Effect of early and current Helicobacter pylori infection on the risk of anaemia in 6.5-year-old Ethiopian children

Background: Epidemiological and clinical studies in high income countries have suggested that Helicobacter pylori (H. pylori) may cause anaemia, but evidence is lacking from low income countries. We examined associations between H. pylori infection in early childhood and anaemia at the age of 6.5 years in an Ethiopian birth cohort. Methods: In 2011/12, 856 children (85.1 % of the 1006 original singletons in a population-based birth cohort) were followed up at age six and half. An interviewer-led questionnaire administered to mothers provided information on demographic and lifestyle variables. Haemoglobin level and red cell indices were examined using an automated haematological analyzer (Cell Dyn 1800, Abbott, USA), and stool samples analyzed for H. pylori antigen. The independent effects of H. pylori infection (measured at age 3.5 and 6.5 years) on anaemia, haemoglobin level, and red cell indices (measured at age 6.5 years) were determined using multiple logistic and linear regression. Results: The prevalence of anemia was 34.8 % (257/739), and the mean (SD) haemoglobin concentration was 11.8 (1.1) gm/dl. Current H. pylori infection at age 6.5 years was positively, though not significantly related to prevalence of anaemia (adjusted OR, 95 % CI, 1.15; 0.69, 1.93, p = 0.59). Any H. pylori infection up to age 6.5 years was significantly associated with an increased risk of anaemia at age 6.5 (adjusted OR, 95 % CI, 1.68; 1.22, 2.32, p = 0.01). A significant reduction in haemoglobin concentration and red cell indices was also observed among children who had any H. pylori infection up to age 6.5 (Hb adjusted β = -0.19, 95 % CI, -0.35 to -0.03, p = 0.01; MCV adjusted β = -2.22, 95 % CI, -3.43 to -1.01, p = 0.01; MCH adjusted β = -0.63, 95 % CI, -1.15 to – 0.12, p = 0.01; and MCHC adjusted β = -0.67, 95 % CI, -1.21 to -0.14, p = 0.01), respectively. Conclusion: This study provides further evidence from a low income country that any H. pylori infection up to age 6.5 is associated with higher prevalence of anaemia, and reduction of haemoglobin level and red cell indices at age 6.5. A detailed description of the original Butajira birth cohort study has been published [27, 30]. Briefly, the birth cohort is nested in the Butajira Demographic Surveillance Site (DSS) which covers a sample of nine rural and one urban administrative units in and around the town of Butajira in Southern Ethiopia [31]. Between July 2005 and February 2006, all women in the DSS aged 15–49 and in their third trimester of pregnancy were identified by the DSS fieldworkers and invited to participate in the study. Of the 1,234 eligible women, 1,065 were recruited (86 % of those eligible) and all live singleton babies born to these women (n = 1006) were followed-up as a birth cohort. After informed consent forms were signed by the mothers, information on demographic and selected lifestyle factors was collected by interviewer-led administered questionnaire during pregnancy: information on mother’s age, place of residence, ethnicity, religion, occupation, education and household income was collected. At birth and during the follow-up visits, the project data collectors visited the child at home and collected information on potential confounders such as birth weight, history of vaccination, household size, vitamin A supplementation, intestinal parasitosis, anthropometric characteristics and sanitary conditions. At follow up visits at ages 3, 5 and 6.5 years, mothers were also asked to collect a faecal sample from their child using a leak-proof plastic container. The samples were then transported for analysis in the Butajira health center laboratory to ascertain the child’s H. pylori and intestinal parasites infection status. Furthermore, at the 6.5 year follow up visit, a blood sample was collected from each child using a vacutainer tube, and transported to Butajira hospital for haematological analysis. H. pylori status was evaluated using the commercially available SD Bioline H. pylori stool antigen test (Standard Diagnostics, Inc) according to the manufacturer’s instructions. A portion of faeces (about 50 mg) from a stool sample was swirled with assay diluent solution at least for ten times, until the sample has been dissolved, and then allowed to settle for 5 min at room temperature. About 100 μL of the prepared sample was placed on the H. pylori Ag examination device. The test results were checked about 15 min later. One red line indicated negative and a double red line indicated an H. pylori positive result. Additionally, all faecal samples were examined qualitatively using the modified formol-ether concentration method to ascertain the child’s intestinal parasites infection status. At the 6.5-year follow-up, a two ml whole blood sample was collected into Ethylene diaminetetraacetic acid (EDTA) tubes between 8:00 and 10:00 am and analyzed on the same day using an automated haematological analyzer (Cell Dyn 1800, Abbott, USA) at Butajira hospital. The analyzer aspirates the blood sample, dilutes and counts leukocytes, erythrocytes and thrombocytes, measures Mean Cell Volume (MCV) and Haemoglobin (Hb), and calculates Haematocrit, Mean Cell Haemoglobin (MCH), and Mean Cell Haemoglobin Concentration (MCHC). This instrument was monitored daily with normal, high and low controls provided by the manufacturer before running the specimen to ensure quality of haematological analyses. The primary study outcome was anaemia at age 6.5, and was defined according to the WHO hemoglobin cutoff: < 11.5 g/dL for children 5–11 years [32]. Data were double-entered into EpiData 3.1 (EpiData, Denmark). The datasets were cleaned, coded and merged ready for analysis using Stata 12 (Statacorp, College Station, Texas, USA). Prior to investigating the association between H. pylori infection and anaemia, univariate analyses were used to identify the possible confounders. Variables that were associated with both exposure and outcome variables in the crude analysis using statistical significance at p value <0.2 were considered to be possible confounders. These included place of residence, ethnicity, religion, maternal education, source of water, crowdedness, sanitary conditions, history of vaccination and history of vitamin A supplementation. Additionally, we included variables previously shown to be associated with anaemia in the literature; child’s sex, anthropometric measures of nutritional status and intestinal parasite status [33]. The hypothesis that infection with H. pylori would be associated with anaemia (a categorical variable) was assessed using univariate and multivariate logistic regression models. To evaluate the effect of H. pylori on haemoglobin and red cell indices as continuous outcomes (Y), regression analyses were performed using generalized linear models (GLIM) with random response variables, Hb, RBC, MCV, MCH and MCHC assuming a gamma distribution for Y. Maximum likelihood allowed estimation of column vector of coefficients, β, in the function f(μY) = Xβ, where f( ) is the link function. We chose the logarithm link (log(μY) = Xβ), based on the optimization of both the Akaike Information Criterion and the Bayesian Information Criterion [34]. We first examined the crude association between child's H. pylori infection (measured at age 3, 5, 6.5 years) and outcome variables (anaemia, haemoglobin and red cell indices) at age 6.5 years. We then repeated the analysis, adjusting for the possible confounders and predictors of outcome listed in Additional file 1: Table S1. The same approach was used in a separate set of analyses to assess the association between H. pylori infection and anaemia, haemoglobin level and red cell indices for all available children at year 6.5, by creating a new exposure variable with categories representing different combinations of H. pylori exposure status at age 3, 5 and 6.5: ‘never infected’ (never infected at any of these three time points) and 'infected any age up to year 6.5’. Covariates were kept in the model if they changed the coefficient of exposure (H. pylori infection) by > 10 % or if they were independently associated with the outcome at p < 0.10. Probability values < 0.05 were considered statistically significant for main effects. Sensitivity analysis was done to compare the distribution of demographic and life style variables between study subject who have complete outcome data (i.e. “complete-case”) and "all respondents" populations. The study was approved by the Institutional Review Board (IRB) of Addis Ababa University, College of Health Sciences, Ethiopia. Written, informed consent was obtained from the mothers after they have been clearly informed about the study, and in keeping with the requirements of the College of Health Sciences IRB all women and their children were reimbursed for health care costs. Children were also requested to give assent and were informed of their right to refuse to participate in the study and to withdraw at any time during the study without jeopardizing their right of access to other health services. Invasive procedures such as collection of blood samples were fully explained to parents and children, and were carried out using sterile disposable materials.