Background
Despite the national
and international efforts to combat HIV, s
ub
-
S
aharan Africa
continues to
suffer the effects of the pandemic. In 2011
in sub
-
Saharan Africa
there were 23.5 milli
on people
living with the virus
and 1.8 million new infections. Swaziland, with a population of
1.1 million
people, has equally shouldered the HIV burden as it has an adult prevalence of 26%, which is
one of the highest in the world. State resources have been affected and the number of orphaned
and vulnerable children continues to rise. Ethiopia on t
he other hand, has a population of 74
million p
eople and is one of the poorest countries
in
the world. However,
the adult
HIV
prevalence remains low
at 1.5% and is lower
than many
s
ub
-
Saharan African countries.
HIV
prevalence
in pregnant
women attending an
tenatal clinic has
declined by
82
% between 2000 and
2010.
The main mode of transmission in both countries is through heterosexual contacts.
Objective
T
he main objective of the study was to identify and compare the main differences in the
determinants of HIV infection in Swaziland (2006) and Ethiopia (2005)
Materials and methods
The
research wa
s a
secondary analysis of a
cross
-
sectional survey
from two
demographic health
survey
s
and studied the effect of the determinants on HIV infection. We constructed a conceptual
framework based on the Mosley and Chen framework, and we grouped the variables into
underlying and proximate factors. The outcome variable w
as HIV status and descriptive statistics
as well as bivariate tests were carried out using Pearson’s Chi
-
square and Fischer’s exact tests for
associations. Univariable and multivariable logistic regression were used to model the explanatory
ii
variables again
st the outcome. Datasets from the two countries were
then
appended and propensity
score matching was done using age, gender and wealth index. Further logistic regression analysis
for the merged data was done for the
proximate
factors and the outcome. The c
onceptual
framework based model was used to model direct and indirect determinants of HIV infection using
a
structural equation model.
Results
In Swaziland, a total of 4
584 females and 3
602 males were tested for HIV, and the adult
prevalence was
26.20%. Females had a higher prevalence of 3
0
.
98
% compared to their male
counterparts who were at 19.5
1
% (p<0.00
0
1). In the multivariable regression analysis the result
show
ed
a greater than 3 times increased likelihood of contracting HIV for those who had
more
than 3 sexual partners in their lifetime, [AOR=3.1
4
, 95% CI (2.55
-
3.8
7
)]. Those with
genital
ulcers were above 2 times more likely to have HIV compared to those who did not, [AOR=2.1
5
,
95% CI (1.7
1
-
2.7
1
)]. Circumcision yielded almost 38% HIV protecti
on in circumcised men
compared to uncircumcised men, [AOR=0.62, 95% CI (0.45
-
0.85)].
In Ethiopia, a total of 5
942 females and 4
630 males aged 15
-
49 years were tested for HIV and
the adult prevalence was 1.97%. Female prevalence was 2.39% compared to 1.
43% for males
(p<0.00
0
1). In the multivariable regression analysis participants with high number of lifetime
partners had almost 6 fold chance of being HIV infected compared to those who had one,
[AOR=5.61, 95% CI (2.7
3
-
11.
5
2)].
Individuals
who had first
sex at age 15
-
19 years were greater
than
4
times more likely to have HIV compared to those who had never had sex, [AOR=
4.20
,
95% CI (
1.90
-
9.32
)].
Circumcision ha
d
HIV benefits as it yielded a 74% protection for the
circumcised, [AOR=0.2
6, 95% CI (0.14
-
0.4
8
)]. The inverse association between Moslems
iii
religion and HIV was also observed in this study as they had a 6
7
% HIV protection compared to
Orthodox, [AOR=0.3
3
, 95% CI (0.21
-
0.5
4
)]. Living in Gambela region was associated with
almost 4 times increased risk o
f HIV infection compared to Tigray, [AOR=4.1
1
, 95% CI (2.0
7
-
8.
17
)].
Analysis using
logistic regression and structural equation modelling adjusting for the country
effect showed a protective effect for those staying in Ethiopia compared to those staying in
Swaziland. The structural equation modelling confirmed findings from the earlier regression
analysis as those with
age at first sex below 15 years were 6.03 times more likely to get HIV,
AOR=
[AOR=6.03, 95% CI (5.40
-
10.07)]
.
Those with three or more sexual
partners were
5.15
times more likely to get
HIV
infection
compared to those with one partner
[AOR=5.15, 95% CI
(3.77
-
7.03)].
Conclusion
and recommendation
This
study was able to demonstrate the
association between HIV acquisition and the
determinants. Most importantly is the ability to clearly point out on the proximate factors in both
countries which happen to be behavioural. The structural equation modelling added value as it
was able to show
the indirect factors and how they impact on HIV infection. This can be
addressed through strengthening of behaviour change programs to reduce risky sexual
behaviours. Programs aimed at delaying sexual debut and reduction of sexual partners should be
streng
thened
in both countries
. Male circumcision programs should be scaled up through social
mobilisation and community involvement
more especially in Swaziland.