Enumeration of hard-to-reach populations for a community-based vaccine distribution study

Abstract

Purpose Estimates of sizes of hard-to-reach (HTR) populations are necessary to estimate disease and intervention success rates in these groups. We aimed to enumerate HTR populations (homeless, sex workers, drug users, elderly shut-ins, immigrants) in 8 neighborhoods of East Harlem and the Bronx as part of community-based influenza vaccine distribution study. Methods Estimation of the number of HTR individuals in the 8 neighborhoods was performed using a modified Delphi technique and venue-based, door-to-door, and capture-recapture sampling methods between January and August 2004. Results Estimates of each of the groups of interest from the four enumeration methods were: 932 – 2620 homeless (mean = 1599, median = 1423); 804 – 275 sex workers (mean = 452, median = 364); 4268 – 7966 drug users (mean =5936, median = 5759); 747 – 4400 elderly (mean = 3101, median = 3629); and 5813 – 13,713 immigrants (mean = 9736, median = 9708). Overall, the most conservative estimates for homeless individuals and drug users were given by the door-to-door sampling method; for sex workers it was the modified Delphi method and for the elderly and immigrants it was the capture-recapture method. Estimates obtained from the venue-based sampling method were most widely divergent from the overall mean for the homeless, sex workers and substance users. The capture-recapture method estimate was the most divergent for the elderly, while for immigrants it was the door-to-door method. Conclusion Enumerating hidden populations represents a special public health challenge. The variety of estimates gathered by using different techniques provides a sense of the complexity in estimating counts of hidden populations. Certain techniques may be more appropriate for estimating the sizes of different of populations; in particular venue-based sampling appears to be the most effective method for reaching individuals who may not be well connected to mainstream society. These techniques can supplement the Census in providing an estimate that likely includes those less likely to respond to an official census.