Reproducibility and agreement of pharyngeal automated impedance manometry with videofluoroscopy

Automated impedance manometry analysis (AIM) measures swallow function variables that define bolus timing, intrabolus pressure, contractile vigor, and bolus presence; these are combined to derive a swallow risk index (SRI) that is correlated with pharyngeal dysfunction and aspiration. We assessed intra-rater and inter-rater reproducibility of AIM analysis-derived variables; the diagnostic accuracy of AIM-based criteria for detecting aspiration was determined by using expertly scored videofluoroscopy as the standard.

Data on 50 bolus swallows of 10 mL each were randomly selected from a database of swallows that were simultaneously recorded with impedance, manometry, and videofluoroscopy. Data were divided into 5 subgroups of 10 swallows for analysis: 10 dysphagic liquid, 10 dysphagic liquid with aspiration, 10 dysphagic semisolid, 10 control liquid, and 10 control semisolid. Repeat analyses were performed by 10 observers with varying levels of expertise in manometry by using purpose-designed software (AIMplot). Swallow videos were scored by 4 experts by using the penetration-aspiration scale (PAS) score. Reproducibility of calculation of swallow function variables and the SRI and PAS was assessed by using intraclass correlation coefficient (ICC). The majority consensus of expert PAS scores was used to dichotomously define aspiration (consensus PAS >3). Observer analyses were compared by Cohen κ statistical analysis.

The intra-rater and inter-rater reproducibility of swallow function variables was high (SRI mean intra-rater ICC, 0.97 and mean inter-rater ICC, 0.91). SRI >15-20 was optimal for detecting the presence of aspiration during liquid bolus swallows with an almost perfect agreement with expert scoring of videofluoroscopy (κ > 0.8).

AIM analysis has high intra-rater and inter-rater reproducibility, and among observers of varying expertise, SRI predicts the presence of aspiration.