Modalities of HCV transmission

Blood-borne infection is the main route of HCV transmission. This occurs when the blood of an infected person comes in direct contact with the blood of a non-infected person. Transmission between sexual partners is very low and favoured by HIV co-infection. Injecting drug use is currently the main source of infection in France. Needles and syringes have the greatest potential for HCV transmission because of their direct contact with blood during the injecting procedure. However, HCV-RNA has also been found in 25–40% of filters, cups and rinse-water samples [Reference Crofts2]. An experimental study showed that HCV maintained outside the human organism in ambient air preserves its infective power for at least 16 h [Reference Kamili3].

Sharing needles and drug paraphernalia between injectors is the dominant mode of transmission in injecting drug users. Unlike HIV transmission, sharing material used to prepare the injection, and in particular the filter, without sharing needles, is a source of HCV transmission between injecting drug users [Reference Hagan4–Reference De6]. Of all subjects who had a seroconversion, the proportion which could be attributed to recipient and/or filter sharing was 13% [Reference Hagan4]. In recent years, sniffing has been considered as a non-negligible risk factor for HCV infection in subjects not using the intravenous route [Reference Tortu7, Reference Martinez and Talal8].

About 75% of infected subjects remain chronic carriers and thus susceptible to transmitting the virus [Reference Micallef, Kaldor and Dore9]. Furthermore, spontaneous or treatment-induced cure of hepatitis C does not lead to sustained immunity. HCV re-infections have been reported in recipients of numerous blood transfusions [Reference Lai10] and in drug users [Reference Bowden11, Reference Micallef12]. Although the issue is controversial [Reference Van De Laar13], patients who have eliminated the virus spontaneously apparently have a lower risk of developing chronic disease after a new infection than those who are HCV naive [Reference Grebely14, Reference Osburn15].

Diagnosis of HCV infection

The diagnosis of HCV infection requires the demonstration of anti-HCV antibodies in the patient's serum. Despite the apparent simplicity of this diagnostic procedure, its implementation in epidemiological studies including drug users is a complex problem for at least three reasons.

First, drawing a venous blood sample is often difficult in drug users because of an altered venous network coupled with non-cooperation. This can bias epidemiological studies by excluding from screening a significant proportion of subjects at risk. Alternative methods such as screening via volunteer reports [Reference Six, Hamers and Brunet16], or using saliva [Reference Lucidarme5, Reference Bello17] or capillary samples [Reference Jauffret-Roustide18] have been applied in many studies to avoid this problem. However, the wide range of assay methods compromises comparisons of HCV prevalence reported in different studies.

Second, screening for HCV infection was not possible before the first-generation of anti-HCV antibodies became available in 1990. Consequently, the incidence before this date has to be reconstructed using mathematical models which take into account epidemiological data available at the time of the study and knowledge about the natural history of the disease.

Third, re-infection can be missed if the diagnosis of spontaneous cure is not established after the first infection [Reference Amin19].

Modalities for an epidemiological evaluation of HCV transmission

The fundamental knowledge which has emanated from epidemiological studies measuring prevalence, incidence and force of infection, particularly regarding the specific aspects of the problem of HCV infection in drug users, should be underscored.

Prevalence counts all cases at a given moment. Prevalence varies when cure rate and deaths do not balance out disease incidence. A decrease in the prevalence of HIV infections in drug users during the 1990s paralleled a fall in the incidence of infection and a rise in AIDS-related deaths. Regarding drug use, initiation of new subjects to injecting practices or a positive migration balance of drug users can lead to an increase in their number. A fall in the number of drug users can result from discontinuation of drug use or injecting practices, or also from user death. In a retrospective cohort of 23 000 persons apprehended for heroin, cocaine or crack abuse in the period 1992–2001, the uncorrected death rate was 7·3 deaths/1000 person-years [Reference Lopez, Martineau and Palle20]. Inversely, if a disease is non-fatal, prevalence generally remains stable for decades even if the incidence has become low or even zero.

Incidence rate is used in epidemiology to describe the dynamics of a given disease in a given population. Variations in incidence in a sub-population do not necessarily reflect variations of incidence in the general population if the incidence in the sub-population is different from that in the general population. The methodological difficulty specific to the study of HCV infection in drug users results from the fact that the endemic situation of the sub-population of drug users has evolved over the last 40 years.

Incidence can be measured prospectively during the course of a longitudinal follow-up study of a cohort of drug users who were seronegative at inclusion. Setting up this type of study is rather difficult because a large number of subjects can be expected to be lost to follow-up at the end of the observation period. Moreover, different biases occur, especially concerning sample representativeness, and also concerning follow-up and adherence, potentially compromising results. Obtaining a representative sample is the most difficult element because the populations at greatest risk of viral infection are also the most difficult to include in epidemiological surveys because of their substance dependence and marginal social status.

Incidence can also be measured indirectly using retrospective studies by dividing the prevalence by the number of years since the first risk event, i.e. the first injection, and the date of the first serology test. This type of measure can be proposed only if the delay is short enough to correspond to a stable incidence over time. As a general rule, sub-populations of users aged <25 years and injectors for <2 years are distinguished in the literature. These two parameters define the category of ‘new injectors’ which can be used to evaluate the renewal of the injecting drug user population.

Force of infection is a measure of the risk of becoming infected, defined as the per capita rate at which susceptible individuals acquire infection [Reference Mathei21]. Force of infection can be estimated from prevalence data as a function of time of exposure to risk. Regarding drug use, the time of exposure is defined as the time from the first injection to the moment of evaluation. The force of HCV infection in injecting drug users is not linear over time. It is very high during the first 2 or 3 years of injecting practices and decreases very rapidly thereafter. This decrease can be explained, at least in part, by changing risk practices, but especially by a saturation phenomenon: the higher the proportion of infected subjects, the lower the proportion of subjects susceptible to becoming infected. Thus, the concept of force of infection is in contradiction to the idea by which the difference observed between the patterns of the HCV and HIV epidemics would be related to the higher prevalence of HCV infection in drug users. It is the change to injecting drug use that is the key element in the HCV epidemic.

Methodology for the present analysis

Two prevalence studies on HCV infection in the general population conducted in France by the InVS in 1994 [Reference Dubois22] and in 2004 [Reference Meffre23] constituted the touchstone data sources for the present analysis. Despite the fact that general population studies can underestimate categories of statistically marginal subjects, the observations made during these two studies lead to hypotheses concerning the dynamics of HCV infection in drug users in France. This approach has the advantage of focusing on two large-scale and methodologically sound studies of representative populations.

The hypotheses constructed on the bases of the general population studies were then examined in light of the multicentre data available for drug-user populations concerning HCV prevalence [Reference Six, Hamers and Brunet16, Reference Jauffret-Roustide18, Reference Valenciano, Emmanuelli and Lert24, Reference Palle and Vaissade25] and incidence [Reference Lucidarme5] as well as modalization work [Reference Mathei21, Reference Downs26, Reference Deuffic27].

Finally, single-centre studies [Reference Bello17, Reference Di Nino28, Reference Lucidarme29] provided complementary insight for understanding the dynamics of HCV infection.

Evolution of the number of HCV-positive drug users from 1994 to 2004

In 2004, the number of persons aged 18–80 years living in metropolitan France who were seropositive for anti-HCV antibodies and who had injected or inhaled illicit drugs at least once in their life was estimated at about 105 000 (95% CI 40 000–154 000) including about 7700±800 who had never injected (sniffers only) [Reference Meffre36].

The Coquelicot multicentre survey conducted in 2004 among drug users (injectors or sniffers at least once in their life) took blood samples to measure the prevalence of viral infections: the HCV prevalence was 59·8% compared to 10·8% for HIV, giving a 5·54-fold difference [Reference Jauffret-Roustide18]. In 2000, the number of HIV-positive subjects among drug users was evaluated at 20 200 (95% CI 14 300–29 100) [Reference Desenclos, Costagliola and Commenges37]. If the 20 200 figure is multiplied by 5·54, the estimated number of HCV-positive drug users reaches 112 000 (95% CI 79 000–160 000). Despite the potential bias of this data source, it can be estimated that the number of HCV-positive drug users in 2004 ranged from 105 000 to 112 000.

The difference in the number of HCV-positive drug users estimated for 2004 (95% CI 105 000–112 000) and 1994 (92 000±14 000) probably corresponds to the in-out balance (new infections minus deaths). The number of deaths in problem drug users can be extrapolated by multiplying the total drug-user population in 1999 (95% CI 142 000–176 000) [38], by the mortality rate of 7·3 deaths/1000 person-years [Reference Lopez, Martineau and Palle20], giving about 1000–1300 deaths annually. Considering that the prevalence of HCV infection in problem drug users was 60% [Reference Jauffret-Roustide18], an estimated 600–800 HCV-positive drug users died annually from 1994 to 2004, or about 6000–8000 deaths for the decade considered. Consequently, the number of incident cases of hepatitis C in HCV-positive drug users between 1994 and 2004 would range from 5000 to 42 000 (taking into account the confidence interval), for an estimated average of 500–4200 new cases per year. An average of 500–4200 new cases of HCV infection per year in drug users from 1994 to 2004 does not mean that the incidence was stable during this period. Since the early 1990s, the trend has been towards a fall in HCV incidence. Consequently, the annual incidence of HCV in 1994 was probably in the neighbourhood of 4200 new cases per year while in 2004 it was closer to the lower figure than the higher one.

Evolution of the number of incident cases of HCV infection from 1999 to 2004

A longitudinal study of HCV infection in the drug-user population was conducted in the North and East of France from 1999 to 2001. The measured incidence was 9/100 person-years [Reference Lucidarme5] and 11/100 person-years in active injectors, i.e. persons who had injected drugs at least once during the 6 months preceding inclusion. Extrapolation of these results to the number of presumed HCV-negative active drug users estimated at that period gave about 3500 (95% CI 2700–4400) new cases of infection in drug users in France annually [Reference Emmanuelli, Jauffret-Roustide and Barin39]. This estimate would be valid for 1999 but not the entire decade since the incidence varied from one year to another.

The survey of HCV prevalence in 2004 [Reference Meffre36] estimated that 7977 (95% CI 303–15 562) subjects aged 18–29 years were HCV-positive. Similarly, in the Coquelicot survey conducted in the same year [Reference Jauffret-Roustide40], subjects born after 1974 and thus aged 18–29 years accounted for only 10·4% of the HCV-positive population, i.e. 11 648 (95% CI 8200–16 600) subjects considering the overall estimate of 112 000 (95% CI 79 000–160 000) persons established earlier from the same study. The range obtained from these two estimates, 7977–11 648, gives an estimate of the number of subjects infected by the virus from 1999 to 2004, accepting the hypothesis that the number of subjects born before 1975 and infected from 1999 to 2004 would be marginal.

Thus, considering that from 1999 to 2004 about 8000 to nearly 12 000 new HCV infections would have occurred, including 2700–4400 for 1999 alone, it can be reasonably estimated that about 1000 (95% CI 500–2000) drug use-related infections occurred in 2004. This very significant decline in HCV incidence in drug users could be nonlinear from 1994 to 2004, with a more marked fall starting in 2000 after the risk reduction policy reached maximum coverage [Reference Emmanuelli and Desenclos41] and is illustrated by the very clear fall in HCV prevalence in the general population for the 20–29 years age group from 1994 to 2004 [Reference Delarocque-Astagneau42].

Evolution of HCV incidence from 2004 to 2010

There was no direct measurement of HCV incidence in drug users in France from 2004 to 2010. However, several indirect evaluations of incidence in new injectors were made using prevalence data from injectors aged <30 years or <25 years, depending on the study. Several surveys conducted in users attending care centres, general medicine clinics [Reference Jauffret-Roustide18, Reference Palle and Vaissade25, Reference Di Nino28] or first-line structures (Centre d'accueil et d'accompagnement à la réduction des risques pour usages de drogues; CAARUD) [Reference Cadet-Tairou43] appear to indicate that the decline in HCV incidence continued.

The HCV prevalence in drug users aged <30 years was 44% in the first Coquelicot study in 2002 in Marseille [Reference Jauffret-Roustide44], 29% in the second study in 2004 in five French large cities [Reference Jauffret-Roustide18], and only 7% in 2007 in the microstructure medical network in Alsace [Reference Di Nino28]. The data from these studies must, however, be interpreted with caution because of the wide geographic and methodological differences of the data sources.

The prevalence figures for positive HCV serology reported by injecting drug users aged <25 years attending care centres with housing accommodation in 1998 [Reference Six, Hamers and Brunet16] and first-line structures in 2006 [Reference Cadet-Tairou43] were 40·8% and 12·2%, respectively. In an exhaustive survey conducted during the same week in 2006 and in 2008, users aged <25 years attending first-line structures (CAARUD) who had already used the intravenous route reported serology results showing a decline in HCV prevalence from 22·5% to 14·3% [Reference Cadet-Tairou43].

Thus the incidence of HCV infection in drug users appears to have declined very significantly over the last 20 years to the point where the current estimates are 10–15% of those established for 1990. However, to this estimate must be added HCV-positive subjects issuing from migrant populations, often from Eastern Europe or Caucasia, whose number is extremely difficult to determine due to the lack of available data.

These findings are quite similar to those reported for a cohort of drug users in Amsterdam where the HCV incidence would have fallen from 27·5% around the end of the 1980s to 2% in 2005 [Reference Van Den Berg45]. In The Netherlands, a policy of risk reduction had been established early in the 1980s.

Decline in use of intravenous injections

The decline in needle sharing may not have been sufficient to alter the course of the epidemic because unlike HIV, HCV can also be transmitted via the material used to prepare the injection. A change in habits in favour of the inhalation route with lesser use of the intravenous route in new users could have been a determining factor. As early as 2003, the TREND system had data to put forward the hypothesis of a decrease in the proportion of injecting users among problem users, particularly heroin addicts [Reference Bello, Cadet-Tairou and Halfen46]. For cocaine, injection is the second most common route of administration, after sniffing, illustrating the fact that the younger generation has preferred, as it has for heroin, inhalation over intravenous injection. Similarly, in users attending first-line structures (CAARUD) the proportion having never used intravenous injection has tended to increase [Reference Cadet-Tairou43]. It can also be noted that injection in a festive environment remains a marginal phenomenon.

The lesser use of the injection route of administration in new users might correspond to later initiation to injection practices in the course of drug use. Evidence from most of the epidemiological studies does not appear to be in favour of this hypothesis since the median age at first injection would be around 20 years [Reference Jauffret-Roustide18, Reference Lucidarme29].

One consequence of lesser use of the intravenous route is a limited renewal of the user population in drug users attending care centres where there has been a very clear decline in the proportion of new injectors. In 1993 and 1998, the proportions of injectors aged <25 years attending care centres with housing accommodation (CSST/CSAPA) were respectively 24% and 16% [Reference Six, Hamers and Brunet16]. In the Coquelicot study of 2004, 8% of subjects had injected or sniffed at least once and only 3% of injectors were aged <25 years [Reference Jauffret-Roustide40, 47]. This phenomenon is not limited to France. In Switzerland, the proportion of drug users who began injecting <2 years before the study fell from 18·7% in 1993 to 3·3% in 2006. An older attending population, with a mean age of 26 years in 1993 and 36 years in 2006, was a consequence of this phenomenon [Reference Dubois-Arber48]. This limited renewal of the drug-user population attending care centres explains a large part of the stability over time of the HCV prevalence in most cross-sectional studies conducted from 1991 to 2004: 50–80% [Reference Bello17, Reference Jauffret-Roustide18, Reference Valenciano, Emmanuelli and Lert24, Reference Lucidarme29, Reference Schmitt, Bertel and Jacob49].

Although the risk of infection is lower for inhalation than for intravenous injection, the proportion of new drug users contaminated by the nasal route appears to be increasing, becoming perceptible [Reference Aaron50].

Decreased force of infection

In intravenous injectors, the second factor would be a decrease in the force of infection with later consequences and less massive HCV infection than in the early 1990s, even if the first years of drug use by intravenous injection remains a high risk period for HCV infection. This phenomenon would result from less frequent injections and better compliance with the rules of asepsis. In 2008, only 20% of users attending first-line structures (CAARUD) who had consumed heroin within the preceding month had consumed daily [Reference Toufik, Escots and Cadet-Tairou33]. Thus the common habit of several injections per day before the advent of substitution treatment was less common, leading to an inevitable drop in injection frequency and thus in the associated risks.

In an incidence study on HCV infection conducted in the North and East of France in 1999, the annual HCV incidence was 17·5% in drug users of <2 years and 6% in users of >2 years [Reference Lucidarme5]. In the 2004 Coquelicot study, the force of infection remained high. Despite a smaller number of subjects concerned, the HCV prevalence in drug injectors aged <25 years was 32·1% [47]. Thus in recent years the cumulative incidence of HCV infection after 5 years of injecting practices would be to the order of 40–50% compared with 90% in the early 1990s [Reference Lucidarme29].

Better access to hepatitis C treatment for drug users

France is the European country which has done the most to provide care for hepatitis C patients [Reference Lettmeier51]. At the end of 2005, an antiviral treatment had been delivered for 16% of the prevalent cases of HCV infection. This diffusion of treatment would be the source of a considerable decline in the proportion of HCV-seropositive subjects with viraemia which was 81% in 1994 [Reference Dubois22] and 57% in 2004 in the 20–60 years age group [Reference Delarocque-Astagneau42]. In drug users, treatment cannot be dissociated from prevention and screening for new infection. Access to treatment has also been the object of much attention. In 2008, 70·5% of drug users attending first-line structures (CAARUD) who were HCV-positive consulted a doctor for this condition and more than one-quarter (28%) were taking treatment, i.e. 5·5 percentage points above the level observed in 2006 [Reference Bello, Cadet-Tairou and Halfen46]. The treatment of drug users is favoured by the presence of a favourable genotype observed in about half of the cases. The results obtained in terms of response to treatment, adverse effects, and premature discontinuation of treatment are globally comparable to those observed in non-drug-user patients [Reference Bruggmann52, Reference Chossegros53]. In this context, multidisciplinary management of addiction-related problems, hepatitis C, and frequently associated psychiatric comorbidity is crucial [Reference Moussalli54]. The proportion of re-infections after treatment to cure is not >10% [Reference Backmund55–Reference Dalgard57] so that there is no reason to proscribe treatment in drug users.