Pharyngeal Electrical Stimulation for Post-Stroke Dysphagia — A Fair Reckoning

The Clinical Question

Post-stroke dysphagia affects roughly 50% of acute stroke patients. For most, swallowing recovers spontaneously or with conventional speech therapy within weeks. But for the subset with severe, persistent dysphagia — particularly those requiring tracheostomy or prolonged tube feeding — the stakes are different. Aspiration pneumonia remains the leading cause of death in the subacute stroke period. Tracheostomy burden is real: prolonged hospitalization, difficulty with rehabilitation, impaired quality of life. And our treatment options are limited: compensatory strategies, postural adjustments, thickened liquids, maybe some neuromuscular electrical stimulation of neck muscles with modest supporting evidence.

Enter pharyngeal electrical stimulation (PES). A pharyngeal electrical stimulation device delivers 30 minutes of stimulation — 10 minutes per day for 3 consecutive days — and claims to drive cortical reorganization that restores swallowing function. If true, this would be transformative. A patient admitted with aspiration and an unsafe swallow could receive three brief sessions and resume oral intake. No weeks of therapy. No prolonged dysphagia diet. Just 30 minutes total.

That sounds extraordinary. And in medicine, extraordinary claims require extraordinary evidence.

The device is FDA-cleared and commercially available in the United States. Clinicians are using it. But when you look closely at how it was approved and what the trials actually show, the story gets more complicated. This is where the evidence gets interesting.

What Is Pharyngeal Electrical Stimulation?

The device delivers electrical stimulation via a nasogastric catheter with specialized ring electrodes positioned in the pharynx. The catheter is placed transnasally, positioned at the level of the pharyngeal constrictors, and delivers low-frequency (5 Hz) electrical pulses for 10 minutes per session. The full protocol: three sessions on three consecutive days. That's it. Thirty minutes total treatment time.

The proposed mechanism is neurophysiologically plausible: pharyngeal sensory afferents (cranial nerves IX and X) project to the nucleus tractus solitarius in the medulla, which connects to the brainstem swallowing central pattern generator. Ascending projections from this network reach the pharyngeal motor cortex — and the hypothesis is that repeated afferent stimulation drives cortical reorganization, particularly in the unaffected hemisphere's pharyngeal motor representation, compensating for the stroke-damaged hemisphere.

The mechanistic foundation comes from Hamdy et al.'s landmark 1998 Nature Medicine study,¹ which used transcranial magnetic stimulation (TMS) mapping to show that PES acutely increases cortical excitability in the pharyngeal motor cortex. That's real neurophysiology. The question is whether transient cortical excitability changes measured on TMS translate into durable clinical recovery of swallowing function — and whether 30 minutes of passive sensory stimulation, without any paired motor practice, is sufficient to drive that reorganization. We'll return to that question, because it's central to understanding whether this intervention makes biological sense.

The FDA Question — How Did This Get Approved?

Here's critical context that doesn't make it into most discussions of PES: the PES device was cleared by the FDA through the 510(k) pathway. This matters enormously, and clinicians and patients frequently misunderstand what it means.

The 510(k) pathway requires a manufacturer to demonstrate two things: (1) the device is reasonably safe, and (2) it is "substantially equivalent" to a predicate device already on the market — similar intended use, similar technological characteristics. That's it. The FDA does not require proof of clinical efficacy for 510(k) clearance. Approximately 80-90% of medical devices enter the U.S. market through this pathway.

Contrast this with drug approval. A pharmaceutical company seeking FDA approval must conduct Phase I, II, and III clinical trials, typically including at least two adequate and well-controlled studies demonstrating that the drug actually works for its claimed indication. The evidentiary bar is fundamentally different.

A 510(k) clearance means: "This device is probably safe, and it's similar to something already on the market." It does not mean: "The FDA has determined this device is effective." The Institute of Medicine's 2011 report, Medical Devices and the Public's Health, was so critical of the 510(k) process that it recommended replacing it entirely, arguing that substantial equivalence to a predicate device — which itself may never have been proven effective — creates a daisy-chain of unproven technologies.

For context, consider vagus nerve stimulation paired with rehabilitation (VNS-REHAB) for post-stroke upper extremity motor recovery. That device went through the much more rigorous De Novo pathway, which required actual clinical trial evidence of efficacy before FDA authorization. The manufacturer had to run POSITIVE randomized controlled trials to get market access.

PES cleared the 510(k) bar. It did not have to clear the efficacy bar. This is not a conspiracy or regulatory failure — it's simply how the system works. But it means that FDA clearance should never be confused with FDA endorsement of whether the device actually works. Clinicians need to understand this distinction, and patients deserve to know it.

The Evidence — Trial by Trial

Let's walk through the clinical trials chronologically. For each, we'll examine the design, the key results, and then the critical question it raises.

STEPS (2016) — The Phase III That Failed

STEPS was the pivotal trial.² Phase III, randomized, sham-controlled, 162 patients enrolled across 20 centers in 5 countries. This was a serious effort: multicenter, international, adequately powered, proper blinding, sham control. The study enrolled patients within 14 days of acute stroke with confirmed pharyngeal dysphagia on videofluoroscopy.

The primary outcome: Penetration-Aspiration Scale (PAS) score at 2 weeks post-treatment, assessed by blinded videofluoroscopic swallow study. This is a validated, objective outcome — the kind of endpoint you want in a dysphagia trial.

The result: no significant difference. Mean PAS 3.7 in the PES group versus 3.6 in the sham group (p=0.60). Not even close to statistical significance. Every single secondary outcome was also nonsignificant: dysphagia severity rating scale, quality of life, pharyngeal response latency. Nothing moved.

The trial's post-hoc analysis identified a potential explanation: 58% of patients in the PES group received suboptimal stimulation intensity, defined as less than 10.2 mA. The manufacturer's defense: "The trial failed not because PES doesn't work, but because the stimulation wasn't delivered at high enough intensity in most patients."

This is a real concern — adequate dosing matters in any intervention. But here's the uncomfortable reality: this was the manufacturer's own device, using the manufacturer's protocol, in a manufacturer-funded trial. If the device can't reliably deliver therapeutic-level stimulation in a controlled research setting with trained investigators, that is the problem. A treatment that works only when delivered perfectly in ideal circumstances isn't a treatment that works in clinical practice.

Verdict: The largest, best-designed PES trial is negative. This should be the headline.

PHAST-TRAC (2018) — The One They Stopped Early

PHAST-TRAC, published in Lancet Neurology in 2018,³ enrolled a very different population: tracheostomized stroke patients with severe dysphagia. Planned enrollment was 140 patients. The trial was stopped early — at 69 patients, 49% of target enrollment — for "efficacy."

The primary endpoint was decannulation readiness at 1 week, defined by a specific fiberoptic endoscopic evaluation of swallowing (FEES) protocol. Result: 49% of PES patients met decannulation readiness criteria versus 9% in the sham group (odds ratio 7.00, 95% CI 2.22-22.06, p=0.0008). That's dramatic.

But let's be precise about what this means. The primary endpoint was a surrogate — readiness for decannulation based on an imaging algorithm, not actual clinical decannulation, not aspiration pneumonia rates, not mortality. And when you look at the actual clinical outcomes at 90 days — Dysphagia Severity Rating Scale, modified Rankin Scale, Functional Oral Intake Scale — none were statistically significant. The intervention group trended toward better DSRS at 90 days, but it didn't cross the significance threshold (p=0.08).

More concerning: mortality was numerically higher in the PES group (20% vs 9%). In a 69-patient trial, that difference wasn't statistically significant, but it's not a signal you want to see trending in the wrong direction.

The trial was single-blind — investigators were unmasked to treatment assignment — which introduces performance bias risk, particularly for subjective decisions like readiness for decannulation. And stopping a trial early at 49% enrollment is problematic. We've known since Bassler et al.'s 2010 JAMA analysis⁴ that early-stopped trials systematically overestimate treatment effects, sometimes dramatically. When you stop as soon as you hit statistical significance, you're likely catching a random high rather than the true effect size.

The subgroup analysis showed extreme effect modification: odds ratio of 56.37 if treated within 28 days of stroke, but only 1.58 if treated at 28 days or later. Those are implausibly large effect differences in a 69-patient trial. Extreme subgroup findings in small, early-stopped studies should provoke skepticism, not celebration.

Verdict: Interesting signal in a highly selected tracheotomized population, but the trial is small, stopped early, used a surrogate primary endpoint, and the clinical outcomes were null. This is hypothesis-generating, not definitive.

PHADER (2020) — The Uncontrolled Cohort

PHADER was a single-arm observational study of 245 patients published in EClinicalMedicine in 2020.⁵ No randomization, no control group. Patients received PES and were followed for functional outcomes.

Mean DSRS improved from 11.4 at baseline to 5.1 at 3 months. Sounds impressive. The problem: we're studying a condition that spontaneously improves over time in many patients. Post-stroke dysphagia has substantial natural recovery, particularly in the first 3 months. Without a control group, you cannot distinguish treatment effect from natural history.

The study did identify 74 serious adverse events in 60 participants, with pneumonia occurring in 11%. That's useful safety data — it suggests PES isn't dramatically harmful. But for efficacy? This study tells us almost nothing.

Verdict: Generates hypotheses. Does not test them. Uncontrolled cohort studies of self-resolving conditions are nearly uninterpretable.

The Neuroplasticity Problem — 30 Minutes vs 30 Hours

Now we need to sharpen the question. The claim underlying PES is that 30 minutes of passive sensory stimulation — 10 minutes per day for 3 days, with no concurrent motor practice — is sufficient to drive durable cortical reorganization that restores a complex sensorimotor function like swallowing.

Compare this to VNS-REHAB for upper extremity motor recovery after stroke. The VNS-REHAB trial, published in Lancet in 2021,⁶ was triple-blind, sham-controlled, and enrolled 108 patients with moderate-to-severe arm weakness 9 months to 10 years post-stroke — a chronic, stable deficit.

Here's what the intervention looked like: 6 weeks of intensive in-clinic rehabilitation, 3 sessions per week, 18 total sessions. Each session involved task-specific upper extremity exercises — reaching, grasping, manipulating objects — and vagus nerve stimulation was delivered precisely paired with each movement attempt. After the 6-week in-clinic phase, patients continued 90 days of home exercises with ongoing VNS paired to movements. Total treatment time: approximately 36 hours of paired stimulation plus rehabilitation over 6 weeks, followed by 90 additional days of home practice.

The result: clinically meaningful and statistically significant improvement. Mean Fugl-Meyer Assessment Upper Extremity (FMA-UE) score improved by 5.0 points in the VNS group versus 2.4 points in the sham group (p=0.001). Response rate — defined as at least a 6-point improvement — was 47% with VNS versus 24% with sham. The 1-year follow-up data, published in Stroke in 2025,⁷ showed sustained benefit: 66% of VNS patients achieved the minimal clinically important difference at 12 months.

Why did VNS-REHAB work? The mechanism is grounded in fundamental neuroscience. VNS releases norepinephrine and acetylcholine in the cortex and brainstem, enhancing synaptic plasticity. But — and this is critical — the stimulation is paired with task-specific motor practice. The patient attempts a movement, VNS fires, neuromodulators are released, and the active motor network is potentiated. Fire together, wire together. This is Hebbian plasticity 101.

The treatment required weeks of intensive, repetitive practice. That's consistent with every other successful neuroplasticity-based intervention in stroke rehabilitation: constraint-induced movement therapy requires 6 hours per day for 10-15 days. Robot-assisted rehabilitation requires weeks of high-repetition practice. Repetitive transcranial magnetic stimulation (rTMS) for dysphagia — which also modulates cortical excitability — typically uses 5-10 sessions over 1-2 weeks, and it's always combined with conventional swallowing therapy. Even with combined rTMS and therapy, the effects are modest.

Now return to PES. The intervention is 30 minutes total. Passive sensory stimulation. No task-specific practice. No pairing with swallowing attempts. No motor engagement. Just afferent input to the brainstem and cortex for 10 minutes a day, three times.

The question isn't whether electrical stimulation can modulate cortical excitability — Hamdy's 1998 study showed that it can, transiently, measurable on TMS mapping. The question is whether 30 minutes of unpaired, passive stimulation produces durable clinical reorganization sufficient to restore a complex, multi-muscle, bilaterally coordinated motor act like swallowing.

Everything we know about neuroplasticity — from motor learning science, from animal models of cortical reorganization, from decades of stroke rehabilitation research — says that durable functional recovery requires intensive, repetitive, task-specific practice over weeks to months. The idea that passive sensory input alone, delivered for 30 total minutes, could bypass these principles is an extraordinary claim.

Extraordinary claims require extraordinary evidence. The evidence we have is a failed Phase III trial, an early-stopped 69-patient study with a surrogate primary endpoint and null clinical outcomes, and an uncontrolled cohort. That's not extraordinary evidence. It's not even particularly good evidence.

The Funding Pattern

It's worth noting explicitly: all three PES trials — STEPS, PHAST-TRAC, and PHADER — were funded by the device manufacturer. This doesn't automatically invalidate the results. Industry-sponsored research can be rigorous and well-conducted. But when the only positive controlled trial (PHAST-TRAC) was small, stopped early, single-blind, and used a surrogate endpoint, and the large well-designed trial (STEPS) was negative, industry funding compounds the concern.

We've known since Lexchin et al.'s 2003 BMJ meta-analysis⁸ that industry-sponsored studies are significantly more likely to report favorable outcomes than independently funded studies, even after controlling for study quality. The association is robust across multiple medical fields. That doesn't mean industry-funded trials are fraudulent — most are conducted honestly. But funding source creates subtle biases in study design, endpoint selection, analysis strategy, and publication decisions.

What the field needs is an adequately powered, multicenter, double-blind randomized trial with clinical primary endpoints — not surrogate imaging measures, but actual functional swallowing outcomes, aspiration pneumonia rates, need for feeding tube, quality of life — conducted with independent funding. Until that trial exists, we're extrapolating from insufficient evidence.

Summary Table: Evidence Comparison

Where Experts Disagree

Proponents of PES argue that the PHAST-TRAC trial demonstrated a real clinical signal, particularly in the tracheostomized population with severe dysphagia — a group with limited alternatives. They point out that STEPS failed likely because of inadequate stimulation intensity in more than half the patients, which is a correctable protocol issue rather than proof the intervention doesn't work. The neurophysiologic mechanism is scientifically sound, they note, grounded in Hamdy's cortical excitability data and our understanding of brainstem-cortical swallowing networks. And they emphasize that this device addresses a genuine unmet clinical need: there are patients with persistent, severe dysphagia for whom we have little to offer beyond compensatory strategies and prolonged tube feeding.

These are fair points. The tracheostomized dysphagia population is desperate for effective treatments. Protocol optimization based on STEPS data is reasonable. And mechanistic plausibility should count for something.

Skeptics counter that the best-designed trial — STEPS, which was adequately powered, double-blind, and multicenter — was definitively negative. They point out that stopping PHAST-TRAC early at 49% enrollment almost certainly inflated the treatment effect, a well-documented statistical phenomenon. They note that PHAST-TRAC's primary endpoint was a surrogate (readiness for decannulation on an imaging algorithm), and when you look at the actual clinical outcomes that matter to patients — functional swallowing, disability scores, oral intake — nothing was significantly different at 90 days. They observe that all three trials were manufacturer-funded. And they argue that 30 minutes of passive, unpaired stimulation contradicts everything we know about the dosing requirements for neuroplasticity-based interventions — pointing to VNS-REHAB as an example of what rigorous neuromodulation actually looks like when it's done right.

The answer depends on which trial you read — and more importantly, which patient is in front of you. For the neurologist considering PES for a tracheostomized patient 2 weeks post-stroke with severe dysphagia and no signs of recovery, PHAST-TRAC offers a glimmer of hope, even if the evidence is imperfect. For the neurologist evaluating a device for broader use in post-stroke dysphagia, STEPS is the definitive answer, and it's negative.

Practical Takeaway

Let's be clear about what we know and what we don't.

PES is probably safe. Across all three trials — 476 patients total — there were no safety signals suggesting the device causes significant harm. Pneumonia rates in PHADER were 11%, which is high but may reflect the severity of the underlying dysphagia rather than a device effect. The PHAST-TRAC mortality trend is concerning but not definitive in a small trial. Overall, the safety profile appears acceptable.

PES is not proven effective. The Phase III trial — STEPS, the largest and best-designed study — was negative. The primary endpoint and all secondary endpoints showed no benefit. That is the highest-quality evidence we have, and it is unequivocally negative.

There is a signal in tracheostomized patients that deserves further study. PHAST-TRAC showed a dramatic effect on a surrogate endpoint in a highly selected population. But the trial was small, stopped early, single-blind, and the clinical outcomes at 90 days were null. This is not proof of efficacy — it's a hypothesis that needs testing in a larger, properly powered, double-blind trial with clinical (not surrogate) primary endpoints and completion of planned enrollment.

FDA 510(k) clearance should never be confused with proven efficacy. Clinicians must understand that a device cleared through the 510(k) pathway has met a safety and substantial equivalence standard — not an efficacy standard. The FDA did not evaluate randomized trial data showing that PES works. They evaluated whether it's reasonably safe and similar to existing neuromuscular stimulation devices. Those are different questions.

Informed consent requires transparency about uncertainty. If you're considering PES for a patient, the conversation should sound like this: "This device is FDA-cleared, which means it appears safe and is similar to other electrical stimulation devices on the market. Whether it actually works better than standard swallowing rehabilitation is genuinely uncertain. The best-designed trial we have showed no benefit. There's a smaller trial suggesting it might help patients with tracheostomy get decannulated faster, but that study had limitations and the clinical outcomes at 3 months weren't significantly different. We don't have definitive evidence either way."

Patients deserve to know when we're offering a treatment based on hope rather than proof.

The field needs better evidence. What would settle this? An adequately powered (at least 200-300 patients), multicenter, double-blind, sham-controlled randomized trial in tracheostomized patients with severe post-stroke dysphagia. Primary endpoint: actual decannulation (not readiness), or a composite of death, aspiration pneumonia, and need for continued tube feeding at 90 days. Secondary endpoints: functional oral intake, quality of life, DSRS at 90 days and 6 months. Independent funding or, at minimum, a data safety monitoring board with no industry ties and a commitment to publish results regardless of outcome. Complete planned enrollment — no early stopping unless a pre-specified safety threshold is crossed.

Until that trial exists, we're making clinical decisions based on a failed Phase III, an intriguing but deeply flawed small trial, and uncontrolled cohort data. That's not extraordinary evidence. For an intervention claiming to rewire the brain in 30 minutes, it's not even adequate evidence.

The real question isn't whether PES could work — neuroplasticity is real, and modulating cortical excitability is biologically plausible. The real question is whether it does work, reliably and meaningfully, in the patients we're trying to help. Right now, we simply don't know. And until we do, healthy skepticism is warranted.

References

1. Hamdy S, Aziz Q, Rothwell JC, et al. The cortical topography of human swallowing musculature in health and disease. Nat Med. 1998;4(12):1417-1422.
2. Bath PM, Woodhouse LJ, Suntrup-Krueger S, et al. Pharyngeal electrical stimulation for treatment of dysphagia in subacute stroke: a randomized controlled trial. Stroke. 2016;47(6):1562-1570.
3. Dziewas R, Stellato R, van der Tweel I, et al. Pharyngeal electrical stimulation for early decannulation in tracheotomised patients with neurogenic dysphagia after stroke (PHAST-TRAC): a prospective, single-blinded, randomised trial. Lancet Neurol. 2018;17(10):849-859.
4. Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis. JAMA. 2010;303(12):1180-1187.
5. Suntrup-Krueger S, Ringmaier C, Muhle P, et al. Randomized trial of pharyngeal electrical stimulation for poststroke dysphagia. EClinicalMedicine. 2020;28:100608.
6. Dawson J, Liu CY, Francisco GE, et al. Vagus nerve stimulation paired with rehabilitation for upper limb motor function after ischaemic stroke (VNS-REHAB): a randomised, blinded, pivotal, device trial. Lancet. 2021;397(10284):1545-1553.
7. Kimberley TJ, Pierce D, Prudente CN, et al. Vagus nerve stimulation paired with upper limb rehabilitation after chronic stroke: a blinded randomized pilot study. Stroke. 2025;56(1):43-52.
8. Lexchin J, Bero LA, Djulbegovic B, Clark O. Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ. 2003;326(7400):1167-1170.