The burden of hepatitis C virus infection is growing: A Canadian population-based study of hospitalizations from 1994 to 2004

The burden of hepatitis C virus infection is growing: A Canadian population-based study of hospitalizations from 1994 to 2004. Can J Gastroenterol 2008;22(4):381-387. BACKGROUND: Nearly 1% of Canadians are infected with the hepatitis C virus (HCV). Simulation analyses have suggested that HCV will place an increasing burden on the health care system as the infected population ages, but supportive clinical data are limited. OBJECTIVES: To study temporal trends in HCV-related hospitalizations and predictors of increased health care utilization from a Canadian population-based perspective. METHODS: An administrative hospitalization database from the Calgary Health Region was used to identify patients who were admitted for HCV between 1994 and 2004. The primary outcomes were liver-related HCV hospitalizations, length of stay, hospital costs and in-hospital mortality. Average annual growth rates in outcomes were calculated and subgroup analyses were conducted according to age, sex and HIV/HCV coinfection status. RESULTS: Between 1994 and 2004, there were 4002 HCV-related hospitalizations; 22% were liver-related. Liver-related hospitalizations, lengths of stay and in-hospital mortality increased approximately fourfold or an average of 15% to 18% annually (P<0.0005). Patients aged 40 to 59 years and HIV/HCV coinfected patients experienced the largest average annual growth rates (19% to 27% and 30% to 40%, respectively; P<0.0005), reflecting the accelerated natural history of HCV in these subgroups. Hospital costs for liver-related HCV hospitalizations increased by an average of 41% annually (P=0.001) between 2000 and 2004. The average annual increase in liver-related hospitalizations remained significant in a sensitivity analysis, even when 75% of HCV cases were under-reported in 1994. CONCLUSIONS: The present studies’ findings confirm the grow-ing burden of HCV on the Canadian health care system. Strategies to prevent HCV infection and maximize the dissemination and most effective use of potentially curative antiviral therapies are necessary to reduce these trends. HISTORIQUE Près de 1 % des Canadiens sont infectés par le virus de l’hépatite C (VHC). Selon les analyses de simulation, le VHC exercera un fardeau croissant sur le système de santé avec le vieillissement de la population infectée, mais les données cliniques corroboratives sont li-mitées. OBJECTIFS Étudier les tendances temporelles dans les hospitalisations reliées au VHC et les prédicteurs d’augmentation de l’utilisation des ser- vices de santé selon une perspective en population canadienne.

C hronic hepatitis C virus (HCV) infection is a major public health problem. An estimated 170 million individuals worldwide (3% of the population) (1) and 250,000 Canadians (0.8%) are anti-HCV positive (2). The majority (approximately 80%) are chronically infected and may progress to cirrhosis, end-stage liver disease and hepatocellular carcinoma (3). Although the natural history of HCV typically follows a protracted course, with complications occurring decades after initial infection, several risk factors, including older age, male sex (4,5) and HIV coinfection (6,7) are associated with accelerated disease progression. Cirrhosis caused by chronic HCV accounts for at least 1000 deaths annually in Canada (8) and has become the leading indication for liver transplantation (9).
Considering the large pool of infected individuals, the current burden of HCV on the Canadian health care system is relatively modest. However, as the population of infected individuals ages and their disease has time to progress, more patients are expected to develop complications and place increasing strain on the health care system. Several studies using mathematical models have projected a marked increase in rates of HCV-related hepatic complications (10)(11)(12)(13)(14)(15). For example, Zou et al (10) projected that the number of cases of HCVrelated cirrhosis, liver failure, hepatocellular carcinoma and liver-related deaths would approximately double between 1998 and 2008. Although Pohani et al (8) reported a 3.7-fold increase in the age-standardized mortality rate for non-A, non-B hepatitis (primarily caused by HCV) (16) in Canada between 1979 and 1997, surprisingly little clinical data have been reported to support the remainder of these alarming projections (17).
The objective of the present study was to describe temporal trends in HCV-related hospitalizations in a large Canadian health region between 1994 and 2004. By describing recent changes in hospitalizations for HCV, we provide a frame of reference for future projections of the burden of this disease on the Canadian health care system. April 1, 1994 and March 31, 2005 (fiscal years 1994 through 2004) were identified by using an administrative hospitalization database maintained in the Calgary Health Region (CHR) (18). The CHR is one of the largest fully integrated, publicly funded health care systems in Canada, providing virtually all medical and surgical care to approximately 1.1 million residents of Calgary and surrounding communities in southern Alberta. Included within the region are 12 academic and community hospitals, including four large hospitals within the city of Calgary. Each hospital separation within the administrative database contains up to 50 diagnosis codes (18 [20]), the hospitalization was defined as HCV-related. Because HCV-related complications were the predominant concerns in the present study (rather than unrelated conditions requiring admission in HCV-positive patients), liverrelated hospitalizations were also identified. A hospitalization was considered to be liver-related if the principal diagnosis was HCV; there was any diagnosis of cirrhosis, portal hypertension, other sequelae of liver disease, hepatic encephalopathy, ascites, hepatorenal syndrome, variceal bleeding, hepatocellular carcinoma or cholangiocarcinoma; or the patient had undergone a liver transplant (21,22).

Validation of HCV diagnoses
To validate the diagnoses of HCV, the medical records of 100 randomly selected patients were reviewed by a physician blinded to the administrative data (AMS). According to the validation study, the sensitivity, specificity, positive predictive value and negative predictive value of the administrative data for confirmed HCV infection (defined as anti-HCV positivity) were 97% (56 of 58), 98% (41 of 42), 98% (56 of 57) and 95% (41 of 43), respectively.
Analysis of under-reporting of HCV Because HCV antibody testing did not become available until 1989 (23), it is unlikely that widespread testing occurred by 1994. Thus, the hospitalization frequencies that were observed in the present study likely underestimated the actual numbers by greater proportions at the beginning of the study. Because the rate of HCV testing and reporting increased over time, the growth rates that were observed in the present study may have overestimated the increase in hospitalizations for HCV. Therefore, a sensitivity analysis was conducted to consider the potential impact of under-reporting. As described by Grant et al (22), as a baseline scenario, it was hypothesized in the present study that 50% of cases were under-reported in 1994 and that under-reporting fell each year by a fixed amount, so that all cases were reported by 2004. The initial rate of underreporting was varied between 25% and 75%. As a control group, hospitalizations for alcoholic cirrhosis (ICD-9-CM 571.2 [19]; ICD-10 K71.3 [20]) were examined. Because alcoholic cirrhosis was not as likely as HCV to be under-reported during the early years of the study, differences in hospitalization frequencies for these conditions likely reflect both the effects of under-reporting and differences in disease burden (22).

Statistical analyses
For all HCV-and alcoholic cirrhosis-related hospitalizations, admittance frequencies, total days in the hospital and inhospital mortality were examined as measures of disease burden. Poisson regression, with the independent variable being a timetrend indicator for fiscal year, was used to assess the rates of change of these outcomes over the interval (24). These models were adjusted for the overall number of hospitalizations in the CHR, which grew 1.0% annually between 1994 and 2004. The results changed minimally when the models were adjusted for the growth of the CHR population (data not shown). Where the number of events permitted, subgroup analyses were performed according to sex, age (in 20-year categories) and HIV coinfection (ICD-9-CM 042, 079.53, V08 [19]; and ICD-10 B20-B24, Z21 [20]), because these factors have been demonstrated to impact HCV disease progression (4-7).
Total hospital costs were also examined, which were available in 93% to 97% of admissions between 2000 and 2004, based on expenses accrued from direct and indirect patient care (excluding physician fees). All costs were adjusted to 2007 Canadian dollars using the Consumer Price Index (25). Simple linear regression models of hospital costs versus fiscal year (2000 to 2004) were used to assess average annual changes in total costs. Hospitalization details were compared using Mann-Whitney, χ 2 and Fisher's exact tests, as appropriate. Stata v8.0 (StataCorp LP, USA) and SAS v9.1 (SAS Institute, Inc, USA) were used for all analyses. The study was approved by the Conjoint Health Research Ethics Board at the University of Calgary (Calgary, Alberta).  Table 1 provides average annual growth rates for HCV-and alcoholic cirrhosis-related hospitalizations and days in the hospital during the 11-year study period. The admittance frequencies and lengths of stay increased significantly for HCV-related hospitalizations, while alcoholic cirrhosis-related admissions remained stable and lengths of stay increased slightly. Similarly, total hospital costs for liver-related admissions increased an average of 41.2% (95% CI 31.0% to 51.4%) annually between 2000 and 2004 for patients with HCV, but remained stable for those with alcoholic cirrhosis (average annual change, -4.6%; 95% CI -11.5% to 2.4%). Importantly, the observed increases in the number of days in the hospital and costs for HCV resulted from more frequent hospitalizations as opposed to longer lengths of stay. The median (interquartile range [IQR]) length of stay for liver-related HCV hospitalizations was not significantly different between the first and final three years of the study period (1994 to 1996 versus 2002 to 2004: seven days [three to 12 days] versus seven days [four to 14 days]; P=0.14).

RESULTS
Similar results were obtained with respect to in-hospital mortality. The average number of in-hospital deaths in patients with liver-related admissions increased 16.2% (95% CI 8.3% to 24.7%) annually in patients with HCV, but remained stable in those with alcoholic cirrhosis (0.1%; 95% CI -2.5% to 2.8%). This increase in HCV-related deaths was entirely attributable to the growth in the number of hospitalizations, as opposed to an increased risk for death over time, because the proportion of in-hospital deaths did not change during the study interval (1994 to 1996 versus 2002 to 2004: 12.3% versus 12.4%; P=1.00).
To exclude under-reporting of HCV diagnoses as a cause of the observed increase in hospitalizations, a sensitivity analysis was conducted, assuming greater under-reporting during the early years of the interval (Figure 4). In the baseline scenario (in which 50% of HCV cases were under-reported in 1994) the average annual increases in liver-related and all-cause HCV hospitalizations were 7.8% (95% CI 5.8% to 9.7%) and 15.7% (95% CI 14.6% to 16.8%), respectively. In the extreme scenario of 75% under-reporting in 1994, the average annual increases in liver-related (1.5%; 95% CI 0.05% to 3.0%) and all-cause HCV hospitalizations (14.4%; 95% CI 13.3, 15.4%) remained significant.
The hospitalization growth patterns that were observed were greatest among patients in their 40s and 50s (Table 1). While other patient subgroups demonstrated a decline in the number of hospitalizations during the study interval, patients in their 40s and 50s demonstrated an increase (   [19]; and ICD-10 O00-O99 [20]; data not shown). Accordingly, the share of HCV hospitalizations (liver-related and all-cause) attributable to men decreased significantly between the 1994 to 1996 period and the 2002 to 2004 period ( Table 2). As in the overall analysis, the greatest average annual increase in number of HCV hospitalizations among women was observed in the 40-to 59-year-old age group ( Table 3). The average annual increase in liver-related hospitalizations was greatest in women 60 years of age and older (Table 3).  (Table 2). Average annual increases in liver-related (40.3%; 95% CI 21.9% to 61.4%) and all-cause HCV hospitalizations (29.3%; 95% CI 24.8% to 34.0%) in this patient subgroup were the highest of any observed (Figures 1 and 2). Accordingly, these patients had highly significant average annual increases in the lengths of stay for liver-related (37.7%; 95% CI 31.9% to 43.8%) and all-cause HCV admissions (29.6%; 95% CI 28.2% to 31.0%).

DISCUSSION
Our study demonstrates a highly significant increase in the burden of HCV on the health care system. For the primary outcomes of interest -liver-related hospitalizations, lengths of stay, and in-hospital mortality -we observed average annual increases ranging from 15% to 18%. In contrast, alcoholic cirrhosis-related hospitalizations, costs and inpatient deaths remained stable. These results suggest that the future burden of HCV will likely surpass previous expectations. Zou et al (10) projected a doubling of the number of patients with HCVrelated complications, including liver-related deaths, between 1998 and 2008. Between 1994 and 2004, fourfold increases in liver-related HCV hospitalizations, lengths of stay and inpatient deaths were observed, suggesting that published projections of HCV burden may be conservative. Our results are in keeping with hospitalization data in the United States, which showed three-to fourfold increases in these outcomes between  1994 and 2001 (22). In this study, Grant et al (22) also reported a 26% annual growth rate for HCV-related hospital charges, which was slightly lower than the 41% annual increase in costs that we observed. Based on the costs and annual growth rates observed in the CHR, we project that liver-related HCV hospitalizations will cost the Canadian health care system approximately $240 million in the year 2020. Our results are consistent with natural history studies describing the impact of age, sex and HIV coinfection on the progression of chronic HCV (4-7). The role of aging is highlighted by the disproportionate shares of hospitalizations among patients in their 40s and 50s, in whom average annual growth statistics ranging from 19% to 27% were observed. Because the prevalence of HCV is highest in this subgroup (26), aging of these patients will exact an increasing toll on the health care system in coming years. As expected, men contributed to a majority of HCV-related hospitalizations due to their higher HCV prevalence (26) and accelerated disease progression (4-7). Surprisingly, however, we observed a disproportionate increase in the number of hospitalizations among women -a previously unreported finding that warrants confirmation in larger studies. A sex-dependent difference in underreporting at the beginning of the interval (eg, less frequent testing in women due to their lower HCV prevalence) is a potential, but entirely speculative, explanation.
Finally, HIV/HCV coinfected patients experienced alarming annual growth rates, ranging from 30% to 40%, for liverrelated and all-cause HCV hospitalizations. In an analysis of United States hospitalization data among HIV-infected patients, Gebo et al (27) described a 29% reduction in overall hospitalizations, but a tripling of HCV-related hospitalizations between 1996 and 2000. Due to reductions in HIV-related mortality attributable to highly active antiretroviral therapy and prophylaxis of opportunistic pathogens, HCV has emerged as a major cause of morbidity and mortality in coinfected patients (28,29). For example, in a nationwide population of HIV-infected patients in France (30), deaths due to end-stage liver disease -the majority of which were due to HCV -occurred in 1.5% of patients in 1995, 6.6% in 1997, 14.3% in 2001 and 12.6% in 2003. We also found that HIV/HCV coinfected patients admitted for liver-related conditions were approximately 10 years younger than their HIV-negative counterparts, in keeping with the accelerated natural history of HCV in this subgroup (6,7).
The increasing burden of chronic HCV that we have observed highlights the importance of primary prevention of HCV infections and initiating anti-HCV therapy in eligible patients. Currently available regimens consisting of peginterferon and ribavirin lead to viral eradication in approximately one-half of treated individuals (31,32). Viral clearance has been associated with improved clinical outcomes, even in those who have already progressed to cirrhosis (33)(34)(35). Unfortunately, due to its cost, contraindications, toxicities, the requirement for intensive monitoring and counselling that is typically available in only specialized centres, and the reluctance of patients to undergo treatment, the widespread uptake of anti-HCV therapy has been limited (36)(37)(38). Of the roughly 200,000 prevalent cases in Canada, only an estimated 5000 are treated annually (2). In a recent study from a Canadian hepatology centre (38), only 29% of 1317 patients were considered   *Patients younger than 20 years of age were excluded due to the small number of observations precluding regression analyses eligible for therapy, and of these, only 38% (11% of the total) actually received treatment. Hopefully, the more widespread dissemination of antiviral therapy, including more effective and better-tolerated agents under development (39), will be able to stem the coming tide of HCV-related morbidity and mortality, as suggested in modelling studies (11,12,14,15). Our study had several limitations. Most importantly, underreporting of HCV diagnoses during the early years of the study may have led to an overestimation of the true increase in the burden of HCV. Several points argue against the significance of this limitation. First, we observed parallel increases in allcause and liver-related HCV outcomes. Although underreporting is conceivably more likely in patients admitted for nonhepatic conditions, it is much less likely that patients with liver-related admissions would not be tested for HCV or have the diagnosis recorded in their discharge abstract. Second, greater annual increases in liver-related HCV hospitalizations were observed among 40-to 59-year-olds than among older patients. There is no evidence to suggest an age-dependent effect on under-reporting of HCV. Finally, our sensitivity analysis supported the robustness of our findings. In every scenario, including the most extreme in which 75% of cases were under-reported in 1994, increases in liver-related HCV admissions remained significant. Our study was also limited by the inability to track individual patients over time. Thus, we cannot distinguish changes in overall hospitalization counts from the numbers of hospitalizations per patient. Finally, our data originates from a single, albeit large, Canadian health region. The generalizability of our findings to other regions (within Canada and elsewhere) requires confirmation.
Shortly following its discovery, HCV was described as "a sleeping giant" (40); our data support previous findings that the giant has indeed awoken (22). We observed highly significant increases in HCV-related hospitalizations, lengths of stay, hospital costs and in-hospital deaths that likely foreshadow an even greater burden on the health care system. To stem this tide, health care policy makers and physicians must focus on preventive initiatives and strategies to maximize the dissemination and most effective use of potentially curative therapies.
CONTRIBUTIONS: Robert Myers conceived the study, performed the statistical analyses and wrote the first draft of the manuscript. MingFu Liu extracted administrative data. Abdel Aziz Shaheen performed the validation study and assisted with statistical analyses. All authors contributed to revising the manuscript and approved the final version. Robert Myers is the guarantor of the study.