Select 2012-2013 survey findings – CBO meeting

Report
Crane Survey
Select Results from Recent Surveys
Kampala, 2012/3
Men Who Have Sex With Men (MSM)
Female Sex workers (FSW)
Drug users
Kampala, December 2013
Crane Survey
• Standing survey platform, Wandegeya
• PEPFAR funded
• Collaborative activity
– Makerere University / School of Public Health
– Ministry of Health
– CDC
Key population HIV surveillance:
Three pillars
• Surveys
– How affected are they?
• Population size estimates
– How many are they?
• Mapping
– Where are they?
Bio-behavioral surveys
How affected are they?
Bio-behavioral survey design
• Sampling: Respondent-driven sampling
– Participants refer others to survey
– Facilitates population estimates: Survey findings
apply to all key population members, not just
those who participated
– Anonymous (no names)
Bio-behavioral survey design (2)
• Data collection: audio-computer-assisted self
interviews, facilitates more candid answers
• Blood draw
– HIV antibodies
– CD4 cell count (disease stage)
– Viral load: Number of viruses (“copies”) in blood
– HIV results returned, HIV-positives referred
2012/3 Crane surveys
• Female sex workers (FSW): 15+ years, sold sex to
men in last 6 months
• Men who have sex with men (MSM): 18+ years, had
anal sex with another man in last 6 months
• Drug users (DUS): 18+ years, habitual consumer of
drugs in last 6 months (by nose, mouth, or injection)
• All: residing in greater Kampala
Number of survey participants
• FSW:
1,490
• MSM:
595
• DUS:
1,170
HIV prevalence estimates
(preliminary)
• FSW*:
33%
(Kampala general women: 9.5%)
• MSM:
13%
(Kampala general men: 4.1%)
• DUS:
17%
(Kampala general pop: 7.1%)
• *) Note: Median duration of sex work at time of survey
participation: 2 years
Select MSM characteristics
•
•
•
•
Syphilis prevalence:
Ever tested for HIV:
HIV prevalence:
Self-reported on ART:
3%
70%
13%
14% (among HIV-pos)
CD4 cell count
Viral load
• Average: 406
• Average: 37,000 copies / ml
• 50% between: 300–520
• 50% between: 8,000–120,000
• CD4<350: ~27%*
• Suppressed (<1000): ~11%*
*) Small numbers – unstable estimates
Select FSW characteristics
• Syphilis prevalence:
• HIV prevalence:
• Self-reported on ART
6%
33%
18% (among HIV-pos)
CD4 cell count
Viral load
• Average: 443 cells / ul
• Average: 3,500 copies / ml
• 50% between 300–600
• 50% between: 100–38,000
• CD4<350: 32% (of these,
86% not on ART)
• Suppressed (<1000): 37%
High rates of gender-based violence
(GBV) among Kampala FSW
• 49% ever experienced rape
– Rape defined as the use of force or threat to use force
– Of those who have ever been raped, on average report
having been raped twice in last 6 months (50% range: 1-3)
• 82% exposed to some form of GBV in last 6 months
– Defined as: physical abuse, verbal abuse, forced sex acts,
paid less or not paid at all after sex act
– 52% experienced three or more GBV events in last 6 months
Conclusion - FSW
• Very high HIV prevalence among FSW: 33%
– 3x that of Kampala women in general
– One in three HIV-pos FSW with CD4 <350
– Short average duration of sex work at time of
survey participation suggests:
• high turn-over
• high HIV incidence
• warrants frequent outreach services
Conclusion – MSM, drug users
• MSM: HIV prevalence 13%
– triple that of Kampala men in general
– low self-reported ART uptake
– high average viral load (high forward transmission
potential)
– small proportion with suppressed viral load
• Drug users: HIV prevalence 17%
– more than double that of general population in
Kampala
Population size estimates
How many are there?
Objective
• To estimate the population sizes of
– Female sex workers (FSW)
– Men who have sex with men (MSM)
in Kampala, Uganda
Method to estimate population size:
Capture/recapture
Universe:
(e.g., all sex
workers in
Kampala)
N
Sex workers
captured twice
(“Match”)
Sex workers
captured, e.g.,
with key chains
(Sample 1)
C1
N = C 1* C 2 / M
M
Sex workers
recaptured,
e.g., by survey
(Sample 2)
C2
Capture 1 – tagging key pop
members with a small gift
• Ordered several hundred tailor-made key chains
– Not available commercially
• Distributed by community members
– 540 key chains among FSW
– 400 key chains among MSM
Capture 2 - survey
• Respondent-driven sampling survey (RDS)
• Start: after key chain distribution
• Asked every participant if s/he can
– show key chain or
– identify key chain (received but doesn’t have)
• Had to choose correct key chain out of 6 similar key
chains
• “Match” defined as survey participant who
show or identified correct key chain
Population size estimates – Match:
presenting or identifying correct key chain
• FSW:
13,200 (95% CI: 10,200 – 16,200)
• MSM:
7,900 (95% CI: 4,400 – 11,300)
• Kampala 15-59 population: ~800,000 (49% male)* +
– Proportion FSW:
3.3% (2.5% - 4.0%)
– Proportion MSM: 2.0% (1.1% - 2.9%)
*): Source: UBOS: 1.73M in 2012, assumed proportion <15 yrs: 49%, assumed sex distribution: 50%.
http://www.ubos.org/onlinefiles/uploads/ubos/pdf%20documents/abstracts/2012%20Statistical%20Ab
stract.pdf
+) : Note: Age bands not identical to MSM survey eligibility age criteria
Conclusion – FSW and MSM size
estimates for Kampala
• Conservative estimate:
– MSM:
8,000 (95% CI: 4,000 - 11,000)
– FSW:
13,000 (95% CI: 10,000 - 16,000)
– Note: FSW highly mobile, both
• Occupationally (in and out migration of sex work)
• Geographically (change in sex work location)
Linking size estimation and
mapping
Where are they?
Crane survey – virtual, electronic
mapping of key populations
• Respondents:
Survey participants (MSM, FSW)
• Interview location:
Survey office
• Software:
Google Earth (offline)
• Mapping results:
Geo-social locations, i.e., where
- FSW respondents meet clients
- MSM socialize
• Note: mapping of (transient) individual geo-social
locations, not primarily venues / buildings
Geo-social
locations
of sex
workers Makindye
“About 200
sex workers
hang out at
G8 club,…,
especially on
the
weekends.”
(Venue
names
changed by
investigators)
Using GIS software to identify
and characterize clusters of
sex workers
Color: population size (darker ~
larger)
Size: geographic extent of
cluster
Projecting Kampala’s FSW population size estimate
(13,000) to the relative sizes of mapped clusters
Thank you
George Lubwama, Enos Sande (SPH / Makerere University)
Rachel Kwezi, Sam Sendagala, Michael Ediau (CDC Uganda)
Steven Borders (GIS Corps)
Wolfgang Hladik, Drew Baughman, Jim Tobias (CDC Atlanta)
Using GIS software to characterize clusters of sex workers
Color: population size (darker ~ larger)
Size: geographic size of cluster
“Mapping and Size Estimation of the Key
Affected Populations in Kampala Capital
City Authority”, 2013
• Method: Census (?)
• FSW: 2,148 (95% CI: 1,611-2,685)
• MSM:
253 (95% CI: 169-338)
• Note: Crane MSM Survey 2012/3, sample size: 595
Commissioned work by D Muhangi and J Kiwanuka for: Alliance of Mayors and Municipal
leaders on HIV/AIDS in Africa
Modes of Transmission study
• UNAIDS commissioned report, 2009
• Country-wide estimates
– FSW (last 12 months): 32,652
– MSM: 3,976
Future size estimation activity:
Network scale-up
• Population size estimation through general
population based survey
– Nation-wide
– Sample size ~5,000 adults
• Size estimates for MSM, MSW, FSW, PWID, ‘mobile
men with money’
• Strata: Kampala, urban, rural
• Start early 2014
Past population size estimates lack quality data
sources: Modes of Transmission study (2009)
National size estimates
• FSW : Sex for money in last 12 months:
– Number:
32,652 (projected % value from
abroad)
– Proportion:
0.3%
• MSM: (No operational definition)
– Number:
4,000 (“MSM study expert guess x 2”)
– Proportion:
<0.1%
Assumptions
1. Closed population
2. Everyone has equal, non-zero sampling probability
3. Capture and recapture constitute independent
samples (no correlation)
4. Matches are accurate
In practice, hard to meet all assumptions
Summary
Capture 1
Capture 2
Matches, total
Matches, present**
Matches, identify***
FSW
540
1567
64
MSM
380 (400)*
372
18
30
34
6
12
*Approximately 20 UMOs distributed “late”, i.e., during
RDS sampling period (excluded from analysis)
** presented UMO
*** did not present, but did identify correct UMO
Population size estimates - Match:
presenting correct key chain
• FSW:
26,000 (95% CI: 18,000 – 35,000)
• MSM:
28,000 (95% CI: 4,000 – 53,000)
*): Source: UBOS: 1.73M in 2012, assumed proportion <15 yrs: 49%, assumed sex distribution: 50%.
http://www.ubos.org/onlinefiles/uploads/ubos/pdf%20documents/abstracts/2012%20Statistical%20Abstract.pdf

similar documents