Breath-Holding Maneuvers and Airway Closure in Healthy Adults and Parkinson’s Disease: Effects of Maneuver Type, Health Status, and Vocal Fold Bowing Authors: James A. Curtis, Alexandra B. Beam, Jina Nam, Michelle S. Troche Published: 2025 ABSTRACT AbstractIntroductionBreath-holding maneuvers are frequently used to improve airway closure and swallowing safety in people with dysphagia. While their effects on airway closure have been studied in healthy adults, little is known about their effects in people with Parkinson’s disease (PwPD). This study aimed to: (1) characterize glottic and supraglottic airway closure during light and hard breath-holding maneuvers in healthy adults and PwPD; (2) determine whether glottic and supraglottic closure changes as an effect of maneuver type (light vs. hard breath-hold) and health status (healthy adults vs. PwPD); and (3) explore the relationship between vocal fold bowing during rest breathing and glottic closure during breath-holding maneuvers.MethodsThis was a secondary analysis of laryngoscopic exams originally collected for prospective research purposes from 35 healthy adults and 21 PwPD. Laryngoscopic exams included rest breathing and light and hard breath-holding maneuvers. Glottic closure, supraglottic closure, and vocal fold bowing were analyzed using visual-perceptual methods and quantitative computerized pixel-based techniques. Statistical analyses included the logistic regression, Chi-Square tests, Fisher’s Exact tests, McNemar’s tests, and a generalized additive model.ResultsNo significant differences were found between healthy adults and PwPD in the frequency of complete glottic (p = 0.86) or supraglottic (p = 0.32) airway closure during light breath-holds. Similarly, no significant differences were observed between healthy adults and PwPD in glottic (p = 0.07) or supraglottic (p = 0.27) airway closure during hard breath-holds. Across all participants, hard breath-holds significantly increased the likelihood of complete glottic (p < 0.001) and supraglottic (p = 0.01) airway closure compared to light breath-holds. Additionally, greater vocal fold bowing during rest breathing was associated with reduced glottic closure during light breath-holds when the vocal folds were fully adducted (p < 0.001), but not during hard breath-holds (p = 0.973).ConclusionThese findings suggest that breath-holding maneuvers may not reliably elicit complete glottic and supraglottic airway closure. Contrary to our hypothesis, no differences were observed between healthy adults and PwPD in airway closure during breath-holds. However, for all participants, hard breath-holds were significantly more effective than light breath-holds at achieving complete glottic and supraglottic airway closure. Importantly, greater vocal fold bowing was associated with reduced glottic closure during light – but not hard – breath-holds. These results emphasize the role for instrumental assessments (e.g., flexible endoscopic evaluation of swallowing) when prescribing breath-holding maneuvers and highlight the potential need for endoscopic biofeedback training to enhance their effectiveness in achieving complete glottic and supraglottic airway closure in dysphagic populations. Introduction Compensatory swallowing strategies are frequently recommended in clinical practice to facilitate immediate temporary improvements in swallowing safety and efficiency for people with dysphagia. One such strategy is a breath-holding maneuver, which is intended to enhance swallowing safety by facilitating more timely and complete glottic and supraglottic airway closure. Light and hard breath-holding maneuvers may be prescribed as standalone strategies or as components of more complex strategy techniques, such as the supraglottic or super-supraglottic swallow [1]. Breath-holding maneuvers may be recommended before a swallow to promote earlier and more prolonged airway closure leading into the swallow to reduce the risk of pre-swallow penetration and aspiration. Breath-holding maneuvers applied immediately after the swallow (before the resumption of respiration) may also be paired with breath-holds before the swallow to prolong airway closure until a post-swallow cough or throat clear is elicited, thereby minimizing the risk of post-swallow aspiration [2, 3]. Although breath-holding maneuvers are widely used in clinical practice, current evidence suggests that these strategies may not consistently achieve complete glottic and supraglottic airway closure, which may negatively impact their effectiveness for providing benefit on swallowing safety. Current research demonstrates that 24–58% of healthy adults fail to achieve complete glottic or supraglottic airway closure during light or hard breath-holding maneuvers [4–7]. This raises concerns about the efficacy of breath-holding maneuvers as compensatory techniques, particularly for people with underlying motor impairments that could further compromise airway protection. Importantly, little is known whether similar patterns of incomplete airway closure during breath-holding maneuvers are also observed in dysphagic populations [8, 9]. Understanding whether people with dysphagia exhibit incomplete airway closure during breath-holding maneuvers is essential for optimizing clinical recommendations and patient-centered care. One population that may be especially vulnerable to incomplete airway closure during breath-holding maneuvers is people with Parkinson’s disease (PwPD). Parkinson’s disease is a progressive neurodegenerative disorder that frequently affects swallowing function due to impairments in motor control. Swallowing dysfunction in PwPD is characterized by multiple deficits, including: (1) prolonged pharyngeal swallow initiation, increasing the risk of pre-swallow penetration and aspiration [10–16]; (2) increased vocal fold bowing and reduced glottic closure, which contributes to penetration and aspiration during the swallow and post-swallow pharyngeal residue [17–20]; and (3) reduced pharyngeal constriction, leading to post-swallow pharyngeal residue and increased risk for post-swallow penetration or aspiration [10, 12, 15, 16, 21, 22]. Given these physiological impairments, it is plausible that PwPD may exhibit a higher prevalence and severity of incomplete glottic and supraglottic airway closure during breath-holding maneuvers compared to healthy adults. If PwPD do not achieve complete airway closure when instructed to perform breath-holding maneuvers, it would raise important clinical concerns about their effectiveness as a compensatory swallow strategy within this population. Furthermore, if PwPD exhibit incomplete airway closure during breath-holding maneuvers at rates similar to or greater than those observed in healthy adults, this would underscore the need for direct visualization of the larynx when prescribing breath-holding maneuvers. Such visualization could help ensure that patients are accurately executing these strategies and, if necessary, inform the use of endoscopic biofeedback training to enhance the precision and consistency of airway closure [23]. Given the above, the aims of this study were to (1) characterize glottic and supraglottic airway closure during light and hard breath-holding maneuvers in healthy adults and PwPD; (2) determine if glottic and supraglottic closure differs significantly between PwPD and healthy adults during light and hard breath-holding maneuvers; and (3) explore the relationship between vocal fold bowing and glottic closure during breath-holding maneuvers. We hypothesized: (1) PwPD would exhibit less frequent glottic and supraglottic airway closure compared to healthy adults during light and hard breath-holds; (2) hard breath-holds would elicit increased frequency of glottic and supraglottic airway closure for both PwPD and healthy adults; and (3) there would be a significant relationship between glottic closure and vocal fold bowing during breath-holding maneuvers. Methods Participants This was a secondary analysis of research data originally collected as part of a larger prospective clinical trial. Recruited participants included healthy adults and PwPD. Participant recruitment occurred from October 18, 2021, to March 28, 2022. Participants included non-dysphagic, community-dwelling, healthy adults, and PwPD. All participants were blinded to the aims of this cross-sectional study. All procedures were performed in accordance with the ethical standards of the Institutional Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Approval was obtained from the Institutional Review Board (IRB No. 21-392). Informed consent was obtained from all participants prior to enrollment in this research study, and all participants were blinded to the study aims. Inclusion criteria for the healthy adult cohort were as follows: (1) 18 years of age or older; (2) no history of medical diagnoses or surgeries associated with oropharyngeal or esophageal dysphagia, including stroke, neurodegenerative disease, neuromuscular disease, head and neck cancer, head and neck surgery, esophageal cancer, lung cancer, respiratory disease/disorders, and autoimmune disease; and (3) no self-reported swallowing difficulties. Inclusion criteria for the PwPD cohort were as follows: (1) 18 years of age or older; (2) diagnosis of Parkinson’s disease from a movement-disorders fellowship-trained neurologist; and (3) no history of medical diagnoses or surgeries (other than Parkinson’s disease with or without deep brain stimulation) associated with oropharyngeal or esophageal dysphagia, including stroke, neurodegenerative disease, neuromuscular disease, head and neck cancer, head and neck surgery, esophageal cancer, lung cancer, respiratory disease/disorders, and autoimmune disease. All PwPD reported having arrived during the “on” state of their Parkinson’s medications. Cognitive function was not used as an inclusion/exclusion criterion for this study. Demographic data (age, sex, gender, race/ethnicity) were collected for all participants. Additionally, Parkinson’s disease-related data were obtained from all PwPD (symptom duration, diagnosis duration, Hoehn and Yahr Stage (HYS), and Schwab and England). Procedures and Equipment Recruited participants underwent transnasal flexible laryngoscopy with simultaneous respiratory inductive plethysmography and nasal airflow monitoring. Laryngoscopy was used to visualize glottic and supraglottic airway closure during light and hard breath-holding maneuvers, and vocal fold bowing during rest breathing. Respiratory inductive plethysmography and nasal airflow monitoring were used to monitor the presence/absence of breathing during the light and hard breath-holds. Laryngoscopic equipment consisted of a 3.0-mm diameter flexible distal chip laryngoscope (ENT-5000; Cogentix Medical, New York, USA) and video system with integrated LED light source and LCD display (DPU-7000A; Cogentix Medical, New York, USA). Laryngoscopic equipment was white balanced at the start of every examination. The flexible laryngoscope was passed transnasally without the use of topical anesthetics or vasoconstrictors. Water was used to lubricate the body of the endoscope. The endoscopists performing the flexible laryngoscopy were speech-language pathology graduate interns who were trained by the first author to perform and interpret flexible laryngoscopy and who were blinded to the aims of this cross-sectional study. Training included hands-on endoscopy passes in normal and disordered adults for performance of flexible endoscopic evaluation of swallowing (FEES) until independence was observed, and didactic lectures of the Visual Analysis of Swallowing Efficiency and Safety (VASES) [24–27], with accompanying interpretation practice, for the interpretation of FEES. Respiratory inductive plethysmography (Ambulatory Monitoring, Inc., Ardsley, NY, USA) was performed using one elastic plethysmography band around the rib cage and another around the abdomen. Nasal airflow was measured using a pneumotachograph (ADInstruments, Dunedin, New Zealand) coupled to a spirometry pod (ADInstruments, Dunedin, New Zealand), PowerLab data acquisition hardware (PL3516 PowerLab 16/35, ADInstruments, Dunedin, New Zealand), and a nasal cannula. Respiratory inductive plethysmography and nasal airflow signals were digitized using PowerLab, with plethysmography signals captured at a sample rate of 2 kHz, and airflow signals captured at a sample rate of 4 kHz. LabChart v8 software (ADInstruments, Dunedin, New Zealand) was used to store and analyze all plethysmography and airflow signals. The plethysmography and airflow signals were synchronized with flexible laryngoscopy by pairing an external camera (Microsoft LifeCam Cinema HD) with the LabChart video capture module. The external camera captured real-time video of the flexible laryngoscopy viewing monitor and the seated participant. These signals were used to confirm the presence of complete respiratory cessation during each breath-holding task. Participants were seated in an upright position during the flexible laryngoscopic procedure and were instructed to remain silent throughout the exam unless explicitly directed otherwise or if they needed to express pain or discomfort. The flexible laryngoscopic procedure began with a 60-s period of resting breathing to allow participants to acclimate to the presence of the flexible transnasal laryngoscope. This period of rest breathing was also used to obtain images of vocal fold bowing during tidal exhalation. Following acclimation, participants performed one trial of a light breath-hold maneuver and one trial of a hard breath-hold maneuver. For the light breath-holding maneuver, participants were instructed as follows: “Hold your breath lightly, whatever that means to you. I’ll tell you when to start and when to stop. Ready?” Participants were then cued to “start” and, after 3 s, were cued to “stop.” Immediately afterward, they were asked, “Did you hold your breath?” to confirm patient-perception of task completion. For the hard breath-hold maneuver, participants were given the following instruction: “Hold your breath hard, as if you are going to the bathroom. I’ll tell you when to start and stop. Ready?” As with the light breath-hold maneuver, participants were cued to “start,” held the maneuver for 3 s, and were then cued to “stop.” They were again asked, “Did you hold your breath?” to verify patient-perception of task completion. All video recordings were stored digitally and analyzed offline in the order in which they were recorded. Endoscopic recordings were evaluated for usability and inclusion in the present study by assessing visualization of rest breathing, light breath-hold, and hard breath-hold trials for each participant. Recordings were deemed usable for the rest breathing video clips if ≥75% of the membranous vocal folds were visible. Recordings were deemed usable for the light and hard breath-holding maneuvers if ≥75% of the glottic or supraglottic structures were visible. Research assistants who were uninvolved in data collection and who were blinded to the study aims and all participant data viewed the simultaneous FEES video and LabChart files and extracted three still images per participant: (1) tidal exhalation during rest breathing; (2) maximal glottic/supraglottic closure during light breath-holding; and (3) maximal glottic/supraglottic closure during hard breath-holding. Outcome Measures Outcome measures were independently analyzed using ImageJ software by two speech-language pathology research assistants who were blinded to patient health status, task, and each other’s ratings [28]. All quantitative outcome measures were averaged between those taken by each independent rater. Any discrepancies between raters in visual-perceptual, categorical measures (i.e., categorical ratings that did not match exactly) were resolved during a consensus meeting with a third expert rater. Glottic closure and supraglottic closure were used to characterize glottic and supraglottic airway closure during light and hard breath-holding maneuvers in healthy adults and PwPD (aim 1) and to determine if glottic and supraglottic closure differed between PwPD and healthy adults during breath-holding maneuvers (aim 2). Total bowing index (BI), glottic angle, and normalized glottic gap area (NGGA) were used to explore the relationship between vocal fold bowing and glottic closure during breath-holding maneuvers (aim 3). These outcome measures are defined below. Glottic Closure Glottic closure was assessed during light and hard breath-holding maneuvers using visual-perceptual ordinal ratings: complete (100%), mostly complete (≥50%), partially complete (<50%), or absent (0%). This was estimated visual-perceptually by assessing the extent to which the vocal folds were obliterating the glottic space. If supraglottic closure obscured visualization of the glottis or vocal folds, the extent of glottic closure was estimated to be equivalent to that of supraglottic closure. Supraglottic Closure Supraglottic closure was also assessed during light and hard breath-holding maneuvers using visual-perceptual ordinal ratings of complete (100%), mostly complete (≥50%), partially complete (<50%), or absent (0%). This was estimated visual-perceptually by assessing the extent to which supraglottic structures (ventricular folds, arytenoids, and epiglottic petiole) covered the midline glottis. Total BI BI is a pixel-based measure that was analyzed using ImageJ software and its straight-line tool function [20, 29, 30]. BI was calculated to quantify amount of vocal fold bowing along the medial edge of the membranous vocal folds using the formula:BI=wL×100where:w = maximum perpendicular distance between L and the superomedial edge of the vocal foldL = length of the membranous vocal fold, measured from the anterior commissure to the vocal fold process w = maximum perpendicular distance between L and the superomedial edge of the vocal fold w = maximum perpendicular distance between L and the superomedial edge of the vocal fold L = length of the membranous vocal fold, measured from the anterior commissure to the vocal fold process L = length of the membranous vocal fold, measured from the anterior commissure to the vocal fold process BI is a dimensionless measure expressed as a percentage of vocal fold length, quantifying the extent of vocal fold bowing. Positive BI values indicate vocal fold bowing, negative values indicate convexity of the vocal folds, and a BI of 0 would indicate a smooth and straight vocal fold edge [31]. BI was measured for the left and right vocal fold. Total BI was then calculated by summing together the left BI and the right BI. Glottic Angle Glottic angle is a pixel-based measure that was analyzed using ImageJ software and its angle tool function [32]. This measure was used to characterize extent of vocal fold adduction and approximation of the vocal fold processes. Measurements were obtained from three anatomical landmarks on each still image: the right vocal fold process, the anterior commissure, and the left vocal fold process. A glottic angle of 0 degrees indicates complete approximation of the vocal fold process. Normalized Glottic Gap Area NGGA is a pixel-based measure that was analyzed using ImageJ software and its freehand tracing tool function [20, 29, 30]. NGGA was used to quantitatively estimate amount of glottic opening/closing and was calculated using the following formula:NGGA=aL2×100where:a represents the glottic area,L2 represents the squared average vocal fold length measured during the rest breathing condition. a represents the glottic area, a represents the glottic area, L2 represents the squared average vocal fold length measured during the rest breathing condition. L2 represents the squared average vocal fold length measured during the rest breathing condition. Glottic area (a) was measured by tracing the visible glottic opening from the anterior commissure, along the superior-medial edges of the vocal folds, and across the interarytenoid space. Vocal fold length (L) was determined during rest breathing (as described in total BI) and squared to serve as an anatomical scalar. Because endoscopic positioning varies across trials and the vocal folds may be obscured during breath-holding, a scaling factor was applied to estimate squared vocal fold length for light and hard breath-holding trials. This factor was calculated as the ratio of squared vocal fold length to epiglottic width during rest breathing and then applied to the epiglottic width measured in each breath-holding trial. Epiglottic width was measured in pixels between the left and right aryepiglottic folds where they merge into the epiglottis, following the Visual Analysis of Swallowing Efficiency and Safety (VASES) methodology [24]. NGGA values are expressed as a percentage, with 0% indicating complete glottic closure and larger values representing greater glottic opening (i.e., less closure). This normalization accounts for individual anatomical differences and endoscopic distance variability, ensuring consistent measurement across trials. Statistical Analysis Statistical analyses were conducted using R version 4.2.3 [33]. An alpha level of p < 0.05 was used to determine the level of statistical significance for each predictor variable. Unadjusted (raw) p values are outlined in the text and tables, with an asterisk in the tables denoting statistically significant findings after adjusting for multiple comparisons. To evaluate the effects of breath-holding maneuvers (light vs. hard) on glottic and supraglottic closure in healthy adults and PwPD, a logistic regression model was initially fit with an interaction term to assess whether the effect of breath-holding maneuver differed by health status. A generalized linear mixed model with a logit link function was attempted to account for the within-subject design by including a random intercept for each participant. However, this model failed to converge due to data sparsity and quasi-separation issues, likely driven by an imbalance in categorical distributions. Given these convergence issues, a standard logistic regression model (glm()) was used instead, treating all observations as independent. The interaction term was not statistically significant for either glottic or supraglottic closure (p > 0.05); therefore, main effects were analyzed separately. The Chi-square tests were then performed to compare the frequency of glottic and supraglottic closure between healthy adults and PwPD during both light and hard breath-holding maneuvers (four models in total). If any expected cell count was <5, Fisher’s exact test was applied instead. Then, McNemar’s tests were conducted to assess within-subject differences in closure patterns across breath-holding maneuvers, with one model evaluating glottic closure and a second model evaluating supraglottic closure. If the chi-Square or Fisher’s exact tests indicated a significant difference in closure between healthy adults and PwPD, McNemar’s tests were planned to be conducted separately within each group. If no significant group differences were found, data from both groups were eligible to be pooled into a single model for breath-holding comparisons. To examine the relationship between total BI and NGGA, a generalized additive model was used to account for potential nonlinear associations. Only trials where the vocal fold processes were completely or near completely approximated were included, as defined by a glottic angle of <5°. The model included breath-holding maneuver (light vs. hard breath-holds) as a predictor, and smooth terms were fit separately for each task to test whether the effect of Total BI on NGGA differed between breath-holding maneuvers. A model comparison was conducted using ANOVA to determine whether including the interaction term between total BI and breathing task improved model fit. Inter-rater reliability was assessed using 100% of the data. Intraclass correlation coefficients estimates and their 95% confidence intervals were used to characterize reliability for all continuous-based outcome measures (NGGA, total BI, and glottic angle), based on a mean rating (k = 2), 2-way, mixed effects model. Weighted Kappas and their 95% confidence intervals were used to characterize reliability of all ordinal categorical variables (glottic closure, supraglottic closure). Results Thirty-five healthy adults and 21 PwPD were included in this study (n = 56; Table 1). Healthy adults included 17 men and 18 women with a mean age of 48 years (SD = 14.3). PwPD included 10 men and 11 women, with a mean age of 70 years (SD = 8.4). Furthermore, the majority of the PwPD presented with mid-stage Parkinson’s disease, characterized by a Hoehn and Yahr stage 2 (n = 10) or 3 (n = 7). Table 1.Demographics​Healthy (N = 35)PD (N = 21)Age, years Mean (SD)49.8 (14.3)70.1 (8.40) Median [Q1, Q3]48.0 [41.0, 60.0]71.0 [65.0, 76.0] Min, Max23.0, 83.053.0, 84.0Sex, n (%) Female18 (51.4)11 (52.4) Male17 (48.6)10 (47.6)Racial/ethnic identity, n (%) Asian2 (5.7)0 (0) Black or African American10 (28.6)2 (9.5) Hispanic or Latino5 (14.3)1 (4.8) White18 (51.4)18 (85.7)Duration of symptoms, years Mean (SD)NA12.1 (7.99) Median [Q1, Q3]NA9.00 [7.00, 14.0] Min, MaxNA2.50, 33.0Duration of diagnosis, years Mean (SD)NA10.7 (7.74) Median [Q1, Q3]NA8.00 [6.00, 12.0] Min, MaxNA1.50, 32.0Hoehn and Yahr Stage (HYS), n (%) HYS 1NA1 (4.8) HYS 2NA10 (47.6) HYS 3NA7 (33.3) HYS 4NA2 (9.5) HYS 5NA1 (4.8)Schwab and England Mean (SD)NA77.6 (19.5) Median [Q1, Q3]NA80.0 [70.0, 90.0] Min, MaxNA10.0, 100Functional Oral Intake Scale (FOIS), n (%) FOIS Level 50 (0)6 (28.6) FOIS Level 63 (8.6)9 (42.9) FOIS Level 732 (91.4)6 (28.6)Self-reported swallow impairment (1–7) Mean (SD)1.43 (1.09)3.00 (1.45) Median [Q1, Q3]1.00 [1.00, 1.50]3.00 [2.00, 4.00] Min, Max1.00, 7.001.00, 6.00Dysphagia Handicap Index (DHI) Mean (SD)5.54 (8.72)25.1 (19.4) Median [Q1, Q3]4.00 [2.00, 7.00]22.0 [10.0, 38.0] Min, Max0, 50.02.00, 68.0 Demographics Twelve breath-holding trials were skipped from the participant’s FEES and were therefore missing from the dataset. Across the remaining 156 laryngoscopic video clips, 120 were deemed usable, yielding an analysis of 55 rest breathing video clips, 35 light breath-hold video clips, and 30 hard breath-hold video clips. The reduction in usable video clips was due to posterior movement of the base of tongue and epiglottis over the laryngeal vestibule which obstructed endoscopic visualization of the glottis and supraglottis. All breath-holding trials resulted in complete respiratory cessation, regardless of the status of glottic and supraglottic airway closure, as confirmed using the respiratory inductive plethysmography and nasal airflow signals. This was in agreement with the participants’ self-report of successfully holding their breath. Descriptive summary statistics are shown in Tables 2–4. In terms of inter-rater reliability, intraclass correlation coefficient estimates and their 95% confidence intervals were observed to be 0.715 (95% CI: 0.586–0.804), 0.724 (95% CI: 0.552–0.830), and 0.99 (95% CI: 0.982–0.994) for the ImageJ analyses of NGGA, total BI, and glottic angle, respectively. Weighted Kappa estimates, and their 95% confidence intervals, were observed to be 0.874 (95% CI: 0.762–0.985) and 0.816 (95% CI: 0.711–0.920) for visual-perceptual assessments of glottic closure and supraglottic closure, respectively. Table 2.Summary statistics of glottic and supraglottic closure​Healthy adultParkinson’s diseaseOveralllight breath hold (n = 18)hard breath hold (n = 15)light breath hold (n = 17)hard breath hold (n = 15)light breath hold (n = 35)hard breath hold (n = 30)Glottic closure, n (%) None2 (11.1)0 (0)3 (17.6)3 (20.0)5 (14.3)3 (10.0) Partially complete5 (27.8)0 (0)1 (5.9)0 (0)6 (17.1)0 (0) Mostly complete2 (11.1)0 (0)5 (29.4)0 (0)7 (20.0)0 (0) Complete9 (50.0)15 (100)8 (47.1)12 (80.0)17 (48.6)27 (90.0)Supraglottic closure, n (%) Complete3 (16.7)9 (60.0)1 (5.9)6 (40.0)4 (11.4)15 (50.0) Mostly complete1 (5.6)1 (6.7)4 (23.5)3 (20.0)5 (14.3)4 (13.3) None11 (61.1)5 (33.3)7 (41.2)3 (20.0)18 (51.4)8 (26.7) Partially complete3 (16.7)0 (0)5 (29.4)3 (20.0)8 (22.9)3 (10.0) Summary statistics of glottic and supraglottic closure Table 3.Summary statistics during rest breathing​Healthy adult (n = 35)Parkinson's disease (n = 20)Overall (n = 55)Total BI, %L Mean (SD)4.56 (6.16)2.22 (7.92)3.73 (6.85) Median [Q1, Q3]3.22 [0, 8.13]0.0762 [−3.03, 7.12]2.20 [−0.0225, 7.83] Min, Max−6.83, 20.8−12.2, 20.2−12.2, 20.8Glottic angle, ° Mean (SD)61.5 (11.4)53.7 (15.7)58.7 (13.5) Median [Q1, Q3]62.6 [52.5, 69.7]52.5 [44.5, 63.7]59.4 [50.3, 67.7] Min, Max43.7, 93.021.3, 89.221.3, 93.0NGGA, %L2 Mean (SD)150 (106)115 (36.4)138 (88.9) Median [Q1, Q3]133 [112, 151]111 [93.2, 121]121 [107, 143] Min, Max81.0, 72663.8, 22163.8, 726 Summary statistics during rest breathing Table 4.Summary statistics of NGGAa during breath-holding tasks​Light breath hold (n = 22)Hard breath hold (n = 27)Overall (n = 49)NGGA, %L2 Mean (SD)1.32 (4.88)0.00893 (0.0447)0.607 (3.32) Median [Q1, Q3]0 [0, 0]0 [0, 0]0 [0, 0] Min, Max0, 22.40, 0.2230, 22.4 Missing, n (%)1 (4.5)2 (7.4)3 (6.1) Summary statistics of NGGAa during breath-holding tasks Glottic Closure in Healthy Adults and PwPD during Light and Hard Breath-Holds Glottic closure did not significantly differ between healthy adults and PwPD during the light breath-holding maneuver (χ2 (1) = 0.03, p = 0.86, VCramer = 0.00, 95% CI: [0.00, 0.27], nobs = 35; Fig. 1). Both groups demonstrated similar distributions, with 50% of healthy adults and 53% of PwPD exhibiting incomplete glottic closure. Similarly, glottic closure did not significantly differ between healthy adults and PwPD during the hard breath-holding maneuver (χ2 (1) = 3.33, p = 0.07, VCramer = 0.28, 95% CI: [0.00, 0.68], nobs = 30; Fig. 2). None of the healthy adults demonstrated incomplete glottic closure, whereas 20% of PwPD exhibited incomplete glottic closure. Stacked bar graph showing the proportion of trials with complete and incomplete glottic closure during a light breath-hold maneuver in healthy adults (left) and PwPD (right). No significant differences in glottic closure frequency were observed between groups (χ2 (1) = 0.03, p = 0.86, Cramér’s V = 0.00, 95% CI: [0.00, 0.27]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Stacked bar graph showing the proportion of trials with complete and incomplete glottic closure during a hard breath-hold maneuver in healthy adults (left) and PwPD (right). No significant differences in glottic closure frequency were observed between groups (χ2 (1) = 3.33, p = 0.07, Cramér’s V = 0.28, 95% CI: [0.00, 0.68]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Across all participants, glottic closure significantly differed between light and hard breath-holding maneuvers (χMcNemar2 (1) = 9.80, p < 0.0001, ĝCohen = 0.35, 95% CI: [0.14, 0.45], npairs = 65; Fig. 3). Specifically, complete glottic closure was observed in 49% of light breath-holding trials but increased significantly to 90% during hard breath-holding, indicating a greater likelihood of glottic closure with a hard breath-hold instruction. Stacked bar graph showing the proportion of trials with complete and incomplete glottic closure during light breath-hold (left) and hard breath-hold (right) maneuvers. Glottic closure was significantly more likely to be complete during the hard breath-hold compared to the light breath-hold maneuver (χmn2 (1) = 9.80, p = 1.75 × 10−3, Cohen’s G = 0.35, 95% CI: [0.14, 0.45]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Supraglottic Closure in Healthy Adults and PwPD during Light and Hard Breath-Holds Supraglottic closure did not significantly differ between healthy adults and PwPD during the light breath-holding maneuver (χ2 (1) = 1.00, p = 0.32, VCramer = 0.00, 95% CI: [0.00, 0.48], nobs = 35; Fig. 4). Incomplete supraglottic closure was observed in 94% of PwPD (94%) and in 83% of healthy adults (83%). Similarly, supraglottic closure did not significantly differ between the healthy adult and PwPD groups during the hard breath-holding maneuver (χ2 (1) = 1.20, p = 0.27, VCramer = 0.08, 95% CI: [0.00, 0.54], nobs = 30; Fig. 5). Specifically, incomplete supraglottic closure was observed in 40% of healthy adults and 60% of PwPD. Stacked bar graph showing the proportion of trials with complete and incomplete supraglottic closure during a light breath-hold maneuver in healthy adults (left) and PwPD (right). No significant differences in supraglottic closure frequency were observed between groups (χ2 (1) = 1.00, p = 0.32, Cramér’s V = 0.00, 95% CI: [0.00, 0.48]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Stacked bar graph showing the proportion of trials with complete and incomplete supraglottic closure during a hard breath-hold maneuver in healthy adults (left) and PwPD (right). No significant differences in supraglottic closure frequency were observed between groups (χ2 (1) = 1.20, p = 0.27, Cramér’s V = 0.08, 95% CI: [0.00, 0.54]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Across all participants, there was a significant effect of breath-holding maneuver on supraglottic closure (χMcNemar2 (1) = 6.37, p = 0.01, ĝCohen = 0.29, 95% CI [0.07, 0.41], npairs = 65; Fig. 6). Specifically, complete supraglottic closure was observed in only 11% of light breath-holding trials, which increased to 50% of hard breath-holding trials, demonstrating an increased likelihood of supraglottic closure with hard breath-holds. Stacked bar graph showing the proportion of trials with complete and incomplete supraglottic closure during light breath-hold (left) and hard breath-hold (right) maneuvers. Supraglottic closure was significantly more likely to be complete during the hard breath-hold compared to the light breath-hold maneuver (χmn2 (1) = 6.37, p = 0.01, Cohen’s G = 0.29, 95% CI: [0.07, 0.41]). Percentages within each bar indicate the proportion of trials classified as complete or incomplete. Relationship between Vocal Fold Bowing and Glottic Gap during Light and Hard Breath-Holds Vocal fold bowing during rest breathing and its relationship to glottic closure during light and hard breath-holds with complete/nearly complete vocal fold adduction are shown in Figure 7. A significant interaction effect was observed between total BI and breath-holding maneuver (p < 0.001), indicating that the relationship between total BI and NGGA differed by breath-holding condition. In the light breath-hold condition, larger total BI was significantly associated with larger NGGA (F = 346.50, p < 0.001), suggesting that vocal fold bowing influenced glottic closure during light breath-holds. However, in the hard breath-hold condition, no significant association was found between total BI and NGGA (F = 0.001, p = 0.973), indicating that vocal fold bowing did not influence glottic closure during hard breath-holds. The top panel displays the raincloud plot distributions of total BI (%) in healthy adults and PwPD using a combination of boxplots, dot plots, and density plots. The bottom panel illustrates the relationship between total BI (%) and NGGA (%) across breath-holding maneuvers. The solid lines represent smooth functions fitted using generalized additive models (GAMs), with shaded areas indicating 95% confidence intervals. Discussion This study investigated glottic and supraglottic airway closure during light and hard breath-holding maneuvers in healthy adults and PwPD and explored the relationship between vocal fold bowing and glottic closure. The primary findings were as follows: (1) no significant differences were present in glottic or supraglottic closure between healthy adults and PwPD during light or hard breath-holding maneuvers; (2) there was a significant increase in the likelihood of complete glottic and supraglottic closure during hard breath-holds compared to light breath-holds across all participants; and (3) greater vocal fold bowing during rest breathing was associated with reduced glottic closure during light breath-holds but not during hard breath-holds. These results partially support our hypotheses. As expected, hard breath-holding maneuvers more reliably elicited complete airway closure compared to light breath-holds, despite that all breath-hold maneuvers resulted in temporary cessation of breathing. However, contrary to our hypothesis, PwPD did not demonstrate significantly different rates of incomplete glottic or supraglottic closure relative to healthy adults. Additionally, while greater vocal fold bowing was significantly associated with reduced glottic closure during light breath-holds, this relationship was not present during hard breath-holds. These findings add to a growing body of literature suggesting that breath-holding maneuvers, particularly light breath-holds, do not consistently achieve complete airway closure, even in healthy individuals. The present results extend this observation to PwPD, suggesting that the prevalence of incomplete closure may be similar across groups when task demands are standardized. Importantly, the observed improvement in airway closure during hard breath-holding indicates that task intensity can meaningfully influence outcomes, regardless of underlying health status. Future research should examine how specific verbal instructions influence glottic and supraglottic airway closure during breath-holding. For example, instructing patients to inhale prior to breath-holding, to perform a breath-hold without qualifiers such as “light” or “hard,” or to “close your voice box” rather than to “hold your breath” may elicit different closure patterns. This distinction is important because, although all participants in this study successfully stopped breathing when asked to “hold your breath,” the extent of glottic and supraglottic closure still differed depending on the type of verbal instruction given (i.e., light vs. hard breath-hold). The novel exploration of vocal fold bowing adds further nuance. Although greater bowing was associated with less complete glottic closure during light breath-holds, the lack of association during hard breath-holds suggests that task effort may, under certain circumstances, be able to compensate for minor structural limitations. One plausible physiologic explanation involves the thyroarytenoid muscle. Contraction of this muscle shortens and tenses the vocal folds, effectively increasing the functional mass of the membranous vocal folds. If hard breath-holds elicit greater thyroarytenoid activation compared to light breath-holds, the resulting increase in vocal fold bulk may help offset the effects of bowing and facilitate more complete closure. This interpretation is consistent with prior work showing that vocal fold bowing does not appear to limit other airway protective behaviors, such as cough, whereby hard breath-holds are frequently used during the compression phase of cough [17]. These findings highlight the importance of directly assessing glottic and supraglottic airway closure when prescribing breath-holding maneuvers in clinical practice. Instrumental assessment, such as FEES or static endoscopic evaluation of swallowing [34], may be necessary to confirm whether patients are achieving effective glottic and supraglottic closure, rather than assuming successful execution based on instruction alone. Furthermore, the results support the potential utility of endoscopic biofeedback training to enhance the effectiveness of breath-holding strategies, particularly for individuals with structural laryngeal changes such as vocal fold bowing [23]. Finally, clinicians may consider instructing patients to perform harder, more effortful breath-holds if incomplete airway closure is observed during light breath-holding tasks. However, individual variability suggests that strategy selection should be tailored based on direct visualization of airway closure rather than using a one-size-fits-all approach. Limitations and Future Directions Several limitations should be acknowledged. First, most of the PwPD in this study (80.9%) were early-to-mid-stage Parkinson’s disease, as noted by a Hoehn and Yahr Stage 2 (47.6% of the study sample) or 3 (33.3% of the study sample). Future studies should include PwPD with more advanced disease staging to verify the generalizability of these findings. Second, breath-holding tasks were evaluated using brief 3-s trials during which the point of maximal closure was used in the analysis; longer durations or more naturalistic swallowing contexts may yield different results and should be explored in future work. Third, in instances where supraglottic closure obscured visualization of glottis, a methodological choice was made to assume that glottic closure was complete inferior to the areas of supraglottic closure. This methodology may have influenced perceptual ratings of glottic closure and NGGA measures. However, it is possible glottic closure and NGGA may have been incomplete despite the presence of supraglottic closure. Fourth, age-related changes in vocal fold structure and function (often termed presbylaryngis) may contribute to bowing, altered vibratory characteristics, and reduced closure capacity in some people. While no structural abnormalities beyond vocal fold bowing were observed in our cohort and all participants exhibited normal vocal fold mobility, the underlying etiology of bowing (age- vs. disease-related) and its subsequent impact on glottic closure could not be determined. Future studies should therefore examine how age-related and disease-related laryngeal changes may interact to influence airway closure during volitional tasks. Finally, while the healthy adult and PwPD groups were comparable in sex distribution, they differed in age by a mean of 22 years. Despite this, no significant between-group differences were observed in airway closure outcomes. The degree of vocal fold bowing was also similar between groups, with post hoc analysis demonstrating no significant difference in total BI between the two groups (p > 0.05). Together, these findings suggest that neither age differences nor disease status meaningfully affected the outcomes of this study. Nevertheless, age-related changes in laryngeal structure and function may interact with disease-related changes, and future work should incorporate age-matched controls or within-group analyses to more thoroughly assess the independent contributions of age and medical status. Future research should aim to assess breath-holding maneuvers during functional swallowing tasks and evaluate whether endoscopic biofeedback training can improve the consistency and accuracy of breath-holding strategies over time. Additionally, exploring other populations with known laryngeal deficits, such as people with severe vocal fold atrophy or vocal fold paralysis, could further inform patient-specific adaptations to airway protection strategies. Conclusions In summary, this study found that hard breath-holding maneuvers were significantly more effective at achieving complete glottic and supraglottic airway closure compared to light breath-holds, with similar outcomes observed between healthy adults and PwPD. Vocal fold bowing was associated with impaired glottic closure during light breath-holds but not during hard breath-holds. These findings suggest that instrumental assessment and targeted biofeedback may be needed when applying breath-holding maneuvers to optimize the accurate performance of breath-holding maneuvers on glottic and supraglottic airway closure. Statement of Ethics This study was performed in accordance with the Declaration of Helsinki. This human study was approved by Institutional Review Board at Teachers College, Columbia University – approval No. 21-392. All adult participants provided written informed consent to participate in this study. Conflict of Interest Statement The authors have no relevant financial disclosure or conflicts of interest. Funding Sources This work was supported by funding awarded to Dr. James Curtis from the Parkinson’s Foundation and the American Brain Foundation in collaboration with the American Academy of Neurology. Author Contributions All authors met the minimum criteria for authorship status, as proposed by the International Committee of Medical Journal Editors (ICMJE). Authorship contributions have been characterized using CRediT (https://credit.niso.org/). Specific authorship contributions are as follow: James A. Curtis: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, visualization, writing – original draft, and writing – reviewing and editing; Alexandra B. Beam: investigation, supervision, writing – original draft, and writing – reviewing and editing; Jina Nam: investigation and writing – reviewing and editing; Michelle S. Troche: resources, supervision, validation, and writing – reviewing and editing.