Infant pigs demonstrate motor adaptation across multiple physiologic functions during feeding in response to dynamic changes in milk flow rate

Abstract

Safe and effective infant feeding requires precise coordination of sucking, swallowing, and breathing, yet disruptions in this coordination affect a significant number of infants. Altering sensory input, such as bottle nipple flow rate, is used to address poor coordination in infants. However, prior studies often compare different nipples used across different feeds, which introduces confounding variables and limits insight into neuromotor responses. To address this, we used infant pigs as a validated animal model to assess real-time neuromotor responses to dynamic changes in flow rate within a single feeding session using a custom, computer-controlled nipple. We collected high-speed biplanar videofluoroscopy, intraoral pressure, and respiratory data to evaluate kinematics, physiology, and coordination patterns. We found that while sucking and breathing rates remained stable, higher flow rates elicited greater tongue, hyoid, and thyroid translations, larger bolus sizes, and lower intraoral pressures. Notably, swallow rate increased resulting in a shift to suck-swallow-breathe coordination, with swallows occurring earlier in the suck cycle and more frequently per breath. These changes suggest that infants rapidly adapt their motor output to changing sensory conditions. This study demonstrates that real-time flow variation significantly impacts feeding mechanics and coordination, highlighting the potential for sensory-based interventions rooted in motor learning and neuromotor rehabilitation principles. Understanding how infants dynamically adjust to sensory changes offers critical insights into feeding development and provides a framework for developing more effective interventions for infants with feeding disorders.

NEW & NOTEWORTHY:

We developed a novel method to dynamically alter milk flow rate mid-feed in infant pigs to assess real-time neurophysiological responses. Unlike prior studies, our approach avoids confounding variables that would result from providing nipples of different flow rates during separate feeding sessions and reveals how motor output adapts to sensory input in real time, offering new insight into the neuromotor control underlying infant feeding.

Introduction

All infant mammals rely on suckling to obtain milk, a highly coordinated process that requires the seamless integration of multiple physiological functions (1). The process begins when an infant creates a secure seal around the nipple, enabling the generation of negative pressure to draw milk into the oral cavity (2–4). To acquire milk, the midline of the tongue moves in a rhythmic, wave-like motion from the anterior end of the mouth posteriorly toward the oropharynx (2–4). Once the milk reaches this space, the swallowing reflex is triggered and food is moved into the esophagus (3–6). Successful feeding relies on the infant’s ability to coordinate these two behaviors (sucking and swallowing) with each other. Swallowing must additionally be coordinated with breathing, as the pathway for air and food cross at the oropharynx in mammals. If swallowing is not properly coordinated with respiratory patterns, food can be aspirated, resulting in nutritional, developmental, and health complications (7–10). Feeding difficulties thus can arise through several mechanisms, including challenges with sucking, swallowing, breathing, or the coordination of those behaviors. As a result, more than 25% of full-term infants and over 80% of preterm infants exhibit some challenge with feeding (5, 11–14).

To effectively and safely perform each of these behaviors, as well as coordinate them, infants must be able to adjust their physiologic function in response to variation in sensation (15–17). Common sensory variation that infants can experience during feeding include flow rate, taste, viscosity, temperature, and smell, all of which can influence motor responses in sucking, swallowing, and aerodigestive coordination (10, 16,18 – 20, 22 – 31). Because of these responses, modifying sensory input is a widely used intervention for infants who experience feeding difficulties and can improve feeding outcomes and promote better oromotor functioning (20). Adjusting milk flow rate is one of the most widely used intervention for infants experiencing feeding difficulties, as it plays a critical role in an infant’s ability to coordinate suck-swallow-breathe patterns, and varies (albeit in different ways) during both breast and bottle feeding (32–34, 36, 37). During breastfeeding, flow rate is influenced by factors such as milk let-down and breast anatomy (4, 23). These natural fluctuations require infants to adjust in real time, fostering the development of oromotor coordination and sensory modulation (4, 24, 25). Understanding these dynamics helps assess how the varying flow profiles that are observed during breastfeeding impact feeding performance and physiological stability. In contrast, bottle feeding offers a less dynamic flow rate, which is affected by bottle and nipple design, milk properties, bottle orientation and volume, and caregiver technique (26–28). It has been demonstrated that significant variability exists among bottle nipples, and labeled flow rates do not always reflect actual performance (27, 29). Given that infants vary widely in anatomy, developmental stage, and feeding readiness, studying flow rate enables clinicians to implement individualized feeding strategies and select appropriate feeding tools (30). This growing recognition of flow rate’s impact on feeding outcomes in both bottle feeding and breastfeeding has prompted extensive research aimed at optimizing feeding interventions and improving infant health and development.

While studies have demonstrated that milk flow can vary dynamically during breastfeeding, and varies across different bottle nipples, we have relatively little insight into how infants respond to these changes in milk flow. Several lines of research have evaluated how infants respond to changes in flow, demonstrating that suction generation, tongue movements, and swallow function are all altered when infants are fed on bottles with different hole sizes (31–35) However, these studies are limited in the need for infants to feed from multiple bottle nipples across feeding sessions. This approach introduces several confounding variables because the infant’s body position, latch onto the nipple, and hunger drive may be altered between feeding from each bottle, which are all known to impact feeding outcomes (36–40). Thus, evaluating the response to flow using separate nipples may limit the ability to understand whether observed differences in feeding performance and mechanics are due to neuromotor responses to changes in milk flow rate, or if they stem from these additional factors (33, 34). Understanding these neurophysiological processes is crucial for developing more effective interventions, as existing strategies primarily address symptoms rather than the origin of feeding difficulties (41–43). Previous research investigating how the infant neuromotor system adapts to changes in flow dynamically have found that at higher flow, swallow volumes increase while ventilatory volumes decreased, demonstrating motor adaptation in feeding (al-Sayed et al., 1994). However, this work was unable to assess several metrics of sucking, swallowing, and respiratory physiology, including how changes in flow may impact their coordination. Thus, there is a need to develop tools that allow us to understand and target the neurophysiological mechanisms that underlie feeding difficulties, without the effects of confounding factors and with the ability to assess multiple physiologic processes in synchrony.

Understanding motor responses to variation in sensation during feeding in infants is further complicated by both the hidden nature of swallowing as well as the fragile nature of infants. Videofluoroscopic Swallow Studies (VFSS) are the gold standard for infant feeding assessment, as they provide a detailed view of movements involved in the oral, pharyngeal, and upper esophageal phases of swallowing (44, 45). However, collecting data in infants presents significant challenges, including the need to minimize exposure to radiation. Healthy infants are rarely included in a VFSS, making it difficult to establish a baseline for what constitutes healthy swallowing (46, 47). Even when infants undergo a VFSS, filming rates are often restricted to 15 frames per second (fps), and several aspects of feeding are unable to be assessed, including the incidence of aspiration, the exact timing of swallowing events, and detailed swallowing kinematics. (45, 48–50). As a result, neurophysiologic responses are generally poorly understood at this developmental stage (46, 51–53).

One potential way to circumvent these challenges is to use an animal model, which allows for improved insights into both healthy and pathophysiologic functioning in infants (54–56). Pigs are a validated animal model for infant feeding and have been used to address several key characteristics of feeding, including how infants latch, suck, and swallow, as well as how their feeding process changes over time as they grow (32–35, 41, 57–59). The use of infant pigs as a model also allows for enhanced control over experimental conditions and offers substantial improvements to data precision and accuracy, with increased temporal (100fps instead of 15fps, e.g. [(60)] and spatial (<1mm precision) resolution of data (61). Additionally, the use of animal models enables researchers to investigate physiological processes that would be difficult to capture in human infants, thereby contributing to a deeper understanding of feeding mechanisms (42, 54, 59, 62).

To evaluate the neurophysiological responses to changes in flow rate, we fed infant pigs using a computer-controlled nipple (Fig.1) that is capable of changing flow rate instantaneously while an infant continues to feed from the nipple. This allowed us to assess the response to changes in flow within a feeding session, without introducing confounding variables that are caused by switching nipples between feeds. We tested how changes in sensory input, specifically flow rate, were associated with changes in the physiological functioning of sucking, swallowing, breathing, and their coordination. We predicted that the pigs would adjust to changes in flow rate by modifying their motor output across multiple behaviors. Specifically, we predicted that at higher flow rates, infants would generate lower intraoral pressure, swallow larger boluses, exhibit larger tongue movements during sucking and swallowing, and experience disruptions in suck-swallow-breathe coordination.

Figure 1:

Diagram of the computer-controlled robotic nipple. Channels (blue) connected the nipple base and bottle. The channels could be opened and closed with the servomotor (black).

Methods

Animal housing and care

All animal care followed standard procedures approved by Northern Arizona University Institutional Animal Care and Use Committee Protocol #22–010 (63–66). Three full-term female infant pigs (Yorkshire/Landrace, Premier BioSource, CA, USA) were obtained 24 hours after birth and were housed in the Northern Arizona University animal vivarium. They were trained to feed on infant milk replacer (Birthright Milk, Ralco Show, Marshall, MN, USA) throughout the experiment and were bottle fed on a custom nipple that was the size and shape of a sow teat and had internal ducts (See (67) for details).

Nipple design

During data collection, we dynamically adjusted flow rate using a custom-designed computer-controlled nipple-bottle system (Cleveland Research Devices, Cleveland, OH) that had the same size and shape as the nipples that the infants were raised on (Fig. 1). The nipple featured three 30 cm parallel ducts of equal diameter (5 mm) and was made with silicone (Ecoflex 00–10; Smooth-On, Inc., Macungie, PA, USA) that was the same as the silicone used for the nipples that the infants were raised on (58, 67). These ducts extended from the nipple and connected to three channels with ¼” inner diameter tubing. The channels could be instantly opened or closed using servo motors controlled via a remote with individual switches for each channel. On the other side of the servo motors, the three channels connected to a bottle containing milk. The system transmitted signals indicating the opening and closing of each channel to PowerLab (ADInstruments, Colorado Springs, CO).

Marker Implantation

At 4–5 days of age, we anesthetized infant pigs using isoflurane anesthesia (2–4%) and used 18-gauge hypodermic needles to implant 0.8mm tantalum radio-opaque markers into the hard palate (submucosally, 5 markers), cranium (subcutaneously, 1 marker), and tongue (5 markers) (68). The tongue markers were injected in a series along the midline from the anterior top of the tongue to its attachment to the hyoid posteriorly. These markers enabled us to track the three-dimensional movements of these structures with biplanar videofluoroscopy using standard marker-based X-Ray Reconstruction of Moving Morphology (XROMM) protocols (59, 61).

Data Collection

At 17–19 days of age (approximately equivalent to a 6- to 9-month-old human infant) we recorded biplanar X-ray video (GE 9400 C-Arm, 71–100 kV, 5.2–8.0 mA) at 100 fps with high-speed video cameras (100fps, 12 MP Redwood, IO Industries, Ontario, CA) synchronized with plethysmography (respiration), and intraoral pressure utilizing a 16-channel PowerLab (10KHz, ADInstruments, Colorado Springs, CO). During suckling, intraoral pressure was measured by inserting a pressure transducer (3.5F Mikro-Tip Catheter Transducer; Millar Inc., Pearland, TX, USA) through the bottle nipple, with 1 cm of the transducer sticking out of the bottle nipple to ensure that the transducer would be in the pig’s oral cavity during feeding. We used a standard grid and calibration cube to undistort and calibrate the images according to standard XROMM data collection procedures (61, 69). Pigs were fed on milk replacer mixed with barium (E-Z-Paque, Radtech X-ray, Vassar, MI, USA) in a radiolucent plexiglass box. Pigs were recorded continuously while feeding from the computer-controlled bottle system as ducts were alternatively switched between three open (high flow) and one open (low flow), typically every 4–5 seconds, until the infants were finished feeding. During data collection, the pigs stood freely without restraint, maintaining similar body positions between trials and individuals. All bottles were filled to maximum capacity, and their angles were kept consistent across all feedings to ensure a continuous milk supply in the nipple. We collected at least 20 swallows per pig per flow rate on the same day, in a randomized order per pig.

We obtained computed tomography (CT) scans of each pig. Two pigs (TD03 and TD05) were scanned using an Aquilion 64 CT scanner (Toshiba, Tokyo, Japan) with a pixel spacing of 0.235 mm × 0.235 mm and a slice thickness of 0.5 mm. One pig (TD11) was scanned using a SkyScan 1273 microCT scanner (Bruker, Billerica, MA, USA) with a pixel spacing of 0.052 mm and matching slice thickness.

Data Processing

We detected when sucks began in the X-ray videos by identifying the frame at which the tongue contacted the hard palate, using standard procedures (70, 71), resulting in 256 sucks identified (N = 125 at low flow, N = 131 at high flow, TD03=86, TD05=82, TD11=88). The end of the suck was identified as the frame before the beginning of the next suck (0.01s earlier). The instantaneous suck rate was calculated as the inverse of the suck duration. On average, it took 1–2 sucks (less than half a second) for the infant pigs to change their behavior in response to the alteration in flow rate (Fig. 2B).

Figure 2:

(A) An X-ray image of an infant pig (TD03) feeding on the computer-controlled nipple with animated mesh models of the cranium (light yellow) and markers (dark blue). The movement of the middle tongue marker during one suck is traced in light pink. (B) Intraoral suction recorded as TD03 fed from the computer-controlled nipple. The dashed line indicates the time when the flow rate switched from high to low.

Swallows were identified as starting at the frame when the bolus accumulated in the oropharynx just before passing the epiglottis, following standard procedures (58, 65, 68). In total, 123 swallows were recorded (N = 54 on low flow, N = 69 on high flow, TD03=42, TD05=48, TD11=33). Swallow rate was calculated as the inverse of the duration between the start of one swallow and the next. Bolus size was measured in the lateral view at the onset of swallowing using ImageJ, with the bottle diameter serving as the scale to convert the images from pixels to mm² (72).

We exported raw intraoral pressure data from LabChart (ADInstruments, Colorado Springs, CO, USA) and converted the data from mV to mmHg by multiplying by 41.3, a conversion factor that was determined empirically. Pressure data were then downsampled using a rolling median from 10,000 Hz to 100 Hz in R. We used a custom MATLAB routine (Version 2021a, MathWorks, Natick, MA) to calculate the peak intraoral pressure generated per suck (Fig. 2B). We also exported plethysmography data from LabChart (ADInstruments, Colorado Springs, CO, USA), and downsampled the data using a rolling mean from 10,000 Hz to 100 Hz in R. We filtered the plethysmograph data using a zero-phase low-pass Butterworth filter at a cutoff frequency of 2.5 Hz. To account for variation in the position and tightness of the plethysmograph band around the ribcage (which affects the magnitude of the plethysmograph data and how much the signal changes as the band is stretched), respiration data were normalized to the maximum value observed in each feeding session for each pig. Breath amplitude was calculated as the difference between the maximum and minimum normalized plethysmograph values during a breath cycle. Breathing rate was calculated as the inverse of the duration between the start of one inspiration to the next. The timing of swallows was calculated as a percentage of the breath cycles, where 0% represents the start of inspiration and 100% represents the end of expiration.

We tracked the positions of the implanted radio-opaque markers in the mediolateral and dorsoventral X-ray videos in XMALab ((69); software and instructions available at https://bitbucket.org/xromm/xmalab). Markers in the hard palate and the skull were utilized as a rigid body. Then, we exported the rigid body transformations of the skull, 3D positions of all markers, undistorted videos, and camera positions using the XROMM_MayaTools scripts (developed by David Baier and Stephen M. Gatsey, available at https://bitbucket.org/xromm/xromm_mayatools/). In Maya (Autodesk, Inc., San Rafael, CA, USA), we created a reference scene for each individual, which contained its skull polygonal mesh and an anatomical coordinate system (ACS) axis that we positioned at the posteriormost end of the hard palate on the midline of the skull. We rotated the ACS so that the x-axis was aligned mediolaterally, the y-axis was aligned dorsoventrally, and the z-axis was aligned anteroposteriorly relative to the hard palate. We created animation scenes in Maya for each trial, in which we imported the individual’s reference scene, 3D marker positions, skull rigid body transformations, undistorted videos, and camera positions (Fig. 2A). Total 3D translations of the middle tongue marker per suck, and the posterior tongue, hyoid, and thyroid markers per swallow, were measured relative to this skull ACS (‘oRel' script, XROMM_MayaTools). This approach allowed us to create a consistent reference for analysis (68). The measured total translations were then aligned to the onset of each suck and swallow for precise temporal correlation. To account for variations in the placement of the markers across pigs, all kinematic data was scaled. In each session, the maximum recorded value was identified, and every data point was then divided by this maximum to produce a scaled measure.

Statistics

All statistical analyses were performed in R (R core team, www.r-project.org, v 4.3.0). We used linear mixed-effects models to test for differences in variables of interest, with flow rate as a fixed effect and individual as random effects (73). Variables of interest included: suck rate, swallow rate, breath rate, breath amplitude, bolus size, intraoral pressure generated per suck, and movement of the hyoid, thyroid, middle aspect of tongue, and posterior aspect of tongue. P-values for the main effects were calculated using the Anova() function on the model in R. In cases where the p-value was significant, we calculated Cohen’s d to determine effect size (74). To evaluate suck-swallow coordination and swallow-breathe coordination, we used a Hermans-Rasson test (using HR_test() from the package ‘CircMLE’, (75)) to determine whether the timing of sucks and swallows deviated from a circular uniform distribution (76, 77). In order to test for a bimodal distribution in the timing of swallows during the breath cycle at high flow, the angles used for this analysis were doubled and the resulting mean angles were reduced to modulo 360° using functions created by (77). Functions from the package ‘circular’ (78) were used to calculate the mean vectors of the distributions and the bootstrap 95% confidence intervals. Plots were generated using the package ‘circular’ and functions created by (77).

Results

The impact of flow on sucking

When infant pigs were exposed to the two different flow rates, there was no statistically significant difference in their sucking rate (p < 0.001, d = 0.43; Fig. 3A, Table 1) or in sucks per swallow (Mean ± SD at high flow: 1.96 ± 0.520, Mean ± SD at low flow: 2.30 ± 0.657). However, infants feeding at the lower flow rate generated significantly greater peak intraoral pressures per suck compared to those feeding at the higher flow rate (p < 0.001, d = −1.20; Fig. 3B, Table 1). Sucking kinematics were also altered as the middle tongue marker underwent larger excursions when feeding at a high flow rate compared to a low flow rate (p < 0.001, d=1.01, Fig.3C, Table 1). In addition, when evaluating tongue kinematics within a suck, the dorsoventral movement of the middle tongue marker resembled a sine wave at high flow, showing relatively continuous motion throughout each suck. In contrast, at low flow rates, the tongue's movement changed — with the middle tongue marker remaining elevated throughout the first approximately 60% of the cycle, before rapidly depressing and then quickly rising just before the next suck began (Fig. S1). Despite, tongue movements being different when infant pigs were exposed to the two different flow rates, we found no statistically significant differences in their jaw movements (χ²= .168, p= 0.682, d= .130, Fig. S2).

Figure 3:

Suck rate (A), peak intraoral pressure generation (B), and kinematics of middle tongue movement (C) in infants feeding under high flow rate (light green) and low flow rate (dark green) conditions. The width of the violin plots at a given value indicates the density (i.e. clustering) of the data points at that value. Solid lines between groups indicate statistically significant differences with a large effect size.

Table 1:

Planned contrast and effect size (Cohen’s d) results from statistical analyses (chi- square statistic, p-value; Cohen’s d)

Variable	X2	P-value	Cohen's d	Mean & SD (High Flow)	Mean & SD (Low Flow)
Suck Rate (Hz)	21.5	p < 0.0001	0.425	3.57, .632	3.50, .616
Peak Intraoral Pressure (mmHg)	105	p < 0.0001	−1.20	197, 59.1	285, 83.5
Middle Tongue perimeter Movement (normalized)	223	p < 0.0001	1.01	.829,.083	.686, .134
Swallow Rate (Hz)	27.4	p < 0.0001	0.668	1.95, .690	1.54, .516
Bolus Size (mm2)	57.6	p < 0.0001	1.16	201, 60.1	140, 42. 5
Posterior Tongue perimeter Movement (normalized)	102	p < 0.0001	1.73	.838, .118	.757,.120
Hyoid perimeter Movement (normalized)	96.8	p < 0.0001	1.65	.835, .124	.628, .094
Thyroid perimeter Movement (normalized)	42.4	p < 0.0001	1.42	.821, .096	.699, .073
Breath Rate (Hz)	2.56	0.110	−0.471	.912,.281	1.07, .413
Breath Amplitude	6.23	p < 0.0001	−0.555	.598,.201	.743, .239
Swallow Timing (% of Suck Cycle)	24.1	p < 0.0001	−0.823	19.1, 10.3	27.2, 8.23

Medium statistically significant values have a p value of p < 0.0001 and a Cohen’s d value of .5–.8. Large statistically significant values have a p value of p < 0.0001 and a Cohen’s d value greater than .8.

Swallow response to change in flow rate

Overall, several aspects of swallowing were significantly influenced by flow rate. When exposed to a higher flow rate, infant pigs exhibited an increased swallowing frequency (p < 0.001, d = 0.67; Fig. 4A, Table 1), larger bolus sizes (p < 0.001, d = 1.16; Fig. 4B, Table 1), and larger excursions of the posterior tongue marker (p < 0.001, d = 1.73; Fig. 5A, Table 1) compared to the low flow condition. Similarly, both the hyoid and thyroid exhibited significantly larger excursions under high-flow conditions (hyoid: p < 0.001, d = 1.65; thyroid: p < 0.001, d = 1.42; Fig. 5A&5B, Table 1).

Figure 4:

Swallow rate (A) and bolus size (B) in infants feeding under high flow rate (light green) and low flow rate (dark green) conditions. The width of the violin plots at a given value indicates the density (i.e. clustering) of the data points at that value. A dashed line between groups indicates a statistically significant difference with a medium effect size, and a solid line between groups indicates a statistically significant difference with a large effect size.

Figure 5:

Kinematics of the posterior tongue (A), hyoid (B), and thyroid (C) during swallows in infants feeding under high flow rate (light green) and low flow rate (dark green) conditions. The width of the violin plots at a given value indicates the density (i.e. clustering) of the data points at that value. Solid lines between groups indicate a statistically significant difference with a large effect size.

Impact of altered flow rate on respiration

There was not a significant difference in breathing rate when infants were feeding at different flow rates (p =.110, d=-.471, Fig.6A, Table 1). However, when infant pigs were feeding at a lower flow rate, breath amplitude was greater than when feeding at a higher flow rate (p < 0.0001, d=-0.555, Fig.6B).

Figure 6:

Breath rate (A) and scaled breath amplitude (B) in infants feeding under high flow rate (light green) and low flow rate (dark green) conditions. The width of the violin plots at a given value indicates the density (i.e. clustering) of the data points at that value. Dashed lines between groups indicate statistically significant differences with a medium effect size.

Coordination response to change in flow rate

Infants' suck-swallow and swallow-breathe coordination were impacted by flow rate. When infant pigs were feeding at a high flow rate they swallowed closer to the beginning of the suck cycle, whereas infants feeding at a low flow rate swallowed later within the suck cycle (p < 0.001, d=-0.820, Fig 7A, Table 1). Since, swallow rate was higher when infants were feeding at high flow, and breathing rate was not different, pigs swallowed more frequently per breath when feeding from high flow than low flow. Specifically, at high flow, infants typically swallowed twice per breath cycle, with one swallow occurring mid-inspiration and one occurring mid-expiration, and the distribution of swallow timing across the breath cycle was significantly different from a uniform distribution (Fig. S3). In contrast, at low flow, infants typically swallowed once per respiratory cycle, typically occurring close to the transition between expiration and inspiration, although this distribution was not significantly different from a uniform distribution (Fig. 7B, Table 1).

Figure 7:

Swallow timing as a percent of suck cycle (A) and swallow timing as a percent of breath cycle (B) in infants feeding under high flow rate (light green) and low flow rate (dark green) conditions. The width of the violin plots at a given value indicates the density (i.e. clustering) of the data points at that value. Arrows indicate clustering in the data at time points when swallows occurred more often. Solid lines between groups indicates a statistically significant difference with a large effect size. The dashed line in (B) indicates the average timing of the end of inspiration and the start of expiration during the breath cycle.

Discussion

Overall, we found that altering flow rate significantly influences multiple aspects of infant feeding across several physiologic functions, including how sucking, swallowing, and respiration are achieved, as well as coordinated, which is a key factor in successful feeding (79). Our data demonstrate that modifying sensory input by changing flow rates leads to substantial changes in physiologic function, ultimately affecting overall feeding performance and mechanics. Notably, infants feeding at a lower flow rate were associated with more stable and coordinated feeding behaviors, suggesting that a reduced flow may better support the development of integrated sensorimotor patterns. These findings imply that a slower milk flow rate may facilitate safer and more effective feeding in early development.

Changes in flow rate impact sucking physiology and performance

The rate of sucking was not impacted by changes in flow (Fig. 3A). This aligns with our expectations, as sucking rate is regulated by a network of neurons (central pattern generators) in the brain stem reticular formation that produce rhythmic motor outputs (80–82). Other studies have also found that sucking rate is invariable despite changes in feeding conditions (35, 83). Recent work has demonstrated that chronic exposure to slower-flow nipples may alter sucking rate to a small degree, although differences were small (less than 0.1Hz, Crandall et al), further suggesting that the central pattern generator controlling the rhythm of suckling is not responsive to changes in feeding condition.

In contrast to the rate of sucking, several physiologic functions that are critical during sucking varied in response to dynamic changes in milk flow. Infants generated greater intraoral pressure when feeding at a low flow rate (Fig. 3B), suggesting a compensatory mechanism to counteract the slower milk influx. This has also been demonstrated in experiments that use different flow rate nipples or use nipples that do not allow for milk to be acquired through expression (21, 33, 58). The regulation of milk acquisition through modulation of pressure generation has been hypothesized to be the primary means by which infants adjust their suckling biomechanics, and it has been suggested that this regulation further results in better suck-swallow-breathe coordination with a reduced likelihood of aspiration (4, 21).

We also found that the middle and posterior tongue markers translated further during feeding at higher flow rates (Fig. 3C & Fig. 5A). These regions of the tongue have been hypothesized to play different roles, with the middle tongue acting as a pump to generate suction and then transport milk to the back of the throat, while the posterior tongue primarily contributes to the swallow (68, 84). Our data suggests that while the function of the middle tongue is to generate suction, greater tongue movement does not correlate to greater suction when the nipple's flow rate is high. Increased translation of the middle tongue at high flow is likely because the high flow of milk equilibrates intraoral suction quickly and, therefore, the tongue can move with less resistance. Essentially, when flow rates are high, suction does not build as quickly for a given amount of tongue movement, as pressure differentials can equilibrate more rapidly. In contrast, increased movement of the posterior tongue at high flow is likely driven by larger volumes of milk being swallowed, as the movement and position of the posterior tongue is the primary means by which changes in bolus volume are accommodated (68).

Previous work has demonstrated that changes in nipple hole size can impact relationships between tongue movement, suction generation, and milk acquisition (33, 34), and our data points to a potential explanation for why those differences may arise. At lower flow, the pattern of tongue movement during suckling was altered, in addition to the magnitude of movement, indicating that infants respond to changes in nipple properties (i.e. hole size) through subtle changes in physiology and mechanics, without altering their overall rate of suckling. Therefore, this understanding is crucial for guiding future investigations into the physiological adaptability of infant feeding and informing the development of targeted interventions for diverse feeding challenges.

Swallow physiology is impacted by flow rate

While changes in nipple properties typically have a direct impact on sucking, they also often have downstream impacts on swallow physiology (33, 34). The increase in milk flow and middle tongue movement during sucking at high flow resulted in increased swallow frequencies (Fig. 4A). The internal branch of the superior laryngeal nerve sends sensory signals from the larynx and upper trachea to the brain, and is responsible for triggering the swallowing reflex (85, 86). The increased swallow frequency at a higher flow rate suggests that the internal branch of the superior laryngeal nerve is responding to the increase in milk intake by swallowing more frequently in an attempt to limit bolus size. In addition to increased swallow rates, we found that infants swallowed bigger boluses when feeding at a higher flow rate (Fig. 4B). This is similar to previous work investigating the impact of nipple hole size on bolus volume and suggests that the increased swallow rate at high flow does not fully compensate for increased milk intake (33). This has implications for infant health, as several studies have demonstrated that infants swallowing larger boluses are at a higher risk of aspiration, as the accumulation of a large bolus of milk may compel the infant to perform each swallow more quickly, resulting in less time to protect the airway (62, 86).

We also found that the hyoid and thyroid translated more during swallowing when flow rate was higher (Fig. 5B & 5C). This increase in hyoid movement stems from the fact that the hyoid bone helps control how the pharynx muscles function during swallowing, likely increasing the opening of the upper esophageal sphincter (UES) to accommodate a larger bolus (87). Movements of the thyroid are often typically coordinated with the hyoid in healthy individuals (88) and are responsible for facilitating the proper closing of the vocal cords and the epiglottis, which directs food and liquid into the esophagus and away from the airway (89). Thus, the increased movements of both the hyoid and thyroid are likely driven by larger bolus volumes per swallow and reflect an active adjustment by infants in order to feed more efficiently and safely. Effective bolus control, in turn, enables the infant to continue sucking without interruption. Therefore, modifications in swallowing behavior directly support and influence the continuity of sucking, demonstrating the temporal and spatial integration between the two behaviors.

Respiratory physiology and coordination

Another important aspect of safe infant feeding is the ability to breathe during feeding by coordinating the respiratory cycle with swallow timing (79, 90). This is especially critical as infants' respiratory control is developing, and infants can be periodic breathers (91). While infants did not respond to changes in flow by altering their respiratory rate (Fig. 6A), we found that they did take deeper breaths when feeding at the lower flow rate, similar to what has been observed at lower flow in human infants ((92); Fig. 6B). As oxygen saturation of the blood can decrease during feeding, especially on a bottle (8), the ability to take deeper breaths at lower flows may result in increased oxygen saturation. Maintaining high oxygen saturation is critical because it enables infants to continue to rhythmically feed and not fatigue (93).

Respiration is not only crucial on its own, but it must also be coordinated with swallowing for feeding to be safe and effective. In addition to impacting sucking, swallowing, and breathing as isolated behaviors, dynamic changes in flow rate also impacted suck-swallow-breathe coordination. Suck-swallow coordination is a key metric in infant feeding because a well-coordinated suck-swallow pattern ensures effective milk extraction (94). While there is limited research on when is the safest time for infants to swallow within the suck cycle, it has been shown that more milk accumulating in the oropharynx can result in an increased risk of aspiration, similar to data demonstrated that larger boluses are correlated with increased risks of aspiration (46). We found that when feeding at high flow, swallows occurred earlier within the suck cycle (Fig. 7A), which is likely because the large size of the bolus triggered the swallow reflux earlier (66).

Because the pathways for food and air cross in mammals, coordinating swallowing with breathing is critical for ensuring that infants can safely consume milk while maintaining proper respiratory patterns (89). Proper swallow-breathe coordination ensures that the airway closes at the right time during swallowing to prevent aspiration, alongside allowing infants to acquire enough oxygen while feeding (36, 99). We found that infants feeding at a high flow rate generally swallowed twice during the breath cycle, one occurring mid-inspiration and the other mid-expiration. In contrast, when flow rate was lower, infants generally only swallowed once within the breath cycle during the transition from inspiration and expiration (Fig. 7B). In adults, swallowing mid-expiration is the safest time to swallow within the breath cycle (96), however, this pattern has yet to be established in infants. Future work should investigate whether the timing of swallowing during the breath cycle in infants dictates the likelihood of aspiration occurring.

Limitations

While these data suggest that changing sensory input during feeding results in changes in physiological function, there are several limitations of this work. First, because we used acute exposure to changes in flow, we were unable to test whether motor learning occurs. Evaluating the potential for motor learning to occur in feeding physiology would demonstrate if this experimental design could be used as an intervention for infants that experience feeding difficulties. Motor learning principles are frequently employed in the context of locomotor rehabilitation, and have been demonstrated to operate in the context of adult feeding (43, 97), but have not been evaluated for infants. Furthermore, this study only utilized healthy infants, which is likely not the population that would benefit from utilizing altering flow rate as an intervention, and future work should evaluate if pathophysiologic populations would exhibit similar responses. For example, preterm infants are neurologically distinct from those born at term, and thus may not respond to interventions or perturbations in the same way (98, 99). Finally, we measured tongue movements at discrete points along the tongue, and future research evaluating movement across the entire tongue might further reveal how this hydrostat changes its function depending on variation in nipple properties (100).

Conclusions

Our investigation into dynamically altering milk flow rate during infant feeding allowed us to assess how the infant’s brain interacts with and responds to changes in sensory input, and revealed significant and multifaceted impacts on sucking and swallowing mechanics, respiratory patterns, and crucially, the coordination between these vital behaviors. While sucking rate remained consistent, likely governed by central pattern generators, we observed significant differences in intraoral pressure and tongue kinematics in response to flow variation. These data highlight the capacity of infants to modulate physiologic function based on sensory input. Furthermore, changes in flow experienced during sucking directly influenced swallow frequency, bolus volume, and hyoid-thyroid movement, underscoring the integration between sucking and swallowing. Although respiratory rate remained stable, infants took deeper breaths at lower flow rates, possibly contributing to better blood oxygenation. Most notably, we found that suck-swallow-breathe coordination was sensitive to flow rate, with lower flows associated with earlier swallows within the suck cycle and distinct swallow-breathe patterns. These findings collectively support the premise that manipulating sensory input through flow rate can induce substantial changes in physiologic functioning and overall feeding mechanics. Our work provides valuable insights into the sensorimotor integration underlying infant feeding and lays the groundwork for investigating the therapeutic potential of dynamic flow rate modulation in vulnerable infant populations.

Supplementary Material

All supplemental material (Figures S1 – S3 and Table S1)are available at fighshare: doi 10.6084/m9.figshare.29666279.

Figure 8.

Data Availability:

Data used in statistical analyses are available at figshare: 10.6084/m9.figshare.29660963.