This paper, which is the first large scale application of Respondent-Driven Sampling (RDS) to non-hidden populations, tests three factors related to RDS estimation against institutional data using two WebRDS samples of university undergraduates. First, two methods of calculating RDS point estimates are compared. RDS estimates calculated using both methods coincide closely, but variance estimation, especially for small groups, is problematic for both methods. In one method, the bootstrap algorithm used to generate confidence intervals is found to underestimate variance. In the other method, where analytical variance estimation is possible, confidence intervals tend to overestimate variance. Second, RDS estimates are found to be robust against varying measures of individual degree. Results suggest the standard degree measure currently employed in most RDS studies is among the best performing degree measures. Finally, RDS is found to be robust against the inclusion of out-of-equilibrium data. The results show that valid point estimates can be generated with RDS analysis using real data, however further research is needed to improve variance estimation techniques.

BackgroundInjection drug use provides an efficient mechanism for transmitting bloodborne viruses, including human immunodeficiency virus (HIV) and hepatitis C virus (HCV). Effective targeting of resources for prevention of HIV and HCV infection among persons who inject drugs (PWID) is based on knowledge of the population size and disparity in disease burden among PWID. This study estimated the number of PWID in the United States to calculate rates of HIV and HCV infection.MethodsWe conducted meta-analysis using data from 4 national probability surveys that measured lifetime (3 surveys) or past-year (3 surveys) injection drug use to estimate the proportion of the United States population that has injected drugs. We then applied these proportions to census data to produce population size estimates. To estimate the disease burden among PWID by calculating rates of disease we used lifetime population size estimates of PWID as denominators and estimates of HIV and HCV infection from national HIV surveillance and survey data, respectively, as numerators. We calculated rates of HIV among PWID by gender-, age-, and race/ethnicity.ResultsLifetime PWID comprised 2.6% (95% confidence interval: 1.8%–3.3%) of the U.S. population aged 13 years or older, representing approximately 6,612,488 PWID (range: 4,583,188–8,641,788) in 2011. The population estimate of past-year PWID was 0.30% (95% confidence interval: 0.19 %–0.41%) or 774,434 PWID (range: 494,605–1,054,263). Among lifetime PWID, the 2011 HIV diagnosis rate was 55 per 100,000 PWID; the rate of persons living with a diagnosis of HIV infection in 2010 was 2,147 per 100,000 PWID; and the 2011 HCV infection rate was 43,126 per 100,000 PWID.ConclusionEstimates of the number of PWID and disease rates among PWID are important for program planning and addressing health inequities.

Over half of HIV infections in the United States occur among men who have sex with men (MSM). Awareness of infection is a necessary precursor to antiretroviral treatment and risk reduction among HIV-infected persons. We report data on prevalence and awareness of HIV infection among MSM in 2008 and 2011, using data from 20 cities participating in the 2008 and 2011 National HIV Behavioral Surveillance System (NHBS) among MSM. Venue-based, time-space sampling was used to recruit men for interview and HIV testing. We analyzed data for men who reported ≥1 male sex partner in the past 12 months. Participants who tested positive were considered to be aware of their infection if they reported a prior positive HIV test. We used multivariable analysis to examine differences between results from 2011 vs. 2008. HIV prevalence was 19% in 2008 and 18% in 2011 (p = 0.14). In both years, HIV prevalence was highest among older age groups, blacks, and men with lower education and income. In multivariable analysis, HIV prevalence did not change significantly from 2008 to 2011 overall (p = 0.51) or in any age or racial/ethnic category (p>0.15 in each category). Among those testing positive, a greater proportion was aware of their infection in 2011 (66%) than in 2008 (56%) (p<0.001). In both years, HIV awareness was higher for older age groups, whites, and men with higher education and income. In multivariable analysis, HIV awareness increased from 2008 to 2011 overall (p<0.001) and for all age and racial/ethnic categories (p<0.01 in each category). In both years, black MSM had the highest HIV prevalence and the lowest awareness among racial/ethnic groups. These findings suggest that HIV-positive MSM are increasingly aware of their infections.

Time-location sampling (TLS) is useful for collecting information on a hard-toreach population (such as men who have sex with men [MSM]) by sampling locations where persons of interest can be found, and then sampling those who attend. These studies have typically been analyzed as a simple random sample (SRS) from the population of interest. If this population is the source population, as we assume here, such an analysis is likely to be biased, because it ignores possible associations between outcomes of interest and frequency of attendance at the locations sampled, and is likely to underestimate the uncertainty in the estimates, as a result of ignoring both the clustering within locations and the variation in the probability of sampling among members of the population who attend sampling locations. We propose that TLS data be analyzed as a two-stage sample survey using a simple weighting procedure based on the inverse of the approximate probability that a person was sampled and using sample survey analysis software to estimate the standard errors of estimates (to account for the effects of clustering within the first stage [locations] and variation in the weights). We use data from the Young Men's Survey Phase II, a study of MSM, to show that, compared with an analysis assuming a SRS, weighting can affect point prevalence estimates and estimates of associations and that weighting and clustering can substantially increase estimates of standard errors. We describe data on location attendance that would yield improved estimates of weights. We comment on the advantages and disadvantages of TLS and respondent-driven sampling.

Respondent-driven sampling (RDS) has become increasingly popular for sampling hidden populations, including injecting drug users (IDU). However, RDS data are unique and require specialized analysis techniques, many of which remain underdeveloped. RDS sample size estimation requires knowing design effect (DE), which can only be calculated post hoc. Few studies have analyzed RDS DE using real world empirical data. We analyze estimated DE from 43 samples of IDU collected using a standardized protocol. We find the previous recommendation that sample size be at least doubled, consistent with DE = 2, underestimates true DE and recommend researchers use DE = 4 as an alternate estimate when calculating sample size. A formula for calculating sample size for RDS studies among IDU is presented. Researchers faced with limited resources may wish to accept slightly higher standard errors to keep sample size requirements low. Our results highlight dangers of ignoring sampling design in analysis.

This study tests the feasibility, effectiveness, and efficiency of respondent-driven sampling (RDS) as a Web-based sampling method. Web-based RDS (WebRDS) is found to be highly efficient and effective. The online nature of WebRDS allows referral chains to progress very quickly, such that studies with large samples can be expected to proceed up to 20 times faster than with traditional sampling methods. Additionally, the unhidden nature of the study population allows comparison of RDS estimators to institutional data. Results indicate that RDS estimates are reasonable but not precise. This is likely due to bias associated with the random recruitment assumption and small sample size of the study. Finally, this article presents methods for testing the validity of assumptions required by RDS estimation.

Background: Persons who inject drugs (PWID) are at increased risk for poor health outcomes and bloodborne infections, including human immunodeficiency virus (HIV), hepatitis C virus and hepatitis B virus infections. Although substantial progress has been made in reducing HIV infections among PWID, recent changes in drug use could challenge this success.

Sexually transmitted diseases (STDs) disproportionately affect gay, bisexual, and other men who have sex with men (MSM) in the United States ( 1 ). Because chlamydia and gonorrhea at extragenital (rectal and pharyngeal) anatomic sites are often asymptomatic, these anatomic sites serve as a reservoir of infection, which might contribute to gonococcal antimicrobial resistance ( 2 ) and increased risk for human immunodeficiency virus (HIV) transmission and acquisition ( 3 ). To ascertain prevalence of extragenital STDs, MSM attending community venues were recruited in five U.S. cities to provide self-collected swabs for chlamydia and gonorrhea screening as part of National HIV Behavioral Surveillance (NHBS). Overall, 2,075 MSM provided specimens with valid results, and 13.3% of participants were infected with at least one of the two pathogens in at least one of these two extragenital anatomic sites. Approximately one third of participating MSM had not been screened for STDs in the previous 12 months. MSM attending community venues had a high prevalence of asymptomatic extragenital STDs. The findings underscore the importance of sexually active MSM following current recommendations for STD screening at all exposed anatomic sites at least annually ( 4 ).