Statistical model helps predict infection risk in pediatric cancer patients

The health of a pediatric cancer patient is a delicate thing. Not just because of the cancer diagnosis itself, but also because of the various complications that pediatric cancer patients are especially vulnerable to.

Among the most common of these complications is fever. Although a number of different factors can potentially cause fever (inflammation, reaction to medications, pulmonary embolism, etc.), in pediatric cancer patients fever is most often a result of infection. This is because many cancers (and some cancer treatments) can dramatically decrease a patient's white blood cell count.

Fortunately, a study submitted in the Journal of Clinical Oncology by researchers at Northwestern University Feinberg School of Medicine suggests that a new statistical model could accurately predict the risk of bloodstream infections in a subset of pediatric cancer patients.

Identifying feverish pediatric cancer patients who are at a higher risk for bloodstream infections has historically proven difficult. Although guidelines for managing fever in pediatric cancer patients with very low white blood cell counts do exist, no such guidelines for pediatric cancer patients with severely low white blood cell counts exist.

Consequently, pediatric cancer patients with fever may be preemptively treated with antibiotics, which could lead to antibiotic resistance later on. That, along with the potential for bloodstream infections in pediatric cancer patients to progress to sepsis, ultimately motivated the team at Northwestern University to develop and test their new predictive statistical model.

"This model has been designed to delineate bloodstream infection risk in these patients at presentation based on a variety of variables and the overall goal is to reduce unnecessary antibiotic use and also identify patients obviously at high risk for a bloodstream infection," study co-author Jenna Rossof explained in a press release issued by Northwestern University.

To test their model, Rossof and her fellow researchers compiled data on fever episodes occurring in pediatric cancer patients from 18 different academic medical centers. They then compared their model's predictions to the seven-day clinical outcomes in each of the 2,500+ cases and found that the model could accurately predict which patients were more likely to experience bloodstream infections.

"Importantly, the paper showed that in the patients whose predicted risk for bloodstream infections using this model was low, there was a very low rate of true bloodstream infections," Rossoff said. "For those few percent of patients who did have a bloodstream infection, there were no severe outcomes."

Rossof, who is also a member of the Robert H. Lurie Comprehensive Cancer Center of Northwestern University, added that further research could help determine the efficacy of the team's statistical model for pediatric cancer patients who've undergone stem cell transplants and other novel therapies. For now, she is hopeful that the newly published findings will help doctors treat feverish pediatric cancer patients more effectively.

"Fevers are a pretty frequent complication during treatment and when our kids don't need antibiotics, we should be avoiding them to prevent antibiotic resistance and disruption of the gut microbiome," Rossof said. "As much as we can safely – 'safely' being the key word – decrease antibiotic administration, that would be a great thing overall."

More information:
Zhiguo Zhao et al, Prospective External Validation of the Esbenshade Vanderbilt Models Accurately Predicts Bloodstream Infection Risk in Febrile Non-Neutropenic Children With Cancer, Journal of Clinical Oncology (2023). DOI: 10.1200/JCO.23.01814

Provided by California NanoSystems Institute