HIV, hepatitis B & C in people who inject drugs in India: A systematic review of regional heterogeneity & overview of opioid substitution treatment

Abstract

Background & objectives:

This systematic review evaluates the human immunodeficiency virus (HIV), hepatitis C virus (HCV) and hepatitis B virus (HBV) burden among people who inject drugs (PWIDs) in India. In addition, we selectively examined research on opioid substitution treatment (OST)-related services due to their role in antiviral treatment uptake and adherence.

Methods:

Data were sourced from peer-reviewed and government publications between 1991 and September 20, 2023, searched in MEDLINE, Scopus and EBSCOhost. English language studies reporting weighted prevalence or raw numbers and recruitment sites were included for review. Quality was assessed using the Joanna Briggs Institute tool. Data synthesis was done in graphs and tables.

Results:

We included 50 reports, yielding 150 HIV, 68 HCV and 24 HBV prevalence estimates across India, revealing significant regional heterogeneity. Notably, 16 States had a single community-based HIV estimate, and 19 States had limited or no HCV data. The highest HIV and HCV prevalence was in Manipur (74.7% and 97.5%, respectively) in 1996. Recent spikes included 50.2 per cent HIV prevalence in Punjab (2010) and 73 per cent HCV in Uttar Pradesh (2021). Nationally, OST coverage in 2020 was under five per cent, with some northeast, north and central States exceeding this, but most others were falling below two per cent. No studies on the cost-effectiveness of directly observed treatment models for OST were identified.

Interpretation & conclusions:

There is a lack of sufficiently granular and generalizable estimates for HIV prevalence and any estimates for HCV and HBV among PWIDs in large parts of the country. Community-based representative studies are required to quantify the prevalence and severity of these diseases and allocate resources.

India has about 8.5 lakh people who inject drugs (PWIDs), most (96%) using opioids1. PWIDs belong to most at risk population (MARP) groups for blood-borne infections (BBIs), among which hepatitis C virus (HCV), hepatitis B virus (HBV) and human immunodeficiency virus (HIV) are of public health concern. India’s National AIDS Control Programme (NACP) and National Viral Hepatitis Control Programme (NVHCP) have been active since 1992 and 2018. Despite overall success of the NACP, its success with PWIDs has been limited. Between 2004 and 2017, HIV seroprevalence decreased by approximately eight per cent among female sex workers (FSWs) and only six per cent in PWIDs2. Against this background, the current article reviews HIV, HBV and HCV disease burden and opioid substitution treatment (OST) services for PWIDs.

Pachuau et al3 and Goel et al4 reviewed HIV and HCV prevalence in PWIDs in 2022. We expand these reviews for three reasons. First, injecting drug use (IDU) occurs in closed groups, causing micro-epidemics of BBI (10-fold difference between two localities of Delhi)5, and neither of these studies focused on regional heterogeneity or temporal trends (doubling of prevalence within five years in the northeastern State of Mizoram)6. “Second, whether enough original research is available to reasonably estimate the prevalence of BBIs among PWID reasonably requires examination”. For example, Goel et al4 reported 46 HCV prevalence estimates, of which some are from the same dataset, like McFall et al7 in 2017 and Solomon et al8 in 2015 (Supplementary File 1, Ref 4 ^{(3.5MB, pdf)} ). This inflates the denominator and gives a false sense of certainty. Finally, one must examine data other than seroprevalence, such as clusters of differentiation 4 (CD4) counts, hepatic fibrosis and viral load, to understand disease burden better.

Public health strategies to reduce new HIV and HCV infections are shifting, emphasizing not just behavioural change but also effective antiviral treatments. For HIV, proper antiretroviral therapy can lower transmission, even with ongoing needle sharing9. Direct-acting antivirals (DAAs) can eradicate the virus, nullifying transmission risk in HCV. However, the success of this treatment-as-prevention (TASP) approach hinges on treatment availability, uptake and adherence. In this context, OST, offering long-acting opioids such as buprenorphine or methadone, is crucial for engaging PWIDs in various treatment services. Overall, OST decreases mortality in PWIDs and improves their quality of life10.

Opioid-using PWIDs typically have opioid dependence, now recognized as a brain disease due to drug-induced neuroplasticity. This affects their ability to prioritize, remember and plan11. OST stabilizes patients, facilitating engagement with treatment services and improving general functioning, which aids in managing BBIs. For instance, OST enhances HCV testing rates and DAA treatment uptake12. Similarly, it boosts Anti-Retroviral Therapy (ART) enrolment, adherence and viral suppression in HIV-positive PWIDs13. In essence, effective treatment for HIV and/or HCV in PWIDs requires OST with antivirals. Research on the availability of these services, uptake and outcomes in India is hence crucial.

Hence this systematic review was undertaken to assess the disease burden of HIV, HCV and HBV (including seroprevalence, viral load, CD4 counts and hepatic fibrosis) among PWIDs in India from 1991 to 2023, with the following components: (i) evaluate regional data quality and availability on HIV, HCV and HBV prevalence at city/district and State levels, (ii) study regional variations in HIV, HCV and HBV prevalence at these levels and (iii) track changes in prevalence for these infections over the study period. The other objective was to provide an overview of OST history, availability, current practices, and barriers, particularly its integration with antiviral treatment.

Material & Methods

Operational definitions: PWIDs was defined as individuals using injectable drugs for non-medical reasons at any time in their lives. HIV-positive status was defined as a positive HIV-1 or 2 screening test and additional confirmatory assay or positive western blot test result. HCV positivity was defined as laboratory test confirming presence of anti-HCV antibody with or without HCV ribonucleic acid confirmation. HBV infection was defined as hepatitis B surface antigen detection through laboratory assay.

Eligibility criteria: Inclusion criteria for this review were peer-reviewed studies and government reports published from January 1, 1991 to August 1, 2023 that (i) reported prevalence of HIV, HBV or HCV among PWIDs in India; (ii) included sufficient data to calculate prevalence or provide weighted estimates, with or without 95 per cent confidence intervals (CIs) around the estimates; (iii) did not focus solely on co-infections; (iv) specified the recruitment location at the city/district or State level and (v) presented new, previously unreported prevalence data. Acceptable study types were cross-sectional surveys, retrospective medical record reviews and cohort reports with cumulative prevalence data.

Information sources and search strategy: We searched MEDLINE (PubMed), Scopus and Psychology & Behavioral Science Collection through EBSCO host between September 10, 2023 and September 20, 2023. Search strategies are detailed in Supplementary Tables I ^{(3.5MB, pdf)} and II ^{(3.5MB, pdf)} . The language was limited to English, and publication dates ranged from January 1, 1991 to August 1, 2023. Additional sources included reference lists from four related systematic reviews3,4,14 and consultations with researchers in the field. We also reviewed reports from the National AIDS Control Organization’s HIV Sentinel Surveillance (HSS) and Family Health International’s Integrated Behavioural and Biological Assessment.

Study selection: We first screened records from the three databases by title and abstract, removing duplicates. Citations without abstracts were tentatively included. Two authors (PS and LS) independently assessed full texts of potential studies. Discrepancies were resolved by a third author (MSK). Regarding data overlap between reports, only the latest one was included if sharing was evident; alternatively the authors were contacted for clarification. Peer-reviewed reports were prioritized over other sources.

Data extraction: Two authors (PS and LS) concurrently extracted data in the first pass and cross-verified each other’s entries in the second. Study characteristics and disease burden variables extracted are detailed in Supplementary Table III ^{(3.5MB, pdf)} .

Quality assessment: We employed the Joanna Briggs Institute (JBI) tool for quality assessment, tailored for prevalence studies15. Decisional rules for using the JBI tool are provided in Supplementary Table IV ^{(3.5MB, pdf)} . Given the report diversity and time span (three decades), we did not exclude studies based on quality. PS and LS independently rated reports in the first pass and resolved discrepancies by consensus in the second.

Data synthesis: We opted not to conduct a meta-analysis for pooled prevalence, deeming it uninformative due to time, region and sampling variations. Instead, we presented trends in study density and reported prevalence in six geographic regions (Supplementary Table V ^{(3.5MB, pdf)} for State lists).

For uncertainty, weighted CIs were used for respondent-driven sampling (RDS) studies when available. In others, we calculated 95 per cent Wilson CIs (covers all sample sizes adequately)16 using reported numerators and denominators.

The year when recruitment ended was used to timestamp prevalence estimates; if unreported, we used submission or publication year as a last resort. We assessed temporal trends using year-wise pooled prevalence (total cases reported/total PWIDs tested) and Wilson’s CIs. Finally, the National Drug Use Survey, 2019 (NDUS 2019)1, informed regarding the ‘at-risk population’ size at the State level. R software was used for all analyses and graphics17.

Overview of OST – history, current practices and integration with antiviral treatments: We searched MEDLINE (PubMed) from September 10, 2023 to September 20, 2023 without date restrictions to capture all OST service related research in India. The search strategy used is in Supplementary Table VI ^{(3.5MB, pdf)} . Annual Reports from the Ministry of Health and Family Welfare were also reviewed for OST coverage data. Of the 214 initially screened records, LS and PS reviewed 23 full-text articles and seven from reference lists. Findings are narratively synthesized around three themes: OST history and current practices in India, availability and integration with antiviral services, and barriers. Our aim was to present an overview of OST in India rather than an exhaustive review; thus, only PubMed and government reports were reviewed.

Results

HIV, HCV and HBV in PWIDs: We included 48 peer-reviewed5,8,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63 and two non-peer-reviewed reports64,65. Figure 1 shows the PRISMA flow chart, and a list of excluded reports with reasons given in Supplementary Table VII ^{(3.5MB, pdf)} . We excluded HSS reports except the one in 200664 due to the lack of State-wise denominators in others. Despite the methodology stating a minimum sample of 250 per site, we did not find this assumption reliable. The Integrated Biological and Behavioural Surveillance (IBBS) showed varied State-wise sample sizes, many falling short of the 250 target65.

Fig. 1.

The PRISMA flow diagram.

Summary of studies: Fifty reports provided 169 estimates of BBI in PWIDs of India (150 HIV, 68 HCV and 24 HBV estimates; Supplementary Table VIII ^{(3.5MB, pdf)} ). The Box summarizes important characteristics of the included studies.

Box.

Important characteristics of included studies

Summary of study characteristics (n=50)

Sampling design: -52 per cent of studies are community based, using methods like time–location clusters (n=13), RDS (n=9), and social network based (n=4).

-36 per cent used treatment-seeking populations. Sample sizes: Community-based studies have larger samples (median=1806, range=91-19,602) than treatment settings (median=150, range=15-12,058)

Large studies: IBBS 2014-2015 covered 29 states/UTs (Aridoss 2022)18 (n=19,602) and YRGCARE (Solomon 2015)8 covered 15 cities (n=14,481)

Gender representation: 62 per cent of studies did not include female PWIDs or failed to report their number. Only 28 per cent discussed BBI in female PWIDs

Clinical metrics: -Only 7 of 42 HIV studies reported viral load or CD4 counts8,23,28,34,37,47,62.

- 6 of 30 HCV studies included confirmatory RNA tests23,28,34,37,52,62.

- Only one reported any measure of hepatic fibrosis (liver stiffness)62 Funding:

- 16 received international funding8,23,24,31,34,36,37,44,45,47,48,52,58,61,62,66 5 received national35,42,43,46,55 and one combined24.

- The rest did not mention a funding source

All nine RDS studies received international funding

RDS, respondent-driven sampling; IBBS: integrated biological and behavioural surveillance; YRGCARE: Y.R. Gaitonde Centre for AIDS Research and Education; UTs, union territories; PWIDs, persons with injection drugs; BBI, blood-borne infection; HIV, human immunodeficiency virus; CD4, cluster of differentiation 4

Quality assessment: We found non-representative sampling frame and non-probability-based sampling in half of the studies included Supplementary Tables IX ^{(3.5MB, pdf)} and X ^{(3.5MB, pdf)} summarise the quality assessment and individual study appraisals.

Regional heterogeneity: Thirty-five studies reported city/district-level estimates, while 15 reported only State-level data. Since only 24 estimates of HBV prevalence were available, we did not examine the regional heterogeneity or temporal trend of HBV prevalence.

Heterogeneity in information density and quality: Figure 2 illustrates the research intensity across States compared to the current estimate of PWIDs. HIV-related studies align well with the PWIDs estimates, but gaps exist, notably in Jammu and Kashmir and southern States. HCV and HBV related data is scarce, lacking data for 15 and 19 States, respectively (Supplementary Table XI ^{(3.5MB, pdf)} ). Generalizability was challenging due to considerable variantion in the methods followed and resulting estimates; for instance, 10 out of 15 studies in Manipur used community-based recruitment (RDS or time–location), whereas only one of the three studies in Karnataka did so (Supplementary Table XI ^{(3.5MB, pdf)} ). State-level summaries and study details are shown in Supplementary File 2 ^{(11.5MB, pdf)} .

Fig. 2.

Research output in HIV, HBV and HCV in persons who inject drugs (PWIDs) and estimated number of PWIDs in different States of India. The colouring scheme uses quantiles for comparison; quantiles are reported in the legend. For example, Tamil Nadu has a smaller number of PWIDs and belongs to Q1 (1500-8283), but HIV in PWID has been studied extensively in the State and thus belongs to Q4 (2766-15426) subjects studied. Figure generated using R Software.

Heterogeneity in reported prevalence: While we decided against meta-analysis, assessing regional heterogeneity in disease burden was crucial. Figure 3 displays region-wise HIV prevalence among PWIDs; the northeast-India consistently showed high rates18,39, and recent data also indicated rising HIV prevalence in Central18,47 and northern part of the country, beyond Punjab50,58. Relatively less heterogeneity was identified in the prevalence of HCV infection, with most studies reporting a prevalence above 50 per cent. However, there was substantial variation in the reported estimates within a region (Fig. 4).

Fig. 3.

Regional heterogeneity in reported prevalence of HIV infection in persons who inject drugs. Points show prevalence, and the vertical lines show 95 per cent confidence intervals. Figure generated using R Software.

Fig. 4.

Regional heterogeneity in reported prevalence of HCV infection in persons who inject drugs. Points show prevalence, and the vertical lines show 95 per cent confidence intervals. Figure generated using R Software.

Temporal trends: Figure 5A reveals a temporal trend in the pooled HIV prevalence; high in the 1990s and stabilizing around 10 per cent in the 2000s. Early studies mainly came from the northeast, east (Manipur and Kolkata) and Chennai33,39,43,46,56, while recent ones are more geographically diverse. There was no clear trend identified for HCV seroprevalence (Fig. 5B); most estimates exceeded 50 per cent.

Fig. 5.

Pooled prevalence of HIV (A) and HCV (B) in persons who inject drugs. The estimate is a black dot; vertical lines show the 95 per cent confidence interval. The blue broken line is a trend curve fitted with a generalized additive model, and the shaded region shows a 95 per cent confidence interval of the curve. This plot does not show the number of subjects contributing to the pooled prevalence in each year. Figure generated using R Software.

Overview of OST in India:

History of OST in India: Kumar et al67, 2012 and Rao et al68, 2017 provide historical context for OST in India. To grasp its current status, we focused on three themes: the economics of large-scale OST medication supply, collaboration between government and non-government organizations (NGOs) and the legal issues surrounding OST.

A Delhi-based NGO, SHARAN, initiated community-based buprenorphine provision in the 1990s, funded by the commission for the European communities. This was replicated in other cities and adopted by the All India Institute of Medical Sciences (AIIMS), Delhi. Later, externally funded NGOs like Emmanuel Hospital Association (funded by Global Fund) expanded such programmes. The NACP phase III, in 2007, joined these efforts. However, community-based organizations were recognised as best suited for IDU-related services. As a result, the National AIDS Control Organization (NACO) currently partners with the Non- Government Organizations (NGOs) to run targeted intervention (TI) sites. However, the provision of OST remains primarily at government hospitals.

Dorabjee and Samson (1998)69 reviewed the SHARAN initiative, and it is poignant to note how innovative community-based interventions for PWIDs started and how are such services managed today. They described a user-friendly drop-in centre running in a slum (active area of drug peddling), providing take-home doses of buprenorphine and holding open discussions about ‘topping up’. The program, now, has moved to hospital-based (exceptions in Delhi and Northeast) directly observed OST70. From using sublingual buprenorphine for patients who were injecting buprenorphine, the approach has now moved to daily dosing of methadone due to cost concerns71 and using tramadol for detoxification for heroin users72.

Current status, availability and barriers: Both buprenorphine and methadone are available in India. The NACO expanded OST centres from 52 in 2007 to 310 by 2020-2021; yet they cover less than 5 per cent (n=41,587) of PWIDs in the country. In the same period, 166,299 PWIDs accessed other services from NACO supported TI-sites and treatment centres, indicating that service uptake or awareness is not a problem; specific service availability is. The increase in centres has been concentrated in the northeast; Supplementary Figs. 1 ^{(3.5MB, pdf)} and 2 ^{(3.5MB, pdf)} show region-wise change in the number of centres and clients. Figure 6 shows State-wise OST coverage for 2020-2021; in most States, OST services cover less than 10 per cent of PWIDs, and in the southern regions coverage falls below 2 per cent73.

Fig. 6.

Percentage of estimated persons who inject drugs (PWIDs) and who received opioid substitution treatment (OST) services during 2020-2021. The State-wise number of PWIDs is taken from NDUS 2019, and the number of OST recipients from the annual report of the National AIDS Control Organization. The report does not mention what qualifies as an OST recipient. Figure generated using R Software.

Legal hurdles restrict OST availability in the private sector. The Drug Controller General of India (DCGI) limits buprenorphine (2 mg) supply to ‘deaddiction centres’ only. Methadone, classified as an essential narcotic, demands extensive documentation68,74. Furthermore, even among psychiatrists, there is an exaggerated fear of diversion and abuse of OST75. This is a difficult barrier to contend with. Finally, there are multiple barriers perceived by PWIDs in accessing OST and other services, such as daily observed treatment coinciding with working hours, services located in multiple disparate locations, stringent registration requirements and unkind attitudes of healthcare providers76,77,78,79,80,81. Supplementary Table XII ^{(3.5MB, pdf)} summarizes published research related to this.

Empirical studies of OST effectiveness and integration: We reviewed the empirical evidence on effectiveness of OST for various outcomes in PWIDs. Most of the studies were conducted with treatment-seeking populations with convenience sampling and showed better treatment retention rates, improved physical and mental health and decreased needle sharing and crime81,82,83,84,85. Supplementary Table XIII ^{(3.5MB, pdf)} summarizes these findings.

Solomon et al86, in a project funded by the National Institutes of Health (NIH, USA), set up Integrated Care Centres (ICCs) in eight Indian cities in 2013. These centres offered co-located services for OST, HIV, HCV, TB screening and more87. Compared to the regular TI/OST services, ICC attendees were more likely to get tested and initiate treatment for HIV and HCV88,89. The study strongly suggests that co-location boosts service utilization.

The northern State of Punjab has a public programme targeting HCV in PWIDs, integrating DAA and OST at the same location. This ‘Punjab model’ showed promising results for integrated HCV treatment. However, the study did not analyze OST’s role in boosting treatment adherence90.

Discussion

We analysed 30 years of academic and government reports on HIV, HCV, and HBV in Indian PWIDs, along with OST research. The key findings were: (i) data quality and volume vary significantly between States, not solely due to varying IDU prevalence; (ii) for HIV, sufficient data exist to identify temporal trends tentatively, but HBV and HCV data are too sparse for regional or temporal analysis, and (iii) the proportion of PWIDs on OST is extremely low and has not improved over time.

Before discussing the findings, we note that studying stigmatized groups like PWIDs presents methodological challenges due to the absence of a sampling frame and infeasibility of random sampling; researchers choose between convenience samples (treatment seeking) or non-random but probability-based sampling. These choices impact the generalizability of investigative findings and bias of estimates. For instance, treatment-seeking PWIDs are likely to differ systematically from the broader population in terms of health awareness or severity of illness. Different sampling methods yield varying levels of rigour and bias. Social network-based methods like RDS offer the least biased estimates91. For example, two studies from Delhi, done within a span of two years, reported HIV prevalence of eight per cent63 and 37 per cent21 due to different sampling designs. Hence, along with availability, the data quality must also be scrutinized.

The State-level data availability in this review was found to be uneven. Sixteen States/UTs, housing 38 per cent of India’s total PWIDs, had only one community-based study (IBBS 2014-2015)65, while Delhi, with 11 per cent of PWIDs, had 10 such studies. This data inequity raises questions about national BBI prevalence estimates in PWIDs. Although historical factors partially explain the focus on high-prevalence States like Manipur, this perpetuates a cycle where only these States are studied, potentially overlooking emerging epidemics in other States or regions.

Uttar Pradesh, with the highest number of PWIDs in India is one such example; while studies in 2006 (HSS) and 2010 reported a five per cent HIV prevalence29,64, in 2013 RDS study by Solomon et al8 reported HIV prevalence of 31 per cent, a figure later supported by other community-based studies, including IBBS18,34,47. In contrast, the 2016-2017 HSS report showed a five per cent prevalence using a random sample of PWIDs receiving HIV prevention services50. This example highlights the need for generalizable, community-based studies across all States for accurate national prevalence estimates. Even then, the limited number of sites within States remains a constraint.

Unlike HIV, a little is known about HCV/HBV prevalence, transmission or disease severity. For example, among 19 States, either no HCV estimate was available (15 States) or data was generated from <500 PWIDs (four States). These States together accounted for 44 per cent of the estimated number of PWIDs in the country. However, some data may exist outside the public domain, notably from antiretroviral therapy centres, especially for patients requiring antitubercular treatment as well. This situation is in stark contrast with the perception in scientific circles. We are aware of unpublished data from Karnataka and Assam reporting 50 to 83 per cent HCV seroprevalence in PWIDS (two authors of this study PS and LS and D. Mukherjee, Tezpur). It is important to note that Assam has no estimates available, and Karnataka has one study of 15 subjects.

Current research on BBI in PWIDs largely overlooks female PWIDs despite their non-negligible presence. For example, Solomon et al8, in 2015, recruited almost 10 per cent females in the sample. Without efforts to achieve representative samples, BBI prevalence in female PWID would remain unknown, with an added risk of mother-to-child transmission. Furthermore, published reports fall short in detailing disease severity or stage of HBV and HCV infection among PWIDs. This limitation hinders the mapping of actual treatment needs, as seroprevalence alone is insufficient to gauge the scope of required interventions.

The history of OST-services in India has been of starts, reverses and off-late retrogression. While, the expansion of OST was concentrated in the northeast States (Supplementary Fig. 2 ^{(3.5MB, pdf)} ), the opening of new centres, as seen in Punjab during 2017-2020, revealed an unmet treatment need, evidenced by robust client uptake despite barriers like loss of working hours79. In addition, there is lack of studies on the cost-effectiveness of the current OST delivery model. The recent availability of methadone, expands OST options but requires evaluation against buprenorphine. Historically, low buprenorphine doses have been used in India68 based on the assumption that higher doses are unnecessary. If this assumption holds, methadone’s greater potency limits its broader application. However, methadone has a high risk of toxicity and thus cannot be given as take-home medication even for short periods. In effect, any savings made on the cost of medication are a transfer of burden to patients who would lose working hours every day to receive directly observed therapy.

This review has certain limitations. First, 11 records that were included in the screening could not be accessed. Five of these were studies conducted in the 1990s and thus may not alter our inference substantially. Second, we could not obtain raw datasets from the NACO for various HSS. Finally, it is likely that a search in Google Scholar and forward and backward citations could have identified some reports.

The findings of this systematic review have implications for policy and research both. Policymaking depends on updated and granular data. We aver that there is lumping to such extreme that the available estimates are unreliable, cities are made representative of States and States are made representative of regions. This is evident in the recent district-level prioritization exercise by the NACO, where the word ‘PWID or IDU’ is not mentioned and ‘HRG’ is mentioned only four times92. All prioritization uses data collected from antenatal surveillance that, by very definition, excludes half of the population. Furthermore, southern States, which have the highest number of people living with HIV, have the least data on BBI prevalence in PWIDs. Therefore, a critical look at the data used for prioritization exercises is necessary. The current infrastructure can be adapted to gather HCV data for better NVHCP implementation, which now operates mainly from tertiary gastroenterology departments. Learning from HIV and tuberculosis programmes, wider availability is crucial for treatment uptake. The NVHCP should therefore extend support to centres more likely to encounter PWIDs, such as addiction treatment facilities. There is also an urgent need for investment in country-wide epidemiological research.

For research, centres capable of data collection should report on HCV and hepatic damage in PWIDs. The current use of pan-genotypic DAAs necessitates surveillance to detect drug resistance. Lastly, initial studies suggest that bundling BBI services with OST is feasible, but more research is needed to inform policymaking.

Financial support and sponsorship

None.

Conflicts of interest

None.