Indan Journal of Medical Research Indan Journal of Medical Research Indan Journal of Medical Research Indan Journal of Medical Research
  Home About us Editorial board Search Ahead of print Current issue Archives Submit article Instructions Subscribe Contacts Login  
  Home Print this page Email this page Small font sizeDefault font sizeIncrease font size Users Online: 1236       

   Table of Contents      
Year : 2012  |  Volume : 136  |  Issue : 7  |  Page : 36-43

Recruitment of at-risk population through network based sampling: experiences from Maharashtra IBBA Round-I survey

1 National AIDS Research Institute (ICMR), Pune, India
2 National AIDS Research Institute (ICMR), Pune; Formally associated with Family Health International, New Delhi, India
3 National AIDS Research Institute (ICMR), Pune; Formally associated with IBBA project in National AIDS Research Institute, Pune, India

Date of Submission03-Aug-2010
Date of Web Publication1-Dec-2012

Correspondence Address:
Ramesh Paranjape
Director, National AIDS Research Institute (ICMR), Plot No. 73, "G" Block, M.I.D.C. Bhosari, Pune 411 026
Login to access the Email id

Source of Support: None, Conflict of Interest: None

Rights and PermissionsRights and Permissions

Background & objectives: Respondent Driven Sampling (RDS) method is increasingly being used for surveys among hard to reach populations. It essentially relies on the networking among the target population. Key behavioural indicators, HIV and sexually transmitted infection (STI) prevalence were measured under Integrated Behavioural and Biological Assessment (IBBA), designed to study the impact of Avahan India AIDS Initiative. RDS surveys were conducted in 2006 and 2007 among high risk populations.
Methods: Four separate surveys covered female sex workers (FSW) (Mumbai, Parabhani), Bar Girl (Mumbai) and injecting drug users (IDU) (Mumbai-Thane) population through RDS. Respondents were recruited through the personal and social networks of the initially selected "seeds". Successive recruitments were done through snowball effect. Duel compensation for participation and recruitment was given. Three uniquely numbered coupons were given to recruit peers from network. Through subsequent waves of recruitment desirable sample was achieved. Informed and voluntary consent was obtained.
Results: A total of 338 bar girls, 403 FSWs (Mumbai), 367 FSW (Parbhani) and 376 IDUs (Mumbai/Thane) were recruited through RDS. The desired samples were achieved within three months period. Average seeds recruited for each survey ranged from 22 to 35. Average reported network size ranged from 4.8 to 13.2. Homophily index for each survey group was found to be 0.0 except for bar girl survey (0.3). Some seeds yielded multiple waves of recruitment while many failed to go beyond first wave.
Interpretation & conclusions: Our findings suggest that RDS does not appear to recruit more marginalized or undercovered section of the targeted group. Greater preparatory activity and better understanding of networks may be required for setting up appropriate venues across geographical boundaries. Further in-depth network analysis is needed in diverse regional structures and population coverage in conducting RDS surveys locally.

Keywords: At-risk population - FSW - HIV - IDU - respondent-driven sampling

How to cite this article:
Deshpande S, Kohli A, Rathod S, Mainkar M, Kazi S, Pardeshi D, Dale J, Aralkar S, Panchal N, Mahajan U, Paranjape R. Recruitment of at-risk population through network based sampling: experiences from Maharashtra IBBA Round-I survey. Indian J Med Res 2012;136, Suppl S1:36-43

How to cite this URL:
Deshpande S, Kohli A, Rathod S, Mainkar M, Kazi S, Pardeshi D, Dale J, Aralkar S, Panchal N, Mahajan U, Paranjape R. Recruitment of at-risk population through network based sampling: experiences from Maharashtra IBBA Round-I survey. Indian J Med Res [serial online] 2012 [cited 2020 Sep 22];136, Suppl S1:36-43. Available from:

Various probability and non-probability based sampling approaches are in use for the surveillance surveys across the world. Time location sampling (TLS) and conventional cluster sampling (CCS), probability based sampling methods are widely used in most of the surveys conducted in at-risk populations. TLS provides a potent solution for reaching out to several mobile sub-groups at different levels of risk who do not associate themselves with any site in a fixed manner. However, both TLS and CCS cannot cover populations that are hard to reach due to the tainted nature of behaviours such as intravenous drug use [1] . TLS and conventional sampling may cause a bias of representativeness of the sample if significant proportion of subpopulation is not frequenting the visible sites.

To overcome the problem of capturing hard-to-reach populations, respondent driven sampling (RDS) approach was introduced [2],[3],[4] . RDS is mainly designed to draw samples from the hidden sub-groups which usually do not come in open due to stigmatized risk behaviour. An underlying assumption in RDS is that it draws samples from personal networks. Since personal networks are interlinked in the population thus within a defined geographic boundary, the sampling universe is aggregated to social network of that survey population. In RDS, sampling begins with a purposive selection of individuals, referred as ''seeds,''. Generally seeds are chosen on the background of their network size and diversity (i.e. they know 5-10 other people from their own community and are from different geographical locations or sites of solicitation or drug use, etc.). Starting with the seeds, each participant is allowed to recruit pre-specified number of respondents within their networks. As the enrollments proceed, sample becomes representative. In RDS, it is assumed that when the equilibrium occurs, the characteristics of the sample no longer change during subsequent waves of enrollment. Another important concept in RDS is "homophily"; which means cohesiveness of the groups within the networks. It is stated that attainment of equilibrium is based on the network homophily [5] . It is argued that the sample becomes increasingly more representative of the underlying population as the successive wave recruitment progresses [6] .

There are a few evidences of successful implementation of RDS in at-risk population in a variety of settings, resulting in the rapid attainment of long and varied recruitment chains [7] . Reports of recruiting and collecting both behavioural and biological data through RDS from sex workers and men who have sex with men (MSM) are increasing [8] . However, issues about the actual implementation of RDS and factors that may influence the application of RDS have not been discussed widely.

In the Integrated Behavioural and Biological Assessment (IBBA) survey an attempt was made to capture hidden but at-risk groups through RDS method. In the first round of IBBA, total four surveys [3 female sex worker (FSW) surveys and 1 injecting drug user survey (IDU)] in Maharashtra State, India were conducted using RDS [9] . We share here our experience and challenges faced in applying RDS in Maharashtra State and discuss the key issues of RDS implementation.

   Material & Methods Top

The [Table 1] describes the details of the RDS surveys conducted in Maharashtra. In the first round of IBBA, RDS surveys were conducted during November 2006 to November 2007.
Table 1: Details of RDS surveys in Maharashtra State, India

Click here to view

A sample size of 400 was estimated for each survey district (a combined sample size of 400 for IDUs was used for Mumbai and Thane referred to as 'Mumbai/Thane'. Pre-survey assessment included a combination of key informant interviews, group discussions with various stakeholders and rapid field visits to assess feasibility of RDS survey in specific geographical region. The pre-survey assessment activities were undertaken to understand the nature of sex work, injecting drug practices and the current socio-political situation around high risk behaviour. For recruiting initial seeds and their consecutive chains and to understand their network size, each of the respondents were asked about the number of people from the community under survey they know personally. A social network was defined as the number of known members indulging in the same risk behaviour, number of members seen in the last month and members living or working in the district.

A dual compensation was paid to the respondents. Primary compensation was given for participating in the survey which essentially covered the wages lost and conveyance cost for participation in the survey and secondary compensation for recruiting peers from the networks. All the participants were given three uniquely numbered coupons to recruit peers from their network. The peers were expected to report at the RDS site with the coupon received by them. Participants were screened for their eligibility and information on how recruiter and recruitment are known to each other was gathered. In addition, information on whether anyone refused to accept coupons was obtained from those who returned to collect their secondary compensation. Through the subsequent waves of recruitment, desirable sample was achieved.

Written informed consent was obtained from all eligible respondents. Respondents were linked anonymously, maintaining the confidentiality. Consenting individuals were interviewed for the behavioural indicators such as demographic profile, injection use and sexual behaviour, self reporting STI, HIV/AIDS related knowledge and exposure to interventions (detailed questionnaire is posted on the website Blood and urine samples were collected from all consenting respondents and ulcer swabs from those who reported genital ulcer on physical examination. Tests for prevalent STIs (syphilis, herpes simplex virus-2, Niesseria gonorrhoeae and Chlamydia trachomatis) were carried out for all participants. Samples from IDUs were additionally tested for hepatitis B and C.

Four RDS surveys with Bar girls, FSWs and IDUs were conducted in Mumbai, Thane and Parbhani in Maharashtra. Total 11 teams were trained for RDS survey. To capture the maximum concentration of the survey population, RDS venues were set up across each of the surveyed districts. RDS venues were set up in consultation with non-government organizations (NGOs) and Community Advisory Board (CAB) and Community Monitoring Board (CMB) members. All RDS venues were set up in public health facility. In some cases, centres were kept mobile to capture all possible areas where key population existed in higher concentration. Two permanent centres were kept open for all working days and one centre mobile with flexibility in timings and days.

RDSAT 5.4 (an RDS analysis tool) was used for preliminary analysis of network and recruitments. Seed network is presented graphically with the help of NetDraw software.

   Results Top

Bar girl (BG) survey (Mumbai): A total of 338 bar girls were recruited in three RDS centres within three months of survey period. Majority of bar girls operated independently, through phones and through their existing client networks. Due to confidentiality issues and inadequate conditions, it was difficult to contact BGs at their residence or bars. But focused group discussion (FGDs) and personal meetings with key informants revealed that BGs have their own groups, stay and work in groups, therefore, decision was taken to cover BGs with RDS in Mumbai.

A total of 26 seeds were deployed during the survey. Of the initial 10 seeds, only four produced any subsequent waves. The remaining 16 seeds were recruited over three months survey time. Eight seeds did single recruitment. The rate of recruitment was fast in the first two weeks of the survey. Due to difficulties in identifying and sustaining "good" seeds, pace of recruitment was disrupted. The average network size reported by BGs was 13.2 (adjusted). [Figure 1] shows the recruitment patterns of seed. Seed 3 recruited more than 50 respondents, reaching till 7 th wave. Whereas most of seeds ended up in zero recruitment or reached up to only 2 nd or 3 rd wave. Homophily index [5] of this group was 0.3 indicating less clustering among same group. Further analysis showed that the comparison of seeds who have produced subsequent recruitment chains and who have not were not different in terms of profile characteristics such as age, education, marital status, duration in sex work, condom use, HIV risk perception and HIV prevalence.
Figure 1: Recruitment of bar girls by seed 3, Mumbai.

Click here to view

All FSW survey (Mumbai): A total of 403 FSWs were recruited in five different centres at Mumbai. These comprised various types of sex workers viz. street based sex workers, call girls, sex workers in massage parlors, bar girls, etc. Diverse seeds of different typologies were employed.

The survey lasted for three months. Initially ten seeds were recruited and 12 more were added later. Of the total 22 seeds, six did not produce any recruitment and seven seeds attained more than 5 waves. The average network size reported was 7.9 (adjusted). Majority of FSWs solicited their clients over phone (34%), followed by public places (28%), home and bar (14 and 10%, respectively).

This was a heterogeneous group surveyed through RDS. It was assumed that different sex workers could be reached through a dynamic seed, but the recruitment was more homogeneous than the heterogeneous. The homophily index was 0.0 meaning that all recruitments were random. Even though recruitments were random, it was observed that same types of sex workers were recruited by the seeds that were of same topology. A street based FSW seed recruited nearly 60 FSWs and reached till 6 th wave. As seen in seed 14, different types of sex workers were recruited as the recruitment wave increased [Figure 2]. This was the exceptional case in all seeds, rest of the seeds have not shown the same trend and failed to even recruit more than 5-8 respondents.
Figure 2: Recruitment of all FSW by seed 14, Mumbai.

Click here to view

FSW survey (Prabhani): A total of 367 FSWs were recruited in three centres. Prabhani is situated in the Marathwada region and is less developed with widely prevalent water scarcity. Number of reported FSWs in a district ranged from 300 to 1500 [9] . There was an increasing reporting of hidden sex workers in the districts. Tamasha theater (a type of folk dance) is a peculiarity of the rural Maharashtra. Parbhani has two main tamasha spots served as solicitation as well as sex point.

Group discussions with key informants and stakeholders reported about mobility and deeper networking of the population, however cross-over of sex workers was not seen in this survey. With an initial seven seeds and two additional seeds, 367 samples were achieved. Attempts to recruit home based sex workers as seed failed primarily because of insufficient and strong network among these secretive sex workers. Of the nine seeds, only three produced successive recruitments and interestingly one single seed contributed 70 per cent of total recruitments. Seed 1 grew till 14 th wave and recruited 267 respondents [Figure 3].
Figure 3: Recruitment of female sex workers seed 1, Parbhani.

Click here to view

The average network size reported was 4.8 (adjusted). Majority of FSWs solicited their clients through brothel (32%), followed by public places (22%), home and rented room (19 and 14%, respectively). This showed that majority of the respondents were affiliated to identifiable venues. Tamasha based FSW seed recruited all Tamasha based FSWs.

IDU survey (Mumbai-Thane): A total of 376 IDUs were recruited through five different centres, across Mumbai and Thane. Volatile nature of IDUs made it hard to capture then on their injecting sites. Discussions with stakeholders and key population revealed that locating any centre near to the hot spots of IDUs would invite unnecessary trouble for the key population.

The survey was started with 14 initial seeds and in due course of time, 23 more seeds were added. It took more than three months to achieve 376 samples. Of the 34 seeds, only five reached till 5 th wave and more than 20 seeds have hardly produced 5 recruitments. One seed had grown up to 13 th wave and recruited more than 90 respondents. The average network size mentioned by the IDUs was approximately 9.2 (adjusted). Network connections were found totally random, i.e. 0.0 homophily index. The proportion of dying of seeds was much higher in IDU survey. Almost 12 seeds had zero recruitment. The same trend of one seed recruiting majority of the recruitments was seen in IDU survey also. Sustainability of seeds and follow up with IDUs were two major hurdles in successful recruitments [Figure 4].
Figure 4: Recruitment of IDU seed 25, Mumbai-Thane.

Click here to view

   Discussion Top

The prerequisite of RDS is that the survey population must be socially networked and has cohesiveness among members [10] . This networked based sampling enhances the chances of hidden individuals being sensitized by their friends/peers to take part in the survey. Thus RDS can make an inroad into hidden populations [11] . Exception to this in Maharashtra, it was observed that most "hidden" populations have not been captured through RDS. The one of the reason could be surveyed populations were less mobile and has no cross-over. This was in contrast to the RDS principle that it is more 'inclusive' in the design and can represent a better cross-section of both accessible and hidden population [12] .

RDS is considered relatively easy to implement and less costly [13] . In contrast to this, RDS surveys in Maharashtra took much longer time than the TLS or CCS. TLS and CCS surveys were finished within 4 - 5 wk time and RDS surveys took more than 12 wk to achieve desired samples with higher cost implications. The approximate cost of collecting single sample in TLS/CCS was 1000-1200 whereas in RDS, it was three times higher i.e. ranged from 2500 to 3500. This was mainly due to duel compensation and duration of the survey period. Another important assumption in RDS is the start-up period of RDS is shorter and less expensive when compared to venue based probability sample surveys. This was not in case of our surveys, it took almost one and half month to start the RDS survey. It involved much more manpower than TLC/CCS. At the same time, in the RDS survey end-time of the survey often remained open-ended as might have slowed down due to saturation of recruitments.

In spite of undertaking extensive pre-survey assessment activity and investing considerable time in community preparation, RDS surveys were found to be more demanding and rigorous. This was in contrast with other RDS surveys conducted with HRGs in various other settings. These studies have shown fast recruitments with a few initial seeds in a minimum time [9],[14],[15] . IBBA surveys conducted in north east (NE) regions of India showed a different picture than Maharashtra [16] . Handful of seeds successively achieved sample size within 8-9 wk time with much lesser incentives given to the respondents (primary compensation 70 and 30 secondary compensation) in NE whereas in Maharashtra more than 30 seeds were recruited and survey lasted for three months and higher compensation paid to respondents.

In some cases, sex workers have reacted favourably to the monetary incentives which were below the rate charged for sex in that particular geographic region. It is evident that the incentives offered are often perceived as being too low or high in such types of surveys [10],[17] . In order to get primary and secondary compensation respondents have to leave their work and travel to and from the fixed RDS centres at least twice and spend more than an hour. The cost of their travel and time may be higher than their usual earnings. This could also have affected speed of recruitments. The duel compensation system worked well in Parbhani district.

Factors like size of the population and spread of their concentration were major concerns for the survey. Estimate of the size of the bar girl population across sub-urban area of Mumbai is more than 10,000 [18],[19] . The number of bars with different types such as service bars, entry bars, and orchestra bars, etc. where girls entertained and served clients, is more than 5000 in Mumbai [20] . In 2004 Maharashtra government banned bars in Mumbai [21] . Another reason for the poor rate of recruitments among more homogeneous groups like IDUs is the characteristics of their network. RDS requires the population from sufficiently large and dense network [22],[23],[24],[25] . Even though reported network size was more in IDUs, the population is very dispersed and scattered across Mumbai and Thane. The spread of population across geographical boundaries needs to be studied more in terms of IDU networks, especially interlinking of individual networks which is essential for RDS success.

Locations of the RDS centres may have played important role in recruitments. Shifting of RDS centres from one place to another after the initial assessment of the survey might have disturbed the flow of the recruitments. Other contributing factors could be heavy rainfall and festive seasons during the survey period. In Mumbai, concentration of subpopulations and spread of the city must have had impact on implementation of RDS. Similarly, an attempt was made in recruiting diverse seeds for maximizing chances of getting diversity and penetrate deeper in to the hidden population. However, survey results did not reflect this. One of the reasons for this could be that initial seeds did not actually have larger and denser networks as they reported. This resulted in larger proportion of non-productive seeds and added burden on research team to add more seeds. Further area of exploration in successful implementation of RDS could be regional variation, proportion of subpopulation and characteristics of the region.

It is assumed that RDS is a flexible and robust method that can produce a sample of representative of the heterogeneity of the target population [26],[27] . In contrast to this, the characteristics of the final samples were not quite different from the initially selected seeds and FSWs covered from TLS or CCS methods. More comparative and in-depth analyses need to be carried out in examining demographic characteristics, risk behaviours and STI and HIV prevalence of FSWs surveyed through TLS, CCS and RDS.

In TLS, teams had flexibility in moving and had surety of covering minimum samples a day. Whereas in RDS, teams were stationed at one place and waiting for the recruitments to come to the centres. Most of the RDS centres ended up in having zero recruitment on a majority of the days. This showed a different pattern of recruitment than the other studies [10] . In our survey, overall speed of recruitment was low and to gear-up the survey lot of efforts were made. In our survey, very few seeds produced subsequent waves. It invites for further in-depth "network analysis" which could probably throw light on the pattern and traits of individual seeds as well as proportion of populations surveyed through RDS.

There is paucity of evidence on the use of RDS amongst hard-to-reach populations in developing countries. RDS alone may not be able to generate rich qualitative information; it should be supported with extensive ethnographic and formative research, which was a limitation of our RDS survey. This could be seen as a disadvantage of a large scale bio-behavioural surveys where objectives of the survey are different than collecting more information on contextual and in-depth knowledge on networks.

Data collected from TLS, CCS and RDS in IBBA did not show significant difference in the characteristics and HIV seroprevalence rates of different survey groups [28] . RDS surveys in Maharashtra provided an opportunity to reach to the more marginalized and inaccessible groups for the first time in the large scale bio-behavioural surveys. This will lead to better understanding of the proportion of hidden population at risk and HIV prevalence. Current programmes are targeting those who are most likely to be visible, and accessible at points but RDS had provided information on the reaching to the at-risk but hard to reach segments of the subpopulation.

   Acknowledgment Top

The IBBA Study Teams

National AIDS Research Institute (NARI), Pune

Abhijit Deshpande, Amey S., Amol Salagare, Arun Risbud, B. Kishorekumar, Deepak More, Geetanjali Mehetre, Jagnnath Navale, Milind Pore, Rahul Gupta, Raman Gangakhedkar, Sachin Kale, Sachin P., Shashikant Vetal, Shradha Gaikwad, Shradha Jadhav, Sujata Zankar, Tanuja Khatavkar, Trupti Joshi.

Family Health International (FHI)

Ajay Prakash, Ashim Chatterjee, Bitra Grorge, Deepak Singh, Gay Thongamba, Katheleen Kay, Prabudhagopal Goswami, Prashant Alur, Rajatshuvra Adhikari, Sharad Malhotra, Shubhra Rehman, Sumita Taneja, Tilak Angra, Tobi Saidel, Umesh Chawala.

TNS Mode Private Limited, Mumbai

Jagdish Krishnappa, Asif Kunwar

Authors acknowledge the Bill and Melinda Gates Foundation for financial support through a grant to Family Health International (FHI).

   References Top

1.Heckathorn DD. Respondent-driven sampling: a new approach to the study of hidden populations. Soc Probl 1997; 44 : 174-99.  Back to cited text no. 1
2.Heckathorn DD. Respondent-driven sampling II: deriving valid population estimates from chain-referral samples of hidden populations. Soc Probl 2002; 49 : 11-34.  Back to cited text no. 2
3.Magnani R, Sabin K, Saidel T, Heckathorn DD. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS 2005; 19 (Suppl 2): S67-72.  Back to cited text no. 3
4.Schwartlander B, Ghys PD, Pisani E, Kiessling S, Lazzari S, Carael M, et al. HIV surveillance in hard-to-reach populations. AIDS 2001; 15 (Suppl 3): S1-3.  Back to cited text no. 4
5.Abdul-Quader AS, Heckathron DD, McKnight C, Bramson H, Nemeth C, Sabin K, et al. Effectiveness of Respondent-Driven Sampling for recruiting drug users in New York city: findings from a pilot study. J Urban Health 2006; 83 : 459-76.   Back to cited text no. 5
6.Salganik MJ. Varience estimation, design effects and sample size calculations for Respondent-Driven Sampling. J Urban Health 2006; 83 : i98-112.  Back to cited text no. 6
7.Platt L, Wall M, Rhodes T, Judd A, Hickman M, Johnston LG, et al. Methods to recruit hard-to-reach groups: Comparing two chain referral sampling methods of recruiting injecting drug users across nine studies in Russia and Estonia. J Urban Health 2006; 83 : i39-i53.  Back to cited text no. 7
8.Saglik MJ, Heckarthon DD. Sampling and estimation in hidden populations using respondent-driven sampling. Sociological Method 2004; 34 : 193-239.  Back to cited text no. 8
9.Saidel T, Adhikary R, Mainkar M, Dale J, Loo V, Rahman M, et al. Baseline integrated behavioral and biological assessment among most at-risk populations in six high-prevalence states of India: design and implementation challenges. AIDS 2008; 22 (Suppl 5): S17-S34.  Back to cited text no. 9
10. 2006.pdf, accessed on February 23, 2010.  Back to cited text no. 10
11.http//www.stat.gamma.rug.nl_Charles Kadushin_basic concepts of networks, accessed on December 2, 2009.  Back to cited text no. 11
12.Semaan S, Lauby J, Liebman J. Street and network sampling in evaluation studies of HIV risk-reduction interventions. AIDS Rev 2002; 4 : 213-23.  Back to cited text no. 12
13.Tiffany JS. Respondent-Driven Sampling in Participatory Research Contexts: Participant-Driven Recruitment. J Urban Health 2006; 83 : i113-24.   Back to cited text no. 13
14.Abdul-Quader AS, Heckathron DD, Sabin K, Saidel T. Implementation and analysis of RDS: Lessons learned from the field. J Urban Health 2006; 83 (Suppl); i1-i5.  Back to cited text no. 14
15.Robinson WT, Risser JM, McGoy S, Becker AB, Rehman H, Jefferson M, et al. Recruiting Injection Drug Users: A three site comparison of results and experiences with Respondent-Driven and Targeted sampling procedures. J Urban Health 2006; 83 : i29-38.  Back to cited text no. 15
16.Mahanta J, Medhi GK, Paranjape RS, Roy N, Kohli A, Akoijam BS, et al. Injecting and sexual risk behaviors, sexually transmitted infections and HIV prevalence in injecting drug users in three states in India. AIDS 2008; 22 (Suppl 5): S59-68.  Back to cited text no. 16
17.Stormer A, Tun W, Guli L, Harxhi A, Bodanovskaia Z, Yakovleva A, et al. An analysis of Respondent Driven Sampling with Injection Drug Users (IDU) in Albania and the Russian Federation. J Urban Health 2006; 83 : i73-82.  Back to cited text no. 17
18. singh, accessed on February 2010.  Back to cited text no. 18
19. shuts dance girls bars_30 march 2005.html, accessed on January 5, 2010.  Back to cited text no. 19
20., accessed on February 17, 2010.  Back to cited text no. 20
21., accessed on March 24, 2010.   Back to cited text no. 21
22.Yeka W, Geraldine MM, Dimitri P, Donn C. Application of Respondent Driven Sampling to collect baseline data on FSWs and MSM for HIV risk reduction interventions in Two Urban Centers in Papua New Guinea. J Urban Health 2006; 83 : i60-72.  Back to cited text no. 22
23.Simic M, Johnston LG, Platt L, Baros S, Andjelkovic V, Novotny T, et al. Exploring barriers to 'Respondent Driven Sampling' in sex workers and drug injecting sex worker populations in Eastern Europe. J Urban Health 2006; 83 : i6-15.  Back to cited text no. 23
24. Interventions/NACP-III.pdf, accessed on March 13, 2010.  Back to cited text no. 24
25.Johnston LG, Sabin K, Hien MT, Huong PT. Assessment of Respondent Driven Sampling for recruiting female sex workers in two Vietnamese cities: Reaching the unseen sex workers. J Urban Health 2006; 83 : i16-28.  Back to cited text no. 25
26.Malekinejad M, Johnston L, Kendall C, Kerr LR, Rifkin MR, Rutherfor GW. Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in International settings: A systematic review. AIDS Behav 2008; 12 : S105-30.   Back to cited text no. 26
27.Mills S, Saidal T, Mahnani R, Brown T. Surveillance and modelling of HIV, STI, and risk behaviours in concentrated HIV epidemics. Sex Transm Infect 2004; 80 (Suppl II): ii57-62.  Back to cited text no. 27
28.Banadakoppa RB, Moses S, Washington R, Isac S, Mohpatra B, Mahagaonkar SB, et al. Determinants of HIV prevalence among female sex workers in four south Indian states: analysis of cross-sectional surveys in twenty-three districts. AIDS 2008; 22 (Suppl 5): S35-44.  Back to cited text no. 28


  [Figure 1], [Figure 2], [Figure 3], [Figure 4]

  [Table 1]


    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

  In this article
   Material & Methods
    Article Figures
    Article Tables

 Article Access Statistics
    PDF Downloaded152    
    Comments [Add]    

Recommend this journal