Correspondence on “Removing systolic blood pressure from the National Early Warning Score (NEWS) for mortality prediction: an observational study” by Zaboli et al

When two models are evaluated on the same cohort of patients, as in the study by Zaboli et al, 1 which compared the National Early Warning Score (NEWS) with and without systolic blood pressure (NEWS-SBP), the resulting performance metrics are inherently correlated. This correlation must be properly accounted for in statistical testing. Misunderstanding this principle can lead to flawed interpretations of study results, a point well raised by recent correspondence2 regarding the aforementioned study.