HIV PREVALENCE AND ASSOCIATED FACTORS 13

Lawrence Marum, James N. Muttunga, Francis M. Munene, and Boaz K. Cheluget

This chapter presents information on the coverage of HIV testing among those eligible, the preva-
lence of HIV in the respondents, and the factors associated with HIV infection in the population. As de-
scribed in Chapter 1, the 2003 KDHS is the fourth survey in the international DHS program to include
HIV testing, and the first to anonymously link the HIV results with key behavioural, social and demo-
graphic factors. The HIV prevalence data provide important information to plan the national response, to
evaluate programme impact, and to measure progress on the National HIV/AIDS Strategic Plan 2000-
2005. The understanding of the distribution of HIV within the population and the analysis of social, bio-
logical and behavioural factors associated with HIV infection offer new insights about the HIV epidemic
in Kenya that may lead to more precisely targeted messages and interventions.

In Kenya, as in most of sub-Saharan Africa, national HIV prevalence estimates have been derived
primarily from sentinel surveillance in pregnant women. Currently, the national sentinel surveillance sys-
tem consists of 42 sites in government and mission health facilities selected to represent the different
groups, regions, and rural and urban populations in the country. For three months each year since 1990,
pregnant women registering their first visit to these antenatal clinics (ANCs) and patients with sexually
transmitted diseases in the sentinel sites have been anonymously tested for HIV and the results entered,
analysed and reported by the National AIDS/STD Control Programme (NASCOP) (Ministry of Health,
2001). The latest round of sentinel surveillance was conducted between May and August 2003, during the
same time period as the KDHS.

While the rate of HIV infection in pregnant women has been shown to be a reasonable proxy for
the level in the combined male and female adult population in a number of settings (WHO and UNAIDS,
2000), there are several well recognized limitations in estimating the HIV rate in the general adult popula-
tion from data derived exclusively from pregnant women attending selected antenatal clinics. First, the
ANC data do not capture any information on HIV prevalence in non-pregnant women, nor in women who
either do not attend a clinic for pregnancy care or receive antenatal care at facilities not represented in the
surveillance system.1 Pregnant women also are more at risk for HIV infection than women who may be
avoiding both HIV and pregnancy through the use of condoms or women who are less sexually active and
are therefore less likely to become pregnant or expose themselves to HIV. In addition, there may be bi-
ases in the ANC surveillance data because HIV infection reduces fertility and because knowledge of HIV
status may influence fertility choices. Finally, the rates among pregnant women are not a good proxy for
male HIV rates. For example, a WHO study of four cities in sub-Saharan Africa, including Kisumu in
Kenya, demonstrated higher risk overall in women compared to men (Buve et al., 2001).

Thus, although the information from the ANC surveillance system has been very useful for moni-
toring trends in HIV levels in Kenya, the inclusion of HIV testing in the KDHS offers the opportunity to
better understand the magnitude and patterns in the infection level in the general reproductive age popula-
tion in Kenya. The KDHS results are in turn expected to improve the calibration of annual sentinel sur-
veillance data, so that trends in HIV infection can be more accurately measured in the intervals between
general population surveys.

1
Nearly 90 percent of pregnant women in Kenya receive antenatal care; however, 21 percent attend dispensaries,
which are not covered in the ANC surveillance system (Chapter 9).

HIV Prevalence and Associated Factors | 217

 13.1 COVERAGE OF HIV TESTING

Table 13.1 presents the coverage rates for HIV testing by the reason for not being tested accord-
ing to gender and residence. HIV tests were conducted for 76 percent of the 4,303 eligible women and 70
percent of the 4,183 eligible men. For both sexes combined, coverage was 73 percent, with rural residents
more likely to be tested than their urban counterparts (79 percent and 62 percent, respectively). There also
were strong differences in HIV testing coverage rates by province. Among both sexes, Nyanza Province,
which as discussed later in the chapter has the highest HIV rate among Kenya’s provinces, had the highest
rate of testing (89 percent), followed by Western (85 percent) and Rift Valley Province (78 percent). Cen-
tral Province (67 percent) and Nairobi (52 percent) had the lowest testing rates. In every province, women
were more likely to be tested than men.

Table 13.1 Coverage of HIV testing by sex and urban-rural residence

Percent distribution of women and men eligible for testing by testing status, according to sex and urban-rural residence, Kenya 2003 (unweighted)
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Residence Province
––––––––––––– –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Rift North
Testing status Urban Rural Nairobi Central Coast Eastern Nyanza Valley Western Eastern Total
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
WOMEN
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Tested 66.2 81.7 54.5 70.7 80.1 76.1 91.1 81.1 88.0 75.6 76.3
Refused 19.2 11.9 21.5 18.8 14.5 15.1 5.4 11.7 8.7 19.9 14.4
Absent for testing 10.6 3.5 19.5 5.1 3.3 4.6 1.9 3.1 2.4 4.5 6.0
Interviewed in survey 5.9 1.7 11.8 3.0 0.8 2.2 1.4 1.3 1.0 0.0 3.1
Not interviewed 4.7 1.8 7.7 2.2 2.5 2.4 0.6 1.9 1.4 4.5 2.8
Other/missing 4.0 2.9 4.5 5.3 2.0 4.2 1.6 4.1 1.0 0.0 3.3

Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Number 1,488 2,815 651 738 488 502 514 702 507 201 4,303
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
MEN
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Tested 58.4 76.7 50.2 62.9 66.2 74.5 87.1 75.7 82.7 74.9 70.3
Refused 16.5 11.2 15.3 16.2 21.4 14.0 3.3 10.0 9.9 13.8 13.0
Absent for testing 20.3 7.9 30.8 13.0 8.8 6.1 6.6 9.4 4.9 11.3 12.2
Interviewed in survey 5.8 2.6 9.8 5.4 3.8 2.8 0.8 1.5 1.1 0.5 3.7
Not interviewed 14.5 5.3 21.1 7.6 5.0 3.4 5.7 8.0 3.9 10.8 8.5
Other/missing 4.8 4.2 3.6 7.8 3.6 5.3 3.1 4.9 2.6 0.0 4.4

Total percent 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Number 1,466 2,717 665 739 444 506 488 679 467 195 4,183
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
TOTAL
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Tested 62.3 79.3 52.4 66.8 73.5 75.3 89.1 78.4 85.4 75.3 73.4
Refused 17.8 11.5 18.4 17.5 17.8 14.6 4.4 10.9 9.2 16.9 13.7
Absent for testing 15.4 5.7 25.2 9.1 5.9 5.4 4.2 6.2 3.6 7.8 9.1
Interviewed in survey 5.9 2.1 10.8 4.2 2.3 2.5 1.1 1.4 1.0 0.3 3.4
Not interviewed 9.6 3.5 14.4 4.9 3.6 2.9 3.1 4.9 2.6 7.6 5.6
Other/missing 4.4 3.5 4.0 6.6 2.8 4.8 2.3 4.5 1.7 0.0 3.8

Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Number 2,954 5,532 1,316 1,477 932 1,008 1,002 1,381 974 396 8,486

Based on the reason for nonresponse, individuals who were not tested are divided into four cate-
gories in Table 13.1:

• those who refused testing when asked for informed consent by the health worker (14 percent
overall)

218 | HIV Prevalence and Associated Factors

 • those who were interviewed in the survey, but who were not at home when the health worker
arrived for testing and were not found on callbacks (3 percent)

• those who were not at home for the testing and were never interviewed (6 percent), and

• those who were missing test results for some other reason, such as they were incapable of giv-
ing consent for testing, there was a mismatch between the questionnaire and the blood sam-
ple, or there was a technical problem in taking blood (4 percent).

Refusal is the most important reason for non-response on the HIV testing component among both
women and men. Among men, absence accounts for almost as much non-response as refusal, while it is
less important among women. The fact that some respondents were interviewed but not subsequently con-
tacted by the health worker is partly due to having only one health worker per team, which contributed to
delays in the time of arrival of the health worker after the interview.

The proportions falling into the four non-response categories vary markedly by urban-rural resi-
dence. Refusal rates are higher among urban women and men (19 percent for women and 17 percent for
men) than among their rural counterparts (12 percent for women and 11 percent for men). Absence was
the main reason for non-response among urban men. Fifteen percent of urban men were not interviewed
nor tested, compared with 5 percent of urban women, while in rural areas, only 5 percent of men and 2
percent of women were neither interviewed nor tested. Similarly, higher proportions of urban women and
men (6 percent among both sexes) were interviewed but were not at home when the health worker visited
to collect the blood sample, compared to rural residents (2 percent of women and 3 percent of men).

Looking at the provincial patterns, Nairobi had the highest rate of refusal in women (22 percent),
the highest proportions absent for the interview (8 percent of women and 21 percent of men), and the
highest level of those interviewed but absent for testing (12 percent of women and 10 percent of men).
Nyanza had the lowest rates of refusal among both women (5 percent) and men (3 percent). Nyanza also
had the lowest proportions of women absent for testing (under 2 percent), while men were least likely to
be absent in Western (5 percent) and Eastern (6 percent) provinces.

Table 13.2 shows coverage rates for HIV testing by age group, education and wealth. If HIV
status influenced participation in the testing, coverage would be expected to rise with age since HIV lev-
els increase sharply with age before leveling off or declining at the older ages. In fact, the coverage rate
for testing among women is consistent across all age groups (range 74 percent to 79 percent). Response
rates are somewhat more variable by age among men (range between 64 percent and 76 percent), but
again they do not rise with age as would be expected if they were influenced by HIV status.

Those with an incomplete primary education are the most likely to have been tested, while men
and women with at least some secondary education were least likely to be tested. Similarly, those in the
highest quintile of the wealth index were the least likely to be tested and had the highest levels of refusal
(18 percent of women and 13 percent of men), absent after interview (6 percent for both men and
women), and absent and not interviewed (5 percent of women and 15 percent of men).

In order to further explore whether nonresponse might have an impact on the HIV seroprevalence
results, an analysis also was undertaken of the relationships between participation in the HIV testing and a
number of other characteristics related to HIV risk. The descriptive tables which were examined in that
analysis are included in Appendix A.

HIV Prevalence and Associated Factors | 219

 Table 13.2 Coverage of HIV testing by age, education, and wealth quintile

Percent distribution of women and men eligible for HIV testing by testing status, by age, education, and wealth quintile, Kenya 2003 (un-
weighted)
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Testing status
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Tested Refused Absent Other/missing
––––––––––––– ––––––––––––– ––––––––––––– –––––––––––––
Not Not Not Not
Age, education, and Inter- inter- Inter- inter- Inter- inter- Inter- inter-
wealth quintile viewed viewed viewed viewed viewed viewed viewed viewed Total Number
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
WOMEN
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Age
15-19 75.1 0.2 13.4 1.2 3.8 3.4 1.6 1.2 100.0 975
20-24 77.2 0.5 11.7 1.7 2.4 2.3 2.4 1.9 100.0 886
25-29 76.1 0.3 11.2 1.7 3.4 3.3 2.4 1.6 100.0 704
30-34 76.1 0.5 13.3 0.8 4.0 1.3 2.1 1.9 100.0 618
35-39 78.9 0.0 11.8 1.8 1.8 3.5 1.3 0.9 100.0 451
40-44 74.2 0.0 15.7 1.5 2.8 3.5 1.0 1.3 100.0 395
45-49 73.7 0.4 16.8 1.8 3.3 2.9 0.4 0.7 100.0 274

Education
No education 74.5 1.0 15.1 1.6 1.6 2.6 1.0 2.6 100.0 689
Primary incomplete 81.4 0.2 10.8 1.2 2.1 1.4 1.7 1.1 100.0 1,259
Primary complete 76.7 0.2 13.7 1.3 3.7 1.7 1.5 1.1 100.0 993
Secondary+ 72.0 0.0 13.4 1.5 4.4 5.0 2.5 1.3 100.0 1,352

Wealth quintile
Lowest 84.3 0.6 10.0 0.2 0.9 2.1 0.9 1.1 100.0 661
Second 86.6 0.4 8.7 0.4 0.4 1.0 0.7 1.6 100.0 677
Middle 81.7 0.5 10.2 0.7 1.9 1.6 1.6 1.6 100.0 732
Fourth 77.9 0.1 12.4 1.6 3.2 1.6 2.1 1.2 100.0 822
Highest 63.2 0.0 18.1 2.9 6.1 5.4 2.7 1.6 100.0 1,411

Total 76.1 0.3 12.9 1.5 3.1 2.8 1.8 1.5 100.0 4,303
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
MEN
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Age
15-19 75.5 0.4 8.2 1.8 3.8 6.0 1.8 2.4 100.0 928
20-24 67.4 0.1 12.0 3.2 3.7 8.0 2.1 3.5 100.0 791
25-29 64.2 0.6 11.5 3.0 3.3 11.9 1.7 3.8 100.0 637
30-34 68.8 1.4 8.8 2.9 3.5 11.7 1.0 1.9 100.0 513
35-39 68.3 0.7 10.9 2.7 5.8 7.8 1.6 2.4 100.0 451
40-44 69.9 0.8 13.8 1.4 2.2 8.0 0.8 3.0 100.0 362
45-49 68.9 0.4 12.0 2.5 3.3 10.4 1.2 1.2 100.0 241
50-54 74.6 0.4 8.5 2.7 4.2 4.6 1.2 3.8 100.0 260

Education
No education 69.3 0.8 10.7 2.5 2.0 7.3 1.4 5.9 100.0 355
Primary incomplete 75.1 0.7 8.9 2.1 3.0 6.2 1.8 2.2 100.0 1,250
Primary complete 70.2 0.2 10.9 3.5 3.9 7.5 1.6 2.2 100.0 939
Secondary+ 65.9 0.6 11.6 2.0 4.5 10.9 1.5 2.9 100.0 1,627

Wealth quintile
Lowest 79.5 0.8 9.1 2.2 1.2 4.7 0.8 1.7 100.0 596
Second 79.3 1.1 7.1 1.0 1.4 5.6 1.3 3.2 100.0 624
Middle 74.1 0.3 9.0 3.0 3.1 6.4 1.3 2.8 100.0 703
Fourth 72.9 0.4 11.0 1.9 3.8 4.9 2.0 3.1 100.0 838
Highest 57.4 0.5 13.1 3.5 6.0 14.6 1.9 3.0 100.0 1,422

Total 69.7 0.6 10.5 2.5 3.7 8.5 1.6 2.8 100.0 4,183
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Total includes 10 women and 8 men not stated as to education.

220 | HIV Prevalence and Associated Factors

 The variation in response rates with these measures is again reassuring as coverage rates are fre-
quently but not uniformly higher among those groups considered to be at higher risk for HIV (Tables A.3-
A.6). For example, response rates are slightly higher among those who have ever had sex than among
those who have not. Similarly, rates are higher among those in polygynous unions and lowest among
those who are not currently in union. Among women, response rates are highest among those who are
widowed, while among men they are highest among those who are divorced or separated. Coverage is
higher among those who reported having had higher risk sex in the 12 months preceding the survey than
among those who had sex but not higher risk sex and those who did not have sex at all in the prior 12
months. Among women, response rates are higher for the small number who report having multiple part-
ners; however, coverage is lower among men with multiple partners.

Among men, the coverage rate for HIV testing is higher among uncircumcised than circumcised
men. Since HIV prevalence is higher among uncircumcised men (see next section), the higher response
rates in the former group again are reassuring. However, men who sleep away from home—a characteris-
tic assumed to be related to higher HIV risk—have lower rates of testing: 75 percent in those who sleep
away more than 5 times a month and 77 percent for those who stay away for more than one month at a
time compared to 81 percent or more among who were never absent or absent less frequently.

Finally, in addition to the examination of the descriptive tables, a multivariate analysis of the de-
terminants of non-response was conducted (ORC Macro, 2004). The results of that analysis confirm that
eligible respondents who were not tested for HIV did not differ in meaningful ways from those tested. In
fact, adjusted prevalence based on the regression analysis would lower prevalence among both women
and men by a fraction of one percent because those not tested have slightly lower behavioural and socio-
demographic risks for HIV.

In summary, the initial descriptive and multivariate examinations of the HIV testing coverage
levels provided no evidence of a consistent relationship between non-response rate and variables associ-
ated with higher HIV risk. Although further analysis is required, this analysis supports the conclusion that
the KDHS prevalence rates are a reasonable measure of the actual levels of HIV prevalence in the popula-
tion.

13.2 HIV PREVALENCE

HIV Prevalence by Socioeconomic Characteristics

Results from the 2003 KDHS indicate that 7 percent of Kenyan adults are infected with HIV.
(Table 13.3). HIV prevalence in women age 15-49 is nearly 9 percent, while for men 15-54, it is under 5
percent. This female-to-male ratio of 1.9 to 1 is higher than that found in most population-based studies in
Africa and implies that young women are particularly vulnerable to HIV infection compared with young
men. Figure 13.1 shows, for example, that 3 percent of women age 15-19 are HIV infected, compared
with less than half of one percent of men 15-19, while HIV prevalence among women 20-24 is over three
times that of men in the same age group (9 percent and 2 percent, respectively). The peak prevalence
among women is at age 25-29 (13 percent), while prevalence rises gradually with age among men to peak
at age 40-44 (9 percent). Only in the 45-49 year age group is HIV prevalence among men (5 percent)
higher than that for women (4 percent).

Since few HIV infected children survive into their teenage years, infected youth represent more
recent cases of HIV infection and serve as an important indicator for detecting trends in both prevalence
and incidence. Overall, prevalence among women age 15-24 in the KDHS is 6 percent, compared with
slightly over 1 percent among men, for an overall prevalence in youth of under 4 percent. These preva-
lence levels will provide a baseline for measuring progress toward the goals of the National HIV/AIDS
Strategic Plan in future surveys.

HIV Prevalence and Associated Factors | 221

 Table 13.3 HIV prevalence by age

Percentage HIV positive among women age 15-49 and men age 15-54 who were tested, by age, Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men Total
–––––––––––––––––– ––––––––––––––––– –––––––––––––––––
Percent Percent Percent
HIV HIV HIV
Age positive Number positive Number positive Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
15-19 3.0 711 0.4 745 1.6 1,456
20-24 9.0 658 2.4 566 6.0 1,224
25-29 12.9 522 7.3 428 10.4 950
30-34 11.7 438 6.6 368 9.4 806
35-39 11.8 345 8.4 321 10.1 666
40-44 9.5 276 8.8 260 9.1 535
45-49 3.9 202 5.2 163 4.4 364
50-54 na na 5.7 193 na na

Total age 15-49 8.7 3,151 4.6 2,851 6.7 6,001

Total age 15-54 na na 4.6 3,043 na na
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
na = Not applicable

Figure 13.1 HIV Prevalence by Age Group and Sex

Percent
14
# Women
12 #
#

10
#
# ,
8
,
, Men
,
6
,
4 #
#
2
,

0
,
. 15-19 20-24 25-29 30-34 35-39 40-44 45-49 .
# Women , Men

KDHS 2003

As Table 13.4 shows, urban residents have a significantly higher risk of HIV infection (10 per-
cent) than rural residents (6 percent). Prevalence in urban women is 12 percent compared with less than 8
percent for rural women, for a 1.6 urban-rural relative risk of HIV infection. For men, the risk associated
with urban residence is even greater; urban men are twice as likely to be infected as rural men (8 percent
and 4 percent, respectively). Since 80 percent of Kenya’s population is categorised as rural, however, the
greatest burden of HIV infection is in the rural population.

The HIV epidemic shows regional heterogeneity. Nyanza Province has an overall prevalence of
15 percent, followed by Nairobi with 10 percent. All other provinces have levels between 4 percent and 6
percent overall, except North Eastern where no respondent tested positive, indicating that the rate is very
low in this province. Gender differences persist in all the regions.

222 | HIV Prevalence and Associated Factors

Table 13.4 HIV prevalence by selected socioeconomic characteristics

Percentage HIV positive among women and men age 15-49 who were tested, by socioeconomic characteristics,
Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men Total
–––––––––––––––––– ––––––––––––––––– –––––––––––––––––
Percent Percent Percent
Socioeconomic HIV HIV HIV
characteristic positive Number positive Number positive Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Residence
Urban 12.3 779 7.5 716 10.0 1,495
Rural 7.5 2,372 3.6 2,135 5.6 4,507

Province
Nairobi 11.9 332 7.8 314 9.9 646
Central 7.6 462 2.0 438 4.9 900
Coast 6.6 236 4.8 197 5.8 433
Eastern 6.1 514 1.5 464 4.0 978
Nyanza 18.3 432 11.6 376 15.1 808
Rift Valley 6.9 747 3.6 691 5.3 1,438
Western 5.8 368 3.8 323 4.9 690
North Eastern 0.0 60 0.0 48 0.0 108

Education
No education 4.4 396 2.7 156 3.9 552
Primary incomplete 9.3 1,052 3.4 982 6.4 2,034
Primary complete 10.6 784 5.9 660 8.5 1,444
Secondary+ 8.2 918 5.2 1,053 6.6 1,972

Employment
Currently working 9.6 1,844 5.9 2,007 7.6 3,851
Not currently working 7.4 1,307 1.5 844 5.1 2,151

Wealth quintile
Lowest 3.9 505 3.4 431 3.6 937
Second 8.5 580 4.2 501 6.5 1,082
Middle 7.1 597 2.2 528 4.8 1,125
Fourth 9.7 663 4.3 624 7.1 1,287
Highest 12.2 806 7.3 765 9.8 1,571

Ethnicity
Embu (2.8) 37 (3.7) 37 3.3 73
Kalenjin 4.9 346 2.0 366 3.4 712
Kamba 8.6 392 1.6 334 5.4 726
Kikuyu 6.6 742 2.8 621 4.9 1,363
Kisii 7.4 171 0.5 163 4.0 334
Luhya 7.9 481 5.1 438 6.6 919
Luo 25.8 361 17.5 341 21.8 702
Maasai 2.8 76 2.2 56 2.5 132
Meru 6.1 172 1.2 165 3.7 337
Mijikenda/Swahili 3.8 137 3.0 116 3.5 254
Somali 0.9 100 1.8 77 1.3 177
Taita/Taveta 11.7 41 7.1 30 9.7 71
Turkana 6.5 39 5.1 45 5.7 84
Kuria * 19 (5.2) 21 2.7 40
Other 6.7 38 5.6 41 6.1 79

Religion
Roman Catholic 8.9 800 4.9 756 6.9 1,556
Protestant/Other Christian 9.2 2,087 4.5 1,729 7.0 3,816
Muslim 2.7 204 3.1 175 2.9 378
No religion 11.1 52 5.5 185 6.7 237

Total 8.7 3,151 4.6 2,851 6.7 6,001

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Figures in parentheses are based on 25-49 unweighted cases. An asterisk indicates that a figure is based on
fewer than 25 unweighted cases and has been suppressed.
na = Not applicable

HIV Prevalence and Associated Factors | 223

 Those who have completed primary school have higher infection levels than those with either less
or more education. Work status is related to the HIV rate among both women and men. Ten percent of
currently working women and 6 percent of currently working men are HIV infected, compared with 7
percent of women and 2 percent of men currently not working. Those in the highest quintile of the wealth
index have the highest rates of HIV infection.

HIV prevalence is substantially higher among the Luo ethnic group than other groups. More than
one in four Luo women and 18 percent of men are HIV positive. The only other group with higher than
average prevalence levels is Taita/Taveta. Women and men who identify themselves as Christian have
rates similar to the national average for each gender, while Muslim women and men both have rates of 3
percent.

HIV Prevalence by Other Sociodemographic Characteristics

As expected, marital status is related to HIV prevalence (Table 13.5). Women currently in a mari-
tal union have a prevalence of 8 percent, only slightly higher than the rate among men who are currently
in union (7 percent). Women who are widowed, divorced, or separated have significantly higher rates (30
percent and 21 percent, respectively) than married women (8 percent).

Table 13.5 HIV prevalence by selected sociodemographic characteristics

Percentage HIV positive among women and men age 15-49 who were tested, by marital status, pregnancy status for
women, and mobility status for men, Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men Total
–––––––––––––––––– ––––––––––––––––– –––––––––––––––––
Percent Percent Percent
Sociodemographic HIV HIV HIV
characteristic positive Number positive Number positive Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Marital status
Currently in union 8.0 1,897 7.0 1,353 7.6 3,250
Widowed 30.2 133 * 18 31.8 151
Divorced/separated 20.9 126 6.4 96 14.6 222
Never in union 5.6 995 1.6 1,384 3.2 2,378
Ever had sex 9.9 480 1.9 932 4.6 1,413
Never had sex 1.6 515 0.9 451 1.3 966

Type of union
In polygynous union 11.4 326 11.9 126 11.6 452
Not in polygynous union 7.2 1,571 6.5 1,227 6.9 2,798
Not currently in union 9.8 1,254 2.4 1,498 5.7 2,752

Currently pregnant
Pregnant 7.3 260 na na na na
Not pregnant/not sure 8.8 2,891 na na na na

Numbers of times slept away

None na na 3.2 1,421 na na
1-2 na na 4.2 655 na na
3-5 na na 5.1 386 na na
5+ na na 9.3 373 na na

Away for more than one month

Away for more than 1 month na na 3.4 470 na na
Away always less than 1 month na na 7.3 944 na na
Never away na na 3.2 1,421 na na

Total 8.7 3,151 4.6 2,851 6.7 6,001

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Total includes cases missing data on number of times slept away and whether away for more than one month.
An asterisk indicates that a figure is based on fewer than 25 unweighted cases and has been suppressed.
na = Not applicable

224 | HIV Prevalence and Associated Factors

 Women who report they ever had sex but have never been in a union have a higher risk than men
in the same category (10 percent and 2 percent, respectively); this and the higher proportion of men who
have never been in a union compared to women accounts for much of the overall female-male difference
in prevalence. Around one percent of individuals who report they have never been in a union and have
never had sex are also HIV-infected, suggesting either reporting errors in the sexual behaviour or non-
sexual transmission of HIV.

Considering the type of current union, women in a polygynous union have a higher prevalence
(11 percent) than those in non-polygynous unions (7 percent). Again the rates for men are similar (12 per-
cent in polyygynous unions and 7 percent for non-polygynous unions).

HIV prevalence among women who are currently pregnant is 7 percent, providing a useful bench-
mark to compare with rates in pregnant women tested during sentinel surveillance.

The survey results show that men who sleep away from home more frequently have higher HIV
prevalence, 9 percent for those who slept away from home five or more times in the previous 12 months
compared with 3 percent for those who did not sleep away from home. Those who are away from home
for short periods of time (always less than one month) have double the risk of HIV infection (7 percent)
than those who are never away and those who are away for more than one month at a time (3 percent for
both groups).

HIV Prevalence by Sexual Risk Behaviour

Table 13.6 examines the prevalence of HIV infection by sexual behaviour indicators among re-
spondents who have ever had sexual intercourse. In reviewing these results, it is important to remember
that responses about sexual risk behaviours may be subject to reporting bias. Also, sexual behaviour in
the 12 months preceding the survey may not adequately reflect lifetime sexual risk.

For women, there is a clear pattern of higher HIV prevalence with earlier sexual debut. This pat-
tern is not evident among men, who have a prevalence of 5 to 6 percent regardless of age at first sex.

Young women (15-24 years) whose first sex was with a man ten or more years older have a
higher prevalence of HIV (10 percent) compared with those whose first partner was less than ten years
older (8 percent). Women who said that a condom was used during their first sexual encounter have a
higher prevalence of HIV (11 percent) than those who did not use a condom (8 percent). Among men, no
significant difference in prevalence can be detected between those who used a condom at first sex and
those who did not.

Seventeen percent of women who had a higher-risk sexual partner (a non-marital, non-cohabiting
partner) are HIV-infected, compared with 8 percent of those who were sexually active but did not have a
higher risk partner. In contrast, men reporting a higher-risk partner in the last year have a lower HIV
prevalence, compared with sexually active men who did not have a higher-risk partner (5 percent and 7
percent, respectively). Among women reporting no sex in the last year, 11 percent are HIV-positive,
compared with 2 percent of men reporting no sex in the last 12 months.

Among women, having more than one partner and having more than one higher-risk partner in
the preceding 12 months are associated with higher HIV prevalence. Among men, however, these vari-
ables are not consistently related to HIV prevalence. Women who exchanged sex for money, gifts, or fa-
vours in the last 12 months have a slightly higher HIV infection level than those who have not (11 percent
and 10 percent, respectively). Among men, those who paid for sex prior to the 12 months preceding the
survey have higher HIV prevalence (8 percent) than either those who have never paid for sex (5 percent)
and, surprisingly, those who paid for sex in the preceding 12 months (4 percent).

HIV Prevalence and Associated Factors | 225

 Table 13.6 HIV prevalence by sexual behaviour characteristics

Percentage HIV positive among women and men age 15-49 who ever had sex and were tested, by sexual behaviour
characteristics, Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men Total
–––––––––––––––––– ––––––––––––––––– –––––––––––––––––
Percent Percent Percent
Sexual behaviour HIV HIV HIV
characteristic positive Number positive Number positive Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Age at first sex
<15 12.4 940 5.1 1,143 8.4 2,083
16-17 9.3 648 5.2 457 7.6 1,106
18-19 9.7 515 4.8 436 7.5 951
20+ 6.0 392 6.0 355 6.0 747

First sexual partner was:1

10 or more years older 10.4 66 na na na na
Other/Doesn’t know 7.7 681 na na na na

Condom use at first sex1

Used at first sexual sex 10.7 93 0.0 108 4.9 201
Did not use at first sex 7.5 654 1.0 674 4.2 1,328

Higher-risk sex in past 12 months

Had higher-risk sex 17.2 392 4.7 812 8.7 1,204
Had sex, not higher risk 8.3 1,833 6.7 1,213 7.7 3,046
No sex in past 12 months 11.1 411 1.9 374 6.7 785

Number of partners in past 12 months

1 9.6 2,166 5.4 1,700 7.7 3,866
2 20.4 53 9.7 262 11.5 315
3+ * 6 3.3 64 5.3 70

Number of higher-risk partners

in past 12 months
1 15.7 361 4.8 632 8.7 993
2+ 34.0 32 4.4 181 8.8 211

Received money/gifts/favours for sex

in past 12 months
Exchanged for sex 11.2 119 na na na na
No exchange 9.8 2,106 na na na na

Paid for sex

In past 12 months na na 4.3 86 na na
Prior to past 12 months na na 8.1 269 na na
Never na na 4.9 2,045 na na

Condom use at last paid sex

Used na na 8.0 173 na na
Did not use na na 6.4 181 na na

Any condom use

Ever used condom 12.9 410 5.0 1,230 7.0 1,640
Never used condom 9.5 2,226 5.5 1,170 8.1 3,396

Condom use at last sex in

past 12 months
Used condom at last sex 15.3 124 4.1 357 6.9 481
No condom at last sex 9.6 2,101 6.3 1,668 8.1 3,769

Total 10.1 2,636 5.2 2,399 7.8 5,036

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Totals include those with missing or inconsistent information on age at first sex. An asterisk indicates that a
figure is based on fewer than 25 unweighted cases and has been suppressed.
1
Refers to those age 15-24 only.
na = Not applicable

226 | HIV Prevalence and Associated Factors

 The relationship between condom use and HIV infection is not uniform. Among women, any
condom use and condom use at the most recent sexual encounter are associated with higher levels of HIV
infection, while for men, condom use is associated with a lower level of infection. Women who know or
suspect they or their partner might be infected would be more likely to use condoms, thus reversing the
expected direction of the relationship of lower HIV prevalence among those who use condoms.

None of the results discussed above demonstrate a consistent relationship between HIV preva-
lence and sexual behavioural risk, particularly among men. However, more detailed analysis will be re-
quired to thoroughly examine these relationships since they may be complicated by other factors such as
age, residence, and educational status that are associated with both the behavioral measures and HIV
prevalence.

HIV Prevalence by Other Characteristics Related to HIV Risk

Table 13.7 presents the variation in HIV prevalence with a number of other characteristics related
to HIV risk among men and women who have ever had sex. As expected, women and men with a history
of a sexually transmitted infection (STI) or STI symptoms have higher rates of HIV infection than those
with none. Among women who reported ever drinking alcohol, HIV prevalence is 19 percent, compared
with 9 percent among those who have never drunk alcohol. Differences for men are much smaller, with a
prevalence of HIV of 6 percent among those who have drunk alcohol compared with 4 percent for those
who have never drunk alcohol.

Table 13.7 HIV prevalence by selected other characteristics

Percentage HIV positive among women and men age 15-49 who ever had sex and who were tested, by whether had a
sexually transmitted infection, drank alcohol, had an HIV test, and perceived risk of getting AIDS, Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men Total
–––––––––––––––––– ––––––––––––––––– –––––––––––––––––
Percent Percent Percent
HIV HIV HIV
Characteristic positive Number positive Number positive Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Sexually transmitted infection
in past 12 months
Had STI or STI symptom 19.0 108 14.8 78 17.2 186
No STI, no symptoms 9.7 2,529 4.9 2,322 7.4 4,850

Use of alcohol
Drank alcohol 18.8 334 6.1 1,353 8.6 1,687
Last month 18.5 145 6.4 808 8.2 953
Ever, not in past month 18.9 189 5.7 545 9.1 734
Never drank alcohol 8.8 2,301 4.2 1,046 7.4 3,348

Perceived risk of getting AIDS

No risk at all 7.0 764 4.3 728 5.7 1,492
Small risk 11.0 1,077 5.8 1,261 8.2 2,338
Moderate risk 11.0 487 4.6 284 8.6 771
Great risk 14.6 267 7.5 118 12.4 385

HIV testing status

Ever tested 12.5 446 7.6 428 10.1 875
Never tested 9.7 2,155 4.8 1,965 7.4 4,119

Total 10.1 2,636 5.2 2,399 7.8 5,036

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Total includes 40 women and 6 men missing data on risk of getting AIDS

HIV Prevalence and Associated Factors | 227

 The relationship between perception of the risk of getting AIDS and actual HIV infection is not
straightforward, especially among men. It is disconcerting to note that 7 percent of women and 4 percent
of men who say they have no risk of getting AIDS are actually HIV positive.

Both women and men who have been tested for HIV in the past are more likely to be HIV in-
fected than those who have never been tested. Among women who have ever had sex, the level of HIV
infection is 13 percent among those who have ever been tested for HIV in the past, compared with 10 per-
cent among those who have never been tested. Among men, 8 percent of those previously tested are HIV
positive, compared with 5 percent of those who have never been tested.

Although the individual’s HIV status is associated with prior HIV testing, the above results indi-
cate that many individuals who are HIV positive have not been tested. Overall, four out of five of those
infected with HIV (82 percent of infected women and 77 percent of infected men) do not know their HIV
status, either because they were never tested or because they were tested and did not receive their results
(Table 13.8). For women, 18 percent of those who are HIV infected have been tested and know their re-
sults for their last test, compared to 13 percent of those who are HIV-negative. For men, there is a similar
pattern; 23 percent of those who are HIV-infected know their results for their last test, compared with 14
percent of those who are HIV negative.

Table 13.8 HIV prevalence by prior HIV testing

Percent distribution of HIV positive and negative women and men age 15-49 by HIV testing status
prior to the survey, Kenya 2003
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Women Men
–––––––––––––––––– ––––––––––––––––
HIV HIV HIV HIV
HIV testing status positive negative positive negative
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Ever tested and know results of last test 18.2 12.9 22.8 13.9
Ever tested, does not know results 2.6 1.4 2.2 1.5
Never tested 79.2 85.7 75.0 84.7

Total 100.0 100.0 100.0 100.0

Number 274 2,877 130 2,720

HIV Prevalence and Male Circumcision

Lack of circumcision is considered a risk factor for HIV infection, in part because of physiologi-
cal differences that increase the susceptibility to HIV infection among uncircumcised men. Several prior
studies in Kenya have shown a significant relationship between male circumcision and HIV risk (Agot et
al., 2004; Auvert et al., 2002; Baeten et al., 2002). The KDHS obtained information on male circumci-
sion status, and these results can be used to examine the relationship between HIV prevalence and male
circumcision status.

As Table 13.9 shows, the majority of Kenyan men (83 percent) are circumcised. However, the
proportion circumcised varies with province and ethnicity, being markedly lower among men in Nyanza
Province (46 percent), and among the Luo (17 percent).

228 | HIV Prevalence and Associated Factors

Table 13.9 HIV prevalence by male circumcision

Among men age 15-54 who were tested for HIV, percentage who are circumcised and percentage HIV positive among circumcised
and uncircumcised men, according to background characteristics, Kenya 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
All men
tested for HIV Circumcised men Uncircumcised men
––––––––––––––––––––– –––––––––––––––––––––– –––––––––––––––––––––
Number Percentage Number of Percentage Number of
Background Percentage of men HIV circumcised HIV uncircumcised
characteristic circumcised tested positive men positive men
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Age
15-19 70.3 745 0.5 524 0.0 221
20-24 89.4 566 1.0 506 14.1 60
25-29 87.3 428 5.2 374 21.7 54
30-34 89.3 368 5.5 329 (16.1) 39
35-39 89.4 321 5.4 287 (33.7) 34
40-44 84.3 260 4.2 219 (33.2) 41
45-49 81.9 163 2.9 133 (15.2) 29
50-54 86.4 193 1.9 167 (29.5) 26

Residence
Urban 82.2 763 5.4 627 16.9 136
Rural 83.8 2,280 2.3 1,911 11.0 369

Province
Nairobi 80.0 336 6.6 269 13.5 67
Central 89.4 476 2.4 425 * 50
Coast 95.6 210 4.1 201 13.4 9
Eastern 96.1 502 1.6 482 * 20
Nyanza 46.4 408 2.1 189 21.1 218
Rift Valley 86.7 718 2.9 623 6.8 95
Western 86.8 339 3.9 295 1.9 45
North Eastern 100.0 55 0.0 55 * 0

Education
No education 86.2 187 2.6 162 (0.0) 26
Primary incomplete 75.7 1,038 2.6 785 7.2 252
Primary complete 84.2 706 3.0 594 21.1 111
Secondary+ 89.6 1,113 3.5 997 19.1 116

Wealth quintile
Lowest 73.9 463 1.3 342 11.4 121
Second 82.9 531 2.8 440 9.6 91
Middle 88.9 558 1.3 496 11.8 62
Fourth 86.9 673 3.5 584 8.3 88
Highest 82.5 819 5.0 676 18.6 144

Ethnicity
Embu 100.0 41 3.3 41 * 0
Kalenjin 90.3 379 2.1 342 (0.0) 37
Kamba 99.4 353 1.7 351 * 2
Kikuyu 92.7 669 3.0 620 0.0 49
Kisii 99.5 172 0.5 171 * 1
Luhya 92.8 460 5.6 427 (0.0) 33
Luo 16.9 367 9.8 62 20.1 305
Maasai 82.5 59 2.2 49 1.4 10
Meru 91.0 187 1.2 170 * 17
Mijikenda/Swahili 100.0 124 2.8 124 * 0
Somali 100.0 86 1.7 86 * 0
Taita/Taveta 96.9 30 7.3 29 * 1
Turkana 44.4 51 0.0 23 (8.1) 28
Kuria 77.3 22 6.2 17 * 5

Religion
Roman Catholic 81.7 821 2.6 670 14.2 150
Protestant/other Christian 82.2 1,836 3.0 1,510 12.7 326
Muslim 100.0 188 2.9 188 * 0
No religion 86.4 192 5.6 166 (3.6) 26

Total 83.4 3,043 3.0 2,538 12.6 505

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: Total includes cases with “other” and missing religion and ethnicity. Figures in parentheses are based on 25-49 un-
weighted cases. An asterisk indicates that a figure is based on fewer than 25 unweighted cases and has been suppressed.

HIV Prevalence and Associated Factors | 229

 Looking at HIV prevalence levels by circumcision status, 13 percent of Kenyan men who are un-
circumcised are HIV infected, compared with 3 percent of those who are circumcised. In Nyanza Prov-
ince, men who are uncircumcised are ten times more likely to be HIV positive than men who are circum-
cised (21 and 2 percent, respectively). Multivariate analysis of these patterns is needed to obtain a better
understanding of the role that the lack of male circumcision may be playing in the susceptibility to HIV
infection in Kenya.

HIV Prevalence among Couples

Over 1,000 cohabiting couples were both tested for HIV in the 2003 KDHS. Results shown in
Table 13.10 indicate that, for 89 percent of cohabiting couples, both partners are HIV-negative while in 4
percent of couples, both partners are HIV positive. Seven percent of couples are discordant, that is, one
partner is infected and the other not. The variation in the level of couple HIV infection by background
characteristics generally conforms to the patterns observed with respect to the variation in individual se-
roprevalence rates, e.g., the infection rate is highest among couples in Nyanza Province.

Looking more specifically at discordant couples, among 3 percent of couples, the man is infected
and the woman uninfected, while in 5 percent of couples, the woman is infected and the man is not. Dis-
cordance is more common among couples in which the woman is age 20-29 or the man age 20-39, cou-
ples whose union is polygynous, urban couples, and couples in Nyanza. The fact that there are twice as
many couples that are discordant for HIV as couples that are both infected represents an unmet HIV pre-
vention need for the country, since the vast majority of these couples do not mutually know their HIV
status. Couple-oriented voluntary counselling and testing (VCT) services, where partners (including those
in polygynous marriages) go together and receive results together, is available throughout the country, but
few VCT clients attend as a couple.

HIV Prevalence among KDHS VCT Clients

As described in the introductory chapter, voluntary counselling and testing (VCT) for HIV was
provided to participants in the KDHS and others in the neighborhood (see Chapter 1). In all, 10,644 cli-
ents came voluntarily for information or counselling and 10,089 chose to be tested for HIV. Those who
came for VCT were self-selected and are not, therefore, representative of the adult population as a whole.
For example, two-thirds of those tested in the VCT component were men.

Among the 3,472 women who were part of the VCT component, 13 percent were HIV infected,
compared with the national rate of 9 percent obtained for women 15-49 in the survey. The higher preva-
lence among the women who came for VCT compared with those tested in the KDHS is consistent with
the KDHS finding that those who are HIV infected are more likely to learn their HIV status than those
who are negative. Five percent of the 6,617 men who were tested in the VCT component were HIV posi-
tive, which is identical to the rate for men tested in the KDHS. The large number who came for the mo-
bile VCT services is testimony to the desire for HIV testing and counselling, especially since the VCT
component was mainly confined to rural areas, and for the importance of offering participants in surveys
an opportunity to learn their HIV status.

230 | HIV Prevalence and Associated Factors

Table 13.10 HIV prevalence among couples

Among cohabiting couples both of whom were tested, percent distribution by HIV test results, according to background characteristics, Kenya
2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Male partner Female partner
Both positive, positive,
Background partners female partner male partner Both partners
characteristic HIV positive negative negative HIV negative Total Number
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Woman’s age
15-19 2.1 0.0 4.4 93.5 100.0 76
20-29 3.9 3.8 6.4 85.9 100.0 457
30-39 4.1 2.6 3.5 89.8 100.0 353
40-49 2.7 2.0 1.6 93.8 100.0 155

Man’s age
15-19 * * * * 100.0 7
20-29 3.7 3.5 5.3 87.5 100.0 244
30-39 3.8 3.3 5.5 87.4 100.0 403
40-54 3.5 2.0 3.2 91.3 100.0 386

Marital status
Married 3.3 3.1 4.4 89.2 100.0 948
Living together 7.7 0.5 6.0 85.8 100.0 92

Type of union
Monogamous 3.1 3.1 3.9 89.9 100.0 913
Polygynous 7.5 1.4 9.0 82.1 100.0 128

Residence
Urban 4.8 3.9 6.4 84.9 100.0 207
Rural 3.4 2.6 4.1 89.9 100.0 833

Province
Nairobi 5.2 4.7 9.4 80.7 100.0 89
Central 1.6 2.1 1.4 94.9 100.0 134
Coast 1.1 3.5 8.8 86.6 100.0 71
Eastern 2.3 0.0 3.5 94.3 100.0 159
Nyanza 9.8 8.7 8.4 73.2 100.0 169
Rift Valley 2.8 0.6 2.8 93.8 100.0 275
Western 2.2 3.3 3.0 91.5 100.0 121
North Eastern 0.0 0.0 0.0 100.0 100.0 25

Woman’s education
No education 1.8 1.2 0.9 96.1 100.0 143
Primary incomplete 4.5 4.7 7.2 83.6 100.0 373
Primary complete 4.3 0.9 2.9 91.8 100.0 273
Secondary+ 2.7 3.1 4.5 89.6 100.0 251

Man’s education
No education 1.7 1.0 1.4 95.9 100.0 94
Primary incomplete 2.7 2.4 4.8 90.1 100.0 289
Primary complete 5.4 2.7 4.4 87.6 100.0 261
Secondary+ 3.7 3.8 5.3 87.3 100.0 397

Wealth quintile
Lowest 2.3 2.2 3.5 91.9 100.0 210
Second 4.0 2.8 4.3 88.9 100.0 211
Middle 2.4 3.3 4.0 90.3 100.0 208
Fourth 2.8 1.5 5.7 90.0 100.0 203
Highest 6.6 4.4 5.3 83.6 100.0 209

Total 3.7 2.8 4.6 88.9 100.0 1,041

––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Note: An asterisk indicates that a figure is based on fewer than 25 unweighted cases and has been suppressed.

HIV Prevalence and Associated Factors | 231

 13.3 DISTRIBUTION OF THE HIV BURDEN IN KENYA

The inclusion of HIV testing in the KDHS provides the basis for a more precise estimate of the
burden of HIV in Kenya and permits the calibration of estimates of HIV prevalence based on sentinel
surveillance in pregnant women. Kenya has a heterogeneous HIV epidemic, with significant differences
by region. Three provinces, containing half of Kenya’s population, have 65 percent of the HIV infections:
Nyanza Province with nearly one-third, Rift Valley with around one-fifth, and Nairobi with one-sixth of
HIV infections in Kenya. Urban residents represent 25 percent of the population age 15-49, but nearly 40
percent of those HIV infected are urban residents. Higher educational level does not protect one from
HIV infection in Kenya; HIV has spread through all regions and strata of society.

The linkage of biological and behavioural data in this survey has strengthened the validity of this
survey by making multivariate analysis possible. The measurement of HIV prevalence in the KDHS has
proven useful in calibrating HIV prevalence estimates of the general population from sentinel surveillance
in pregnant women and has resulted in downward projections of the severity of the epidemic in Kenya.
These adjustments arise from a better understanding of rural-urban population distribution, from a recog-
nition that rural pregnant women who do not seek ANC care have lower rates than those who do, and,
most importantly, from acknowledgement of the high ratio of 1.9 women infected for every man.

This linkage between HIV test results and demographic and behavioural data also enhances the
understanding of the distribution, patterns, risk factors for HIV in Kenya, with the potential for improved
planning and implementation of programs as a result of this information. The higher rate of HIV in uncir-
cumcised men supports the need to evaluate possible causal links between male circumcision and HIV.
Finally, the prevalence of couples that are discordant for HIV underscores the need for knowledge of both
one’s own HIV status and that of one’s partner in order to prevent the continued spread of the HIV epi-
demic.

232 | HIV Prevalence and Associated Factors