Variability in Swallowing Biomechanics in Infants with Feeding Difficulties: A Videofluoroscopic Analysis

Swallowing is a complex process that requires the precise coordination of 26 groups of muscles and the integration of the respiratory, gastrointestinal, masticatory and neurological systems [1‐3]. Liquid feeds in young infants show significant physiological differences to eating and drinking in older children and adults. The reflexive, motor pattern of suckling in an infant is a result of the coordination of the lips, tongue, jaw, palate, hyoid bone, and pharynx and, it is understood that the suckling reflex diminishes around 4–6 months of age as textures are introduced and swallowing comes under more volitional control [1‐3]. Whilst infant feeding is considered primarily reflexive, the process of swallowing has inherent variability with changes in hunger / thirst, state, viscosity, flow rate, volumes, general fatigue, as well as transience in reflexes and the masticatory demand over a mealtime as a child introduces textures [4, 5]. The presence of adaptability in suckling and sucking patterns is supported by work in breastfeeding which found wide ranges in measures within and between breastfeeds [6‐11]. This suggests that cortical modulation is possible and one interpretation is that this variability in swallows during breastfeeding is an adaptation to the variable flow from the breast [6‐11]. With this inherent adaptability, swallow to swallow differences are to be expected. Variability or cortical modulation is considered a common feature of healthy swallowing in adults and a sign of a flexible system that responds to bolus properties [12‐14]. Yet, in paediatric feeding, clinicians often report swallow variability or inconsistency in infants as a sign of concern in feeding evaluations, describing the swallow as ‘incoordinated’. In infants, where milk suckling is a more reflexive, rhythmical process, can variability be excessive and therefore pathological with an impact on safety? Or, similar to adults, is a lack of variability a sign of inability to adapt and modulate and so a sign of safety risk? Given that these two scenario’s represent opposite interpretations, a greater understanding of feeding biomechanics and in particular variability in infants would greatly improve the accuracy of assessment and guide tailored interventions.

Videofluoroscopic swallowing study (VFSS) is one of the most common instrumental feeding assessments across the lifespan because it allows for the visualization of all phases of swallowing. By using objective measures, we are able to capture patterns of movement in a reliable, repeatable manner. Our lab has previously validated these measures in infants and children [15‐19]. These measures allow between infant, and within-infant variability to be measured quantitatively. This study explored variability in swallowing biomechanics in infants with feeding difficulties using videofluoroscopic analysis. Our research questions were as follows: Is it possible to assess swallow to swallow variability in an infant with feeding concerns? Is there variability in pharyngeal timing and displacement in infants referred for videofluoroscopy, and is variability associated with aspiration risk?

This study received appropriate regional ethics and locality approval by The University of Auckland Human Patients Ethics Committee New Zealand (9263). Infants consecutively referred to the radiology department at Starship Childrens’ Health or Middlemore Hospital with concerns regarding feeding were included. Inclusion criteria were under 9 months old, full 20-s mid-feed loop, ability to view oropharyngeal anatomy and physiology. Studies that needed to be terminated prior to 20 s due to aspiration or non-compliance were excluded. Information regarding medical history, demographics, presence of a tracheostomy, history of lower respiratory tract infections (LRTI) and oxygen requirement were collected. Infants were broadly categorised by their corrected chronological age (group 1: 0–2 months, group 2: 3–5 months, group 3: 6–9 months) and grouped by their primary medical conditions into: neurological (including hypoxic ischaemic encephalopathy and neonatal drug exposure), respiratory (chronic lung disease and bronchiolitis), cardiac (valve surgery), anatomical (cleft palate), chromosomal (Down Syndrome) and unknown (typically developing with feeding concerns with no known medical aetiology) groups.

VFSS was conducted in the radiology suite on a Siemens Sireskop radiographic unit (Siemens, Munich, Germany). Either breast milk or recommended formula was used to prepare materials for the study, according to the particular infant’s needs. We used Varibar barium sulphate contrast (40% w/v) (E-Z-EM Canada Inc, Quebec, Canada) in 50:50 to create Level 0 Thin fluids (International Dysphagia Diet Standardization Initiative, IDDSI, 2016). Infants were placed at their usual or recommended feeding posture with the support of a caregiver/ parent and used their usual nipple and bottle. 20-s video loops of ‘midfeed’ bottle feeding were recorded. ‘Midfeed’ was defined as midway through the feed, ensuring that infants had established their stable, functional feeding pattern. The timing of recording the midfeed loop was determined by the attending speech pathologist and was recorded on a USB external drive in.avi file format at 30 frames per second (f/s) for frame-by-frame analysis.

The videos were analysed using Swallowtail™ (Belldev Medical, Chicago, USA). Each swallow was analysed utilizing the timing, line and area features in the program. The embedded timing and displacement measures were developed by Leonard and Kendall [20] and adapted by members of our lab to reflect the paediatric population including the addition of number of swallows and suck to swallow ratios. All measures have proven reliability and validity [15‐18]. Other key clinical notations were measured in binary form including the presence of residue, nasopharyngeal regurgitation and presence of gastroesophageal reflux. The Penetration-Aspiration Scale [21] was used to reflect current clinical practice and to draw on functional relevance in identifying airway safety during swallowing. Each swallow was given a Penetration-Aspiration Scale score. Airway risk was also rated for each swallow following the rating system of scores > 3 on the Penetration-Aspiration Scale described by Steele and colleagues [22]. A binary measure of aspiration was calculated for each individual (PAS < 3 versus PAS 3 +). All swallows were measured twice by two blinded independent raters. Raters were experienced speech pathologists in infant VFSS and had received training, and had measured > 100 videos using objective quantitative parameters. Hyoid measures were not taken as the hyoid bone has been proven to be difficult to visualize in those < 9 months old [19]. Where the airway was difficult to view, or the infant did not return to resting position with open airway between swallows, airway measures were not taken (Fig. 1).

Infants swallowed between 1 and 15 times during the 20-s mid-feed loop with a total of 349 swallows for analysis from the 50 selected infants. There were three outliers who only swallowed once within the 20-s loop. These infants were excluded from further analyses in view of the study aim to investigate swallow variability within an infant and to reduce data skewness. The characteristics of the 50 infants are displayed in Table 1.

The objective measures gained from the cohort are displayed in Table 2, comparing values across all infants (excluding 3 outliers who swallowed only once in the 20-s loop). Inter-rater reliability was substantial across all measures. All binary subjective measures and the Penetration-Aspiration Scale achieved 100% agreement using Cohen's κ test. A moderate-strong degree of reliability was found between raters on all timing measures (ICC = 0.78, 95% CI 0.73-0.82 p < 0.001). A high degree of reliability was also found between raters on all displacement measures (ICC = 0.82, 95% CI 0.78-0.84, p < 0.001).

There was significant within-infant variability in both timing and displacement measures (Figs. 2, 3). Within swallow timing and displacement measures correlated with one another (Table 3). There was no significant correlation between age and variability in objective swallowing measures (p > 0.05). There were no statistically significant differences in timing and displacement measures between diagnostic groups (U = 0.001, p = 0.317). Suck to swallow ratio was higher in those with neurology as their primary medical diagnosis (Median 4:1 suck to swallow ratio) when compared to those with a respiratory diagnosis (Median 2:1 suck to swallow ratio) (U = 36.00, p < 0.05).

This exploratory study examined variability in objective videofluoroscopic timing and displacement measures in 50 infants referred with feeding concerns. Objective measures and variability in measures were recorded successfully with moderate-strong inter-rater reliability. By choosing a mid-feed loop, variability associated with ‘warming up’ was considered to be reduced [23], leaving mid-feed biomechanics to be observed. Interestingly, whilst variability was seen in timing and displacement in this cohort, measures did not correlate with aspiration. Aspiration was, however, associated with a higher suck:swallow ratio, particularly if suck:swallow ratio was > 3 sucks per swallow, and with variability in suck:swallow ratio within a 20-s mid-feed loop. The population frequency of aspiration observed in this study (32%) is congruent with other patient cohort studies [15]. Aspiration was not a constant with some infants aspirating only once, and others aspirating across more swallows. No infant aspirated on all swallows. This has also been seen in adult populations with dysphagia [24].