Study design and population
This study used cross-sectional design and conducted among mining workers in Sali traditional gold mining site southwest Ethiopia. It is estimated that 12,000 people living in the site. There is only one health post and ten private clinics serving the mining site population and no any HIV prevention program yet implemented in the site. Additionally, there is no communication, road facility and school in the site yet.
Participants and sampling
The source populations were all gold miners in the site and study populations were all sampled gold miners working in site and who were happen to be present during study period. Individuals who stayed in the site for more than three month were included. While who were unable to communicate due to severe illness or deafness excluded from the study.
The sample size was calculated assuming proportion of 50% for consistent condom use, confidence interval of 95% and 5% margin of error (d). The final sample size was determined to be 422 after adding 10% for possible non-response.The sampling frame was prepared with respective of their small site where they specifically work. Then using the simple random sampling technique the individual to be involved in the study was selected.
Operational definitions
HIV preventive behavior- if an individual is abstaining from sexual intercourse in the last one year until the time of study period, using condom consistently or have only one partner in the past one year, having only one sexual partner and tested for HIV before their first sexual relation in the last six month, tested for HIV infection in last three month of the study period and consistently use condom it was said to be in HIV preventive behavior.
Data collection procedure
The study was conducted from Jan 15, 2012 to Feb 15, 2012. The questionnaire was initially prepared in English and then translated into Amharic. The Amharic version was again retranslated back to English by other individual to check for consistency of meaning. The translated Amharic version questionnaire was pre-tested prior to the actual data collection on 5% of the sample size outside of the study area from similar population. The trained six data collectors who had completed 10 to 12 grades and two supervisors who were first degree holder health profession were recruited. Those randomly selected individual was found by arranging with site leaders to know their respective small site where they specifically work. Finally the data was collected at small site where they were working.
Instruments
The data were collected using interviewer administered structured questionnaire adapted from other related behavioural studies [10, 15, 16, 18]. It comprised of socio demography characteristics, HIV preventive behaviour (9 items), perceived susceptibility (6 items), perceived severity (6 items), perceived benefit (16 items), perceived barrier (16 items), self efficacy (16 items) and cue to preventive action (12 items). Every constructs of HBM items response were elicited using five-point likert scale: strongly disagree (1),” “disagree (2),” “neither agree nor disagree (3),” “agree (4),” “strongly agree (5).” Negatively worded items were reversely coded during analysis. Reliability of each items of HBM constructs were checked using cronbach’s alpha (α). After dropping one item, the alpha of perceived susceptibility was α = 0.73, perceived severity α = 0.71 having all items, perceived benefit α = 0.87 without dropping any item, perceived barrier α = 0.75 after dropping four items, self efficacy α = 0.84 after dropping one item and cue to preventive action α = 0.82 by including all items. Finally the sum of each items remained in each scale were used as a construct of HBM.
Statistical analysis
The data was entered using EPI data version 3.1 and analyzed using SPSS Statistics software version 17.0. Data cleaning and assumption checking were performed prior to proceeding to analysis. Descriptive statistical analysis was done, T-test and chi square was used as needed to check the association between HIV preventive behavior, constructs of HBM and modifying factors (socio demographic characteristics). Multiple logistic regression analysis was used for prediction of HIV preventive behavior in three different models. In the first regression model, the effect of the socio demographic characteristics was assessed. While in the second model the constructs of HBM was assessed, in the final model those independent variables which had statistical significant association with the outcome variable in the two model above were entered the final regression model. To claim statistically significant effect, crude and adjusted odds ratio with 95% confidence interval (CI) was employed.
Ethical consideration
The study protocol was approved by the Ethical Clearance Committee of Jimma University and written informed consent was sought from each respondent before the interview. The data obtained in due course were confidentially stored.