Introduction There has been widespread concern about the slow uptake of voluntary counseling and testing (VCT) in many parts of sub-Saharan Africa [ 1, 2]. VCT is a cornerstone of cost-effective HIV prevention and linkage to expanding HIV treatment in low-resource settings [ 3, 4]. Some of the most significant barriers to HIV testing identified in sub-Saharan Africa include lack of access to VCT and to high quality clinical services, lack of access to antiretroviral (ARV) therapy, and HIV-related stigma [ 1, 5, 6]. With a seroprevalence of 37% of adults ages 15–49 [ 7, 8], Botswana established universal access to antiretroviral treatment (ART) beginning in 2002 for all patients with CD4 counts less than 200 or with an AIDS-defining illness [ 9– 11]. By January 2004, however, only 17,500 patients were enrolled in the Botswana National Treatment Program out of an estimated 110,000 eligible individuals [ 9]. Slow enrollment in HIV treatment was thought to be due in part to underutilization of HIV testing [ 9, 11, 12]; by mid-2003, only 70,000 tests in total had been performed in Botswana out of a population of 1.7 million [ 13]. HIV stigma was identified by government and press sources as one possible impediment to HIV testing and hence to the success of the new ART program, in that individuals may avoid testing and treatment facilities to avoid potential stigma and discrimination [ 8, 11, 13]. We previously reported that social stigma and fear of positive test results significantly delayed testing among a group of patients treated in the private sector in 2000 [ 14]. In an attempt to increase the uptake of HIV testing and ART, the Botswana government introduced the policy of routine HIV testing in early 2004, whereby nearly all patients would be tested as a routine part of medical visits unless they explicitly refused [ 13, 15]. While this approach to testing is provider-initiated, all patients should receive essential information about HIV testing and be informed of their right to refuse. In addition, there is typically greater emphasis on post-test compared with pre-test counseling [ 16]. Studies in resource-rich settings have shown that routine HIV testing can be cost-effective and life-saving, both by increasing the life expectancy of individuals with HIV and by reducing the annual HIV transmission rate [ 17– 21]. In June 2004, as part of a change in testing policy recommendations, UNAIDS and the World Health Organization recommended the routine offer of HIV testing by healthcare providers in a wide range of clinical encounters based in part on the Botswana experience [ 22, 23]. The goal of routine testing is to increase the proportion of individuals aware of their status, and thereby reduce “HIV exceptionalism,” lessen HIV-related stigma, and provide more people access to life-saving therapy [ 16, 24]. While provider-initiated approaches to testing are gaining popularity, there have been concerns that routine testing policies are potentially coercive, that counseling will no longer be practiced, that people may be dissuaded from visiting their doctors for fear of being tested, and that this policy may increase testing-related partner violence [ 15, 25– 27]. As routine testing is increasingly being recommended as an option in other countries [ 17, 18, 28– 30], it is important to improve our understanding of the consequences and specific human rights concerns associated with implementation of this policy in Botswana. We therefore assessed: 1) knowledge of and attitudes toward routine testing in Botswana with a focus on human rights concerns associated with this policy; 2) factors associated with whether respondents had heard of routine testing, and with positive attitudes toward this policy; and 3) the prevalence and correlates of HIV testing, barriers and facilitators to testing, and reported experiences with testing 11 months after the introduction of routine testing in Botswana. Methods In November and December 2004, we conducted a cross-sectional study using structured survey instruments among a probability sample of 1,268 adults selected from the five districts of Botswana with the highest number of HIV-infected individuals. These districts included Gaborone, Kweneng East, Francistown, Serowe/Palapye, and Tutume, and cover a population of 725,000 out of a total population of 1.7 million individuals in Botswana. We used a stratified two-stage probability sample design for the selection of the population-based sample with the assistance of the Central Statistics Office at the Ministry of Finance and Development Planning in Botswana. In the first stage of sampling, 89 enumeration areas were selected with probability proportional to measures of size, where measures of size are the number of households in the enumeration area as defined by the 2001 Population and Housing Census. At the second stage of sampling, households were systematically selected in each enumeration area by trained field researchers under the guidance of field supervisors. With a target sample of 1,200 households, and 15% over-sampling for an anticipated 85% response rate, 1,433 households were selected. Within each household, random number tables were used to select one adult member who met our inclusion criteria, and up to two repeat visits were made. Participants were excluded if they were older than 49 or younger than 18 years old, if they had cognitive disabilities, or if there was inadequate privacy to conduct the interviews. The 45- to 60-minute survey was conducted in either English or Setswana in a private setting, and written consent was obtained from all study participants. Our structured survey instrument and consent forms were pilot-tested among 20 individuals from Gaborone, and then translated into Setswana and back-translated into English. All study procedures were approved by the Human Subjects Committee at the University of California San Francisco (San Francisco, California, United States), as well as by the Botswana Ministry of Health Research and Development Committee. Measures Domains of inquiry for our 234-item survey ( Protocol S1) included demographics, HIV knowledge, experiences with HIV testing, barriers and facilitators to HIV testing, attitudes toward routine testing, HIV risk behaviors, HIV-related stigma, depression, beliefs about gender roles and gender discrimination, and measures of healthcare access and utilization. Based on an extensive literature review [ 2, 6, 31– 37] and discussions with key informants, we developed a conceptual model that guided the selection of variables for our multivariate model for correlates of testing, as shown in Figure 1. Relevant variables are explained below. Knowledge of and attitudes toward routine testing. Participants were asked whether they had heard of routine testing and were given a detailed explanation of this policy (see Table 1). Participants then indicated the extent to which they are in favor of routine testing and whether they think this policy affects HIV-related stigma, barriers to testing, violence against women related to testing, and uptake of ARVs. From questions assessing attitudes toward routine testing ( Table 1), we constructed an ordinal outcome of positive attitudes toward this policy. Participants were categorized as having zero to one, two, three, or four positive views toward routine testing. (See Tables 1 and 2 for specific items.) HIV testing. Participants were asked whether they had ever been tested for HIV (by either VCT or routine testing). If so, they were asked detailed questions about their experiences with pre-test and post-test counseling, confidentiality, facilitators to testing, and personal repercussions of testing. If not, they were asked a series of questions related to barriers to testing adapted from the CDC HIV Testing Instrument, version 9.00, and about their intention to be tested within the next six months. HIV status was not asked in order to maximize response rate and hence the generalizability of the population-based sample. HIV-related stigma. Respondents were asked seven questions representing potential stigmatizing attitudes adapted from the UNAIDS general population survey and the Department of Health Services AIDS module, which have been used successfully in previous studies in Botswana [ 38]. Anyone who reported a discriminatory attitude on any of four principal questions was registered as having stigmatizing attitudes per the UNAIDS scoring system. Since participants may not always openly endorse stigmatizing views toward people living with HIV and AIDS (PLWA) due to social desirability bias, as an additional measure of stigma, we also asked individuals to project the type of responses they would anticipate from others if they were to test positive for HIV and divulge their status to others. We converted this information to a nine-item index on “projected HIV stigma” with higher scores associated with a greater number of reported adverse social consequences associated with testing positive. This index had high internal reliability with a Cronbach alpha of 0.77. HIV knowledge. Participants were asked 15 questions about their knowledge of HIV transmission and prevention, based on questions modified from the UNAIDS General Population Survey and the Department of Health Services AIDS module. Using the UNAIDS knowledge indicator scoring system, individuals were scored as having HIV knowledge if they correctly identified the two most common modes of HIV prevention in Botswana. Depression. As depression is known to impede access to care and to worsen HIV outcomes in Western settings, we included depression in our analysis [ 39– 41]. Symptoms of depression were measured using the 15-item Hopkins Symptom Checklist for Depression [ 42] which has been validated previously in locations in Africa and elsewhere [ 43]. Analysis We used standard procedures for data entry and quality control. All data were analyzed with S TATA statistical software. Outcomes of interest included: a) having heard of routine testing; b) number of positive attitudes toward routine testing (categorized as an ordinal variable consisting of the following categories: zero to one, two, three, and four positive statements about routine testing); c) self-reported HIV testing (by either VCT or routine testing); d) having been tested under routine testing; and e) planning to test within the next six months (among people who had not tested). The following covariates were included in our analyses: 1) age (continuous); 2) sex; 3) income (≥population mean,
