Intra- and interrater reliability of the Chicago Classification of achalasia subtypes in pediatric high-resolution esophageal manometry (HRM) recordings

Abstract

Background

Subtyping achalasia by high-resolution manometry (HRM) is clinically relevant as response to therapy and prognosis have shown to vary accordingly. The aim of this study was to assess inter- and intrarater reliability of diagnosing achalasia and achalasia subtyping in children using the Chicago Classification (CC) V3.0.

Methods

Six observers analyzed 40 pediatric HRM recordings (22 achalasia and 18 non-achalasia) twice by using dedicated analysis software (ManoView 3.0, Given Imaging, Los Angeles, CA, USA). Integrated relaxation pressure (IRP4s), distal contractile integral (DCI), intrabolus pressurization pattern (IBP), and distal latency (DL) were extracted and analyzed hierarchically. Cohen's κ (2 raters) and Fleiss' κ (>2 raters) and the intraclass correlation coefficient (ICC) were used for categorical and ordinal data, respectively.

Results

Based on the results of dedicated analysis software only, intra- and interrater reliability was excellent and moderate (κ=0.89 and κ=0.52, respectively) for differentiating achalasia from non-achalasia. For subtyping achalasia, reliability decreased to substantial and fair (κ=0.72 and κ=0.28, respectively). When observers were allowed to change the software-driven diagnosis according to their own interpretation of the manometric patterns, intra- and interrater reliability increased for diagnosing achalasia (κ=0.98 and κ=0.92, respectively) and for subtyping achalasia (κ=0.79 and κ=0.58, respectively).

Conclusions

Intra- and interrater agreement for diagnosing achalasia when using HRM and the CC was very good to excellent when results of automated analysis software were interpreted by experienced observers. More variability was seen when relying solely on the software-driven diagnosis and for subtyping achalasia. Therefore, diagnosing and subtyping achalasia should be performed in pediatric motility centers with significant expertise.

Differences in the software-generated diagnosis of achalasia and achalasia subtypes explained by variation in landmark placement, leading to differences in (average values of) esophageal pressure topography metrics between analysts.

http://ift.tt/2syEFBm