Bridging the Gap: Conference Record [Abstract book, International Conference on AIDS (12th: 1998: Geneva, Switzerland)]

12th World AIDS Conference

Abstracts 43483-43487 941

43483 Sex worker intervention project becomes replicable
model and leads to plateauing in Mumbai
Vinay Vasta', I.S. Gilada', S.I. Gilada1. 'Indian Health Orgn. Muncipal School
Bldg. J.J. Hospital Compd. Bombay-400008 MH, India
Background: After initial experience of working among CSWs for STD control
and their empowerment since 1982, prior to HIV/AIDS awareness or detection in
India, we started intervention project termed as 'Saheli Project' in 1991, based
on peer education principle. CSWs have been traditionally blamed as reservoirs
of STDs/HIV and infecting the community. With the objective to study the impact
of our ongoing project different hard and soft parameters have been looked into.
Methods: Three tier of peer-leaders to cover hierarchy of the CSWs, a Saheli
(friend) for 25 SWs; Tai (sister) to supervise 10 sahelies and a Bai (mother) to
oversee 10 tais, from the brothel owners were chosen and trained. Initially they
worked for distributing condoms and educating their peers about STDs/HIV and
client negotiation. Leaders with aptitude and interest were selected and trained
as Health Care Givers. Making 1991 a baseline, parameters of number of SWs,
their clients, condom usage rate and HIV prevalence are compared with 1997.
Results: Saheli project has now gone beyond its initial goal. STDs have come
down by 75%, clients reduced by 60%, abortions have been minimal, condom
usage has gone up by 1900% among sahelies and 1200% among control group,
HIV case doubling rate widened. Sahelies have evolved into educators, Health
Care Givers and a force to reckon with for AIDS control and role model for others.
('Transmission possibility of 0.2% from female to male.)

SWs No. HIV+ Clients/ Total
SW/day clients
1991 100,000 32000 (32%) 5 500,000
1997 (total) 60,000 42,000 (70%) 3 180,000
1997 (saheli) 5,000 2,100(42%) 3 15,000

Condom
usage
25,000 (5%)
126,000 (60%)
14,250 (95%)

HIV
infections/day
304
102
3

Conclusion: Contrary to popular belief, HIV infection rate is plateauing in
Bombay for 2 years. This has been made possible because of HIV control among
CSWs and their extended role as health educators and trainers. Saheli Project
has emerged as a formidable strength and role model for others in India and its
neighbouring countries. If replicated elsewhere without any delay, India could be
saved from a major disaster in the ofing.
43484 Incorporating age effects and case definition change into
empirical Bayesian back-calculation (EBBC): European
Union (EU) action on multinational AIDS scenarios
Angela M. Downs', S.H. Heisterkamp2, L. Rava3, H. Houweling4, F.F. Hamers',
C. Rossi5, J.C. Jager4. 'European Centre for Monitoring of AIDS, 14 Rue Val
D'osne 94415 St. Maurice Cedex, France; 2University of Amsterdam,
Amsterdam; 4Nat. Instit. of Public Health (RIVM) Bilthoven, Netherlands;
3Reference Centre for AIDS, Rome; 5University For Vergata, Italy
Objectives:To improve EBBC in Europe by incorporating a minimum age at
infection, age-specific disease progression and effects of the 1993 revision of the
AIDS case definition, and to provide estimates by birth cohort.
Methods: EBBC used Markov disease progression models fitted to Dutch
cohorts of homo/bisexual men (HBM) and injecting drug users (IDU); the stage
prior to AIDS was subdivided to allow development of 1 of the 3 new diseases
in the revised definition at rates depending on the proportion of AIDS cases
satisfying only the new definition. The IDU model allows for pre-AIDS death. The
HBM model was used for heterosexually infected persons (HET). Using AIDS
data through 1995 (i.e. prior to effective therapies), EBBC was implemented in
45 groups (HBM, IDU, HET in 15 EU countries): 1) with no age effects (BC1), 2)
by 5-year birth-cohort with minimum age of 13 at infection (BC2) and 3) as BC2,
with age-dependent progression rates (BC3).
Results: In 16 groups with >1000 AIDS cases, estimates of the early HIV
curve and year of peak infection (if any) were similar in all 3 BC versions. Peaks
were generally lower, but subsequent declines in incidence slower, in BC2 and
BC3 than in BC1, so that cumulative HIV incidence in BC2 and BC3 progressively
surpassed that in BC1; BC3 usually gave higher estimates than BC2. No clear
peaks were estimated among HBM in Italy, IDU in Portugal or HET anywhere.
Within each transmission group, patterns by birth-cohort showed marked similarities between countries: the 1960-64 birth-cohort was generally most affected,
but more recent infections occurred predominantly among those born since 1964.
In smaller groups, much greater differences were found between BC versions:
both BC2 and BC3 yielded unrealistically high HIV incidence in the early years
and loss of peaks found by BC1, reflecting the inevitable predominance of the
adopted neighbour prior when data become too sparse.
Conclusions: EBBC by birth-cohort can be particularly useful to detect continuing transmission in younger cohorts when overall HIV incidence is declining.
Feasibility depends on the size and stage of the epidemic. For homo/bisexual
men, IDU and heterosexually infected persons in EU countries, 1000 cases usually sufficed; age-specific progression rates are recommended (BC3). For smaller
groups, version BC1 (no cohorts) is preferred. Any future EBBC will also need to
incorporate the effects of new therapies delaying disease progression.

S43485 Parametric and non-parametric estimation of the mean
ages of the vertically infected HIV survivorship and
cleansing the maternally derived antibody against
anti-HIV
Hyun Mo Yang, Y.C. Lian, M. Della Negra, W. Sao Paulo, A.P.V. Gomes. Instituto
de Infectologia Emilio Ribas Sao Paulo, Brazil
Objectives: The estimation of the mean age of cleansing the maternally derived
antibodies against HIV, and the mean survival age of HIV infected children.
Design: Retrospective study.
Methods: We developed a parametric probability distribution based on Poisson
process. The resulting probability, which is in function of age, is parametrized in
terms of the mean age a (survival or cleansing) and the degree of the heterogeneity /. This probability function is similar to the Fermi-Dirac distribution function. We
assessed both the mean age and heterogeneity degree by applying the maximum
likelihood estimation method, based on a sample of 583 vertically exposed to
HIV followed up children, collected from December, 1985 until September, 1996.
We compared the parametric estimation of the mean age with the Kaplan-Meier
non-parametric approach. We presented only the estimation of the parameters
based on the raw data. The other estimations of the parameters a and P can be
done by subdividing the entire population with respect to gender, maternal risk
and drug treatment.Results: The fitted parameters by the likelihood method are:
a = 11.5 ~ 1.4 months and / = 3.65 ~ 0.18 months, for serorevertion, and a =
69.5 ~ 9.7 months and / = 28.9 ~ 1.8 months, for survivorship. The Kaplan-Meier
estimation of the mean age are:, = 11.1 months, for serorevertion, and a = 91.2
months, for survivorship.
Conclusion: Both the parametric and non-parametric estimation methods provided us with the same mean age of cleansing the maternally derived antibody
anti-HIV, but the latter estimation resulted in a greater mean survival age of HIV
infected children.
43486 Partial herd immunity in a high-prevalence HIV
subcommunity: "Firewalls of seropositives" as a
protective network factor
Samuel R. Friedman', B. Jose', A. Neaigus', D.C. Des Jarlais' 2, R. Curtis'.
SNDRI, 2 World Center 16 fl. New York, NY; 2Beth Israel Medical Center, New
York, NY USA
Objectives: In many cities, HIV prevalence among IDUs (or other communities
at risk) stabilizes at high levels (30%+). Insofar as we are aware, no such stabilization period has been followed by a large later increase, for example from
40% to 60%. Among New York City IDUs, seroprevalence was stable at about
50% from 1982-91 without evidence of a subsequent breakout. Both high viral
loads among newly-infected persons and epidemic modeling results suggest that
primary-infection-based epidemic breakouts can occur. Question: Can social network structures help explain the lack of primary-infection-driven later breakouts"?
Methods: 767 street-recruited IDUs were interviewed about their networks in
1991-1993; 687 were tested for HIV. Two subjects are considered 'linked' if either
named the other as someone he/she had injected drugs with or had sex with in the
prior 30 days and if this was confirmed by face-to-face contact, field observation,
or computer matching of personal descriptors. A 'network' is defined as a set
of subjects linked to each other directly or indirectly in a chain-linkage. ANALYSES: 1. Risk behavior by seronegatives was assessed. Network analyses, using
UCINET software, assessed: 2. Was there a large network that might have rapidly
spread HIV across the IDU community? Was it 'saturated'? 3. Are subnetworks
of linked seronegatives (defined by ignoring their links to seropositives) small
enough to limit subsequent primary-infection outbreaks (since seropositives are
not susceptible to primary infection)?
Results: 1. 36% of seronegatives had injected with someone else's syringe in
the prior 30 days; 36% had passed a used syringe to another IDU. 2. There was
a large network of 230 members; HIV prevalence in it was 45%-which raises
the question of why it is so far from network saturation. 3. In the seronegative
subset, the largest network had 18 members; the next largest 7 members; and 2
networks had 6 members each. Even if all untested subjects were HIV-negative,
the largest seronegative sub-network would have only 39 members.
Conclusions: If network sizes have not decreased, a single new infection
in a large network early in the epidemic could precipitate rapid HIV spread to
many of its other 229 members (and their unmeasured contacts). Independent of
behavioral risk reduction, 'firewalls' of seropositives (not susceptible to primary
infection) may limit later HIV spread within large networks to far smaller 'islands
of seronegatives.' Still needed: multicity studies of generalizability & simulation of
how these processes contribute to stabilization.
43487 Joint use of empirical Bayesian back-calculation and
Mover-Stayer model to estimate the size of the HIV/AIDS
epidemic in Italy
Lucilla Rava1, A.M. Downs2, S.H. Heisterkamp3, H. Houweling4,

C. Pasqualucci5, C. Rossi5, G. Schinaia6. 'Ctr. Rif AIDS IRCCS L. Spallazani
via Portuense 292 00149 Rome; 5University of Rome Tor Vergata, Rome;
6University of Rome La Sapienza, Italy; 2European Ctr AIDS Monitoring,
Paris-Saint Maurice, France; 3University of Amsterdam, Amsterdam; 4Rivm,
Bilthoven, The Netherlands
Study for the EU Concerted Action on Multinational AIDS Scenarios
(BMH1-CT94-1723).