Background
Worldwide, about 350 million persons have chronic hepatitis B virus (HBV) infection, and about 125 million have been infected with hepatitis C virus (HCV), putting viral hepatitis B and C amongst the world's greatest infectious disease health problems [
1‐
3]. Many patients are not aware of their chronic disease, and this is also true for another chronic liver disease, namely hereditary haemochromatosis, which is the most common autosomal recessive genetic disorder in Europe [
4]. Viral hepatitis B and C, and haemochromatosis have in common that they may lead to liver cirrhosis and subsequently to liver failure and hepatocellular carcinoma, and that these complications can be prevented in most patients if diagnosis is made early and patients receive adequate treatment. Therefore these liver diseases are attractive candidates for public health measures aiming at prevention, early diagnosis and treatment. In this context not only information on the general population, but also on selected populations with a potentially higher risk is important. The setting of emergency rooms is particularly attractive for screening because patients can be directly educated about viral hepatitis, and treatment is readily available if necessary. A US based study determined the prevalence of HBV and HCV infection in an emergency room patient population in the early 1990s [
5], but no comparable studies were done recently or in Europe. Likewise no previous studies have looked at the prevalence of elevated transferrin saturation (TFS) and its association with clinical symptoms of haemochromatosis and other liver diseases in such a population. We therefore aimed to study the epidemiology of these liver diseases in patients admitted to a university hospital emergency room.
Methods
Setting
The study was conducted in the emergency room (ER) of the University Hospital Berne where patients are admitted 24 hours per day, 365 days per year. The ER admits about 20'000 patients per year from urban Berne as well as from the surrounding rural areas. These include any medical, surgical and subspecialty emergencies, except paediatric, gynaecologic and obstetric emergencies, which are generally admitted to other specialised clinics. The study was approved and supervised by the regional ethics board (Kantonale Ethikkommission Bern).
Subject recruitment and data collection
The a priori target was to include a representative sample of 5'000 patients from the population of all patients admitted to the ER. Patients were recruited between February 2003 and September 2004. There were no formal exclusion criteria, but any individual patient could only be included once. During an initial pilot phase of about two months we asked only a sample of patients admitted to the ER to participate in the study (increasing rate from 1 in 10 to 1 in 2 patients, selected by random sampling). Thereafter all patients were asked to participate in the study. Upon arrival and admission to the ER, patients were given a patient information sheet, an informed consent sheet, and a study-specific questionnaire, which was available in German, French, Italian and English (Table
1). Treating physicians and nurses provided additional information, answered any study-related questions and obtained a 10 ml venous blood sample. Demographic data were collected through the ER's electronic patient data system.
Table 1
Study-specific patient questionnaire
Have you been vaccinated against hepatitis B ? | Yes/No/I don't know |
Do you have any known hepatitis B infection ? | Yes/No |
Do you have any known hepatitis C infection ? | Yes/No |
Have you ever had a yellowing of the skin (known as jaundice) ? | Yes/No |
Have you ever had a blood transfusion ? | Yes/No, if yes until/after 1990 ? |
Do you have a tattoo ? | Yes/No |
Do you have any piercing ? | Yes/No |
Do you drink alcohol ? | Never/1–10/>10 glasses/week |
Have you ever taken illegal intravenous drugs ? | Yes/No |
Do you live with a partner in a steady relationship ? | Yes/No |
Do you work in the public health services with patient contact ? | Yes/No |
Have you been suffering from stomach aches without clear cause ? | Yes/No |
Have you been suffering from tiredness without clear cause ? | Yes/No |
Have you been suffering from muscle pains without clear cause ? | Yes/No |
Have you been suffering from joint pains without clear cause ? | Yes/No |
What is your ethnic origin ? | White/Black/Asiatic/other, mixed |
Were you born in Switzerland ? | Yes/No, if no specifiy: |
Do you permanently reside in Switzerland ? | Yes/No, if no specifiy: |
We obtained complete and reliable individual patient level information (age, sex, country of residence, health insurance, admission specialty, admission time and date) on all patients admitted to the ER during the study period from the hospital's electronic data system. This information was used to compare our study population with the source population, and to standardize prevalence to the source population.
Laboratory analyses
Anti-HBc antibodies were determined with a diagnostic kit from Ortho-Clinical Diagnostics (Vitros Immunodiagnostic Products Anti-HBcAg Reagent Pack). Anti-HCV antibodies were determined using a second generation ELISA (Ortho-Clinical Diagnostics, Vitros Immunodiagnostic Products Anti-HCV Reagent Pack), measuring anti-HCV-C100-3, anti-C22c and anti-C33c. Serum alanine aminotransferase (ALT) activity, serum iron and total iron binding capacity were determined using the laboratory's routine methods. Transferrin saturation (TFS) was calculated as the ratio of serum iron and total iron binding capacity.
Data analysis
Unadjusted prevalence was calculated directly, and 95% confidence intervals were based on a binomial distribution. For anti-HBc and anti-HCV antibodies prevalence was also adjusted by direct standardization to the source population's demographic and admission variables. Univariable (unadjusted) prevalence ratios (PR) were calculated for several independent variables (factors). When confounding by other variables was suspected adjusted relative prevalence rartios were calculated in a restricted population or as a pooled estimate after stratification over categories of potentially confounding variables according to the method of Mantel and Haenszel [
6]. Thereafter multivariable logistic regression was used for the simultaneous modelling of independent associations with several variables. We used log-binomial regression as an alternative to logistic regression when log-binomial model fitting could be achieved, in order to estimate prevalence ratios instead of odds ratios [
7]. All analyses were done with STATA Version 8.2 for MacOS X (Stata Corporation, College Station, Texas, USA).
Discussion
The current study presents the first data on prevalence and associated factors of liver diseases in a large sample of patients admitted to the emergency room of a European university hospital. The prevalence of anti-HBc and anti-HCV was 6.7% and 2.7%, respectively. Factors independently associated with positive anti-HBc were IVDA, foreign country of birth, non-white ethnicity and high age. Positive anti-HCV was associated with IVDA, blood transfusion and chronic abdominal pain. Patients demonstrated poor awareness and knowledge of liver disease. Transferrin saturation was elevated in 3.3% and was associated with chronic fatigue.
The obtained data from a large sample of patients admitted to the ER were essentially complete, plausible, and generally in accordance with previously available information, supporting the overall validity of our study. The current study also had important limitations: first, we studied only 14.9% of our target population, and the possibility of a recruitment bias has to be addressed. Therefore, we obtained demographic and admission characteristics on all patients admitted to the ER during the study period, and a comparison with this information demonstrated that our study population can be considered as a reasonably representative sample of the target population with regard to these variables, and allowed us to standardise prevalence estimates. Second, we were not able to perform additional laboratory tests of interest including HBsAg, anti-HBs, and confirmatory secondary testing in all anti-HCV positive patients. However, like in other studies, the primary aim of our study was to determine past infection with HBV and HCV. Also, further confirmatory testing was done in a small sample of anti-HCV positive patients, and the observed predictive value positive of 95% suggests that among the 135 anti-HCV positive patients 129 had a previous HCV infection. Assuming a 94% sensitivity of the study's enzyme immunoassay [
8], about 137 patients in the tested population would then have had a previous HCV infection, which is very similar to the number without these corrections. And although HCV-RNA was not determined in all patients, one can expect that approximately 90% of infected patients had a chronic infection [
9]. Third, we were not able to do genotyping for mutations associated with haemochromatosis. Finally, the observed associations are descriptive, and do not allow causal inferences.
Several studies from different countries have reported prevalence estimates of viral hepatitis in selected or more general populations as e.g. in the Italian Dionysos study [
10]. However the only previous study from a developed country that determined the prevalence of viral hepatitis in an emergency room population was conducted in Baltimore, USA in the early 1990s, i.e. before the introduction of widespread HBV vaccination programs [
5]. There, the prevalence was 5% for HBsAg, indicating active and not previous HBV infection, and 18% for anti-HCV. In the general Swiss population an estimated 3% have been infected with HBV [
11,
12], and 0.75% with HCV [
13]. Our results therefore indicate that the prevalence of previous infection is about twice as high for HBV and three times higher for HCV compared to the general population. Possible explanations for the higher prevalence are medical conditions associated with viral hepatitis that lead to ER admission and overrepresentation of high-risk groups such as intravenous drug users, patients not enrolled in regular health plans and a primarily urban population. Amongst intravenous drug users, anti-HBc prevalence was 42% compared to 60–80% amongst other Swiss IVDA cohorts in the 1980s and 20% in the early 1990s [
14,
15]. The higher prevalence of previous HBV infection in immigrants is in accordance with frequent perinatal transmission in countries with high endemicity, and with results from a previous Swiss study in pregnant women [
12]. 60% of intravenous drug users were anti-HCV positive, which is similar to the results of two Swiss studies published in 1990 and 2000 in IVDA cohorts [
16,
17].
The decreasing prevalence of HBV infections in intravenous drug users demonstrates the success of HBV vaccination programs that target this high-risk group, and immunization of all adolescents (implemented in Switzerland in 1998), is expected to be eventually highly efficacious in reducing HBV infections in the general population. However it will require decades until these measures will reduce the population burden of HBV infection, and adults at high risk including immigrants are difficult to reach through national vaccination programs and public health campaigns. Particularly this high-risk subpopulation is well represented in ER departments. Further, 74% of the anti-HCV positive patients admitted to the ER were less than 60 years old, which is of particular relevance since treatment with pegylated interferon and ribavirin should be instituted in patients at risk for late complications at a young age. Based on a chronic course in approximately 10% of anti-HBc and 70–80% of anti-HCV positive patients, about 130 patients with chronic HBV infection and 350–400 patients with chronic HCV infection are seen in the studied ER population during one year. However 42% had not been vaccinated against HBV, 33% did not know whether they had been vaccinated, only less than half of all anti-HCV positive patients knew about their infection, and half of all anti-HCV positive patients reported regular alcohol consumption, which is strongly discouraged as it promotes HCV-related fibrogenesis [
18]. Therefore our results indicate that intense efforts to diagnose viral hepatitis and educate patients in ER have the potential to prevent new infections and treat complications.
Elevated TFS is an important marker of disease in C282Y homozygote patients, who account for about 90% of all clinical cases of haemochromatosis [
19,
20]. Although the prevalence of this mutation is known to be about 0.4 to 0.5% in Caucasian populations [
4,
21], until recently there has been few data on the penetrance of C282Y in the general population, which is critically important for the controversial discussion on the utility of population screening [
22]. In a large international study Adams et al. reported a prevalence of 0.44% for C282Y homozygotes in a white primary care population, and amongst those TFS was >50% in 84% of men and >45% in 73% of women [
21]. Based on this data we would expect 20 C282Y homozygous amongst 4'616 white patients (60% men) in our study population, and 16 of them with elevated TFS as defined above. We identified 151 patients with such an elevated TFS amongst 4'616 white patients (3.3%), which allows us to estimate a specificity of 97% and a positive predictive value of 11% regarding the identification of C282Y homozygote patients with elevated TFS in our study population. This implies that 89% of patients with an elevated TFS in white patients of our study population were not C282Y homozygote and therefore limits the power and specificity of our results with regard to haemochromatosis. Other studies in primary care populations were able to identify patients with C282Y homozygosity through genetic screening and reported an association with liver diseases. However, they also failed to observe an association between C282Y homozygosity with elevated TFS and the rather non-specific symptoms of haemochromatosis that are highly prevalent in such a population [
21,
23,
24]. Further, our study had the advantage that patients with high alcohol consumption and HCV infection could be excluded in a subanalysis, and the higher prevalence of fatigue in patients with elevated TFS even within this restricted population may indicate the presence of other chronic liver diseases including haemochromatosis and non-alcoholic steatohepatitis.
Recent data suggest that even amongst C282Y homozygotes with elevated TFS only a minority will develop frank clinical haemochromatosis, and consequently no clear indication for population-based screening for haemochromatosis has been established [
22,
24]. Nevertheless, increased TFS is a useful sign of chronic liver disease including haemochromatosis and non-alcoholic steatohepatitis, and our results confirm that chronic fatigue is a possible symptom of increased TFS and chronic liver disease.
Authors' contributions
SR lead the study coordination and the work of the Clinical Investigation Unit, contributed to the study design, undertook all statistical data analyses, and wrote the first draft of the manuscript. EAD contributed to the data collection, organized the raw data files and undertook the initial descriptive data analyses. HZ and JR initiated and designed the study, revised the manuscript and obtained the funding. HZ lead the contributions of the emergency room staff. GP was responsible for all laboratory sample analyses. SH contributed to the initial study design and planning. All authors read and approved the final manuscript.