Mathematical modelling and its potential to inform MPT development
Published: July 18, 2018
One of the ongoing challenges in the MPT field is that, because MPT products are not yet commercially available (aside from male and female condoms), we do not have ‘real world’ data to inform product development, introduction, and investment decisions. Given this, MPT product developers, researchers, and funders must grapple with critical questions, such as: How impactful do these products need to be for me to consider funding or starting a product development program? Is that impact potential achievable, and under what conditions? How efficacious would the product need to be? What is the market for the product? How many people would need to use the product correctly and consistently? What associated costs will make a product cost-effective or cost-saving?
Particularly in the limited resource context of global health research and development, mathematical modelling can provide estimates and projections to help answer some of these questions. Mathematical models combine real data and assumptions to create a representation of a system. In a global health system, a model can be used to estimate the population-level impact of an intervention under certain conditions, and inversely, what conditions need to be met to achieve a given population-level impact. In the MPT field, some of this work is already underway. The IMPT Secretariat recently conducted a landscape review, including a literature review and key informant interviews, of MPT-relevant modelling projects. The small but trailblazing body of work focuses on the context of MPTs in South Africa and reports estimates on potential MPT demand, cost-effectiveness, and health impact that could support the field’s progress.
Two discrete choice experiments conducted in South Africa suggest that many women want to use MPTs and would prefer to use an MPT over a product that only protected against HIV. The first, led by the London School of Hygiene and Tropical Medicine (LSHTM), used discrete choice survey data collected in 2005 to estimate relative preference between products (i.e., microbicide, diaphragm, and female condom), effectiveness against HIV, effectiveness against pregnancy, price, and interactions between product and effectiveness. Notably, the study predicts that the addition of pregnancy prevention to a microbicide product could increase the uptake of that product by up to 65% assuming high product efficacy. A second discrete choice experiment, also led by LSHTM, used survey data collected in 2015 among adolescent and adult women as well as female sex workers to estimate relative preference between product attributes. The discrete choice model estimates that demand for HIV prevention products would increase among all groups if products also include STI and/or unintended pregnancy protection.
Building from this demand, another model suggests that MPTs could have meaningful public health impact under certain epidemiological, product uptake, and product efficacy conditions. Imperial College London developed a complex contraceptive and HIV preventive intravaginal ring (IVR) model for South Africa that compartmentalizes HIV transmission, fertility, contraception, and pregnancy outcomes, and then stratifies estimates by age, sex, behavioral risk group, male circumcision status, ART use, and MPT use. The model generated results estimates based on assumptions of efficacy through perfect product use (estimated through single-indication product trial data as 56% – 75% efficacy for HIV and 92% – 97% efficacy for pregnancy) and 10% uptake among youth aged 15 – 19 years and other women at risk. With this modest uptake, the model projects that with this MPT IVR, 4.5% and 3.66% of new HIV infections could be prevented among youth and other women at risk respectively, 4.41% and 1.43% of unintended pregnancies, and 4.8% and 1.81% maternal deaths averted.
The same MPT IVR model from Imperial College London also suggested that these products have the potential to be cost-effective and even cost-saving if low cost and high efficacy can be achieved. Given the status of this product’s development, realistic unit costs are not well understood for MPT IVRs. However, from estimated cost-effectiveness thresholds of $8,909 per disability-adjusted life year (DALY) for cost-effective and $2,221 per DALY for very cost-effective, MPT IVR developers can set reasonable targets for unit costs in the context of the estimated efficacy described above. Notably, the first cost-effectiveness estimates to be published on a range of candidate MPT products, led by the LSHTM and based on its discrete choice experiment described above, were published after the IMPT’s landscape assessment. This work suggests that MPTs that combine HIV prevention and contraception could be more cost-effective than single-indication HIV prevention products among younger women and female sex workers.
As the MPT field advances, assumptions around product uptake, adherence, and costs will be better defined, but the foundation set by the existing body of MPT modelling work is an important step forward. It would be useful for future models to clarify the impact of other potential MPT product configurations and in other geographic settings. See the full IMPT Secretariat landscape review report and a live webinar recording for more information on this emerging space of MPT modelling.
This Secretariat would like to thank Chelsea Polis for her critical contributions to the content of this post.