4. 55%
HIV Prevalence by Age and Gender among South African youth
50% age 15-24 years
45%
40% Males
Females
35%
31.2%
95% Confidence Interval
28.9%
30%
26.3%
25.0%
25%
20%
14.4%
15% 13.8%
12.2%
11.0%
9.4%
10% 7.9%
4.1% 6.0% 5.8%
4.0%
5% 2.3%
3.6%
2.1% 2.6%
4.1%
2.0%
0%
Age Age Age Age Age Age Age Age Age Age
15 16 17 18 19 20 21 22 23 24
Pettifor A, et al. AIDS 2005, 19: 1525-1534.
5. High risk behaviors not observed
• We found that young South African women do not report
engaging in “high risk” sexual activity, despite the incredibly
high incidence and prevalence of HIV among young women.
• The median age of first sex in our national sample of 15-24 year
olds was 17 for young women and 16 for young men
• 7.8% of young women reported having sex before the age of 15
• Condom use at last sex was reported by 48% of young women and
57% of young men.
• Young women reported a mean of 2.3 lifetime partners and young
men 4.9 partners.
• Among sexually active young women, the mean and median age of
first pregnancy was 18 years.
• Only 2% of young women reported ever having engaged in
transactional sex.
6. Determinants of Infectiousness and
Susceptibility in Young Women
• Behavioral/Contextual/Structural
– Gender power inequities
• Less able to negotiate condom use (Unprotected sex)
• Less able to refuse unwanted sex (Forced sex)
– Older male partners
– Sugar Daddies? Transactional sex?
– Education/Poverty
• Biological
– Anatomy
– Higher Transmission Probability from Men to Women
– Use of Hormonal Contraception?
– STIs
– Pregnancy?
7. Underlying Proximate Biological Health Demographic
determinants determinants determinants outcomes outcome
New sex partner
“c”
Context Coital frequency
Exposure of
Socio-economic Concurrency
susceptible to
Sociocultural including Abstinence
infected person
gender Sexual mixing
Demographic
Education
Intervention Condom use
programs Other STIs “β” HIV
Circumcision Efficiency of Disease Mortality
HIV testing &counseling infection
STI control Type of sex transmission
Education for Biological per contact
knowledge & changing susceptibility
attitudes
Treatment w/ARVs
“D”
Treatment of OIs
Duration of
infectivity
Proximate-determinants framework for factors affecting the risk of sexual
transmission of HIV (Boerma & Weir JID 2005)
8. Education in young women in
South Africa
• Among young women with one lifetime
partner, those who had not completed high
school were almost 4 times more likely to
be HIV infected compared to those that
had completed HS (AOR 3.75 95% CI
1.34–10.46) (Pettifor A et al. IJE 2008)
9. Education and HIV:
protection or risk?
• Data from early in the epidemic suggested
that more education was associated with
increased risk of HIV infection
• Two recent reviews on HIV and education
indicate a protective association between
higher education and HIV infection,
particularly as epidemics mature
(Hargreaves et al. AIDS 2008, Jukes et al.
AIDS 2008)
10. Promising data from longitudinal
studies
• In Zambia, young women with more education were less
likely to be HIV infected than those with less education,
and declines in infection rates from 1995-2003 were
greatest in young women with the most education
(Michelo C et al. AIDS 2006)
• In Uganda, HIV infection rates declined most rapidly in
young women with a secondary school education (de
Walque D et al. TMIH 2005)
• In South Africa, participants were 7% less likely to
become infected with HIV for each year of education
they had completed (Barnighausen T, et al AIDS 2007)
11. A Social Vaccine?
HIV prevalence by education category, Rural Uganda,
1990-2001. Individuals aged 18-29
De Walque and J Whitworth, MRC Uganda (2002)
12. Percentage of young women and men age 15-24 who
reported being in school (primary or secondary) in South
Africa 2003
13.
14. Percentage of young women and men 15-24 completing
grade 12 (High School (HS)) in South Africa 2003
15. Association between school attendance and HIV awareness,
sexual behavior and HIV infection among 3682 females aged 15-19
from South Africa, 2003
Currently Dropped out Adjusted for
Unadjusted
Variable Attending or Never age, province
Model
School Attended and urban/rural
N % N % POR (95% CI) aPOR (95% CI)
Total N 2800 882
Knows ways to prevent
2672 95.5 814 92.3 0.97 (0.93, 1.01) 0.97 (0.94, 1.00)
HIV infection
Lack of Parental
1448 52.0 503 56.5 1.09 (0.92, 1.28) 1.02 (0.84, 1.23)
Communication
Not Had an HIV test 2484 89.2 650 74.4 0.83 (0.75, 0.93) 0.85 (0.74, 0.96)
Ever had sex 1075 36.5 666 76.3 2.09 (1.83, 2.38) 1.47 (1.32, 1.62)
Unwanted sex at
77 7.1 37 6.2 0.87 (0.46, 1.66) 0.93 (0.50, 1.73)
debut*
More than 1 partner
161 17.0 94 11.6 0.68 (0.42, 1.12) 0.83 (0.55, 1.26)
during previous year*
Early Debut* 159 13.2 115 16.3 1.24 (0.69, 2.24) 2.40 (1.66, 3.48)
Ever pregnant 207 17.6 351 54.1 3.07 (1.90, 4.96) 2.41 (1.70, 3.42)
Ever pregnant before
150 11.3 221 33.0 2.93 (1.28, 6.23) 3.19 (2.06, 4.96)
age 18
Partner Age
Difference*
105 8.6 121 15.8 1.84 (1.28, 2.66) 1.68 (1.14, 2.49)
(partner >3 years vs. <2
years older)
No condom use at least
428 34.7 400 58.2 1.67 (1.22, 2.31) 1.38 (1.10, 1.72)
sex*
HIV-positive 180 5.8 121 11.4 1.97 (1.38, 2.80) 1.24 (0.83, 1.86)
16. Association between school attendance and HIV awareness,
sexual behavior and HIV infection among 3555 males aged 15-19
from South Africa, 2003
Currently Dropped out or Adjusted for age,
Unadjusted
Variable Attending Never province and
Model
School Attended urban/rural
N % N % POR (95% CI) aPOR (95% CI)
Total N 2836 719
Knows ways to prevent 2684 92.4 645 82.6 0.89 (0.80, 1.00) 0.87 (0.80, 0.96)
HIV infection
Lack of Parental 1831 62.5 438 64.1 1.03 (0.81, 1.30) 1.16 (0.92, 1.46)
Communication
2569 91.1 605 90.5 0.99 (0.94, 1.05) 0.99 (0.96, 1.03)
Not Had an HIV test
Ever had sex 1257 46.9 503 59.0 1.26 (0.90, 1.75) 1.01 (0.88, 1.15)
Unwanted sex at debut* 7 0.3 4 0.9 2.84 (0.57, 14.2) 2.06 (0.66, 6.48)
More than 1 partner
439 42.1 181 45.5 1.08 (0.87, 1.34) 1.09 (0.88, 1.34)
during previous year*
Early Debut* 292 29.3 88 18.8 0.64 (0.49, 0.84) 1.16 (0.71, 1.89)
Partner Age
Difference*
4 0.5 7 0.6 1.36 (0.23, 8.02) 1.02 (0.06, 16.24)
(partner >3 years vs. <2
years older)
No condom use at least 473 43.7 201 42.2 0.97 (0.72, 1.30) 1.20 (0.94, 1.54)
sex*
HIV-positive 77 2.5 27 2.4 0.97 (0.52, 1.78) 0.96 (0.51, 1.80)
17. Proportion of young South African women ages 15-19 who
dropped out of high school by current sex partner age
difference (n=1750)
18. Proportion of young South African women ages 20-24 who
reported completing high school by the age difference of
their most recent sex partner (n=1848)
19. Reduced risk of sexual behaviors
associated with school attendance
• Study of ~2000 14-25 year olds in rural South Africa
• Young female students less likely
– To have 2 or more partners
– To have a partner > 3 yrs older
– To have had unprotected sex in the last year
• Young male students less likely
– To have three or more partners
– To have HIV
Hargreaves et al. J Epi Comm Health 2008.
20. Impact of education extends beyond
HIV to almost every health and
development outcome
• Better educated women are more likely than
their less educated peers to delay coital debut,
use condoms more often, delay marriage and
childbearing, have fewer children and healthier
babies, and enjoy better earning potential.
• In 17 countries in Africa and 4 in Latin America,
better-educated girls were found to delay age of
first sex and were more likely to use condoms.
World Bank. Education and HIV/AIDS a window of hope. Washington D.C.: World Bank; 2002.
21. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Exposure to HIV
messages
22. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Better
understanding of
messages
23. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Increased self-
esteem/self
efficacy/decision
making power
24. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Different social/sexual
networks
25. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Structured time in
school
26. Education and HIV risk:
causal mechanism
BLACK BOX
Reduced risk of
Education
HIV infection
Socio-economic status
27.
28. Why target girls?
• HIV incidence highest in young women
• Barriers to school attendance and drop out
appear greater for girls than boys
• Observed effects on education and HIV
are greater for girls than boys
• Programs that have reduced barriers to
education have had greater effects for girls
than boys
29. Barriers to Education
• For many poor families the costs associated with school
make it an economic impossibility.
• School fees/costs of education represent an important
barrier to school attendance in South Africa and are
often cited as reasons for not attending an educational
institution and for not furthering education.
• In South Africa, 65% of young people who were not in
school indicated that they did not have enough money to
continue their education.
• The second most common reason for school drop-out for
girls is pregnancy; 30% of girls who were not in school
indicated that they left school due to pregnancy.
Samson 2004
30. Barriers to education (2)
• Young women are often taken out of
school to find employment to support the
family or to care for sick family members.
• Family commitments was cited as the
reason for not attending an educational
institution by 9% of non-school attending
South African females, as opposed to <1%
of non-attending males
31. Cash Transfers to keep young women
in school
• Children in South African
households that receive
government social welfare
grants are more likely to attend
school and the observed
effects are greater for young
women than young men
(Samson 2004)
– the greatest benefit of social
welfare grants on educational
outcomes appears to be for
young women from the
poorest households
32. Cash Transfers increase school
attendance
• In Mexico, the Progresa program, which provides
conditional cash transfers to poor families to send
their children to school, has found that the program
increases school enrollment, particularly for girls
(Schultz T IFPRI 2000)
• Evaluations also find positive impacts on girls school
enrollment and/or attendance at primary and/or
secondary levels in Nicaragua, Ecuador, Colombia,
Brazil, Jamaica, Bangladesh, Cambodia, Turkey,
and Pakistan (Adato and Bassett 2008)
33. Do reducing barriers to education
reduce HIV risk?
• One RCT conducted in Kenya where
school uniforms were provided free of
charge, a reduction in self-reported
pregnancy was documented (Duflo E et al.
2006)
34. Study Design
CM Yes CM No
CCT Yes CCT+ CM CCT Only
CCT No CM only No
intervention
35. Study Site
• Agincourt area,
Mpumalanga
• DHSS site – population
~84,000
• ~2,500 HH with girl aged
14-16 years
• Wealth quintile
associated with school
attendance
• 21 secondary schools
• High HIV prevalence
36. Study Objectives
1. Impact of CCT on girls’ HIV risk
compared to control group
2. Impact of community mobilization on
general population aged 18-35 years
compared to control communities
3. Comparison of HIV risk among young
women receiving CCT in CM villages
with girls receiving CCT alone
37. Study Assessments
COMMUNITY (18-34 years)
•ACASI assessments at
baseline & 36 months
(n=1200)
•Ethnography - mobilization
HOUSEHOLD
ACASI assessments at
baseline & 36 months
GIRLS
•ACASI assessments at baseline,
12, 24, 36 months
•Ethnographic Research quarterly
•School Attendance monitoring
38. Study Measures
ACASI
•Unprotected sex in past 3 months
•Pregnancy
GIRLS
•Age difference with partner
•School attendance
•Sexual debut
•HIV
•HSV2
39. Study Measures
Ethnography
•How did CT work?
•Did parents ensure school
attendance?
•How did it change life?
GIRLS •Did it reduce barriers to
education?
•Impact on unprotected sex?
•Did it reduce HIV risk and how?
• What did HH do with money?
41. Study Measures
COMMUNITY (18-34 years)
•Gender norms
•Social Mobilization
•Collective Efficacy
HOUSEHOLD •Unprotected Sex
•Concurrency
•Intimate Partner
Violence
GIRLS
42. Formative Research
•Confirm poverty as
main reason for drop
out
IDIs
•Food
FGDs •Pregnancy a major
factor (10/15)
•Decision made by girl
•Parents think
education important
BUT don’t act on this
belief
•Educators report lack
of parental
involvement
•Limited feedback on
intervention
43. Formative Research
• Next phase
– Better information on planned intervention
using scenarios in FGDs
– School survey to better understand
implementation of programme in schools
