Helicobacter pylori infections in Ethiopia; prevalence and associated factors: a systematic review and meta-analysis

Background Helicobacter pylori (H.pylori) infections are prevalent and recognized as major cause of gastrointestinal diseases in Ethiopia. However, Studies conducted on the prevalence, risk factors and other clinical forms of H.pylori on different population and geographical areas are reporting conflicting results. Therefore, this review was conducted to estimate the pooled prevalence of H.pylori infections and associated factors in Ethiopia. Methods PubMed, Embase, Google scholar, and Ethiopian Universities’ repositories were searched following the Preferred Items for Systematic review and Meta-analysis (PRISMA) guideline. The quality of included studies was assessed using the Newcastle-Ottawa Scale in meta-analysis. Heterogeneity between studies was assessed using Cochrane Q test and I2 test statistics based on the random effects model. Comprehensive meta-analysis (CMA 2.0) and Review Manager (RevMan 5.3) were employed to compute the pooled prevalence and summary odds ratios of factors associated with of H.pylori infection. Results Thirty seven studies with a total of 18,890 participants were eligible and included in the analysis. The overall pooled prevalence of H.pylori infection was 52.2% (95% CI: 45.8–58.6). In the subgroup analysis by region, the highest prevalence was found in Somalia (71%; 95% CI: 32.5–92.6) and the lowest prevalence was reported in Oromia (39.9%; 95% CI: 17.3–67.7). Absence of hand washing after toilet (OR = 1.8, 95% CI; 1.19–2.72), alcohol consumption (OR = 1.34, 95% CI; 1.03–1.74) and gastrointestinal (GI) symptoms (OR = 2.23, 95% CI; 1.59–3.14) were associated with H.pylori infection. The trend of H.pylori infection showed a decreasing pattern overtime from 1990 to 2017 in the meta-regression analysis. Conclusion The prevalence of H.pylori infection remains high; more than half of Ethiopians were infected. Although the trend of infection showed a decreasing pattern; appropriate use of eradication therapy, health education primarily to improve knowledge and awareness on the transmission dynamics of the bacteria, behavioral changes, adequate sanitation, population screening and diagnosis using multiple tests are required to reduce H.pylori infections. Recognizing the bacteria as a priority issue and designing gastric cancer screening policies are also recommended. Electronic supplementary material The online version of this article (10.1186/s12876-018-0927-3) contains supplementary material, which is available to authorized users.


Background
Helicobacter pylori have been found to infect about half of the world's population [1][2][3][4][5]. The prevalence of H.pylori infection varies globally with a greater prevalence generally reported from developing countries. The global estimate of H.pylori infection was reported at 48.5% while continental reports were 69.4% in South America, 37.1% in North America, 24.4% in Oceania, 54.6% in Asia, 47.0% in Europe and 79.1% in Africa [2,6]. This difference has been related to geography, age, ethnicity, socioeconomic factors, and methods of diagnosis and eradication therapy [1,6,7]. Diseases associated with H.pylori infections are commonly occur at earlier ages in developing countries [1][2][3][8][9][10].
In 2017, World Health Organization (WHO) has published lists of 16 bacteria that pose the greatest risk for human health. H.pylori was thus categorized as a high priority pathogen for research and development of new and effective treatments [35]. In addition, recommendations are emerging to change approaches to management of H.pylori due to increased drug resistance [30,31,[36][37][38]. The success of these developments needs knowledge of prevalence of H.pylori.
In Ethiopia, the prevalence of H.pylori infection ranged from 7.7% [39] to 91% [40]. It is highly prevalent and recognized as major cause of gastrointestinal diseases. Studies conducted on the prevalence, risk factors and other clinical forms of H.pylori on different population and geographical areas are reporting conflicting results. Socioeconomic factors, sanitation, crowded living conditions, unsafe food and water, ethnicity as well as poverty can contribute to H.pylori infections [41]. Studies published on the prevalence of H.pylori in Ethiopia dated back to the 1990's [42]. However, comprehensive review has not been done on its prevalence and associated factors in Ethiopia. Therefore; this study was done to estimate the pooled prevalence of H.pylori infection and associated factors in Ethiopia.

Data bases and search strategy
PubMed, Embase, and Google scholar were searched to identify potential articles on H.pylori infections in Ethiopia. To include unpublished studies, Ethiopian University repositories were searched and reference lists of eligible studies were searched to maximize inclusion of relevant studies. The search was conducted following the PRISMA guideline and checklists ( [43], Fig. 1). The following terms with MeSH (Medical Subject Headings) and Boolean operators were used to search PubMed; Helicobacter pylori OR H.pylori OR Campylobacter pylori OR C.pylori OR gastritis OR gastric cancer OR gastric carcinoma OR peptic ulcer disease OR PUD OR duodenal ulcer OR dyspepsia OR mucosa associated lymphoid tissue OR MALT AND Ethiopia. The search was limited to English language publications and done independently by each reviewer to minimize bias and the missing of studies. Search results were combined in to EndNote X6 file (Clarivate Analytics USA) and duplicates were removed. All articles published up to June 30, 2018 were included in the review if fulfilled the eligibility criteria (Table 1).

Quality assessment
The quality of included studies was assessed by using the Newcastle-Ottawa Scale (NOS) (Additional file 1). Two authors independently assessed the quality of each study. Disagreements between authors was resolved by discussion and articles were included if agreement was reached between the authors.

Data extraction
Data were extracted into customized Microsoft Excel. Data extracted from each included study include; author, publication year, study area, study period, study design, study population, sample size, laboratory tests used, prevalence and/or number of H.pylori cases. We have also contacted corresponding authors of the included studies for missing data though no one responded.

Data analysis
Studies providing data on crude prevalence of H.pylori or numbers of cases and study participants were included in the meta-analysis. Prevalence for individual studies was determined by multiplying the ratio of cases to sample size by 100. The estimation of pooled prevalence and summary odds ratios of H.pylori infection was done using CMA 2.0 and RevMan 5.3 softwares. Subgroup analyses were done by study period, study region, study design, laboratory tests used (types and numbers) and publication history. With the assumption that true effect sizes exists between eligible studies, the random effects model was used to determine the pooled prevalence, summary odds ratios and 95% CIs. Significant association between H.pylori infections and potential factors was declared at p-value < 0.05. Heterogeneity was evaluated using the Cochran's Q test and I 2 statistics. Significant heterogeneity was declared at I 2 > 50% and Q-test (P < 0.10).

Publication bias and sensitivity analysis
Funnel plots were drawn to assess the possibility of publication bias. We plotted the studies' logit event rate and the standard error to detect asymmetry in the distribution. A gap in the funnel plot indicates potential for publication bias. In addition, Begg's adjusted rank correlation and Egger's regression asymmetry tests were used to assess publication bias, with P < 0.05 considered to indicate potential publication bias. Sensitivity analysis, leave-one-out analysis was done to assess the prime determinant of the pooled prevalence of H.pylori infection and to detect the possible causes of heterogeneity between studies.

Pooled prevalence of H.pylori
A total of 18,890 Ethiopians were participated in the study; out of which 8979 were infected with H.pylori in the period under review giving an overall pooled prevalence of 52.2% (95% CI: 45.8-58.6; I 2 = 51.05%, p = 0.503) (Fig. 2). Sensitivity analysis revealed no significant  Unpublished difference both in the pooled prevalence and heterogeneity. When one study was excluded from the analysis step-by-step, the pooled prevalence was between 50.5 and 53.5% while heterogeneity was similar (I 2 = 53.5%). Drawing of funnel plot supported with Egger's regression (p = 0.172) and Begg's correlation (p = 0.367) tests showed no evidence of significant publication bias (Fig. 3).

Socio-demographic risk factors
Available sociodemographic data (by age, sex, residency and level of education) were extracted and analyzed to determine their possible association with H.pylori infection but none of these variables had significant association. Even though not significant; male participants (OR = 1.07; 95% CI: 0.93-1.23; p = 0.33) and urban residents (OR = 1.04; 95% CI: 0.74-1.74; p = 0.83) were more likely to be infected with H.pylori than their counter parts (Fig. 4).

Environmental factors
Sources of drinking water, hand washing before meal and after toilet were the environmental factors assessed for their possible association with H.pylori infection. Participants who were not washing their hands after toilet were more likely to be infected with H.pylori (OR = 1.8; 95% CI: 1.19-2.72; p = 0.005). Other variables had no significant association (Fig. 5).

Behavioral factors
Chat chewing, cigarette smoking and drinking alcohol were analyzed for any association with H.pylori infection.  (Fig. 6).

Clinical factors
Gastrointestinal (GI) symptoms, allergic reactions, hyperemesis gravidarum, HIV and TB infections were some of the clinical factors reported with H.pylori infection by included studies. Because studies on allergic reactions, hyperemesis gravidarum, HIV and TB infections are small enough to compute pooled summary of odds ratios, we analyzed only the association between GI symptoms and H.pylori infection. Hence; participants who had GI symptoms (including dyspepsia, gastritis, peptic ulcer and related) were more likely to be infected with H.pylori (OR = 2.23; 95% CI: 1.59-3.14; p < 0.00001) (Fig. 7).

Meta-regression
Meta-regression was done to explore the trend of prevalence of H.pylori by year of study and sample size of the included studies. A significant downward trend of H.pylori infection was observed from 1990 to 2017 (B = − 0.067, p = 0.00004). However; there was no significant association between prevalence of H.pylori and sample size of the studies even though there was a slight decrease of prevalence of H.pylori with increased sample size (B = − 0.00079; p = 0.193) (Fig. 8).

Discussion
The global prevalence of H.pylori infection was estimated at 48.5% in 2017 [6]. The World Gastroenterology Organization (WGO) in its 2011 global guideline    and Estonia (82.5%) [6]. Some reasons may explain the lower prevalence in Ethiopia. Firstly; the number of studies and participants included; pooled from 37 studies and 18,890 participants in our analysis compared to fewer number of studies and participants. Second, our analysis included recent reports which showed a decreasing trend to recent times. Third, most of the laboratory tests used by studies included in our review were based on stool antigen for detection of H.pylori; which has low isolation rate.
However, our estimate is higher than other countries; 22.1% in Denmark, 43.6% in Thailand, 46 . This decrement might be related with relative improvements in sanitation, water access, life style and behavioral changes, quality of life and socioeconomic status, and increased awareness on the transmission, diagnosis, eradication therapy, prevention and control of H.pylori infection.
Regional estimates of H.pylori infection in the subgroup analysis showed a lower prevalence of 39.9% in Oromia and higher prevalence of 71.0% in Somalia region. This regional difference can be attributable that in Somalia, study participants were University students with known gastritis which is a known risk factor for H.pylori infection. Other explanations could be related with sociodemographic, socioeconomic, environmental, clinical, behavioral factors and number of studies included in each category.
The prevalence of H.pylori infection differs on the bases of laboratory tests used. Higher prevalence was observed when detection is supplemented with sensitive tests including PCR, culture, rapid urease test and histopathology as shown in (Table 3). This is supported by studies comparing diagnostic tests for H.pylori that sensitive tests improve the detection rate of H.pylori infections from clinical samples [74][75][76][77]. In addition, combination of at least two diagnostic methods is recommended to increase the validity of results [16,20,28,48,63,[78][79][80][81][82], but only ten studies used multiple tests to make a definitive diagnosis of H.pylori in our analysis.
Our subgroup analysis confirms that the pooled prevalence of H.pylori infection when multiple tests are used is higher than the pooled prevalence when a single test is used to detect H.pylori infection (62.9% Vs 48.1%). Several risk factors for H.pylori infection were identified and reported usually with conflicting result. These factors are to be analyzed and pooled to have a summary of effect sizes. The results of this meta-analysis showed that participants with gastrointestinal (GI) symptoms were more likely to be infected with H.pylori (OR = 2.23; 95% CI: 1.59-3.14). This could be due to the effect of GI symptoms providing a growing medium (changing pH, thinning of gastric wall, gastric ulceration, change in gut microbiota) for the bacteria.
Although the type and level of alcohol, amount and frequency of consumption were not described, individuals taking alcohol were more likely to be infected with H.pylori infection than those who did not consume alcohol. This result is inconsistent with previous studies [83][84][85][86] reporting that alcohol consumption has either a protective effect or has no any relation with H.pylori infection. Other study [87] reported alcohol as a risk factor for H.pylori supporting our analysis. As alcohol is known to directly damage the gastric mucosal layer, it is theoretically possible that alcohol can provide ways for H.pylori infection. In addition, heavy drinking can possibly predispose consumers to social contacts that favor transmission of H.pylori infection. Other mechanism may be involved in the synergistic effect of alcohol including bacterial adherence and host factors in facilitating infection among drinkers. Further longitudinal and epidemiological studies are needed to test these explanations.
Previous studies [1,86,88] have reported that the prevalence of H.pylori infection seems to increase with age but the increment with age is assumed most likely due to cohort effect. As most infections are acquired early in life; the bacteria usually persists indefinitely unless treated with specific antibiotic. In our analysis, consistent with the assumption, participants in the age of <20 years were less likely to be infected when compared to above 20 years of age but the association is not significant (OR = 0.85; 95% CI: 0.70-1.03).
Poor sanitation and unsafe food and water were repeatedly reported as risk factors contributing for H.pylori infection. In our analysis, individuals who were not washing their hands after toiled were more likely to be infected with H.pylori. This is in agreement with previous studies [1,41] and can be explained that H.pylori is largely transmitted through feco-oral or oral-oral routes. Lack of proper sanitation and basic hygiene after toilet, therefore; can be source of infection and increase the chance of acquiring H.pylori. World Health Organization (WHO) has also identified it as one of the greatest risks for human health and categorized it as a high priority pathogen for research and development of new and effective treatments [35].
In the meta-regression analysis, the prevalence of H.pylori infection showed a significant downward trend from 1990 to 2017 (B = − 0.067, p = 0.00004). This can be due to relative improvements in sanitation, water access, lifestyle and behavioral changes, quality of life and socioeconomic status, increased awareness on the transmission, diagnosis, eradication therapy, preventive and control of H.pylori infection. On the other hand; a slight decrement of prevalence of H.pylori was observed with increased sample size (B = − 0.00079; p = 0.193) but the association was not significant. This might be related with the numbers of studies done on large sample sizes in that only few studies were done when compared to numbers of studies done on small sample sizes.

Strengths and limitations
The strength of this study is that it included relatively larger numbers of published and unpublished studies without time limit of publication year. It has provided the national pooled prevalence, trends of infection and identified factors associated with H.pylori infection. The study included sub-group analysis and meta-regression of differences of study area, laboratory tests, study period, sample size and study designs. In few regions of Ethiopia, no study was found from and most identified studies were hospital based which could affect the generalisability of our findings. Moreover, there was lack of data sets to investigate the association between H.pylori infection and possible risk factors and the outcome variables may be affected by other cofounders.

Conclusion
The prevalence of H.pylori infection remains high; more than half of Ethiopians were infected and significant association was observed between H.pylori infection and absence of hand washing after toilet, alcohol drinking and presence of gastrointestinal (GI) symptoms. These findings strengthen the action to implement the control and prevention of H.pylori infection more effectively to prevent gastric cancer and other related complications in Ethiopia. Although the trend of infection showed a decreasing pattern; appropriate use of eradication therapy, health education primarily to improve knowledge and awareness on the transmission dynamics of the bacteria, behavioral changes, adequate sanitation, population screening and diagnosis using multiple tests are required to reduce H.pylori infections.. Recognizing the bacteria as a priority issue and designing gastric cancer screening policies are also recommended.