Hepatitis C diagnosis and treatment, impact on engagement and behaviour of people who inject drugs, a service evaluation, the hooked C project

There is emerging evidence that Hepatitis C (HCV) treatment engagement is associated with change in drug behaviours and reduced drug‐related death rates among people who inject drugs (PWID). The project aims to investigate whether HCV diagnosis and treatment engagement reduces all‐cause mortality and drug‐related death, and whether any effect is dependent on treatment regimen and intensity of engagement with staff. Case‐control studies comparing: PWID with active HCV infection (PCR positive) to PWID HCV infected but spontaneously resolved (PCR negative); PCR‐positive patients who engaged with treatment services to nonengagers; and patients who received interferon vs direct‐acting antiviral (DAA) based treatment. No differences in risk of all‐cause mortality or drug‐related death between PCR‐negative controls and PCR‐positive cases were detected. The odds of all‐cause mortality was 12.2 times higher in nonengaging persons compared to treatment engaging cases (aOR 12.15, 95% CI 7.03‐20.99, P < .001). The odds of a drug‐related death were 5.5 times higher in nonengaging persons compared with treatment engaging cases (aOR 5.52, 95% CI 2.67‐ 11.44, P < .001). No differences in risk of all‐cause mortality or drug‐related death between interferon‐treated cases and DAA‐treated controls were detected. HCV treatment engagement is significantly protective against all‐cause mortality and drug‐related death. This engagement effect is independent of treatment regimen, with the introduction of DAA therapies not increasing risk of drug‐related death, suggesting intensity of HCV therapy provider interaction is not an important factor.


| INTRODUC TI ON
Hepatitis C (HCV) is a blood-borne virus and affects up to 1% of the Scottish Population. 1 Around 90% of those infected with HCV acquire the virus through injecting drug use behaviour. 2 HCV related liver disease is a primary contributor to morbidity and mortality among people who inject drugs (PWID). 3 HCV is preventable, treatable and curable, with research supporting the treatment of active injecting drug users for Hepatitis C. 4 The efficacy of pan-genotypic direct-acting antivirals (DAA) provides an excellent opportunity to scale up HCV diagnosis and treatment, ultimately achieving the WHO target of HCV elimination by 2030. 5,6 There is evidence that HCV care engagement is associated with change in behaviours among PWID. Studies have demonstrated the positive impact of HCV status notification on reduction in injecting behaviour among PWID. 7,8 Furthermore, a systematic review highlighted the positive impact of HCV treatment on patients' injecting and sharing behaviour. 9 The causes of death among PWID are strongly associated with active drug use. 10 Scotland has observed a twofold increase in drug-related deaths between 2008 and 2018, with Tayside experiencing the highest number of drug deaths ever recorded in the region in 2018. 11,12 It is vital that informed action is urgently undertaken to reverse this trend.
The introduction of Multidisciplinary Managed Care Networks (MCN) in HCV treatment has increased access to services and reduced all-cause mortality. 13 The associated improvement in access into care and HCV treatment may have led to a greater degree of engagement with health services and may have had a stabilizing effect on drug using behaviour. However, there is concern around the potential impact of reduction in intensity of staff contact when transitioning from the interferon era to the DAA era of treatment.
Interferon based treatment required a greater intensity of staff to patient engagement due to adverse side effects and long treatment duration. Contrastingly, DAA based treatment has minimal side effects and higher cure rates (in excess of 95%). 14 Thus, treatment pathways are streamlined and arguably provide less opportunity for patients to develop a therapeutic relationship with healthcare professionals involved in their care, and therefore reduced opportunities to facilitate change in patients' drug use behaviour, and lower risk of mortality.
The aims were to investigate whether HCV diagnosis and engagement in treatment services reduced all-cause mortality and drug-related death, and whether any effect was dependent on treatment regimen or intensity of engagement with staff. A series of retrospective case-control studies were carried out. Initially, comparing PWID with active HCV infection (PCR positive) vs PWID who were HCV infected but cured spontaneously (PCR negative), to elucidate whether knowledge of HCV infection status impacted risk of mortality. Secondly, comparing PCR-positive patients who engaged vs did not engage with treatment services to assess if outcomes were dependent on engagement. Finally, comparing interferon treated patients vs DAA-treated patients, exploring the effect of intensity of HCV therapy provider interaction on outcomes.

| Data sources and data linkage
The main data source utilised was the Tayside Hepatitis C Clinical Database which records patients tested for Hepatitis C, awaiting treatment, on treatment, cured and re-infected in Tayside

| Identification of selected cohort
From the Tayside clinical database, a cohort of individuals was identified whose risk factor for HCV was intravenous drug use. Therefore, the definition of PWID in our study is people who have "ever injected" drugs, with no differentiation between recent/active and former PWID. Individuals with other risk factors, such as transfusion of blood products or maternal transmission were excluded as we were specifically investigating the impact of HCV treatment on the behaviour of PWID. Individuals with non-Tayside postcodes were excluded as drug-related death outcomes would not be registered for non-Tayside individuals on the Tayside Drug Deaths Database.
Individuals co-infected with other blood-borne viruses were excluded from the selected cohort as these individuals would have differing mortality rates and treatment experiences to those only infected with HCV. Individuals who were tested or initiated on treatment before January 2008 were excluded as the MCN for HCV care in Tayside was introduced in 2008 and this substantially changed the care pathways. Lastly, individuals who were tested or initiated treatment after November 2017 were excluded to allow for a minimum of one year of follow-up.
For each analysis, cases and controls were defined differently, although derived from the same cohort previously described. For analysis 1, all individuals who tested HCV antibody positive were identified. Cases were defined as PWID with active HCV infection (PCR positive), and controls were defined as PWID who were HCV infected but cured spontaneously (PCR negative). For analysis 2, all individuals who tested HCV PCR positive were identified. Cases were defined as PCR-positive patients who engaged with treatment services, and controls were defined as PCR-positive patients who did not engage with treatment services. For analysis 3, all individuals who were PCR positive and engaged with treatment were identified.
Cases were defined as pegylated interferon alpha treated patients, and controls were defined as DAA-treated patients. For all analyses, each case was matched with one control by age group (20-35, 36-51, 52-67, 68-83, 84+) and sex. Controls from the respective categories were randomly selected using an online random number generator.

| Definition of drug-related death
The definition of a drug-related death is a death where the underlying cause is as follows: drug abuse or drug dependence; or drug poisoning (intentional or accidental) that involves any substance controlled under the Misuse of Drugs Act 1971. 11 The National Records of Scotland uses the ICD 10 classification system to identify cases of drug-related death once a death certificate has been issued.

| Treatment engagement
"Treatment engagement" was defined as engaging with healthcare professionals and commencing treatment. All patients who commenced treatment were classified as "treatment engagers", irrespective of how many days/weeks of treatment they completed, whether they completed their entire course of treatment or not, and the outcome of their treatment, for example if a sustained viral response (SVR) was achieved. Correspondingly, patients who did not commence treatment were classified as "treatment non-engagers".

| Opioid substitution therapy (OST)
Data were collected on individuals' OST status around the time of testing or treatment. Specifically, for analysis 1 (PCR negative vs PCR positive) and analysis 2, (treatment engagers vs nonengagers), data were collected on whether individuals were on OST at the time of HCV RNA PCR testing, ±6 months. For analysis 3 (interferon vs DAA-treated patients), data were collected on whether individuals were on OST at the time of treatment commencement, ±6 months.

| Cirrhosis
Data were collected on individuals' cirrhosis status. Individuals were classified as being cirrhotic if their liver stiffness (FibroScan) score was 12.5 kPa or above, or their FIB-4 score was 3.25 or above. 15

| SVR
Data were collected on individuals' sustained virologic response (SVR) status. SVR was defined as absence of detectable HCV RNA at 24 weeks after cessation of treatment. Kaplan-Meier survival analysis was performed to investigate differences in the rates of all-cause mortality and drug-related deaths between cases and controls. Comparison of survival curves was performed using log-rank tests. Binary logistic regressions were used to compare the odds of all-cause mortality and dying of a drug-related death among cases with those among controls. We estimated odds ratios (ORs) and 95% confidence intervals for all comparisons and adjusted all models for the matching variables; age and sex. A number of other covariates were also included in certain models; SVR, OST and cirrhosis.

| Analysis 1 -PCR positive vs negative
A total of 3431 individuals who tested HCV antibody positive were identified. Of these, 386 PCR-negative controls and 918 PCRpositive cases met the inclusion criteria (see Figure 1) and were randomly matched by age group and sex, leading to 386 PCR-negative controls and 386 PCR-positive cases included in the study.
Baseline characteristics of cases and controls are presented in For all-cause mortality, the survival distributions for the two groups were not significantly different, χ 2 (2) = 0.425, P =.515. No difference in risk of all-cause mortality between PCR-negative controls and PCR-positive cases was detected (aOR 1.18, 95% CI 0.80-1.73, P = .40), after adjustment for age and sex (see Table 2).
For drug-related deaths, the survival distributions for the two groups were not significantly different, χ 2 (2) = 0.291, P =.590. No difference in risk of drug-related death between PCR-negative controls and PCR-positive cases was detected (aOR 1.19, 95% CI 0.71-2.00, P = .512), after adjustment for age and sex (see Table 3).

| Analysis 2 -PCR-positive treatment engagers vs PCR-positive treatment nonengagers
A total of 2499 individuals who tested HCV PCR positive were identified. Of these, 267 treatment nonengaging controls and 650 treatment engaging cases met the inclusion criteria (see Appendix S1) and were randomly matched by age group and sex, leading to 263 treatment nonengaging controls and 263 treatment engaging cases included in the study (successful matching was not possible for four controls).
Baseline characteristics of cases and controls are presented in For all-cause mortality, the survival distributions for the two groups were significantly different, with nonengaging controls at a significantly higher risk of all-cause mortality, χ 2 (2) = 91.395, P = <.001 (see Figure 2). The odds of all-cause mortality were 12.2 times higher among treatment nonengaging controls, (aOR 12.15, 95% CI 7.03-20.99, P < .001) compared with treatment engaging cases, after adjustment for age, sex and OST (see Table 2).
For drug-related deaths, the survival distributions for the two groups were significantly different, with nonengaging controls at a significantly higher risk of drug-related death, χ 2 (2) = 32.364, P = <.001 (see Figure 3). The odds of a drug-related death were 5.5 times higher among treatment nonengaging controls, (aOR 5.52, 95% CI 2.67-11.44, P < .001) compared to treatment engaging cases, after adjustment for age, sex and OST (see Table 3).

| Analysis 3 -Interferon treated vs DAA treated
A total of 1664 PCR-positive individuals who engaged with treatment were identified. Of these, 380 interferon treated cases and 270 directing acting antiviral-treated controls met the inclusion criteria (see Appendix S1) and were randomly matched by age group   Table 2). Note, 28 individuals were omitted from the regression analysis due to missing data on cirrhosis; eight controls and 20 cases.
For drug-related deaths, the survival distributions for the two groups were not significantly different, χ 2 (2) = 0.281, P =.596. No difference in risk of drug-related death between DAA-treated controls and interferon-treated cases was detected (aOR 2.06, 95% CI 0.80-5.23, P = .13), after adjustment for age, sex, SVR, OST and cirrhosis (see Table 3). In the second analysis, we looked at PCR-positive patients who engaged vs did not engage with treatment services to explore whether self-selecting engagement behaviour accounts for the perceived difference in mortality. Our findings provide evidence that HCV treatment engagement is a significant protective factor against both all-cause mortality and drug-related death among PWID, with nonengaging PCR-positive individuals having 12 times higher odds of all-cause mortality and 5 times higher odds of drug-related death, in comparison to PCR-positive treatment engaging persons. These findings confirm previous research that engaging in Hepatitis C treatment leads to a reduction in all-cause mortality. 13 It is important to note that our cohort was selected from a population that has a high testing and diagnosis rate, nearly reaching WHO 2030 targets This finding also has significant implications for addressing ongoing concern around the change in intensity of staff contact when transitioning from the interferon era to the DAA era of treatment.

| D ISCUSS I ON
In addition, it is important to consider that DAA-treated patients are arguably more unstable than interferon treated patients as many would have been deemed to be unsuitable for interferon therapy due to associated adverse side effects. Indeed, it has been hypothesised that DAA-treated patients might have worse outcomes than interferon treated patients given the less intensive support during therapy. Thus, the fact that we observed no difference in risk of all-cause mortality or drug-related deaths between the two groups is evidence that intensity of staff engagement is not an important protective factor. Consequently, current treatment practice does not need to implement an increase in intensity of staff contact.

| Limitations
The predominant limitation of the current study was the retrospective study design, with substantial limitations in quality of available data. Data on a number of meaningful variables were not available.
For instance, OST data were not attainable for PCR-negative individuals and therefore could not be included as a predictor variable in the regression model in analysis 1. Moreover, available OST data indicated whether individuals were on OST at the time of diagnosis, but not whether they were on OST at the time of their death, which could have given more insights. Data on history of nonfatal overdoses would also have been advantageous, as previous research has demonstrated that nonfatal overdose is classified as a risk factor for ensuing fatal overdose in PWID. 18 Other unattainable data which could have been beneficial were injecting history, injecting status, change in injecting behaviours and other significant comorbidities.
Furthermore, data on unmeasured potential confounding variables which may explain the association between engagement and decreased risk of mortality is lacking, with further research needed to elucidate the complex reasons that lead to nonengagement.
Another limitation to the current study is the lack of differentiation of individuals in analysis 2 (treatment engagers vs nonengagers). Specifically, engagers were not differentiated by a more specific measurement of treatment engagement, for example how many weeks of treatment they completed and/or whether they completed their full course of treatment. Equivalently, nonengagers were not differentiated by the reason for their nonengagement. For instance, a minority of patients may have not started treatment due to concerns around treatment contra-indications or age. This is particularly relevant for patients treated in the interferon treatment era due to higher incidence of associated adverse side effects compared with DAA based treatment regimens. Arguably such differentiation may provide greater insight into the impact of treatment engagement on subsequent risk of death, and whether, for example, completion of treatment potentiates the engagement effect.

| CON CLUS IONS
In conclusion, a series of case-control studies were conducted to investigate the impact of HCV diagnosis and engagement in treatment services on risk of all-cause mortality and drug-related death among PWID. No difference in risk of all-cause mortality or drug-related death was observed between PWID with active HCV infection (PCR positive) and HCV infected but cured spontaneously (PCR negative). HCV treatment engagement is significantly protective against all-cause mortality and drug-related death, with nonengaging PCR-positive individuals 12 times higher odds of all-cause mortality and five times higher odds of drug-related death, in comparison to PCR-positive treatment engaging persons. This engagement effect is independent of treatment regimen, with no difference in risk of all-cause mortality or drug-related death between interferon treated patients and DAA-treated patients, suggesting intensity of engagement with staff is not an important factor.
These findings provide further evidence of the importance of HCV diagnosis and treatment engagement among PWID, reducing their risk of mortality, beyond liver-related outcomes.

ACK N OWLED G EM ENTS
We would like to thank NHS Tayside Sexual Health & Blood Borne Virus Managed Care Network (MCN) for funding this project.

CO N FLI C T O F I NTE R E S T
No authors declare any conflicts of interest.