Variations in assessing results of extended pH monitoring (EPM): Is there a single best method?

Gastroesophageal reflux (GER) in infancy is commonly evaluated with EPM. Criteria for distinguishing between physiologic and pathologic GER are variable, however, most centers use reflux index (RI), i.e., total % of time distal esophageal pH is < 4.0 as a major determinant. Other factors affecting RI include type and frequency of feedings, position, activity and technical equipment. We prospectively analyzed EPM recordings on 81 infants (45 males, 36 females) between 1-12 months in age by determining RI, Euler and Byrne Score (EBS); [number of reflux episodes-x and the number of episodes > 5 minutes-y, EBS = x + 4y] and area under pH 4.0 (AUP) on formula feedings using Synectics® digitrapper Mark III. Our aim was to analyse and compare the results of these 3 different methods to determine their reliability. Pathologic GER was defined as RI ≥ 5%, or EBS of ≥ 50 or AUP of 25/>. We selected RI, EBS and a combination of both as gold standards and compared them against each other and then to AUP. Optimal thresholds for discriminating between physiologic and pathologic GER were evaluated by Bayesian analysis and receiver operating characteristic curves (ROC). Specificity and sensitivity for each individual value was plotted. Using RI and EBS, overlap in standard deviations occurred. However, ROC analysis of AUP was superior to distal RI or EBS in making the distinction between physiologic and pathologic GER. As shown in the table, a significant number of patients would have been classified as having physiologic GER based on interpretation using RI and a smaller number would have been missed even with EBS. Physiologic Pathologic Sensitivity Specificity EBS 38 43 0.89 0.91 RI 56 25 0.93 0.84 AUC 35 46 0.91 0.91 Our data suggest that with computerized readout of EPM, it is easy to assess all 3 parameters i.e., RI, EBS and AUP. All 3 parameters should be evaluated to determine the normalcy of the test, as otherwise a significant number of patients will be misdiagnosed.