Skip to main content

20-06-2024 | Pediatric Cancer | Editor's Choice | News

Phoenix Sepsis Score supported for assessing infection risk in children with cancer

Author: Lynda Williams


medwireNews: Children undergoing cancer treatment who have a suspected infection should be assessed using the Phoenix Sepsis Score, suggest study findings indicating the model can accurately predict the risk for definitely attributable mortality in this patient population.

Joshua Wolf (St Jude Children’s Research Hospital, Memphis, Tennessee, USA) and co-workers compared the newly developed Phoenix Sepsis Score for children with cancer with standard assessments, namely Pediatric Risk of Mortality 3, the Pediatric Sequential Organ Failure Assessment (pSOFA), the quick SOFA, and the Paediatric Logistic Organ Dysfunction 2.

The Phoenix Sepsis Score determined the risk for attributable or definitely attributable mortality from sepsis on the basis of life-threatening cardiovascular, respiratory, neurological, or coagulation dysfunction in 143 children who were admitted to the intensive care unit (ICU) at the St Jude Children’s Research Hospital between 2013 and 2019.

At the time, the children (58% male, median age 10.3 years) were undergoing treatment for hematologic (56.6%) or solid malignancies, and had not received hematopoietic stem cell transplantation as their latest therapy.

During 171 identified episodes of suspected sepsis, 9.9% of patients died and 21.6% had a prolonged ICU stay of at least 7 days; the deaths were attributable or definitely attributable to sepsis in 9.9% and 7.6% of patients, respectively.

Analysis demonstrated a significant correlation between the 24-hour Phoenix Sepsis Score and attributable mortality, definitely attributable mortality, all-cause mortality, and the need for a prolonged ICU stay, the researchers report in a research letter to JAMA Network Open.

The area under the receiver operating characteristic curve for the Phoenix Sepsis Score was numerically better at predicting both attributable and definitely attributable mortality than the other models, they continue.

Specifically, at a cutoff threshold of 2 points or above, the Phoenix Sepsis Score accurately identified 75% of episodes as sepsis, with a sensitivity for attributable mortality of 77% and a specificity of 25%, and a respective sensitivity and specificity of 89% and 25% for definitely attributable mortality.

When the cutoff threshold was raised to 4 points or above, the corresponding values for attributable mortality were 69% and 72%, and for definitely attributable mortality, they were 89% and 72%.

Wolf et al comment that the “[r]elatively poor performance of sepsis scores for attributable mortality in this population may be caused by delayed death and/or preexisting subclinical multiorgan dysfunction from cancer or chemotherapy,” but emphasize that “the temptation to focus only on definitely attributable mortality should be resisted as it might underestimate true sepsis-related mortality.”

The authors conclude that the “Phoenix Sepsis Score accurately classified the risk of definitely attributable mortality in children receiving treatment for cancer who were admitted to the ICU with suspected infection.”

But they add: “More research is needed to validate these findings, evaluate population-specific cutoffs, and identify approaches that predict delayed attributable mortality.”

medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2024 Springer Healthcare Ltd, part of the Springer Nature Group

JAMA Networ Open 2024; doi:10.1001/jamanetworkopen.2024.15917


Related topics