SAN DIEGO – A simple three-variable prediction rule can accurately predict which febrile infants younger than 2 months with fever who present to the emergency department have serious bacterial infections.
Bacterial meningitis, urinary tract infections, and bacteremia are considered serious bacterial infections; many young infants with these infections are difficult to identify, and current laboratory protocols for identifying them include urinalysis, white blood cell counts, band counts, and sometimes cerebrospinal fluid.
Dr. Nathan Kuppermann
If validated in larger studies, the new prediction rule – which does not require cerebrospinal fluid – could limit lumbar punctures, antibiotic use, and unnecessary hospitalizations among infants at negligible risk of serious bacterial infections (SBI).
In a presentation at the annual meeting of the Society for Academic Emergency Medicine, Dr. Nathan Kuppermann of the University of California, Davis, demonstrated results from a large prospective cohort study of 1,821 febrile infants 60 days old and younger conducted at 21 emergency departments. Infants with underlying congenital anomalies or critical illness were excluded from the study.
The investigators evaluated eight variables as potential predictors: age, temperature, Yale Observation Scale score, and clinician suspicion of SBI, along with four laboratory variables (urinalysis, white blood cell count, absolute neutrophil count [ANC], and procalcitonin). Band counts were available at some, but not all of the study centers, and therefore were not evaluated.
Dr. Kuppermann and his colleagues found that positive urinalysis, ANC of 4.09 x 1,000/mm3 or higher, and serum procalcitonin of 1.71 ng/mL or higher were, taken together, 98%-99% sensitive and about 60% specific, in predicting SBI in the cohort. Negative predictive values approached 100%. The rate of SBI was 9.3% of the cohort, in keeping with expected rates.
“The SBI positive group was more likely to have higher clinician suspicion of SBI, and all of the lab markers were more elevated in the SBI positive group than the SBI negative group,” Dr. Kuppermann said at the meeting. However, after the researchers considered all of the predictors as a group, only the urinalysis, the ANC, and the procalcitonin remained important. Only 3 of 170 (1.76%)* infants with SBIs were missed when the three-variable prediction tool was used.
To make the prediction rule easier to remember and use, the investigators evaluated lower, more standard, and easier to remember thresholds for the three variables. They reanalyzed their data using a lower cutoff point for ANC of 4.00 x 1,000/mm3, and a lowered procalcitonin cutoff of 0.5 ng/mL. The rule performed almost identically as the original rule, and did not miss any more patients with SBIs beyond the original three.
Dr. Kuppermann described the three-variable rule as “simple, objective, and highly accurate” in predicting or ruling out SBI. Nonetheless, it requires external validation in a large cohort, he acknowledged. He said he would continue to routinely perform lumbar punctures in infants younger than 30 days with fever until the findings could be further validated.
“But personally, in that second month of life, I would use these data to decide who actually needs a lumbar puncture and hospitalization,” he said.
Dr. Kuppermann also noted as a limitation of the study that the cohort included few infants with bacterial meningitis.
The study was conducted in the Pediatric Emergency Care Applied Research Network (PECARN) and supported by grants from the Health Resources and Services Administration, the Maternal and Child Health Bureau, the Emergency Medical Services for Children program, and the National Institutes of Health. Dr. Kuppermann disclosed no relevant conflicts of interest.
*A correction was made to this article 6/24/15