ࡱ > [ ^ X Y Z [ 1 bjbj 44 ΐ ΐ h h l = < = y ! ! ! ! n" " " J= L= L= L= L= L= L= @ %C v L= 9 " n" n" " " L= ! ! = $ $ $ " ! ! 7 ` $ " J= $ $ V + 8 - ! .: "$ >- 7 = 0 = J- T C $ C - - p C . " " $ " " " " " L= L= $ " " " = " " " " C " " " " " " " " " h : @ Supplementary Material
1.0 Sensitivity Analyses Substituting the VH weights for the Giles SS weights did not change the significance of any of the variables in our multivariate models for the full population and the HIV-positive, unaware subsample (data not shown). The odds ratios changed by less than 5% for each (data not shown). In the full sample, the complete-case analysis ignored observations in varying numbers due to missing data at each level of the continuum of care. Additionally, excluding non-productive seeds from our analysis did not change our final results (data not shown).
2.0 Survey Instrument A multi-disciplinary team of medical, public health and social science researchers with expertise in HIV research and prevention, social networks, survey research, sexuality, and community violence in minority communities worked for over a year developing the survey design and instrument. The questionnaire covered demographic and background information, family life, Black and MSM community involvement, sex and sex-drug risk/reduction practices, exposure to violence and the criminal justice system, STI/HIV testing and treatment, social media use, and numerous self-administered scales including depression, knowledge and attitudes toward safe sex and risk. During this period, pilot interviews and cognitive testing of questions was conducted by the Survey Research Lab at the University of Chicago.
3.0 RDS Sampling, Weights and Implementation RDS is a widely used approach that provides a design for sampling using seeds and subsequent recruits, and a methodology for estimating statistical properties of the target population ADDIN EN.CITE Heckathorn DD199715[1]151517Heckathorn DD,Respondent-driven sampling: a new approach to the study of hidden populations.SOCIAL PROBLEMS-NEW YORKSOCIAL PROBLEMS-NEW YORK1997[1]. The strengths of RDS have prompted its wide use in public health studies, especially when populations are stigmatized or not accessible through traditional rosters ADDIN PAPERS2_CITATIONS EA216E86-F69D-46DE-8D9E-36659966567A099199700001200000000200000Respondent-driven sampling: a new approach to the study of hidden populations9CDA2F6C-311A-4707-839A-3FD09529EE19400400Heckathorn:1997vwhttp://socpro.oxfordjournals.org/content/socpro/44/2/174.full.pdfSOCIAL PROBLEMS-NEW YORK--100-100D1C9B7B5-E56F-4996-9F2F-239A665D405DDDHeckathorn99201400001200000000200000Evaluating Outcome-Correlated Recruitment and Geographic Recruitment Bias in a Respondent-Driven Sample of People Who Inject Drugs in Tijuana, Mexico55271D39-2C3A-4A3E-8514-527453431D6A400400Rudolph:2014uchttp://link.springer.com/article/10.1007/s10461-014-0838-4Springer USAIDS and Behavior-100-100CC5A9B55-E9A7-4D59-B189-0447E30805EBAERudolphTLGainesRLozadaAVera1BDB31F5-4ECC-4FE1-9AF3-3641DD25423B1810.1007/s10461-013-0640-8236699201412001200000000220000http://link.springer.com/10.1007/s10461-013-0640-8Jenness:2014by400Spatial recruitment bias in respondent-driven sampling: Implications for HIV prevalence estimation in urban heterosexuals.124002373Springer USAIDS and Behavior-100-100CC5A9B55-E9A7-4D59-B189-0447E30805EBSamuelMJennessAlanNeaigusTravisWendelCamilaGelpi-AcostaHollyHaganC1605219-6696-4715-B7EA-24F9EA1C4C958910.1136/sextrans-2012-05055715699201303001200000000220000http://sti.bmj.com/cgi/doi/10.1136/sextrans-2012-050557Hakre:2013iu400Prevalence of HIV and other sexually transmitted infections and factors associated with syphilis among female sex workers in Panama.Department of Epidemiology and Threat Assessment, United States Military HIV Research Program, Henry M. Jackson Foundation for the Advancement of Military Medicine, Bethesda, Maryland 20817, USA. shakre@hivresearch.org2400164Sexually transmitted infections-100-100CAE519D2-1821-4B12-BBCB-6A6E1950E6B6ShilpaHakreGriseldaArteagaAurelioENezChristianTBautistaAimeeBolenMariaVillarroelSheilaAPeelGabrielaPaz-BaileyPaulTScottJuanMPascalePanama HIV EPI Group378AA10F-6DF3-4D35-8F32-980659E27EDF2310.1097/EDE.0b013e31823ac17c13899201201001200000000220000http://content.wkhealth.com/linkback/openurl?sid=WKPTLP:landingpage&an=00001648-201201000-00021McCreesh:2012ck400Evaluation of respondent-driven sampling.Department of Infectious Disease Epidemiology, Faculty of Epidemiology & Population Health, London School of Hygiene and Tropical Medicine, UK.1400147Epidemiology (Cambridge, Mass.)-100-1005834E32E-1D8D-4E81-B7AF-9EFC2ADF347CNickyMcCreeshSimonD WFrostJanetSeeleyJosephKatongoleMatildaNTarshRichardNdunguseFatimaJichiNatashaLLunelDermotMaherLisaGJohnstonPamSonnenbergAndrewJCopasRichardJHayesRichardGWhiteF91F55DC-5C1D-41F1-B536-877271AD022A99920120810120000000022200010.1186/1477-7517-9-373799201200001200000000200000http://www.harmreductionjournal.com/content/9/1/37Morineau:2012ju400HIV prevalence and risk behaviours among injecting drug users in six indonesian cities implications for future HIV prevention programs.992011060812000000002220001FHI Asia Pacific Regional Office, 19th floor, Tower 3, Sindhorn Building; 130-132, Wireless Road, Lumpini, Phatumwan, Bangkok, 10330, Thailand. gmorineau@fhi360.org.400BioMed Central LtdHarm reduction journal-100-100A15013A6-6C88-4C0A-97F7-792ECAE8C2BAGuyMorineauLiesbethJmBollenRizkyIkaSyafitriNurjannahNurjannahDyahErtiMustikawatiRobertMagnaniD73E7498-FE47-464C-9F8E-18014F10623C129920080523120000000022200010.1007/s10461-008-9421-1S10599200807001200000000220000http://link.springer.com/10.1007/s10461-008-9421-1Malekinejad:2008gm400Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in international settings: a systematic review.992007121812000000002220004 SupplSchool of Public Health, University of California, Berkeley, Berkeley, CA, USA. mmalekinejad@psg.ucsf.edu40030Springer USAIDS and Behavior-100-100CC5A9B55-E9A7-4D59-B189-0447E30805EBMohsenMalekinejadLisaGrazinaJohnstonCarlKendallLigiaRegina Franco SansigoloKerrMarinaRavenRifkinGeorgeWRutherford19 Suppl 299200505001200000000220000Family Health International, Arlington, VA, USA.S67Review of sampling hard-to-reach and hidden populations for HIV surveillance.6E2B02B6-3E74-422B-8B95-CAA9AD80589F40072400Magnani:2005vzhttp://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&id=15930843&retmode=ref&cmd=prlinksAIDS-100-10099A44DD8-E69C-446A-82E9-12EF7BAF6F25RobertMagnaniKeithSabinTobiSaidelDouglasHeckathorn02C68DE3-9235-406A-AA4D-9860C5BEB8364010.1111/j.1467-9531.2010.01223.xrespondent-driven sampling: an assessment28599201008001200000000220000http://smx.sagepub.com/lookup/doi/10.1111/j.1467-9531.2010.01223.xGile:2010bq400Respondent-Driven Sampling: An Assessment of Current Methodology.Nuffield College, University of Oxford.1400327Sociological methodology-100-10003CBC71B-46A1-4DBD-B3D6-588BDF447203KristaJGileMarkSHandcock ADDIN EN.CITE ADDIN EN.CITE.DATA [1-9]. Inference from RDS data requires strong assumptions to treat the sampling scheme as probabilistic, and may be biased due to a number of underlying assumptions, which are discussed in detail elsewhere ADDIN EN.CITE ADDIN EN.CITE.DATA [10, 11]. Methodological research to improve estimation of parameters and standard errors from RDS studies has received much recent attention ADDIN EN.CITE ADDIN EN.CITE.DATA [12-16].
Recent recommendations for estimation of population parameters from RDS data include an estimator that treats the RDS recruitment process as a successive sampling (SS) process when the size of the population is known or can be reliably estimated ADDIN EN.CITE Gile KJ201432[17, 18]323217Gile KJ, Johnston LG, Salganik MJ.Diagnostics for respondent-driven sampling.Journal of the Royal Statistical Society: Series A (Statistics in Society)Journal of the Royal Statistical Society: Series A (Statistics in Society)241-26917812014KJ.201131313117Gile KJ.Improved inference for respondent-driven sampling data with application to HIV prevalence estimation.Journal of the American Statistical AssociationJournal of the American Statistical Association2011[17, 18]. Previously, the size of young Black MSM in the South Side of Chicago and adjacent suburbs was estimated to be 5500 ADDIN EN.CITE ADDIN EN.CITE.DATA [19]. To assess the sensitivity of our results to this estimator, we compared our results with weights generated by another commonly used estimator ADDIN EN.CITE Volz E200830[16]303017Volz E, Heckathorn DD.Probability based estimation theory for respondent driven sampling.Journal of official statistics.Journal of official statistics.2008[16].
Since our data were generated using RDS, we used Giles Sequential Sampling estimator to weight our sample to enable probabilistic inference from our data ADDIN EN.CITE Gile201023[2]232317Gile KJ, Handcock MSNuffield College, University of Oxford.Respondent-Driven Sampling: An Assessment of Current MethodologySociol MethodolSociological methodologySociol MethodolSociological methodologySociol MethodolSociological methodology285-3274012010Aug0081-1750 (Print)
0081-1750 (Linking)22969167http://www.ncbi.nlm.nih.gov/pubmed/22969167343733610.1111/j.1467-9531.2010.01223.x[2]. The Sequential Sampling estimator is recommended when the population size is reliably known. Sensitivity to this weighting scheme was assessed by using an alternate estimator proposed by Volz and Heckathorn (VH). ADDIN EN.CITE Volz E200830[16]303017Volz E, Heckathorn DD.Probability based estimation theory for respondent driven sampling.Journal of official statistics.Journal of official statistics.2008[16] All weights were computed using the RDS package ADDIN EN.CITE Handcock MS201454[20]545413Handcock MS, Gile KJ, Fellows IE, Neely WWRespondent-Driven Sampling2014http://cran.r-project.org/web/packages/RDS/index.html.[20].
The implementation of respondent driven sampling began by purposively selecting a diverse set of YBMSM to serve as initial respondents or seeds. The research team gathered a group of twenty community partners to help with identifying potential seeds. A meeting was convened with the community partners in May 2013 to explain the goals and importance of the study and to ask attendees to invite up to three eligible candidates who were social in the community and likely to bring others into the study. Supplementary Table 1 summarizes the main types of methods and venues from which our team members and partners were recruited, including sources of seeds. Our goal was to recruit a diverse group of seeds who would be interested and willing to participate in the study and who would be able to recruit others like themselves. Ten seeds were recruited via connections to the House/Ball Community. Most of the rest of the seeds were recruited via avenues that were less specific, e.g., via web sites, Facebook postings, community events, college campuses etc. for a total of 62 seeds.
(Supplementary Table 1About Here)
Each participant was invited and given instructions about recruiting other eligible MSM they knew using six coupons with unique ID numbers. The distribution of successful referrals were 0 (6.4%), 1 (44.3%), 2 (24.4%), 3 (15.1%), 4 (5.9%), 5 (2.5%) and 6 (1.4%). This was repeated for each new participant whether they were a seed or a sprout. Each respondent, seed and recruit, were offered $60 for their participation in the interview and were told they would receive an additional $20 for each recruit who participated. Sensitivity analyses that excluded referrals of 4, 5 and 6 did not change our main findings. In addition, in our primary analysis we included all seeds because there is no current consensus as to whether seeds should be included or excluded from analyses (1, 2). The ratio of seeds to recruits of approximately 10% is consistent with other RDS samples of YBMSM (3). Our main findings were robust and did not change when analyses included all seeds, productive seeds, or no seeds. Levels of CJI were pretty stable across RDS waves with a range of only 5% (46%-51%) and by wave 10 was stable within 1% across waves.
4.0 Additional covariates measured Criminal justice involvement, defined as having ever previously been detained, arrested, or spent time in jail or prison, represented the main independent variable. In addition, we asked How much total time, over the course of your life, have you spent in a jail, prison, juvenile detention center or correctional facility? and On how many separate occasions have you been detained, arrested, or spent time in jail or prison? to assess for duration and frequency of CJI. In addition we included, based on community input, factors especially salient for YBMSM in low-resourced communities that move away from traditional White/Black disparities analytic frameworks. We included the following variables: membership of a House/Ball or gay family, community venues frequented, community violence, connectedness to the Black community, having a mother figure and openness about sexuality, all as covariates. Additional covariates measured include: education, employment, housing instability, insurance status, alcohol use, other substance use, and depression. Education was defined as less than high school education, completed high school education, some college education, Associates degree, Bachelors degree, and Masters degree. Employment status was defined as currently working full-time (> 30 hours/week), part-time (< 30 hours/week), or unemployed. Housing instability was defined as self-reported homeless status in the past twelve months. Insurance status was defined as currently having health insurance coverage or currently having no coverage. Alcohol use was defined as never, one to two per year, one to two per month, or one to two per week/everyday over the course of the participants lifetime. Finally, depression was analyzed using the Brief Symptom Inventory 18-question survey (BSI-18), a self-report measure of psychological symptoms over the previous week. Raw scores for the BSI-18 were converted to T-scores and defined as present if the T-score was greater than 62.
5.0 Details of the uConnect cohort can be found in SupplementaryTable 2.
Supplementary Table 1. Sources of seedsProductive Seeds, n(%)All Seeds, n(%)Source(n=37)(n=62)Federally Qualified Health Center Clinic10 (27.0)17 (27.4)Community Based Organization focusing on HIV Prevention6 (16.2)12 (19.4)Direct Community Contact - unaffiliated 6 (16.2)7 (11.3)House/Ball Weekly Meeting Group4 (10.8)5 (8.1)Black MSM Community Based Organization3 (8.1)4 (6.5)Facebook3 (8.1)4 (6.5)Website2 (5.4)5 (8.1)Ball Community Member unaffiliated1 (2.7)5 (8.1)HIV Support Group1 (2.7)1 (1.6)College0 (0.0)1 (1.6)Unknown1 (2.7)1 (1.6)
Supplementary Table 2. Characteristics of a population-based sample of younger black MSM by HIV serostatus andcriminal justice involvement (CJI), uConnect 2013-2014.
Sample ProfileTotalaCJIUnweighted N (Weighted %)N=622N=286SociodemographicsAge at interview (years) 16-1851 (10.4)16 (6.3) 19-20117 (18.6)44 (15.0) 21-24280 (42.1)135 (43.2) 25+174 (29.0)91 (35.6)Education Less than high school76 (14.3)32 (12.0) High school graduate or equivalent165 (26.2)84 (27.2) Some college or higher380 (59.5)169 (60.7)Sexual identification Gay414 (64.3)171 (57.5) Bisexual167 (28.8)93 (35.5) Straight22 (4.6)12 (3.7) Something else18 (2.4)9 (3.3)Current employment Full Time152 (24.0)61 (21.8) Part Time172 (23.8)76 (24.9) Not employed298 (52.2)149 (53.4)Housing instability in past year157 (23.8)85 (28.9)Behavior & HealthVenues to meet other men Ballroom165 (22.4)82 (21.4) Clubs/Bars377 (52.4)187 (60.5) Gym69 (11.7)29 (10.4) Public spaces286 (41.1)130 (41.2)Gender of sexpartnersa Exclusively men523 (82.5)231 (78.1) Both men and women62 (10.5)40 (17.0) Exclusively women18 (4.1)8 (2.2) No sex partners19 (2.9)7 (2.8)Drug useb124 (16.3)77 (26.9)Health care coverage339 (50.0)152 (53.1)Symptoms of psychological distressc107 (14.8)57 (16.6)Use of ARVs as TasP66 (9.4)37 (12.7)Social FactorsCriminal justice involvement286 (40.8)--Have a mother/mother figure562 (90.2)256 (89.4)Open about sexual identity551 (85.3)247 (79.5)Member of ballroom house or gay family207 (27.5)104 (28.6)Feels close to African-American community538 (87.1)249 (89.4)Ever witness to violent assault(s)443 (68.9)232 (80.5)Years since HIV diagnosisd214 (2.15)115 (2.19)HIV CareHIV seropositive214 (31.6)115 (42.8)HIV seropositive aware164 (23.6)89 (32.9)Linked to caree125 (19.1)72 (29.5)Retained in caref86 (13.1)53 (23.0)Adherence to ARVsg69 (10.6)38 (17.5)Viral suppressionh52 (8.1)29 (14.0)ARVs, antiretrovirals; MSM, men who have sex with men; TasP, treatment as prevention.
aSelf-report of all anal, oral and vaginal sex partners in past six months.
bSelf-report of recreational drug use in past year including cocaine, ecstacy, heroin, methamphetamines, molly, psychadelics, prescription painkillers and volatile nitrates (poppers); marijuana and alcohol use not included in measure.
cAssessed via the Brief Symptom I n v e n t o r y ( B S I ) - 1 8 w i t h T - s c o r e s e"6 2 s i g n a l i n g p s y c h o l o g i c a l d i s t r e s s o r d i s o r d e r ( D e r o g a t i s , 2 0 0 0 ) .
d M e a n n u m b e r o f y e a r s b e t w e e n H I V d i a g n o s i s a n d s t u d y p a r t i c i p a t i o n .
e H a v i n g a t l e a s t o n e H I V m e d i c a l c a r e v i s i t w i t h i n 6 m o n t h s o f d i a g n o s i s .
f H a v i n g t w o or more HIV medical care visits within 90 days in the previous 12 months.
gMissing HIV medications on fewer than 4 days in the previous month (i.e. at least 85% adherence).
hHaving a viral load of <2000 copies of human immunodeficiencyvirus ribonucleic acid (HIV-RNA) per milliliter of whole blood.
Supplement References
ADDIN EN.REFLIST 1. Heckathorn DD. Respondent-driven sampling: a new approach to the study of hidden populations. SOCIAL PROBLEMS-NEW YORK 1997.
2. Gile KJ HM. Respondent-Driven Sampling: An Assessment of Current Methodology. Sociol Methodol 2010,40:285-327.
3. Hakre S, Arteaga G, Nunez AE, Bautista CT, Bolen A, Villarroel M, et al. Prevalence of HIV and other sexually transmitted infections and factors associated with syphilis among female sex workers in Panama. Sex Transm Infect 2013,89:156-164.
4. Jenness SM, Neaigus A, Wendel T, Gelpi-Acosta C, Hagan H. Spatial recruitment bias in respondent-driven sampling: Implications for HIV prevalence estimation in urban heterosexuals. AIDS Behav 2014,18:2366-2373.
5. Magnani R, Sabin K, Saidel T, Heckathorn D. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS 2005,19 Suppl 2:S67-72.
6. Malekinejad M, Johnston LG, Kendall C, Kerr LR, Rifkin MR, Rutherford GW. Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in international settings: a systematic review. AIDS Behav 2008,12:S105-130.
7. McCreesh N, Frost SD, Seeley J, Katongole J, Tarsh MN, Ndunguse R, et al. Evaluation of respondent-driven sampling. Epidemiology 2012,23:138-147.
8. Morineau G, Bollen LJ, Syafitri RI, Nurjannah N, Mustikawati DE, Magnani R. HIV prevalence and risk behaviours among injecting drug users in six indonesian cities implications for future HIV prevention programs. Harm Reduct J 2012,9:37.
9. Rudolph AE, Gaines TL, Lozada R, Vera A, Brouwer KC. Evaluating outcome-correlated recruitment and geographic recruitment bias in a respondent-driven sample of people who inject drugs in Tijuana, Mexico. AIDS Behav 2014,18:2325-2337.
10. Khanna AS, Michaels S, Skaathun B, Morgan E, Green K, Young L, et al. Preexposure Prophylaxis Awareness and Use in a Population-Based Sample of Young Black Men Who Have Sex With Men. JAMA Intern Med 2015:1-3.
11. Khanna AS MS, Skaathun B, Morgan E, Green K, Young LE, Schneider JA. Low preexposure prophylaxis (PREP) awareness & uptake in a population-based sample of young Black MSM in Chicago. . In: International AIDS Society. Vancouver, Canada; 2015.
12. DD H. Respondent-driven sampling II: deriving valid population estimates from chain-referral samples of hidden populations. Social problems 2002.
13. Goel S, Salganik MJ. Respondent-driven sampling as Markov chain Monte Carlo. Stat Med 2009,28:2202-2229.
14. Goel S, Salganik MJ. Assessing respondent-driven sampling. Proc Natl Acad Sci U S A 2010,107:6743-6747.
15. Salganik MJ HD. Sampling and Estimation in Hidden Populations Using Respodent-Driven Sampling. Sociological Methodology 2004,34:193-239.
16. Volz E HD. Probability based estimation theory for respondent driven sampling. Journal of official statistics. 2008.
17. Gile KJ JL, Salganik MJ. Diagnostics for respondent-driven sampling. Journal of the Royal Statistical Society: Series A (Statistics in Society) 2014,178:241-269.
18. KJ. G. Improved inference for respondent-driven sampling data with application to HIV prevalence estimation. Journal of the American Statistical Association 2011.
19. Livak B, Michaels S, Green K, Nelson C, Westbrook M, Simpson Y, et al. Es 2 V
o
~
} ~ M N Q R A A A ǹyl[Q[B[l[l[Q h;h OJ QJ ^J mH nH uh;h OJ QJ ^J !j h;h h0@ OJ QJ U^J h;h hED? OJ QJ ^J -h;h hED? OJ PJ QJ ^J eh r h;h h7 OJ QJ ^J h;h hED? 5OJ QJ ^J h;h h,{ 5OJ QJ ^J h;h h7 5OJ QJ ^J hA h7 ^J h1% h7 ^J h1% h7 5^J h1% h7 5OJ QJ ^J h1% hO 5OJ QJ ^J W
C N A[ _ _ d d v d `gd7 $d `a$gd7 d `gd;h d `gd,{ d gdED? d gdED? d gd7 >! d $d %d &d