Groome and Richardson have pointed out that the exposure frequency distribution provided in Table 1 of our article is nested.1 This observation is somewhat correct but not completely so. The study was carried out in villages in rural Vietnam, where people frequently keep their own poultry and also have cross-household poultry contacts. Some persons therefore had direct contact with sick or dead poultry when helping or visiting the households of friends and neighbors, without having sick or dead poultry in their own household. Indeed, 412 of the 6702 persons reporting direct contact with sick or dead poultry had that contact within the households of others. For this reason, we analyzed the association between the 3 types of poultry exposure both separately and in the same model. As expected, the separate models gave higher odds ratios for all 3 levels of exposure: 1.15 vs 1.04 for raising and keeping poultry in the household; 1.41 vs 1.14 for reporting sick or dead poultry in the household; and 1.93 vs 1.73 for direct contact with sick or dead poultry. We chose to present the conservative model to not risk overestimating the association. We hope this clarification will be helpful and apologize for not being clear enough in the article. As for the confounders, socioeconomic status was the most important one, and a clear strength of our study was the possibility to adjust for this factor.
Thorson A, Petzold M, Chuc NTK, Ekdahl K. Flulike Illness and Exposure to Sick or Dead Poultry—Reply. Arch Intern Med. 2006;166(13):1421. doi:10.1001/archinte.166.13.1421-a