How Do We Know Swallowing Has Failed — An Introduction to Oropharyngeal Dysphagia

If your arm cannot go up, it shows on the outside. Not being able to walk also shows. Yet when swallowing goes wrong, there might be no sound at all, and food and saliva slip into the airway in the meantime. Among the functions covered in rehabilitation, swallowing is one where the very fact that it has failed may not be visible on the outside.

First, let us clarify a few terms. Dysphagia is a state where the act of swallowing is not working properly or is misaligned. Swallowing is divided into three stages—the oral, pharyngeal, and esophageal phases—depending on where the food bolus is located. Inside the throat, the food path passes right behind the entrance to the airway, and that entrance is the larynx. The vocal cords that produce voice are inside the larynx.

Aspiration is when things like food, water, or saliva enter the airway below the vocal cords. Silent aspiration is aspiration that occurs without outward signals like coughing, throat clearing, or voice changes. It is also called asymptomatic aspiration. These terms and the three stages of swallowing were first established by the primer "Swallowing, Speech and Language — The Crossing Where the Food Path Meets the Airway".

That primer also outlined the flow for finding dysphagia in three steps. Simple checks first screen out who needs a closer look, healthcare professionals examine those who are caught directly, and if necessary, the act of swallowing is directly viewed through imaging or endoscopy. This article follows those three steps one by one. We look at what each step catches and what it misses, who goes up to which step, and how reliable the score is that makes the final judgment at the end.

Why We Must Look — It Is Common and Moves Together with Other Outcomes

Right after a stroke, 42% of acute patients experience dysphagia, and pneumonia and death are more frequent on the side with dysphagia (PMID 35562660). This is the number covered in the primer. Dysphagia does not end with swallowing; it moves together with other outcomes like pneumonia and death. This is an association of appearing together, and how much aspiration causes pneumonia is examined in "Is Aspiration Pneumonia Really Caused by Aspiration?".

The question of this article is already contained within the value of 42%. In the same analysis, studies that found dysphagia using direct tests such as imaging or endoscopy showed a higher prevalence (PMID 35562660). This means that the more closely one looks, the more one finds. The answer to how common dysphagia is depends on what it was found with, and therefore knowing how to look comes first.

What Is Being Measured — How Far Into the Airway It Went, What Is Being Eaten

There is one more step before aspiration in the terms used to judge swallowing. Penetration is when food or water has entered the larynx but stays above the vocal cords. When it goes below the vocal cords, it is aspiration. The Penetration-Aspiration Scale (PAS) is a score from 1 to 8 that rates how deep the entered material went and whether it was pushed back out. The top score of 8 is an aspiration that went below the vocal cords with no cough reflex, which is a silent aspiration (PMID 25782622).

If the PAS measures what happened during a single swallow, there is also a score that measures what is being eaten day by day. The Functional Oral Intake Scale (FOIS) is a score that divides the level of eating by mouth from 1 to 7. Levels 1–3 are a state of receiving nutrition through a tube, levels 4–5 are a state of eating by mouth by changing the firmness and thickness of food, and levels 6–7 are a state of eating by mouth without changing the thickness of food (PMID 31866129). The PAS and FOIS answer different questions. This difference becomes important again later when reading the effects of treatment.

We also need to decide what to consider a disease. Presbyphagia is a change in which the functional reserve of swallowing naturally decreases as we age, and it is different from dysphagia (PMID 27049901, PMID 19483069). There are also conditions tied to age. Sarcopenic dysphagia is a state where swallowing becomes difficult due to sarcopenia, in which the skeletal muscles of the whole body and the swallowing muscles decrease together. The way swallowing muscles decrease with age is especially pronounced in the geniohyoid muscle connecting the chin and the hyoid bone, and in the tongue (PMID 27049901). Therefore, saying "that is just how it is when you get old" is only half right. A misalignment that goes beyond the changes of aging is a disease, and disease can be measured.

Screening — It Is Better to Do It, But Look at the Type of Evidence

Screening is a short, simple check to filter out who needs a closer look first. Pneumonia occurred less often in hospitals that had a structured screening process in place. In 2,532 stroke cases across 15 acute hospitals, the pneumonia rate was 2.4% in hospitals with a formal screening procedure and 5.4% in those without (P=0.0016). The difference remained even after adjusting for stroke severity (PMID 16109909).

Screening early points in the same direction. In data from 12,276 patients with acute ischemic stroke, the pneumonia rate was 29.7% in patients with dysphagia and 3.7% in those without. An odds ratio (OR) is a value comparing the likelihood of an event occurring between two groups, and a value less than 1 means it happened less often in that group. Early screening within 24 hours of admission was associated with a reduced risk of pneumonia (OR 0.68; 0.52–0.89) (PMID 27074007).

Guidelines split this flow into two tiers. The 2021 guidelines from the European Stroke Organisation and the European Society for Swallowing Disorders recommend swallowing screening for all stroke patients to prevent post-stroke pneumonia and early death, and rate the quality of that evidence as moderate. They then suggest a more detailed swallowing evaluation for patients who screen positive for dysphagia risk, stating that the level of evidence for this suggestion is low (PMID 34746431). Everyone goes through screening, and those caught there move on to the next step.

However, the numbers above are all observational data. Observational data is data gathered by watching and collecting what has already happened, whereas a randomized controlled trial is a study where people are randomly divided like a draw of lots, with the intervention applied to only one side and compared. A review systematically gathering literature up to November 2016 for the American Heart Association's 2018 guidelines found 3 randomized controlled trials addressing screening procedures, and concluded that randomized trial data were insufficient to determine whether screening reduces pneumonia, death, or dependency (PMID 29367332). Why it is difficult to read things moving together as cause and effect is covered in "If People Who Take It Live Longer, How Were They Different to Begin With? — Confounding". While observational data accumulates toward the side that screening is better and European guidelines recommend it with moderate-level evidence, the randomized trials gathered by this review alone could not determine whether screening reduces pneumonia.

Bedside Evaluation — How Much Do Eyes and Ears Catch?

How well a test catches something is measured by two ratios. Sensitivity is the proportion of people who actually have the disease whom the test correctly flags as positive, and specificity is the proportion of people who do not have the disease whom the test correctly flags as negative. Those two terms are covered in detail in "What a Test Actually Tells You — An Introduction to Sensitivity, Specificity, and Pretest Probability".

When relying on ad-hoc notes left without a set procedure, most cases are missed. When swallowing-related notes left by nurses and doctors in the medical records of stroke patients at one tertiary hospital were compared with a speech-language pathologist's evaluation, the sensitivity was 36.7% and the specificity was 94.2% (PMID 29497830). Out of ten people with dysphagia, fewer than four were caught in the records, and it was rare to mistakenly write that someone without dysphagia had it. The authors recommend using a formal screening procedure whose accuracy and consistency have been validated as a measurement tool (PMID 29497830).

A clinical swallow examination is an evaluation where medical staff observe the patient at the bedside without instruments, looking for signs such as coughing after swallowing or changes in voice. Even this evaluation performed by experts is not perfect. In 49 acute stroke patients within 24 hours of onset, six clinical indicators were compared with the results of directly viewing the inside of the throat with an endoscope. The six indicators were voice abnormalities, articulation abnormalities, abnormalities in the gag reflex where the throat is stimulated and causes retching, abnormalities in intentional coughing, coughing after swallowing, and changes in voice after swallowing. The results showed a sensitivity of 86%, a specificity of 30%, and a false-positive rate—the proportion of people without aspiration risk who were mistakenly judged to be at risk—of 70% (PMID 12140648). This means that while most dangerous cases were caught, most non-dangerous cases were also judged to be dangerous. The authors concluded that because of this limitation, the majority of acute stroke patients must be able to receive reliable instrumental tests in a timely manner (PMID 12140648).

Aspiration Without Cough — How Do You Know If No Alarm Sounds?

In the previous study on clinical swallowing examinations, the main pattern of errors was overdiagnosis—judging a person not at risk to be at risk. Another thing that bedside evaluations easily miss is silent aspiration. Silent aspiration has been reported in 2% to 25% of acute stroke patients (PMID 16362510). Because silent aspiration has been identified using various methods, it is difficult to compare studies (PMID 16362510). Proposed mechanisms include weakness or misalignment of throat muscles, reduced sensation in the throat, and impairment of the ability to reflexively cough. The same review notes a caveat as well. Even healthy people can aspirate somewhat silently at night, and while many studies found that silent aspiration is associated with poor outcomes, not all studies did (PMID 16362510).

Measured right after a stroke, the proportion is not low. In a small single-hospital feasibility study of hyperacute patients within 72 hours of onset who even underwent endoscopy, 7 out of 23 patients who completed the examination (30%) showed silent aspiration (PMID 37158000).

The hyperacute study mentioned earlier notes, as an established fact, that clinical swallowing examinations are not reliable for detecting silent aspiration and that there is no consensus on which examination items best catch silent aspiration (PMID 37158000). Other external bodily signals have also been tested. When 189 stroke patients had their blood oxygen saturation measured with a sensor clipped to the finger during swallowing, not a single one of the 6 patients who silently aspirated experienced a drop in oxygen saturation. The authors concluded that bedside evaluation and oxygen saturation measurement, either alone or combined, are insufficient to screen out aspiration compared to an X-ray visualization of swallowing (PMID 17095740). In this study, neither bedside evaluation nor the external signal of oxygen saturation was enough to catch aspiration.

Two paths remain. One is to intentionally trigger and test the alarm system of the cough reflex, and the other is to look directly inside the airway. The first path is not yet simple. The citric acid cough test is an examination that has people inhale citric acid as a mist to see if a cough is triggered. In 134 subacute stroke patients, a period immediately following the acute phase, the sensitivity of this test in screening out silent aspiration was merely 0.19, leading to the conclusion that it is difficult to use as a standalone screening tool (PMID 25782622). The brief cough test measures the time until the first cough appears after inhaling the same mist. For this test, the sensitivity for distinguishing silent aspiration among patients with confirmed aspiration was 87.1%, and the authors suggest that this test could be a useful way to screen for aspiration regardless of whether a cough is present (PMID 25229026). The fact that two tools based on the same concept yield such different numbers represents the current standing of this field. However, the two figures cannot be directly compared because the comparison groups are different. The former is the value for screening out silent aspiration among all subacute stroke patients suspected of having dysphagia, while the latter is the value for distinguishing people who aspirate without a cough from among patients with brain conditions including stroke who already have confirmed aspiration. And in one of the randomized trials found by the American Heart Association review mentioned earlier, adding a cough reflex test to routine screening yielded no evidence of reduced pneumonia (PMID 29367332). According to the preliminary results of the hyperacute feasibility study mentioned earlier, the sensitivity of the clinical swallowing examination may actually be higher than that of the cough reflex test, and the authors note that a larger study comparing the two is indeed feasible, and that further research is needed to determine the best method and the respective sensitivities and specificities (PMID 37158000).

Instrumental Assessments — Videofluoroscopy and Endoscopy

Instrumental assessments are tests that directly view the act of swallowing using imaging or an endoscopy, and there are two main types. Videofluoroscopic swallowing study (VFSS) uses X-ray imaging to watch food mixed with a contrast agent—meaning a substance that shows up clearly on X-rays—pass through. Fiberoptic endoscopic evaluation of swallowing (FEES) uses a thin endoscope to directly look inside the pharynx and larynx. The pharynx is the space inside the back of the throat, extending downward from behind the nose and mouth, where both food and air pass through.

The gold standard refers to the most reliable test used as a benchmark when comparing other tests. It was not clear which of the two instrumental assessments is the gold standard. In a study where 49 patients with oropharyngeal dysphagia underwent both tests simultaneously, the ratings for penetration, aspiration, and pharyngeal residue showed a high correlation between the two tests. The authors concluded that the two can be regarded equally as the gold standard for these three aspects (PMID 34678269). The same study also shows what each test misses. Instances where the videofluoroscopy score was higher occurred when there was an event during the moment of swallowing that was not visible on endoscopy, or when videofluoroscopy lost lateral-view information and misjudged it. Instances where the endoscopy score was higher typically occurred when the endoscopy showed the structures inside the throat more clearly (PMID 34678269).

In reality, the same swallow does not receive the same score. When 15 evaluators across 12 institutions scored simultaneous recordings of both tests for 15 patients, the PAS score from the endoscopic video was significantly higher than that from videofluoroscopy—meaning by an amount too large to be considered mere chance—with a mean difference of 1.15 points (PMID 17906496). This means that when comparing two PAS scores, one must first verify whether the scores came from the same test method. While the two instrumental assessments are used as standard tests to directly observe aspiration, they do not assign the exact same score to the same swallow. The tests themselves also entail burdens. In the aforementioned hyperacute-phase study, anxiety about the endoscopic exam was the primary barrier to participation, and patients did not always tolerate endoscopy well (PMID 37158000).

Imaging also reveals things beyond the pharynx. In a videofluoroscopy assessment of 57 patients with a first ischemic stroke, 57.9% showed abnormalities in clearing the esophagus (PMID 36179218). However, the authors note that simply observing the esophagus during this test cannot diagnose esophageal abnormalities, and it can be used as a clue to screen out individuals who need a separate esophageal examination (PMID 36179218). This means that ending the test after looking only at the pharynx can leave some areas missed.

Superimposing a breathing record onto the imaging makes it possible to measure when a person swallows. Because swallowing and breathing share the same pathway, the exact moment of breathing at which one swallows is tied to protecting the airway. In an exploratory study where videofluoroscopy and respiratory monitoring were performed simultaneously in 31 Parkinson's disease patients with dysphagia, these patients scored worse than existing data from 84 healthy controls in the movement that elevates the larynx and the movement that closes the laryngeal vestibule. Furthermore, in 26% of all trials, the swallow was initiated while inhaling, and such swallows were associated with impairments in the movement initiating the pharyngeal phase and the movement closing the laryngeal vestibule. The authors note that this needs to be reconfirmed in a larger sample (PMID 41329938). This coordination also serves as a target for training (PMID 41466097).

There are also findings showing that aspiration seen on imaging does not directly dictate long-term survival. In a retrospective cohort study following 182 post-stroke dysphagia patients, aspiration confirmed by videofluoroscopy after the acute phase did not predict long-term survival over a median of 30.8 months. The HR, meaning the hazard ratio, is a value comparing the rate at which that event occurs at any given time between two groups. Factors that independently predicted survival were older age, recurrent stroke, the need for tube feeding (HR 2.07), and wheelchair dependency (HR 2.83) (PMID 15742979). However, the outcome examined by this retrospective study was solely long-term survival. The fact that it did not predict survival does not mean it is unrelated to the risk of pneumonia. Silent aspiration has been linked—though not in all studies, in many—to increased morbidity and mortality (PMID 16362510), and is known to increase the risk of pneumonia in the early period following a stroke (PMID 37158000).

The Scores Used to Judge Also Fluctuate

Even when instrumental tests show images, reading those images as scores remains a separate task. A systematic review examining the measurement properties of scales that read videofluoroscopic and endoscopic images looked at 45 studies and 39 scales. Its conclusion was that the evidence was insufficient to recommend any individual scale as valid and reliable for interpreting these images (PMID 30019178). Although the two tests are commonly considered the gold standard, the starting point of the same review is that there is no consensus on the standard scales for analyzing those images (PMID 30019178).

The most widely used PAS is no exception. A single lump of food or water swallowed at one time is called a bolus. When 38 people who received head and neck cancer treatment were made to swallow the same type of bolus twice in succession and compared, there was no difference in the average of the whole group, but the scores fluctuated significantly twice within the same person (in the three boluses with the largest differences, the within-individual coefficient of variation comparing score fluctuation to the mean was 0.458–0.759). The authors suggest using the highest PAS score for each bolus type when assessing swallowing safety in head and neck cancer patient studies (PMID 28593510). This means it is difficult to judge a person's swallowing based on a single swallow result. Screening tools are similar, and the sensitivity of 8 studies covering screening tools that nurses can use varied from 29–100% (PMID 28757720).

Even when there are tests that directly view aspiration, the scores reading those images are not yet sufficiently validated.

What You Measure With Also Changes the Answer to Treatment

The entirety of the treatment evidence is based on a 2026 Cochrane review (181 trials, 11,500 people), which is a systematic review that gathers and evaluates studies using a set procedure, and was synthesized by "Swallowing, Speech and Language — The Crossing Where the Food Path Meets the Airway" (PMID 42573047). This article adds just one thing. Even with the same treatment, the answer varies depending on what was used to measure it.

In the 2018 Cochrane review, swallowing treatment tended to reduce the length of hospital stay, the proportion of people left with dysphagia at the end of the trial, and chest infections, and the certainty of the evidence for these three ranged from moderate to very low. By contrast, death or dependency, which that edition chose as its primary outcome, could be measured by only a single trial (306 people), and showed no significant effect (OR 1.05; 0.63–1.75). The penetration-aspiration score also did not decrease significantly (standardized difference −0.37) (PMID 30376602). The 2026 edition set swallowing impairment itself as the most important outcome. Here, electrical stimulation to the neck and applying a weak electrical current to the brain from outside the head appeared to reduce swallowing impairment with moderate certainty, and behavioral therapy pointed the same way but with very low certainty (PMID 42573047).

The risk ratio is the value obtained by dividing the proportion of people who experienced the event when followed through to the end. In an analysis pooling 44 randomized trials involving acute and critical care patients, neurostimulation treatments applying electrical or magnetic stimulation to the ear, neck, or brain improved swallowing function immediately after treatment, increased the proportion of people who could eat by mouth again (risk ratio 1.39; 1.12–1.74), and were also associated with a reduction in pneumonia (risk ratio 0.62; 0.39–0.98). However, they did not significantly reduce aspiration itself (risk ratio 0.67; 0.36–1.26), and there were no significant differences in swallowing function at 3 months and the length of hospital stay. The authors conclude that various neurostimulation treatments are effective therapies for improving swallowing function in these patients (PMID 41772815). Since some indicators improve together while others do not even with the same treatment, when reading treatment effects, look first at what was used to measure them. The finding that the aspiration score did not decrease significantly while pneumonia trended downward is an example of this.

The decisions made after a test detects something each have their own questions. Whether the consistency that best eliminated aspiration in the laboratory also outperformed the others regarding pneumonia at three months is examined by "The Decision to Change Viscosity — What Thickening Actually Changes" using two paired trials. Whether fasting—waiting without eating—is safe is covered by "Is Aspiration Pneumonia Really Caused by Aspiration?", and the decision to deliver nutrition through a tube is covered by "When Eating Is Not Possible — Facing the Tube-Feeding Decision".

Honest Unresolved Questions

• The best composition of screening has not been established. It remains unclear which items to bundle or through which procedure to have someone swallow (PMID 22308250), and sensitivity varies from 29–100% depending on the tool (PMID 28757720). There were only 3 randomized trials that examined whether screening changes outcomes (PMID 29367332).
• The best way to find silent aspiration has not been established. Only when detection methods improve can their frequency and prognostic significance be determined (PMID 16362510), and more studies comparing clinical swallow examinations and cough reflex tests in the hyperacute phase are needed (PMID 37158000).
• The scales for reading imaging have not been validated. Among 39 scales, none can be recommended (PMID 30019178), and even the PAS fluctuated in head and neck cancer patients when swallowing the same bolus twice (PMID 28593510).

To Go Deeper

• When reading papers — first separate whether the outcome indicator is swallowing physiology (such as the PAS), the feeding method (FOIS), or a clinical event (pneumonia or mortality). An example is how the 2018 Cochrane review showed that the proportion of remaining dysphagia decreased, but no significant effect was seen in mortality or dependence measured by a single trial (PMID 30376602).
• When reading assessments — look at what test was used, what bolus, and how many times they were made to swallow. The testing method moved the PAS by more than 1 point (PMID 17906496), and in head and neck cancer patients, the second swallow of the same bolus also came out differently (PMID 28593510).
• If you remember only one thing — coughing is a sign of aspiration, not the definition of aspiration.

Source Papers

• Prevalence of dysphagia and risk of pneumonia and mortality in acute stroke patients: a meta-analysis. PMID 35562660 (2022)
• Respiratory-Swallowing Phase Pattern and Swallowing Impairment in Individuals With Parkinson's Disease. PMID 41329938 (2026)
• Respiratory-swallow coordination training using bimodal signal biofeedback for patients with post-stroke dysphagia: a randomized controlled trial. PMID 41466097 (2026)
• Dysphagia and tube feeding after stroke are associated with poorer functional and mortality outcomes. PMID 31866129 (2020)
• Presbyphagia and Sarcopenic Dysphagia: Association between Aging, Sarcopenia, and Deglutition Disorders. PMID 27049901 (2014)
• Senescent swallowing: impact, strategies, and interventions. PMID 19483069 (2009)
• Formal dysphagia screening protocols prevent pneumonia. PMID 16109909 (2005)
• Dysphagia in Patients with Acute Ischemic Stroke: Early Dysphagia Screening May Reduce Stroke-Related Pneumonia and Improve Stroke Outcomes. PMID 27074007 (2016)
• European Stroke Organisation and European Society for Swallowing Disorders guideline for the diagnosis and treatment of post-stroke dysphagia. PMID 34746431 (2021)
• Effect of Dysphagia Screening Strategies on Clinical Outcomes After Stroke: A Systematic Review for the 2018 Guidelines for the Early Management of Patients With Acute Ischemic Stroke. PMID 29367332 (2018)
• Screening for Dysphagia in Adult Patients with Stroke: Assessing the Accuracy of Informal Detection. PMID 29497830 (2018)
• Aspiration risk after acute stroke: comparison of clinical examination and fiberoptic endoscopic evaluation of swallowing. PMID 12140648 (2002)
• Can pulse oximetry or a bedside swallowing assessment be used to detect aspiration after stroke? PMID 17095740 (2006)
• Comparison of Simultaneous Swallowing Endoscopy and Videofluoroscopy in Neurogenic Dysphagia. PMID 34678269 (2022)
• Assessing penetration and aspiration: how do videofluoroscopy and fiberoptic endoscopic evaluation of swallowing compare? PMID 17906496 (2007)
• Abnormal Esophageal Clearance Identified During Modified Barium Swallow Study in an Acute Poststroke Cohort. PMID 36179218 (2022)
• Psychometric Properties of Visuoperceptual Measures of Videofluoroscopic and Fibre-Endoscopic Evaluations of Swallowing: A Systematic Review. PMID 30019178 (2019)
• Within-Bolus Variability of the Penetration-Aspiration Scale Across Two Subsequent Swallows in Patients with Head and Neck Cancer. PMID 28593510 (2017)
• Validity and reliability of swallowing screening tools used by nurses for dysphagia: A systematic review. PMID 28757720 (2016)
• Valid items for screening dysphagia risk in patients with stroke: a systematic review. PMID 22308250 (2012)
• Silent aspiration: what do we know? PMID 16362510 (2005)
• Screening for silent aspiration in hyperacute stroke: A feasibility study of clinical swallowing examination and cough reflex testing. PMID 37158000 (2023)
• Usefulness of citric cough test for screening of silent aspiration in subacute stroke patients: a prospective study. PMID 25782622 (2015)
• Usefulness of the simplified cough test in evaluating cough reflex sensitivity as a screening test for silent aspiration. PMID 25229026 (2014)
• Swallowing therapy for dysphagia in acute and subacute stroke. PMID 42573047 (2026)
• Swallowing therapy for dysphagia in acute and subacute stroke. PMID 30376602 (2018)
• Neurostimulation for Dysphagia in Acute and Critical Care: A Systematic Review, Pairwise and Network Meta-Analysis. PMID 41772815 (2026)
• Can the aspiration detected by videofluoroscopic swallowing studies predict long-term survival in stroke patients with dysphagia? PMID 15742979 (2004)