Prevalence of shigellosis in the U.S.: consistency with dose-response information

Every year there are estimated 300 000 cases of Shigella in the United States (Bennett et al., 1987, Am. J. Prev. Med. 3, 102–114). A beta-poisson model was fit to human dose-response information on pathogenic Shigella using the Maximum Likelihood Estimation technique (Haas, 1983, Am. J. Epidemiol. 118, 573–582). Pooled and separate data sets for the Shigella species were fit to the beta-Poisson model and 95% confidence limits and regions were calculated. Shigella dysentariae and Shigella flexneri confidence regions and limits overlapped with each other and with the pooled data set, suggesting that this model can describe Shigella in general. The pooled Shigella model as well as the upper and lower confidence limits of the three data sets showed average exposures based on the estimated U.S. caseload of pathogenic Shigella of 0.01 to 0.014 organisms (confidence limits 0.001–0.05) for a 7-day per annum period of exposure and ranges from 0.07 to 0.1 organisms (confidence limits 0.006–0.4). for a 1-day per annum period of exposure. The plausibility of the pooled dose-response model was then evaluated by comparison with two known cruise ship outbreaks. The pooled model estimated that the two outbreaks studied could have been due to ingestion of 344 (confidence limits 72–915) Shigella cells per meal and 10.5–12 (confidence limits 1–44) Shigella cells per glass of water by passengers.