Section 4: Modelling the Incidence and Prevalence of Hepatitis C Infection and its Sequelae in Canada, 2007 – Discussion

Discussion

We carried out a modelling study to characterize the epidemiology of hepatitis C infection in Canada in 2007. Specifically, we wished to estimate the incidence and prevalence of hepatitis C infection in Canada from 1967 to 2027 stratified on gender, exposure category and place of birth. We also modelled the prevalence and incidence of serious sequelae of hepatitis C infection for this 60-year period. To accomplish these objectives, we adapted and refined the actuarial model developed for estimating HCV infection in Canada in 2002, which was completed in 2003.

We concluded that, as of December 2007, 242,521 persons in Canada were infected with hepatitis C. HCV infections in Ontario represent approximately 42% of all infections in Canada. The highest HCV prevalence rates were observed in the Yukon, Northwest Territories and British Columbia. With respect to HCV incidence, an estimated 7,945 persons were newly infected in Canada in 2007.

Of the 242,521 HCV-infected persons in Canada, 52,512 (22%) were active IDUs and 87,452 (36%) past IDUs, 25,905 (11%) had been infected through blood transfusion, 861 (0.36%) were hemophilia patients and 75,790 (31%) were infected through other modes of transmission, including sexual, occupational, nosocomial and vertical transmission. Sixty-one percent of HCV-infected persons were male.

We also estimated the prevalence of infection by sex and age group. HCV prevalence was 64% greater in males than in females (0.95% versus to 0.61%). Not surprisingly, HCV prevalence generally, though not always, increased with increasing age. On the other hand, HCV incidence in 2007 was greatest among persons aged 20 to 39 years, with the highest incidence of 0.110% among those 25 to 29 years old. HCV incidence among men overall was 0.034%, almost twice that among females with 0.018%.

We found that HCV prevalence varied considerably by province and territory. The largest number of HCV-infected persons (102,858, or 42%) were residents of Ontario, with the next highest being British Columbia (53,254 or 22%) and Quebec (37,505, or 16%). The highest HCV prevalence rates were observed in the Yukon (3.87%), Northwest Territories (1.83%) and British Columbia (1.21%).

The estimated number of HCV-infected persons in Canada in 2007 was slightly lower than the 251,000 HCV-infected persons we estimated for 2002. The difference is accounted for by more precise estimates of mortality in the five exposure categories examined as well as incorporation of the natural loss of HCV virus and antibody that is not negligible.

The impact of the sequelae of hepatitis C infection on the health of persons in Canada appears to be considerable. In 2007, 15,814 persons were living with cirrhosis, 5,495 were in liver failure, 338 had been diagnosed with hepatocellular carcinoma and 1,187 were post-transplant patients. The annual incidence of new cases of cirrhosis appeared to plateau in 1997 through 2007 but, according to the results of our model, the incidence of the more serious outcomes of HCV infection will continue to rise, at least until 2027, unless modified by treatment. Liver deaths, for example, were estimated to increase from only 24 in 1967 to 613 in 2027, increasing approximately 27% from 2007 to 2027.

Overall, a cumulative total of 221,198 cases of hepatitis C have been diagnosed and reported in Canada, with the time trend of HCV reported cases suggesting we are probably not close to “exhausting the prevalent pool” of HCV-infected persons. Our estimate of the proportion of HCV diagnosed is subject to uncertainty. Applying annual mortality rates from 1991 to 2007, we estimated that 192,225 of these diagnosed cases were still alive as of December 2007. This represents 79% of modelled HCV infections, though the proportion varied considerably by province/territory.

We were unable to determine to what extent there may be residual duplicates in the national database. If there were significant numbers of unrecognized duplicates in the database, the proportion of HCV infections that have been diagnosed would be lower.

There are several assumptions and limitations to the methods we used in this analysis. With regard to demographic characteristics, data on births, deaths and census populations were relatively precise and therefore not subject to significant uncertainty. modelled population counts initially agreed very closely with Statistics Canada data in 2001, in part because the model was adjusted to fit known population counts. To determine the prevalence of HCV infection in immigrants, we relied on HCV in the country of origin as reflected by the occurrence of HCV among persons not born in Canada in the EHSSS database. It is unclear how precise the country birth data is in this database.

For the sake of simplicity, we assumed that all hemophilia patients were born in Canada. This was done for two reasons. First, given the limited medical care available in many of the countries from which immigrants come to Canada, many persons with hemophilia may not survive long enough to be able to emigrate to Canada. Also, some hemophilia patients would not be considered eligible for immigration to Canada, given that persons with this condition may incur excessive medical costs. Finally, we do not have data on the distribution of hemophilia patients by country of birth. Thus, it was simpler to model this group entirely within the Canadian-born stratum. Nevertheless, it is likely that at least some of the hemophilia patients categorized as Canadian in our analysis were, in fact, born elsewhere.

The model we developed incorporates assumptions about the likely incidence and prevalence of HCV in key populations. The results of our model are based on data derived from a review of epidemiologic studies and from the EHSSS surveillance project operated by the Public Health Agency of Canada. However, only one reasonably population-based large study on HCV prevalence has been carried out in Canada. This is the study of outpatients attending sentinel hospitals in Quebec in 1990-92¹⁶. The EHSSS also has limitations related to incomplete and biased data. These are discussed more fully in the report of our study of HCV infection in Canada in 2002². Risk factor data was not available from cases routinely reported for surveillance purposes, which would have helped to assess indirectly the relative infection rates in the principal groups at risk for HCV.

In the present study, we also reviewed published studies to identify additional information on the parameters to quantify progression of hepatitis C through its clinical stages and incorporated updated data into our Markov model. Nevertheless, there are substantial uncertainties about the values of these parameters. First, there was considerable variation in the parameters found in different studies and this was only partly controlled by stratification by HIV status and gender. It is unclear to what extent these parameters are applicable to the Canadian population; therefore, the estimates of the incidence and prevalence of HCV sequelae are subject to uncertainty.

We determined the prevalence of hepatitis C infection in each province/territory by using reported cases of hepatitis C as weights and interpolating the HCV infections for each province/territory from the Canadian total using these weights. This approach assumes that the diagnosis and reporting of hepatitis C infection is similar in all jurisdictions in Canada; this may not be the case. It is difficult to determine the strength of direction of biases that may have been introduced by differential testing and reporting.

In this study, we examined HCV prevalence among two key populations, namely incarcerated individuals and the Aboriginal population. Among incarcerated populations, we were able to obtain information from several studies on the proportion of prisoners who had injected drugs and the prevalence of hepatitis C infection stratified by IDU history.

Nevertheless, the populations studied may not have been representative of all incarcerated populations in Canada, and there are regional and institutional differences in HCV prevalence across Corrections Services Canada⁵⁹. Recent data from Corrections Services Canada suggest that the overall HCV prevalence in federal penitentiaries was approximately 29% in 2005 (slightly higher than our estimate)⁶⁰. Estimated self-reported HCV prevalence among those who do not report a lifetime history of IDU may also be higher than the estimated 1% in this report, as a result of self-reported risk behaviours, including slashing/fighting, sex with an IDU, and sex trade involvement⁵⁹. Prevalence greater than 1% in non-IDUs could also be due, in part, to under-reporting of injection drug use in these studies.

We also examined HCV infection among the Aboriginal population. However lack of data limited the validity of this process. Few studies of hepatitis C infection in this population were available. In fact, few studies of hepatitis C in population-based samples were found. We based our estimate on the relative incidence of AIDS among IDUs in Ontario and incorporated results from Manitoba. Thus, there is considerable uncertainty about our estimate of the 7,200 HCV-infected Aboriginal persons in Ontario. There is uncertainty with respect to the HCV prevalence rate in Aboriginal populations, as it is not clear whether the definition of Aboriginal is the same in different databases. We used the figures for the Aboriginal population from the 2006 Canadian census; however, ethnicity in the census is self-identified and therefore may not conform to other definitions of Aboriginal status, such as those used in epidemiologic studies. An additional source of uncertainty with respect to HCV infection in Aboriginal populations is the observation by Minuk that a higher proportion of persons in this population with anti-HCV antibody may be HCV-RNA negative (i.e. not actively infected) compared to other populations39. The reasons for this apparent difference remain unclear.

With regard to active HCV infection status in general, the estimates presented in this report indicate the number of persons who have HCV RNA, HCV Ab or both. In fact, not all of these persons are actively infected. We estimated from an extraction of the model examining only those with HCV RNA, that 205,338, or 85%, of these persons were HCV RNA-positive, that is have active HCV infection. Only these persons can transmit HCV and are candidates for antiviral therapy.

In our model, we found that 31% of prevalent HCV infections were due to modes of transmission other than drug injection, blood transfusion and clotting factors. This is likely to be a substantial overestimate of the importance of other modes of transmission. These other modes of transmission may include sexual, occupational, nosocomial and vertical (i.e. mother to baby), as well as transmissions by unsterile practices in body piercing and tattooing. Though hepatitis C may certainly be transmitted through all these other modalities, it is unlikely that they are highly efficient (and therefore frequent) modes of transmission. The proportion of other modes of transmission is based on data from the EHSSS database. Due to the stigmatization of injection drug use, it is likely that this behavior is underestimated in this surveillance system, thus overestimating the importance of the other modes of transmission. Participation bias may also result in unrepresentative data. IDU cases are more difficult to reach and many in fact may not be reached. In some EHSSS sites, patient consent is required for interview and IDU may be less likely to consent.

We received and reviewed data on reported cases of hepatitis C in Canada. Nevertheless, as noted above, it is unclear whether a significant number of duplicate cases remained in the database. In addition, it appeared that acute cases were not systematically differentiated from chronic infections nor were risk factor data systematically collected. Ensuring these functions would improve the quality and utility of the hepatitis C surveillance data. It would also be extremely helpful to include enhanced surveillance data from all major cities in Canada.

In our projections of HCV infection and their sequelae to 2027, we did not take into account the possible impact of the expanded use of currently available antiviral medications (e.g. pegylated interferon and ribavirin) nor the potential impact of new, more effective regimens. Both these developments could have considerable impact on our future projections. Also, if new medications are identified that are effective and widely available, this could have a considerable impact on our future projections. Similarly, if effective HCV prevention programs for IDUs, both to reduce transition from non-injection to injection and reduce the extent of sharing of equipment, are developed and widely implemented, the incidence and prevalence of HCV projected to 2027 may be less than that estimated by our model.