Introduction: Ultrasound assessment of swallowing has been documented as reliable in both healthy and dysphagic participants. In addition, there is evidence of good correlation with ‘gold standard’ videofluoroscopic swallowing study (VFSS). Despite this, ultrasound has not translated into clinical practice. This may be due to the cost and accessibility of ultrasound devices as well as the time required to analyse images offline. Recent innovations have produced inexpensive, wireless, portable ultrasound technology, which has the potential for increased access and immediate results. This project explored a number of components of inter- and intra-rater reliability using portable ultrasound. Reliability of measures, from images acquired, selected and measured online in a pressured clinical environment, was compared with reliability of measurement of pre-selected images offline. The project additionally made preliminary assessments of the validity of ultrasound against the gold standard of VFSS.

Methods:

Participants: Eight patients, aged 33-96 with mixed aetiologies were recruited following referral for a clinical VFSS. Instrumentation: A curvilinear Clarius™ultrasound device, wirelessly connected to an iPad, was used to acquire images during dynamic swallowing gestures - hyoid excursion and thyrohyoid approximation as well as images for measures of tongue thickness at rest. A linear Clarius™ transducer was used to collect measures of cross-sectional area of submental muscles at rest. Data acquisition and measurement: Ultrasound data were independently collected by two investigators within the same day. The primary investigator completed ultrasound concurrently with VFSS for the purposes of validity assessment. Subsequent ultrasound analysis was completed by a co-investigator immediately following. Online measurements of ultrasound images were completed during the exam, using Clarius™software on an iPad. Offline analyses of ultrasound were completed by two raters with a minimum of eleven days between measures. VFSS measures were completed offline by rater one, using ImageJ software on a large screen. Reliability assessment: Inter-rater reliability was calculated with intraclass correlation coefficient (ICC) based on linear mixed effects model analyses (in R software). Effect of data acquisition on reliability was explored by calculating online inter-rater ICC and comparing with offline inter-rater ICC. Effect of environmental, equipment and time constraints on online measurement was explored by calculating ICC of online and offline measurement of the same pre-selected acquired images. Validity assessment: Hyoid excursion and thyrohyoid approximation during liquid and puree swallowing were concurrently assessed using ultrasound and the ‘gold-standard’ instrumentation, VFSS. Pearson correlation coefficients were calculated in order to make a preliminary assessment of correlation between assessment methods.

Results:

Reliability: Inter-rater reliability of online acquisition and measurement ranged from poor (< .50) to moderate (.50 –.75). ICC values for online and offline measurement of the same images were moderate (.50 –.75) for dynamic measures, and excellent (>.90) for static measures. Inter- and intra-rater reliability for offline measures was good (>.75) to excellent (>.90) for hyoid excursion and static morphometry measures and moderate (.50 –.75) for thyrohyoid approximation.

Validity: Pearson coefficient of correlation calculations for hyoid excursion were moderate (r=0.76; p=0.001) for puree bolus and excellent for liquid bolus (r=0.92; p=0.03). Thyrohyoid approximation was found to have a moderate but insignificant, relationship between modalities for both puree and liquid bolus (r=0.61; p=0.11).

Conclusion: The high reliability for offline measurement of ultrasound images is comparable to previous studies using sophisticated instrumentation. Reduction in reliability is noted when measuring the images online within the context of a clinical environment compared with offline measurement. Online data analyses may be affected by the pressure and lighting of a clinical environment paired with lower resolution of the device, size of the screen and use of a touch screen for measurement. Further reduction in reliability of dynamic swallowing measures is noted when data acquisition is added, this may be due to different techniques by examiners as well as variance in patient performance. The findings suggest that it is important to further explore methods of improving reliability of data acquisition as well as immediate online analysis before clinical translation of ultrasound assessment of swallowing is achieved. Preliminary data on validity of the portable ultrasound device indicates high correlation between assessment methods (ultrasound and VFSS) for hyoid excursion only. Analysis of a larger cohort is required to provide a robust assessment of the validity of ultrasound images collected with this technology for both hyoid excursion and thyrohyoid approximation.