We used a cross-sectional study design, sampling 1242
poultry and 807 wild birds in 16 villages across four
governorates in Lower Egypt from 2010 to 2012 (Table 1).
We sampled wild birds and asymptomatic and symptomatic poultry. The governorates sampled were selected
based on epidemiological and logistical considerations. To
increase the likelihood of detecting H5N1, we selected
governorates that had reported a human case of H5N1 in
the past year. Of these governorates, we sampled those in
which staff from the local branch of the General
Organization of Veterinary Services, the Egyptian agency
responsible for veterinary health, were available to
participate in the planning of field campaigns, liaise with
farmers, and take part in field work.
The number of villages sampled per governorate was
determined by the governorate’s land cover types, defined
as ecological regions with characteristic vegetation. We
sampled one village in every major land cover type, to test
whether land cover affected the prevalence of H5N1 in
wild birds. Within a land cover type, villages were selected
to be at least 30 km apart by road, but blindly with respect
to the presence of H5N1. For example, in Damietta
governorate, which has three land cover types – saltwater
wetlands bordering the Mediterranean, freshwater wetlands bordering Lake Manzala, and farmland in the Nile
Delta – we sampled Ezbet Sita village in a saltwater
wetland, Enania village in a freshwater wetland, and Sheta
village in farmland.
We sampled houses opportunistically inside each
village without inquiring in advance about whether the
household flock was infected with H5N1. We sampled all
individuals of all wild birds species mist netted at each
household. Due to logistic constraints, we could not
sample all individuals of all domestic species. Instead,
we sampled an average of 4 birds per household and an average of 20 households per village (range: 15–22
households). We administered one questionnaire per
household. Each village had up to 1000 backyard birds.
A previous study estimated the prevalence of H5N1 in
backyard poultry in Egypt at 10.5% (El-Zoghby et al., 2013).
Given our sample size and this prevalence, we could be 99%
confident of detecting H5N1 if present in backyard flocks
(Fosgate, 2009). We took samples (cloacal swabs, tracheal
swabs, and serum) from the asymptomatic and symptomatic poultry present in each house. Wild birds were
captured with mist nets (12 m 35 mm mesh) open from
5:30 to 9 a.m. Cloacal and tracheal swabs were placed in
viral transport media (VTM), stored in ice in the field, and
transported to the National Laboratory for Veterinary
Quality Control on Poultry Production where a cold chain
was maintained until testing. Serum samples were stored
in phosphate buffered saline, centrifuged, and cold chain
maintained following established protocols