Factors influencing duration of neonatal cranial ultrasound: A pilot study of retrospective data

Abstract

Thank you to the authors for their commentary and raising the important point of appropriate statistical model. Our project accurately identified significant differences between number of sonographers undertaking neonatal cranial ultrasound and overall scan duration, as well as between scan duration for babies with pathology present compared to those who had no pathology detected. We would like to reiterate that the impetus for this study was to conduct a single site examination of retrospective data to describe factors that influenced scan time in our context. We did receive advice from a statistician and study limitations are clearly reported in our published article. As suggested by the author of the letter to the editor, we conducted further analysis of our data. We agree that using general linear models (GLM) are useful for taking into consideration aspects of the data, which are problematic to standard regression analysis. As per our published paper we entered the variables of number of scan operators and presence of pathology into the model as these were two variables significantly correlated with the outcome variable of estimated scan duration. For the GLM, the variables were entered as main effects and with consideration of any interaction effect between presence of pathology and number of operators. The categorical variables had been recoded with dummy variables and were entered as fixed factors. The overall corrected model remained significant: R = .32; adjusted R = .31 (F = 30.95, df = 3, p = <.001). The GLM identified that when two operators conducted the scan, the average scan duration took significantly longer compared to