Study population, setting and design
The current study focuses on a population under transition; the Israeli Arab population. This population has unique characteristics, in terms of infrastructure, health care and education systems which are similar to those existing in developed countries, while the rates of H. pylori infections and anemia are comparable to those reported from developing countries. The Israeli Arab population comprises 20% of the Israeli population . The Israeli Arabs reside mostly in separate locations than the Jewish population, and usually in rural areas. The Israeli Arab population has lower educational levels and socioeconomic status as compared with the Jewish population , nevertheless this population is in positive transition, with ongoing improvement of the educational level and medical system. Israeli Arabs have mandatory health insurance according to the national health insurance law. The vaccination coverage in this population is over 95%.
This retrospective cohort study was conducted in 2007-2009, among children who participated in a previous project on H. pylori infection in 2004, when they were 3-5 years of age. Fifty percent of the children were H. pylori positive at this age . Families of these children live in three villages in northern Israel. There are about 150,000 Muslim Arab inhabitants living in this region, with 3914 live births in 2007 . Two of the villages have approximately 10,000 residents, and the third one is inhabited by about 14,000 residents. According to the Central Bureau of Statistics, one village belongs to cluster 2-socioeconomic status (SES), one belongs to cluster 3-SES, and the third village belongs to cluster 4-SES (for more details on the study villages see additional file 1). The clusters are on a scale of 1-10, the lower the index, the lower the SES . At the national level, these villages are of low and intermediate SES levels , but given the variation among them, they were labeled in the present study as low, intermediate and high SES village. Drinking water supply in these villages is piped, and all households are connected to the national electricity company similarly to the rest of the country. Connection to the cable television and internet networks is also available. The educational system in these villages includes kindergartens, primary and high schools. The three villages were selected to represent different socioeconomic background within the Arab population. The characteristics of the selected villages are similar to the Israeli Arab population. For example the median age in the Israeli Arab population is 20 years , as compared with 18-21 years in the three villages . 34% of the families in the Israeli Arab population have ≥6 persons, and 21% of the women hold a job , as compared with 33% and 24%, respectively in the study sample. The mean number of rooms per a household is 3.7, and the median year of schooling is 11.3-12.0 in the Israeli Arab population [22, 26], as compared with 3.8 and 10 years, respectively in the study sample.
In the original study, we used cluster sampling procedure, in which 9 kindergartens (3 per village) were sampled from the kindergartens in each village. Parents of all children from each selected kindergarten were offered to participate in the study, through personal meetings at the candidates' homes.
In the current study, children born at a gestational age of 34 week or more and a birth weight of 2 kg or more were eligible to participate in the study. Among 289 participants of the 2004 study, 3 relocated their residence place, 5 could not be located, 1 child deceased due to cancer, 2 could not participate since their mothers deceased during the study period, 7 children were excluded due to birth weight of less than 2 kg or birth week less than 34. Nine additional children were excluded due to thalassemia minor (3 children), type-1 diabetes (1 child), Glucose-6-phosphate dehydrogenase deficiency with anemia (1 child), major heart defect (1 child), panhypopituitarism (1 child), hemophilia (1 child), and significant developmental delay requiring therapy (1 child). These conditions might affect cognitive function directly or might be associated with other conditions related with cognitive function e.g. hemoglobin levels. Among parents of 263 eligible children who were contacted through home visits, 41 refused to participate in the study and 222 consented, of these, 200 complied with the study procedures (i.e. compliance rate of 76%).
The Institution Review Boards of Tel Aviv University and of Hillel Yaffe Medical Center approved the study. Written informed consent was obtained from the parents' participants.
Additional independent variables
Current hemoglobin levels
Blood collected by finger lancing was used for hemoglobin measurement employing a portable hemoglobinometer (Hemocue Hb 201+, Sweden).
Hemoglobin levels in early childhood
Infants in Israel are screened for iron deficiency anemia at the age 9-18 months, and the results of the participants' tests were collected from medical records.
Anthropometric measurements were performed by specially trained registered nurses. Body weight was measured to the nearest 0.1 kilogram using an analog scale (calibrated before use), and height (to the nearest 0.1 centimeter) with a stadiometer. Information on anthropometric measurements in early childhood (ages 18-30 months) was obtained from medical records. Z scores of height for age (HAZ), weight for height (WHZ), and Body Mass Index for age (BMIZ) were calculated using Epi/Info software (Center for Disease Control and Prevention, Atlanta, Georgia (CDC)). The calculations were based on the 2000 CDC growth reference curves, which were primarily based on the US National Health Examination (NHES) and the National Health and Nutrition Examination Surveys (NHANES). BMI was calculated as: weight (kg)/height (m)2.
Socioeconomic status (SES)
SES was assessed by several parameters: (1) community SES rank as classified by the Israel Central Bureau of Statistics, (2) household socioeconomic characteristics: (a) maternal education, (b) paternal education, (c) crowding index, and (d) reported family income.
In addition, a composite variable of individual level SES was created using the parameters: maternal education, paternal education, monthly family income, and crowding index. The summative scoring of this composite index was as following: each child was accredited one point if maternal education level was ≥10 years and 0 points if maternal education level was <10 years, one point if paternal education level was ≥10 years and 0 points if paternal education level was <10 years, one point if the monthly family income was >4000 New Israeli Shekels (NIS) and 0 points if the monthly family income was ≤4000 NIS, one point if the crowding index was below the median level (1.61 persons/room) and 0 if the crowding index was ≥1.61. The higher the summative score, the better the socioeconomic status. Scoring below the median level was defined as low socioeconomic status, while scoring the median level or higher was classified as high socioeconomic status.
Differences between the villages in the independent and the outcome variables were examined using Chi square test and one way analysis of variance (ANOVA). The difference in the mean IQ levels between H. pylori infected children and uninfected ones was examined using Student t test. Student t test was also used to examine the difference in IQ scores in relation to sex and categorical socioeconomic characteristics. Pearson coefficients were calculated to examine the correlations between IQ levels and independent continuous variables (current hemoglobin levels, hemoglobin levels in early childhood, HAZ and WHZ scores in early childhood, and current BMIZ scores). Multiple linear regression models were used to obtain adjusted β coefficients of effect estimates, while controlling for other covariates in the models. Variables that were associated with IQ scores in the univariate analysis (P < 0.1) were included in the multivariate analysis. Additional multivariate analyses were performed, while including in the model H. pylori infection, the composite SES index, hemoglobin levels and current BMIZ score as a measure of nutritional status. Since socioeconomic features might affect cognitive function and given the differences in socioeconomic status among the three villages, we hypothesized that IQ scores might also differ among the villages. In addition, the three villages differed significantly in the prevalence of H. pylori infection, being highest in the low SES village . Thus the statistical analyses were stratified by village of residence. In all analyses two tailed P < 0.05 was considered statistically significant. Data were analyzed using SPSS software (SPSS Inc, Chicago, IL) version 17.