Abstract Book Vol. 2 [International Conference on AIDS (14th: 2002: Barcelona, Spain)]

450 Abstracts ThPeC7450-ThPeC7453

XIV International AIDS Conference

system plus related data from the blood bank and HIV and AIDS case reporting
system were analyzed to extract inputs for AEM. The outputs generated included
new, cumulative and current HIV/AIDS/death and routes of transmission for each
year from the beginning of the epidemic to year 2010.
Results: The projections showed there were approximately 240,000 cumulative
HIV infections as of 2000: 190,000 people living with HIV, 10,500 new infections,
20,000 new AIDS cases and 17,000 AIDS deaths. Changes in sexual behaviorespecially increased condom use in commercial sex-have slowed the epidemic
enormously. If these reduced risk behaviors are maintained, the prevalence of
HIV will continue to decline, with projected infections for years 2005 and 2010 of
115,000 and 47,500 respectively. New AIDS cases will exceed 20,000 a year during the early 2000s, gradually decreasing to around 10,000 new cases in 2010.
While most new infections in the early part of the epidemic occurred among clients
visiting female sex workers, the major transmission route after 1997 was husband
to wife and then mother-to-child transmission from 2001 forward.
Conclusions: The model supports the role of behavior change in producing
prevalence declines in Cambodia. Sustaining sex work prevention efforts remains
essential, but expanded strategies to address increasingly important transmission
modes such as husband to wife and mother to child are urgently needed.
Presenting author: Hor Bun Leng, c/o Family Health International/Cambodia, 14,
Street 278, Boeng Keng Kang I, Sangkat Chamkarmon, Phnom Penh, Cambodia,
Tel.: [phone], Fax: [phone], E-mail: [email protected]
ThPeC7450 Patterns of sexual behaviour and the risk of HIV
infection in rural Zimbabwe
J.J.C. Lewis, S. Chandiwana, C. Nyamukapa, G.P. Garnett, C.A. Donnelly,
S. Gregson. Imperial College of Science, Technology and Medicine, Department
of Infectious Disease Epidemiology, Imperial College of Science, Technology and
Medicine, St Mary's Campus, Norfolk Place, London W2 1PG, United Kingdom
Background: HIV-1 has spread widely throughout rural Zimbabwe reaching unprecedented prevalences. A wide range of social, economic, demographic and
behavioural variables contribute to the spread of HIV within populations. Detailed
quantitative data on patterns of risk are necessary to understand the course of
this rural epidemic.
Methods: A cross-sectional survey of almost 10 000 people in rural Zimbabwe
was undertaken from July 1998 to January 2000. Quantitative data was collected
on both current and long-term sexual behaviour, HIV/AIDS awareness and a large
variety of background characteristics.
Results: Univariate and multivariate logistic regressions for risk factors for HIV
infection show that the number of lifetime partners is a key predictor of infection
(p-value < 0.001 in all cases). Additionally, the number of years of sexual activity is positively correlated with HIV infection after controlling for age for men and
some of the short-term behavioural measures are negatively correlated with HIV
infection for women. When questioned about their level of HIV awareness, 42.9%
of respondents thought that "... all people infected with HIV (could) be identified by
looking at them". A Mann-Whitney test suggested that this was positively related
to the number of new partners over the last 12 months (p-value = 0.009).
Conclusions: The sexual behaviour variables that reflect a long period of exposure are the most reliable predictors of HIV infection. The respondents who
thought they could identify infected people on sight seem to have a higher partner acquisition rate and provide a target for future educational programmes.
Presenting author: James Lewis, Department of Infectious Disease Epidemiology, Imperial College of Science, Technology and Medicine, St Mary's Campus,
Norfolk Place, London W2 1PG, United Kingdom, Tel.: [phone], Fax:
[phone], E-mail: [email protected]
ThPeC7451I Who infects who? HIV discordance and the
directionality of spread among migrant and
non-migrant couples in South Africa
M.N. Lurie1, B.G. Williams2, G.P. Garnett3, D.M. Mkaya1, M.D. Sweat4,
N. Dladla1, J. Gittelsohn4, S.S. Abdool Karim5. 1Medical Research Council, 164
Summit Aye, Providence, RI, 02906, South Africa; 2 World Health Drganisation,
Geneva, Switzerland; 3lImperial College of Medicine, London, United Kingdom;
4Johns Hopkins University, Baltimore, United States; 5 University of Natal
Durban, South Africa
Background: Migration has played an important role in the spread of HIV in
South Africa. It has long been assumed that the direction of spread is largely
from returning male migrants to their rural partners, but few studies have collected data that allows for an analysis of the directionality of spread in the context
of migration.
Methods: A cross-sectional study with male migrants at two common migration
destinations, their rural partners, and a group of non-migrant couples, with behavioural and biological outcomes. A mathematical model was constructed to
estimate the likelihood of men and women being infected from within primary re
lationships versus from outside.
Results: 69.6% of couples were negatively concordant for HIV, 9.5% were positively concordant and 20.8% were discordant. Migrant couples were more likely
than non-migrant couples to have one or both partners infected (35% versus 19%;
p = 0.026; OR=2.28) and to be HIV-1 discordant (27% versus 15%; p=0.066;
OR=2.06). In 71.4% of discordant couples, the male was the infected partner;
this did not differ by migration status.

In the mathematical model, migrant men were 26 times more likely to be infected
from outside their regular relationships than their female partners (RR=26.3;
p=0.000); while non-migrant men were 10 times more likely to be infected from
outside their regular relationships than from inside (RR=10.5; p=0.00003).
Conclusions: Migration continues to play an important role in the spread of HIV-1
in South Africa. The direction of spread of the epidemic is not only from returning
migrant men to their rural partners, but also from women to their partners. Prevention efforts will therefore need to target both migrant men - who are clearly at
very high risk of infection - and also women who remain at home.
Presenting author: Mark Lurie, 164 Summit Ave, Providence, RI, 2906, United
States, Tel.: [phone], Fax: [phone], E-mail: mark lurie@brown.
edu
ThPeC7452 Individual and network characteristics associated
with return of syringes to a syringe exchange
program
M.E. Shepherd', S.G. Sherman', T.W. Valente2, R.S. Brookmeyer', D. Vlahov3,
J.C. Bareta1, S. Huettnerl, S.A. Strathdeel. 'Johns Hopkins Bloomberg School
of Public Health, Johns Hopkins Bloomberg School of PH, 615 N. Wolfe St, Rm
W4510, Baltimore MD 21205, United States; 2 University of Southern California,
Alhambra, United States; 3New York Academy of Medicine, New York, United
States
Background: Minimizing the circulation time of used syringes amplifies the preventive benefits of syringe exchange programs [SEPs] at the population level. We
identified individual and network factors associated with return of used syringes
to offer insight into optimizing the impact of SEP in Baltimore, USA.
Methods: All SEP syringes distributed from 08/94 - 02/97 were barcoded. Information obtained at SEP registration, including the results of a voluntary HIV test,
were linked to barcoded syringe data to identify individual and network characteristics of those more likely to return syringes. Syringe circulation time was modeled
using Cox proportional hazards regression analysis.
Results: 746029 barcoded syringes were distributed and 197216 barcoded and
512801 non-barcoded syringes were returned to the SEP during the study period.
The median circulation time was 10 days(range 0-899). Syringes more likely to be
returned were those obtained by: women (Adj Hazard Ratio [AHR]=1.37;p<.001),
clients >34 years old(AHR=1.14; p<.001), those residing near an exchange site
(AHR=1.19;p<.001), those injecting for >25 years (AHR=1.17;p<.001), those
who began injecting >17 years old(AHR=1.09; p<.001). Additionally, syringes
were more likely to be returned if they were obtained by secondary exchangers
(AHR=1.68;p<.001), and those with direct (AHR=1.43;p<.001) network ties to
secondary exchangers. Syringes acquired by HIV+ exchangers were less likely
to be returned (AHR=0.86;p<.001), as were those distributed to more frequent
heroin injectors (2x/day or more: AHR=0.93; p<.001).
Conclusions: Interventions to improve syringe return are needed, especially
among high risk and HIV+ injectors. Secondary exchangers play a key role in
reducing syringe circulation time, which in turn attenuates epidemics of HIV and
other blood-borne pathogens. Syringe disposal systems are also needed to reduce the circulation of potentially contaminated syringes in the community.
Presenting author: Mary Shepherd, Johns Hopkins Bloomberg School of PH,
615 N. Wolfe St., Rm W4510, Baltimore MD 21205, United States, Tel.: +410 592
3382, Fax: [phone], E-mail: [email protected]
ThPeC7453 HIV-1 population genetics in human populations
with structured mixing
S.M. Goodreau. University of Washington Center for AIDS and STD, Seattle,
WA, United States
Social network analysis has proven successful in uncovering previously overlooked patterns of sexual behavior that have major effects on the trajectory of
an HIV or STD epidemic. I extend this line of work into the question of HIV microevolution, the pattern of neutral genetic diversity that arises among seroposifive individuals in an infection network. Most population genetic methods for analyzing this variation make the untenable assumption that the partnership networks generating it were random. I present a general framework for simulating
viral spread and evolution in populations with structured mixing. This framework
involves actor-based microsimulation, a necessary component when tracking viral evolution that is so rapid as to create unique genetic sequences among every seropositive individual. It is also dynamic in that individual partnerships form
and dissolve over time. The framework is applied to five mixing models: purely
random, assortative within two clusters, assortative within eight clusters, divided
into core and periphery, and divided into sex workers, clients, and clients' wives.
The same framework is applied to data on injection needle sharing in Colorado
Springs, US. The model populations yield significant differences in prevalence
that are in agreement with previous studies; core and sex-worker populations
yield higher prevalence than random networks with the same mean rate of con
tact, and clustered populations yield lower. Prevalence in the Colorado Springs
network is strongly dependent on the position in the network where infection begins. Once prevalence is controlled for, the models also display significantly different patterns of genetic diversity; the core, sex-worker and Colorado Springs networks have lower diversity than a random population with the same prevalence,
and clustered models have higher. I discuss possible applications of this work to
the reconstruction of transition dynamics and to drug resistance monitoring.