Innovative Strategies for Remotely Sampling Hard-to-Reach Populations: Assessing Phone Versus Internet Respondent-Driven Sampling Approaches Among Venezuelan Refugees and Migrants in Colombia.

It is challenging to quantitatively measure the health vulnerability and risk factors of refugees and migrants residing outside of formal settlement settings. For hard-to-reach populations without available sampling frames, researchers have increasingly turned to novel sampling and statistical methods, like respondent-driven sampling (RDS). "Standard" RDS is typically conducted face-to-face at fixed sites. However, during the coronavirus disease 2019 (COVID-19) pandemic, face-to-face survey methods and recruitment approaches posed high potential risk of virus transmission and infection, making remote RDS approaches optimal. In this paper, we explore the feasibility of implementing telephone and Internet RDS strategies to assess challenges faced by Venezuelan refugees and migrants in the city of Bogotá, Colombia's capital, and the department of Norte de Santander, the main Venezuelan-Colombian border crossing site. We describe RDS assumptions, survey design, formative research, and the implementation of both strategies and present diagnostics for determining whether assumptions are met. Phone-based recruitment strategies in both locations and the Internet strategy in Bogotá achieved their calculated sample size; however, the Internet strategy in Norte de Santander did not. Most RDS assumptions were sufficiently met at sites where sample sizes were reached. These surveys provide valuable lessons for implementing innovative remote strategies with which to study hard-to-reach populations such as refugees and migrants.

The association between HIV self-test awareness and recent HIV testing uptake in the male population in Gambia: data analysis from 2019-2020 demographic and health survey.

BACKGROUND: The Gambian Ministry of Health is supportive of HIV self-testing (HIVST) and HIVST initiatives are being piloted as an additional strategy to increase HIV testing for individuals not currently reached by existing services, particularly men. This study aimed to determine awareness of HIVST among Gambian men, and whether prior awareness of HIVST is associated with recent HIV testing uptake. METHODS: We used men's cross-sectional data from the 2019-2020 Gambian Demographic and Health Survey. We employed design-adjusted multivariable logistic regression to examine the association between HIVST awareness and recent HIV testing. Propensity-score weighting was conducted as sensitivity analyses. RESULTS: Of 3,308 Gambian men included in the study, 11% (372) were aware of HIVST and 16% (450) received HIV testing in the last 12 months. In the design-adjusted multivariable analysis, men who were aware of HIVST had 1.76 times (95% confidence interval: 1.26-2.45) the odds of having an HIV test in the last 12 months, compared to those who were not aware of HIVST. Sensitivity analyses revealed similar findings. CONCLUSION: Awareness of HIVST may help increase the uptake of HIV testing among men in Gambia. This finding highlights HIVST awareness-raising activities to be an important intervention for nationwide HIVST program planning and implementation in Gambia.

Alarmingly high prevalence of high-risk drug use among palestinian males: a cross-sectional study.

BACKGROUND: The unique socioeconomic context in Palestine, characterized by political and economic tensions, creates conditions that facilitate the spread of illicit drug use among Palestinians. This paper presents findings from a 2017 survey of high-risk drug use (HRDU) among males in four regions in Palestine: the West Bank (north, middle, and south) and the Gaza Strip. These findings are essential for developing effective policies to respond to the increasing use of drugs among Palestinians. METHODS: Eligible participants were males aged 15 years and above who used at least one drug other than non-synthetic hashish or marijuana during the previous week. Participants underwent a face-to-face interview and had their drug use verified by urinalysis. Data were collected using respondent-driven sampling and data were analyzed using the successive sampling estimator. Multivariate regression analysis was conducted to examine factors associated with ever seeking rehabilitation services for illicit drug use in the West Bank and the Gaza Strip. RESULTS: A total of 400 males who use drugs were sampled in Gaza, plus 299 in the south, 300 in the north, and 299 in the middle region of the West Bank. It is estimated that there are 26,500 male HRDUs in Palestine comprising 1.8% of the male population aged 15 and above. Findings indicate that polydrug use is a serious issue in Palestine, especially in the West Bank, and that synthetic marijuana is prevalent among teenagers and young adults. CONCLUSIONS: Palestine must strengthen its national efforts to scale up harm reduction and treatment and care options for people suffering from drug use disorders, especially those involved in polydrug use. Additional measures are needed to prevent substance use among children and youth, support the families of people who use drugs, and ensure the continuity of HRDU services during emergencies.

Trends and Factors Affecting Knowledge of and Stigma and Violence Towards Female Sex Workers in Morocco.

BACKGROUND: We investigate factors affecting HIV knowledge, stigma, and violence among female sex workers (FSW) in Agadir, Rabat, Fes, and Tangier, Morocco, over three rounds of HIV surveillance surveys (2012, 2016, and 2019) conducted using respondent-driven sampling (RDS). Multivariable analyses from RDS studies examining the relationships between variables are under-utilized, particularly analyses that combine multiple locations and years of data together in a principled manner. METHOD: We fit three weighted logistic regression models for HIV transmission knowledge, having been denied service (experienced stigma), and having been hit (experienced violence) in the last 12 months, and perform model selection using 41 possible explanatory variables. RESULTS: Variables significantly associated with higher risk included reasons for sex work, how FSW solicit clients, if female family members are also involved in sex work, ever being forced to have sex, and ever being arrested or jailed. There were also significant differences between cities and in trends over time, with HIV transmission knowledge increasing and having been denied health services and having been hit decreasing. CONCLUSION: We found associations indicating that some particularly vulnerable FSW may be subject to multiple intersecting risks related to HIV knowledge, stigma, and violence which may decrease their agency to receive proper HIV testing, care, and treatment. Although Morocco has made excellent progress in providing HIV services to vulnerable populations, this additional analysis will be useful as Morocco makes programmatic decisions related to ending HIV by 2030.

Hidden population size estimation and diagnostics using two respondent-driven samples with applications in Armenia.

Estimating the size of hidden populations is essential to understand the magnitude of social and healthcare needs, risk behaviors, and disease burden. However, due to the hidden nature of these populations, they are difficult to survey, and there are no gold standard size estimation methods. Many different methods and variations exist, and diagnostic tools are needed to help researchers assess method-specific assumptions as well as compare between methods. Further, because many necessary mathematical assumptions are unrealistic for real survey implementation, assessment of how robust methods are to deviations from the stated assumptions is essential. We describe diagnostics and assess the performance of a new population size estimation method, capture-recapture with successive sampling population size estimation (CR-SS-PSE), which we apply to data from 3 years of studies from three cities and three hidden populations in Armenia. CR-SS-PSE relies on data from two sequential respondent-driven sampling surveys and extends the successive sampling population size estimation (SS-PSE) framework by using the number of individuals in the overlap between the two surveys and a model for the successive sampling process to estimate population size. We demonstrate that CR-SS-PSE is more robust to violations of successive sampling assumptions than SS-PSE. Further, we compare the CR-SS-PSE estimates to population size estimations using other common methods, including unique object and service multipliers, wisdom of the crowd, and two-source capture-recapture to illustrate volatility across estimation methods.

Alarmingly High HIV Prevalence Among Adolescent and Young Men Who have Sex with Men (MSM) in Urban Indonesia.

Indonesia's HIV epidemic is concentrated among key populations. While prevalence among men who have sex with men (MSM) is high, transmission among young MSM (15-24-years-old) remains poorly understood. We conducted a respondent driven sampling survey of 211 young MSM in urban Bandung, Indonesia in 2018-2019 to estimate HIV prevalence and associated risk factors. Thirty percent of young MSM were HIV antibody positive. This is nearly 100-fold greater than Indonesia's population prevalence and sevenfold higher than average estimates for young MSM across Asia and the Pacific Region. Individual risk factors associated with HIV infection were being 20-24 years old, having a steady partner and preferring the receptive position during sex. Issues of stigma, discrimination and social exclusion were common. Few young MSM who were open with friends and family members about their sexual identity. Among those that were, close to half reported experiencing feelings of aversion from these groups. Wider structural factors that reduce social tolerance, restrict the rights of young MSM and compel concealment of sexual identity are likely to fuel high-risk behaviors and limit access to essential testing care and support services including pre-exposure prophylaxis which is not yet widely available. Urgent health, social, legal and political actions are required to respond to these factors and reduce the disproportionate contribution of young MSM to Indonesia's HIV epidemic.

Estimating the Population Size of Males Who Inject Drugs in Myanmar: Methods for Obtaining Township and National Estimates.

Estimating the sizes of key populations at risk for HIV is crucial for HIV prevention and treatment. We provide findings of population size estimates (PSE) of males who inject drugs (MWID) in Myanmar, provide an intuitive method for countries to extrapolate subnational estimates into national estimates and provide guidance on how to maximize the utility of current PSE techniques. We used unique object and service multipliers, and successive sampling PSE in conjunction with a respondent driven sampling survey of MWID in ten Myanmar townships in 2014. Township estimates were assessed at a stakeholder meeting for biases and coded into ranges of high, medium and low MWID prevalence areas. Using the sampled townships as benchmarks for a range of MWID proportion estimates, national level MWID size estimates were derived by multiplying the adult male population for all townships with their corresponding proportion estimates. Final PSE ranged from high (4.12%), medium (1.02%) and low (0.11%), with the final agreed national point estimate of 83,000 MWID. Using estimates from survey data, this can translate into actual numbers of MWID living with HIV and practicing risky injecting and sexual behaviors. Although PSE are vital for monitoring HIV epidemics, no guidance exists for interpreting results of different PSE techniques or for extrapolating these results into national estimates. Assessing bias and gaining consensus on township level estimates and deriving ranges of MWID PSE throughout the country using stakeholder input is intuitive and accessible to countries.

Prevalence of HIV and other infections and injection behaviours among people who inject drugs in Addis Ababa, Ethiopia.

BACKGROUND: Ethiopia is one of the sub-Saharan African countries most affected by HIV/AIDS. However, the country lacks data describing the extent of the epidemic among people who inject drugs (PWID). Thus, a bio-behavioural study was conducted in 2015 to generate strategic information on the magnitude of HIV, hepatitis B virus (HBV), hepatitis C virus (HCV), syphilis and related risk behaviours among PWID in Addis Ababa. METHODS: A cross-sectional study using respondent-driven sampling was conducted among people reported to have injected illicit drugs within 6 months before the study. Males and females aged 15 years or above and who were resident in Addis Ababa were included in the study between 26 March and 22 May 2015. Data was analysed using respondent-driven (RDS) Analyst software. RESULTS: A total of 237 participants, including 6 seeds, enrolled in the study; most of the PWID were males (96%) with a mean age of 26 years. Most (79%) of the PWID reported injecting heroin but also reported using non-injecting drugs, including marijuana or ganja (47%) and/or khat (31%). Forty per cent of PWID reported ever sharing needles and 56% reported sharing other injecting equipment. However, only 14% reported injecting daily, and 49% reported injecting only 1 to 3 times a month. HIV prevalence was 6%, HBV was 5.1%, HCV was 2.9% and syphilis 5.1% among PWID. Among HIV-positive PWID, 60% reported sharing a needle the last time they injected. CONCLUSION: Even though the prevalence of HIV among drug users is not much higher than in the general population in Addis Ababa, the needle sharing prevalence was high. Thus, this baseline study shows the need to establish harm reduction programmes and prevention strategies for the PWID in Addis Ababa.

Associations of HIV Testing, Sexual Risk and Access to Prevention Among Female Sex Workers in the Dominican Republic.

The Caribbean region has one of the highest proportions of HIV in the general female population attributable to sex work. In 2008 (n = 1256) and 2012 (n = 1525) in the Dominican Republic, HIV biological and behavioral surveys were conducted among female sex workers (FSW) in four provinces using respondent driven sampling. Participants were ≥15 years who engaged in intercourse in exchange for money in the past 6 months and living/working in the study province. There were no statistically significant changes in HIV and other infections prevalence from 2008 to 2012, despite ongoing risky sexual practices. HIV testing and receiving results was low in all provinces. FSW in 2012 were more likely to receive HIV testing and results if they participated in HIV related information and education and had regular checkups at health centers. Further investigation is needed to understand barriers to HIV testing and access to prevention services.

Correlates of Forced Sex Among Young Men Who Have Sex With Men in Yangon and Monywa, Myanmar.

Forced sex at an early age is associated with a variety of negative factors including increased illness, high-risk sexual and substance-use behaviors, and mental and psychological stress. These sequelae may be compounded for men who have sex with men (MSM), especially young MSM and those with feminine gender identity and expression. This survey examined the prevalence and associations of forced sex among young MSM in two cities in Myanmar. In 2013-2014, surveys using respondent-driven sampling collected data on 200 young MSM in Yangon and 200 in Monywa. One quarter of young MSM in Yangon and 21 % in Monywa reported ever experiencing forced sex. In a multivariable model, having problems with family members and having any MSM friends with many partners had higher odds of experiencing forced sex. Having maternal acceptance of same-sex attraction (compared to acceptance by both parents) and becoming aware of their same-sex attraction at or above the age of 16 had lower odds of experiencing forced sex. Focused research is needed to understand the family and other social dynamics affecting vulnerability to forced sex, as well as specific sexual risks associated with forced sex among young MSM, including HIV acquisition and transmission risks.

Respondent-driven sampling as a recruitment method for men who have sex with men in southern sub-Saharan Africa: a cross-sectional analysis by wave.

OBJECTIVES: Respondent-driven sampling (RDS) is a popular method for recruiting men who have sex with men (MSM). Our objective is to describe the ability of RDS to reach MSM for HIV testing in three southern African nations. METHODS: Data collected via RDS among MSM in Lesotho (N=318), Swaziland (N=310) and Malawi (N=334) were analysed by wave in order to characterise differences in sample characteristics. Seeds were recruited from MSM-affiliated community-based organisations. Men were interviewed during a single study visit and tested for HIV. χ(2) tests for trend were used to examine differences in the proportions across wave category. RESULTS: A maximum of 13-19 recruitment waves were achieved in each study site. The percentage of those who identified as gay/homosexual decreased as waves increased in Lesotho (49% to 27%, p<0.01). In Swaziland and Lesotho, knowledge that anal sex was the riskiest type of sex for HIV transmission decreased across waves (39% to 23%, p<0.05, and 37% to 19%, p<0.05). The percentage of participants who had ever received more than one HIV test decreased across waves in Malawi (31% to 12%, p<0.01). In Lesotho and Malawi, the prevalence of testing positive for HIV decreased across waves (48% to 15%, p<0.01 and 23% to 11%, p<0.05). Among those living with HIV, the proportion of those unaware of their status increased across waves in all study sites although this finding was not statistically significant. CONCLUSIONS: RDS that extends deeper into recruitment waves may be a promising method of reaching MSM with varying levels of HIV prevention needs.

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

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.

Dans les enquêtes bio-comportementales évaluant la prévalence de l'infection par le virus de l'immunodéficience humaine (VIH), il conviendrait de demander aux répondants le résultat de leur dernier test de dépistage du VIH. Mais nombre d'autorités gouvernementales, d'organisations non gouvernementales, de chercheurs et d'acteurs de la société civile sont défavorables à ce que l'on demande aux répondants de déclarer leur statut VIH dans ce type d'enquêtes. Parmi les raisons invoquées pour ne pas demander le statut VIH figure le fait que les réponses peuvent être inexactes et qu'une telle demande pourrait transgresser les droits humains des répondants et accentuer les problèmes de stigmatisation et de discrimination. Néanmoins, nous soutenons qu'aujourd'hui, à l'ère de la thérapie antirétrovirale, le fait de demander leur statut VIH aux répondants des enquêtes bio-comportementales est essentiel à des fins d'évaluation des services de soins, traitement et prévention de l'infection à VIH et pour en optimiser l'accès et la couverture. Cela est également fondamental pour évaluer la véritable ampleur des besoins non satisfaits dans la gestion de l'épidémie de VIH et pour correctement interpréter les comportements associés à l'acquisition et à la transmission du VIH. Les données disponibles indiquent que la plupart des participants aux enquêtes liées à la santé acceptent de répondre à une question sur leur statut VIH, parmi d'autres questions sensibles concernant les comportements sexuels et l'usage de drogues. Enfin, une normalisation de l'auto-déclaration du statut VIH pourrait aider la communauté mondiale à dépasser ce que l'on appelle « l'exceptionnalisme ¼ pour ouvrir une ère de déstigmatisation, ce qui permettrait d'améliorer la réponse face à l'épidémie dans le monde entier.

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.

A Systematic Review of Published Respondent-Driven Sampling Surveys Collecting Behavioral and Biologic Data.

Reporting key details of respondent-driven sampling (RDS) survey implementation and analysis is essential for assessing the quality of RDS surveys. RDS is both a recruitment and analytic method and, as such, it is important to adequately describe both aspects in publications. We extracted data from peer-reviewed literature published through September, 2013 that reported collected biological specimens using RDS. We identified 151 eligible peer-reviewed articles describing 222 surveys conducted in seven regions throughout the world. Most published surveys reported basic implementation information such as survey city, country, year, population sampled, interview method, and final sample size. However, many surveys did not report essential methodological and analytical information for assessing RDS survey quality, including number of recruitment sites, seeds at start and end, maximum number of waves, and whether data were adjusted for network size. Understanding the quality of data collection and analysis in RDS is useful for effectively planning public health service delivery and funding priorities.

Recent HIV Testing Among Young Men Who Have Sex with Men in Bangkok and Chiang Mai: HIV Testing and Prevention Strategies Must Be Enhanced in Thailand.

HIV infection among men who have sex with men, particularly in Thai urban settings and among younger cohorts, is escalating. HIV testing and counseling (HTC) are important for prevention and obtaining treatment and care. We examine data from a 2013 survey of males, 15-24 years, reporting past-year sex with a male and living in Bangkok or Chiang Mai. Almost three quarters of young MSM (YMSM) in Bangkok and only 27 % in Chiang Mai had an HIV test in the previous year. Associations for HIV testing varied between cities, although having employment increased the odds of HIV testing for both cities. In Bangkok, family knowledge of same sex attraction and talking to parents/guardians about HIV/AIDS had higher odds of HIV testing. Expanded HTC coverage is needed for YMSM in Chiang Mai. All health centers providing HTC, including those targeting MSM, need to address the specific needs of younger cohorts.

Estimating the Size of Hidden Populations Using Respondent-driven Sampling Data: Case Examples from Morocco.

BACKGROUND: Respondent-driven sampling is used worldwide to estimate the population prevalence of characteristics, such as HIV/AIDS and associated risk factors in hard-to-reach populations. Estimating the total size of these populations is of great interest to national and international organizations; however, reliable measures of population size often do not exist. METHODS: Successive sampling-population size estimation (SS-PSE) along with network size imputation allows population size estimates to be made without relying on separate studies or additional data (as in network scale-up, multiplier, and capture-recapture methods), which may be biased. RESULTS: Ten population size estimates were calculated for people who inject drugs, female sex workers, men who have sex with other men, and migrants from sub-Saharan Africa in six different cities in Morocco. SS-PSE estimates fell within or very close to the likely values provided by experts and the estimates from previous studies using other methods. CONCLUSIONS: SS-PSE is an effective method for estimating the size of hard-to-reach populations that leverages important information within respondent-driven sampling studies. The addition of a network size imputation method helps to smooth network sizes allowing for more accurate results. However, caution should be used particularly when there is reason to believe that clustered subgroups may exist within the population of interest or when the sample size is small in relation to the population.

Strategic information is everyone's business: perspectives from an international stakeholder meeting to enhance strategic information data along the HIV Cascade for people who inject drugs.

People who inject drugs (PWID) are at increased HIV transmission risk because of unsafe injecting practices and a host of other individual, network, and structural factors. Thus, PWID have a great need for services within the Cascade of HIV prevention, diagnosis, care, and treatment (HIV Cascade). Yet the systems that monitor their progress through the Cascade are often lacking. Subsequently, fewer reliable data are available to guide programs targeting this key population (KP). Programmatic data, which are helpful in tracking PWID through the Cascade, also are limited because not all countries have harm reduction programming from which to estimate Cascade indicators. Also, due to stigma and the illegal nature of drug use, PWID may not disclose their drug use behavior or HIV status when accessing services. Consequently, PWID appear to have low HIV testing rates and, for those living with HIV, lower access to health services and lower viral suppression rates than do other KP groups. This commentary, based on outcomes from an international stakeholder meeting, identifies data gaps and proposes solutions to strengthen strategic information (SI), the systematic collection, analysis, and dissemination of information, to optimize HIV prevention, care, and treatment programming for PWID.

Incorporating the service multiplier method in respondent-driven sampling surveys to estimate the size of hidden and hard-to-reach populations: case studies from around the world.

BACKGROUND: Estimating the sizes of populations at highest risk for HIV is essential for developing and monitoring effective HIV prevention and treatment programs. We provide several country examples of how service multiplier methods have been used in respondent-driven sampling surveys and provide guidance on how to maximize this method's use. METHODS: Population size estimates were conducted in 4 countries (Mauritius- intravenous drug users [IDU] and female sex workers [FSW]; Papua New Guinea-FSW and men who have sex with men [MSM]; Thailand-IDU; United States-IDU) using adjusted proportions of population members reporting attending a service, project or study listed in a respondent-driven sampling survey, and the estimated total number of population members who visited one of the listed services, projects, or studies collected from the providers. RESULTS: The median population size estimates were 8866 for IDU and 667 for FSW in Mauritius. Median point estimates for FSW were 4190 in Port Moresby and 8712 in Goroka, Papua New Guinea, and 2,126 for MSM in Port Moresby and 4200 for IDU in Bangkok, Thailand. Median estimates for IDU were 1050 in Chiang Mai, Thailand, and 15,789 in 2005 and 15,554 in 2009 in San Francisco. CONCLUSION: Our estimates for almost all groups in each country fall within the range of other regional and national estimates, indicating that the service multiplier method, assuming all assumptions are met, can produce informative estimates. We suggest using multiple multipliers whenever possible, garnering program data from the widest possible range of services, projects, and studies. A median of several estimates is likely more robust to potential biases than a single estimate.

An empirical examination of respondent driven sampling design effects among HIV risk groups from studies conducted around the world.

For studies using respondent driven sampling (RDS), the current practice of collecting a sample twice as large as that used in simple random sampling (SRS) (i.e. design effect of 2.00) may not be sufficient. This paper provides empirical evidence of sample-to-sample variability in design effects using data from nine studies in six countries among injecting drug users, female sex workers, men who have sex with men and male-to-female transgender (MTF) persons. We computed the design effect as the variance under RDS divided by the variance under SRS for a broad range of demographic and behavioral variables in each study. We also estimated several measures for each variable in each study that we hypothesized might be related to design effect: the number of waves needed for equilibrium, homophily, and mean network size. Design effects for all studies ranged from 1.20 to 5.90. Mean design effects among all studies ranged from 1.50 to 3.70. A particularly high design effect was found for employment status (design effect of 5.90) of MTF in Peru. This may be explained by a "bottleneck"--defined as the occurrence of a relatively small number of recruitment ties between two groups in the population. A design effect of two for RDS studies may not be sufficient. Since the mean design effect across all studies was 2.33, an effect slightly above 2.00 may be adequate; however, an effect closer to 3.00 or 4.00 might be more appropriate.

Respondent driven sampling for HIV biological and behavioral surveillance in Latin America and the Caribbean.

Since 2005, respondent driven sampling (RDS) has been widely used for HIV biological and behavioral surveillance surveys (BBSS) in Latin America and the Caribbean (LAC). In this manuscript, we provide a focused review of RDS among hard-to-reach high-risk populations in LAC and describe their principal operational, design, and analytical considerations. We reviewed published and unpublished reports, protocols, and manuscripts for RDS studies conducted in LAC between January 1, 2005 and December 31, 2011. We abstracted key operational information and generated summary statistics across all studies. Between 2005 and 2011, 87 RDS studies were conducted in 15 countries in LAC (68 % in South America, 18 % in Mexico and Central America, and 14 % in the Caribbean). The target populations were primarily men who have sex with men (43 %), sex workers (29 %), and drug users (26 %). Study considerations included establishing clear eligibility criteria, measuring social network sizes, collecting specimens for biological testing, among others. Most of the reviewed studies are the first in their respective countries to collect data on hard-to-reach populations and the first attempt to use a probability-based sampling method. These RDS studies allowed researchers and public health practitioners in LAC to access hard-to-reach HIV high-risk populations and collect valuable data on the prevalence of HIV and other infections, as well as related risk behaviors.

Evaluation of respondent-driven sampling.

BACKGROUND: Respondent-driven sampling is a novel variant of link-tracing sampling for estimating the characteristics of hard-to-reach groups, such as HIV prevalence in sex workers. Despite its use by leading health organizations, the performance of this method in realistic situations is still largely unknown. We evaluated respondent-driven sampling by comparing estimates from a respondent-driven sampling survey with total population data. METHODS: Total population data on age, tribe, religion, socioeconomic status, sexual activity, and HIV status were available on a population of 2402 male household heads from an open cohort in rural Uganda. A respondent-driven sampling (RDS) survey was carried out in this population, using current methods of sampling (RDS sample) and statistical inference (RDS estimates). Analyses were carried out for the full RDS sample and then repeated for the first 250 recruits (small sample). RESULTS: We recruited 927 household heads. Full and small RDS samples were largely representative of the total population, but both samples underrepresented men who were younger, of higher socioeconomic status, and with unknown sexual activity and HIV status. Respondent-driven sampling statistical inference methods failed to reduce these biases. Only 31%-37% (depending on method and sample size) of RDS estimates were closer to the true population proportions than the RDS sample proportions. Only 50%-74% of respondent-driven sampling bootstrap 95% confidence intervals included the population proportion. CONCLUSIONS: Respondent-driven sampling produced a generally representative sample of this well-connected nonhidden population. However, current respondent-driven sampling inference methods failed to reduce bias when it occurred. Whether the data required to remove bias and measure precision can be collected in a respondent-driven sampling survey is unresolved. Respondent-driven sampling should be regarded as a (potentially superior) form of convenience sampling method, and caution is required when interpreting findings based on the sampling method.