A combination of data collection tools were used to gather the information in this study. Tools used for data collection from ARV users were exit interviews, semi-structured questionnaires, adherence tool and FGDs. These tools were originally developed at a multi-country workshop held in Bagamoyo in February 2005, and were then adapted after pilot testing. They are included in this report as Annexes. The following tools were used with other population groups: semi-structured interviews with staff, observation, pharmacy stock controls and key informant interviews with identified community members.

Instruments used to measure adherence were:

• Two-day recall, involving the use of 'sun and moon' charts to measure the consistency of the times respondents took their medicine over the previous two days (Annex 4).

• Visual analogue, in which ARV users were asked to pour beads from one glass into another to indicate the number of pills they would have taken over a one-month period. The remaining beads in the first glass were deemed to be the pills which the ARV user had forgotten to take during the course of the month. Adherence was recorded on the basis of the line marked from 0-10 on the first glass (Annex 4).

• Pill count method undertaken by the dispensing pharmacist, in which the numerator was the number of pills supposed to have been taken over a given period minus the number of pills missed and the denominator was the total number of pills supposed to have been taken (Annex 4).

• Health workers' assessment, comprising the estimate of the level of adherence to treatment by ART clients over a period of not less than three months (Annex 6).