Use of audio computer-assisted self-interviews to assess tuberculosis-related risk behaviors

The objective of this study was to compare self-reported tuberculosis and h uman immunodeficiency virus (HIV) risk factors obtained from computer-assis ted questionnaires and interviewer-assisted questionnaires among participan ts of a needle exchange program. Between June 1998 and May 1999, needle exc hange program participants requesting tuberculosis screening underwent inte rviews regarding demographics and risk factors for tuberculosis and HIV inf ection. The first 190 participants underwent traditional interviewer-assist ed questionnaires, whereas the remaining 92 underwent computer-assisted que stionnaires. Data were analyzed by interview technique using odds ratios (O R) and multiple logistic regression. Among 282 participants, demographic ch aracteristics, health status, HIV serostatus, visits to homeless shelters, alcohol intake, and cigarette smoking were all similar by interview techniq ue. However, respondents receiving computer-assisted questionnaires were mo re likely than those receiving interviewer-assisted questionnaires to repor t smoking marijuana (OR = 5.56), crack (OR = 1.88), and heroin (OR = 2.60); as well as sharing cocaine smoking equipment (OR = 4.49), sharing heroin s moking equipment (OR = 2.85), "shotgunning" (OR = 4.48), and visiting crack houses (OR = 4.39). In the final multivariate model, respondents receiving computer-assisted questionnaires were more likely to report "shotgunning" and visiting a crack house relative to respondents receiving interviewer-as sisted questionnaires. In conclusion, increased odds of high-risk behaviors for tuberculosis and HIV infection among computer-assisted questionnaire r espondents support the use of computer-assisted questionnaires to ascertain risk behavior data for both tuberculosis and HIV.