Three mathematical models presented at the Eighteenth International
AIDS Conference in Vienna have found that circumcision and a microbicide, used
separately or together, could produce modest reductions in HIV incidence.
The models, by Andrew
Cox of the London School of Hygiene and Tropical Medicine and Hiam Chemaitelly
of Weill Cornell Medical College in Qatar, broadly agreed that a realistic
level of microbicide and circumcision adoption would produce an approximately
20% reduction in new HIV infections within 15 to 20 years. Adding comprehensive antiretroviral
therapy (ART) into this mix would reduce new infections considerably more.
A third model, by Geoffrey Garnett of Imperial College in
London, was cited by Bill Gates
in his conference talk on Monday. This showed that adding in microbicides
and pre-exposure prophylaxis (PrEP) to currently available prevention interventions –
excluding ART – would reduce HIV infections by an additional 15 to 18%.
Andrew Cox’s model used as its baseline the current
situation in Uganda and forecast changes for the next 15 years.
He initially assumed a rise in the proportion of men in the
population who were circumcised would rise from 18%, as it is now, to 48% over
the next ten years, and then stay the same. He assumed that circumcision
stopped six out of ten infections in men. For microbicides, he initially assumed
an efficacy of 60% (rather on the high side for the microbicide in the CAPRISA study) , that women
would use them 80% of the time when they had them, and that they would be
available to 30% of the population after ten years.
Under these circumstances, circumcision (if provided alone) would
produce a 12% decline in new infections in the whole population and
microbicides alone an 8.4% reduction, but they would produce a nearly 20%
reduction if used together.
Various combinations of microbicide and circumcision would
produce a 20% reduction, but one combination cited by Cox was 53% circumcision coverage
combined with 35% microbicide coverage – both higher than his initial
assumptions. A 40% reduction in incidence would require probably unrealistic
coverage rates of 81% for circumcision and 63% for microbicides, and a 30%
reduction in incidence would require 68% coverage of circumcision and 46% coverage of microbicides use.
Inevitably, circumcision would benefit men more and
microbicides would be of greater benefit to women. If circumcision was the only intervention used, 70% of the
reduction in infections seen would be amongst men and if microbicides alone
were used, 65% of the reduction would be in women, but if used together, the
sexes would share the benefit equally.
Hiam Chemaitelly looked at the effect of circumcision in
serodiscordant couples and weighed it up against the impact of treating the
HIV-positive partner; treating the negative partner (as PrEP); and 100% condom
use. Her model did not include microbicides as it was developed before the
CAPRISA result came out, but she commented that a microbicide appeared about as
effective as male circumcision.
She based her modelling on an existing cohort of 1003
serodiscordant couples who are enrolled in a study in the town of Bushenyi in
Amongst the parameters used in a complex model were: a
circumcision efficacy of 58%; that, if supported by voluntary counselling and
testing (VCT), condoms would be 80% effective and would be used 57% of the time;
that PrEP would be 60% efficacious at stopping infections and that adherence to
it would be 72.5%; and that antiretroviral treatment would be 92% efficacious
and that adherence to it would be 80%.
Importantly, she assumed in her initial model that everyone
who was eligible for a specific intervention would be offered it.
Under these circumstances, she found that antiretroviral
therapy (ART) used alone would produce a 69% reduction in infections between
the partners; condoms (with VCT) and
PrEP would reduce new infections by 37%; and circumcision by 19%.
Greater efficacies would result from combining them, thus:
Condoms and ART, 82%
PrEP and ART, 82%
Circumcision and ART, 75%
Condoms and PrEP, 62%
Circumcision and PrEP, 50%
Circumcision and condoms, 49%.
Finally, if all four interventions were used together, the
reduction in infections would be 92%.
Chemaitelly emphasised the importance of condoms being
supported by VCT; if people used condoms without counselling and testing
support, she found much lower rates of efficacy for them. Similarly, if the
other three interventions were in short supply, reductions in infections would
be much lower. Hr model also assumed universal provision of ART to the positive
partner; if it was only provided to people with CD4s under 350 or 200, lower
effectiveness rates resulted.
She pointed out that her model showed that there was no ‘magic
bullet’ in prevention and that impressive reductions in the infection rate
between couples was only achievable by using the range of interventions available in combination.