Twenty healthy Asian subjects (9 men and 11 women) with a mean age of 21.3 years (range, 20–23 years) were recruited for this study. Exclusion criteria were previous history or current symptoms of swallowing difficulty, gastrointestinal disease, or neuromuscular disease. The Institutional Review Board of the Kosin University Gospel Hospital approved this study protocol (IRB No. 11-44). Informed consent was obtained from all participants.

A thin liquid of diluted barium sulfate (Solotop Suspension 140, Taejoon Inc., Seoul, Korea) with a final concentration of 14 g/dL was administered to the participants, in 2-mL doses using a spoon or syringe, in UP and right LD postures, respectively. To minimize radiation exposure, only one consistency was used. Participants sat on a chair with a backrest and assumed a comfortable upright head posture by gazing into an indicator at chest level on the wall in front. While in the LD posture, the head rested on a pillow to maintain a comfortable head and neck posture. Participants held the bolus after the barium delivery and swallowed on receiving a verbal command. Three UP-LD cycles were repeated sequentially (UP-LD-UP-LD-UP-LD), totaling 6 swallows per person. An exact lateral view instead of a slight oblique projection maximized the overlap of bilateral mandibular shadows. Continuous mode lateral fluoroscopy (Uromax EX, Shimadzu, Kyoto, Japan) was employed to record video clips at 60 frames per second.

For subsequent motion analysis, the recorded videos were trimmed using Pinnacle Studio 12.0 (Corel Corp, Ottawa, Canada). Each video clip included frames from 0.5 seconds before the start of the bolus head reaching the lower mandibular margin, to 0.5 seconds after the end of swallowing when the bolus tail completed its passage through the upper esophageal sphincter. A coin measuring 24 mm in diameter attached to the midline under each subject’s chin served as a reference ruler.

A researcher digitized the position of the anatomical structures and the food bolus using the Ariel Performance Analysis System (Ariel Dynamics Inc., Trabuco Canyon, CA, USA) frame by frame. Ten video clips of videofluoroscopic swallowing analyses obtained from 10 different participants and copies with slightly modified overall brightness were evaluated for inter-rater and intra-rater reliability. To assess reliability, two researchers with 3 years of experience, who were blinded to information on postures, analyzed these video clips. The anterior-inferior margin of the hyoid bone, the head of the food bolus, and the anterior edge of the subglottic air column were marked. The superior margin of the subglottic air column was selected as a proxy for the vocal folds. The coordinate axes are shown in Fig. 1. All values were calculated using in-house MATLAB script and Statistics Toolbox (R2014b; MathWorks, Natick, MA, USA). We analyzed all the 6 video clips for each subject (3 swallows per posture in 2 postures) and averaged all the kinematic parameters per posture for statistical analysis. As a result, a dataset consisting of 20 pairs was used for each statistical analysis. The coefficient of variation was calculated for each 3-cycle set of every kinematic parameter to evaluate measurement reliability.

The Penetration-Aspiration Scale (PAS), which is an 8-point observational scale, was used to determine the depth of penetration into the airway [16]. The location of each structure when the bolus head reaches the lower mandibular margin was defined as the initial position. The peak position refers to each structure at the highest (the peak vertical) and the anterior most (the peak horizontal) position. The calculated kinematic parameters included the horizontal and vertical initial positions (HIP, VIP), horizontal and vertical peak positions, maximal horizontal and vertical displacement of the hyoid bone and the vocal folds (Fig. 2A). The maximal horizontal displacement was calculated from the posteriormost to the anteriormost point of the structure during swallowing plus or minus 0.5 seconds. The maximal vertical displacement was defined as the distance between the lowest and the highest positions during swallowing plus or minus 0.5 seconds (Fig. 2A). We measured the location of the bolus head at a time point half-way between the initial and peak vertical positions of the vocal folds. The approach was adopted because, at the peak vertical position of the vocal folds, the bolus head was already in the esophagus in many cases, which was outside of the recorded video clips (Fig. 2B). Velocity was calculated for each set of adjacent frames. Time to peak position was calculated from the point when the bolus head reached the lower margin of the mandible to when the hyoid bone or vocal folds reached the highest position in the vertical or the anteriormost position in the horizontal direction. The pharyngeal transit time (PTT) is the time determined starting from the head of the bolus reaching the lower margin of the mandibular bone to the point when the bolus tail passed through the lower end of the upper esophageal sphincter.

To generate an average trajectory of the hyoid bone and vocal folds, subject trajectories were standardized by the temporal normalization employed in our previous study and then pooled [17]. Briefly, each subject’s swallowing cycle was normalized temporally by interpolating for 100 time steps. The pooled average trajectory and data dispersion were presented at each time point as a series of mean values and band ranges obtained from the elliptical distributions within 95% confidence intervals (mean±2 SEM) in both the x- and y-axes at each time step. As normalization was used only for figure generation, non-normalized raw data were used for all statistical analysis.

A Wilcoxon signed-rank test was used to compare the maximal displacement, maximal velocities, and PTT between the UP and LD postures, because of the small sample size and non-normal distribution of data. Inter-rater and intra-rater reliability was assessed using an intraclass correlation coefficient (ICC) for each pair of measurements from two copies of the same video clip. The coefficient of variation was calculated for each 3-cycle set of every parameter. To compare PAS distribution between postures, χ² test was used. Differences were considered significant if the p-value was less than 0.05. Analysis of all data was conducted using SPSS version 14.0 (SPSS Inc., Chicago, IL, USA). Cohen’s d was calculated using R [18] and the package “effsize” [19].

The resting positions of hyoid bone and larynx were higher and more forward in the LD posture when compared with the UP posture in the present study. However, the resting position of the hyoid bone moves more posteriorly and inferiorly in the supine posture than in the UP posture [13]. This finding suggests a fundamental difference in the effect of supine and LD postures on swallowing kinematics. First, supine and LD postures impose different directions of gravitational force on the hyoid bone, larynx, and food bolus. Supine posture leads to a posterior gravitational pull on the tongue and food bolus, which can ease tongue base retraction and strengthen oral propulsive pumping [10]. Right LD posture, however, does not exert backward or downward gravitational pull, leaving right-directed force on the coronal plane, resulting in minimal effects on the movement in lateral fluoroscopic projection. Thus, LD posture results in minimal gravitational effect on swallowing-related structures, unlike the supine posture. Second, the downward pull of the hyolaryngeal complex by the lung and trachea differs between the supine and LD postures. Since the larynx is connected to the lung via the trachea, and to the hyoid bone by ligamentous structures and muscles [20], the compliance and flexibility of the lung, trachea, and adjacent soft tissues affects hyolaryngeal elevation during swallowing. Adhesion and fibrosis of surrounding soft tissues induce fatigue during swallowing in many head and neck cancer survivors [21]. Additional studies are required to investigate the thoraco-abdominal biomechanics in the supine and LD postures. Third, pre-swallow jaw motion or tongue muscle activity affects the resting position of the hyoid bone and larynx [18]. Compared with the UP posture, the LD posture increases the difficulty in holding the food bolus still, which increases the effort of the tongue just before swallowing. Further research is required to compare the resting hyoid position with and without food bolus in the two body postures.

Most studies investigating the kinematic effect of the swallowing posture relate to the supine posture [12,13,15,22]. However, the effect of supine posture on hyoid bone displacement during swallowing is controversial. While ultrasonography showed no significant difference in hyoid maximal displacement [12], another study reported that the resting hyoid position moves to a more posterior position in the supine posture resulting in a greater anterior displacement during swallowing [13]. The present study reaffirms that 2D displacement of the hyoid bone remains unchanged in the LD posture. Furthermore, we report a novel finding that the hyoid bone trajectory during swallowing has a different shape, greater upward and smaller forward displacement in the LD posture compared with the UP posture. Ultrasonography might be less precise in evaluating vertical and horizontal hyoid displacement separately due to difficulties in controlling head motion, probe angle, and pressure on the contact surface. Greater vertical displacement in the LD posture can be explained by changing the direction of the gravitational force and the reduction of downward force on the hyoid bone and larynx or strenuous tongue activity in order to compensate for less effective bolus transport.

Another novel finding is that the hyoid reached the peak superior position earlier in the LD posture compared with the UP posture. The PTT was comparable, which is consistent with previous reports [11,22]. It may be an epiphenomenon related to increased tongue activity in oral propulsion to accommodate the different direction of gravitational pull. However, considering that temporal patterns of the swallowing-related muscle electromyographic activities are comparable between upright and supine postures [11], a shorter time to peak position is probably related to mechanical changes due to decreased downward force than changes in neuromuscular control. In other words, the LD posture facilitates early peak position of the hyoid bone during swallowing by diminishing the opposing downward force, which is a combination of gravitational pull and traction from the soft tissue, lung, and trachea. Although laryngeal vertical peak velocity was reduced in the LD posture, the mean vertical velocity until the peak position was reached remained comparable between the UP and LD postures (p=0.903). The lower peak vertical velocity, with no difference in mean velocity of the larynx, suggests an upward acceleration of the larynx in the early phase of swallowing, which might also be related to decreased opposing downward force in an LD posture.

In the present study, the LD posture yielded less frequent penetration (PAS 2) compared with the UP posture (Table 2). It is essential not to overestimate this result because PAS 2 is considered normal in that up to 20% of apparently healthy young adults reportedly showed PAS 2 penetration [23]. This observation is in line with the aforementioned findings, such as higher location of the food bolus during laryngeal ascension, higher resting and peak position of the hyoid and larynx, and shorter time for the hyoid to reach its peak vertical position. The hyoid showed smaller anterior and greater superior displacement during swallowing in the LD posture. In addition, the resting position of the hyoid and larynx moved farther upward and forward. Molfenter and Steele [24] suggested that a higher peak vertical position of the hyoid bone is related to a decreased risk of penetration. Therefore, enhanced vertical displacement and elevated resting and peak hyoid position might facilitate airway protection. Although the difference was small, expedited excursion of the hyoid bone to its peak vertical position may not endanger safe swallowing. Therefore, we suggest that consuming small amounts of liquid in the LD posture does not increase the risk in terms of hyolaryngeal positioning and movement. However, previous studies reported that the LD posture compromises respiratory function in certain patient populations [25,26]. Therefore, further studies are required to determine if LD posture was safe for consumption of larger volume of liquid as well as other diet with different textures. In addition, the oral and esophageal phases of swallowing require evaluation to obtain a complete safety profile.

Understanding the changes in the magnitude and direction of the motion of the hyoid bone and larynx with different body postures provides a foundation for personalized treatment in patients who are unable to maintain upright posture. The horizontal travel distance of the hyoid was smaller, but the peak position moved farther forward due to the altered resting hyoid position in the LD than in the UP posture. These results suggest that different body postures change the environment for swallowing exercises, linked to different treatment effects. Thus, planning for anti-gravity exercises in dysphagia management requires consideration of posture effects on hyolaryngeal movement.

First, the pre-swallow tongue activity might differ between the two postures, particularly because different food delivery methods (spoon vs. syringe) were used. Most participants, however, stayed still while waiting to swallow, which stabilized the hyoid bone position for the initial half second before swallowing (Fig. 2B). In future investigations, a consistent delivery method such as straw drinking will be a more viable option in order to minimize this concern. Second, the findings in young and healthy volunteers require caution until there is clearer evidence of the benefits in older population. Third, the head-to-head comparison between the UP and LD postures while maximizing the statistical power, excluded the supine posture from the study design and may have limited the richness and variety of clinical inferences. Finally, a potential carry-over effect occurred because of the unbalanced design, with which all participants started with UP and ended with LD. A balanced, random sequence design will be more appropriate for the comparison.