Hepatitis viruses in Ethiopia: a systematic review and meta-analysis

Background The existing seroepidemiological data on viral hepatitis in Ethiopia showed a wide variation in prevalence pattern and the clinical and public health burden have been underestimated. The aim of this systematic review and meta-analysis was to provide a clear and comprehensive estimation of viral hepatitis epidemiology and the potential clinical burdens in Ethiopia. Methods A comprehensive literature search was carried out from five decades (1968–2015) published studies from biomedical databases; PubMed, Google scholar, Medline and Web of Science. Results The overall pooled prevalence of hepatitis B virus (HBV) was 7.4% (95%CI: 6.5–8.4). The pooled prevalence among subgroups showed 5.2% (95%CI: 3.7–7.4) in human immunodeficiency virus (HIV) infected individuals, 8.0% (95%CI: 5.9–10.7) in community based studies, 8.4% (95%CI: 5.4–12.7) in blood donors, 11.0% (95%CI: 7.5–15.9) in immigrants and 6.9% (95%CI: 5.6–8.5) in other groups. Among study parameters considered during meta-regression analysis, only study years were associated with a decreasing HBV prevalence rate over time. The overall pooled prevalence of anti-hepatitis C virus antibody (anti-HCV) was 3.1% (95%CI: 2.2–4.4). Unlike HBV, the anti-HCV prevalence in HIV infected individuals was higher (5.5%, 95%CI: 3.8–7.8%, p = 0.01) than the prevalence observed in the other subgroup of study population. Although relatively few data were available, hepatitis virus A (HAV), D (HDV) and E (HEV) were also circulated in Ethiopia. Conclusions This review indicates that all types of viral hepatitis origins are endemic in Ethiopia. Adapting a recommended diagnostic and treatment algorithm of viral hepatitis in the routine healthcare systems and implementing prevention and control policies in the general population needs an urgent attention. Electronic supplementary material The online version of this article (doi:10.1186/s12879-016-2090-1) contains supplementary material, which is available to authorized users.


Background
Hepatitis viruses, such as HAV, HBV, HCV, HDV and HEV cause potentially life-threatening inflammation of the liver, which is characterized by acute and chronic forms of liver disease. According to 2013 World Health Organization (WHO) global health impact report of viral hepatitis [1], more than 240 and 150 million populations were affected by chronic liver disease due to HBV and HCV, respectively. Africa has the second largest number of chronic HBV carriers after Asia and is considered as a region of high endemicity [2]. Despite its high prevalence and highly infectious nature, HCV remains under-diagnosed and under-reported in most African countries [1]. Regarding HAV, Africa is known by high prevalence rate and nearly all older children and adults in the region acquired anti-HAV antibody immunity [3]. HEV is also one of the leading causes of major outbreaks of acute viral hepatitis worldwide, especially in developing countries [4]. Similarly, HDV is prevalent worldwide and associated with the most severe form of viral hepatitis [5].
In Ethiopia, an old clinical study showed that liver disease accounted for 12% hospital admissions and 31% hospital mortality [6]. Moreover, in Ethiopia and neighbouring Kenya more than 60% of chronic liver disease and up to 80% of hepatocellular carcinoma (HCC) are due to chronic HBV and HCV infections [6,7]. Unlike HAV, HDV and HEV, which are not extensively studied, several HBV and HCV seroepidemiological studies were available in the country. However, the majority of the reports showed epidemiological variations of 2.1 to 25.0% over time and across geographical areas as well as the same localities [8][9][10][11][12][13][14][15][16][17]. More importantly, because of HIV pandemic and possible epidemiological overlap as the result of shared transmission ways and risk factors, viral hepatitis-HIV co-infection and subsequent severe forms of clinical complications could be potentially high in the country. Nevertheless, the clinical and public health burdens due to viral hepatitis in general are still given no emphasis in the country's health system. For instance, a recent report showed the presence of very limited knowledge, minimal awareness and underestimation of the viral hepatitis prevalence and disease burden in the country, which have resulted insufficient budgetary and organizational focus [7]. Moreover, according to the WHO report, Ethiopia is regarded as a country with no national strategy for surveillance, prevention and control of viral hepatitis, but the country is classified under the geographical regions with intermediate to hyperendemic viral hepatitis infections [1]. Thus, the lack of a consolidated epidemiological data on the burden of viral hepatitis in Ethiopia might be responsible for the absence of practical action at the policy level. Therefore, this systematic review and meta-analysis was made using data published in the last five decades  to provide a quantified estimate of the problem as a step toward for a better understanding of the viral hepatitis epidemiology, clinical burden and the situation of human immunodeficiency virus (HIV) co-infection in Ethiopia.

Search strategy
A comprehensive literature search was carried out from biomedical databases; PubMed, Google scholar, Medline and Web of Science according to Operations Manual of the Global Burden of Diseases for systematic epidemiological reviews of targeted diseases or condition frequency [18] and Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) [19]. The search focused on all published studies with epidemiological and/or clinical data on the seroprevalence of hepatitis viruses (HAV, HBV, HCV, HDV, and HEV) in Ethiopia from the first scientific description (1968) to 2015. During the search, the following keywords (terms) were used; ["hepatitis A" AND (seroprevalence OR prevalence) AND "Ethiopia"], ["hepatitis B" AND (seroprevalence OR prevalence) AND "Ethiopia"], ["hepatitis C" AND (seroprevalence OR prevalence) AND "Ethiopia"], ["hepatitis D" AND (seroprevalence OR prevalence) AND "Ethiopia"], ["hepatitis E" AND (seroprevalence OR prevalence) AND "Ethiopia"]. The search was repeated by replacing the full written phrases using the abbreviated form of each respective hepatitis virus. Similarly, key terms like "Epidemiology", "Viral liver disease", "Viral hepatitis and HIV coinfections" were used in place of the terms "Seroprevalence" and "Prevalence".

Inclusion and exclusion criteria
Abstracts were screened to determine the relevance of each study. All studies that reported "seroprevalence" and/or "clinical isolation of viral hepatitis" were selected for inclusion according to PRISMA Flow Diagram [19] (Fig. 1). The inclusion was restricted to original research articles published in English language. The studies also included Ethiopian Jews who immigrated to Israel and screened for hepatitis viruses at the time of their arrival.

Data source
The screened articles were organized and subgrouped into five major study population types: community based, blood donors, symptomatic liver disease patients, HIV co-infected individuals and various other groups (healthcare workers and medical waste handlers, out and inpatient department patients, street dwellers, pregnant women, diabetic patients, HIV voluntary counselling and testing (VCT) centres clients and commercial sex workers).

Data extraction and quality assessment
In each abstract and/or full text of the article which was considered to be eligible, information about the study area, year of sample collection and publication, study population, sample size, type of screening kits used and other useful variables were recorded (Additional files 1 and 2). The studies were assessed to identify 'seroprevalence' as an outcome of serological evidence for each of the hepatitis viruses. The quality and eligibility of the screened articles were assessed using a 12 point scoring system based on modified Downs and Black [20] check lists. Special emphases were given for clarity of objective, method, study population characteristics and finding presentations. Articles with low quality score were excluded during the meta-analysis (Additional files 1 and 2).

Data analysis
The extracted data were systematically restructured among viral origin, study area, study groups and study year during the systematic review. The subgrouped data were analysed using simple descriptive statistics during the systematic review. Accordingly, a median and inter quartile range (IQR) values across each subgroup (community based, blood donors, symptomatic liver disease patients, HIV infected individuals, and various other groups of the society were calculated because of expected heterogeneity in terms of the study area, time, sample size and the study design. During the meta-analysis, all selected studies were combined and a random effects model by DerSimonian and Laird [21] were used to estimate overall pooled prevalence. Since the risks of viral hepatitis exposure are potentially high in symptomatic liver disease patients, the metaanalysis was done independently in order to avoid the potential bias on the outcome of pooled estimation. Cohran's Q (low P value indicates presence of heterogeneity) and and I squared (I 2 ) (with I 2 > 50% denoting substantial heterogeneity statistical tests) were conducted to test heterogeneity [22]. The subgroup analysis was performed among the community based studies, blood donors, HIV infected individuals and other groups. The subgroup analysis was not considered for the screening kits used because of the observed homogeneity in the types of the kits (which include an immunoassay based screening kits such as enzyme linked immunoassay, radioimmunoassay and chromatographic immunoassay) (Additional files 1 and 2). Meta-regression analysis was used to determine potential confounders such as mean age groups, study population types, year of study and geographical zones. The mean age group meta-regression was considered for those studies which only reported the mean age (Additional files 1 and 2). The overall fixed and random effects model with 95% confidence intervals (95% CIs) were calculated and illustrated using a forest plot graph presentation. The meta-analysis was not considered for data extracted for HAV, HDV and HEV since the number of available studies was very small. Compressive meta-analysis software version 3.3., 2014 (www.meta-analysis.com) was used during the metaanalysis.

Demographic and study population characteristics
The population of Ethiopia was 22 million at the time (1960s) when the first HBV seroprevalence data were reported. After 30 years, when many of the seroepidemiological and few clinical reports of hepatitis viruses were available, the population steadily increased to 48 million [23]. Currently, with a population of 94 million, the country is the second most populous in Africa. The total study population size screened for hepatitis viruses and involved in this systematic review and meta-analysis were 79,931. Among these, 62,955 were screened for hepatitis viruses from the general population. About 5,229 were from symptomatic patients with acute (867) and chronic (1020) liver diseases, and outpatient department attendants (3,342). The rest 11,747 were from HIV infected individuals. Geographically, the majority of population screened for hepatitis viruses were from central Ethiopia (mainly in Addis Ababa) (45,037), northern Ethiopia (16,071), southern Ethiopia (17,207) and from Ethiopian immigrants to Israel (1616). Overall, 68 studies were eligible for the review and among these, 26 studies reported both HBV and anti- HCV seroprevalence and the rest 6 and 36 studies reported only HCV and HBV, respectively. The mean age group of the study population screened for HBV and HCV was 28.9 ± 6.1 (range 14-48) and 28.2 ± 7.3 (range 16-55) years old, respectively. The peak prevalence was reported between the group of 24-39 and 20-37 years old, respectively to HBV and HCV (Additional files 1 and 2). All of the selected studies were used immunoassay based kits of various types as their primary screening test. In addition, only four (6.1%) studies (two each for HBV and HCV) used PCR for measuring viraemia level (Additional files 1 and 2).
Fourty two (67.7%) out of the 62 studies which reported HBsAg prevalence (Additional file 1) were used in the meta-analysis among the general population. The Cohran's Q and I 2 statistic for HBsAg was 509.7 and 91.9% (p < 0.001), respectively which showed the studies were heterogeneous. The overall pooled prevalence of HBV was 7.4% (95%CI: 6.5-8.4) (Fig. 3). The subgroup meta-analysis of study parameters such as the mean age group, study population types, year of study and geographical zones showed significant variations in the HBV prevalence (Table 1). In particular, the pooled prevalence among  In some studies, along with HBsAg other HBV markers; anti-hepatitis B surface antigen (anti-HBs), anti-hepatitis B core antigen (anti-HBcAg) and hepatitis e Ag (HBeAg) and anti-hepatitis e antigen (anti-HBeAg) were also determined ( Table 3). The peak prevalence rate (82%) for any markers was reported among chronic liver disease patients, whereas the lowest rate (42.2%) was in the general population [52]. The anti-HBs antibody was recorded at 11.6% in HIV infected individuals [53] and the highest rate (73.6%) was among healthcare workers [25]. Similarly, the HBeAg prevalence rate ranges from 0.6 to 44.0% (Table 3).

Epidemiology of HAV, HDV and HEV
An early report of HAV antibody prevalence rates was 84 and 50% [46,59], respectively among the general population and children. In regard to HDV, available scientific reports in the 1980s showed from 5.8 to 9.6% of anti-HDV antibody prevalence rates among the general population [59,60].
HEV was also reported 59% in acute pregnant women [61] and 93% in acute patients and 3% in healthy controls [47].

Epidemiology of viral hepatitis among HIV infected individuals
Cases-control studies of HBsAg among HIV infected individuals showed a record of 14% (vs. 4.3% in the control) in Shashemene (Southern Ethiopia) [62] and 5.4% (vs. 3.9% in the controls) from the central Ethiopia in Addis Ababa [42]. Studies which included HIV infected individuals alone reported 2.3% HBsAg prevalence rate in Woldiya [63] and 5.6% in Gondar [64] and 6.9% in Hawassa [65] from the north, northwest and south part of Ethiopia, respectively. Overall, for the last 10 years, the median HBsAg seroprevalence among HIV infected individuals was 6.0% (IQR = 3.0-6.0) with an increased trend prevalence (X 2 = 15.1, p < 0.001) when compared to HIV negative controls. In the subgroup meta-analysis, HBV prevalence in HIV co-infected individuals was 5.2% (95% CI 3.7-7.4, p < 0.001) ( Table 1). Includes: Nationwide studies (n = 3), North (n = 1), North east (n = 2), South east (n = 5) and South west (n = 1). d It includes: Community based studies, blood donors, HIV co-infected and immigrants. e one, three, one, and three studies were from nationwide, north, northeast and southwest Ethiopia, respectively The prevalence rate of HCV antibody in HIV infected individuals was 4.5%, which was higher than HIV negative controls (0.8%) in Addis Ababa [55]. A study from southern Ethiopia in Hawassa [11] showed a 10.5% prevalence of HCV in HIV infected individuals compared to 6.0% in the HIV negative control group. In this report and another similar study in Addis Ababa [65], HCV viraemia was reported 9.1 and 25.4%, respectively from anti-HCV antibody positive individuals. Overall, the HCV prevalence showed an increasing trend prevalence (X 2 = 18.7, p < 0.001) in HIV infected individuals than noninfected counterparts with a median prevalence of 6.5% (IQR = 4.2-9.9) [11,12,53,55,64,[66][67][68][69]. In the subgroup meta-analysis, the prevalence of anti-HCV antibody in HIV co-infected individuals was 5.5% (95% CI 3.8-7.8, p < 0.001), respectively (Table 1 and Additional file 4). Unlike mean

Discussion
This is the first detailed systematic review and metaanalysis study of viral hepatitis in Ethiopia over a large period of time. The findings clearly show the burden of viral hepatitis in the country in general and endemicity of HBV in particular. The findings were also consistent with the estimate of 5-20% of most African countries [70] and the WHO HBV endemic definition (5-7% HBsAg prevalence) in the general population of a defined geographical area [1]. Similarly, HBV markers other than HBsAg ever reported in the country have Includes antenatal care attendants, HIV voluntary counselling and testing (VCT) clients (HIV negative), VCT clients, diabetic patients, non-diabetic patients (controls for diabetic patients), out and inpatient department clients, healthcare workers and medical waste handlers, and VCT clients been in agreement with data from most African countries and support the endemicity of HBV estimated from the HBsAg prevalence.
HBV seroprevalence studies in blood donors were merely geographical representative, which covered north, northwest, southwest and central parts of Ethiopia. However, the HBsAg prevalence in this group was a bit higher (8.4%) than the community based studies (8.0%), but in agreement with reports on blood donors of other African countries [70][71][72]. Nevertheless, in Ethiopia, studies in this group showed a marked variation even in the same locality with exceptionally highest [8,14] or lowest [13,15] prevalence rates over time. Apart from geographical location and study time differences, the discrepancies in this group in particular and other groups in general could be due to methodological drawbacks such as small sample size seen from southwest Ethiopia [14] and ill-defined study design from northwest Ethiopia [50]. For instance, the latter study failed to exclude patients who were screened HBsAg for differential diagnosis of other illness, but rather reported 14.4% HBsAg prevalence by considering the subject as 'clinically suspected viral hepatitis patients'. The other source of the discrepancy might be due to the potential variability in sensitivity and specificity of the commercially available test kits used in each study, although the studies used in this review were based on an immunoassay based screening kits with more or less similar principle of antibody detection.
From early 1980s to the present time, more than 120,000 [73] Ethiopian Jews immigrated to Israel. Because of the low standards of socio-economic and healthcare status of these migrants, they attracted the focus of public health intervention and were screened for infectious disease agents including hepatitis viruses [74,75]. The 11.0% HBsAg prevalence rate observed were higher than the blood donors, the general population and other groups. This could be due to the fact that the immigrates were originally from rural Ethiopians communities where risky traditional unsafe practices (tattooing, ear piercing, tonsillectomy, circumcision, ritual scars, traditional surgery, unsterilised shaving at the barber shop, dental extraction at home, home delivery by traditional birth attendants, unsafe abortion practices) are commonly practiced. The finding in this group, however, could represent and reflect HBV prevalence in the rural settings of Ethiopia since the majority of the studies concentrated in urban settings. Moreover, the prevalence observed in these immigrants also could indirectly reflect the potential distribution dynamics of hepatitis viruses to the less endemic parts of the world.
In Ethiopia, reports on HBsAg prevalence were also common from groups such as pregnant women, diabetic patients, street dwellers, HIV VCT clients and commercial sex workers. The HBsAg prevalence was hyperendemic among healthcare professionals [29] and medical waste handlers [30] in particular might be associated with occupational risk exposures since they deal with all Fig. 4 The meta-analysis and forest plot presentation of anti-HCV seroprevalence from 1993 to 2015. (Citations of studies used in the analysis from top to bottom: [9-12, 28, 32, 39, 49, 53-58, 64, 66-68, 98, 99]) sorts of the infected samples with no HBV vaccination, which is the simplest protection and standard practice elsewhere. A recent study showed that although knowledge about availability of HBV vaccine was 62% among healthcare workers, fully vaccinated professionals were only 5.4% [76]. Another study conducted on Ethiopian surgeons showed that most of them were not vaccinated because of negligence and lack of knowledge about the recent introduction of the vaccine, despite their strong belief HBV vaccine is useful [77]. This could explain the reason that the relatively higher prevalence rate of HBsAg and anti-HBS antibody level among healthcare workers in Ethiopia since the exposure and subsequent natural immune development could be expected relatively higher in this group. As the result, to minimize at least an occupational exposure of HBV infection for such risky groups, an urgent need for the HBV vaccination scheme, awareness creation and attitudinal health education on the universal precaution and safety of blood born infection transmissions are crucial in Ethiopia.
The disease burden caused by HBV and other forms of hepatitis is undocumented in Ethiopia, although acute viral hepatitis, chronic hepatitis, cirrhosis of the liver and HCC accounted for significant hospital admissions and mortality rate [78]. In late 1990s, the most common tumour in medical units in Ethiopia was associated with a 50% carrier rate of HBsAg [7,78]. In line to this, the overall HBsAg prevalence in the symptomatic liver disease patients estimated 20.0 and 21.2% from the metaanalysis and systematic review analysis, respectively.
Regarding HCV, the prevalence rate in different time zones was estimated in Ethiopia before as 1.9% [79], 1.3% [80] and 2.7% [81] as part of the HCV epidemiological estimation in sub Saharan African countries. However, the above estimates were limited in terms of numbers of studies included, study time, study groups and geographical representation of the studies available throughout the country. For instance, the former two estimates were made from four studies [9,17,28,56] in which two of them were from blood donors of the same study area. However, in the current study, a comprehensive estimate analysis was made from three groups independently analysed as the general population (includes community based, blood donors, healthcare workers, etc.), symptomatic liver disease patients and HIV infected individuals representing a vast geographical area of the country and time. The overall 3.1% pooled prevalence estimate and the magnitude of anti-HCV antibody from 2.7 to 37.8% in chronic liver disease patients [44,49], therefore, might show the true epidemiological picture and the real burden of HCV in Ethiopia. This is also supported by the level of HCV viraemia reported in 25.4% of HCV antibody positive individuals [66]. The pooled prevalence in this study was higher than the recent reports from neighbouring countries of 0.3% in Djibouti, 0.9% in Somalia, and 1.0% in Sudan [82] but in line with a 3.0% prevalence estimate of most other African countries [79,83]. The risk factors discussed for transmission of HBV in Ethiopia also might play their own roles in the transmission of HCV, although the degree of transmission potential varies for both viruses.
Similarly, unlike HBV and HCV, prevalence studies of HAV, HDV and HEV among the general population and liver disease patients were very few in Ethiopia. The data available so far [47,60,61,84] showed a considerate prevalence particularly for HAV and HEV infections. These viruses are transmitted by the faecal-oral route and many of the environmental and socio-economic factors foster the transmission routes. In Ethiopia, the exposure to HAV at least once in every individual was common and the prevalence rates range from eight to 99% in the general population [48,85]. However, unlike the blood born viral hepatitis, HAV and HEV do not commonly exist in a chronic carrier state [84] and they are very different in exposures/transmissions. In Africa, HEV infection is widespread and is expected to be a threat to numerous lives, especially for those pregnant women and their foetuses [4]. A serological data from Egypt showed that the seroprevalence of anti-HEV can reach close to 100% in the general population [4]; similar to anti-HAV prevalence in Ethiopia and elsewhere in Africa [84,86]. It was also reported that in many countries like Sudan, Chad, Uganda, Kenya, Somalia and Ethiopia a number of large HEV outbreaks occurred at different times [4]. For instance, in Uganda alone, the mortality rate among children younger than 2 years of age was 8% [4]. Similarly, in Ethiopia, the HEV antibody seroprevalence observed in jaundice patients was 93% [48] but the least from acute sporadic patients (33%) [48], pregnant women (59%) and healthy adults (3%) [48]. According to this report, in the early 1990s, HEV was a common cause of acute sporadic viral hepatitis in Ethiopian patients. In particular, the occurrence during pregnancy was associated with high maternal and foetal morbidity and mortality [48]. Although there is no current study available to compare, the late 1980s HDV reports ranged from 5.8 to 9.6% could classify Ethiopia as one of an endemic country. The worldwide estimation also indicated that 5% of HBsAg carriers were infected with HDV [5]. In Ethiopia, an old report showed that the distribution of HBV markers was similar to anti-HDV positive and anti-HDV negative individuals due to the relatively young age of the population and/or hyperendemic condition of the area [60].
Although an increasing pattern of viral hepatitis in general is predictable from HIV infected individuals in Ethiopia, unlike HCV (5.5%), the co-prevalence estimate of HBV (5.2%) with HIV were lower than the estimate of other groups. Similarly, the meta-regression analysis showed a decreasing pattern of overall HBsAg prevalence rate over study year. The intensive prevention and control measures done after the era of HIV pandemic might be responsible for HBV exposure reduction. Therefore, this could might encourage a need for integrating HBV management to the HIV prevention and control programs in Ethiopia. However, unlike HBV, the HCV seroprevalence analysis in the HIV infected individuals was higher up to two folds than HIV negative controls. Regarding the HAV, HDV and/or HEV co-infections with HIV in Ethiopia, there are no data available to compare with monoinfection of each hepatitis virus.

Conclusion
Given the potential limitations arising from small sample size study effects, inconsistency of the screening kits used in the studies as well as study design, time and area heterogeneity, the findings from the systematic review and meta-analysis were in agreement with each other and indicates all forms of viral hepatitis infection are common in Ethiopia. In particular, HBV and HCV and/ or their respective co-infection with HIV are prevalent and endemic. These could be taken as one of the nation's public health problems and potential healthcare threats especially in compromising HIV management. Recent studies on the magnitude of HAV, HDV and HEV are lacking, hence, the need for further studies to generate up-to-date data is recommended. Developing viral hepatitis diagnosis algorithms, treatment schemes and an enhanced HBV vaccination program for children and risk groups in Ethiopia are strongly recommended. The HBV/HCV co-prevalence with HIV supports the need for preventive measures and screening of viral hepatitis in Ethiopia and could be integrated into the country's existing HIV prevention and control programs.