Proxy measures for HIV incidence trends

1
Proxy measures for HIV incidence trends

• Dr. Ly Penh Sun
• Cambodia National Center for
• HIV/AIDS, Dermatology, and STDs

The 2nd Global HIV/AIDS Surveillance
Meeting Bangkok, Thailand 2-5 March 2009
2
Introduction

• Methods to assess trends in HIV incidence
  include
• Longitudinal cohort studies
• Statistical modeling
• Testing for detection of recent HIV infections
• Using HIV prevalence among the youngest age group
  as a proxy for HIV incidence
• i.e., prevalent cases in the youngest age group
  probably represent recently acquired HIV
  infection assuming recently initiated risk
  behavior
• Because Cambodia collects data about duration of
  brothel employment from brothel-based female sex
  workers, it was suggested that duration might
  also be used as a proxy indicator for HIV
  incidence

3
Objective

• To compare results of stratifying HIV prevalence
  data by age and by duration of brothel employment
  with trends estimated from detection of recent
  infections (using the BED assay)

4
Methods

• HIV sentinel surveys (HSS) have been conducted in
  Cambodia since 1996
• Brothel-based female sex workers (BFSWs) have
  been included as a sentinel group in all 8 HSS
  rounds
• The BED assay has been used to estimate HIV
  incidence in 4 of the last 5 surveys
• In 2003, serum specimens archived from the 1999,
  2000, and 2002 surveys were tested in batch
• As part of the 2006 HSS, dried blood spot
  specimens were collected for quality control
  retesting and HIV incidence testing

5
Methods
Sample size by survey year
Cambodia has a total of 24 provinces and
municipalities
6
HIV prevalence among female sex workers in
Cambodia, by survey year
7
HIV prevalence among female sex workers in
Cambodia, by age group and year
Age in Years
Adjusted for results of quality control
8
HIV prevalence among female sex workers in
Cambodia, by duration of sex work and survey year
Duration of sex work
Adjusted for results of quality control
9
HIV prevalence among female sex workers in
Cambodia, by duration of sex work, age, and
survey year
Adjusted for results of quality control
10
HIV incidence among FSWs by survey year
Estimated based on detection of recent infection
by BED assay. Corrections (Hargrove and McDougal
) not required because of the very low proportion
of HIV-positive specimens that tested "recent ."
11
ResultsComparison of HIV incidence with
age-stratified prevalence
Incidence based on detection of recent
infections using the BED assay
12
ResultsComparison of HIV incidence with duration
of sex work
Uncorrected incidence
13
ResultsComparison of HIV incidence with age and
duration of sex work
Uncorrected incidence
14
Recommendations for Data Use

• Duration of sex work may not be a good proxy for
  HIV incidence if
• this group includes women who have been sexually
  active prior to working in brothels and
• information on time sex work was initiated is not
  collected (short duration may not be the same as
  newly initiated)
• In Cambodia, age may be a better proxy for
  initiation of sexual activity and HIV incidence
  than duration of sex work

15
Discussion

• Brothel-based FSWs are at high risk for HIV
  because of the large number of clients
  (concurrent sex partnerships)
• Cannot assume
• their only risk is from selling sex
• they were not sexually active or not at risk
  before working in a brothel
• those "new to sex work" or those who worked for a
  brief time were not already infected with HIV
• Duration as measured by HSS is associated with
  ageFSWs who have worked in brothels for more
  than 2 years are older on average than those who
  have worked a short time, but not all FSWs who
  have worked for lt1 year are young
• Although most of the recent infections occurred
  in those working lt1 year, prevalent infections
  were also common
• 10.1 of those who worked 1-6 months
• 9.2 among those who worked 7-11 months
• 20.5 of those who worked 2 years
• In comparison, only 2.8 of those aged 16-19 had
  prevalent infections
• In some countries, age data may be more reliable
  than data on duration of employment

16
Thank you for your kind attention
17
Results HIV incidence among FSWs in 2006
With 95 confidence interval
Hargrove and McDougal adjustments not valid
when percent of recent infections 5.7 of all
HIV positive
18
ResultsComparison of HIV incidence with
age-stratified prevalence
Uncorrected incidence