RESUMO
BACKGROUND: Studies from low- and middle-income countries (LMIC) indicate that the use of audio computer-assisted self-interviewing (ACASI) is associated with more accurate reporting of sensitive behaviors (e.g. substance use and sexual risk behaviors) compared with interviewer-administered questionnaires. There is a lack of published information on the process of designing, developing, and implementing ACASI in LMIC. In this paper we describe our experience implementing an ACASI system for use with a population of orphans and vulnerable children in Zambia. METHODS: A questionnaire of mental health, substance use, and HIV risk behaviors was converted into an ACASI system, tested in pilot and validity studies, and implemented for use in a randomized controlled trial. Successes, barriers, and challenges associated with each stage in the development and implementation of ACASI are described. RESULTS: We were able to convert a lengthy and complex survey into an ACASI system that was feasible for use in Zambia. Lessons learned include the importance of: (1) piloting the written and electronic versions; (2) proper and extensive training for study assessors to use ACASI and for those doing voice recordings; and (3) attention to logistics such as appropriate space, internet, and power. CONCLUSIONS: We found that ACASI was feasible and acceptable in Zambia with proper planning, training, and supervision. Given mounting evidence indicating that ACASI provides more accurate self-report data and immediate data download compared with interview-administered measures, it may be an effective and economical alternative for behavioral health research studies in LMIC.