Swallow Reaction Time in Healthy Adults

Abstract

Purpose:

Swallow onset is commonly characterized by bolus location. Quantifying this swallow event using swallow reaction time (SRT) may further inform swallow assessment, but few studies have established normative reference values for SRT using a large sample size and well-tested, standardized, and validated videofluoroscopic swallow study (VFSS) approach. The purpose of this study was to investigate SRT in a large cohort of healthy adults and the effects of demographic (age, sex) and bolus (viscosity, volume) characteristics on SRT using the Modified Barium Swallow Impairment Profile approach.

Method:

Archival VFSSs of 195 healthy adults (21–89 years) were analyzed to obtain SRT across seven standardized swallow tasks ranging from thin to extremely thick liquid and a solid. Descriptive statistics were generated for age, sex, viscosity, and volume. Generalized estimating equation models were computed to explore the effects of demographic and bolus characteristics on SRT.

Results:

Variability in SRT was observed among healthy adults across all swallow tasks. Only viscosity significantly influenced SRT. Specifically, thicker viscosities resulted in longer SRT.

Conclusions:

Wide variability in SRT was observed in this large cohort of healthy adults across various liquid swallow tasks and a solid task. Thicker viscosities, in particular, yielded longer SRT and should be considered a normal variant. This study further provides normative SRT data related to a commonly used VFSS approach, which clinicians can use as reference values for their patient data.

Swallow onset is a sensory-dependent event that demarcates the time and point when the intricate and overlapping motoric swallow response triggers within the upper aerodigestive system (Steele & Miller, 2010). This phenomenon can be visualized in videofluoroscopic swallow studies (VFSSs) as the first anterosuperior movement of the hyoid (Dodds et al., 1990; Herzberg et al., 2019; Kendall & Leonard, 2001; Martin-Harris et al., 2007). Historically, a qualitative description of the location of the bolus head relative to the hyoid “burst” has been used to differentiate between a timely and delayed swallow. The literature often describes a “delayed” swallow occurring when the bolus head passes the mandibular ramus prior to swallow onset (Kim et al., 2005; Robbins et al., 1992). However, the literature also reports variability in bolus location at the time of swallow onset in healthy adults and can be influenced by a variety of factors, such as age and bolus task (Bhutada et al., 2020; Humbert et al., 2018; Martin-Harris et al., 2007; Steele et al., 2019). A more advanced position of the bolus head at the point the hyoid burst (HYB) occurs, however, does not appear to impact swallowing safety in healthy adults (Bhutada et al., 2020; Garand et al., 2019; Herzberg et al., 2019; Martin-Harris et al., 2007; Steele et al., 2019; Stephan et al., 2005) but is associated with increased airway invasion in patient populations (Bingjie et al., 2010; Kim & McCullough, 2007).

Using biomechanical metrics like temporal measures on VFSS enhances the quantification and measurement of the type and degree of swallowing impairment (Martin-Harris et al., 2020; Molfenter & Steele, 2014; Wei et al., 2022). However, discrepancies exist in how swallow onset has been measured, including which anatomical landmark to use as the starting point. For example, some studies use the bolus passing the posterior nasal spine (Kendall & Leonard, 2001; Kendall et al., 2000, 2001; Leonard & McKenzie, 2006), while others utilize the mandibular ramus (Humbert et al., 2018; Steele et al., 2019), or even the hypopharynx (Power et al., 2007). While subtle, these differences in anatomical starting points directly affect the time to swallow onset, such that more caudally oriented landmarks will increase this measure. Thus, comparing data to the different normative values in the literature might reveal variability based solely on methodological differences, making clinical interpretation difficult. The terminology used to indicate this timing measure is also variable, including “pharyngeal delay time” (Logemann et al., 2000), “stage transition duration” (Robbins et al., 1992), “initiation of the pharyngeal swallow” (Martin-Harris et al., 2007), “swallow response time” (Power et al., 2007), and “swallow reaction time” (SRT; Humbert et al., 2018; Steele et al., 2019). Henceforth, the term SRT will be used for this article as it is the most recently used term in the literature.

Clinically, delayed swallow onset remains a critical component contributing to airway invasion for patients with dysphagia. Vose et al. (2018) reported that most speech-language pathologists (SLPs; 88%, N = 267) accurately identified SRT as the primary physiological swallowing impairment when shown an “easy” VFSS, defined as only having one physiological problem. Yet, only 6% (N = 11) of 191 SLPs who completed a “moderately” complex VFSS with multiple physiological problems correctly identified SRT as the main impairment, which they defined as airway invasion caused by swallow onset occurring “outside the reported upper 95% confidence interval for SRT (.54 seconds).” The authors postulated that there might be several reasons for SLPs’ lack of uniformity in diagnostic and treatment selection, such as an excessive focus on bolus flow instead of temporal and kinematic swallowing outcomes (Vose et al., 2018). Thus, employing a temporal-based, quantitative method for measuring swallow onset may be more clinically useful than a qualitative description of bolus depth at the time of swallow onset. Using SRT would likely decrease diagnostic inaccuracies, particularly the confusion of when pharyngeal swallow onset is “delayed,” by providing clinicians with normative data, they can compare to patient values. Few large-scale studies have established normative SRT data. In particular, two of the most recognized and recent datasets for SRT are derived from Humbert et al. (2018) and Steele et al. (2019).

Humbert et al. (2018) reported the following normative SRT values, which are provided in milliseconds (ms): 212.12 (± 579.54) for 10 ml of ultrathin liquid, 1,101.31 (± 1,590.06) for a teaspoon of barium ice chips, 827.97 (± 1,448.47) for a teaspoon of room-temperature pudding, 1,070.12 (± 1,410.51) for a teaspoon of frozen pudding, and 1,064.09 (± 2,791.52) for a teaspoon of ultrathin liquid with chocolate chips. In addition, the authors explored the effects of age, sex, and bolus type, with effects found across all factors. Specifically, SRT was longer in older than younger adults, males compared to females, and with mixed consistencies compared to ultrathin liquid (Humbert et al., 2018).

Steele et al. (2019) collected the following SRT values for uncontrolled volumes of various liquid consistencies based on the International Dysphagia Diet Standardisation Initiative, ranging from thin to extremely thick liquids (in ms): 109 (± 177) for thin liquid; and 178 (± 269) for slightly thick, 198 (± 278) for mildly thick, 307 (± 402) for moderately thick, and 347 (± 471) for extremely thick liquids. The investigators explored the influence of sex, viscosity, and volume. They failed to find significant effects of sex and volume. However, viscosity effects were found, notably that moderately and extremely thick liquids significantly increased SRT compared to the other bolus consistencies (Steele et al., 2019). Results of other studies that have explored influential factors on SRT in healthy adults are summarized in Table 1. Note that there are discrepancies in results across each factor examined, which may be attributed to methodological constraints. This may include different samples sizes; varying swallow tasks; inconsistent operational definitions of SRT; and a lack of transparency in defining all aspects of VFSS analysis, such as how bolus aggregation of highly viscous consistencies was handled as this phenomenon has shown to increase the latency of swallow onset (Matsuo & Palmer, 2008; Palmer, 1998). Inconsistent study procedures make it challenging to systematically compare the results of independent studies and determine the consistency and strength of effects.

Table 1.

Factors explored in SRT.

Factor	Impact on SRT	Citation
Age	Increased with age	Byeon and Koh (2016); Humbert et al. (2018); Kendall and Leonard (2001); Kim et al. (2005); Leonard and McKenzie (2006); Mancopes et al. (2021); Nagaya et al. (1998); Robbins et al. (1992)
	Increased with age (all participants were males)	Logemann et al. (2000)
	No significant difference between younger and older adults (all participants were females)	Logemann et al. (2002)
	No significant difference between younger and older adults	Kang et al. (2010)
Sex	Increased with males compared to females	Humbert et al. (2018)
	Increased with females compared to males when manometric catheter in situ, but no significant difference without catheter	Robbins et al. (1992)
	No significant difference between males and females	Kim et al. (2005); Mancopes et al. (2021); Molfenter and Steele (2013); Nagy et al. (2013); Steele et al. (2019)
Viscosity	Increased with mixed consistency compared to thin liquid	Humbert et al. (2018)
	Increased with semisolids compared to liquids	Robbins et al. (1992)
	Increased with thicker liquids	Steele et al. (2019)
	No significant difference between thin liquid and paste	Kendall et al. (2001)
	No significant difference between thin liquid and mildly thick liquid	Oommen et al. (2011)
Volume	Increased with larger volume	Chi-Fishman and Sonies (2000)
	Decreased with larger volume	Kendall and Leonard (2001)
	No significant effect of volume	Kendall et al. (2000); Mancopes et al. (2021); Oommen et al. (2011); Power et al. (2007); Steele et al. (2019)
Cueing	Decreased with verbal cues to swallow	Daniels et al. (2007)
	No significant difference between cued and noncued swallows	Nagy et al. (2013)
Temperature	Decreased with room temperature bolus compared to cold bolus	Humbert et al. (2018)

Note. SRT = swallow reaction time.

The studies by Humbert et al. (2018) and Steele et al. (2019) provide a good foundation of established normative reference values. However, some constraints exist among these studies. Both studies had relatively small sample sizes, with Humbert et al. (2018) including 41 healthy adults, while Steele et al. (2019) had 38 healthy adult participants. Small sample sizes can limit the ability to extrapolate these findings. In the study by Humbert et al. (2018), participants were clustered into two broad age categories, younger (18–59 years) and older (60 years or older) adults, with the younger adults having a wider age range than their counterparts. The broad age categories may not capture slight but meaningful differences associated with stages of aging. Regarding Steele et al. (2019), the normative reference data were based only on younger to middle-aged adults; thus, this dataset cannot be extended to older adults. In a subsequent report by Mancopes et al. (2021), the authors compared VFSS data (including SRT) of 38 older adults to the normative findings from Steele et al. (2019) to determine age effects. The authors found that SRT increased with age (Mancopes et al., 2021). However, similar limitations are apparent, including a small sample size, an upper age range of 82 years, and only uncontrolled thin liquids being tested.

Differences in swallow task protocols should also be considered. Humbert et al. (2018) used a novel, standardized swallow task protocol that focused on frozen and mixed consistencies but also included boluses at room temperature: 10-ml ultrathin liquid and a “heaping teaspoon” of pudding, meaning they approximated the volumes of pudding instead of measuring precise bolus weight. The authors also administered the ultrathin liquid via a straw, which could have resulted in a more posterior placement of the bolus upon entry into the oral cavity despite applying a cued oral bolus hold. Furthermore, Humbert et al. (2018) randomized the swallow task order, but individuals who received more viscous boluses early during the VFSS may have been influenced by pharyngeal residue. Steele et al. (2019) also used a unique standardized protocol, where the researchers presented thin liquid and slightly and mildly thick liquids by cup, and moderately and extremely thick liquids by teaspoon in order of ascending viscosity. However, they only used uncontrolled volumes across swallow tasks (Steele et al., 2019). Although choosing not to control volume may be viewed as a relative strength as it simulates real-world oral consumption, it may be difficult to delineate the effects of viscosity and volume when no bolus volumes are controlled.

Therefore, it may be beneficial to examine and compare SRT across a range of swallow tasks using both fixed and uncontrolled conditions using the Modified Barium Swallow Impairment Profile (MBSImP) protocol, a validated and standardized VFSS procedure that is widely accepted by clinicians and researchers internationally (Clain et al., 2022; Martin-Harris et al., 2008, 2017; Northern Speech Services [NSS], 2020). Normative SRT data have yet to be established using the MBSImP protocol, which could promote the use of commonly used barium consistencies, volume conditions, and presentation methods. Adding these data would allow for increased reproducibility of findings and exploration of the relationship between swallowing biomechanics and MBSImP components. It would also extend the range of findings to controlled and uncontrolled thin and mildly thick liquid volumes, and controlled moderately and extremely thick liquid volumes. Descriptive data for the standardized MBSImP solid task would further be a valuable contribution, as this bolus type is rarely examined in the literature (Kagaya et al., 2016).

The gaps in the existing evidence related to normative SRT, the discrepant findings regarding influential factors, and the limitations previously described indicate that further research contributions are warranted. Therefore, this study aimed to contribute a normative SRT reference value dataset using the MBSImP in a large sample of healthy adults. A secondary aim of this study was to explore the potential impact of demographic (age, biological sex [hereafter referred to as “sex”]) and bolus (viscosity, volume) characteristics on SRT. The authors hypothesized that SRT would increase with age and viscosity, but there would be no sex and volume effects.

Method

Participants

Data were derived from an existing normative VFSS dataset of 195 healthy individuals (109 females, 86 males) obtained from a study that received institutional review board (IRB) approval and was conducted at the Medical University of South Carolina. Participants ranged from 21 to 89 years (M = 47, SD = 17.4) and were classified into three age groups (21–39 years [younger adults], 40–59 years [middle-aged], and 60 years and older [older adults]). Participants were included if they (a) were at least 21 years old, (b) presented with grossly intact cognition (judged subjectively by study personnel or achieved a score ≥ 26 on the Montreal Cognitive Assessment; Nasreddine et al., 2005), (c) had no history of dysphagia, (d) consumed an unrestricted diet, and (e) signed informed consent approved by the university’s IRB. Excluded from the study were individuals who presented with a current or history of one or more of the following: (a) neurological condition, (b) pulmonary disease, (c) head and neck cancer, (d) hiatal hernia (> 2 cm), (e) barium allergy, (f) pregnancy or suspected pregnancy, and (g) surgery of the head and/or anterior neck. Participant demographics can be found in Table 2.

Table 2.

Participant demographics organized by age group.

Variable	21–39 years (n = 70)	40–59 years (n = 70)	≥ 60 years (n = 55)
Age (years), M (SD)	28.2 (4.6)	48.9 (6.2)	68.7 (8)
Sex
Female	36 (51)	35 (50)	38 (69)
Male	34 (49)	35 (50)	17 (31)
Race
White/Caucasian	50 (71)	44 (63)	50 (91)
African American	16 (23)	23 (33)	5 (9)
Other	4 (6)	3 (4)	0 (0)

Note. Values provided are measured in frequency (percent) unless otherwise reported.

VFSS

Each participant completed a VFSS, which adhered to the MBSImP protocol. Varibar barium sulfate (Bracco, E-Z-EM, Inc.), 40% weight/volume (w/v), was used to promote uniformity of barium opacification and concentration of viscosities for the VFSS. Seven swallow tasks viewed in the lateral plane were analyzed: 5 ml and cup sip (approximately 20 ml) of thin liquid (< 15 cP [cps]), 5 ml and cup sip (approximately 20 ml) of mildly thick liquid (< 150–450 cps), 5 ml of moderately thick liquid (< 800–1,800 cps), 5 ml of extremely thick liquid (4,500–7,000 cps), and ½ of a Nabisco Lorna Doone cookie coated in 3 ml of extremely thick barium (Martin-Harris et al., 2008, 2017). Thin liquid and mildly thick liquid were presented via teaspoon and a self-administered (uncontrolled) cup sip. Moderately and extremely thick liquids were administered via teaspoon. Participants were cued to hold the 5-ml and cup sip liquid boluses in their mouth until instructed to swallow when ready, except for the extremely thick liquid and solid tasks, whereby participants were not cued to perform an oral hold and were simply instructed to swallow when ready. The MBSImP incorporates a cued oral hold for selected liquid swallow tasks previously listed as this step is of diagnostic value, allowing clinicians to judge the patient’s oral bolus control before the first productive anteroposterior tongue movement oriented toward swallow initiation (Martin-Harris et al., 2008). Clinicians can therefore differentiate between oral tongue control and swallow onset, especially for consistencies prone to naturally escape posteriorly into the pharynx. Sequential swallow tasks were excluded from the current analysis as these are part of a separate ongoing study aimed at establishing comprehensive normative biomechanical measures for sequential swallowing, which is largely understudied and may best be defined in the context of its own varying motor strategies distinct from discrete swallows as suggested by Chi-Fishman and Sonies (2000).

A fluoroscopic device ascertained images under continuous fluoroscopy, and a digital recording device (Digital Swallow Workstation, Model 7100, Kay Electronics Corp.; TIMS Medical) was used to record swallowing images at a rate of 30 frames per second. A total of 1,361 swallows were eligible for analysis. Three swallows were missing, and one swallow was excluded because of poor image quality. Each swallow was analyzed using ImageJ by a single, trained rater (K.R.A.) blinded to bolus type and demographic characteristics.

Operational Definitions

Consistent with recent studies (Humbert et al., 2018; Steele et al., 2019), SRT was defined as the timeframe between the bolus passing the mandible (BPM) and the HYB. This definition is further specified using the parameters of the ASPEKT (Analysis of Swallowing Physiology: Events, Kinematics, and Timing) Method (Steele et al., 2019), which are briefly described below:

• BPM: This is the first frame where the bolus head contacts or passes the mandibular ramus leading to a swallow (see Figure 1). For swallow tasks that demonstrated bolus aggregation (i.e., typically highly viscous consistencies and the solid), the first frame where the bolus escaped into the pharynx (at or below the mandibular ramus) was considered the BPM frame because it is a tongue-driven phenomenon (Palmer, 1998). If a double mandibular shadow was observed (see Figure 2), the inferior edge of the more superior mandibular ramus was used as the BPM frame. Residue from a previous swallow task was not considered, and only material from the current task was considered the BPM frame.

• HYB: This is the first frame where the hyoid makes the initial brisk movement toward maximal anterosuperior displacement during the swallow (see Figure 3). Any other minimal movements, such as those typically observed during bolus manipulation, mastication, or bolus aggregation, were not considered the HYB frame.

Figure 1.

Bolus past mandible. Bolus past mandible frame = 50. Yellow line shows the first frame where the bolus passes the mandibular ramus.

Figure 2.

Double mandibular shadow.

Figure 3.

Hyoid burst. Hyoid burst frame = 49. Yellow box depicts initial hyoid movement in an anterosuperior trajectory, resulting in blurring of the structure.

SRT Calculation

The BPM and HYB frames were recorded. SRT was initially calculated in frames by subtracting the BPM frame from the HYB frame. To convert SRT from frames to ms, the SRT in frames value was divided by 30 and multiplied by 1,000. A negative SRT value refers to the HYB occurring before the BPM, while a positive value indicates that the HYB occurred after the BPM. When BPM and HYB co-occur, the SRT value is zero.

Statistical Analyses

For reliability, Rater 1 (K.R.A.) analyzed 20% of the swallows at two points separated by 1 month to establish intrarater reliability. A second trained rater analyzed the same subset to determine interrater reliability. An intraclass correlation coefficient (ICC) analysis was performed for all analyzed swallows. Descriptive statistics were calculated for all measures of interest. Based on continuous, repeated measures within subjects, a generalized estimating equations (GEE) framework was used to estimate the effects of age, sex, viscosity, and volume (Pekár & Brabec, 2018). For the demographic GEE model, age, sex, and all swallow tasks were factored to account for potential confounding. For the viscosity GEE model, volume was collapsed, meaning only 5-ml tasks were included (thin liquid and mildly, moderately, and extremely thick liquids). By only comparing swallow tasks that can be equally matched for volume differences (5-ml and cup sip thin liquid and mildly thick liquid tasks), the potential of viscosity being a confounder was considered for the volume GEE model. Thus, the volume GEE model was only tested with thin liquid and mildly thick liquid. The solid task was excluded from GEE models that examined bolus characteristic effects to control for significant expected differences between solids and liquids. An exchangeable structure for the working correlation matrix was assumed for all GEE models. Pairwise comparisons were calculated for any significant variables found in the GEE models to explore the relationship between predictor variable levels further. Using Spearman rho, a post hoc analysis was performed to explore the relationship between SRT derived across swallow tasks from this study and previously reported scores for MBSImP Component 6 (initiation of the pharyngeal swallow; Bhutada et al., 2020) and the Penetration Aspiration Scale (PAS; Garand et al., 2019). A Bonferroni-corrected alpha (.01) was used to adjust the alpha level for multiple comparisons. All statistical analyses were conducted using the Statistical Package for Social Science Version 28 (IBM Corp, 2021).

Results

Reliability

Based on the ICC criteria suggested by Koo and Li (2016), excellent reliability was achieved for intra- (.99) and interrater (.97) reliability on the analyzed swallows.

Descriptive Statistics

Wide SRT ranges were observed for all variables explored. This trend can be seen in the SRT values (in ms) for the aggregate data, ranging between −566.67 and 28,866.67, with an average of 651.51 and a large variance (SD = 1911.04). Complete descriptive statistics for the aggregate data can be found in Table 3. Variability in descriptive SRT values was also noted when the data were analyzed by age group, sex, viscosity, and volume, which are displayed in Table 4. Additional descriptive information for swallow tasks and swallow tasks factored by age and sex can be found in the appendices.

Table 3.

Descriptive overview of aggregate SRT data.

M ± SD	Mdn	Range (minimum, maximum)	95% CI (lower, upper)
651.51 ± 1911.04	33.33	−566.67, 28866.67	549.89, 753.13

Note. Data are reported in milliseconds (ms). SRT = swallow reaction time; CI = confidence interval.

Table 4.

Descriptive overview of SRT across predictor variables.

Variable	Parameter	M ± SD	Mdn	Range (minimum, maximum)	95% CI (lower, upper)
Age group (years)	21–39	519.22 ± 1516.68	33.33	−200.00, 10666.67	[384.46, 653.98]
	40–59	702.60 ± 1757.43	66.67	−266.67, 13900.00	[546.28, 858.91]
	60+	755.03 ± 2464.46	66.67	−566.67, 28300.00	[507.76, 1002.31]
Sex	Female	724.95 ± 2132.98	66.67	−566.67, 28300.00	[572.96, 876.93]
	Male	558.91 ± 1584.54	33.33	−266.67, 13900.00	[432.08, 685.75]
Viscosity	Thin liquid	161.61 ± 375.30	33.33	−266.67, 2833.33	[124.20, 199.02]
	Mildly thick liquid	142.55 ± 340.92	.00	−566.67, 2500.00	[108.48, 176.62]
	Moderately thick liquid	240.68 ± 589.52	.00	−333.33, 4400.00	[157.42, 323.95]
	Extremely thick liquid	462.74 ± 705.54	66.67	−233.33, 3366.67	[363.09, 562.38]
	Solid	3238.46 ± 4039.01	1833.33	−166.67, 28300.00	[2668.00, 3808.92]
Volume	5 ml	127.05 ± 274.93	33.33	−566.67, 1600.00	[99.57, 154.52]
	Cup sip	177.04 ± 424.52	33.33	−266.67, 2833.33	[134.72, 219.35]

Note. Data are reported in milliseconds (ms). For viscosity, volume (5 ml and cup sip) was collapsed for thin liquid and mildly thick liquid. For volume, thin liquid and mildly thick liquid were collapsed for 5 ml and cup sip. SRT = swallow reaction time; CI = confidence interval.

GEE Analyses

Influence of Demographic Factors (Age, Sex) on SRT

Age.

There were no significant differences in SRT between age groups, Wald χ²(2, N = 1,361) = 4.76, p = .093. Employing age as a continuous variable also failed to reach significance, Wald χ²(1, N = 1,361) = 5.53, p = .019.

Sex.

Sex did not reveal a significant main effect, Wald χ²(1, N = 1,361) = 2.62, p = .11. The parameter estimates for age and sex are reported in Table 5.

Table 5.

GEE parameter estimates for age group and sex.

Parameter	Intercept	Standard error	95% Wald CI (lower, upper)	Wald chi-square	df	p value
(Intercept)	57.82	67.37	[−74.23, 189.87]	.74	1	.39
60+ years	209.49	121.29	[−28.23, 447.21]	2.98	1	.084
40–59 years	184.36	104.58	[−20.62, 389.34]	3.11	1	.078
21–39 years	0a
Male	−147.24	90.99	[−325.59, 31.10]	2.62	1	.10
Female	0a
Solid	3120.31	286.06	[2559.64, 3680.98]	119.98	1	< .001
5-ml extremely thick liquid	344.58	44.61	[257.16, 432.01]	59.68	1	< .001
5-ml moderately thick liquid	122.53	36.93	[50.14, 194.92]	11.01	1	< .001
Cup sip mildly thick liquid	30.49	26.11	[−20.69, 81.67]	1.36	1	.24
5-ml mildly thick liquid	18.03	20.58	[−22.31, 58.37]	.77	1	.38
Cup sip thin liquid	87.32	30.82	[26.91, 147.73]	8.03	1	.005
5-ml thin liquid	0a

Note. Superscripted a is set to zero as this parameter is the referent level and would be redundant to report. GEE = generalized estimating equations; CI = confidence interval.

Influence of Bolus Characteristics (Viscosity, Volume) on SRT

Viscosity.

Viscosity significantly influenced SRT, Wald χ²(3, N = 777) = 60.36, p < .001 (see Table 6). Specifically, moderately and extremely thick liquids significantly increased SRT (p < .001) compared to thin liquid, with a 122.23-ms mean difference between moderately thick liquid and thin liquid and a 344.28-ms mean difference between extremely thick liquid and thin liquid. Extremely thick liquid resulted in the most prolonged SRT (M = 462.74 ms) compared to the other liquid viscosities (in ms): thin liquid (M = 118.45), mildly thick liquid (M = 135.47), and moderately thick liquid (M = 240.68). Pairwise comparisons revealed significant differences (p < .01) between all liquid viscosity combinations, except for the comparison between thin liquid and mildly thick liquid (p = 1.00) and moderately and mildly thick liquids (p = .031). Overall, thicker viscosities generally increased SRT compared to less viscous liquids. Note that age and sex were excluded from this model, given their lack of significance from the prior demographic model.

Table 6.

GEE parameter estimates for viscosity.

Parameter	Intercept	Standard error	95% Wald CI (lower, upper)	Wald chi-square	df	p value
(Intercept)	118.46	16.78	[85.57, 151.35]	49.83	1	< .001
Extremely thick liquid	344.28	44.61	[256.86, 431.70]	59.57	1	< .001
Moderately thick liquid	122.23	36.96	[49.78, 194.68]	10.94	1	< .001
Mildly thick liquid	17.02	20.69	[−23.52, 57.56]	.68	1	.41
Thin liquid	0a

Note. Superscripted a is set to zero as this parameter is the referent level and would be redundant to report. Data are based only on 5-ml swallow tasks. GEE = generalized estimating equations; CI = confidence interval.

Volume.

No significant difference was observed between 5 ml and cup sip tasks, indicating no volume effects on SRT, Wald χ²(1, N = 776) = 5.62, p = .018. For the same reason as the viscosity model, age and sex were excluded. Parameter estimates for volume are reported in Table 7.

Table 7.

GEE parameter estimates for volume.

Parameter	Intercept	Standard error	95% Wald CI (lower, upper)	Wald chi-square	df	p value
(Intercept)	127.07	16.79	[94.16, 159.97]	57.27	1	< .001
Cup sip	50.15	21.16	[8.67, 91.64]	5.62	1	.018
5 ml	0a

Note. Superscripted a is set to zero as this parameter is the referent level and would be redundant to report. Thin liquid and mildly thick liquid were collapsed for 5 ml and cup sip. GEE = generalized estimating equations; CI = confidence interval.

Relationship Between SRT and MBSImP Component 6 and PAS Scores

There was a significant positive correlation between SRT and MBSImP Component 6 scores across all swallow tasks, all p < .001. The positive correlation suggests that SRT increases with more advanced positions of the bolus head at the time of swallow onset. However, no significant correlations were detected between SRT and PAS scores across all swallow tasks, all p > .01. Spearman rho values for all swallow tasks are provided in Table 8.

Table 8.

Correlations between SRT and MBSImP Component 6 and PAS scores.

Variable	Swallow task	Spearman rho	p value	N
MBSImP Component 6 Scores	5-ml thin liquid	.66	< .001	191
	Cup sip thin liquid	.61	< .001	195
	5-ml mildly thick liquid	.66	< .001	191
	Cup sip mildly thick liquid	.53	< .001	193
	5-ml moderately thick liquid	.61	< .001	194
	5-ml extremely thick liquid	.61	< .001	194
	Solid	.37	< .001	195
PAS Scores	5-ml thin liquid	−.001	.98	194
	Cup sip thin liquid	−.079	.28	195
	5-ml mildly thick liquid	.084	.24	193
	Cup sip mildly thick liquid	.062	.40	193
	5-ml moderately thick liquid	.078	.28	195
	5-ml extremely thick liquid	.12	.091	194
	Solid	.051	.48	195

Note. SRT = swallow reaction time; MBSImP Component 6 = initiation of the pharyngeal swallow on the Modified Barium Swallow Impairment Profile; PAS = Penetration Aspiration Scale.

Discussion

SRT is a common temporal measure of clinical interest since any delay can result in the bolus approaching an open airway, increasing the risk of airway invasion (Matsuo & Palmer, 2008). This study examined SRT in a large cohort of healthy adults across seven standardized swallowing tasks commonly employed during VFSS. All but the hypothesis regarding age effects were confirmed.

Variability in SRT Values

Inspection of SRT across all variables explored demonstrates the inherent wide variability in swallow onset among healthy adults. Current findings support previous works that also observed variability in SRT for discrete swallows (Humbert et al., 2018; Kendall & Leonard, 2001; Kendall et al., 2000, 2001; Leonard & McKenzie, 2006; Steele et al., 2019) and solid swallowing tasks (Kagaya et al., 2016). Ultimately, the variability found helps to elucidate differences in the perceived norm of pharyngeal swallow onset, where confusion frequently exists between a timely and delayed swallow. The variability in SRT suggests that healthy adults across demographic profiles can accommodate different extrinsic factors (i.e., viscosity and volume) or swallow task demands. This further demonstrates the robustness of the healthy physiological swallowing mechanism and its ability to respond to various conditions. However, this may pose challenges for older, disordered populations, who are typically associated with changes in swallow onset that may place them at greater risk for airway invasion (Gandhi et al., 2021; Kim & McCullough, 2007; Mancopes et al., 2020; Oommen et al., 2011; Power et al., 2007; Waito et al., 2018).

Influence of Demographic Factors (Age, Sex) on SRT

Age

Comparable to Kang et al. (2010) and Logemann et al. (2002), SRT was not significantly different between age groups. However, an inspection of the data in this study showed that the mean SRT values appeared to increase with each age group. Several previous studies observed significantly longer SRT in older adults (Byeon & Koh, 2016; Humbert et al., 2018; Kendall & Leonard, 2001; Kim et al., 2005; Leonard & McKenzie, 2006; Mancopes et al., 2021; Nagaya et al., 1998; Namasivayam-MacDonald et al., 2018; Robbins et al., 1992). Logemann et al. (2000) also reported longer SRT with increased age, but this finding was observed in male participants only. A direct comparison between this study and previous reports cannot be easily made due to various methodological differences, such as sample size, age distribution, operational definitions of SRT, swallow task protocols, and statistical analyses employed. For example, this study offers the largest sample of participants (N = 195) compared to previous studies, and larger sample sizes tend to reflect less variability (Well et al., 1990). Yet, Leonard and McKenzie’s (2006) study also included a relatively large sample size (N = 151) and observed significant differences between age groups. However, their swallow tasks were limited to 20-ml thin liquid, and 1- and 3-ml paste boluses (Leonard & McKenzie, 2006). Studies also defined age categories differently or had varying age ranges, such as dichotomizing into younger (under 60 years) and older (60 years or older) groups (Humbert et al., 2018; Mancopes et al., 2021), while this study accounted for younger, middle-aged, and older adults.

Sex

Sex failed to reach significance as an influential factor of SRT. The current findings align with previous works (Kim et al., 2005; Mancopes et al., 2021; Molfenter & Steele, 2013; Nagy et al., 2013; Steele et al., 2019), although they disagree with results from Humbert et al. (2018) and Robbins et al. (1992). Humbert et al. (2018) revealed that significantly longer SRT was associated with males. While Robbins et al. (1992) found a similar pattern, these observations were found in females. The mixed findings may be due to variations in swallow tasks between these studies. Robbins et al. (1992) utilized only two swallow tasks in smaller volume amounts (2-ml thin liquid and 2-cm semisolid). Interestingly, the sex effect in the Robbins et al. (1992) study was only observed when a pharyngeal manometer catheter was in situ and, therefore, may have influenced the sex difference reported. The novel swallow tasks employed by Humbert et al. (2018) included mixed consistencies and boluses of different temperatures.

Influence of Bolus Characteristics (Viscosity, Volume) on SRT

Viscosity

Compared to thin liquid, moderately and extremely thick liquids significantly increased SRT. These results corroborate the findings by Steele et al. (2019). Oommen et al. (2011) also found comparable results in the lack of significant difference between thin liquid and mildly thick liquid, but in a cohort of patients poststroke. They postulated that thin liquid and mildly thick liquid might be more similar in sensory properties compared to more prominent differences between mildly thick liquid and even thicker viscosities that influence sensory stimulation and the bolus flow rate (Oommen et al., 2011). Similar to previous studies, bolus aggregation of highly viscous consistencies in the pharynx may explain the longer, but normal variants in SRT ranges found across these bolus tasks (Matsuo & Palmer, 2008; Palmer, 1998).

In contrast, Kendall et al. (2001) failed to observe significant differences in SRT between thin liquid and paste, but the latter resulted in a longer SRT mean value than the former. The disagreement between studies is again likely because of methodological variations, specifically, differences in VFSS analysis procedures, swallow tasks utilized, and sample size. In their cohort of 60 healthy adults, Kendall et al. (2001) used the posterior nasal spine to demarcate the entry of the bolus into the pharynx but did not describe how they handled bolus aggregation. However, this study defined this point as the mandibular ramus and specified that bolus aggregation was considered an intentional tongue-driven phenomenon (Matsuo & Palmer, 2008; Palmer, 1998). Furthermore, Kendall et al. (2001) only utilized thin liquid and paste, compared to the more comprehensive range of viscosities employed in this study and by Steele et al. (2019).

Volume

No significant differences were found between the 5-ml and cup sip conditions in this study, supporting previous findings regarding the lack of volume effects on SRT (Kendall et al., 2000; Mancopes et al., 2021; Oommen et al., 2011; Power et al., 2007; Steele et al., 2019). Nonetheless, a ceiling effect may have been reached because the cohort comprised only healthy adults. In the absence of perturbation, healthy adults do not depend on shorter SRT to facilitate safe and efficient swallowing. Because healthy adults can modulate swallow timing as needed to accommodate different bolus volumes, the mean SRT was generally longer, albeit nonsignificant, for the cup sip compared to the 5-ml tasks. The ability of healthy adults to modulate swallow onset in order to accommodate larger volumes has also been noted by Chi-Fishman and Sonies (2000), who found that larger volumes significantly increased SRT and were often associated with distal pharyngeal bolus positions at swallow onset during a 150-ml sequential swallow task. However, ingesting larger volumes has been significantly associated with shortening SRT in clinical populations with dysphagia due to increased sensory input provided to the oropharynx (Lazarus et al., 1993; Park et al., 2016). Thus, clinical populations may experience protective benefits from applying bolus volume modulation as a compensatory strategy. It is important to note that volume was uncontrolled for the cup sip condition in this study, with larger ingested volumes presumed for this task compared to the 5-ml condition. This assumption was based on previous research supporting an approximate 20 ml of liquid ingested for the former condition using the MBSImP approach (Martin-Harris et al., 2017) and a typical cup volume range of 10–14 ml reported by other studies (Bennett et al., 2009; Steele et al., 2019).

Relationship Between SRT and MBSImP Component 6 and PAS Scores

Among the clinically relevant results of this study was the significant positive correlation between SRT and MBSImP Component 6. These findings support the notion that visuo-perceptual analysis of swallow onset can provide relatively reliable clinical information since it was well correlated to pixel-based ASPEKT measurements of SRT. This positive relationship may be impacted by the rigorous, standardized, and systematic training necessary to use the MBSImP approach, which carefully reviews each swallowing component and requires a threshold of user reliability. Comments regarding correlations to other visuo-perceptual analysis methods using different operational definitions of swallow onset cannot be made as they are understudied. Nonetheless, pixel-based measurements of SRT still serve an important tool for clinicians to capture and monitor changes in swallowing function. Compared to pixel-based measures, ordinal measures of swallow onset, such as those used by the MBSImP, may not be as sensitive to detecting subtle changes.

Lastly, SRT did not correlate with PAS scores across all swallow tasks as anticipated. These finding are likely due to the cohort being made up of only healthy adults. Previous research has shown that healthy adults infrequently demonstrate airway invasion, especially aspiration (Garand et al., 2019). Although healthy adults normally protect their airways adequately, persons with dysphagia may experience greater challenges with appropriate timing and coordination of swallow onset, which needs to be carefully considered in such cases.

Limitations

The swallow protocol was limited in the number of trials per swallow (one each), making it difficult to assess potential variability and the integrity of the findings within subjects. The different cueing conditions in this study may have also impacted variability. For instance, participants were cued to perform an oral hold for the 5 ml and cup sip of thin liquid and mildly thick liquid and 5-ml moderately thick liquid tasks, while participants were not cued for the extremely thick liquid and solid tasks. Differences in cueing conditions across tasks may have affected the statistical comparisons generated for liquids, as prior research has shown that cued swallows tend to yield shorter SRT (Daniels et al., 2007). Lastly, the thin liquid and mildly thick liquid cup sip tasks were uncontrolled for volume despite analysis assuming such volumes were larger than the 5 ml task. However, the uncontrolled MBSImP swallow tasks have been previously associated with consuming larger volumes as previously specified (Martin-Harris et al., 2017).

Study Implications

Despite such limitations, only a few large-scale studies have explored SRT in healthy adults. This study is a helpful contribution as it demonstrates the replicability of some aforementioned results found by prior hallmark studies despite methodological differences and the use of independent data. Replicability in science is critical for traversing the literature when limited information exists, or it lacks consensus, such as the evidence regarding influential factors of SRT, as replicability strengthens the consistency and validity of results (Nosek et al., 2022). Compared to the two most recent and widely accepted normative SRT reference values (Humbert et al., 2018; Steele et al., 2019), this study adds the largest normative SRT dataset to the literature. Assuming the underlying populations sampled for each study were equivalent and that the same statistical methods were used to generate the results of each study, it is reasonable to assume that this study, with its sample size of 195 participants, will provide a more accurate representation of the larger population. Whereas Humbert et al. (2018) and Steele et al. (2019) only provided aggregate SRT reference values, this study generated specific descriptive values across various variables in addition to aggregate data, including a wide range of younger to older adults, sex, viscosity, volume, and swallow tasks organized by age and sex. Thus, as opposed to making comparisons based solely on decontextualized aggregate norms, the current data enable clinicians to locate SRT ranges specific to certain demographic and bolus characteristics.

Another important implication unique to this study is that the normative data are based on the MBSImP protocol, which has been shown to appropriately challenge the swallowing mechanism and differentiate impairment from typical functioning using a set of well-tested swallow tasks found to impact swallowing function (Clain et al., 2022; Hazelwood et al., 2017; Martin-Harris et al., 2008). Unfortunately, this cannot be definitively said for the swallow task protocols used by Humbert et al. (2018) and Steele et al. (2019), albeit they were also standardized and included a different range of swallow tasks. A unique feature of this study is that both controlled and uncontrolled volumes and a solid trial were used, which have not yet been explored and compared by the aforementioned studies. Furthermore, the results can be directly compared to the MBSImP components across several standardized swallow tasks, promoting increased quantification of these observations. Since the MBSImP is widely used by over 6,000 clinicians across several countries (NSS, 2020), including the facilities in which this study was conducted, the data provide clinically useful information that clinicians can readily apply to improve the diagnostic sensitivity of a commonly misdiagnosed physiological swallowing component. While specific cutoff ranges are suggested to determine impairment, clinicians should always consider the unique clinical presentation of each patient.

Conclusions

Current study findings support wide variability in SRT among healthy adults. Although age, sex, and volume were not observed to influence SRT significantly, viscosity effects were detected. Specifically, thicker liquid viscosities resulted in longer SRT than less viscous liquids. Clinicians can use this normative information to directly compare patient data, enhancing clinical decision-making. Future studies should explore SRT in additional tasks and conditions commonly employed during VFSS, such as sequential swallowing and head positioning modifications, and further establish reference values for disordered populations.

Appendix A

Table A1.

Descriptive Overview of SRT Across Swallow Tasks

Swallow task	M ± SD	Mdn	Range (minimum, maximum)	95% CI (lower, upper)
5-ml thin liquid	117.53 ± 234.85	33.33	−133.33, 1300.00	[84.27, 150.78]
Cup sip thin liquid	205.47 ± 471.19	66.67	−266.67, 2833.33	[138.78, 272.16]
5-ml mildly thick liquid	136.61 ± 310.38	.00	−566.57, 1600.00	[92.55, 180.68]
Cup sip mildly thick liquid	148.45 ± 369.52	.00	−233.33, 2500.00	[96.13, 200.78]
5-ml moderately thick liquid	240.68 ± 589.52	.00	−333.33, 4400.00	[157.42, 323.95]
5-ml extremely thick liquid	462.74 ± 705.54	66.67	−233.33, 3366.67	[363.09, 562.38]
Solid	3238.46 ± 4039.01	1833.33	−166.67, 28300.00	[2668.00, 3808.92]

Note. Data are reported in milliseconds (ms). SRT = swallow reaction time; CI = confidence interval.

Appendix B

Table B1.

Descriptive Overview of SRT Across Swallow Tasks Organized by Age Group and Sex

Age group (years)	Sex	Swallow task	M ± SD	Mdn	Range (minimum, maximum)	95% CI (lower, upper)
21–39	Female	5-ml thin liquid	62.04 ± 143.46	16.67	−100.00, 600.00	[15.83, 108.24]
		Cup sip thin liquid	140.74 ± 390.86	33.33	−66.67, 2100.00	[14.85, 266.63]
		5-ml mildly thick liquid	134.29 ± 252.50	33.33	−133.33, 1000.00	[50.72, 215.10]
		Cup sip mildly thick liquid	65.74 ± 157.26	.00	−133.33, 600.00	[15.09, 116.39]
		5-ml moderately thick liquid	229.63 ± 374.83	.00	−200.00, 1133.33	[108.90, 350.36]
		5-ml extremely thick liquid	457.41 ± 603.72	200.00	−166.67, 2100.00	[262.95, 651.86]
		Solid	2720.37 ± 3078.21	1366.67	−166.67, 10366.67	[1728.90, 3711.84]
	Male	5-ml thin liquid	40.20 ± 113.94	.00	−133.33, 433.33	[2.47, 77.93]
		Cup sip thin liquid	174.51 ± 497.20	33.33	−66.67, 2833.33	[9.86, 339.16]
		5-ml mildly thick liquid	20.59 ± 107.63	.00	−100.00, 400.00	[−15.05, 56.23]
		Cup sip mildly thick liquid	44.12 ± 162.62	.00	−66.67, 833.33	[−9.73, 97.97]
		5-ml moderately thick liquid	201.96 ± 642.65	−16.67	−133.33, 2933.33	[−10.85, 414.77]
		5-ml extremely thick liquid	321.57 ± 720.64	0.00	−133.33, 3200.00	[82.93, 560.21]
		Solid	2634.31 ± 3190.23	1166.67	−133.33, 10666.67	[1577.87, 3690.76]
40–59	Female	5-ml thin liquid	163.81 ± 288.83	100.00	−133.33, 1300.00	[69.50, 258.12]
		Cup sip thin liquid	240.95 ± 470.72	66.67	−66.67, 1700.00	[87.25, 394.65]
		5-ml mildly thick liquid	164.71 ± 323.17	.00	−100.00, 1466.67	[56.57, 268.13]
		Cup sip mildly thick liquid	139.22 ± 257.35	50.00	−233.33, 1033.33	[52.52, 221.20]
		5-ml moderately thick liquid	284.76 ± 790.93	33.33	−133.33, 4400.00	[26.50, 543.02]
		5-ml extremely thick liquid	709.52 ± 790.03	600.00	−100.00, 3233.33	[451.56, 967.49]
		Solid	3130.48 ± 3245.33	2100.00	−66.67, 13633.33	[2070.79, 4190.17]
	Male	5-ml thin liquid	179.05 ± 281.11	33.33	−66.67, 933.33	[87.26, 270.84]
		Cup sip thin liquid	297.14 ± 581.49	100.00	−266.67, 2500.00	[107.27, 487.01]
		5-ml mildly thick liquid	212.38 ± 388.82	33.33	−66.67, 1600.00	[85.42, 339.34]
		Cup sip mildly thick liquid	191.43 ± 415.34	.00	−100.00, 1966.67	[55.81, 327.05]
		5-ml moderately thick liquid	391.43 ± 755.30	33.33	−100.00, 3033.33	[144.80, 638.06]
		5-ml extremely thick liquid	434.29 ± 665.71	33.33	−133.33, 2133.33	[216.91, 651.66]
		Solid	3265.71 ± 3867.19	2166.67	−100.00, 13900.00	[2002.97, 4528.46]
60+	Female	5-ml thin liquid	123.42 ± 210.07	33.33	−100.00, 866.67	[55.11, 187.80]
		Cup sip thin liquid	156.14 ± 373.77	50.00	−100.00, 2166.67	[38.87, 273.41]
		5-ml mildly thick liquid	170.78 ± 404.54	16.67	−566.67, 1566.67	[43.26, 297.09]
		Cup sip mildly thick liquid	203.51 ± 410.29	33.33	−66.67, 1766.67	[74.79, 332.23]
		5-ml moderately thick liquid	180.70 ± 391.17	33.33	−333.33, 1433.33	[57.98, 303.42]
		5-ml extremely thick liquid	497.37 ± 804.06	66.67	−233.33, 3366.67	[245.11, 749.63]
		Solid	5101.75 ± 6094.96	3183.33	−166.67, 28300.00	[3189.54, 7013.97]
	Male	5-ml thin liquid	154.90 ± 338.90	.00	−133.33, 1200.00	[−1.39, 311.19]
		Cup sip thin liquid	252.94 ± 553.28	33.33	−100.00, 2133.33	[−2.21, 508.10]
		5-ml mildly thick liquid	86.27 ± 182.60	33.33	−100.00, 666.67	[2.07, 170.48]
		Cup sip mildly thick liquid	339.22 ± 746.59	33.33	−100.00, 2500.00	[−5.09, 683.52]
		5-ml moderately thick liquid	74.51 ± 252.91	.00	−166.67, 700.00	[−42.12, 191.14]
		5-ml extremely thick liquid	229.41 ± 407.55	.00	−133.33, 1200.00	[41.46, 417.36]
		Solid	1545.10 ± 1821.55	766.67	−166.67, 5100.00	[705.06, 2385.14]

Note. Data are reported in milliseconds (ms). SRT = swallow reaction time; CI = confidence interval.

Data Availability Statement

All data generated or analyzed during this study are included in this published article (and/or its supplemental material files, if applicable). Sharing of raw data are available upon written request to the corresponding author.