Linking Surveillance and M

1
Linking Surveillance and MEKaren A.
StaneckiUNAIDS

• SI/ME Field Officer Orientation
• August 2004
• Atlanta, GA

2
Overview

• New data from national surveys in countries with
  generalised epidemics
• ME Coverage estimates used in prevalence
  estimation
• Behavioral data

3
New dataNational population-based surveys
4
Limitations of ANC-based Estimates

• Prevalence equivalence of ANC and general
  population may not apply equally well in all
  countries, e.g. Kenya.
• But other new studies support the assumption,
  including Zambia DHS Kisesa, Tanzania
• Under-representation of smaller rural sites in
  surveillance systems (except South-Africa PPS
  sampling).
• But many countries have recently expanded the
  sentinel surveillance system to include more
  rural sites
• Adjustments for rural under-representation may
  not apply equally well in all countries

5
National population-based surveys

• Direct estimate of national prevalence
• Recently increased implementation of surveys
  (since 2000) Mali DHS, Dominican Republic DHS,
  Zambia DHS, Kenya DHS, Ghana DHS, Burkina Faso
  DHS, South-Africa HSRC survey, South-Africa
  loveLife survey (15-24 only), Burundi, Niger,
  Zimbabwe YAS (15-29 yrs only)
• Many more to come Cameroon, Tanzania, Uganda,
  Swaziland, Haiti, etc

6
Comparison of HIV Prevalence in Zambia DHS
CatchmentClusters to ANC Results
7
ANC versus general population

• ANC prevalence underestimates community
  prevalence among women
• ANC prevalence overestimates community prevalence
  among men

8
Individual Non-Response Rate ()
South-Africa additional household non-response
of 29 Non-response rates not available for
Burundi, Niger
9
Coverage Rates for HIV Testing by Urban-Rural
Residence and gender Zambia, Kenya, South-Africa
10
Using (adjusted) survey-based estimates to
calibrate ANC-based estimate
11
Improving ANC HIV Surveillance

• Collect a minimum set of socio-demographic data
  at ANC clinics
• Age, parity, schooling, length of time in the
  area
• Use Satellite rural areas to increase coverage
• Collect residence information
• Description of the population served by ANC
  clinic

12
Analysis of population surveys to inform
ANC-based Estimates

• Compare ANC and population survey estimates
• ANC vs. both sexes combined for nearby clusters
• Urban ANC with urban survey clusters
• Rural ANC with nearby rural clusters
• Regional prevalence ranking
• Compare women in surveys who delivered in the
  last 2 years or who are pregnant with ANC women

13
ME Coverage Estimates

• Anti-retroviral Therapy
• Number on ART
• Impact on mortality estimates
• Number in need
• PMCT
• Reduction MTCT
• Impact on Children infected with HIV and orphan
  estimates

14
Behavioral Data
15
Integrated analysis

• Behavioral data still not adequately collected
• Who is becoming infected and how
• Dont be fooled by numerical thresholds
• Concentrated vs generalized (Ukraine, Thailand,
  Cambodia)
• Important shifts in patterns of transmission do
  take place
• Thailand

16
Distribution of new HIV infections by type of
exposure in selected countries, 1998-2002
Source Pisani E. et al. Back to basics in HIV
prevention focus on exposure, BMJ 2003
3261384-7 (Fig 1)
17
Estimated number of new HIV infections in
Thailand by year and changing mode of transmission
160
140
Spouse 50 IDU 20 SW 15 MTCT 15
SW 90 Spouse 5 IDU 5
120
New HIV infections (number of people, in
thousands)
100
80
60
40
20
0
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Year




Spouse heterosexual transmission of HIV in
cohabiting partnerships SW HIV transmission
through sex work IDU HIV transmission through
injecting drug use MTCT mother to child
transmission of HIV

Source Thai Working Group on HIV/AIDS
Projections, 2001
2004 Report on the Global AIDS Epidemic (Fig 4)