The importance of assessing self-reported HIV status in bio-behavioural surveys

Abstract In bio-behavioural surveys measuring prevalence of infection with human immunodeficiency virus (HIV), respondents should be asked the results of their last HIV test. However, many government authorities, nongovernmental organizations, researchers and other civil society stakeholders have stated that respondents involved in such surveys should not be asked to self-report their HIV status. The reasons offered for not asking respondents to report their status are that responses may be inaccurate and that asking about HIV status may violate the respondents’ human rights and exacerbate stigma and discrimination. Nevertheless, we contend that, in the antiretroviral therapy era, asking respondents in bio-behavioural surveys to self-report their HIV status is essential for measuring and improving access to – and coverage of – services for the care, treatment and prevention of HIV infection. It is also important for estimating the true size of the unmet needs in addressing the HIV epidemic and for interpreting the behaviours associated with the acquisition and transmission of HIV infection correctly. The data available indicate that most participants in health-related surveys are willing to respond to a question about HIV status – as one of possibly several sensitive questions about sexual and drug use behaviours. Ultimately, normalizing the self-reporting of HIV status could help the global community move from an era of so-called exceptionalism to one of destigmatization – and so improve the epidemic response worldwide.


Introduction
Community-based surveys such as population-based household and bio-behavioural surveillance surveys are primarily intended to assess the magnitude of epidemics of human immunodeficiency virus (HIV) infection and to evaluate national responses. These surveys are conducted among general populations and key populations at higher risk of HIV exposure and acquired immunodeficiency syndrome (AIDS) -e.g. men who have sex with men, people who inject drugs, and sex workers and their clients. After a decade of scaling-up antiretroviral therapy (ART) and growing evidence of ART's preventive effects on onward HIV transmission, 1 evaluation of national responses requires data on each point in the cascade of engagement in HIV-related care. 2 Data are also required to determine if countries have achieved -or are likely to achieve -the 90-90-90 targets of the Joint United Nations Programme of HIV/AIDS (UNAIDS). UNAIDS set a goal that, by 2020, 90% of people living with HIV will know their HIV status, 90% of people with diagnosed HIV infection will receive sustained ART and 90% of people receiving ART will have viral suppression. 3 To inform the cascade of engagement in care and monitor progress towards the 90-90-90 targets, it is essential to assess self-reported HIV status in bio-behavioural surveys. We need to know if respondents in such surveys know their serostatus and are enrolled and retained in HIV-related care and treatment. The failure to ask many survey participants their current HIV status -i.e. the results of their most recent HIV test -limits our ability to monitor the cascade of HIVrelated care. 4,5 Population-based household and bio-behavioural surveys 6 of key populations are important for monitoring HIV prevalence and risk behaviours. Data from such surveys can improve the design and evaluation of HIV intervention programmes. If survey participants are offered rapid tests for the detection or confirmation of HIV infection, they can be referred to existing HIV services for the confirmation of infection or directly to HIV prevention, care and treatment services. 7,8 The behavioural questionnaires found in most population-based and bio-behavioural surveys on HIV ask sensitive questions on both sexual practices -e.g. on the number of sexual partners and condom use -and behaviours that may be illegal in the study country -e.g. on drug use, selling sex and buying sex. Most such surveys also ask participants a question for UNAIDS Global AIDS Response Progress Reporting: "Have you had an HIV test in the past 12 months and, if so, did you receive the test result?" 4 However, many survey interviewers do not ask respondents to share the results of any HIV tests they have had and therefore cannot ask about respondents' engagement in care and treatment services. Surveys on HIV must be approved by national and often international human subjects' review boards and usually record no personal identifying data. To facilitate analysis of the determinants associated with prevalent HIV infections, any self-reported behavioural risk data and HIV test results that are collected are linked by non-identifying reproducible codes. For bio-behavioural surveys among key populations, community advisory groups may be engaged to assure respect for the community 6 However, most of these surveys are unable to identify any of the impor-Abstract In bio-behavioural surveys measuring prevalence of infection with human immunodeficiency virus (HIV), respondents should be asked the results of their last HIV test. However, many government authorities, nongovernmental organizations, researchers and other civil society stakeholders have stated that respondents involved in such surveys should not be asked to self-report their HIV status. The reasons offered for not asking respondents to report their status are that responses may be inaccurate and that asking about HIV status may violate the respondents' human rights and exacerbate stigma and discrimination. Nevertheless, we contend that, in the antiretroviral therapy era, asking respondents in bio-behavioural surveys to self-report their HIV status is essential for measuring and improving access to -and coverage of -services for the care, treatment and prevention of HIV infection. It is also important for estimating the true size of the unmet needs in addressing the HIV epidemic and for interpreting the behaviours associated with the acquisition and transmission of HIV infection correctly. The data available indicate that most participants in health-related surveys are willing to respond to a question about HIV status -as one of possibly several sensitive questions about sexual and drug use behaviours. Ultimately, normalizing the self-reporting of HIV status could help the global community move from an era of so-called exceptionalism to one of destigmatization -and so improve the epidemic response worldwide.

Asking HIV status in surveys
Lisa G Johnston et al.
tant correlates of having unrecognized HIV infection or to make distinctions between the behaviours of those who are aware of their HIV-positive serostatus and those who are unaware. Unless we can analyse the factors associated with undiagnosed infection and HIV-related care and treatment uptake, we will fail to understand the potential for the expansion of either the HIV epidemic among a given population or the coverage of related care and treatment services. Early treatment -which is only possible with early diagnosis -improves the health and long-term prognosis of people living with HIV and decreases the risk of onward sexual transmission of the virus. 2,[9][10][11][12] To determine the HIV serostatus of survey participants accurately and inform progress towards the target of 90% of people living with HIV knowing their status, surveys need to ask each respondent who has been tested for HIV about the result of their most recent HIV test. This information is important for planning targeted interventions -including prevention messaging and strategiesand in the exploration of respondents' experiences of HIV-related stigma and discrimination. Surveys that record self-reported HIV status should include community consultation processes to inform survey implementation and ensure respect of human rights, confidentiality and redress.
The 2012 Kenya AIDS Indicator Survey 13 is an example of a populationbased household survey in which participants were asked their HIV status -providing the opportunity to describe the community-based cascade of care among people living with HIV. This survey found that, of 363 respondents who claimed to be HIV-positive, 89.3% were enrolled in HIV care programmes, 84.5% of those eligible for treatment were receiving ART and 78.5% of those on ART who had not missed taking a pill were virally suppressed. 13 These results contributed to reprioritization in the HIV response, with greater focus placed on identifying HIV-infected individuals by expanding all testing modalities -particularly among men and children -with a direct linkage between the point of HIV diagnosis and HIV care services. 14 Despite precautions taken to protect respondents during population-based household and bio-behavioural surveys, many government authorities, nongov-ernmental organizations, researchers and other civil society stakeholders currently oppose asking respondents to self-report their HIV status. Guidelines for AIDS indicator and bio-behavioural survey questionnaires do not recommend questions on HIV status. 5,6,8 Resistance to the inclusion of data collection on self-reported HIV status has been based on numerous concerns, including that collection of such data could: (i) violate the respondents' human rights, by inadvertent release of data; (ii) exacerbate stigma and discrimination; and (iii) lead to inaccurate estimates of HIV prevalence. Below, we explore and respond to each of these concerns and present recommendations on how best to include self-reported HIV status in survey questionnaires.

Protecting human rights
The response to HIV in the 1980s and 1990s led to the advent of so-called AIDS exceptionalism -in part due to human rights abuses and discrimination against people living with HIV. 15 The initial responses to AIDS included potential and real violations of privacy -including discussions of compulsory quarantine for those at high risk of HIV infection and isolation for those already infected. While stigma and discrimination remain, the need for an exceptional response as it applies to surveys -i.e. the exclusion of questions on HIV status -has decreased because of the introduction of safeguards to respect individual human rights, the availability of ART and the ability of ART to reduce onward HIV transmission risks and improve health outcomes when initiated early after infection. 9,16-18 By asking respondents to state their HIV status -if known -it becomes possible to check if respondents living with HIV are accessing relevant care and treatment services. Individual autonomy can be maintained because respondents can refuse to answer any survey questions or stop the interview at any time, without repercussions. The study protocols and informed consent procedures and forms used in any survey need to be tailored to the HIV-related legal, policy and social context in which the survey is implemented and to the type of HIV testing and procedures for the return of test results and counselling used in the survey area.

Stigma and discrimination
Key populations and people living with HIV often face layered stigma, discrimination and -depending on their country of residence -the consequences of legislation outlawing behaviours that may increase their risk of HIV infection -e.g. drug use, sex work, sex with sex workers and male-male sex. 19 Those who oppose asking survey participants to reveal the result of their most recent HIV test are concerned that respondents will suffer additional stigma and discrimination if their self-reported HIV statuses are inadvertently disclosed. To help ensure the confidentiality of test results and minimize the respondent's risk of further stigma and discrimination, interviewers are often asked to preface any question about a respondent's HIV test with the words "I do not want to know the results of the test …" Any post-survey stigma and discrimination must be minimized through confidentiality procedures for the survey participants, ethical reviews and staff training on confidentiality. Clear penalties for the disclosure of information on HIV status must be set and should apply to all staff involved in the survey.
Internalized stigma is more difficult to measure and may worsen for survey respondents asked about their HIV status. Bio-behavioural surveys on HIV infection often include questions that most would consider to be sensitive -e.g. on sexual intercourse, condom use and illicit drug injection. Participants are often asked about being arrested, whether they have been forced to have sex and whether they have been discriminated against because of their behaviours. Interviewer training should address how to respond to sensitivities that can vary among individual respondents -including those who know they are HIV-infected.
A review of results from selected bio-behavioural surveys in which respondents -from different key populations in low-and middle-income settings -were asked to state their HIV status -if known -revealed that the proportion of respondents with an HIV testing history who refused to reveal their HIV status varied greatly (Table 1). In a survey of people who injected drugs in Nador, Morocco, none of 277 respondents refused to self-report their HIV status although 4% refused to answer questions on other sensitive topics, such as needle sharing. 22 In contrast, 15-16% Policy & practice Asking HIV status in surveys Lisa G Johnston et al.
of 6106 people who injected drugs in Ukrainian cities refused to share their HIV status with interviewers. 25 In a survey in Ghana, where male-male sex remains illegal, only 2% of 456 interviewed men who had sex with men refused to respond to a question about HIV status. 20 Data we collected on response rates (Table 1) indicate that, in general, survey respondents do not consider a question on HIV status any more sensitive than several other questions included in bio-behavioural studies on HIV.

Inaccuracy of self-reported status
Many who believe survey respondents should not be asked to reveal their HIV status also believe that the results of such questioning are likely to be inaccurate because of the effects of social desirability bias and post-test seroconversion. The accuracy of self-reports of the most recent HIV test results depends on numerous factors -e.g. to whom and in which context the information is being reported and the perceived benefit or harm of sharing the information. Inaccuracies may arise when respondents fail to understand the meaning of a positive or negative test result, 27 fail to remember the result or -perhaps because of low social desirability for frequent testing in a particular community 28 -have not been tested for years. Detailed analysis of data from household surveys in Malawi 29 and Uganda 30 indicated that, even after adjusting for expected seroconversions, about one-quarter to one-third of HIV-positive respondents intentionally misreported their HIV status as negative. Conversely, of 4125 participants who provided their serostatus in a 2011 Ukraine survey of people who inject drugs, 85 (2%) who claimed to be seropositive for HIV were found to be seronegative when tested using a rapid diagnostic assay during the survey. 25 In the 2007 Kenya AIDS Indicator Survey, 92.2% of the 181 respondents who reported that they were HIV-positive were found to be seropositive when tested in the survey and 93.8% of the 4886 respondents who reported that they were HIV negative were found to be seronegative when tested during the survey (Andrea A Kim, United States Centers for Disease Control and Prevention, unpublished data, 2015).
Knowledge of HIV serostatus is important for gauging the level of status awareness among those infected and uninfected with HIV and permits better assessment of the uptake of appropriate services based on HIV status. Although all so-called sensitive questions are subject to response biases, such biases can be minimized by training interviewers to build rapport, assess responses, probe for more valid answers and behave ethically. In some settings, the impact of such biases can be assessed by testing for antiretroviral drugs in the blood. In the 2012 Kenya AIDS Indicator Survey, testing for such drugs indicated that 10% more respondents were aware that they were HIV-infected than indicated by the self-reporting. 13 In household surveys in Malawi and Uganda, similar testing for antiretroviral drugs indicated similar levels of misreporting of serostatus. 29,30 Such evaluation of bias permits improvement of HIV prevalence estimates and suggests that eliciting honest responses to questions about HIV status requires more work to build trust between interviewees and interviewers.
Population-based household surveys increasingly rely on rapid HIV testing in the household -rather than central laboratory testing -for estimating national and subnational HIV prevalence. Accurate data on reasons for refusing rapid HIV testing -e.g. prior known HIV-positive status -  Have available information about -and referrals for -local care and treatment. Data management and confidentiality Do not collect any personal identification. Link all behavioural data, including self-reported HIV status, using codes. Once completed and reviewed, keep all documents, including those with self-reported HIV status, in a secure and locked location that is only accessible to designated personnel.

Surveys in small communities
When surveys take place in small communities, where survey staff might know participants, measures in addition to the ethical considerations mentioned above may be needed. For example, consider having interviewers and other staff recuse themselves if they know the participant. Where possible, bring interviewers and other staff members from other communities and/or use computer-assisted self-interviewing techniques.

Data use Estimates
Estimate the total size of the epidemic from the numbers of reported and unrecognized cases. Also estimate the percentage of HIV-infected participants who were unaware of their positivity at the time of testing in the survey, and the percentage of HIV-infected individuals who are not receiving care.

Identification
Identify population subgroups and risk behaviours associated with unrecognized or undiagnosed infection to improve the prioritization and targeting of care and treatment services. Comparison Compare self-reported HIV status from surveys with any additional and relevant data that are available -e.g. viral load, blood levels of antiretroviral medications and case surveillance data -to inform the interpretation of the true levels of HIV status awareness.
HIV: human immunodeficiency virus.
Policy & practice Asking HIV status in surveys Lisa G Johnston et al.
respondent's HIV status. Discrepancy between self-reported HIV status and the respondent's perceived status has been reported. 28 While asking for known HIV status remains crucial, the resultant data have to be compared with any additional relevant data so that discrepancies between actual, self-perceived and selfreported HIV status can be evaluated. 31 A deeper understanding of the context in which HIV status is correctly known and reported can be used to enhance survey and questionnaire design and to improve HIV testing and counselling programmes.

Need for self-reported HIV status
When institutional review boards approve surveys, they expect adherence to informed consent with appropriate protection of human participantsincluding protections of anonymity, confidentiality and awareness of the right to refuse to respond. 32 Surveys can include linkages, where needed, with psychosocial support counsellors. We assert that the benefits of asking respondents to self-report their HIV status outweigh the potential risks of asking this question.

Treatment and care coverage
Questions about knowledge of status provide an introduction to further questions about linkages to HIV care services and ART. In the era of ART, it is especially crucial to assess the reach and coverage of ART programmes and identify -and refer for care -any seropositive individuals not currently in an HIV care programme. Bio-behavioural surveys have been critical in measuring the coverage of HIV prevention programmes. The omission from such surveys of self-reported HIV status represents a missed opportunity to measure treatment coverage, particularly among key populations. Measuring HIV care cascades -in terms of knowing an individual's HIV status, assessing seropositive individuals for treatment eligibility, enrolment and retention of eligible individuals in pre-ART or ART-based care, and the achievement of viral suppression -is essential to the accurate monitoring of global efforts to slow the HIV epidemic and increase healthy outcomes for people living with HIV. [33][34][35]

Treatment needs and transmission
HIV case reporting captures the number of diagnosed people living with HIV in a population -when mortality and migration are known. However, to estimate the total number of people living with HIV -i.e. total treatment need -the number of undiagnosed cases of HIV infection is required. 36 Individuals who are unaware of their HIV infection are more likely to be associated with continued HIV transmission. [37][38][39][40][41] Comparison of the characteristics of people aware and unaware of their HIV status can lead to more efficient programme design and implementation. Status awareness is also a key parameter in the modelling of HIV transmission rates.
Unrecognized HIV infections are an important factor in onward transmission of the virus and the number of infected individuals who are unaware of their HIV status is a crucial indicator of potential HIV spread. 42,43 Infected individuals who believe themselves to be uninfected may use harm-reduction strategies -e.g. seroadaptive behaviours during unprotected sex -incorrectly. 44,45 A growing number of HIV interventions -e.g. early ART for serodiscordant couples and pre-exposure prophylaxis -require awareness of HIV status. 1,46 Measures of risk behaviours also increasingly require the context of serostatus -of both the individual of interest and their sex partner(s) -to interpret the potential for transmission. Knowledge of which population segments are most likely to be HIV seropositive -and unaware of it -is useful for prioritizing and tailoring intervention programmes for the people most in need 42,47 and for forecasting financial needs for the HIV response. Finally, self-reported HIV statuses can be used to monitor HIV testing uptake and efficiency. 33,38,48 Recommendations Table 2 summarizes recommendations for survey designers, data analysts, ethical review committees and civil society. For all surveys, it is essential to receive approval of institutional review boards and to pilot questions to ensure they are designed to respect participants while capturing needed information. Respondents who have never had an HIV test may be asked what they think their HIV status is. Although trained survey staff are usually equipped to ask sensitive questions, in some circumstances it might be better to have an HIV counsellor ask and record questions about HIV status. Surveys in which respondents are asked their HIV status should include strategies for the provision of information on HIV and the referral of seropositive respondents for HIV care and treatment. Follow-up questions and additional analyses are useful for verifying responses and optimizing the usefulness of findings for prioritizing and targeting care and treatment services.

La importancia de evaluar el VIH autodeclarado en encuestas bioconductuales
Durante la realización de encuestas bioconductuales que miden la prevalencia del contagio del virus de la inmunodeficiencia humana (VIH), los encuestados deberían entregar los resultados de la última prueba de VIH a la que se hayan sometido. No obstante, muchas autoridades gubernamentales, organizaciones no gubernamentales, investigadores y otras partes interesadas de la sociedad civil han declarado que las personas encuestadas implicadas en dichas encuestas no deberían tener que realizar una autodeclaración de su estado con relación al VIH. Las razones expuestas para no pedir a los encuestados que informen de su condición son que las respuestas pueden ser inexactas y que preguntarles por su estado en relación al VIH puede violar sus derechos humanos y exacerbar la estigmatización y la discriminación. Sin embargo, se sostiene que, en la era de terapias antirretrovirales, solicitar a los encuestados en encuestas bioconductuales que declaren su estado en relación al VIH es fundamental para medir y mejorar el acceso a, y la cobertura de, servicios para la atención, el tratamiento y la prevención del contagio del VIH. También es importante para estimar el verdadero alcance de las necesidades insatisfechas a la hora de abordar la epidemia de VIH, así como para interpretar las conductas asociadas a la adquisición y transmisión del virus de forma adecuada. Los datos disponibles indican que la mayor parte de los participantes en encuestas relacionadas con la salud están dispuestos a responder una pregunta sobre su estado en relación al VIH como una de las muchas posibles preguntas delicadas sobre comportamientos sexuales y de consumo de drogas. Por último, normalizar las declaraciones sobre el estado en relación al VIH podría ayudar a la comunidad mundial a pasar de una época caracterizada por el "excepcionalismo" a una caracterizada por la "desestigmatización", y, de este modo, mejorar la respuesta frente a las epidemias a nivel mundial.