All case–control studies are susceptible to selection and
information biases. To minimize these biases, we employed
unambiguous case and control definitions, trained the interviewers,
and pre-tested the questionnaire. Also, food items
consumed by controls were ascertained for the same time period
as for the matched case. A major limitation of the study, however,
was a delay in initiating the investigation, resulting in delayed
identification of the vehicle and testing of suspected food items,
including raw eggs. Another limitation was that serotyping of
bacteria was not performed by laboratories because of the limited
capabilities of provincial laboratories as well as a lack of
knowledge and skills of local health workers in specimen collection
during the investigation. Training in sound epidemiological
investigation methods and laboratory capacities for health
personnel are therefore critical in resource-poor settings.