Clinical utility of voluntary cough peak expiratory flow outcomes to screen aspiration status in head and neck cancer survivors

Abstract

Purpose

Peak expiratory flow of voluntary cough (PEF) is a physiologic metric indexing an individual’s physiologic capacity to defend the airway by expelling tracheal aspirate. PEF can be obtained in an objective, quick and inexpensive manner in busy clinical settings, with demonstrated discriminant ability to detect unsafe swallowing in neurogenic populations. This study examined the discriminant ability of voluntary cough PEF to classify aspiration status in head and neck cancer (HNC) patients post-radiotherapy (RT).

Methods

This was a secondary analysis of a prospective clinical trial of disease-free HNC survivors. Patients referred for modified barium swallow (MBS) studies ≥ 3 months post-RT underwent voluntary cough PEF assessment. Aspiration was defined as a worst Penetration-Aspiration Scale (PAS) score ≥ 6. Logistic regression and ROC assessed associations between maximum PEF (PEFmax) and aspiration. Optimal thresholds were determined using Youden Index and classification accuracy.

Results

Among 152 participants, mean PEFmax was 447.6 ± 118.5 L/min. Aspirators (51/152; 34%) exhibited 13% lower PEFmax than non-aspirators (mean difference: 61.8 L/min). Lower PEFmax was significantly associated with aspiration, with a modest 5% increase in odds per 10 L/min decrease (OR = 1.005; 95% CI: 1.002–1.008; p = 0.003). The optimal threshold (≤ 475 L/min) achieved 72.5% sensitivity and 48.5% specificity; individuals below this threshold had 2.5-fold greater odds of aspiration.

Conclusion

Voluntary cough PEF may serve as an adjunctive rapid and pragmatic objective clinical metric for consideration in aspiration screening in HNC survivors. While moderate specificity limits stand-alone use, its simplicity makes it valuable in triage and surveillance in both resource-rich and resource-limited settings.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00405-026-10102-5.

Introduction

Head and neck cancers (HNCs), encompassing malignancies of the oral cavity, pharynx, larynx, and salivary glands, account for approximately 4% of all cancer diagnoses in the United States, and the seventh most common cancer worldwide, responsible for over 660,000 new cases and 325,000 deaths annually [1, 2]. The global incidence of HNC has been rising, with projections suggesting a 30% increase by 2030 [3]. Advances in surgery, radiotherapy, chemotherapy, and targeted therapies have improved survival, with current five-year relative survival rates approaching 69% [4]. However, this growing population of long-term survivors faces significant treatment-related complications. One of the most serious is aspiration pneumonia—a potentially life-threatening condition caused by the entry of food, saliva, or other substances into the respiratory tract, leading to infection and inflammation. In patients with HNC, aspiration pneumonia is a well-recognized acute and late effect of radiotherapy [5, 6].

US-based population studies have shown that the cumulative incidence of aspiration pneumonia in patients with HNC is estimated at 23.8% within five years after receiving chemo-radiotherapy [6]. Etiology of aspiration pneumonia is multifactorial; however, aspiration related to radiation-associated dysphagia (RAD) plays a central role [7, 8]. Early identification of individuals at high risk for aspiration and subsequent aspiration pneumonia is therefore crucial to ensure optimal pulmonary function and safe oral intake [9].

The utility of screening non-swallowing behaviors or functions is being tested and adopted in populations at risk for dysphagia and aspiration [10]. Voluntary cough airflow measures are associated with radiographically-confirmed swallowing safety in neurogenic populations, including stroke, Parkinson’s disease and amyotrophic lateral sclerosis [9, 11–17]. Not surprisingly, cough has been suggested as a useful metric for integration into the clinical swallowing evaluation [18]. Although imaging examinations of swallowing (i.e., videofluoroscopy and/or endoscopy) remain the gold standard to detect aspiration and directly view the swallowing mechanism, not all centers have immediate access to imaging studies, or have the capacity to perform in every patient. Thus, a brief clinical metric to help identify HNC survivors at increased risk of aspiration might serve as a pragmatic and cost-effective approach to enable risk stratification and triaged care models.

Airway protection is a complex process spanning a continuum of behaviors, with swallowing preventing material from entering the airway and coughing expelling aspirated material when compromise occurs. Both behaviors rely on highly coordinated sequences of structural movements and reconfiguration of the ventilatory pattern, sharing overlapping sensory and motor control mechanisms across the peripheral and central nervous systems [19–21]. Deficits in one behavior can therefore impact the other [19]. Historically, clinical assessment primarily focused on airway protection during swallowing, with less attention to airway defense strategies, such as cough, that are arguably easier to test directly in clinical settings. Not surprisingly, dysphagia and dystussia are noted to co-occur across a number of neurologic and HNC patient populations [22, 23] and has brought into attention the potential utility of cough testing to assess aspiration risk during the clinical swallowing evaluations [9, 17, 24, 25]. Measures of voluntary cough, such as peak expiratory flow (PEF), serve as practical clinical screens for airway protection in patients with neuromuscular disease. For example, a PEF below 270 L/min—associated with increased secretion retention and infection risk—represents a meaningful threshold for identifying individuals who may benefit from cough augmentation techniques [26].

We have previously reported associations between voluntary cough PEF and aspiration status in post-radiation HNC survivors [22]. Specifically, we noted that PEF was on average 14.9% lower (approximately 60 L/min difference) among aspirators compared to non-aspirators after adjusting for age and sex, highlighting a clinically meaningful reduction in expiratory force in patients with aspiration. The potential role of a voluntary cough PEF threshold to detect aspirators after HN radiation, however, was not previously investigated. Thus, the purpose of the current study was to evaluate the ability of voluntary cough PEF thresholds to detect aspiration status among HNC patients post radiation. We hypothesized that similar to previous findings in neurogenic patient populations, a voluntary cough PEF threshold could discriminate aspirators versus non-aspirators in post-radiation HNC survivors and serve as a useful clinical tool.

Materials and methods

This cross-sectional study enrolled disease-free HNC survivors with a history of curative-intent radiotherapy for primary HNC (in the past 15 years). Patients referred to the Section of Speech Pathology and Audiology at MD Anderson Cancer Center (Houston, TX, USA) for swallowing evaluation between March 2016 and November 2017 were screened for enrollment. Patients that were excluded at the time of enrollment had a history of (i) recurrent disease, (ii) second primary cancer of head and neck, central nervous system, or thoracic cavity, (iii) prior head and neck surgery. The procedures followed were in accordance with the ethical standards of the responsible institutional committee on human experimentation. The analysis was conducted within a larger study approved by the Institutional Review Board of MD Anderson Cancer Center (MDACC; OHRP IRB Registration IRB 00000121; Submission ID: 2015-0238-MDACC). All subjects provided informed consent. Following consent, participants attended a single testing session that included maximum voluntary cough testing and a standardized swallow study. Reporting of this study adhered to the Standards for Reporting of Diagnostic Accuracy Studies (STARD) checklist [27].

Voluntary cough testing

All patients completed maximum peak expiratory flow (PEFmax) of voluntary cough testing at the time of enrollment. Voluntary PEFmax was measured using the Mini-Wright Peak Flow Meter (KW-Med, Inc., Antioch, IL, U.S.A.). Patients while standing were instructed to inhale to total lung capacity (“Take a deep breath”) and to “cough hard like there is something stuck in your throat.” The examiner modeled an example of a strong cough to ensure that the patient understood task requirements. Voluntary PEFmax was calculated from the best of three trials within a 10% variance.

Swallowing evaluation

Swallowing evaluations included a standardized modified barium swallowing (MBS) study acquired and recorded at 30 frames per second. Patients were administered two trials each of 5-mL, 10-mL, and cup sips of Varibar thin liquid barium (Bracco Diagnostics Inc.), teaspoon presentation of Varibar pudding, and a ¼ cracker coated in Varibar pudding in a mid-sagittal lateral plane, and single presentation of barium in an anteroposterior coronal plane.

Each study was subsequently graded by a blinded, trained laboratory rater using the Penetration-Aspiration Scale (PAS) [28]. The PAS is a validated eight-point ordinal scale that ranks the safety of the swallow by the depth of bolus entry into the airway and the patient’s response (0 = no airway entry, 8 = silent aspiration) [28]. The maximum PAS score across all bolus trials was used for analysis, as this parameter is associated with an elevated risk of aspiration pneumonia [8]. Patients with maximum PAS scores of 1 to 5 were classified as non-aspirators, and those with PAS scores of 6 to 8 were as aspirators.

Statistical analysis

Demographic and clinical characteristics, along with cough testing and swallow evaluation results, were summarized using descriptive statistics. Continuous variables were reported as mean (SD) or median (range) and categorical variables as frequency (percentage). Associations between PEFmax and aspiration status were analyzed both as a continuous variable and dichotomized at < 270 vs. ≥ 270 (based on clinically relevant thresholds [26]. Comparisons were performed using Student’s t-test, Mann–Whitney U test, or Chi-square test, as appropriate. To evaluate the diagnostic performance of PEFmax for detecting aspiration, a receiver operating characteristic (ROC) curve was generated using a binary logistic regression model. Optimal cutoff points were identified based on three criteria: (i) the highest Youden Index, (ii) the highest Youden Index among thresholds with at least good sensitivity (≥ 70%), and (iii) the maximum correct classification rate. Diagnostic accuracy was quantified using sensitivity (SN), specificity (SP), and positive/negative predictive values (PPV/NPV), categorized as poor (< 50%), moderate (50–70%), good (70–90%), or excellent (> 90%) [29]. Odds ratios for aspiration status at the selected PEFmax thresholds were calculated, and odds ratios and diagnostic estimates were presented with 95% confidence intervals.

A sensitivity analysis was also performed recalculating all metrics using an alternative PAS dichotomization (0–4 vs. ≥ 5) to classify aspiration status. A two-sided p-value < 0.05 was considered statistically significant.

Analyses were performed using StataNow/SE 18.5 (1985–2023 StataCorp LLC) and R version 4.4.1 (The R Foundation for Statistical Computing).

Results

Patient characteristics

A total of 152 eligible participants consented and were included in this analysis. Mean age was 61.6 years (SD: 8.9), and males represented 90% (n = 137) of the sample (Table 1). Most (87%) participants had an oropharyngeal tumor, and 43% T-stage 3 or 4 disease. At the time of measurement, median time since RT completion was 10 (range: 3-162) months (Table 1). PAS scores ranged from 1 to 8, with a median of 3. Among participants, 46 (33.5%) were classified as aspirators (PAS ≥ 6), and 106 (66.5%) as non-aspirators (PAS < 6).

Table 1.

Characteristics of the study population (n = 152)

Variable	N (%)
Sex
Female	15 (9.9)
Male	137 (90.1)
Age, mean (SD)	61.6 (8.9)
Smoking status at diagnosis
Current	6 (4.0)
Former	50 (32.9)
Never	96 (63.2)
T-Stage
T0	2 (2.6)
T1	28 (18.4)
T2	53(35.6)
T3	31(20.8)
T4	33 (22.2)
N-Stage
N0	16 (10.7)
N1	12 (8.0)
N2	114 (76.0)
N3	8 (5.3)
Subsite
Oropharynx	132 (86.8)
Nasopharynx	1 (0.7)
Hypopharynx	4 (2.6)
Larynx	10 (6.6)
Unknown primary	5 (3.3)
Treatment
CRT	106 (69.7)
Induction + CRT	28 (18.4)
Induction + RT	6 (4.0)
RT alone	3 (2.0)
Surgery + POCRT	6 (4.0)
Surgery + PORT	3 (2.0)
Time since last treatment
< 1 year	87 (57.2)
1 to 3 years	47 (30.9)
> 3 years	18 (11.8)
PAS score
1	27 (17.8)
2	37 (24.3)
3	24 (15.8)
4	3 (2.0)
5	10 (6.6)
6	5 (3.3)
7	8 (5.3)
8	38 (25.0)

CRT chemoradiotherapy, POCRT postoperative chemoradiotherapy, PORT postoperative radiotherapy, RT radiotherapy

Cough profiles and associations with aspiration status

PEFmax values varied between 165 and 785 L/min, with a mean of 447.6 (SD 118.5); 6.6% of the sample had a PEFmax below 270 L/min (Table 2). PESmax distribution by PAS score is shown in Fig. 1. Aspirators exhibited a significantly lower mean PEFmax (406.6, SD 121.0) compared to non-aspirators (468.3, SD 112.3). The proportion of aspirators was significantly greater among those with PEFmax < 270 L/min (13.7% vs. 3.0%, p = 0.012) (Table 2).

Table 2.

Comparison of maximum peak expiratory flow of voluntary cough (PEFmax) between aspirators and non-aspirators (n = 152)

PEFmax (L/min)	Total sample (n = 152)	Aspirators (n = 51)	Non-aspirators (n = 101)	Difference
Mean (SD) [range]	447.63 (118.5) [165; 785]	406.6 (121.0) [165–695]	468.3 (112.3) [185–785]	61.8 [95% CI: 22.6; 100.9] p = 0.002
Cutoff: 270 L/ml				P = 0.012
<270, n (%)	10 (6.6)	7 (13.7)	3 (3.0)
≥270, n (%)	142 (93.4)	44 (86.3)	98 (97.0)

Fig. 1.

Maximum peak expiratory flow of voluntary cough (PEFmax) distribution by PAS score

Clinical utility of PEFmax in discriminating aspiration status

Logistic regression analysis demonstrated that lower PEFmax was significantly associated with increased odds of aspiration. Specifically, each 10-unit decrease in PEFmax corresponded to a modest 5% increase in the odds of aspiration (OR = 1.005; 95% CI: 1.002–1.008; p = 0.003).

The ROC curve for this model is shown in Fig. 2, with an AUC of 0.65, with cutoff points that optimized diagnostic performance shown in Table 3. The cutoff ≤ 350 L/min yielded the highest correct classification rate (70.4%), with good specificity (86.1%) but low sensitivity (39.2%). The ≤ 385 L/min cutoff had the highest Youden Index (0.29), improving sensitivity to 49.0% while maintaining good specificity (80.2%). The cutoff with the highest Youden Index among those with at least sensitivity ≥ 70% was ≤ 475 L/min, offering 72.5% sensitivity and 48.5% specificity. Individuals with PEFmax ≤ 475 L/min had 2.5 times greater odds of aspiration compared to those above this threshold (Table 3).

Fig. 2.

ROC curve for peak expiratory flow (PEFmax) of voluntary cough in detecting aspiration (PAS ≥ 6)

Table 3.

Diagnostic characteristics of optimal Peak Expiratory Flow (PEFmax) of Voluntary Cough cutoffs to indicate aspirator status (PAS ≥ 6)

Cutoff	Threshold established in other populations	Best correct classification rate	Best Youden Index	Best Youden Index for SN ≥ 70%
	≤ 270 L/min	≤ 350 L/min	≤ 385 L/min	≤ 475 L/min
Aspiration (PAS ≥ 6) detection rate	7/152 (4.6%)	20/152 (13.2%)	25/152 (16.5%)	37/152 (24.3%)
Sensitivity (95% CI)	13.7% (5.7–26.3)	39.2% (25.8–53.9)	49.0% (34.8–63.4)	72.5% (58.3–84.1)
Specificity (95% CI)	97.0% (91.6–99.4)	86.1% (77.8–92.2)	80.2% (71.1–87.5)	48.5% (38.4–58.7)
PPV (95% CI)	70.0% (34.8–93.3)	58.8% (40.7–75.4)	55.6% (40.0-70.4)	41.6% (31.2–52.5)
NPV (95% CI)	69.0% (60.7–76.5)	73.7% (64.8–81.4)	75.7% (66.5–83.5)	77.8% (65.5–87.3)
Correct classification	69.1%	70.4%	69.7%	56.5%
Youden Index	0.14	0.25	0.29	0.21
ORs (95% CI)	5.2 (1.4–19.3)	4.0 (1.8–8.8)	3.9 (1.9–8.1)	2.5 (1.2–5.1)

The sensitivity analysis using PAS ≥ 5 as the cutoff showed similar results, with 385 L/min again yielding the best correct classification rate and Youden Index, and 475 L/min remaining the optimal cutoff for achieving sensitivity ≥ 70%. Diagnostic metrics for this sensitivity analysis are reported in Supplementary Material 2, eTable 1.

Discussion

Voluntary PEF is an inexpensive quantitative measure that can be obtained in less than five minutes in a clinic. The primary objective of this project was to examine the discriminant ability of voluntary PEFmax to detect radiographically confirmed aspiration status in post-RT HNC survivors. A voluntary cough PEFmax threshold of 475 L/minute represented the optimal sensitive threshold to differentiate radiographically confirmed aspirators vs. non-aspirators in post-RT HNC survivors. To our knowledge, this is the first study to evaluate the clinical utility of voluntary cough PEF thresholds to model airway compromise during swallowing in post-RT HNC survivors. The voluntary cough airflow testing in the current investigation is an inexpensive hand-held peak cough flow meter capable of immediate analog or digital readouts that can be performed easily in the clinic and at the patients’ bedside, therefore offering a useful adjunct measure not requiring expensive equipment, software or timely analysis.

As hypothesized, cough PEFmax was a significant predictor of aspiration status in post-RT HNC survivors, with lower cough expulsion force related to poor airway safety during swallowing. Specifically, post-RT HNC survivors who aspirated demonstrated cough PEFmax that was 13% lower than non-aspirators (mean difference = 61.8 L/min), with ranges spanning 165–695 L/min and 185–785 L/min, respectively. While statistically significant, the increase in aspiration risk per 10 L/min decrease in PEFmax was modest (OR = 1.005), a finding that is not surprising, given that aspiration reflects impaired airway protection mechanisms rather than reduced voluntary cough strength itself. In comparison to normative values for healthy adults (440–1200 L/min), both groups demonstrated reduced expiratory performance. The neuromuscular threshold of 270 L/min is widely considered critical for identifying patients at risk of ineffective airway clearance and for initiating cough augmentation strategies [26, 30], a significantly higher proportion of aspirators (13.7%) fell below this cutoff compared to non-aspirators (3.0%). These findings suggest that severely depressed cough flows, while not exclusive to aspirators, are disproportionately present in this subgroup. However, 270 L/min as an actionable threshold fails to optimize detection of potential aspirators in the HNC population.

A cough PEFmax threshold of ≤ 475 L/min offered the best performance for the goal of screening HNC survivors after RT for aspiration. This threshold correctly identified 72.5% of aspirators (i.e. sensitivity) and 48.5% of non-aspirators (i.e. specificity). Notably, the area under the ROC curve (AUC = 0.65) indicates only moderate discriminative ability, reflecting that PEFmax alone is a limited predictor of aspiration. Nonetheless, these results, add to a growing evidence base noting clinical utility of voluntary cough to screen for unsafe swallowing in PD, ALS, and stroke [24, 31]. These previous works have noted, using more comprehensive cough spirometry methods, that voluntary peak expiratory flow rate predicts penetration/aspiration status with 74%-86% sensitivity in individuals with PD [11, 24], stroke [12], or ALS [9], with cutoffs of 174 L/min, 182 L/min, and 240 L/min, respectively. The substantially higher cutoff identified in the current HNC cohort (≤ 475 L/min) suggests that cough weakness and aspiration may share fewer underlying mechanisms (such as concurrent neural injury to both cough and swallow substrates) compared with other clinical populations. Another important point is that most aspirators in our cohort were silent (PAS 8), indicating impaired airway sensation and absent reflexive responses are a major contributor to impaired airway protection in addition to insufficient expulsive force as a fundamental issue. Thus, cough strength alone is not expected to be a stand-alone marker of airway safety in HNC, underscoring the multifactorial nature of aspiration risk in this group and the need for population-specific thresholds when using cough metrics for screening. Although voluntary cough PEFmax reflects the motor, rather than sensory, component of airway protection, it nonetheless demonstrated a significant association with aspiration status. This suggests that reduced expulsive capacity represents one measurable pathway contributing to aspiration in post-RT HNC, even if PEFmax cannot capture sensory impairment directly.

Two alternate PCF cutoffs (i.e., ≤ 350 L/min and ≤ 385 L/min) were identified, which demonstrated better specificity (86.1% and 80.2%, respectively) and higher Youden Index values compared to ≤ 475 L/min. However, both showed poor sensitivity (39.2% and 49.0%, respectively), which limits their utility for screening purposes. For a screening tool, high sensitivity is essential to minimize the risk of false negatives and avoid missing patients at risk of aspiration, particularly in a condition like aspiration pneumonia, which carries significant morbidity and mortality. From a clinical standpoint, the threshold of ≤ 475 L/min, despite its lower specificity (48.5%), aligns better with the intended goal of screening because it correctly identified 72.5% of aspirators, missing approximately 1 in 4 (false-negative rate of 27.5%). In contrast, using the ≤ 350 L/min threshold would miss more than half of aspirators. Nevertheless, the lower specificity of the ≤ 475 L/min cutoff introduces a considerable false-positive burden (48.5%), meaning that more than half of non-aspirators would be incorrectly flagged as at risk. Suboptimal specificity reduces efficiency in high-resource settings. However, in low-resource or high-demand environments where instrumental assessments such as MBS or FEES cannot be performed universally, a screening tool that prioritizes sensitivity can still be valuable for triaging patients who warrant further testing. In practice, we do not foresee PEFmax being used in isolation but rather as an adjunct to the standard clinical swallowing assessment, which remains the primary determinant for referral to instrumental evaluation. In addition, emerging work suggests that other low-cost markers (such as patient-reported outcomes with validated thresholds [32, 33]) may serve as complementary triggers for instrumental assessment. Combining PEFmax with these clinical or patient-reported indicators may offer a more robust multi-parameter screening approach, which warrants future investigation. Additionally, determining thresholds for initiating swallowing intervention was not within the scope of this study, and triggering treatment based solely on worst MBS aspiration status would be overly simplistic and inconsistent with clinical decision-making, which integrates physiologic, efficiency, and patient-reported considerations. Future work should evaluate whether PEFmax, in combination with clinical or patient-reported measures, could contribute to multi-parameter pathways for follow-up or triage, with possible implications for more efficient use of healthcare resources.

The present study focused on cough PEFmax to screen for aspiration status in the post-RT setting and the cohort’s predominance of male and oropharyngeal cancer survivors reflects the current epidemiology of HNC typically seen in our post-RT surveillance clinic. Nonetheless, generalizability to women, to other HNC subsites (e.g., larynx, hypopharynx), and to surgically treated patients may be limited. These findings may also not extend to other clinical timepoints such as pre-treatment or postoperative screening. In addition, this study was designed to examine the prediction of aspiration status, but there are many other critical factors that instrumental swallowing testing derives that may be missed by screening only for aspiration. These include, among others, measures of efficiency and pathophysiology of RAD.

Conclusion

The current findings suggest that voluntary cough PEFmax could serve as an adjunctive rapid and pragmatic objective clinical metric of a patient’s physiologic capacity to defend the airway. While the association with aspiration is statistically significant, the modest effect size and moderate discriminative ability indicate that PEFmax should be considered an adjunctive rather than stand-alone metric. As such, it could serve as a valuable complementary measure to the bedside or clinical swallowing evaluation, helping to classify patients at risk for aspiration after RT for HNC.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

In accordance with the Contributor Roles Taxonomy (CRediT, https://credit.niso.org/), the contributing authors have designated responsibilities and individual author attribution. The corresponding author (KAH) assume responsibility for role assignment, and all contributors have been given the opportunity to review and confirm assigned roles. BM: Data curation, Formal analysis, Interpretation, Writing – original draft. LBS: Data curation, Interpretation, Writing – original draft. CLW: Data curation, Formal analysis. GAE, EKP: Conceptualization, Methodology, Writing – review & editing. KH: Conceptualization, Funding acquisition, Writing – original draft, review & editing, Supervision.

Funding

The parent trial for this work was supported by a Survivorship Seed Money Award from the MD Anderson Institutional Research Grant Program. BM is supported by the UTHealth Houston Innovation for Cancer Prevention Research Training Program Postdoctoral Fellowship (Cancer Prevention and Research Institute of Texas grant #RP210042). The content is solely the responsibility of the authors and does not necessarily reflect the official views of the National Institutes of Health or the Cancer Prevention. All authors have no conflicts of interest to declare.

Data availability

The datasets generated during and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare that they have no conflicts of interest related to this work.

Previous meetings

Preliminary data from this study were presented at the 2018 American Head & Neck Society 2018 Annual Meeting April 18-19, 2018, National Harbor, MD