VOL. 133 | NO. 19 | Thursday, January 25, 2018
Study Finds Black Children More Likely To Die Post-Surgery Than White Children
By Andy Meek
A research team from the University of Tennessee Health Science Center and Le Bonheur Children’s Hospital has published a study that found black children are more than twice as likely to die after surgery than white children.
The team zeroed in on those findings after developing race-specific models to predict surgical outcomes. Their study is also set to be published in the February issue of the publication “Pediatrics.”
Dr. Oguz Akbilgic, an assistant professor at UTHSC with an expertise in statistical modeling and machine learning, was a corresponding author on the study, which he said has already led to a next step of looking for patterns in the data that can be used to help develop interventions.
“Previously, we were trying to come up with a clinical-decision support system that can help clinicians, surgeons, better evaluate children before they undergo surgery, to see whether there’s a risk of death after surgery,” Akbilgic said. “So we came up with a model, a very simple, easy-to-use decision model.”
It included a series of basic yes-or-no questions, like is the child receiving oxygen support? Is there ventilator dependency? Does the child have sepsis? And based on the answers to those questions, a risk of death was identified on a sliding scale.
Something materialized, however, in data from that model. In every risk category, according to Akbilgic, the risk of death was overestimated for white children and underestimated for black children.
“For example, in the highest risk category, the risk of death was 38 percent,” he said. “But when we look at specific African-American children, the risk of death was actually 42 percent, and for white children it was 34 percent. Then we said, ‘OK, apparently the dynamics of the risk of death work differently for white and black children.’
“So we decided to do separate studies, separate risk models, for African-American and for white children. And we showed that race-specific risk models work better than non-race specific models.”
For the just-released study, he worked with Dr. Max Langham, a pediatric surgeon at Le Bonheur Children’s Hospital and professor of surgery and pediatrics vice chairman of the Department of Surgery at UTHSC, and Dr. Robert Davis, the governor’s chair and professor and founding director of the UTHSC-Oak Ridge National Laboratory Center for Biomedical Informatics.
They examined 30-day mortality rates post-surgery for more than 260,000 surgical procedures performed for children from 2012 to 2015.
The data were compiled from the American College of Surgeons’ National Surgical Quality Improvement Program database and include patients from children’s hospitals from across the nation.
Among the findings were a higher occurrence of risk factors for poor outcomes in black children, risk factors that include ventilator use, oxygen support, wound infections, transfusions and neonatal status.
Dr. Jonathan McCullers, pediatrician-in-chief at Le Bonheur, said that data and models surrounding the study will be critical tools to identity interventions that could hopefully eliminate “some, if not all, of the disparities that arise from the poverty and limited access to comprehensive care for which black race is a marker, not a cause.”
The study authors said that the use of race-specific models could more accurately identify patients at high risk for death following surgery as compared with models that examine all races grouped together. They also suggest that interventions to decrease the risk of death after surgery should be tested within the context of race-specific risk classifications.
“Right now, we cannot tell why African-American children are at higher risk,” Akbilgic said. “Our next step is what we’re doing right now to analyze notes in the electronic medical records to see whether we can find some socioeconomic risk factors, key words that are associated with high mortality in African-American children. Our next step is to find these causality patterns so we can develop interventions.”