External validation can provide the clinician with IOS equations that best predict the risk of uncontrolled asthma in their population

Authors:
Publisher website
Google Scholar
Abstract
Background: Peripheral airway impairment (PAI) has been shown to have a close association to risk of uncontrolled asthma in children. However, clear methods have not been established for the clinician to select impulse oscillometry (IOS) reference equations best suited for their population. Our study aimed to develop a practical external validation analytic approach for the clinician to determine which of the available reference equations best predicts uncontrolled asthma for their patients.Methods: This is a post hoc analyses of data collected at baseline in a randomized controlled study that occurred from March 2016 to 2018. The study population consisted of 227 children, ages 4−18 years, with moderate to severe asthma. Discrimination and calibration predictive performance of available and suitable IOS equations were assessed by using uncontrolled asthma as the criterion outcome. Discrimination statistics of accuracy, sensitivity, and specificity served as the primary performance indicators. Rank scores were determined by the number of acceptable limit thresholds met for these measures (≥60%, ≥50%, and ≥60%, respectively) across IOS metrics (R5, R5‐R20, AX, and X5) resulting in a total possible score of 12.Results: External validity assessment determined the rank order of best to worst equations as being Gochicoa‐Rangel (rank score = 10) > Nowowiejska (rank score = 9) > Assumapcao (rank score = 6) > Amra (rank score = 2). Gochicoa‐Rangel reference equations provided the best option for universal application with accuracy of 73.1%, 72.2%, 76.7%, and 66.2% for R5, R5‐R20, AX, and X5, respectively.Conclusions: External validation, particularly discrimination in asthmatic children, offers the clinician a practical approach to selecting the most suitable predictive equations for their patients.

This publication has 26 references indexed in Scilit: