The proportions of employed men and women in our sample were lower than in the general population according to the 2001 census, so the findings will tend to under-represent the views of employed people, especially men. However, there were no differences between employed and non-employed people in their perceptions of their HIV risk, their views about RHT, or the proportion who had been tested for HIV in the last 12 months. So the lower proportion of employed people in our sample is unlikely to have introduced an important bias in the findings. Some 29% of the households approached could not be included in the survey because no one was present and in a further 9% no one over 18 years old was present to be interviewed. Although this means we were unable to include over a third of the initially approached households, we have no particular reason to believe these "absent" households would have responded differently to the survey questions; in most cases the houses were empty because their occupants were away at their "lands".
RHT is regarded as cost-effective in resource-rich settings, even when the HIV prevalence is relatively low [16, 17]. Revised guidelines from the Centers for Disease Control in the USA now recommend a routine offer of HIV testing in the majority of health care settings . Some authors have stressed the need to increase the rate of HIV testing in Africa as a means of dealing with the AIDS epidemic from a public health standpoint [19, 20]. Others have raised concerns that RHT might be coercive and lead to testing without consent , might lead to people avoiding using health care facilities because of fear of testing , and might lead to increased partner violence against women [21, 22]. Our study suggests that RHT has been largely successful in Botswana, achieving a high rate of HIV testing without alienating users of government health facilities, and advocating for ART.
Over one half (55%) the respondents in our survey reported having an HIV test in the last 12 months. And of these, half said they had the test under the scheme of routine testing when they visited a government health facility. This is a big increase since 2005, when only 15% of those tested for HIV had been tested under the routine testing scheme . HIV testing is indeed being offered to people visiting government health facilities: one half of those who attended in the last 24 months were offered testing on their last visit. The reported rate of being asked to have an HIV test was 60% among people aged 25–44 years. Perhaps of concern, the offer rate was only 45% for users aged 18–24 years, but we do not know how many of them had already been tested recently.
The RHT scheme seems to reach women more than men. Women used government health facilities more than men and, on top of this, female service users were more likely than male service users to be offered testing and to go on to be tested. This higher rate of testing under the routine testing scheme is one reason women were more likely than men to have been tested for HIV in the last 12 months. It is not clear why women visiting government clinics are more likely to be offered testing under the RHT scheme but their higher rate of acceptance of the offer is in line with previous studies about VCT. Women's take-up of VCT is generally higher than that of men and studies of different groups have reported higher HIV testing rates among women, in countries without RHT on offer [23–25]. One study in a township in Cape Town did not find a higher rate of HIV testing among women . In Botswana, routinely collected data show that more women come forward for VCT than men  and a study of students found females were more willing than males to be tested for HIV .
In our study, rural dwellers were both more likely to think themselves at risk of HIV and more likely to report having an HIV test in the last 12 months. This finding may be a consequence of the widespread availability of HIV testing in Botswana: in addition to the routine testing offered in government clinics, VCT is widely available in rural areas, often attached to the government clinic in rural communities. The relative availability of ART within reach of even rural communities is probably also a factor increasing the rate of HIV testing. A recent study from Tutume in Botswana reviewed records and reported a big increase in numbers coming forward for HIV testing once ART became available locally . We found an association between a positive view about ART and being tested for HIV. The association between being in favour of RHT and being tested for HIV is interesting; it could mean that people who favour RHT deliberately visit clinics so that they will be offered an HIV test.
Few people expressed concerns about the routine HIV testing policy introduced into government clinics. Indeed, nearly all respondents knew of the policy and were in favour of it. Those few people who needed an explanation of RHT were given a description of a "routine offer" approach. Some of the majority of respondents who already knew about RHT may even have believed that an "opt-out" approach was operating but nearly all of them nevertheless approved routine testing. Nearly all respondents (90%) were also comfortable about the confidentiality of the information about them in government facilities. For the 10% who were not comfortable that the information about them was kept private and confidential we have no evidence of an actual breach of confidentiality, or that their concerns were about HIV or AIDS information. Nevertheless any breach of confidentiality even in rare cases would be a serious matter given the continuing stigma around HIV and AIDS. Our findings do not suggest that people are avoiding using government facilities for fear they might be coerced into being tested for HIV: the use of government facilities is high (much higher than the use of private services) and shows no signs of reducing; and nearly all respondents said they would go to a government facility if they had an illness they thought could be due to AIDS, many specifically saying this was because they would be tested for HIV in the facility and could have access to ART. The association between visiting a government health facility in the last 12 months and being tested for HIV in this period could be partly because some people visited the government facility because they thought themselves at risk of HIV and wanted to be tested. The high level of reported satisfaction with the service and perception of being treated with respect in government health facilities contrast with our findings from household surveys in other countries in southern and east Africa asking similar questions [10, 30, 31].
There is a potential risk that instead of being a "routine offer" of testing, the routine testing becomes an opt-out process, whereby one is tested unless specifically requesting not to be so. This seems to be only rarely the case in Botswana. Very few people thought they were tested despite not being asked. Some people (20 of the 52 tested "without consent") apparently went to the facility specifically to request testing (essentially using the clinic for VCT) and reported that they were not asked for testing. Others were older individuals who thought they would have been tested anyway and reported being given a test result, when they simply had blood taken and were given the results of a different type of test. There could be potential for false reassurance in such cases, and it is important that people do not believe they have tested negative for HIV when they have not in fact been tested. Most (83.5%) of those who were offered HIV testing said they had the test; women in particular rarely refused the testing offer. This high rate of acceptance of the HIV testing offer could reflect a tendency to do what health care workers instruct, which has been raised as a concern about the routine testing system [5, 21]. It is also possible that some people who were asked about having an HIV test declined to have the test but were nevertheless tested against their will. However, we had no indication from any respondent that this happened and we believe it is unlikely.
Nearly everyone who reported being tested for HIV when they visited a government clinic also reported being given the result. The few who did not get a result could have been waiting to receive it after the blood was sent away for testing. The usual practice for HIV testing in government health facilities is to use a rapid testing method and give the result immediately. However, if the facility runs out of rapid testing kits, they send the blood to be tested elsewhere. At least one young woman who reported being tested but not getting the result admitted to the interviewer that she "ran away" because she found she could not face getting the result.
We found that both men and women were more likely to think themselves at risk of HIV if they had suffered violence from their partner in arguments in the last year. This is in line with studies that have reported a higher rate of HIV infection among women who have suffered gender-related violence . It has been reported that women known or suspected to be HIV positive can suffer violence and abuse as a result  and fear of violent reactions from partners can be a barrier to women being tested for HIV . We did not find any association between suffering partner violence and having had an HIV test in the last 12 months. In this cross-sectional study we would not have been able to say which came first, the testing or the violence. We cannot rule out that women who experienced violence are less likely to visit clinics (and be tested) but at the same time women who are tested are more likely to experience partner violence afterwards.
According to government figures, in February 2006 some 61,981 people in Botswana were receiving ART, with 51,203 of them receiving treatment through the public sector and 2,460 out-sourced from the public sector . According to a WHO report, by the end of 2005, some 85% of people in Botswana in need of treatment were receiving ART . By comparison, in South Africa in 2005, less than 20% of the almost one million people in need of ART were receiving it . Our study suggests that the Botswana public is now fully convinced of the value of ART. This seems to be in large part because the widespread treatment provision means many people have seen relatives or friends "come back from the dead" while taking ART. Word of mouth testimony spreads quickly in Botswana, where there are strong family and community ties within the small population.
Corruption in health services is a well-known problem in many developing countries. In this study we did not ask about unofficial payments to health workers as part of the individual questionnaires, but payments to health workers or health workers stealing the medicines were not raised at all as concerns when the focus groups discussed problems of access to ART. Only one youth focus group mentioned that some people sold their ARVs in order to get money for alcohol or drugs.
The ART programme in Botswana is not confined to urban or peri-urban areas. Our study found knowledge and approval of ART to be high even in remote communities where people have to travel long distances to get their ART. The main complaint about ART was the long travel some people had to make to get their treatment, and this is especially a problem for people in remote rural communities.
One concern about ART and a very high prevalence of HIV infection is that people taking ART remain sexually active and may continue to spread the virus , whereas people sick with AIDS may be too ill to be sexually active. There is a common belief, voiced in the focus groups in this study, that ART actually stimulates sexual drive. However, other authors have reported a reduction in risky sexual behaviour in a programme of ART combined with counselling in Uganda .
Despite these successes in encouraging HIV testing and advocating and implementing ART, the HIV epidemic in Botswana remains a crucial public health challenge: unless the HIV incidence falls significantly the number of people living with HIV will actually increase as ART prolongs the life of those already infected. While VCT and RHT are clearly important for tertiary prevention, the role of VCT for secondary prevention is less clear, since behaviour change among people coming forward for VCT is mostly limited to those who test positive [38, 39]. The government of Botswana now needs to build on its successful tertiary prevention programme, which has given it considerable credibility with the public, to face the challenges of secondary, and particularly primary, prevention.