Patterns of knowledge and condom use among population groups: results from the 2005 Ethiopian behavioral surveillance surveys on HIV

Background Behavioral surveys help interpret the magnitude of HIV/AIDS. We analyzed indicators of knowledge on HIV/AIDS and condom use among sub populations selected for behavioral surveillance in Ethiopia. Methods We used 2005 HIV/AIDS behavioral data from ten target groups. These were female sex workers, defense forces, police force, pastoralists, truck drivers, intercity bus drivers, road construction workers, teachers, factory workers and people in ANC catchment areas. Results Data from 14,524 individuals were analyzed. The majority were males (63.6%). Overall, knowledge of the three preventive methods, misconceptions and comprehensive knowledge was 57%, 75% and 18.5%, respectively. Female sex workers and the defense force showed some behavioral change in using a condom during the most recent sexual encounter and consistently used a condom with non-regular sexual partners and paying partners. Women, pastoralists and the illiterate were less likely to use condom. Conclusion Misconceptions about the transmission of HIV were high and comprehensive knowledge about HIV & AIDS was low, particularly among pastoralists. Consistent condom use and condom use during the last sexual encounter were high among both female sex workers and defense force employees, both with paying and non-regular sexual partners. This might be a positive sign, though a considerable proportion in each target group did not report using a condom during sex with non-regular partners.


Background
More than two decades have passed since the evidence of HIV in Ethiopia [1,2]. In spite of the indications that the HIV prevalence is leveling off, Ethiopia has remained one of the countries with a high number of people living with HIV/AIDS [3]. Because of the significant variation of prevalence estimates of HIV between the 2005 Ethiopian Demographic and Health Survey (1.4%) [4] and the 2005 ANC sentinel surveillance report (3.5%) [5], the Ministry of Health had to synthesize the findings in both and agree on a single point national estimate of the HIV adult prevalence which was 2.1% with 7.7%, and 0.9% urban and rural prevalence, respectively [6]. One of the major challenges in this regard was lack of serial quality data on high-risk and sexual behaviors as related to HIV/AIDS and sexually transmitted infections (STIs) that would contribute to the interpretation of the patterns of magnitude of HIV/AIDS [7].
Among the earliest groups that were identified and followed for HIV surveillance were female sex workers, long distance truck drivers, uniformed services and mobile groups [8][9][10]. In addition to the high risk-groups mentioned above, youth, teachers, road construction workers, the pastoralist community along transport corridors, people in Ante Natal Care (ANC) catchment areas, and other mobile populations such as intercity bus drivers were added to the list of potential risk groups for behavioral HIV surveillance. The selection of the target groups was based on emerging data and recommendations by the national technical working group for HIV surveillance.
Conventional household surveys usually fall short of penetrating high-risk groups which often shape the HIV epidemic. Behavioral Surveillance Surveys (BSS) employ sampling methods that better capture hard-to-reach populations through the use of standardized methodologies and tools that help comparison and monitoring of risk behaviors over time [11,12]. Ethiopia has already conducted two rounds of behavioral surveys, one in 2002 and another in 2005. Both rounds of surveys revealed a high level of awareness about HIV/AIDS [13,14]; however, the level of comprehensive knowledge was very low. For example, the level of comprehensive knowledge among long truck drivers and intercity bus drivers where the same methodology was used was lower in 2005 than in 2002 (42.8% vs. 29% and 31.1% vs. 23.2% respectively). Comprehensive knowledge was measured by six questions that indicate knowledge of the three preventive methods and absence of misconceptions. Out of the three common misconceptions one was universally recommended while two of them were specific to Ethiopia.
Changes in behaviors of people measured by behavioral surveys may indicate the effects of the several interventions aimed at reducing high-risk behaviors. One of the important assumptions in studying knowledge and misconceptions is people who are informed and equipped with necessary skills make better decisions in protecting themselves against HIV infection. Hence, along with other data sources, information garnered from behavioral surveillance surveys can illustrate trends in risk behaviors among various segments of the population and thereby better inform program interventions [12].
The objective of this analysis is to measure the level of essential knowledge on HIV/AIDS and condom utilization among several Ethiopian subpopulations selected for behavioral surveillance. By analyzing the pattern of behaviors and knowledge among these subgroups, practitioners and decision makers will be able to better calibrate interventions to reach these populations.

Methods
This study used behavioral data on HIV/AIDS/STIs conducted in 2005 in Ethiopia. The data was obtained from cross-sectional surveys on ten target populations for HIV/ AIDS/STIs behavioral surveillance surveys. BSS are repeated cross-sectional surveys that use probability sampling mehods on selected population groups that are hard-to-reach by conventional househod surveys. The BSS guideline provides a standardized methodology and generic questionnaires that are adapted to local situations [11]. For example, mobile populations (Long distance drivers) are sampled using time-location sampling through extensive mapping and listing subject locations, as well as peak and low congregation times of the target group. On the other hand, for sampling people in ANC catchement areas a houdehold survey was used. This was done after the ANC catchement areas were categorised as high, medium and low prevelence sites for HIV based on ANC surveillance reports.

Target populations and sites
Selection of target groups was based on the distribution of HIV in different subpopulations and the relevance of the groups in spreading and maintaining HIV/AIDS sexual networks based upon location, mobility or priorities of interventions. Table 1 shows the target populations, number of sites, size of the sample and sampling methods.

Sample sizes, sampling and survey procedures Sample size
In calculating the sample size for each target group assumptions and findings of the first round BSS were used. Sample size determination and sampling procedures followed guidelines for BSS. Design effect and power were considered in estimating the sample size for each corresponding target group.

Instruments and Data Collection procedures Instrument
The BSS uses standardized questionnaires that are pretested and adapted to the Ethiopian situation. Almost all the questions are the same for all targets except a few that are specific to each target group. All the questions were closed ended with options provided. The questionnaires were translated into major languages namely Amharic, Oromifa, and Afarigna and back translated to check for consistency with the English version. The questionnaires were field tested before use.
The data collectors had completed high school and most were from the Central Statistical Agency (CSA). In order to ensure confidence and convenience, the interviewers were sex matched with interviewees and were selected from different areas of the same region speaking the same language as the interviewees. This arrangement was preferred to avoid possibilities of acquaintance between the interviewer and interviewee.
Variables a) Explanatory variables: Age, sex, education and target group b) the outcome variables included: knowledge of prevention methods, misconception, comprehensive knowledge, consistent condom use with non regular sexual partner and paying sexual partner; condom use during last sexual encounter with non regular sexual partner and paying sexual partner.

Data Quality
The BSS used highly supervised data collection and data handling procedures that followed manuals of the CSA. One supervisor supervised a maximum of four to six data collectors and each questionnaire was checked for completeness and consistency on the spot. All data collected were handled and processed by CSA.

Data analysis
Data was analyzed using two statistical softwares: SPSS version 13 and Stata version 9. The analysis focused on assessing knowledge and condom use variables as related to the different target groups. We used bivarite analysis and then logistic regression controlling for age, sex and education levels. Adjusted Odds Ratios and 95% Confidence Intervals were calculated to indicate statistical associations at 5% significance level. A person was classified as knowledgeable on prevention if s/he mentioned all three methods.

Misconceptions to three
Knowledge on misconceptions was measured if a person rejected three common misconceptions. The first two of the misconceptions are specific to Ethiopia: 1. Can a person get the HIV virus from eating uncooked eggs from a chicken that swallowed a condom?
2. Can a person get the HIV virus from eating raw meat prepared by a person infected by HIV?
3. Do you think that a healthy-looking person can be infected with HIV, the virus that causes AIDS?
Comprehensive knowledge this index was built based on the answers to six questions i.e. three questions on knowledge of prevention and three questions on common misconceptions. A person was considered as having had comprehensive knowledge if s/ he knew all the three preventive methods and rejected the three misconceptions.

People in the ANC catchment area
Refers to the general population living in areas where ANC surveillance is taking place in the country. It included both rural and urban populations living in ANC surveillance sites.

Paying sexual partner
Having sex with someone in exchange for money.

Non-regular sexual partner
Non-commercial sexual partnership for unmarried or non-cohabitating partners.

Defense forces
Includes ground forces and air force

Uniformed services
Includes members of the police force and defense forces.

Long distance drivers
Includes long distance truck drivers and intercity bus drivers. Table 2 shows the characteristics of the study sample (N = 14,524). Of the total sample, 35.8% were selected from population in ANC catchment areas and 11.17% were road construction workers, 11.2% were pastoralists, 9.6% were female sex workers, 9% were police, and 7.4% were members of the defense force. The remaining groups accounted for less than 5% each. The majority of the people in the sample were males (63.6%). Some groups did not have females in the sample (Long distance drivers and to some extent the uniformed services and road construction workers).

Target groups and background variables
Knowledge of the three preventive methods was 57% of all target groups. Level of misconceptions was measured using two indicators i.e. rejecting three common misconceptions and rejecting six misconceptions. The proportion of people who had at least one misconception out of the three commonly existing misconceptions was very high (75%). Moreover, only one out of five people rejected all six misconceptions. Overall, the proportion of people with comprehensive knowledge was 18.5% (8.9% in females and 24% in males). Table 3 indicates distribution of knowledge indicators by sex and target groups. The outcomes included knowledge of prevention methods, level of misconceptions, and level of comprehensive knowledge.

Knowledge and misconceptions
Knowledge of the preventive methods ranged between 39.3% in pastoralists to 80.0% in the members of defense forces in males. Among females, pastoralists ranked the lowest for knowledge of preventive methods (18.5%) and female members of the police force ranked the highest (77.8%). Knowledge of preventive methods was consistently low for both females (18.5%) and males (39.3%) in pastoralist communities. Generally, 43.8% of females and 64.8% of males were knowledgeable about preventative methods.   (70%). In males, the highest proportion of consistent condom users was among intercity bus drivers (83.7%) and the lowest was among pastoralists (21.4%).
Condom use during last sexual encounter with non-regular sexual partners was 66% for females and 76.6% for males. Condom use at last sexual encounter was lowest among female pastoralists (none reported use but numbers were small) and the highest was amongst female sex workers (78.3%). In males, condom use at last sexual encounter was lowest in male pastoralists and highest among intercity bus drivers (89.5%).

Condom use with paid partner
Among males, consistent condom use was highest (90.4%) in members of the defense forces and lowest among factory workers (62.5%). In general, consistent condom use with paying sexual partners was high (93.0% for females and 86.2% for males). This was skewed towards females because of the female sex workers.
Condom use at last sex with a paying partner was 97.5% for females and 96% for males. Condom use during last sexual encounter was lowest for female pastoralists and factory workers (numbers were too small) while the highest was for female sex workers. Among males, condom use during last sexual encounter was highest among members of the defense force (98.3%) and lowest for factory workers (62.5%).     Table 6 shows condom use with non regular and paying sexual partners against the different target groups. In our analysis, age, sex and education were controlled while the people who live in ANC catchment areas was used as the referent group. .85) were more likely than people in ANC catchment areas to report condom use during last sexual encounter with paying partners. Female sex workers were five times more likely than the people in ANC catchment areas to report consistent condom use with paying sexual partners.

Discussion
This analysis provides a better understanding of the level of knowledge of HIV among important sub-populations in Ethiopia that are considered to be primary HIV intervention targets for the Ethiopian government.

Knowledge of preventive methods
Generally, knowledge of the preventive methods was fairly high; however, considerable knowledge gaps were observed among rural, females, people 35 years and older and the illiterate segments of the studied population. The Ethiopian DHS conducted in 2005 revealed that knowledge of preventive methods for limiting sex to one uninfected partner and use of condoms was 65.5% for urban women and 28.0% for rural women, and was 75.5% for urban men and 53.3% for rural men [4]. In Jimma town and its neighboring communities, nearly two-thirds of the respondents earned an adequate knowledge score [15]. Although the figures were not the same due to the difference in the nature of the studies, the pattern in this study was similar -women and groups with less access to information were less knowledgeable on ways to prevent HIV. Knowledge of preventive methods exceeded 66% among groups that are generally stigmatized for maintaining the HIV epidemic in Ethiopia such as long distance drivers, female sex workers, and uniformed services that expressed established sexual networks. Groups such as construction workers also demonstrated considerable knowledge which may be due to policy influences as a result of mainstreaming of HIV in sectors and labor laws that demand HIV intervention activities in all development sectors in Ethiopia [16,17].
Further, our analysis indicates that knowledge of preventive methods was significantly higher in groups that had better access to information and interventions; however it was lower among pastoralists who had limited access to modern media such as printed material, radio and television. This demonstrates the need to design alternative strategies to improve access to correct information to communities such as the pastoralists who are largely mobile. Programs should also implement tailored strategies to address the needs of the non-literate and women.
Our study revealed that knowledge of preventive methods was lower among people 35 years and older compared to people 15-24 years of age. From the different studies con-  ducted on adolescents and youth, knowledge was considered as adequate in terms of knowing the three programmatically important preventive methods [18][19][20].

Misconceptions
The magnitude of misconceptions was high in almost all groups but very high in groups that did not score well in levels of knowledge of preventive methods. Other authors  [18]. Other segments of the population such as people in the informal business sector in Addis Ababa have considerable misconceptions about the transmission of HIV [19]. In Northwest Ethiopia, it was reported that although knowledge on HIV was reasonably high, considerable misconceptions were noted among the study population [20]. To this end, it was also surprising to see high levels of misconceptions among female sex workers, as their knowledge of preventive methods was relatively high. It is important to note that female sex workers and long distance truck drivers were the first groups put under biological as well as behavioral surveillance in Ethiopia [21,22]. However, it is likely that the prevention programs focused on prevention methods and paid little attention to address misconceptions.
The proportion of female sex workers that had comprehensive knowledge was very low. This was attributed to the presence of widespread misconceptions about the transmission of HIV infection [13,14]. Nevertheless, the major difficulty in comparing knowledge across studies is the variability in definitions as well as the breadth of measurement regarding HIV knowledge. The scope of measurement extends from assessing simple awareness to the development of an index measuring the different domains of knowledge about HIV & AIDS. Behavioral surveillance surveys define comprehensive knowledge by combining knowledge of preventive methods with the ability to reject prevailing common misconceptions. In this regard, authors have also indicated the importance of addressing incorrect information in promoting condoms and building self-efficacy [23,24].

Condom use
Condom use in every penetrative sexual act has remained one of the major indicators of behavior change in HIV/ AIDS preventions. In this study condom use was measured using two indicators: 1) report of condom use during the last sexual encounter with non-regular sexual partners or paying sexual partners; and 2) pattern of condom use in any sexual relationship in the last 12 months (does not include condom use among married couples), which was described as consistent condom use.
The findings on female sex workers in terms of reporting condom use during last sexual encounter with a paying partner and consistent condom use with a non-regular sexual partner was consistently high. This might indicate changes in behavior among high-risk groups such as female sex workers. Consistent condom use and condom use during last sexual encounter with paying sexual partners did not show statistically significant difference in all the groups except female sex workers. This might mean that the pattern of condom use with high-risk partners such as female sex workers was similar across all study groups. In a randomized controlled trial in high-risk settings in Nicaragua, it was demonstrated that condom use was significantly higher in paid sex than non-commercial sex [25]. Similarly, a higher proportion of female sex workers reported consistent condom use with paying clients (77%) in Assossa, Ethiopia [26]. However, these studies are at least three to four years old compared to the findings of our study. On the other hand, other studies have shown that consistent condom use was low among brothel-based female sex workers in India and Ethiopia [27][28][29].
In our study condom use among long truck drivers was high compared to other reports. Among 343 long distance truck drivers in Bangladesh less than a third (31%) had ever used a condom which was very low [30]. Among female sex workers, the practice of condom use varied by type of partner with a tendency of low utilization within sexual relationships to long-term partners (as compared to casual partners). This was also reported by earlier studies [31]. Others have indicated that condom use with casual partners was determined by the interpersonal interaction between partners and less by attitudes and beliefs (unlike with established partners) [32]. The negotiations whether to request a partner to use or not to use condoms relies on several factors including the type of partner, access to condoms, a priori knowledge and risk perception. Long term sexual relationships tend to build norms and understanding among partners. In the behavioral surveillance surveys one of the major reasons for not using condoms was trusting a partner [13].

Conclusion
Knowledge of preventive methods was fairly high in almost all the target groups. However, there exists high level of misconception about HIV/AIDS among the target groups selected in this study. The misconceptions were greatest among groups with less access to media such as pastoralists which indicates a need to devise special strategies to address misconceptions among those without access to conventional forms of media.
The level of comprehensive knowledge was very low across all target groups but severe among pastoralists, women, people above the age of 35, as well as those who cannot read and who lives in rural areas. These deficiencies also highlight the limited efforts taking place to address in misconceptions and teach these sub-populations about prevention methods.
Consistent condom use and condom use during last sexual encounter was consistently high among female sex workers both with paying sexual partners as well as with non-regular sexual partners. Members of the defense force show a similar pattern indicating possible positive outcome of the HIV/AIDS interventions. Consistent condom use with paying partners was not statistically significant among the different target groups with the exception of female sex workers. This also might be a positive sign, though considerable proportions did not use condoms with non-regular partners. The HIV/AIDS prevention program must continue to encourage people to use condoms during sexual relations with different types of sexual partners.