Changes in HIV testing in Brazil between 1998 and 2005

OBJECTIVE: To analize changes in HIV testing, reasons reported by those who were tested or not and received counseling. METHODS: Cross-sectional studies conducted in both men and women aged 16 to 65 years based on representative samples of urban Brazil in 1998 (n=3,600) and 2005 (n=5,040). Sociodemographic, sexual, reproductive characteristics, life experiences and health data were collected and analyzed. Potential differences in the distribution of variables was analyzed using Pearson’s chi-square and design-based F test (α<5%). RESULTS: In 1998 and 2005, 20.2% and 33.6% of interviewees had been tested, respectively. A total of 60% women aged 25–34 years were tested, but those who reported sexual initiation before the age of 16 and four or more sexual partners in the fi ve years prior to the interview were less tested. There was no signifi cant increase in testing among men, except among those aged 55– 65 years, per capita income between 1–3 and 5–10 monthly minimum wages, retired, historical Protestant and followers of African-Brazilian religions, living in the North/Northeast region and who reported homosexual/bisexual partners or no sexual relationship in the fi ve years prior to the interview. Testing rates did not increase in those who self-reported as high risk for HIV. Among women, prenatal testing rate increased while work-related testing decreased among men. In 2005, half of those who were tested did not receive any advice before or after testing. CONCLUSIONS: HIV testing scaling up was unequal and was mostly seen among women at childbearing age, adults and those better off. There seems to be an increase in testing rates in Brazil but without regard for people’s right to free choice and without offering more widely and better quality counseling. DESCRIPTORS: Acquired Immunodefi ciency Syndrome, diagnosis. HIV. Socioeconomic Factors. Health Knowledge, Attitudes, Practice. Health Inequalities. Cross-Sectional Studies. Population Studies in Public Health. Brazil. Cross-sectional studies.


INTRODUCTION
HIV testing was fi rst approved in 1985 for blood and blood product control.a,b In São Paulo, southeast Brazil, HIV testing became available in 1986 as a result of civil society's pushing the State House of Representatives for a Ministério da Saúde.Coordenação Nacional de DST/AIDS.Aconselhamento: um desafi o para prática integral em saúde -avaliação das ações.Brasília; 1999.Between 1987 and 1988, there were created Centers for Testing and Counseling (CTA), then called Centers for Serological Advice and Support (COAS) that mainly provided free, confi dential and anonymous testing to the so-called "risk groups," i.e., male homosexuals, sex workers and intravenous drug users.c The introduction of highly active antiretroviral therapy (HAART) in 1996, as a right in Brazil, made testing a diagnostic tool to detect those who needed and were eligible to receive HAART.As a preventive action, early diagnosis aims at providing comprehensive medical care to reduce disease burden and mortality and thus reduce vertical, sexual and parenteral transmission by decreasing viral load and HIV circulation in the blood stream. 8ven the notorious stigma and discrimination accompanying HIV/AIDS since the beginning of the epidemic, 22 scaling up HIV testing was based on the following regulating principles: counseling and information on HIV/AIDS before and after testing; testee's explicit voluntary and informed consent; and confi dentiality of testing results.d It is thus intended to integrate freedom (private autonomy) and equity (health and education) rights. 12e purpose of the present study was to analyze changes in HIV testing, the reasons reported by those who were tested or not tested and received counseling.

METHODS
The analyses refer to fi ndings of the survey "Comportamento Sexual e Percepções da População Brasileira sobre HIV/Aids" e (Sexual behavior and perceptions of the Brazilian population regarding HIV/AIDS), carried out in 2005, compared with a similar survey carried out in 1998.f Both surveys consisted of representative samples of Brazilian urban population based on the microareas defined by the Instituto Brasileiro de Geografia e Estatística (IBGE -Brazilian Institute of Geography and Statistics).Using a stratifi ed multi-stage sampling, census tracts, private households and individuals aged between 16 and 65 years were randomly and successively drawn in each microregion.
For the 2005 survey, the criterion for microregion selection was modifi ed to include more urban areas compared to the 1998 sample. 6e fi nal 1998 and 2005 samples consisted of 3,600 and 5,040 subjects, respectively, comprising both men and women aged between 16 and 65 years.Sociodemographic, sexual, reproductive characteristics, life and health experiences were the dependent variables; HIV testing, reasons for being tested and counseling were analyzed as outcomes.
For time analysis, double-entry expectancy tables were compiled by gender and main sociodemographic variables: age, skin color, full years of schooling, per capita family income, Brazilian macroregion, marital status, current occupation, current religion, sexual and reproductive practices, age at sexual initiation, condom use at fi rst sexual intercourse, type of sexual partner and number of sexual partners in the last fi ve years prior to the interview, prior sexually transmitted disease (STD), number of children, HIV-related life experiences and health, self-assessment of HIV risk, and belief about mandatory testing.Study variables were defi ned and categorized as to allow comparability between both 1998 and 2005 surveys.
In the 2005 survey, there were included questions on reasons for being tested or not related to their last testing and whether individual or group counseling was offered before and after testing.Assuming that most Brazilians are not familiar with the term counseling, the following question was asked: "Did you talk about it before your last testing?".Data were adjusted by weight, primary sampling unit, and strata for complex sample designs (Stata 8.0). 6Differences between 1998 and 2005 were analyzed using Pearson's chi-square and design-based F test at a 5% signifi cance level.
The project of the survey "Comportamento Sexual e Percepções da População Brasileira Sobre HIV/Aids" was approved by the Ethics Committee of Faculdade de Saúde Pública of Universidade de São Paulo.

RESULTS
In 1998 and 2005, 20.2% (95% CI: 16.2;24.3)and 33.6% (95% CI: 31.7;35.4) of all interviewees had been tested for HIV, respectively.However, these rates included blood donor testing.After excluding blood donors, 15.3% and 28.6% of all interviewed had been tested in 1998 and 2005, respectively.When in addition prenatal testing was excluded, access to HIV testing felt to 13.5% and 20.8%, respectively.Table 1 illustrates changes in HIV testing between 1998 and 2005 among men and women, according to sociodemographic variables.
Among women, both in 1998 and 2005, there were seen lower testing rates at younger and older ages (16-19 and 55-65 years), in the North/Northeast region, and in certain occupations (household maid, liberal professional, retired, student and homemaker).In 2005, signifi cant differences were seen with lower testing among Black women who were illiterate or had elementary schooling, income less than three monthly minimum wages (MMWs), living in the North/Northeast and Central-West/Southeast regions, single, retired, students and followers of Catholic and Protestant religions.
A comparison between 1998 and 2005 data show increased testing rates among women in almost all categories studied, reaching 60% in those aged 25-34.No signifi cant increase was seen in women aged 16-19 years, illiterate, per capita family income between 5-10 MMWs and certain occupations (private sector employees, unemployed and students).The number of female interviewees was small in some categories (liberal professionals, business owners, followers of African-Brazilian religions), which prevented further comparisons.
As for men, both surveys showed lower testing rates at younger and older ages (16-19 and 55-65 years), among those illiterate or who had elementary schooling, and income less than 3 MMWs.Signifi cant testing differences were seen in 2005 with lower rates among those living in the North/Northeast and Central-West/ Southeast regions, single, unemployed and students.However, a comparison of testing rates between 1998 and 2005 showed no increase except in those aged 55-65 years, per capita income between 1-3 and 5-10 MMWs, retired, followers of historical Protestantism and African-Brazilian religions, living in the North/ Northeast region, and those who reported homosexual practices or not having sexual intercourse in the last fi ve years prior to the interview.Among those more frequently tested, higher rates were found only among those followers of African-Brazilian religions and who reported homosexual practices.
Table 2 shows lifetime testing rates for men and women according to sexual and reproductive variables.
Differences between men and women were identifi ed in both surveys.In 1998, women who had one or no sexual partner sexual during the fi ve years prior to the interview and men who reported homosexual and bisexual partners were less frequently tested.
As for women, in 2005, signifi cant differences were seen in all variables studied.Lower testing rates were found among those who reported sexual initiation between 16 and 23 years of age, no condom use at fi rst intercourse, no sex in the last fi ve years prior to the interview, heterosexual partners, no STDs and no children.
As for men, in 2005, testing was signifi cantly lower among those heterosexual, with no past history of STDs and no children.
A signifi cant increase in testing was seen between 1998 and 2005 among women in almost all categories, except in those who had their sexual initiation before the age of 16 or four or more sexual partners in the last fi ve years prior to the interview.However, among men, a signifi cant increase in testing was seen among those who reported having, during the fi ve years prior to the interview, homosexual or bisexual partners, no sexual intercourse, two to three sexual partners and past history of STD.
When testing was analyzed by life and health experiences (Table 3), there were marked changes between 1998 and 2005 among women, except in those who were self-assessed as high risk for HIV infection.On the other hand, no signifi cant increase in testing rates was seen in any of the variables studied.Signifi cant differences persisted in 2005, which were identifi ed in 1998 as well, regarding higher testing rates among women and men who were close to an HIV-positive person.Among women, in 1998, those self-assessed as high risk were more frequently tested, while, in 2005, those self-assessed as low or intermediate risk were more often tested.
In 2005, testing was less frequent among women who believed that consent is necessarily required for HIV testing.Men who were self-assessed as low risk were more often tested than those who were self-assessed as intermediate risk or no risk at all.
Reasons for getting tested are displayed in Table 4.There was signifi cant increase of testing during prenatal care among women, and decrease in "work-related reasons," particularly among men.Signifi cant differences in reasons for testing persisted between men and women in 2005.In both 1998 and 2005, even after excluding those women who reported prenatal care testing, the most common reason was medical indication (44% and 35%, respectively).Among men, the most common reason was blood donation (39% and 36%, respectively).Among those who had never been tested, 72% reported that they were not likely to be exposed to HIV, 4.7% did not know where they could get tested, 2.5% did not want to think about HIV, 2% said they were afraid to fi nd out they were HIV-positive, and 1.7% claimed to be afraid of needles.Additionally, 30 interviewees (0.7%) said To be continued they believed the results would not be kept confi dential, 18 claimed to be afraid of losing their jobs, insurance, house, family and friends, and fi ve out of a total of 5,040 said they were afraid their names would be reported to the authorities in case of positive results.
In 2005, more than half of men and women did not get any advice before or after testing.Of those who did receive it, advice was given in an individual session.Between 1% and 2.5% did not know they were being tested.

DISCUSSION
The comparison between cross-sectional studies with similar methodological approaches allows to identifying changes in in-between years and to distinguish particular generational, social, regional, ethnic/racial and gender inequalities in HIV testing. 7e proportion of those ever tested signifi cantly increase between 1998 and 2005 (from 20% to 33.6%), meaning that almost 27 million, or after excluding blood donors, 22.7 million people (28.6%) 6  25% of non-virgin women (88%) reported ever being tested, projecting a coverage of at least 22% compared to 38.2% in 2005.There is thus evidence of increasing testing rates in Brazil.
Current testing rates are similar to those reported in the US (34%) 16 and Switzerland (40%) 25,33 in 1997-1998 and Canada (34.9%) in 1995-1996. 15However, these studies excluded blood donation testing and, after excluding blood donors, testing rates in Brazil in 2005 are signifi cantly lower.
In the US, the Centers for Disease Control and Prevention (CDC) have, since 1973, periodically collected data on health and, since 1995, on HIV testing through the National Survey of Family Growth.These surveys show increasing testing rates among American women, from 34.5% in 1995 to 54.9% in 2002 (excluding blood donations). 3In Italy, there was found a testing rate of 39.3% among people from four different provinces in 2002.Quota sampling (with 40% refusal rate) and different collection approaches do not allow to inferring an increase in testing rates in Italy between 1998 and 2002. 29sting rates in Brazil seems higher than in Greece (10.1%),Italy (15.5%), and Norway (17.4%) in 1997-1998. 25According to Jeannin et al 17   This selective scaling up is consistent with the historical medicalization of the female body while the male body has not systematically been an object of intervention in Brazilian health settings.The example of the Women's Comprehensive Health Care Program (PAISM) is illustrative.Established in 1980s, this program has privileged women as a sexual/reproductive being. 9In Brazil, HIV testing has been linked to between prenatal care with routine testing of all pregnant women with no consent required or adequate advice offered, leading to an increase in testing rates to 60% among women aged between 25 and 34 years.
In the US, compared to 1998, 16 women were more often tested than men in 2002. 2 In Italy, HIV testing is less common among women, while in Greece and Norway there are no gender-related differences. 26ere is evidence that a reduction of vertical transmission comes after a policy of screening and treatment of HIV-positive pregnant women. 5However, for achieving vertical transmission reduction ethical principles concerning pregnant women's absolute choice on testing must not be disregarded.
In the present study, 55% of interviewees did not get any counseling and 1.6% were unaware they were being tested.In other words, 26.7 out of 79.5 millions who were ever tested, more than 14.7 millions were tested unadvised and more than 420,000 were unaware they were being tested.
Goldani et al 13 (2003) claim that testing pregnant women has represented rather a mandatory than voluntary strategy.They studied 1,658 pregnant women of three public maternity hospitals in Porto Alegre (RS) and found 59.2% had not received any prior counseling, 18.1% were unaware they were being tested and 3.2% believed testing was mandatory.Morimura et al 19 (2006), while studying pregnant women in a school maternity hospital in Recife, also reported no pre-testing counseling (52% during prenatal care and 90% in rapid testing in the maternity hospital), and diffi culty to receive test results as well.Segurado et al 28 (2003) interviewed women living with HIV/AIDS in reference services in São Paulo.They found 42% received pre-testing and 62.5% post-testing counseling.These studies indicate that testing has been performed during pregnancy care and other care settings without women's consent, either they were HIV-positive or not, and without offering any counseling.An US study demonstrated that HIV testing was recommended based on the provider's perception of pregnant women's risk behaviors, suggesting judgmental decision making. 4sting rates seems to be increasing in Brazil without showing proper concern to people's right to autonomous decision and without offering wider and quality counseling.Being tested can be part of what Paiva et al 21 defi ned as "the right to prevention": promotion of access to prevention materials (condoms, syringes), information, education and quality counseling even when the interface with care is greater as having access to quality STD treatment, sexual and reproductive health care or prevention of transmission vertical.
As for HIV testing, not all health-related difference means inequality.Health inequality means unequal differences that "besides being systematic and signifi cant, they are preventable, unfair and unnecessary as well". 7 * Fifteen interviewees gave more than one reason besides their own initiative.
Lower testing rates can be associated to reduced likelihood of early identifying certain population groups with prevention and treatment needs.The present study documents that young men and women aged between 16 and 19 have been less frequently tested, consistently shown in 1998 and 2005, without any signifi cant changes in in-between years.These differences may be unequal by systematically affecting less economically favored social groups.
Similarly, testing rates were lower in those aged 16-19 years in the US (12.2% in 1998, 16 and 18.7% in 2002 3 ), Canada (18%, 1995-1996 15 ), Italy (4.9% of men; 6.3% of women 26 ) and in other European countries. 25e literature confi rms that young people may resist HIV testing if they fi nd health services unfriendly. 18,20frican studies have reported that services friendly to young people are those that assure confi dentiality, and provide well-trained and non-stigmatizing counselors, different strategies for the integration between family and social networks and young populations.a,b In the US, although young people credit being tested due to provider's recommendation, less than half of health providers do that, disregarding the American Academy of Pediatrics recommendations. 4Low testing rates among those aged 16 to 19 may suggest that prevention actions targeting this population are basically limited to information and male condom distribution at schools. 21other population segment that did not show any progress in testing was illiterate men and women.This fi nding corroborates other studies in Greece, Italy, Switzerland and Norway that showed lower testing rates among low schooling people, 25 as well US studies. 16Illiterate people's health status suffer from major disadvantages in Brazil. 1 Men and women with less than three per capita MMWs still had the lowest testing rates in 2005.Considering data on illiteracy and low income, it can be said that there persists socioeconomic inequalities in access to testing.Despite higher testing rates in the North/Northeast regions, there were still regional inequalities in 2005.Compared to White, Black women were less often tested for HIV, which corroborates the Ministry of Health's study fi ndings.a Other major results of the present study were testing stabilization among private sector employees, business owners, unemployed and students, as well among those who reported sexual initiation before the age of 15.The literature did not have any other studies on HIV testing in these subpopulations.
Women who reported sexual initiation before the age of 15 were more frequently tested.The literature has established an association between age at sexual initiation and HIV infection. 14,23In Brazil, D'Oliveira et al  described that one out of three women in the city of São Paulo, Southeast region, and Zona da Mata area in Pernambuco, Northeast region, reported forced sexual initiation before the age of 15.Forced early sexual initiation can increase the incidence of HIV infection and make more women seek testing.
Between 1998 and 2005, there was no increase in testing rates among women who reported four or more sexual partners in the fi ve years prior to interview or who were self-assessed as high risk for HIV infection.This trend requires attention since these groups are more likely to get HIV infection. 4 Cock et al 10 (2006) stated that for equitably scaling up HIV testing innovative approaches are required including offering new methods, such as saliva or fi ngertip testing, as well as actions for testing entire families at health facilities, at home or in community settings.However, these strategies can be effective only along with strategies for fi ghting stigma associated to HIV.
In fact, Brazilian population segments (illiterate and poor men) excluded from HIV testing scale-up are showing persistent increase in AIDS incidence since 1999. 11,27cioeconomic, generational, regional, ethnic/racial inequalities in testing are consistent with diffi culties faced by other countries such as Italy, Canada and the US. 15,16,25,26 the present study, non-voluntary (blood donation and work-related) testing rates decreased in 2005, especially among men, while prenatal care testing increased in settings suggesting non-voluntary testing without counseling.In 1998, 80% of urban Brazilian population believed HIV testing should be mandatory to everyone regardless of their HIV vulnerability.a There can be seen in this scenario the introduction of effective practices for mandatory testing without counseling among women during prenatal and delivery care for prevention of vertical transmission. 13,19,28This is a contradictory approach for scaling up coverage as it does not safeguard women's right to autonomy and body integrity.Increased coverage and mandatory testing are relevant issues that should be discussed along with the role of counseling.The CDC 4  In the 2005 survey, a higher number of individuals were sampled.However, some populations remained in small numbers and showed wide confi dence intervals (CIs) such as women and men in certain occupations (public sector employees, unemployed, students, liberal professionals and business owners); followers of spiritism and African-Brazilian religions; those having homosexual or bisexual partners or who were self-assessed as high risk for HIV infection.Widowed men and women with more than six sexual partners still showed wide CIs.Further survey studies should include larger samples, similar to international designs on sexuality and risk perception including samples of more than 10,000 interviewees.
In the analysis of testing during lifetime, it should be taken into consideration the likelihood of inconsistent answers due to respondents' inaccuracies or forgetfulness, even regarding testing in the year prior to the interview.In the US, an analysis of national surveys identifi ed different reporting by the same individual regarding being tested in a year time. 24However, this study did not manage to measure the direction of bias (under or overestimation) besides the fact that it was conducted between 1990 and 1992 before HAART introduction.
Population-based studies, on the other hand, may underestimate response rates of sensitive questions that may stir up stigma and discrimination.The Brazilian Institute of Public Opinion and Statistics (IBOPE) was responsible for data collection in the 2005 survey and sought to apply potentially mitigating procedures, e.g., interviewees were interviewed by same-sex interviewers.Also, to prevent underestimation, it was explicitly explained to interviewers and interviewees that test results would not be asked during data collection.
To strengthen testing as a preventive strategy, there is a need to increase offer along with counseling.This combination has the potential of breaking off the HIV transmission chain as it allows people to know their HIV status and ponder on infection risks and prevention through behavioral changes.a,b It is estimated that new infections could be reduced by 30% a year if all infected people would know their HIV status. 4 Campaigns such as "Fique Sabendo" c (Did you know?) are opportune but should focus on those segments that have been systematically excluded and be implemented to assure the rights of those being tested.Future awareness campaigns need to reinforce that HIV testing should always be voluntary, confi dential and offered along with high-quality counseling.Access to such testing is a citizen's right and a provider's duty in both public and private settings.

Table 5 .
Counseling characteristics of HIV testing reported by urban Brazilians men and women aged 16 to 65. Brazil, 2005 Referral and counter-referral mechanisms should also be established, notably where they are defi cient: blood banks, private laboratories, basic health units, among others.