Content uploaded by Brian Gerard Williams
Author content
All content in this area was uploaded by Brian Gerard Williams on Jun 08, 2020
Content may be subject to copyright.
The Impact of Migration on HIV-1 Transmission in South Africa
A Study of Migrant and Nonmigrant Men and Their Partners
MARK N. LURIE, PHD,*
†
BRIAN G. WILLIAMS, PHD,
‡
KHANGELANI ZUMA, MA,*
§
DAVID MKAYA-MWAMBURI, MD,*
GEOFF P. GARNETT, PHD,
储
ADRIAAN W. STURM, MD,
†¶
MICHAEL D. SWEAT, PHD,
#
JOEL GITTELSOHN, PHD,
#
AND SALIM S. ABDOOL KARIM, MDChB, PHD**
††
Background: To investigate the association between migration and
HIV infection among migrant and nonmigrant men and their rural
partners.
Goal: The goal was to determine risk factors for HIV-1 infection in
South Africa.
Study Design: This was a cross-sectional study of 196 migrant men
and 130 of their rural partners, as well as 64 nonmigrant men and 98
rural women whose partners are nonmigrant. Male migrants were
recruited at work in two urban centers, 100 km and 700 km from their
rural homes. Rural partners were traced and invited to participate.
Nonmigrant couples were recruited for comparison. The study in-
volved administration of a detailed questionnaire and blood collection
for HIV testing.
Results: Testing showed that 25.9% of migrant men and 12.7% of
nonmigrant men were infected with HIV (Pⴝ0.029; odds ratio ⴝ2.4;
95% CI ⴝ1.1–5.3). In multivariate analysis, main risk factors for male
HIV infection were being a migrant, ever having used a condom, and
having lived in four or more places during a lifetime. Being the partner
of a migrant was not a significant risk factor for HIV infection among
women; significant risk factors were reporting more than one current
regular partner, being younger than 35 years, and having STD symp-
toms during the previous 4 months.
Conclusion: Migration is an independent risk factor for HIV infection
among men. Workplace interventions are urgently needed to prevent
further infections. High rates of HIV were found among rural women,
and the migration status of the regular partner was not a major risk
factor for HIV. Rural women lack access to appropriate prevention
interventions, regardless of their partners’ migration status.
SOUTH AFRICA is experiencing one of the most rapidly growing
HIV epidemics in the world. Among women attending antenatal
clinics nationwide, the prevalence of HIV infection increased from
0.76% in 1990 to 24.5% in 2000.
1
Among the nine South African
provinces, KwaZulu/Natal has consistently had the highest ante-
natal HIV prevalence, which in 2000 was 36.2%. As in the rest of
sub-Saharan Africa, the predominant mode of transmission is
heterosexual intercourse.
Over the past century, migration became common among rural
men seeking employment in urban and mining centers, and this
persists today. In the Hlabisa District of rural KwaZulu/Natal
South Africa, the site of this study, for example, 62% of adult men
spent the majority of nights away from their rural homes.
2
Men
also migrate to South Africa from neighboring countries, and there
are an estimated 2.5 million legal and many more undocumented
migrants in South Africa today.
3
Twenty years ago the gold mines
employed approximately 500,000 people, of whom about half
were South African and the rest were from neighboring countries,
including Botswana, Lesotho, Mozambique, and Malawi.
4
While
the number of men employed in the gold mines has fallen to about
300,000, the southern Africa region is still linked by extraordinar-
ily high levels of migration.
5
Although there are many different types of migration, the pre-
dominant mode of migration in southern Africa is “circular” or
“oscillating” migration, in which young men leave their rural
partners to work in urban areas and return home periodically,
depending on the distances involved.
The roots of migrant labor in South Africa run deep and can be
traced to the discovery of gold on the Witwatersrand in 1886 and
the associated demand for labor. The system of migrant labor was
a cornerstone of apartheid policy, in which the movement of South
Africa’s black population was strictly controlled so as to maintain
a separation of the races while ensuring a steady supply of labor-
ers, who were prohibited from settling permanently in “whites-
The authors thank Nozizwe Dladla and all of the members of the
Migration Project Team, without whom this project would not have been
possible, as well as Mervyn Susser, Peter Lurie, Jonathan Zenilman, David
Celentano, Abigail Harrison, Thomas Painter, Zena Stein, and David
Wilkinson, each of whom contributed to this project in important ways.
This study was part of the Africa Centre for Population Studies and
Reproductive Health in Mtubatuba, South Africa and was supported by the
Wellcome Trust (Grant #050517/z/97abc) and the South African Medical
Research Council. This publication was made possible in part through the
support of a training grant awarded by the National Institute of Drug Abuse
to The Miriam Hospital (Grant #5 T32 DA13911) and a Fogarty AIDS
Training Grant (Grant #Two-0321).
Reprint requests: Mark Lurie, PHD, Brown University School of Med-
icine, Department of Medicine, 164 Summit Avenue, Providence, RI
02906. E-mail: Mark_Lurie@brown.edu
Received March 12, 2002, revised June 26, 2002, and accepted June 28, 2002.
From the *South African Medical Research Council, HIV
Prevention and Vaccine Research Unit, Durban, South Africa;
†
Brown University School of Medicine, Department of Medicine,
and Miriam Hospital, Providence, Rhode Island;
‡
Communicable
Diseases, World Health Organization, Geneva, Switzerland;
§
Department of Statistics, University of New Zealand, Waikato,
New Zealand;
储
Department of Infectious Disease Epidemiology,
Faculty of Medicine, Imperial College, London, UK;
¶
Department
of Medical Microbiology, **University of Natal, Durban, South
Africa;
#
Department of International Health, Johns Hopkins
University School of Hygiene and Public Health, Baltimore,
Maryland; and
††
Columbia University, New York, New York
149
only”areas. Patterns of migration have changed dramatically,
however, in the last decade. With the lifting of apartheid laws, the
emergence of trade unions that were able to negotiate more flexible
work contracts, and the rapid development of an extensive, infor-
mal, but efficient transport infrastructure, people are able to move
more freely than before, and HIV, like other infectious diseases
that spread from person to person, follows the movement of
people.
6
Migration is one of many social factors that have contributed to
the AIDS epidemic.
7,8
Several studies have shown that people who
are more mobile or who have recently changed residence tend to
be at higher risk for HIV and other sexually transmitted diseases
(STDs) than people in more stable living arrangements.
9–12
In
Uganda, people who had moved within the last 5 years were 3
times more likely to be infected with HIV than those who had lived
in the same place for more than 10 years,
13
and in South Africa,
people who had recently changed their residence were 3 times
more likely to be infected with HIV than those who had not.
14
Decosas and others have argued that it is not so much the move-
ment itself but rather the “conditions and structure of the migration
process” that put people at risk for HIV and other STDs.
7
The role of migration in the spread of HIV has been described
primarily as a result of men becoming infected while they are away
from home and infecting their wives or regular partners when they
return. In a study of seasonal migration in Senegal, Pison argued that
the virus was “mainly transmitted first to adult men through sexual
contacts met during their seasonal migration and second to their wives
or regular partners once they are back home.”
9
Other studies have
shown that men who live away from their wives or regular partners
are more likely than those who live with their wives or regular
partners to have additional sex partners
11
and are therefore more likely
to become infected with HIV
11
or other STDs.
8
However, the precise way in which migration contributes to the
spread of STDs is complex and not well understood. Previous
studies have focused on the destinations of migrants, or, less often,
on the areas from which migrants come
15
; few studies have con-
sidered both ends of the migration process—those who leave home
as well as those who remain behind. These studies therefore tend
to give a static view of what is essentially a complex and dynamic
process. Understanding both ends of migration routes is essential
if targeted interventions are to be successfully implemented.
Methods
This study tested the hypothesis that migrants and their partners
are at greater risk for HIV infection than are nonmigrants and their
partners, and we investigated potential risk factors for HIV infec-
tion. We measured the prevalence of HIV-1, syphilis, chlamydia,
and gonorrhea (although here we report only on HIV-1) among
migrant men and their rural partners, as well as among nonmigrant
men and their rural partners. We also conducted a behavioral
survey with the same study participants to identify social, behav-
ioral, and biomedical risk factors associated with HIV infection.
Between October 1998 and November 2000, male migrants
from two adjacent rural districts (Hlabisa and Nongoma) were
recruited at two migration destinations, Carletonville and Richards
Bay (Figure 1), 700 km and 100 km away, respectively, from their
rural homes. These sites were chosen because they are common
destinations for migrant men from rural KwaZulu/Natal
2
and be-
cause they represent the two common types of migration prevalent
in the area: long-distance migration with infrequent trips home
(Carletonville) and short-distance migration with more frequent
trips home (Richards Bay). Carletonville is a gold-mining town
southwest of Johannesburg with a population of roughly 220,000
people, of whom 80,000 are migrant men living in single-sex
hostels and working in the gold mines. Because of the distances
involved, these men tend to return home only three to four times
a year. Richards Bay, an industrial town on the north coast of
KwaZulu/Natal, is also a common migration destination for these
rural men, but because of the proximity to their rural homes, they
are able to return home more frequently, on average at least once
a month.
Three gold mines in Carletonville and three factories in Rich-
ards Bay were selected because they employ large numbers of
people from the Hlabisa and Nongoma districts. Lists of workers’
origins were generated through a census in Richards Bay and
through a list provided by the Employment Bureau of Africa, the
agency responsible for recruiting men to work in the gold mines.
Men from the Hlabisa and Nongoma districts were invited to the
project offices, where the purpose of the study was explained and
they were invited to participate. Men were included only if they
were from the Hlabisa/Nongoma districts, if they had been a
migrant for at least 6 months, and if they had at least one “regular”
partner living in Hlabisa/Nongoma. A regular partner was defined
through prior focus group discussions
16
as a stable sex partner with
whom one envisions a future (maqondana, in Zulu). Those who
were eligible and who agreed to participate were administered a
detailed questionnaire and offered voluntary counseling and test-
ing for HIV and STDs.
In addition, migrant men were asked a series of questions in
order to locate and identify their rural partners. These included
questions about the name of the head of the rural household, the
nearest clinic and school, and specific directions to the household
of the migrant’s rural partner. This information was sent to the
project field office, where field-workers visited these women and
invited them to participate in the study. Once a participating
partner of a migrant man was identified, a nonmigrant couple
living within a radius of one kilometer of each migrant household
was identified and invited to participate in the study. A nonmigrant
man was defined as a person who spends most nights at home and
who had not been a migrant for a total of more than 6 months over
the last 5 years. All women were residents of the Hlabisa/Non-
goma districts and none were migrants. The number of men and
women participating in the study was not equal because of refusal
to participate and inability to trace some partners.
Structured, face-to-face interviews were held with each partic-
ipant and included socioeconomic and demographic questions,
Fig. 1. Map of South Africa with study sites.
150 Sexually Transmitted Diseases ●February 2003LURIE ET AL
migration histories, details of stable (regular) and casual sex part-
nerships, condom use, age at sexual debut, and history of and
health-seeking behavior for current or previous urogenital disease
symptoms.
All participants were offered pretest and posttest HIV counsel-
ing, free condoms at each visit, and free treatment for symptomatic
or laboratory-diagnosed STDs. Participants were encouraged to
receive their HIV test results but were also given the option of not
receiving them should they so desire.
17
Trained nurses treated
symptomatic STDs at the time of enrollment, using the KwaZulu/
Natal provincial syndromic management guidelines.
18
Laboratory-
diagnosed syphilis, chlamydia, and gonorrhea were treated at
10-day follow-up visits. The presence of symptomatic STDs is a
major risk factor for HIV transmission,
19
and treatment therefore is
likely to confer some protection against HIV infection. Those who
agreed to participate were followed-up every 4 months. In this
article we present an analysis of the cross-sectional enrollment
data.
Specimen Collection and Processing
Two milliliters of venous blood were taken from those who
consented to participate and were tested for HIV-1 by means of the
Determine Rapid Test (Abbott Diagnostics). Positive tests were
confirmed by two additional ELISAs (HIV 1.2.0, Abbott/Murieux;
Vironosticka HIV uniform 2 ⫹0, Omnimed). A random sample of
10% of the specimens that were negative on the Determine Test
were also subjected to ELISA confirmation to validate the speci-
ficity of the testing methodology; all of those tests remained
negative by ELISA. Specimens collected in Richards Bay and the
Hlabisa and Nongoma districts were transported daily from the
Hlabisa office to the laboratory in Durban. Specimens collected in
Carletonville were flown nightly via courier to the same Durban
laboratory.
The study was approved by the human subjects committees of
The Johns Hopkins University School of Hygiene and Public
Health and the University of Natal, Durban.
Data Management and Statistical Analysis
Data were double-entered and analyzed with SAS version 6.12
(SAS Institute, Cary, NC). The primary outcome was HIV-1
infection. The analysis was done separately by gender. Differences
in quantitative variables were assessed with Student ttest. Tests of
significance for categorical variables were based on chi-square test
or Fisher exact test, as appropriate. Logistic regression was used to
estimate the adjusted odds ratios for HIV seropositivity. Confi-
dence intervals are given as 95%. All Pvalues were derived from
two-sided tests. A Pvalue of ⱕ0.05 was considered statistically
significant.
Results
Between October 1998 and November 2000, 260 men and 228
women were recruited for the study; 196 migrant men from the
Hlabisa/Nongoma districts were recruited at their workplaces, 64
nonmigrant men were recruited in Hlabisa/Nongoma, and 130
female partners of migrants and 98 female partners of nonmigrants
were recruited in the Hlabisa/Nongoma districts. Not all study
participants were matched to a partner because some partners
refused to participate, and in some cases it was not possible to find
the partner.
The overall prevalence of HIV-1 infection was 20.1%. Preva-
lence among men was not significantly different from that among
women (22.7% versus 19.1%, respectively; P⫽0.34; OR ⫽1.2;
95% CI ⫽0.80–1.93). The prevalence of HIV-1 among migrants
and their partners, however, was significantly higher than among
nonmigrants and their partners (24.0% versus 15.0%, respectively;
P⫽0.02; OR ⫽1.8; 95% CI ⫽1.1–3.0). Results are presented by
gender.
Males
The sociodemographic and biomedical data for migrant and
nonmigrant men are shown in Table 1. The median age was 39.1
years (mean ⫽37.4; SD ⫽8.4) and migrants were, on average, 6
years younger than nonmigrants (P⬍0.001). Most men had some
education, and migrants tended to be better educated than nonmi-
grants. Almost 40% of nonmigrants but only 20% of migrants had
never attended school, while less than 20% of nonmigrants and
nearly 30% of migrants had attended secondary school. Nearly all
men were either married or living as married, with similar propor-
tions among migrants and nonmigrants. Four men were widowed,
divorced, or separated, and 14% were single, although, because of
the study’s enrollment criteria, the single men had to have at least
one regular partner in Hlabisa/Nongoma. Migrant men were sig-
nificantly more likely than nonmigrant men to derive an income
from formal employment; all of the migrant men but only 43% of
nonmigrant men had a formal income.
Almost all of the nonmigrant men lived with their wives or
regular partners most of the time, while very few of the migrant
men did. In Carletonville, all but three of the men lived in single-
sex hostels provided by employers (data not shown), while in
Richards Bay only three men lived in employer-provided accom-
modation, and the majority lived either alone (36%), with other
workers (17%), or with relatives (22%).
Most men reported only one current regular sex partner, but
about 30% of both migrant and nonmigrant men said that they had
two or more regular partners. Nonmigrant men were more likely to
have regular partners in Hlabisa/Nongoma, while migrant men
were more likely to have regular partners outside of Hlabisa/
Nongoma, mostly at their migration destination. Migrant men were
significantly more likely (P⫽0.02) than nonmigrant men to have
at least one current casual partner, but only 20% of migrant men
and 6% of nonmigrant men reported having one or more casual
partners. Most of the men who had casual partners were migrants
younger than 35 years of age. The median reported age of sexual
debut for migrant men was 18 years, and for nonmigrant men, 19
years. Nonmigrant men reported a significantly higher number of
lifetime partners than did migrant men, although this may be partly
confounded by age.
Condom use was low; fewer than 20% of men in both groups
reported that they had ever used a condom. Men who were younger
than 35 years of age were significantly more likely than older men
to have ever used a condom (OR ⫽2.4; 95% CI ⫽1.2–4.6), and
men who reported having many casual partners were more likely
than men who reported few casual partners to have ever used
condoms. Compared to men who had no casual partners, the odds
of ever having used a condom was 1.7 (95% CI ⫽1.1–2.7) for
those who had one casual partner and 8.4 (95% CI ⫽1.5–49.0) for
those who had four casual partners. Nonmigrant men were more
likely than migrant men to have used condoms in regular relation-
ships (10.9% versus 23.7%; P⫽0.04).
Approximately one quarter of men said that they ever had a
genital ulcer and 35% said they had ever experienced genital
discharge. Approximately 7% of men said that they were experi-
encing ulcers, discharge, swollen testes, or swollen lymph nodes at
the time of the survey. These symptoms were equally common
among migrant and nonmigrant men.
The prevalence of HIV among migrant men was significantly
higher than among nonmigrant men (25.9% versus 12.7%; P⫽
Vol. 30 ●No. 2 151IMPACT OF MIGRATION ON HIV TRANSMISSION IN SOUTH AFRICA
0.03; OR ⫽2.4; 95% CI ⫽1.1–5.3), and the prevalence was
higher among migrant men than among nonmigrant men when
stratified according to age (Table 2), although the individual with-
in-age-group differences were not statistically significant because
of the limited sample size.
Table 3 shows univariate analyses for risk factors associated
with HIV infection. The most important risk factors for HIV
among men were being a migrant, being ⬍35 years old, having
one or more casual partners, having symptoms of STDs in the last
4 months, and ever having used a condom. Those with current
TABLE 1. Selected Sociodemographic and Sexual Behavior Variables for Migrant and Nonmigrant Men, Partners of Migrants, and
Partners of Nonmigrants*
Variable
Men
PValue
Women
PValueMigrant Nonmigrant
Partners of
Migrants
Partners of
Nonmigrants
Mean age (y) 37.4 43.6 34.2 39.1
SD 8.4 9.5 ⬍0.001 8.4 11.0 ⬍0.001
n 196 64 130 98
Level of education
(completed) n ⫽192 n ⫽63 n ⫽128 n ⫽97
None 39 (20.3%) 23 (37.5%) 22 (19.2%) 30 (30.9%)
Grades 1–5 83 (43.2%) 27 (42.2%) 0.02 62 (48.4%) 45 (46.4%) 0.05
Grades 6–9 55 (28.6%) 12 (18.8%) 34 (25.6%) 19 (19.6%)
Matric; Matric ⫹Cert/Dip 15 (7.8%) 1 (1.6%) 10 (7.8%) 3 (3.1%)
Current marital status n ⫽191 n ⫽62 n ⫽127 n ⫽98
Married—civil 55 (28.8%) 21 (33.3%) 40 (31.5%) 18 (18.4%)
Married—traditional 80 (41.9%) 23 (38.1%) 55 (43.3%) 54 (55.1%)
Unmarried but committed
or living as married 23 (12.0%) 11 (17.5%) 0.59 8 (6.3%) 19 (19.4%) 0.001
Widowed/divorced/separated 3 (1.6%) 1 (1.6%) 0 0
Single 30 (15.7%) 6 (9.5%) 24 (18.9%) 7 (7.1%)
Total no. of current regular
partners n ⫽193 n ⫽64 n ⫽130 n ⫽98
1 133 (68.9%) 40 (62.5%) 130 (100%) 97 (98.9%)
2 42 (21.8%) 16 (25%) 0.54 0 1 (1.1%) 0.25
ⱖ3 10 (5.2%) 6 (9.4%) 0 0
Refused 8 (4.2%) 2 (3.1%)
Total no. of current casual
partners n ⫽195 n ⫽64 n ⫽130 n ⫽98
0 155 (79.9%) 60 (93.8%) 0.02 127 (97.6%) 98 (100%) 0.13
1 16 (8.3%) 3 (4.7%) 3 (2.3%)
ⱖ2 23 (11.8) 1 (1.6%)
Age (y) at first sex n ⫽158 n ⫽55 n ⫽94 n ⫽102
Mean 18.2 18.7 0.07 17.6 17.1 0.19
SD 2.9 3.8 2.6 2.4
No. of lifetime partners n ⫽121 n ⫽56 n ⫽96 n ⫽91
Mean 13.4 18.2 ⬍0.0001 1.8 2.0 0.36
SD 13.1 23.4 1.3 1.8
Condoms
Ever used 32/182 (17.6%) 14/63 (22.2%) 0.41 14/123 (11.4%) 11/93 (11.8%) 0.92
Ever used with wife 6/92 (6.5%) 5/56 (8.9%) 0.59 10/96 (10.4%) 6/76 (7.9%) 0.57
Ever used in regular
relationship 14/129 (10.9%) 9/38 (23.7%) 0.04 3/25 (12%) 5/17 (29.4%) 0.16
Ever used in casual
relationship 12/56 (21.4%) 0/6 (0%) 0.21 2/2 (100%) 0/0
STD history: ulcer
Currently 7/191 (3.7%) 1/63 (1.6%) 0.41 5/128 (3.9%) 3/96 (3.9%) 0.76
In last 4 mo 21/194 (10.8%) 4/63 (6.3%) 0.29 16/124 (12.9%) 5/90 (5.6%) 0.07
Ever 45/192 (23.4%) 21/63 (33.3%) 0.12 34/128 (26.6%) 20/98 (20.4%) 0.28
STD history: discharge
Currently 2/192 (1.0%) 0/64 (0%) 0.41 11/127 (8.7%) 11/96 (11.5%) 0.49
Last 4 mo 10/193 (5.2%) 1/64 (1.6%) 0.38 39/122 (31.9%) 15/88 (17.1%) 0.02
Ever 82/191 (42.9%) 23/64 (35.9) 0.33 64/128 (50%) 35/97 (36.1%) 0.04
One or more STD
symptoms
†
Currently 15/192 (7.8) 3/64 (4.7) 0.39 26/128 (20.3%) 21/98 (21.4%) 0.84
Last 4 mo 34/194 (17.5) 6/64 (9.4) 0.12 56/130 (43.1%) 28/98 (28.6%) 0.03
Ever 95/192 (47.9) 29/64 (45.3) 0.56 84/128 (65.6%) 50/98 (51.0%) 0.03
*Participants with unknown response, with “don’t know” response, or who refused to answer were excluded.
†
One or more of the following symptoms: ulcer, discharge, swollen testes, swollen lymph nodes.
152 Sexually Transmitted Diseases ●February 2003LURIE ET AL
STD symptoms, symptoms in the last 4 months, or a history of
STD symptoms were more likely to be HIV-infected than those
who had never had STD symptoms. Those who had ever used
condoms were more likely to be HIV-positive than those who had
not. The probability of being infected with HIV was not signifi-
cantly associated with income, education, lifetime number of part-
ners, age at sexual debut, or the number of places lived over the
course of a lifetime.
A multivariate, forward-stepwise logistic regression was carried
out, including all those variables that were found to be significant
in the univariate analysis, as well as other variables of potential
importance, leading to the model given in Table 4. In the multi-
variate analysis the risk of HIV infection remains higher among
migrants than among nonmigrant men (OR ⫽2.65; P⫽0.026),
among those who report recently having STD symptoms (OR ⫽
2.09; P⫽0.029) and among those who have lived in more than
four places (OR ⫽3.56; P⫽0.001) rather than only one place.
Having lived in four or more places was not significant on biva-
riate analysis but became significant in the multivariate model.
Those who said that they have ever used condoms were also at
greater risk of HIV infection than those who said that they had not
(OR ⫽2.18; P⫽0.045), but this is confounded by the fact that
those who reported having used condoms were also likely to have
had more casual partners than those who said that they had never
used condoms.
Females
Of the 228 women recruited for the study, 130 were partners of
migrants and 98 were partners of nonmigrants (Table 1). Because
of the study design, none of the women were migrants. The women
were, on average, about 4 years younger than their male partners.
The level of education of women was similar to that of their male
partners; a quarter of women had had no formal education, and
23.5% had had at least some secondary education. Partners of
migrants were significantly more educated than the partners of
nonmigrants (P⫽0.05). Few women in either group were for-
mally employed. The partners of migrants were significantly more
likely than partners of nonmigrants to receive financial support
from their partners, which is to be expected since men still migrate
largely for economic reasons. Nevertheless, only half of the part-
ners of migrant men said that they received financial support from
their partner.
As with the men, most of the women were married or living as
married; 19% of the regular partners of migrants and 7% of the
regular partners of nonmigrants said that they were single (P⫽
0.01). Only one woman said that she had more than one regular
partner, and only three women said that they had any casual
partners. The median age at sexual debut, 17 years, was 1 year
younger for women than for men. Women reported having, on
average, only two lifetime partners, fewer than reported by the
men, suggesting that they had ever had only one partner apart from
their current regular partner.
Reported use of the male condom was lower among women than
it was among men (P⫽0.07); almost 90% of women said that they
had never used a condom. Women who reported ever having used
a condom had slightly more lifetime partners than women who had
never used a condom (1.9 versus 2.0; P⫽0.096).
STD symptoms were also common among women, with 24%
saying that they had ever had a genital ulcer and 44% that they had
ever experienced a discharge. Two thirds of all women said that
they had experienced discharges, ulcers, and/or swollen lymph
nodes, and partners of migrants were more likely to have experi-
enced these symptoms than partners of nonmigrants (P⫽0.03).
HIV infection was more frequent in partners of migrants than
partners of nonmigrants (21.1% and 16.5%, respectively), al-
though these differences were not statistically significant (P⫽
0.39). Among the youngest group of women (Table 2), HIV
infection prevalence was higher for partners of nonmigrants
(34.5%) than partners of migrants (25.7%); again, this difference
was not significant (P⫽0.39). In the two older age groups,
partners of migrants had a higher prevalence of HIV infection than
partners of nonmigrants; these differences were not significant.
Table 3 shows risk factors for HIV infection among women. The
strongest association was with the number of lifetime partners:
women who reported having had more than one lifetime sex
partner were five times more likely to be infected with HIV than
women who said that they had only had one lifetime partner (OR
⫽5.1; 95% CI ⫽2.2–11.5). Age was also a significant risk factor
for HIV, with younger women more likely to be infected than older
women (OR ⫽2.3; 95% CI ⫽1.2–4.5). Women who reported
having sexual intercourse for the first time at or before the age of
17 years were more likely to be HIV-positive (24.5%) than those
who reported a later age at sexual debut (14.3%), although this was
only marginally significant (P⫽0.07; OR ⫽2.0; 95% CI ⫽
1.0–4.1).
The prevalence of HIV among women was not significantly
associated with being the partner of a migrant, receiving financial
support from the husband or regular partner, level of education,
STD symptoms, or ever having used a condom. Women who had
used a condom were as likely to be HIV-infected as those who had
not.
Table 4 shows the results of the multivariate forward-stepwise
logistic regression model for women. Young women, and those
who had had more than one lifetime partner, were at particularly
high risk of infection.
Discussion
The exceptionally high prevalence of HIV in most southern
African countries has raised important and complex questions
about the factors that have contributed to the rapid spread of HIV
in the region and about the eventual prevalence the epidemic might
reach. This cross-sectional, community-based study of migrant and
nonmigrant men and their rural partners has revealed a very high
TABLE 2. Age-Specific HIV Prevalence (%) for Migrant and Nonmigrant Men, Partners of Migrants, and Partners of Nonmigrants
Group % HIV
⫹
(n)
Men: Age (y) 22–34 35–49 50–66
Migrant 33.8% (77) 21.2% (99) 17.7% (17)
Nonmigrant 22.2% (9) 13.8% (36) 5.6% (18)
Women: Age (y) 18–34 35–49 50–66
Partners of migrants 25.7% (70) 15.1% (53) 20% (5)
Partners of nonmigrants 34.5% (29) 7.6% (53) 14.3% (14)
Vol. 30 ●No. 2 153IMPACT OF MIGRATION ON HIV TRANSMISSION IN SOUTH AFRICA
TABLE 3. HIV Prevalence (%) in Relation to Important Risk Factors, for Men and Women*
Variable n % HIV
⫹
POR (95% CI)
Men
Migrant
Yes 193 25.9 0.03 2.4 (1.1–5.3)
No 63 12.7
Monthly income ($)
0–2000 163 23.3 0.73 0.9 (0.5–1.7)
2000⫹79 25.3
Age (y)
ⱕ35 95 30.5 0.02 2.0 (1.1–3.6)
⬎35 161 18.0
Level of education
†
None 61 18.0 0.73 0.87 (0.4–1.9)
Grades 1–5 110 26.4 0.43 1.3 (0.7–2.5)
Grades 6–10⫹81 22.2 1
No. of lifetime partners
ⱕ5 52 15.4 0.23 0.6 (0.3–1.4)
⬎5 124 23.4
No. of regular partners
1 170 22.9 0.81 1.1 (0.6–2.0)
ⱖ2 74 24.3
No. of casual partners
0 212 19.8 0.001 2.7 (1.4–5.3)
ⱖ1 43 37.2
Age at first intercourse (y)
ⱕ17 81 25.9 0.33 1.4 (0.7–2.7)
⬎17 129 20.2
STD symptoms
Ever 131 27.5 0.07 1.7 (0.95–3.1)
Never 122 18.0
STD symptoms currently
Yes 18 38.8 0.09 2.3 (0.9–6.2)
No 238 21.4
STD symptoms, last 4 mo
Yes 40 42.5 0.001 3.2 (1.6–6.3)
No 216 19.0
Condom use
Ever 36 32.6 0.03 2.3 (1.1–4.8)
Never 196 19.8
Number of places lived, lifetime
⬍4 150 22.0 0.77 1.3 (0.7–2.3)
ⱖ4 106 23.6
Women
Partner of Migrant
Yes 128 21.1 0.39 1.4 (0.7–2.7)
No 97 16.5
Monthly income from husband
Yes 101 14.9 0.18 1.6 (0.8–3.2)
No 123 21.9
Age (y)
ⱕ35 110 25.5 0.02 2.3 (1.2–4.5)
⬎35 114 13.2
Level of education
†
None 50 18.0 0.42 0.7 (0.3–1.7)
Grades 1–5 107 16.8 0.23 0.6 (0.3–1.4)
Grades 6–10⫹66 24.2 1
No. of lifetime partners
1 92 7.6 0.001 5.0 (2.2–11.5)
⬎1 93 29.0
Age at first intercourse (y)
ⱕ17 102 24.5 0.07 2.0 (1.0–4.1)
⬎17 92 14.3
STD symptoms
Ever 133 20.3 0.56 1.2 (0.6–2.4)
Never 92 17.4
STD symptoms currently
Yes 47 19.1 0.99 1.0 (0.4–2.3)
No 178 19.1
STD symptoms last 4 mo
Yes 84 23.8 0.17 1.6 (0.8–3.1)
No 141 16.3
Condom use
Ever 25 28.0 0.27 0.6 (0.2–1.5)
Never 188 18.6
*Participants with unknown response, with “don’t know”response, or who refused to answer were excluded.
†
Reference group: grades 6–10⫹.
154 Sexually Transmitted Diseases ●February 2003LURIE ET AL
prevalence of HIV among both men and women. The study pro-
vides evidence of the importance of migration in the spread of HIV
in southern Africa and shows that migration is a significant risk
factor for HIV-1 for men.
For men, being a migrant and having lived in four or more
places were independent and significant risk factors for HIV-1
infection. Thus, not only is labor migration—with its associated
separation of families—an important risk factor for HIV-1 trans-
mission, but so too is the social disruption caused by repeated
relocation, in some cases forced as a result of apartheid policies
and political violence.
These findings are particularly interesting, given the mature
stage of the southern African HIV/AIDS epidemic. It is likely, for
example, that the role of migration in the spread of HIV was more
important—and more easily measured—in the early stages of the
epidemic than in the later stages.
20
Indeed, isolating a single causal
factor in a mature epidemic, when prevalence is already very high,
was likely to be difficult. The fact that the odds of a migrant man
being infected was 2.4 times the odds of a nonmigrant man being
infected, even at this advanced stage of the epidemic, highlights
the importance of migration as one explanation for the size and
rapidity of spread of the southern African epidemic.
For women, being a partner of a migrant confers a slight but not
statistically significant risk for being HIV-infected. There are
several possible reasons why the higher rates of HIV observed
among migrant men versus nonmigrant men do not translate into
significantly higher rates among their rural partners. The study was
not powered to measure a 5% difference in HIV prevalence be-
tween partners of migrants and partners of nonmigrants, but
whether a larger sample size would confirm a difference is unclear.
The fact that, for young women, HIV prevalence was higher
among partners of nonmigrants than among partners of migrants
shows that there is transmission occurring in the rural areas,
regardless of the migration status of the women’s partners. This
conclusion is supported by the patterns of HIV-1 discordance
among these same couples.
21
In addition, the fivefold increased
risk for HIV infection among women reporting more than one
lifetime partner suggests that women are more likely to be infected
by someone other than their regular partner.
Since most research on migration and AIDS has taken place
only at male migration destinations and has excluded the rural end
of the migratory routes, there has been a suggestion that interven-
tions for migrants should be targeted at male migration destina-
tions. Indeed, operational issues, including the ease of finding and
following people, make this an attractive option. Our findings,
however, demonstrate the complexity of HIV transmission in the
presence of large-scale male migration and the need to address the
spread of disease, especially among young rural women—not just
women living in migrant relationships. What has not been ac-
knowledged to date is the role of local, rural transmission in this
complex epidemic. The findings of this study show that it is
important to include rural areas if HIV treatment and prevention
programs are to succeed in reducing the spread of HIV. In addi-
tion, further work is necessary to more fully explore the complex
patterns of sexual networking, particularly among women in rural
areas. Some of this work is under way within the context of the
current project.
16
By design, this study included only women who were not
migrants. This was partly for operational reasons, since tracing
women to many different rural districts would have been logisti-
cally challenging. Nevertheless, it raises important questions about
whether female migrants are at increased risk for HIV infection
and the extent to which nonmigrant, rural women who are infected
became infected as a result of contact with returning migrants as
opposed to contact with men who are resident in the rural com-
munities. The latter question cannot be answered with the available
data, but in a study carried out in a township near Carletonville,
women who were self-identified as being migrants were 1.6 times
more likely (95% CI ⫽1.1–2.3) to be HIV-positive than women
who were self-identified as not being migrants.
22
This study also shows that migrant men were significantly more
likely than nonmigrant men to have casual sex partners and to be
HIV-positive. That more men than expected reported currently
having no casual partners may indicate underreporting or that
casual relationships are of short duration. For women, there was a
marked reluctance—for obvious social reasons, including the fear
of violence—to admit to having additional sex partners. It is likely
that, in keeping with the findings of other behavioral surveys,
23
women in this study underreported the extent of their own sexual
networks. The reluctance of women to speak openly about whether
they had casual relationships—even in qualitative interviews—
has already been documented in this setting.
2,16
For example,
Dladla found that women spoke of others taking on additional sex
partners, although few would acknowledge having done so them-
selves. It is likely that this reluctance would be further exacerbated
in the more formal setting of a survey. Further research and
perhaps the development of additional methods for the study of
female sexual behavior in rural areas are urgently needed to shed
more light on social arrangements that underlie the complex epi-
demiologic patterns identified in this study.
Findings in this study about the age at sexual debut and the
number of lifetime partners were consistent with those of other
South African studies.
24,25
A community-based survey in Carle-
tonville
25
found the age at first sex to be slightly younger (a year
for girls, a year and a half for boys) than in our own study, but this
may be a result of the urban composition of the former study’s
sample. The Carletonville study also found similar high rates of
reported STD symptoms and numbers of lifetime and casual
partners.
25
The high rates of self-reported STD symptoms may highlight a
possible target for intervention strategies. Successful syndromic
management of symptomatic STDs can significantly reduce the
incidence of HIV
19
and should be a central component of HIV
infection prevention programs in this setting.
26
In addition, pre-
sumptive STD treatment among sex workers in some South Afri-
can gold mines has been reported to reduce the prevalence of STDs
among miners.
27
Although this study focused only on male circular migration
within South Africa, from the perspective of two rural health
districts, circular migration is in fact extremely common through-
TABLE 4. Multiple Logistic Regression Model for HIV Prevalence
Among Men and Women
Variable POR 95% CI
Men*
Migration status 0.026 2.65 1.12–6.26
STD symptoms in last 4 mo 0.029 2.09 1.69–7.53
Lived in 4 or more places 0.001 3.56 1.08–4.06
Ever used a condom 0.045 2.18 1.02–4.67
Women
†
No. of lifetime partners ⬎1 0.033 1.26 1.02–1.58
Age ⱕ35 y 0.036 2.46 1.07–5.65
STD symptoms in last 4 mo 0.042 2.29 1.03–5.11
*Hosmer and Lameshow Goodness of Fit statistic ⫽4.7304 with 7
df (P⫽0.693).
†
Hosmer and Lameshow Goodness of Fit statistic ⫽9.2394 with 6
df (P⫽0.1606).
Vol. 30 ●No. 2 155IMPACT OF MIGRATION ON HIV TRANSMISSION IN SOUTH AFRICA
out southern Africa. It is important to recognize, however, that
other types of migration do exist and may play an important role
in facilitating the dissemination of HIV throughout the southern
African region. Further studies that focus on other types of migra-
tion—particularly female migration—are urgently needed.
Finally, the value of HIV prevalence data in isolating risk
factors is limited, given the difficulty of interpreting the complex
temporal relationship. Incidence data from this ongoing cohort
study will therefore be useful in validating the current findings.
The high prevalence of HIV among migrant men indicates that
this group is an appropriate target for focused intervention strate-
gies. At the same time, migrant interventions that concentrate
exclusively at the workplace are likely to have only limited suc-
cess, given that a significant amount of HIV transmission among
rural women occurs irrespective of the migration status of a
woman’s partner. Interventions are most likely to be effective if
they include both men at the workplace and women in rural
communities.
Despite the fact that migrancy is acknowledged to be a major
determining factor in the social conditions in the region,
28
few
studies have explicitly considered the impact that migrancy has on
the health of people, even though the health consequences of
migration may be critical to health outcomes. This study highlights
the importance of migrancy as a risk factor for HIV and probably
other diseases
28
and the need to fully incorporate a good under-
standing of public health in studies on migration.
It is ironic that the lifting of apartheid laws has led to increased
mobility throughout southern Africa and has contributed to the
spread of HIV in the region. However, while migration spreads
disease, it can also be used to spread messages and interventions
that can positively impact on the epidemic. Unless ways are found
to deal with the combined effects of HIV and migration, it is
unlikely that HIV transmission in southern Africa will be substan-
tially reduced.
References
1. Department of Health, South Africa. National HIV and Syphilis
Sero-Prevalence Survey of women attending Public Antenatal
Clinics in South Africa. Pretoria, South Africa: Department of
Health, 2000.
2. Lurie M, Harrison A, Wilkinson D, Abdool Karim SS. Circular mi-
gration and sexual networking in rural KwaZulu/Natal: implications
for the spread of HIV and other sexually transmitted diseases. Health
Transition Rev 1997; 7(suppl 3):15–24.
3. Zwi A, Bachmayer D. HIV and AIDS in South Africa: what is an
appropriate public health response? Health Pol Plan 1990; 5:316–
326.
4. Crush J. Mine migrancy in the contemporary era. In: Crush J, James
W, eds. Crossing Boundaries: Mine Migrancy in a Democratic South
Africa. Cape Town, South Africa: IDASA/IDRC, 1995.
5. McDonald D, ed. On Borders: Perspectives on International Migration
in Southern Africa. New York: St Martin’s Press, 2000.
6. Quinn TC. Population migration and the spread of types 1 and 2
human immunodeficiency virus. Proc Natl Acad Sci 1985; 91:2407–
2414.
7. Decosas J, Adrien A. Migration and HIV. AIDS 1997; 11(suppl
A):S77–S84.
8. Mabey D, Mayaud P. Sexually transmitted diseases in mobile pop-
ulations. Genitourin Med 1997; 73:18–22.
9. Pison G, Le Guenno B, Lagarde E, Enel C, Seck G. Seasonal migra-
tion: a risk factor for HIV in rural Senegal. J AIDS 1993; 6:196–200.
10. Legarde E, Pison G, Enel C. A study of sexual behaviour change in
rural Senegal. J AIDS 1996; 11:282–287.
11. Mbizvo MT, Machekano R, McFarland W, et al. HIV seroincidence
and correlates of seroconversion in a cohort of male factory workers
in Harare, Zimbabwe. AIDS 1996; 10:895–901.
12. Brewer TH, Hasbun J, Ryan CA, et al. Migration, ethnicity and
environment: HIV risk factors for women on the sugar cane planta-
tions of the Dominican Republic. AIDS 1998; 12:1879–1887.
13. Nunn AJ, Hand-Ulrich W, Kamali A, et al. Migration and HIV-1 sero-
prevalence in a rural Ugandan population. AIDS 1995; 9:503–506.
14. Abdool Karim Q, Abdool Karim SS, Singh B, Short R, Ngxongo S.
Seroprevalence of HIV infection in rural South Africa. AIDS 1992;
6:1535–1539.
15. Lurie MN. Migration and AIDS in Southern Africa: a review. South
Afr J Sci 2000; 96:343–347.
16. Dladla N, Hiner C, Qwana E, Lurie M. Rural South African women
talk about their partnerships [abstract ThPeD5523]. XIII Interna-
tional AIDS Conference. Durban, South Africa, July 4–7, 2000.
17. Mkaya-Mwamburi D, Qwana E, Lurie M. HIV status in South Africa:
who wants to know and why? [abstract MoPeC2376]. XIII Interna-
tional AIDS Conference, Durban, South Africa, July 4–7, 2000.
18. Department of Health, KwaZulu/Natal Province, South Africa. Syn-
dromic Management of STDs. Durban: Department of Health, STD
Coordinating Committee, 1995.
19. Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment
of sexually transmitted diseases on HIV infection in rural Tanzania:
randomised controlled trial. Lancet 1995; 346:530–536.
20. Coffee M, Garnett G, Lurie M. Modelling the impact of circular
migration on the rate of spread and the eventual scale of the HIV
epidemic in South Africa [abstract WePpC1393]. XIII International
AIDS Conference, Durban, South Africa, July 4–7, 2000.
21. Lurie M, Williams BG, Sturm AW, et al. HIV discordance among
migrant and non-migrant couples in South Africa [abstract We-
OrD519]. XIII International AIDS Conference, Durban, South Af-
rica, July 4–7, 2000.
22. Zuma K, Gouws E, Williams BG, Lurie M. Risk factors for HIV
infection among women in Carletonville, South Africa: migration,
demography and sexually transmitted diseases. Int J STD AIDS. In
press.
23. Cleland J, Ferry B, eds. Sexual Behaviour and AIDS in the Developing
World. London: Taylor and Francis, 1995.
24. Department of Health, South Africa. South Africa Demographic and
Health Survey Preliminary Report. Pretoria, South Africa: Depart-
ment of Health, 1998.
25. Williams B, Gilgen D, Campbell C, Taljaard D, MacPhail C. The
Natural History of HIV/AIDS in South Africa: A Biomedical and
Social Survey in Carletonville. Johannesburg, South Africa: Centre
for Scientific and Industrial Research, 2000.
26. Harrison A, Abdool Karim SS, Floyd K, Lurie M, Ntuli N, Wilkinson
D. Syndrome packets and health worker training improve sexually
transmitted disease management in rural South Africa: randomised
controlled trial. AIDS 2000; 14:2769–2779.
27. Steen R, Vuylsteke B, DeCoito T, et al. Evidence of declining STD
prevalence in a South African mining community following a core-
group intervention. Sex Transm Dis 2000; 1:1–8.
28. Packard RM. White Plague, Black Labor: Tuberculosis and the Polit-
ical Economy of Health and Disease in South Africa. Berkeley:
University of California Press, 1989.
156 Sexually Transmitted Diseases ●February 2003LURIE ET AL
