Excluding group with potentially higher rates of infection with H pylori could bias estimated odds ratio

EDITOR—We are writing in response to the paper by Danesh et al on infection with Helicobacter pylori and early onset myocardial infarction.1 We believe that this thorough study is of interest because of its large sample size, and inclusion of sibling pairs and young people. We would, however, like to address the following points. Firstly, Danesh et al did not explain clearly how the controls were selected for the early onset case-control study. We concluded that they were chosen from the pool of spouses and relatives of the cases, which would make them unrepresentative of the general population.

Secondly, Danesh et al say that exclusion of cases with any history of gastrointestinal bleeding or peptic ulceration (both of which could be caused by certain cytotoxic strains of H pylori) would not have caused any underestimation as controls who reported these conditions were also excluded. Excluding this group with potentially higher rates of infection with H pylori could, however, bias the estimate of the odds ratio. For example, considering the case-control study, if we assume the same numbers (183) of controls as cases were excluded, and that 50% of the excluded cases and controls were positive for H pylori, the odds ratio calculation would be as in the table.