美国伤寒分析_培训课件.ppt

* The primary measure of association was a matched odds ratio. Attributable risk was estimated by the following formula. We assumed that the odds ratio was a reasonable approximation of the risk ratio and, since ours was a population based study,that the controls were representative of the Fayoum population. * Patients and controls were similar in respect to several demographic characteristcs, including: -median age -the percent of females in each group -the number of people living in the house -the percent of heads of households who were farmers -and the percent of houses with a TV We therefore concluded that the two groups were comparable. * There were no food exposures that were significantly associated with illness. This table shows examples of several of the many individual and grouped food items that were evaluated. Leafy vegetables include lettuce, cabbage, and spinach. Melons include watermelon and cantaloupe. These two categories represent items grown low to the ground which we hypothesized are more likely to come into contact with contaminated irrigation water. Neither milk nor ice cream was found to be associated with illness. Questions about shellfish were not included as there was no shellfish consumption reported on hypothesis generating interviews. Cases were more likely to eat food from a street vendor than controls, though the association was not statistically significant. So although we looked, we found no evidence that food was a significant risk factor for illness. * In contrast, several water related exposures were significantly associated with illness. Patients were less likely than controls to have piped water to their homes. They were more likely to store drinking water in a container at home and more likely to consume ice. Patients were also more likely than controls to wash their utensils in an open canal and less likely to have soap for hand washing at home. * We examined these exposures in a conditional logistic regres