Participant selection and sample size
The desired sample size was calculated at 800 for Livingstone and Chipata combined, plus 400 for Chirundu. This was to allow for comparison over time at Livingstone and Chipata combined. Chirundu would be compared combined to the other sites of the Corridor of Hope Regional Project. The sample size was calculated assuming that each site had a minimum of 400 sex workers. However, during the mapping it became clear that this was not the case and a "take all" approach was used. Since the survey period was relatively short, we can assume that few sex workers would have moved from site to site during this period and we have not sampled a significant number of sex workers twice. Only 624 participants were finally recruited, 481 in Livingstone and Chipata combined and 143 in Chirundu. This sample may be sufficient to compare changes for most behavioural indicators and STD prevalence rates over time at all sites combined, and for Livingstone and Chipata combined. It will not be sufficient to detect changes in Chirundu separately.

The 'take all' approach aimed to recruit as many sex workers as possible among the sex workers operating at the site at the time of the survey. It is difficult to estimate the success of this approach. The mapping assured that most of the places where sex workers are active were visited. However, some places were overlooked. At the Livingstone site, for example, interviewers were not able to visit the motel where a significant number of sex workers are based and where a large number of commercial sexual contacts take place, because of the motel owner's refusal to participate in the study.

In addition, a certain number of women who are not sex workers were mistakenly recruited. The incentives given, and particularly the opportunity offered for a free physical exam and treatment, may have encouraged women to pretend that they were sex workers. At the beginning of the study, some women were discovered pretending and measures were taken to prevent this. However, it is possible that some women, particularly in Chipata, were still erroneously recruited.

Refusal to participate is reported to have been extremely low (0.5%). It is possible that not all women who refused were recorded and that the real response rate was lower. Because of the reasons mentioned above, it is possible that the sex workers recruited do not represent completely all sex workers at the sites. However, we do believe that the sample is sufficiently valid.

Socio-demographic profile of the study population
As expected, the sex workers interviewed were generally young. There was a very large proportion of teenagers (37%). This proportion is much higher than what is generally reported in other sex worker surveys in Africa. The survey was not able to identify a clear cause for this high proportion, although some contributing factors can be suggested. (1) It is possible that, as the HIV epidemic advances, there is a shift to younger women because of increased mortality in the older women; (2) there is anecdotical evidence that some clients look for younger sex workers believing that the risk for HIV and other sexually transmitted infections is less; and (3) the HIV epidemic may have changed the economic needs, forcing teenage girls into sex work.

The educational, religious and ethnic profile of the study population is probably similar to that of the general Zambian female population of that age in those areas. Compared to the data of the 1998 Demographic and Health Survey (DHS), the study population has a similar proportion of women who completed primary school, but fewer who completed secondary school. This could be due to the high proportion of teenagers.

About half of the women reported having been married, but only 6.8% said that they were currently married. An extremely high level of separation after marriage probably explains this. Similar results were found in the survey among sex workers in Ndola in 1998, where 43% of the women reported to be divorced, separated or widowed. The high proportion (69%) of married women who reported that their husband has other wives is probably due to a misunderstanding of the question (having considered any other partners as wives) rather than to a high number of polygamous marriages.

As expected, this is a highly mobile population with about half of the women originating from another province, and a median period of residence of only six years. This is comparable to survey findings among sex workers along truck stops in Tanzania⁵.

Sexual behaviour and sex work characteristics
The women reported starting sex work at a very young age (median 17 years). The reported median age at first sexual intercourse was 15 years, which means that the women engage in sex work very soon after becoming sexually active. The median period between sexual debut and sex work activities was one year, and almost half of the women (46%) reported starting sex work in the same year that they became sexually active (results not shown).

The median period of sex work was three years (results not shown), which is similar to findings in Ndola⁶ and in other surveys. The median period of sex work at the current site was 1.7 years, which confirms their high mobility. Almost all women reported that they had done sex work elsewhere before. However, the option 'no sex work elsewhere' was not listed included on the questionnaire. Livingstone sex workers appeared to be more stable than those in Chipata and Chirundu. Sex workers came from as far away as the Copperbelt area.

The median number of clients per week (2.8) is similar to survey findings in Ndola (3 per week) and in some other African towns, but is much lower than what is found in sex worker populations in big African cities, such as Kinshasa, Nairobi and Abidjan. Only a small proportion of the women can be considered as 'hard core' sex workers with at least three clients on a working day. It is not surprising to see that one-third reported that they had an economic activity apart from sex work. Similar to what is found in other surveys in Africa, more than half of the women (58%) reported that they support not only themselves, but others as well. This is not unusual in Africa where every individual has a responsibility to take care of the other members of the family network.

Most women (71%) were living without a partner, but 50% had a sexual partner in the past week who was not required to pay. For most of these women, this was a partner they saw regularly (86% reporting sexual intercourse with this partner at least twice in the past month). It is disturbing that more than one-third (37%) of these women reported that they had been forced to have sexual intercourse during the past year. This figure is high in comparison with reports from other similar surveys⁷.

Condom use
Condom use with clients, as reported by the women, was markedly higher than condom use reported by the sex workers in Ndola. Fifty-four percent of respondents said they had used a condom during their last sexual contact, while in Ndola only 28% had used condoms. Almost 20% claimed that they always used a condom with clients. Condom use with clients could therefore be considered better than expected, particularly since specific interventions have not been undertaken yet with this population. Nevertheless, condom use is still far from satisfactory, with plenty of room left for improvement. As expected, condom use with non-paying partners was less frequent than with clients. However, the difference is less than what is generally found in other surveys, which is encouraging.

About one-third of the women who had not used condoms at the last sexual contact (with a client or non-paying partner) cited partner's objection as their reason for not using condoms. This confirms that interventions with the clients and partners are crucial to enhance condom use. However, a still greater proportion of women said that they themselves didn't like condoms, didn't think of it, or thought it was not necessary.

A lack of availability of condoms was only mentioned by a small proportion of women, and almost all women said they could obtain condoms, if needed, in less than one hour. The presence of the Society for Family Health's (SFH) social marketing programme at all three sites and the public health services' condom distribution efforts explain the easy access to condoms reported by the women. Easy access to condoms is one of the indicators defined by the WVI project, and it may be difficult to further improve this indicator.

Female condoms were reasonably well known and used by the studied population, particularly in Chirundu. More than half of the population had heard of them and 11% had used them, probably because female condoms are also socially marketed by SFH. They are not used widely enough to replace the male condom, however.

Sexually Transmitted Diseases
Almost all of the women had heard about STDs, but when asked to list symptoms, 15% could not mention any. Only 10% mentioned spontaneously all four major STD symptoms in women (genital discharge, genital ulcers, pelvic pain and inguinal swelling).

About one-quarter (24%) reported having a genital discharge in the past year. This is lower than expected and lower than findings from other surveys. One explanation could be that the translation in local language used for 'genital discharge' may have been interpreted as 'genital lesions' instead of 'genital discharge'.

Care-seeking behaviour for STD symptoms was similar to behaviour reported in other studies of STD care seeking in African women. Still one-quarter of the symptomatic women do not seek care at a health facility. Traditional healers remain important care providers for genital symptoms. The majority of women do not inform their partners or change their sexual behaviour when symptomatic.

Family planning
The women were asked about their family planning practices. As expected, family planning practices were poor. More than half of the women did not use any FP method, and of those who did, a large proportion used condoms as the sole method. The women were not asked if they ever had an unwanted pregnancy or induced abortion because this question was considered too sensitive. Instead interviewers asked the women if they ever lost a pregnancy. About one-third responded 'yes'. Because of the young age of the respondents and the poor family planning practices, it is possible that a proportion of these lost pregnancies were induced abortions. It is extremely important to include FP activities with STD/HIV interventions undertaken with this population.

Knowledge and attitudes related to HIV
The women's knowledge of modes of transmission and methods of prevention was generally good. At this stage, almost all women know about HIV/AIDS, know that healthy looking persons can be infected, and most know that infection can be prevented by abstinence, faithfulness or condom use. The percentages of women who knew these three ways to prevent HIV infection were much higher than the percentages found in the 1996 DHS, where only 28.6%, 48.7%, and 38.4% of women knew that they could avoid HIV by abstinence, faithfulness or condom use respectively. This could be explained by our study population being more at risk and therefore better informed, but also by the fact that the questions in our questionnaire were prompted, while the questions in the DHS survey require a spontaneous response. This could also explain why a large proportion reported that HIV could be transmitted by mosquito bites or sharing meals. The proportion of women that can cite two HIV prevention strategies measures the WVI project's indicator on HIV knowledge. This proportion is already high at baseline (85%) and significant improvement may be difficult to achieve. But, as the Ndola study shows, a high knowledge of HIV does not necessarily mean that the women consider themselves at risk.

While most of the women knew that an infected pregnant woman or a breastfeeding mother can infect her child, only a small proportion of the women knew that treating pregnant women could prevent these infections. This is probably because the therapy for preventing mother-to-child transmission is still not widely available in Zambia.

HIV Voluntary counselling and testing
HIV Voluntary Counselling and Testing (HIV VCT) facilities are still relatively new in Zambia and not yet widely available. Even so, two-thirds of the women said that they had access to confidential HIV testing. The response was the same for women from Chirundu, where there is no facility offering voluntary HIV testing. Probably the women were referring to possibilities for testing in Lusaka, or another big city. Twelve percent of the women said they had already been tested once, but almost half said it was not on a voluntary basis, and one-quarter never found out the test result. The questionnaire did not ask for the reasons for testing if not voluntary, or why the women never found out the result. More in-depth qualitative research will be needed to explore these issues further.

Prevalence of sexually transmitted infections
The prevalence of sexually transmitted infections is extremely high in this population. Almost one quarter of the women is infected with either gonorrhoea or genital chlamydia infection, almost one third with syphilis and almost half with trichomoniasis. Particularly the women surveyed in Livingstone are highly affected. Prevalence rates reached 23% for gonorrhoea and 36.5% for syphilis.

The levels of sexually transmitted infections found in our survey are similar to what is found in other sex worker populations in Africa. In comparison to the population surveyed in Ndola, our population has a similar level of trichomoniasis infection, a somewhat lower level of genital chlamydia and syphilis infection (the Ndola survey found prevalence rates of respectively 9% and 42%) and a higher level of gonorrhoea infection (15% in Ndola).

The high STI prevalence rates conflict somehow with the relatively low level of number of clients and high level of reported condom use. They indicate that these women remain at extremely high risk for STI and HIV and that appropriate prevention interventions and STD treatment services are urgently needed.

Generalizability to other sites
For logistic and budget issues, only three of the five project VWI sites were included in the sample, so we cannot make any conclusions about characteristics of sex workers at the other sites. In general, socio-demographic characteristics and characteristics of sex work were quite similar between the three sites and the characteristics measured in Ndola in 1998. Thus we believe that this sample is representative -- in terms of these characteristics for sex worker populations -- of places of similar size along major transport routes in Zambia.

Condom use and knowledge, attitudes and behaviour related to HIV, STD and FP were often significantly different between sites. These characteristics are more dependent on exposure to several types of interventions and, therefore, more site specific.

Comparison with results from truck driver survey
Simultaneously with the sex worker's survey, investigators interviewed truck drivers passing through at the same border post, using a similar questionnaire. The results of this survey are presented in a separate report (Round 1 Behavioral Surveillance Survey Zambia 2000: Long Distance Truck Drivers

Reports from sex workers and truck drivers were quite different regarding condom use in commercial sex acts. The truck drivers reported a 92% use at the last commercial sex contact, while sex workers only reported a 54% use. Ninety-six percent of truck drivers also claimed that they 'always used' or 'almost always used' condoms for commercial sex, while only 25% of sex workers reported this. Only 7% of the truck drivers reported that it was the sex worker who suggested condom use, while 62% of the sex workers said it was their suggestion. Comparison is risky because truck drivers are only a part of the sex worker's clientele (particularly in bigger towns such as Livingstone and Chipata), and truck drivers frequent sex workers at sites other than the site where they were interviewed. However, it is likely that at least part of the discrepancy is explained by reporting bias by one or both groups.

Both groups answered similarly on knowledge and availability of the male condom. Sex workers seemed to procure condoms more often at the market, while truck drivers got condoms from pharmacies and friends. A similar proportion of truck drivers and sex workers knew about STDs and, not surprisingly, sex workers knew more about the symptoms in women, while the truck drivers knew more about symptoms in men. The sex workers had a better STD care seeking profile than drivers did. They more often sought advice from a health facility, told their partner and more frequently changed their behaviour when experiencing symptoms.

Knowledge, opinions and attitudes related to HIV were similar among these two groups, although sex workers more often reported that breastfeeding can transmit HIV and that treating a pregnant woman can prevent infection. The truck drivers more frequently reported having access to confidential HIV testing than the sex workers. They also were voluntarily tested for HIV and found out their result more often than the women. This can be explained by the higher mobility of the truck drivers who have access to testing facilities in a wider area.

Conclusion

The sex worker population found at the three sites was young, with a high proportion of teenagers, mostly living alone and similar to the general population in terms of educational level, religion and ethnicity. Most of the women had only a few clients a week and were quite mobile. Further ethnographic work is warranted to understand the young age of entry into sex work and the reported low numbers of partners.

Knowledge and behaviour related to HIV, STDs and family planning often varied between sites. In general, knowledge related to HIV was good, and condoms were available. Reported condom use, both with clients and non-paying partners, was higher than what is often found in similar settings elsewhere in Africa, but is still far from sufficient. Knowledge and behaviour related to STD symptoms can be improved. Family planning practices were poor, and the availability and use of counselling and testing facilities needs to be explored further. The extremely high levels of sexually transmitted infections, similar to what is found in other sex worker populations in Africa, confirm the continued high vulnerability of this population and the need for strengthening interventions.

The WVI project should focus on enhancing condom use, both male and female, through peer and other education programs with the sex workers, their clients and their regular partners. Knowledge and care seeking for STD symptoms could be improved by regular visits to trained health care providers in selected health facilities where access barriers have been removed. These health facilities should also develop activities to enhance sex workers' use of family planning. The possibility of offering HIV VCT facilities should be explored further.