Courtesy of the School of Medicine
The National Institute of Health recently awarded a $2.4 million Research Project Grant, known as an R01, to researchers at the Yale School of Medicine and Yale School of Nursing to study the incidence of fragility fractures in patients with human immunodeficiency virus, or HIV.
The two principal investigators for this research are Evelyn Hsieh, assistant professor of medicine, and Julie Womack, associate professor of nursing. Together, they will lead a team of medical researchers, clinicians, computer scientists and statisticians in analyzing data from the Veterans Aging Cohort Study of HIV-positive patients, according to Hsieh. The goal is to create a prediction model that takes into account different patients’ medical histories and outputs a risk assessment for bone fractures.
“Falls and fractures are something everyone who is aging has to think about,” Amy Justice, professor of medicine and professor of public health, said. “Everyone has to deal with bone fragility as they get older, but HIV is almost a model for aging with vengeance.”
According to Justice, there are two main categories of determinants when considering a person’s risk factor for falls and fractures. There are clinical factors such as the patient’s bone structure, bone density and ability to maintain balance, as well as environmental factors such as the patient’s lifestyle, medications and exercise regime. Genetics are only a small part of tailoring a patient’s care in order to avoid adverse health outcomes, Justice added.
HIV, as an autoimmune deficiency disease, influences the severity of these clinical factors due to chronic inflammation and reduced immune function. HIV patients also spend decades taking several medications, all of which have side effects that can affect balance or bone fragility. One goal of this study is to use the database from the Veterans Aging Cohort Study, which has 60,000 HIV-positive patients and 120,000 HIV-negative patients, to determine the risk of fractures in HIV-positive patients, Justice said.
“Patients with HIV at baseline seem to have higher risk for osteoporosis compared to those without HIV, and there are different reasons for that,” Hsieh said. “Some studies suggest that increased inflammation … which is common in patients with chronic infections, is linked to impact on the bone.”
The incidence of fractures are difficult to study with small data sets, and they become even more difficult to pinpoint as the study population’s age decreases. To look at fractures within a younger population, researchers need an even larger data set than normal to notice a true risk for fractures, added Hsieh.
According to Justice, one of the unique facets of this study is its interest in identifying and managing risk for middle-aged patients, especially those living with HIV. In their grant application, Hsieh and Womack mentioned the fact that bone compromise begins in middle age, yet most of the studies concerning falls focus on patients over the age of 65.
“One of the questions we want to answer is what are all the factors that should be considered when predicting risk for a patient with HIV?” Hsieh said. “How important are HIV medications in this picture?”
Another feature of the study is the diversity in the data set’s sample, according to Hsieh, Womack and Justice. They said that the 180,000 veterans included in the study reflect the U.S.’s population more accurately than any other data set in terms of racial diversity. This has been an issue in past studies on falls that focused primarily on white patients over the age of 65. The researchers hope that using a data set that includes Black and Latinx patients will lead to more accurate findings and better health outcomes for those not typically represented in the samples.
Womack also mentioned the variety of expertise within the research team’s personnel. According to her, medicine, computer science and data science experts will come together to analyze unstructured data from clinical notes and identify outcomes of interest. The breadth of the research project will also include qualitative data, such as information on alcoholism or mental health, thereby affording a much broader base from which to draw information.
Samah Jarad-Fodeh, assistant professor of emergency medicine and of biostatistics, is one of the computer scientists on the team. Her goal is to take the data set and create a predictive model that will eventually be applicable to any patient.
“What I will be doing is natural language processing,” Jarad-Fodeh said. “We need to identify the language, and, once we have this lexicon of terms, we will use it to automatically measure the risk.”
In this case, “language” refers to a specific lexicon that will be used to code the model. In a previous similar study focused on pain, the language Jarad-Fodeh developed included vocabulary such as intensity, persistence and aggravators. The goal is to develop a similar lexicon for the case of bone fragility where each word is a specific class of risk factor, added Jarad-Fodeh.
Once the model can accurately predict the risk for the study’s data set, it will be employed at the Yale New Haven Hospital and Boston Medical Center to test for accuracy. If this testing phase is successful, then the model can then be used on a larger scale, according to Womack.
A similar risk assessment tool already exists for osteoporosis but does not work perfectly for HIV-positive patients. The bone fragility model is meant to help clinicians decide on appropriate treatment for their patients and also to help patients make changes to their lifestyle based on their risk factors.
The five-year study began in July of this year and will run through June of 2026.