Current prognostic models could not reliably differentiate Kawasaki Disease (KD) from paediatric inflammatory multisystem syndrome (PIMS) temporally associated with COVID-19, reveals a study.
“We could not reproduce a distinguishability using the models or a correlation between the utilized parameters and PIMS except for one,” the investigators said. “Henceforth, research will need to focus on nonphenotypical properties for differentiation of KD and PIMS. [Pediatr Infec Dis J 2026;45:942-946]
This study applied two externally developed prognostic models to two datasets, one consisting of 26 KD and PIMS patients from Bayreuth, Germany (cohort 1), and the other comprising 632 patients of both diagnoses reported by paedatricians from all over Germany and Innsbruck, Austria (cohort 2).
The investigators then assessed the potential of these models to distinguish the diagnoses on a phenotypic level based on sensitivities and specificities, κ-coefficients by Cohen, and correlation analysis.
The model by Starnes and colleagues assigned the KD diagnosis to 22 patients and the PIMS diagnosis to three in cohort 1, with one left undiagnosed. Overall, 21 patients had been diagnosed with KD by their physicians in accordance with the model, which achieved a sensitivity of 90.5 percent and a specificity of 60.0 percent. [J Hosp Med 2024;19:175-184]
The other model by Godfred-Cato and colleagues diagnosed 17 patients with KD and three with PIMS, with six unassigned. Of the 26 patients in cohort 1, only 15 were identified in accordance with the diagnosis of their physicians, while five were assigned with another diagnosis. This model had a sensitivity of 66.7 percent and a specificity of 80.0 percent. [Pediatr Infect Dis J 2022;41:315-323]
Distinguishability
Among 632 patients with data from the German Society of Paediatric Infectiology (Deutsche Gesellschaft für Pädiatrische Infektiologie) in cohort 2, 493 were diagnosed with KD and 139 with PIMS. The model by Godfred-Cato and colleagues assigned 331 individuals in agreement with the diagnoses of their physicians, while 120 were left unassignable, and 181 were subordinated to a different diagnosis.
A total of 233 KD and 279 PIMS patients were identified. The model demonstrated a sensitivity of 44.6 percent and a specificity of 20.1 percent.
Comparing the two models, the κ-coefficient brought about small values of 0.347 and 0.059 for Starnes and Godfred-Cato, respectively.
“The sensitivity and specificity of both prognostic models did not display a reliable distinguishability between KD and PIMS,” the investigators said. “Moreover, correlation analysis cannot confirm the correlation of most parameters utilized in the models with the PIMS diagnosis.”
Notably, the patients’ age was the only parameter that had a positive association in both datasets.
“However, this parameter must be viewed with caution, since the median age and mostly affected age groups have been shifting with the different SARS-CoV variants,” the investigators said. “The number of hospitalized infants has been rising, and fewer teenagers have been admitted to the hospital.” [J Glob Health 2025;15:01002]
Current research suggests a theory of disease spectrum, including major hyperinflammatory syndromes, such as KD and PIMS. [Sci Rep 2023;13:1173; Int J Mol Sci 2025;26:8491]
“Nevertheless, the assessment of [the present] study is limited to the respective phenotypes,” the investigators said. “While phenotypically the results do not oppose such theory, on their own, they are insufficient to establish causality.”