JMIR Publications

ABSTRACT

Background:

Population size estimates for hidden populations at increased risk of HIV, including female sex workers (FSW), are important to inform public health policy and resource allocation. The service multiplier method is commonly used to estimate the sizes of hidden populations. We used this method to obtain population size estimates for FSW at nine sites in Zimbabwe and explored methods for assessing potential biases that could arise in using this approach.

Objective:

To provide guidance on the assessment of bias that arise when estimating the population sizes of hidden populations using the service multiplier method based on respondent-driven sampling.

Methods:

We conducted respondent-driven sampling (RDS) surveys at nine sites in late 2013 where the Sisters FSW programme, which collects programme visit data, was also present. Using the service multiplier method, we obtained population size estimates for the FSW in each site by dividing the number of FSW who attended the Sisters programme, based on programme records, by the RDS-II weighted proportion of FSW who reported attending this programme in the previous six months in the RDS survey. Both the RDS weighting and the service multiplier method make a number of assumptions, potentially leading to biases if the assumptions are not met. To test these assumptions, we used convergence and bottleneck plots to assess seed dependence of the RDS-II proportion estimates, the chi-squared test to assess if there was an association between characteristics of women and knowledge of FSW programme existence, and logistic regression to compare the characteristics of women attending the programme with those in RDS data.

Results:

The population size estimates ranged from 194 (95% CI: 62-325) to 805 (95% CI: 456-1142) across the nine sites for the period May to November 2013. The 95% CIs for the majority of sites were wide. In some sites, the RDS-II proportion of women who reported programme use in the RDS survey may have been influenced by the characteristics of selected seeds and we also observed bottlenecks in some sites. There was no evidence of association between characteristics of FSW and knowledge of programme existence, and in the majority of sites there was no evidence that the characteristics of the populations differed between RDS and programme data.

Conclusions:

We used a series of rigorous methods to explore potential biases in our population size estimates and found that we were able to identify these as well as the potential but not ultimate direction of bias in our estimates. We have some evidence that the population size estimates in most sites may be biased, some suggestion that the bias is toward underestimation and this should be considered if the population size estimates are to be used. These tests for bias should be included when undertaking population size estimation using the service multiplier method combined with RDS surveys.