|
" Assessment of the use of respondent-driven sampling for studying populations most at risk of HIV infection "
Mohsen Malekinejad
M. Potts
| Document Type
|
:
|
Latin Dissertation
|
| Language of Document
|
:
|
English
|
| Record Number
|
:
|
52593
|
| Doc. No
|
:
|
TL22547
|
| Call number
|
:
|
3353467
|
| Main Entry
|
:
|
Mohsen Malekinejad
|
| Title & Author
|
:
|
Assessment of the use of respondent-driven sampling for studying populations most at risk of HIV infection\ Mohsen Malekinejad
|
| College
|
:
|
University of California, Berkeley
|
| Date
|
:
|
2008
|
| Degree
|
:
|
Dr.Ph.
|
| student score
|
:
|
2008
|
| Page No
|
:
|
168
|
| Abstract
|
:
|
This dissertation assesses the use of respondent-driven sampling (RDS) in HIV epidemiological surveys among most-at-risk populations (MARPs, e.g., injection drug users [IDUs], men who have sex with men) in international settings. Recruiting representative samples of MARPs for HIV surveys is challenging but crucial for generating accurate data for informing prevention and treatment programs. Developed in the US for sampling IDUs, RDS is a recent chain-referral sampling method theoretically capable of producing generalizable estimates of the parameters of target populations. My colleagues and I conducted three studies to address two outstanding concerns about RDS: its applicability to sampling MARPs other than those studied in initial RDS surveys and the lack of a gold-standard sampling methodology for validating RDS results. The first study reviewed methodological characteristics of RDS-based HIV surveys completed before October 2007 among MARPs outside the US in 28 countries across five continents. Of the 155 surveys identified, 123 (86%) had properly fulfilled RDS methodological requirements. We examined factors predicting successful recruitment and found that RDS was effective across various MARPs, particularly IDUs. The second study examined the effectiveness of RDS in recruiting IDUs for an HIV bio-behavioral survey in Iran, where accurate HIV surveillance data are limited, drug use is illegal, and HIV epidemics are growing. Following a slow start to the sampling process, 548 demographically-diverse IDUs were recruited after we increased the amount of incentives, number of recruitment sites, and collaboration with organizations trusted among IDUs. However, certain subclasses of IDUs (e.g. women, Afghan immigrants) were not reached. HIV prevalence was 26% (21.3%-32.1%). In the third study, we examined IDU peer-recruitment patterns in a 2005 RDS-based HIV behavioral survey in San Francisco, comparing the results with those of a concurrent IDU survey that employed targeted sampling. RDS generated a diverse sample of IDUs and reached subpopulations that had not been accessed by the targeted sampling study. This dissertation shows that RDS is an effective recruitment technique for studying MARPs and a robust sampling method capable of generating diverse samples. Further research is needed to establish if RDS-generated samples are representative, an assertion supported by current RDS theory.
|
| Subject
|
:
|
Health and environmental sciences; Respondent-driven sampling; HIVAIDS; Most-at-risk population; Injection drug users; Surveillance; Sampling; Public health; Epidemiology; 0573:Public health; 0766:Epidemiology
|
| Added Entry
|
:
|
M. Potts
|
| Added Entry
|
:
|
University of California, Berkeley
|
| |