Public health interventions, priority populations, and the impact of COVID-19 disruptions on hepatitis C elimination among people who have injected drugs in Montreal (Canada): A modeling study

Background In Montreal (Canada), high hepatitis C virus (HCV) seroincidence (21 per 100 person-years in 2017) persists among people who have injected drugs (PWID) despite relatively high testing rates and coverage of needle and syringe programs (NSP) and opioid agonist therapy (OAT). We assessed the potential of interventions to achieve HCV elimination (80% incidence reduction and 65% reduction in HCV-related mortality between 2015 and 2030) in the context of COVID-19 disruptions among all PWID and PWID living with HIV. Methods Using a dynamic model of HCV-HIV co-transmission, we simulated increases in NSP (from 82% to 95%) and OAT (from 33% to 40%) coverage, HCV testing (every 6 months), or treatment rate (100 per 100 person-years) starting in 2022 among all PWID and PWID living with HIV. We also modeled treatment scale-up among active PWID only (i.e., people who report injecting in the past six months). We reduced intervention levels in 2020–2021 due to COVID-19-related disruptions. Outcomes included HCV incidence, prevalence, and mortality, and proportions of averted chronic HCV infections and deaths. Results COVID-19-related disruptions could have caused temporary rebounds in HCV transmission. Further increasing NSP/OAT or HCV testing had little impact on incidence. Scaling-up treatment among all PWID achieved incidence and mortality targets among all PWID and PWID living with HIV. Focusing treatment on active PWID could achieve elimination, yet fewer projected deaths were averted (36% versus 48%). Conclusions HCV treatment scale-up among all PWID will be required to eliminate HCV in high-incidence and prevalence settings. Achieving elimination by 2030 will entail concerted efforts to restore and enhance pre-pandemic levels of HCV prevention and care.


Introduction
If untreated, chronic hepatitis C virus (HCV) infection can cause cirrhosis, liver cancer, and death ( Manns et al., 2017 ). Direct-acting antivirals (DAAs) cure HCV in over 95% of those treated, regardless of co-infection with HIV ( Sulkowski et al., 2015 ). The World Health Organization 's (WHO) targets for HCV elimination as a public health threat include reductions of 80% in chronic HCV incidence and 65% in HCV-related mortality between 2015 and 2030 ( World Health Organization, 2016 ). In Canada, 85% of new HCV infections are diagnosed among people who have injected drugs (PWID) -a priority population for elimination ( Jacka et al., 2020 ;Lanièce Delaunay, et al., 2021a ).  ( Greub et al., 2000 ). By the time they develop end-stage liver disease, most PWID have ceased injecting ( Manns et al., 2017 ;Montain et al., 2016 ) and ex-PWID should be considered for preventing HCV-related deaths.
In Montreal (Quebec, Canada), 82% of active PWID (i.e., people who report injecting in the past six months) use needle and syringe programs (NSP) and 33% are on opioid agonist therapy (OAT) -close to the 40% recommended coverage ( Canadian Network on Hepatitis C, 2019 ). They are tested for HCV every 14 months on average (annual testing is recommended) ( American Association for the Study of Liver Diseases, 2021 ; Leclerc et al., 2021 ). DAAs are universally accessible in Quebec and, in 2017, 33% of active PWID seropositive for HCV and 39% of active PWID seropositive for HCV and HIV had ever been treated for HCV infection ( Leclerc et al., 2021 ).
HCV seroincidence remains high among active PWID in Montreal (21 per 100 person-years [PY] in 2017), and seroprevalence is higher among active PWID living with HIV (86% in 2018) than among all active PWID (69%) ( Leclerc et al., 2021 ). However, those living with HIV are mostly engaged in care, and thus more easily reached by interventions -in 2018, 83% of active PWID living with HIV were on antiretroviral treatment (ART) in Montreal ( Leclerc et al., 2021 ).
Traditionally, HCV prevention for PWID has relied on harm reduction programs such as NSP and OAT ( Platt et al., 2017 ). Following the introduction of DAAs, studies have suggested that HCV "treatment-asprevention " can reduce HCV incidence and prevalence among PWID ( Iversen et al., 2019 ;Montaner, 2011 ;Pitcher et al., 2019 ). Results are context-dependent, but regular testing, broad DAA access, and highcoverage harm reduction programs are likely essential to any HCV elimination response ( Pitcher et al., 2019 ).
However, few studies have explored elimination strategies implemented among and for PWID living with HIV specifically. One modeling study conducted in Spain showed that increasing DAA uptake among PWID living with HIV was unlikely to achieve the WHO incidence reduction target due to ongoing HCV transmission from HIV-negative PWID ( Skaathun et al., 2018 ). Further, the effects of focused interventions on active and/or ex-PWID on HCV-related mortality are unknown.
Finally, starting in 2020, the COVID-19 pandemic has disrupted HCV prevention and care services for PWID across Canada and exacerbated the pre-existing overdose crisis ( Canadian Centre on Substance Use & Addiction, 2020 ; Public Health Agency of Canada, 2020 ; Institut National de Santé Publique du Québec, 2022 ). The impact of these disruptions could have compromised recent progress towards elimination.
Informed by robust data, dynamic models of disease transmission can simulate the course of an epidemic under different "what if " scenarios and estimate the direct and indirect effects of interventions ( Jit & Brisson, 2011 ). Our overall objective was to assess the potential of various intervention scenarios to achieve HCV elimination among all PWID and PWID living with HIV by 2030 in Montreal. Multiple local initiatives aim to eliminate HCV and we can leverage population-based surveys to inform elimination efforts ( McGill University Health Centre Research Institute, 2019 ;Klein et al., 2010 ;Leclerc et al., 2021 ). To strengthen this evidence-base we: 1) examined different elimination strategies for PWID subgroups (all PWID or PWID living with HIV; active and/or ex-PWID); 2) evaluated how likely these strategies are to achieve both incidence and mortality reduction targets; and 3) explored how COVID-19-related disruptions could affect elimination efforts.

Model structure
We used a dynamic, deterministic, sex-stratified compartmental model of HCV and HIV transmission via injection drug use among PWID. Key model parameters are presented in Table 1 , and the model's full description can be found elsewhere ( Lanièce Delaunay, et al., 2021 ).
Briefly, PWID are modeled from their first drug injection until death. We explicitly model three causes of death: background mortality among all PWID, liver-related mortality among those chronically infected with HCV, and AIDS-related mortality among PWID living with HIV. The rate of recruitment into the model replicates the observed decline in the active PWID population size from 11,700 in 1996 ( Remis et al., 1998 ) to 3910 in 2010 in Montreal ( Leclerc et al., 2014 ). PWID population size estimates after 2010 for either the whole province or neighboring cities are highly heterogenous and uncertain. As such, we assumed a stable population size of PWID in Montreal after 2010 ( Agence de la Santé Publique et des Services Sociaux de Montréal, 2013 ; Centre Intégré de Santé et de Services Sociaux de Laval, 2021 ; Centre Intégré Universitaire de Santé et de Services Sociaux de la Capitale Nationale, 2022 ; Jacka et al., 2020 ).
We modeled three concurrent dynamics: HCV transmission, HIV transmission, and injection behaviors ( Fig. 1 ). Seven compartments represent HCV infection and the care cascade: 1) susceptible to HCV infection (HCV-seronegative; all PWID enter the model through this HCV compartment); 2) acute infection (primary infection); 3) susceptible to HCV re-infection (HCV-seropositive); 4) acute infection (re-infection); 5) undiagnosed chronic HCV infection; 6) diagnosed chronic HCV infection; 7) under treatment. Seven compartments describe HIV infection and care cascade: 1) susceptible to HIV infection (all PWID enter the model through this HIV compartment); 2-4) living with HIV, not on ART, stratified by CD4 cell counts ( > 350 cells/mm 3 , 200-350 cells/mm 3 , < 200 cells/mm 3 ), and 5-7) living with HIV, on ART, stratified by CD4 cell counts. Injecting behavior dynamics are modeled through three compartments: 1) active PWID not on OAT; 2) active PWID on OAT; 3) ex-PWID (regardless of OAT status). We assumed disassortative mixing by sex ( Smith, et al., 2018 ) and proportional mixing by HCV status. Mixing patterns are allowed to vary from proportional to fully assortative by HIV status, between people susceptible to HIV infection or living with HIV and undiagnosed on one hand, and people living with HIV and diagnosed on the other hand ( Lanièce Delaunay, et al., 2021 b).
For both viruses, we used time-varying forces of infection that depend on prevalence among injecting contacts (depending on mixing by sex, HIV status, and injecting behaviours), NSP and OAT coverage, and treatment status of injecting partners ( Lanièce Delaunay, et al., 2021 b). Although HIV can be sexually acquired and transmitted by PWID and their partner, the preeminent risk of HIV transmission among PWID is through multi-person use of contaminated syringes or needles ( Baggaley et al., 2006 ) ( Boily et al., 2009 ). Given the low HIV incidence among PWID in Montreal ( Leclerc et al., 2021 ), the high ART coverage among PWID living with HIV ( Leclerc et al., 2021 ), the small role of sexual HIV transmission to HCV dynamics among PWID, and the uncertainty in sexual mixing patterns in this population, we did not model sexual transmission of HIV.
We modeled the impact of HIV infection and linkage-to-care on HCV natural history and care: people living with HIV are less likely to spontaneously clear HCV ( Smith et al., 2016 ); those not on ART have higher HCV-related mortality rates ( Hayashi et al., 2014 ); and those on ART have facilitated access to HCV testing and treatment ( Bartlett et al., 2019 ).

Model parametrization and calibration
To inform model parameters, we used data from SurvUDI , the HIV/HCV bio-behavioural surveillance network among active PWID in Quebec ( Leclerc et al., 2021 ) and the Canadian Co-infection Cohort , a prospective study of HIV-HCV co-infected people ( Klein et al., 2010 ). We retrieved complementary information from the literature -primarily from systematic reviews and meta-analyses ( Table 1 ).
We used the Bayesian sampling importance resampling algorithm for model calibration ( Rubin, 1987 ). First, we used Latin hypercube sampling to sample 70,000 parameter sets from our prior distributions The recruitment parameter is estimated from the yearly reduction in the population of active people who inject drugs and the yearly number of deaths. The male to female population ratio is maintained constant across time. ( McKay & Conover, 1979 ). For each parameter set, we ran the model to equilibrium using baseline parameter values. We used the population distribution across compartments at equilibrum to simulate epidemics from 2003 to 2018. In that period, we used the following annual estimates from SurvUDI as calibration targets: HCV seroprevalence, HCV seroincidence, HIV prevalence, HIV incidence, joint prevalence of anti-HCV antibodies and HIV, and ART coverage among those living with HIV. Based on these calibration targets, we estimated the likelihood of each sampled parameter set. From the initial 70,000 parameter sets, we resampled 350 sets with replacement using sampling weights proportional to their likelihood values. We used the resulting parameter sets for our simulations, thereby propagating parameter uncertainty to model outcomes.
As the last calibration data available were in 2018, we assumed constant parameter values from 2018 to March 2020. From March 2020 to May 2021, we modeled COVID-19-related disruptions by reducing access to HCV services. Based on emerging Canadian literature and information provided by local community organizations, we modeled a 50% reduction in HCV testing and treatment rates and a 15% reduction in NSP and OAT coverage ( Canadian Centre on Substance Use & Addiction, 2020 ; Public Health Agency of Canada, 2020 ; Friesen et al., 2021 ;Van Gennip et al., 2021 ). We assumed that pre-COVID-19 intervention levels (i.e., 2018 parameter values) were restored in June 2021.

Intervention scenarios and model outcomes
Between 2022 and 2030, we simulated eight intervention scenarios, including a status quo scenario with unchanged intervention levels ( Table 2 ). In other scenarios, intervention levels increased linearly to reach values provided in Table 2 in 2024, and then remained constant until 2030. We set a limit of 95% for HCV diagnosis and treatment coverage because existing interventions make it difficult to reach higher coverage. To isolate the impact of each intervention, we modeled them separately in the main analyses. We obtained intervention parameter values by triangulating information about current intervention levels in Montreal, Canadian public health guidelines, results from empirical and modeling studies, and consulting HCV care providers ( Canadian Network on Hepatitis C, 2019 ; Cousien et al., 2017Cousien et al., , 2018Lanièce Delaunay et al., 2022 ;Leclerc et al., 2021 ). For each scenario, we generated point estimates using medians and 95% credible intervals (95%CrIs) for all outcomes described in Table 3 .

Sensitivity analyses
We modeled scenarios combining previously described interventions. Due to uncertainty around the magnitude and duration of COVID-19-related disruptions in HCV services, we simulated the eight scenarios described above assuming no COVID-19 disruptions in 2020-2021. We also ran scenario 7 (increased treatment for all PWID; Table 2 ) with the following changes in COVID-19-related parameter values: 1) extending the same level of reduction in access to HCV services as initially modeled for an additional six months (i.e., 50% reduction in HCV testing and treatment rates; 15% reduction in coverage of NSP and OAT, March 2020-November 2021); 2) a smaller reduction in HCV services than initially modeled (40% reduction in testing and treatment rates; 10% reduction in harm reduction coverage); 3) a greater reduction in HCV services (60% reduction in testing and treatment rates; 20% reduction in harm reduction coverage); 4) a reduction in HIV services (50% reduction in HIV testing and treatment rates) in addition to reductions in HCV services and harm reduction.

Ethics
This study was approved by the McGill University Health center Research Ethics Board (REB#: MP-37-2019-4700). Our analyses were con-  b People who report injecting drugs in the past six months.HCV: hepatitis C virus; PWID: people who have injected drugs. ducted using publicly available data sources. Hence, consent was not necessary for this study.

Impact of COVID-19 disruptions
Between March 2020 and May 2021, reductions in access to HCV services could have led to small temporary increases in chronic HCV incidence (from 8.0 to 8.5 per 100 PY) and prevalence (from 30% to 31%).

Interventions among PWID living with HIV
Increasing harm reduction coverage (scenario 2) or HCV testing (scenario 3) among those living with HIV made little additional difference compared to status quo: incidence targets were not met more frequently among all active PWID or those living with HIV, and it was uncertain whether the mortality target could be achieved among all PWID ( Fig. 2 ). Decline in chronic HCV prevalence was also similar to that modeled under the status quo. The increases in proportion of cumulative chronic HCV cases and HCV-related deaths averted (2022)(2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030) were negligible in comparison to status quo ( Table 4 ). Scenarios 2 and 3 led to the same number of treatment initiations among all PWID between 2022 and 2030 as the status quo scenario (treatment initiation ratios of 0.99 (95%CrI: 0.99-1.00) and 1.00 (95%CrI: 1.00-1.00), respectively).
When increasing treatment among PWID living with HIV (scenario 4), the posterior probability of reaching incidence targets was 40% among all active PWID and 60% among those living with HIV ( Fig. 2 ). Under this scenario, mortality targets were achieved in most simulations (94%) among all PWID and all of them among PWID living with HIV. Chronic HCV prevalence decreased by 73% (95%CrI: 65-80) between 2015 and 2030 among active PWID and by 91% (95%CrI: 88-93) among those living with HIV. Approximately one in four cumulative  ( Table 4 ). One in six HCV-related deaths projected under status quo was prevented among all PWID, and over half among PWID living with HIV. Scenario 4 also led to 1.04 (95%CrI: 0.97-1.19) times more HCV treatment initiations among all PWID between 2022 and 2030, as compared to status quo.

Interventions among all PWID
Extending enhanced harm reduction (scenario 5) or testing (scenario 6) to all did not substantially impact chronic HCV incidence, prevalence, or mortality ( Fig. 3 ). Yet, compared to status quo, increasing NSP and OAT coverage prevented 12% of cumulative chronic infections between 2022 and 2030 among all active PWID and 11% among those living with HIV ( Table 4 ). Higher harm reduction coverage also prevented 3% of cumulative HCV-related deaths among PWID and 4% among PWID living with HIV. Scenario 5 led to 0.98 (95%CrI: 0.96-0.99) times less HCV treatment initiations among all PWID between 2022 and 2030. In scenario 6, the ratio of treatment initiations was 1.04 (95%CrI: 1.03-1.04) as compared to the status quo.

Increased treatment among active PWID
With regards to HCV incidence and prevalence, increasing treatment among active PWID only (scenario 8) yielded the same results as scenario 7 (scaled-up treatment for all PWID). This intervention achieved HCV elimination targets for both PWID and PWID living with HIV. Yet, smaller fractions of HCV-related deaths were prevented between 2022 and 2030: 36% among PWID and 39% among PWID living with HIV. Scenario 8 also led to 1.19 (95%CrI: 1.06-1.47) times more HCV treatment initiations among all PWID between 2022 and 2030.

Sensitivity analyses
Jointly modeling interventions did not affect our results: sizeable effects were consistently driven by increased treatment: increases in testing or harm reduction coverage had little impact (Additional file 1; Figs. 2 and 3). Assuming no disruption in HCV services during the COVID-19 pandemic did not affect the qualitative ranking of scenarios (Additional file 1; Figs. 4 and 5). Similarly, varying the magnitude and duration of COVID-19-related reductions in HCV services or reducing access to HIV testing and treatment did not affect the outcomes of our scenario increasing HCV treatment among all PWID (Additional file 1; Fig. 6).

Principal findings
DAA scale-up to levels beyond those reached in 2020 among all PWID diagnosed with HCV, irrespective of HIV status or time since last injection, was key to reaching HCV elimination targets by 2030 in Montreal. Focusing treatment efforts on PWID living with HIV -who can be reached by interventions and have high HCV acquisition and transmission risks-prevented many HCV-related deaths, especially among those living with HIV. This strategy led to substantial progress towards elimination but was insufficient to achieve it. Ex-PWID no longer contribute to onward HCV transmission via injection drug use. Yet, including them in treatment efforts is important to prevent HCV-related deaths. HCV transmission is high in Montreal and modeling COVID-19-related reductions in access to HCV services led to a modest, yet rapid epidemic resurgence among PWID. In addition to scaling-up treatment, it is essential that pre-COVID-19 levels of HCV testing and harm reduction are rapidly restored to ensure elimination by 2030.

Contextualization of results
Harm reduction coverage and HCV testing rates are already relatively high in Montreal. It is therefore not surprising that further increases in coverage had little impact. This is especially true for HCVrelated deaths since our study period (2022)(2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030) is too short to observe the prevention benefits of harm reduction on HCV mortality. Previous modeling studies have also concluded that harm reduction strategies alone were unlikely to eliminate HCV among PWID ( Pitcher et al., 2019 ). Nevertheless, NSP and OAT are cost-effective, prevent multiple drug-related harms, and may be instrumental to sustaining HCV elimination targets post-2030 among PWID ( Lanièce Delaunay, et al., 2021 b;Wilson et al., 2015 ). Harm reduction should remain at the core of elimination strategies for PWID. Bottlenecks can occur at different stages of the HCV care cascade. In settings where few PWID are aware of their HCV infection, increasing testing can reduce HCV prevalence ( Blake & Smith, 2021 ). In settings like Montreal, where most PWID know their HCV status, further reducing time from infection to diagnosis may have little impact on HCV incidence or mortality ( Cousien et al., 2018 ). These differences advocate for tailoring HCV elimination strategies to local challenges.
Evidence from modeling studies suggests that HCV treatment rates below 10 per 100 PY could achieve elimination among PWID in several settings ( Pitcher et al., 2019 ). In Montreal, however, HCV seroincidence and seroprevalence remain high despite relatively high engagement in HCV prevention and care: higher treatment uptake would be necessary to reduce transmission. A modeling study suggests that HCV treatment rates of 200 per 100 PY could be cost-effective among PWID in France, where baseline chronic HCV incidence levels were comparable to those estimated with our model for Montreal ( Cousien et al., 2018 ). Nevertheless, HCV treatment cost may vary widely across countries, and these results may not be generalizable to Canadian settings. In real-world programs like Iceland, treatment rates above 150 per 100 PY have been attained among PWID ( Olafsson et al., 2019 ). We used a treatment rate of 100 per 100 PY as it was deemed achievable in our setting based on existing literature. Nevertheless, extensive public health efforts will be required to massively engage PWID in DAA treatment given current rates are approximately 10-30 per 100 PY. Local empirical data are warranted to determine how to reach and sustain this level of treatment.
Few studies have examined interventions to eliminate HCV among PWID while incorporating HIV transmission dynamics Skaathun et al., 2018 ). Our results suggest that while not sufficient to meet the WHO incidence reduction target among all PWID, increasing HCV treatment among PWID living with HIV could lead to substantial progress toward this goal. Because our model tracks PWID who have ceased to inject, we could additionally show that focusing treatment efforts on PWID living with HIV could achieve the WHO mortality reduction target for all PWID.

Potential implications
Scaling-up DAA uptake among PWID is necessary to eliminate HCV in this population, and this could be achieved by macro-level policies such as funding health services, care models offering alternatives to hospital-based specialized HCV care, interventions facilitating health service utilization by patients and providers, and HCV education ( Ortiz-Paredes et al., 2022 ). HCV treatment-as-prevention has generated optimism that we can reach and sustain HCV elimination as a public health threat among PWID ( Lanièce Delaunay, et al., 2021 b). Yet, the COVID-19 pandemic could delay HCV elimination in many countries ( Blach et al., 2021 ). Disruptions in harm reduction, HCV testing, and HCV treatment services have been observed across Canada ( Public Health Agency of Canada, 2020 ;Van Gennip et al., 2021 ). In Montreal, HCV incidence among active PWID is high but declining. Nevertheless, our results show that reduced access to HCV services could have reversed this trend. In Germany, models also showed that COVID-19-related disruptions in HCV care may cause an increase in HCV incidence and compromise elimination among men who have sex with men living with HIV ( Marquez et al., 2022 ). Staying on track for HCV elimination will require refocusing public health efforts towards HCV prevention and care, which may prove difficult in the context of an aggravated overdose crisis ( Friesen et al., 2021 ).

Strengths and limitations
Our results need to be interpreted considering some limitations. First, potential sources of heterogeneity in HCV risks and unmet prevention needs are not explicitly modeled, which could lead to underestimating efforts required for HCV elimination ( Baral et al., 2019 ). For instance, Indigenous peoples, who are disproportionately affected by HCV in Canada, are not explicitly represented ( Canadian Network on Hepatitis C, 2019 ). A minority of PWID self-identify as Indigenous in Montreal, nevertheless, their specific health needs should be addressed ( Fayed et al., 2018 ;Lanièce Delaunay et al., 2022 ). Second, we assumed that COVID-19-related disruptions in HCV services ceased as of June 2021. Our projections were robust to changes in the scale and duration of these disruptions, yet we may have underestimated the efforts needed to stay on track for HCV elimination by 2030.
Our findings help understand how heterogeneity in HCV acquisition/transmission risk (by HIV status) and needs (between active and ex-PWID) can be factored in the design and implementation of strategies to reduce HCV burden among PWID. These results can inform HCV elimination strategies in settings comparable to Montreal. This is one of few studies to explore the implications of the COVID-19 pandemic on the HCV elimination agenda for priority populations.

Conclusions
In settings with elevated HCV incidence and prevalence, high DAA uptake by all PWID will be required to eliminate HCV as a public health threat among all PWID and PWID living with HIV. Ex-PWID should be included in treatment efforts to avoid preventable HCV-related deaths. The COVID-19 pandemic has impacted HCV prevention and care services for PWID, and concerted efforts to restore and scale-up treatment are urgently needed to ensure that HCV elimination is reached by 2030.

Declarations of Interest
CLD, AG, BL, and CD have no conflict of interest to declare. MK reports grants for investigator-initiated studies from ViiV Healthcare, Abbvie , and Gilead ; research grants from Janssen , and consulting fees from ViiV Healthcare, AbbVie , and Gilead , all outside the submitted work. JC has received institutional funding for investigator-sponsored research from ViiV Healthcare and Gilead Sciences . He has also received remuneration for advisory work and/or travel support from ViiV Healthcare, Gilead Sciences and Merck Canada . NK reports research funding from Gilead Sciences, McGill Interdisciplinary Initiative in Infection and Immunity, Canadian Institutes of Health Research , and Canadian Network on Hepatitis C ; reports advisory fees from Gilead Sciences, ViiV Healthcare, Merck , and AbbVie ; and reports speaker fees from Gilead Sciences, Abb-Vie , and Merck , all outside of the submitted work. MM-G reports grants from the Canadian Institutes of Health Research and the Canadian Foundation for AIDS Research , and contractual arrangements from both the World Health Organization and the Joint United Nations Programme on HIV/AIDS (UNAIDS), all outside of the submitted work.

Acknowledgments
This work was supported by the investigator-initiated study funded through the Gilead Sciences LEGA-C Program [grant number IN-US-334-4492 , June 1st, 2018]. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. CLD received a PhD trainee fellowship from the Canadian Network on Hepatitis C. The Canadian Network on Hepatitis C is funded by a joint initiative of the Canadian Institutes of Health Research ( NHC-142832 ) and the Public Health Agency of Canada. CLD also received a doctoral training award from the Fonds de recherche du Québec -Santé. MK is supported by a Tier I Canada Research Chair . NK is supported by a career award from the Fonds de Recherche Québec-Santé (FRQ-S; junior 1). CMD is supported by the Fonds de recherche du Québec -Santé (FRQS). MM-G's research program is supported by a Tier II Canada Research Chair in Population Health Modeling .

Ethical approval
This study was approved by the McGill University Health center Research Ethics Board (REB#: MP-37-2019-4700). Our analyses were conducted using publicly available data sources. Hence, consent was not necessary for this study.

Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.drugpo.2023.104026 .