Table 5. Sampling methods being used in each site.

Country—site	Sampling of localities	Sampling of households within localities
Benin	We will randomly select a total of 80 villages in proportion to their population size, from a total of 103 villages in Tchaourou, Parakou and Natitingou districts.	To include about 63 participants (5000/80), approximately 12 households should be included in each village, based on available demographic data. We will record the boundaries of each village using GPS. We will use satellite imagery and a random selection algorithm to select 12 dwellings in each village, plus six dwellings in reserve. The dwellings will be visited consecutively using a smartphone equipped with a GPS application until approximately 63 (+/- 3) participants are included. If this number is not reached after visiting the 12 dwellings (e.g. dwelling unoccupied or refusal of the household), the dwellings from the reserve list will be visited.
Guinea	Thirty-two villages in the region of Faranah will be selected purposively by a commission based on four criteria: i) incidence of fever from an unknown origin higher than 4% ii) between 1000 and 2000 habitants iii) not located on main roads and iv) less than 45 from Faranah city by car (for logistic reasons).	We will calculate the number of participants required in each village in proportion to the size of the village. We will obtain an enumeration of households in each selected village from the previous household census conducted for bed net distribution. Within each village, we will select households by simple random sampling until the total number of eligible household members reaches the target number.
Liberia	We will purposively select three documented Lassa fever hotspot communities in Bong County.	We will calculate the number of participants required in each community in proportion to the size of the community and convert this into the number of households using available demographic data (average household size estimated at five members). We will record the boundaries of each community using GPS. We will use satellite imagery and a random selection algorithm to select the desired number of dwellings in each village, plus 50% in reserve. We will visit these dwellings consecutively using a smartphone equipped with a GPS application until the target number of participants is included, using the dwellings from the reserve list if necessary.
Nigeria—Abakaliki	We will select ten documented Lassa fever hotspot communities.	We will calculate the number of participants required in each community in proportion to the size of the community and convert this into number of households using available demographic data (average household size estimated at five members). We will conduct systematic random sampling from the exhaustive enumeration of all households in the community obtained from a previous immunization programme. We will continue recruitment until the target number of participants is reached.
Nigeria–Irrua	We will purposively select seven documented Lassa fever hotspot communities.	We will calculate the number of participants required in each community in proportion to the size of the community and convert this into the number of households using available demographic data (average household size estimated at five members). We will conduct systematic random sampling from the exhaustive enumeration of all households in the community obtained from a mini census. We will continue recruitment until the target number of participants is reached.
Nigeria—Owo	We will purposively select ten documented Lassa fever hotspot communities.	We will calculate the number of participants required in each community in proportion to the size of the community and convert this into number of households using available demographic data (average household size estimated at five members). We will conduct systematic random sampling from geographical data, including locations (streets), as well as demographic information on those geographical locations (number of households). We will continue recruitment until the target number of participants is reached.
Sierra Leone	We will randomly select 18 villages from the 3782 villages in four districts surrounding Kenema General Hospital, from where the number of Lassa cases have been previously admitted to the Lassa fever ward. The 3782 villages are distributed over 385 sections. To maximise representativeness, we will use a two-stage cluster sampling method: the first stage will be to sample the sections with probability proportional to size and the second stage will be to select randomly one village in each section. Owing to the geography of the villages (too far apart to be adequately grouped in the same cluster), we will not include settlements with fewer than 20 dwellings.	Given the small size of the villages (on average about 500 inhabitants) and to facilitate acceptance of the study, we will include all households in a village. We will enrol villages until the target population is reached.