How reliable is the statistical evaluation using the 'mean' in an abnormally distributed dataset
Yükleniyor...
Tarih
2024
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
SpringerLink
Erişim Hakkı
info:eu-repo/semantics/embargoedAccess
Özet
We read with great interest the article by Tourneux et al. [1] regarding the predictors of continuous positive airway pressure (CPAP) failure in moderate-to-late preterm infants. CPAP failure, requiring mechanical ventilation or surfactant administration, was detected in 12% of their study group. They reported that the strongest factor associated with CPAP failure was a low cut-off of 23% fraction inspired of oxygen (FiO2) at 3 h of life, which was significantly lower than in previous studies (29%) [1]. There is a major concern about the main outcome of this study, the difference in maximum FiO2 values between CPAP failure and CPAP success groups at 3 h after birth. The lowest value of FiO2 is estimated as 18.6% and 18.5% in the CPAP success and CPAP failure groups, respectively, if the authors gave the mean and standard deviation (mean±SD) of maximum FiO2 values (22.7±4.1 for the CPAP success group and 34.4±15.9 for the CPAP failure group). FiO2 values below 21% cannot be administered in any scientific study performed on human newborns. FiO2 values that are not compatible with reality are a consequence of the abnormal distribution of the data in the study. Besides, the authors have already stated the lowest FiO2 value as 22% in their supplementary data.
Açıklama
Anahtar Kelimeler
Statistical Evaluation, Distributed Dataset
Kaynak
European Journal of Pediatrics
WoS Q Değeri
Scopus Q Değeri
Q1
Cilt
183
Sayı
2
