Statistical modelling of the relationship between intimate partner violence and HIV infection among women in Zimbabwe.
Zimbabwean women between the ages of 15-49 years are among the women most affected by HIV and Intimate Partner Violence in the world. The high rates of HIV infection among women have raised an alarm and stimulated research on the problem of violence against women. Intimate Partner Violence (IPV) is a well-known violation of human rights and is a problem in public health. It usually overlaps with the HIV/AIDS epidemic and has been reported to be a determinant of women's risk for HIV. The present study explored relevant statistical methods in modelling the relationship between Intimate Partner Violence (IPV) and HIV in Zimbabwe. The data used in the current research is from a Demographic and Health Survey (DHS) conducted in Zimbabwe for year 2005 - 06. The study aimed at analysing the relationship between IPV and HIV using the following explanatory variables: age; marital status; religion; education; wealth index; region; decision making; media exposure; STI; physical and sexual violence. Principal Component Analysis was used to create indices of IPV, media exposure and decision making among women in the age group 15 - 49. Survey Logistic Regression models accounting for multi-stage survey design was also used to adjust for socio-demographic and socio-economic factors. In order to explore the relationship between IPV and HIV prevalence among women, a generalised linear mixed model was adapted, controlling for socio-demographic variables and treating DHS survey clusters as random effects. Since IPV takes up more than two categories, Multinomial Logit Modelling was used to analyse the relationship of IPV with socio demographic and socio-economic variables. The results from the survey logistic regression modelling were as follows: unadjusted odds ratios (OR) for sexual or physical IPV ranged from 0:91 - 1:09 and 95% confidence intervals (CI) were (0:72; 1:14) for sexual and (0:92; 1:28) for physical violence. The adjusted odds ratios for sexual violence 0:82 [95%CI : 0:63; 1:06] and physical violence 1:12 [95%CI : 0:97; 1:36]. Both survey logistic regression models and generalised linear mixed models found no association between HIV and IPV among women in Zimbabwe. This study provides further evidence that IPV and HIV are not associated. In addition, the analysis revealed that the covariates which were associated with HIV and IPV were age, education, marital status, STI, religion and wealth index. As a result the study recommends that more research is required to find the situations or circumstances under which IPV is associated with HIV prevalence.