**3.1.2 Sampling**

The sample size estimation for a prevalence study was calculated on the basis of previous prevalence estimation studies, where available (Folch et al., 2005). According to this calculation, a total of 2,800 persons (400 per city) were planned to be included in the data collection process.

Time-Location (or Time-Space) Sampling (TLS) was used to recruit representative samples of men visiting the gay scene in each city. This specific approach was used in previous studies (Williamson & Hart, 2007; Stueve et al., 2001; MacKellar et al., 2007; Gallagher et al., 2007; Muhib et al., 2001), and has proven to be a good and reliable method for gathering both behavioural and biological data in hidden or hard to reach populations. According to the TLS method, spaces (locations) are venues attended by the target population, in this case MSM, while times refer to specific days and time periods when the target population attends each identified space or setting. This sampling approach allows to construe a sample with known properties, enabling statistical inferences to be made to the larger population of venue visitors (ideally, all MSM attending the gay scene in each city).

In order to identify timing and settings, a specific formative research was conducted in each city. The first step in the formative research was the identification and selection of all gay venues in the given city. Bars, discos, saunas, cruising venues, sex-shops, sex-clubs were therefore identified. All venues were mapped and visited when information was not sufficient. This process allowed to identify the potential TLS units (the attendance time frame, opening days and hours) and to create a specific data collection calendar. The spaces and their associated days were divided into standardised time segments (four-hour periods). MSM were then enrolled over the entire TLS unit time period. During the data collection sessions, information regarding the number of refusals per unit was registered. In order to include all possible time and location units that did not occur frequently but that may attract members of the target population, also settings or special gay events were identified and mapped, creating a "special events" category. A list of TLS units was created according to this sampling approach, where the TLS units were the "primary sampling units" (PSU). In addition, in creating the final list of TLS units, safety and feasibility of data collection in different settings and at different times (especially in cruising venues) was taken into account. At this stage, the PSU were randomly selected from complete list of eligible TLS list in each participating city.
