Diagnostic interpretation of extended pH monitoring: is there a single best method?

Our aims were to compare the specificity and sensitivity of the reflux index (RI), Euler and Byrne score (EBS), and area under the curve (AUC) at pH < 4.0 in infants 1 year old for identifying pathologic gastroesophageal reflux (GER) and to identify new cutoffs on formula feedings. We undertook a prospective investigation of extended pH monitoring (EPM) in 117 infants with symptoms of GER. Infants were categorized as having pathologic reflux by three measures: RI > or = 5%, EBS > or = 50, or AUC pH < 4.0 or < or = 21.3, as determined by receiver operating characteristic (ROC) curve analysis. Using the RI as a criterion, 27 infants (23%) had pathologic reflux. When EBS and AUC were used, 65 (56%), and 67 (57%) respectively, were classified as having pathologic reflux. The specificities of RI, EBS, and AUC were 97.8, 100.0, and 100.0%, respectively, and the sensitivities were 93.0, 91.5, and 94.4, respectively. A new cutoff value for RI of 2.1% was determined using ROC curve analysis to improve the specificity and sensitivity of RI to the above values. The number of infants with pathologic GER is likely to be significantly less using the traditional RI alone on formula feedings, however, the revised RI cutoff value and AUC analysis by computerized pH tracings can enhance the accuracy on formula feedings.