| Introduction. Injecting drug users (IDUs) drive the HIV epidemic in China. Female injecting drug users who are sex workers (IDU-FSWs) is a strategic "bridge population" for HIV transmission from the IDU to non-IDU populations. Background characteristics, health behavioral theories (e.g. the Theory of Planned Behavior, TPB), drug dependence, economic pressure, psychological problems, social support and gender power are potential predictors of condom use during commercial sex among IDU-FSWs. Most of these associations have not been investigated in China or elsewhere, and the TPB has not been applied to HIV-vulnerable populations in China. A knowledge gap exists.;Objectives. This study validated two instruments measuring severity of drug dependence. The prevalence of inconsistent condom use among IDU-FSWs and its associations with the aforementioned variables were investigated. The hypotheses that different blocks of variables would have independent effects on condom use during commercial sex, and the effects of TPB-related variables on condom use would be moderated by some external variables (e.g. severity of drug dependence) were tested.;Subjects and methods. Two cross-sectional studies were conducted. In Study I, 178 non-institutionalized drug users were interviewed in Dazhou, Sichuan. In Study II, 281 non-institutionalized IDU-FSWs were interviewed in Dazhou, Sichuan and Hengyang, Hunan, using snowballing method and face-to-face interviews. Statistical methods such as hierarchical and interaction modeling, stratification analysis, ROC method were used in this study.;Results. The Opiate Addiction Severity Inventory-Revised (OASI-R) was fully validated in the Study I. Around 6.8% of IDU-FSWs were HIV positive and respectively 48% and 64% of them practiced needle sharing and unprotected commercial sex (last six months). After adjusting for significant background variables, the five TPB-related variables (AOR=0.43 to 1.92, p<0.001), severity of drug dependence (AOR=1.05, p<0.01), economic pressure (AOR=1.07, p<0.05) and all studied psychosocial variables (e.g. depression, social support and gender power; AOR=0.70 to 1.67, p<0.05) were significantly associated with condom use during commercial sex.;The final hierarchical model predicting condom use during commercial sex included variables coming from four blocks of independent variables, with ROC area = 94% and sensitivity/specificity = 0.84/0.91. A "Wellbeing Status Index" moderated the associations between some of the TPB-related variables and condom use during commercial sex.;Conclusions. The "bridging effect" for HIV transmission is evident. Variables predicting condom use during commercial sex are multidimensional. Integrative programs are hence required. Methadone clinics may offer a platform for offering such services to IDU-FSWs. The performance of the TPB could be enhanced by addition of external variables; its applicability varies according to the wellbeing status of the IDU-FSWs. Future randomized control studies are warranted to design effective evidence-based programs targeting IDU-FSWs. |
