Comparison of surveillance sample demographics over two cycles of the National HIV Behavioral Surveillance Project, Houston, Texas

Assessing reliability of naïve respondent-driven sampling samples by using repeated surveys among people who inject drugs (PWID) in New Jersey

INTRODUCTION

Respondent-driven sampling (RDS) is widely used to sample populations with higher risk of HIV infection for whom no sampling frames exist. However, few studies have been done to assess the reliability of RDS in real world settings.

METHODS

We assessed the reliability of naïve RDS samples using five rounds of the National HIV Behavioral Surveillance - People Who Inject Drugs surveys in Newark, New Jersey from 2005 to 2018. Specifically, we compared the distributions of time-insensitive demographic characteristics in temporally adjacent RDS samples with Monte Carlo Two-Sample Kolmogorov-Smirnov Test with 100,000 replicates. The distributions of time-sensitive demographic characteristics were also compared as sensitivity analyses.

RESULTS

The study showed that repeated RDS samples among people who inject drugs in the greater Newark area, New Jersey were reliable in most of time-insensitive demographics and recruitment homophily statistics. Sensitivity analyses of time-sensitive demographics also presented consistencies in most of temporally adjacent samples.

CONCLUSIONS

In conclusion, RDS has the potential to provide reliable samples, but demographic characteristics of RDS samples may be easily biased by homophily. Future studies using RDS may need to pay more attention to potential homophily bias and consider necessary diagnostic procedures and sample adjustments.

Original title: HIV infection and unawareness among men who have sex with men in Puerto Rico: data from The National HIV behavioral surveillance system 2011-2014

BACKGROUND

Unawareness of HIV infection is a public health challenge that needs to be addressed, particularly in the case of men who have sex with men (MSM), since recent data are reporting increasing rates of HIV in this population in Puerto Rico.

OBJECTIVES

We examined differences in the prevalence of HIV infection and unawareness among MSM in 2011 and 2014 using data from the National HIV Behavioral Surveillance System, 2011 and 2014.

METHODS

Bivariate analyses was used to compare demographical and behavioral characteristics in both cycles (2011 and 2014). Prevalence ratio (PR) was assessed with Poisson regression models to determine changes in HIV prevalence and unawareness across cycles, using the 2011 NHBS-MSM cycle as reference group.

RESULTS

A stable rate in HIV prevalence was observed in 2011 and 2014. There was a higher prevalence in 2014 than in 2011 in multiple behavioral characteristics such as age at sexual initiation, the number of sexual partners in the 12 months prior to being interviewed, HIV testing in the year prior to being interview, and the disclosure of sexual orientation to a healthcare provider. A significant decrease in HIV unawareness was reported (76.67%, 2011; 46.51%, 2014). Age-adjusted regression models showed a marginal reduction of 55% in HIV unawareness for men who disclosed their sexual orientation to their healthcare providers.

CONCLUSION

Behavioral surveillance systems in groups such as MSM in on the island will aid to monitor prospectively the effectiveness of HIV testing outreach and engagement, as well as capacity capacity-building strategies targeted towards health care providers, aimed to increase HIV testing and awareness among this group.

Design-based inference in time-location sampling

Time-location sampling (TLS), also called time-space sampling or venue-based sampling is a sampling technique widely used in populations at high risk of infectious diseases. The principle is to reach individuals in places and at times where they gather. For example, men who have sex with men meet in gay venues at certain times of the day, and homeless people or drug users come together to take advantage of services provided to them (accommodation, care, meals). The statistical analysis of data coming from TLS surveys has been comprehensively discussed in the literature. Two issues of particular importance are the inclusion or not of sampling weights and how to deal with the frequency of venue attendance (FVA) of individuals during the course of the survey. The objective of this article is to present TLS in the context of sampling theory, to calculate sampling weights and to propose design-based inference taking into account the FVA. The properties of an estimator ignoring the FVA and of the design-based estimator are assessed and contrasted both through a simulation study and using real data from a recent cross-sectional survey conducted in France among drug users. We show that the estimators of prevalence or a total can be strongly biased if the FVA is ignored, while the design-based estimator taking FVA into account is unbiased even when declarative errors occur in the FVA.