The present study was approved by the Research Ethics Boards of the University of Toronto (Toronto, Ontario) and Mount Sinai Hospital (Toronto, Ontario). Blood samples from the Ontario Familial Colon Cancer Registry (OFCCR) were used to estimate the prevalence of H pylori
infection. The OFCCR collects family history information, epidemiologic data and blood samples from a population-based sample of colorectal cancer patients and controls. Population controls are identified using random selection through Info-direct (Bell Canada, Canada), a service of Bell Canada that provides a listing of residential telephone numbers in Ontario. If there is more than one eligible household member (matched by sex and five-year age group with OFCCR case distribution), then one person is randomly selected and asked to participate. The methodology for the OFCCR has been described in detail elsewhere (14
). For the purpose of the present study, the blood samples taken from the control population, which were stored in the biospecimen repository of Mount Sinai Hospital, were used. Ten microlitres of plasma were aliquoted from each of the 1306 samples (adults aged 50 to 80 years). H pylori
-specific immunoglobulin G antibody titres were measured by a validated ELISA using the DRG kit (DRG International Inc, USA) in the robotics laboratory at Mount Sinai Hospital. Performance data for this kit showed a sensitivity and specificity of 99% and 97%, respectively (A Azad, personal communication).
Using family history and epidemiological questionnaires, information was obtained on past and present medical history, smoking and drinking habits, socioeconomic status, number of siblings, education level, ethnicity and place of birth, as well as the consumption of fruits, vegetables and meat. Nonwhites included blacks (from Africa, the Caribbean and North America) and those from the Middle East and Asia. Low education levels corresponded to completion of high school or lower, moderate education levels corresponded to completion of technical school or college, and higher education levels corresponded to completion of bachelor’s degree or higher. Data on fruit, vegetable and meat intake referred to patient diets two years before completion of the questionnaire. The alcohol consumption for patients between 30 and 40 years of age, 41 and 59 years of age and 60 years of age or older referred to the consumption during their 20s, 30s and 40s, and since they had turned 50 years of age, respectively.
The prevalence of H pylori infection was estimated separately for each sex. Weighted prevalence estimates were obtained using sampling weights calculated as the inverse of the sampling fractions to correct for the sampling strategy. In weighting, the distribution of the 2003 Ontario population by sex and five-year age group was used as a reference (Statistics Canada, 2003).
The relationships between the prevalence of infection and the various sociodemographic factors were assessed by calculating age-adjusted ORs and 95% CIs, using logistic regression with incidence of infection as the outcome measure. In all the regression analyses, age was used as a categorical variable because the reference population that was used for weighting the estimates was grouped by age and sex.
In multivariate analyses, the dependent variable was seropositivity for H pylori, and covariate variables included available sociodemographic and lifestyle factors. The stepwise solution was used, which combined forward and backward solutions and therefore overcame the limitations associated with each. Stratified analyses suggested potential effect modification by sex on the association between H pylori seropositivity and various factors; thus, interaction terms for such factors were tested in multivariate logistic models. Intervariable correlations were evaluated before modelling. Place of birth and age at immigration were significantly correlated (r2=0.98) and were combined. The new variable had three categories: born in Canada, immigrated to Canada at younger than 20 years of age and immigrated to Canada at 20 years of age or older. The associations between H pylori infection and the various factors considered in the present study were not weighted.
Most of the data analyses were performed using SPSS version 12.1 (SPSS Inc, USA). STATA version 8.0 (StataCorp LP, USA) was used to calculate the weighted prevalence estimates and their 95% CIs.