – review & editing. Glick Sara N.: Writing – review & editing. Kapadia Shashi N.: Writing – review & editing. Perlman David C.: Writing – review & editing. Schackman Bruce R.: Funding acquisition, Project administration, Supervision, Writing – review & editing.

Funding

Dr. Kapadia has received research grants to his institution from Gilead Sciences Inc, unrelated to the current work. The other authors have no conflicts of interest to disclose. This research was supported by the National Institute of Drug Abuse (R01DA027379, P30DA040500, P30DA011041, and K01DA048172). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the views of the funding agencies or the U.S. government.

Declaration of Competing Interest

Dr. Kapadia has received research grants to his institution from Gilead Sciences Inc, unrelated to the current work. The other authors have no conflicts of interest to disclose.

Appendix A. Supplementary material

Supplementary material