There is no cure for human immunodeficiency virus, which affects millions of people across the globe. But understanding the sustainability of long-term HIV care and treatment strategies - particularly in low- to middle-income countries where the continuity of care is difficult - is critical to helping those individuals lead long and healthy lives.
Constantin Yiannoutsos, a professor of biostatistics at the Indiana University Fairbanks School of Public Health at IU Indianapolis, has spent much of his career focused on compiling, analyzing and understanding data around HIV and AIDS. He makes sense of data and translates it into actionable information for decision-makers. His work has informed evidence-based decision- and policy-making for some of the most influential organizations around the world, including the World Health Organization.
Constantin Yiannoutsos is a professor of biostatistics at the Indiana University Fairbanks School of Public Health at IU Indianapolis. Photo courtesy of Constantin Yiannoutsos
"So much policy is enacted without evidence and because it's the right thing to do, but the right thing to do is not always the best thing to do," Yiannoutsos said. "Society has limited resources, and so the right thing to do is to help everybody. Our work is allowing people to have the evidence necessary to make the best decision for society in terms of allocating scarce resources most efficiently."
Yiannoutsos, along with IU School of Medicine's Kara Wools-Kaloustian and Aggrey Semeere of the Infectious Diseases Institute at Makerere University College of Health and Sciences in Uganda, leads the East Africa region of the International Epidemiology Databases to Evaluate AIDS. Known as IeDEA, this multidisciplinary consortium of seven international data centers is funded by the National Institutes of Health. The centers collect observational data representing people living with and at risk for HIV within their geographic regions. These data are used to generate large databases that can be merged across regions to address important questions about the scope and impact of HIV around the world.
Now in its 19th year, the East Africa cohort is nearing its $100 million mark in grant funding, including awards from several NIH institutes. Built on IU's success with AMPATH, the Academic Model Providing Access to Healthcare, the cohort operates in Kenya, as well as Tanzania and Uganda. They use data collected during routine health care in those countries and leverage long-standing partnerships with programs and clinicians on the ground in East Africa. The group's expertise in statistics, mathematics and machine learning methods allows them to assess program effectiveness, evaluate the impact of policies and international guidelines and identify groups at highest risk for disengagement from care and adverse health outcomes.
The team also collaborates with partners at the other regional IeDEA centers to collate data globally. They hope to provide critical information to clinicians, programs, governments and international organizations to support their long-term strategies for the treatment and care of individuals with HIV.
"We want to make the world a better place for people who live in some of the most destitute countries around the world," Yiannoutsos said. "These are countries with unbelievable poverty, and having HIV and dealing with stigmatization on top of that is really difficult for people. We want to improve their condition any way we can."
One challenge to the use of observational data is that biases can be easily introduced due to gaps in the data collected through routine patient care. Eliminating or adjusting for potential biases is at the center of Yiannoutsos' research, and specifically ones that occur in low- to middle-income countries that lack the infrastructure to document patient deaths.
A common example is program clients who no longer attend clinic visits and are no longer present in the data. Because they are commonly at the highest risk for adverse outcomes, estimates of program effectiveness would be over-estimated if conclusions were based only on those who remain in care. Yiannoutsos and his colleagues have developed pioneering statistical methods, based on strategically augmenting available data with surveys of dropouts, to adjust the naïve estimates of patient outcomes that would otherwise be generated.
Eliminating biases is critical to ensuring that the data provided to decision-makers across the globe is accurate. If, for example, the data shows an incorrect amount of people with HIV in a specific region, a country may end up without enough medicine for those who need it. This is particularly the case with respect to pediatric formulations of HIV drugs. IeDEA, and Yiannoutsos in particular, have been advising international organizations on the number of children living with HIV who need these medicines to control their infection. These estimates have been included in meetings sponsored by the Vatican in its efforts to support children with HIV.
Yiannoutsos and his colleagues work primarily with UNAIDS, the joint United Nations Programme on HIV/AIDS, which uses mathematical modeling to inform policy worldwide. IeDEA has also provided data on patient mortality and continuity of care to the World Health Organization. That data informed updated guidance known as "test and treat" in 2015, which provides immediate HIV care to all individuals as soon as they are diagnosed.
"Updating guidance on such a large scale costs billions and billions of dollars, so having some semblance of evidence to inform those decisions is very important," Yiannoutsos said. "In low- to middle-income countries, there were often no people, no training, no systems and no drugs in place. We have to prove that the benefits outweigh the challenge."
With artificial intelligence and machine learning methods maturing, Yiannoutsos and his team are looking to expand the biostatistics research they do, ultimately creating new systems that will improve patient management and maximize the impact of limited resources. They hope to one day develop simple algorithms that run on a phone or tablet that will alert providers, for example, when people most at risk of dropping out of care enter a clinic. Providers could triage them immediately, even going into their communities to give the necessary medication and eliminating the stigma of needing to enter the HIV clinics.
"Low- to middle-income countries are pressured to treat and manage chronically millions and millions of people," Yiannoutsos said. "We want to figure out how to do that by optimizing their limited resources. We are at the cusp of doing really important work, and I am so proud to be involved in it."
