Vagus Nerve Stimulation (VNS)

Overview

Vagus nerve stimulation is, in many ways, the strangest member of the neuromodulation family. It requires surgery to implant, then works slowly over months, and once it's working, the benefit keeps accruing for a year or two. That profile makes it genuinely useful for a specific population — patients with chronic, markedly treatment-resistant depression who have tried nearly everything else — and genuinely mismatched for acute presentations where speed matters.

This page is written for both patients considering VNS and clinicians supporting or managing VNS patients. The distinction between those audiences isn't as sharp as it is for TMS or ECT; VNS is enough of a commitment that patients considering it tend to want the same level of detail their clinicians do.

What VNS is

VNS uses a small implanted pulse generator — about the size of a cardiac pacemaker — to deliver regular electrical pulses to the left cervical vagus nerve. The generator sits under the skin below the left collarbone. A thin wire (the "lead") runs from the generator, tunneled beneath the skin of the chest and neck, to the vagus nerve itself, where helical electrodes wrap around the nerve and deliver the stimulation. The left side is used because the right vagus carries the main supply to the heart's sinoatrial node, and stimulating it risks slowing the heart rhythm (Mandalaneni & Rayi, StatPearls 2023).

Once turned on, the device runs continuously on a preset schedule — for example, stimulating for 30 seconds every 5 minutes, 24 hours a day. Patients don't have to do anything to activate it. The device doesn't need to be charged. The battery is not rechargeable and lasts several years (published estimates range from about five to more than ten, depending on the model and the stimulation settings) before requiring a minor outpatient replacement procedure.

The implanted system

Illustration of the implanted system: generator in the upper chest, lead running up the left neck to the vagus nerve. Image: "Vagus nerve stimulation" by Manu5 / Scientific Animations, via Wikimedia Commons, licensed CC BY-SA 4.0. Unmodified.

What it looks like on X-ray

Radiograph of an implanted VNS system (two views): the generator sits below the left clavicle; the lead is coiled for strain relief and ends at the helical electrodes in the neck. This image is from a patient treated for epilepsy; the hardware is the same as that used for depression. Image: "Vagal nerve stimulator VNS bei Epilepsie 56W – CR – 001" by Hellerhoff, via Wikimedia Commons, licensed CC BY-SA 4.0. Unmodified.

How VNS was discovered for depression

VNS has been used for epilepsy since the 1990s (FDA approval for seizures came in 1997). The antidepressant effect was noticed clinically — epilepsy patients treated with VNS reported that their mood improved, out of proportion to their seizure control. This led to dedicated open-label trials for depression beginning in the late 1990s (Rush 2000; Sackeim 2001; Marangell 2002), and eventually to FDA approval for treatment-resistant depression in 2005.

How it works in the brain

The vagus nerve is a bidirectional information highway between the body and the brainstem. About 80% of its fibers are afferent — they carry signals up from the body to the brain — and only about 20% are efferent (Bonaz 2018; StatPearls 2023). VNS therefore mostly activates the sensory side of the nerve. Those afferents terminate in the nucleus tractus solitarius (NTS) in the medulla, where glutamate is the principal transmitter at the first synapse (Berthoud & Neuhuber 2000). From the NTS, signals are relayed to the locus coeruleus and the raphe nuclei — the brain's main norepinephrine and serotonin sources — as well as the parabrachial nucleus, hypothalamus, thalamus, amygdala and, through these, the hippocampus and prefrontal cortex (Henry 2002; Ruffoli 2011; Bonaz 2018; Austelle 2022; StatPearls 2023).

The mechanism is thought to involve gradual changes in noradrenergic and serotonergic signaling, plus effects on neurotrophic factors (BDNF) and neuroplasticity that accumulate over months (Carreno & Frazer 2017). The slow onset almost certainly reflects the fact that these are adaptive changes that take time to develop, rather than acute effects that could be measured immediately. It is worth saying plainly that the precise mechanism by which VNS helps patients who did not respond to conventional antidepressants is not established (Carreno & Frazer 2017); what is established is the clinical time course described in the evidence section.

How VNS is different from other treatments

VNS has an unusual therapeutic profile that sets it apart from TMS, ECT, ketamine, and pharmacotherapy:

• Slow onset. Benefits typically emerge over 3–12 months of continuous stimulation, with full effect often taking 12–24 months. This is dramatically longer than any other antidepressant intervention. Patients considering VNS need to understand they are making a long-horizon decision.
• Durable response. When VNS works, the benefit tends to accumulate rather than wane. In the 5-year registry the gap between VNS and treatment-as-usual widened over time rather than narrowing (see the curves in The evidence), and in RECOVER about 80% of patients with a meaningful benefit at 12 months still had it at 24 months (Conway 2026, IJNP).
• Continuous, not episodic. Unlike TMS or ECT treatment courses, VNS is always on. There is no "index course" to complete, no tapering schedule, no maintenance sessions to schedule.
• Complementary. VNS is typically used alongside ongoing pharmacotherapy and sometimes alongside maintenance ECT or TMS, rather than as a replacement. It's additive rather than substitutive.
• One-time procedure. Unlike TMS (requires 30+ daily visits) or ECT (requires acute course + ongoing maintenance), VNS requires one surgical procedure and then background operation. The tradeoff: the procedure is surgical rather than outpatient clinical.

Who VNS is for

VNS is FDA-approved for adults in a major depressive episode with chronic or recurrent depression who have not had an adequate response to four or more adequate antidepressant treatments (FDA 2005). In practice, the typical VNS candidate has:

• Long-standing depression (years, often decades)
• Multiple failed medication trials across drug classes
• Often prior experience with ECT, with partial benefit but unable or unwilling to continue maintenance ECT indefinitely
• Capacity to tolerate a slow, uncertain response curve and continue with ongoing support while waiting for benefit

A 2026 international expert consensus statement (32 VNS experts, modified Delphi; McAllister-Williams 2026) reached agreement on the same profile: four or more adequate antidepressant trials in the current episode, an episode of at least two years or more than three episodes in five years, and a need for long-term maintenance rather than acute rescue. The same panel agreed that medication should not be reduced for at least a year after implantation, and then only gradually. The statement was funded by LivaNova, and about half the panel report financial relationships with the company; the recommendations are consistent with, and do not extend beyond, the FDA label and the trial entry criteria.

Where those four failed treatments sit on the broader ladder of treatment resistance — and why "four" is the threshold rather than two or eight — is the subject of the next section.

The ECT-dependent patient: a particularly good fit

One specific population stands out as especially well-suited to VNS: patients who have responded well to ECT but are dependent on ongoing maintenance ECT to stay well. Maintenance ECT is effective but logistically heavy — ongoing procedures under general anesthesia, accumulating cognitive effects, time off work, transportation needs, periodic relapses when sessions are missed or spaced too far apart.

The Aaronson 2017 registry specifically examined this group and found that patients with a history of ECT response had the highest 5-year cumulative response rate of any subgroup reported (71.3% with adjunctive VNS vs. 56.9% with treatment as usual alone). Notably, ECT non-responders also did better with VNS than without it (59.6% vs. 34.1%). The clinical reasoning is that ECT responsiveness identifies a depression that responds to neuromodulation broadly, not just to the specific mechanism of seizure induction. VNS appears able to step into that role — offering a less disruptive, less cognitively demanding, "always on" approach to keeping these patients well.

A 10-case series of patients receiving both maintenance ECT and VNS (Aaronson 2021) and a French case series of 7 patients (Aboubakr 2023) both reported that patients were able to substantially reduce the frequency of maintenance ECT after VNS implantation, with a subset able to discontinue ECT entirely (4 of 7 in the French series). These are small, uncontrolled case series, not randomized comparisons; the signal is consistent with the larger registry data but should be weighed accordingly.

When VNS isn't the right fit

VNS is less appropriate for patients who need rapid response to acute suicidality or crisis — by the time VNS is meaningfully working, months have passed. For those situations, ECT or rapid-acting ketamine are much better fits. VNS is also not usually first-line for moderate treatment-resistant depression where TMS hasn't been tried; the cost, surgical nature, and slow timeline of VNS make it a later-in-the-sequence option rather than an early one.

Patients should also be in a position to commit to long-term follow-up. VNS isn't a procedure you get and then forget about — parameter adjustments and monitoring are part of the treatment, particularly during the first 1–2 years.

Staging treatment resistance: when is enough enough?

"Treatment-resistant" is used loosely. The honest version is a ladder: each additional failed treatment lowers the odds that the next medication will work and raises the odds of relapse if it does. Knowing where a patient stands on that ladder is what tells you when an interventional treatment like VNS becomes reasonable.

What STAR*D showed

The Sequenced Treatment Alternatives to Relieve Depression trial (STAR*D) was a federally funded, real-world study of 4,041 outpatients with major depression who were treated in up to four sequential steps: a first antidepressant (citalopram), then switches or augmentations at each subsequent step for those who had not remitted (Rush 2006). Its central finding is the one that matters here — the remission rate fell sharply after the second step, and the relapse rate in the year after remission rose with every step it had taken to get there:

Conway, George and Sackeim: a two-failure definition

In a 2017 JAMA Psychiatry viewpoint titled "Toward an evidence-based, operational definition of treatment-resistant depression: when enough is enough," Charles Conway, Mark George and Harold Sackeim proposed that the field stop using "treatment-resistant" to mean anything from one to eight failed trials and instead adopt a single operational definition: failure to respond to two antidepressant trials delivered at an adequate dose for an adequate duration (Conway 2017). They further proposed staging TRD into two tiers — patients with two failed adequate trials, and patients with three or more — arguing that the STAR*D data support two failures as the point at which the probability of remission with further conventional pharmacotherapy falls markedly, and therefore the point at which more intensive, interventional treatments should enter the conversation rather than a third or fourth medication switch.

Two features of that proposal are worth holding onto clinically:

• "Adequate" has to be documented. A trial only counts if the dose and duration were adequate and the patient actually took it. The Antidepressant Treatment History Form (ATHF) and its short form (ATHF-SF) are the standard instruments for rating this (Sackeim 2019); the RECOVER trial used the ATHF-SF and an independent eligibility committee to confirm every prior failure (Conway 2024). Other staging tools exist — the Thase–Rush and Massachusetts General Hospital models (reviewed in Fava 2003) and the Maudsley Staging Method (Fekadu 2009), which adds illness duration and severity to the count of failed treatments.
• Resistance is a spectrum, not a switch. Two failures is the threshold at which "treatment-resistant" becomes a defensible label; it is not the threshold for VNS. The FDA label, the Medicare coverage definition and the RECOVER entry criteria all sit further up the ladder, at four or more failures with a chronic course.

Where VNS enters

The practical takeaway: a patient with two documented adequate failures is treatment-resistant and should be hearing about interventional options. A patient with four or more, a chronic course, and — typically — prior ECT, TMS or esketamine is in the population in which VNS was actually studied. That is the population described in the next section, and it is much sicker than the phrase "treatment-resistant" usually conveys.

The evidence

The VNS evidence base has been complicated and is worth walking through in detail. Both the strengths and the limitations matter for understanding when VNS makes sense.

Early evidence (2000s) and FDA approval

Early open-label trials in the 2000s (Rush 2000; Sackeim 2001; Marangell 2002) showed promising response rates of 30–50% over 12 months in patients with treatment-resistant depression. These trials established the pattern that has been consistently observed since: slow onset, cumulative benefit over months to years, modest response rates within the first 3 months but meaningful gains by 12 months and beyond.

The pivotal sham-controlled RCT that preceded FDA approval (Rush 2005a) was a 10-week acute trial, and it did not separate active VNS from sham on its primary endpoint. A companion naturalistic 12-month study (Rush 2005b) showed substantially better outcomes in the VNS group than in a comparable treatment-as-usual cohort. The FDA approved VNS for treatment-resistant depression in July 2005 on the basis of the longer-term data, but the failure of the short-term RCT created a credibility problem that has shaped the field's reception of VNS ever since.

On May 4, 2007, the Centers for Medicare & Medicaid Services (CMS) determined that VNS was "not reasonable and necessary" for treatment-resistant depression and issued a national non-coverage determination, citing the 2005 RCT as insufficient evidence (CMS CAG-00313R). Many private insurers followed CMS's lead, which is the principal reason VNS has remained substantially underutilized clinically relative to its evidence profile.

Aaronson 2017: the 5-year observational registry

The single most important piece of long-term evidence for VNS is the Treatment-Resistant Depression Registry (ClinicalTrials.gov NCT00320372), published by Aaronson and colleagues in the American Journal of Psychiatry in 2017. This was a 5-year, prospective, open-label, nonrandomized observational study conducted at 61 US sites with 795 enrolled patients (494 with adjunctive VNS, 301 with treatment as usual), sponsored by the device manufacturer (Cyberonics, now LivaNova).

The patient population was severely treatment-resistant: all participants were in a major depressive episode of at least 2 years' duration or had three or more lifetime episodes, all had failed four or more depression treatments, and the registry's posted baseline data show a mean of 7.9 prior failed treatments (8.0 in the VNS arm, 7.3 in the treatment-as-usual arm). Mean baseline MADRS was 33.1 in the VNS arm and 29.3 in the treatment-as-usual arm — both in the moderate-to-severe range. Patients with psychosis or rapid-cycling bipolar disorder were excluded (Aaronson 2017).

The cumulative figures above answer "how many patients ever responded." The curves below answer a different and arguably more useful question — "what fraction of patients were in response at each visit" — using the outcome data the sponsor posted on ClinicalTrials.gov for this registry. The two arms separate by 3 months and the gap does not close:

Several aspects of the registry results are clinically meaningful:

• The benefit accrued over time. The gap between VNS and treatment-as-usual was visible by 3 months, never closed, and was widest in the final two years — the opposite of the pattern most psychiatric treatments show, where response is highest early and tends to attenuate with time.
• Suicidality outcomes favored VNS. Both arms improved on suicidality measures from baseline, but VNS produced significantly greater reductions on two of three measures (QIDS-SR item 12 and the investigator-completed suicidality assessment), with a strong trend on the third (MADRS item 10) (Aaronson 2017). The posted safety data show 2 suicides in each arm over 5 years, with roughly twice the person-years of observation in the VNS arm.
• Bipolar and unipolar patients both benefited. A subanalysis of the bipolar patients showed similar efficacy, with effects evident by 12 months and continued through 5 years (McAllister-Williams 2020).
• The design has real limitations. This was an open-label, non-randomized observational registry; patients and clinicians were not blinded; patients who chose VNS may have differed from those who did not; the VNS arm started with higher baseline MADRS scores; and the number of patients contributing data fell over time (in the treatment-as-usual arm, from 249 at 3 months to 116 at 60 months). The registry was sponsored by the device manufacturer. These limitations are real, and any honest reading of the data has to account for them.

Even acknowledging those limitations, the size, duration, and consistency of the registry findings make Aaronson 2017 the most informative single long-term dataset on VNS for treatment-resistant depression. It was not, however, the kind of evidence CMS had asked for. When LivaNova requested reconsideration of the 2007 decision in October 2017, CMS responded on February 15, 2019 not with coverage but with Coverage with Evidence Development: Medicare would pay for VNS for treatment-resistant depression only inside a CMS-approved, double-blind, randomized, placebo-controlled trial with at least one year of follow-up (CMS CAG-00313R2). That trial is RECOVER.

The RECOVER trial (2019–2025)

RECOVER (Conway et al., Brain Stimulation 2025; ClinicalTrials.gov NCT03887715) was designed to answer CMS's evidence requirements. It was sponsored by LivaNova and conducted under the CMS coverage-with-evidence-development framework, with Dr. Charles Conway of Washington University as principal investigator. Enrollment began in September 2019 at US sites (84 sites contributed to the unipolar analysis), and the trial enrolled 493 adults with unipolar major depression who had failed at least four adequate antidepressant treatments in the current episode, each failure verified with the ATHF-SF by an independent eligibility committee (Conway 2024).

How ill were the RECOVER patients?

The investigators describe this as a profoundly treatment-resistant cohort, the most resistant we are aware of in a randomized depression trial, and the numbers deserve to be stated exactly, because they define the population to whom the results apply (all figures from Conway 2024, Tables 1–3, unless noted):

What had they already tried? In the current episode alone, the medication classes used by the RECOVER cohort were (Conway 2024, Table 3):

Every patient had received pharmacotherapy in the current episode; 66.1% had received some form of neuromodulation and 75.1% some form of psychotherapy. The baseline paper reports classes tried rather than a count of concurrent medications; the durability paper tracked the number of concurrent psychotropic medications in the second year and found that changes in medication did not account for the durability of benefit (Conway 2026).

Trial design

• All 493 randomized patients were scheduled for implantation (485 were implanted; 15 withdrew before surgery)
• 249 randomized to active stimulation; 244 to sham (device implanted but not activated)
• Active and sham conditions maintained for 12 months, triple-blind
• Primary endpoint: percent of time in MADRS response (≥50% reduction from baseline) across months 3–12, rated by offsite blinded raters
• Secondary endpoints reported by the trial included clinician-rated global improvement (CGI-I), patient self-report (QIDS-SR), offsite blinded clinician ratings (QIDS-C), partial response (≥30% improvement), and quality-of-life and daily-function instruments
• 88.4% of participants completed the 12-month trial
• Parameter adjustments were limited during the blinded period to protect the blind

What RECOVER found

The headline result: the trial missed its primary endpoint. Percent time in MADRS response was 18.9% with active VNS and 16.3% with sham, a difference that was not statistically significant (p=0.137) (Conway 2025, as summarized by Philip & Brunoni 2025). This was a real disappointment for the field and for advocates expecting RECOVER to produce a clean coverage-justifying result.

However, the trial's secondary endpoints told a more complex story. Active VNS produced statistically significant advantages on (Conway 2025):

• Clinician-rated improvement (CGI-I) — both percent time in response (p=0.004) and percent time in partial response (p<0.001)
• Patient self-report (QIDS-SR) — percent time in response (p=0.049)
• Offsite blinded clinician ratings (QIDS-C) — percent time in partial response (p=0.006)
• Quality of life and daily function — detailed below (Rush 2025)

Quality of life and daily function in RECOVER (Rush 2025)

The quality-of-life paper is worth reading in full, because it contains both the positive and the null results, and a fair account has to include both:

Source: Rush AJ et al., Brain Stimulation 2025;18(3):690–700 (abstract). Baseline scores in this cohort were extremely poor: median Q-LES-Q-SF 34 of 70 and median WHODAS 50 of 100 (Conway 2024). The authors' conclusion was that VNS "has a broader therapeutic impact than symptom improvement alone in patients with marked psychosocial impairment."

The pattern across these results is informative: three independent rater perspectives — the treating clinician, the patient, and an offsite blinded rater — all detected an antidepressant signal favoring active VNS, even when the primary MADRS-based metric did not. Partial response (≥30% improvement) was the most sensitive measure, which makes clinical sense in a population where full 50% response and remission are exceptionally hard endpoints to hit. Remission was rare in both arms and did not differ between them.

Why might the primary endpoint have failed?

Several factors plausibly contributed to the sham-arm improvement and the resulting primary endpoint miss (these are the investigators' and commentators' explanations, and they are interpretations rather than established facts):

• Procedural placebo effect. Every participant underwent real surgery and then lived for 12 months with an implanted device that might or might not be on. Patients who volunteer for that are highly invested in their care and receive enhanced clinical attention as part of the trial protocol — all of which can produce substantial improvement on patient-reported and clinician-rated measures.
• Enhanced treatment as usual. All RECOVER participants received structured ongoing care that was likely more intensive than what most of these patients had received in routine practice.
• The MADRS as a metric. Percent time in MADRS response across a 9-month window may simply be too narrow a metric to capture the kinds of partial, gradually-accumulating improvements that VNS produces in this population — which is exactly what the partial-response secondary endpoints showed (Sackeim 2025 examines the symptom-level data).
• Blinding-driven parameter restrictions. Real-world VNS practice involves periodic parameter optimization (current, pulse width, duty cycle) over the first 1–2 years to find the dose that works for an individual. RECOVER constrained these adjustments to maintain blinding, which may have reduced active-arm efficacy.
• Time. Most of the improvement in the active arm emerged in the final three months of the blinded year, consistent with everything known about VNS's slow onset (Washington University news release, 2024). A 12-month window may have captured the beginning of the effect rather than its full extent.

The second year: durability

Participants originally randomized to active VNS were followed for a further 12 months of open-label stimulation (n=214). Among those with at least a meaningful benefit at 12 months, a median of 83.1% still had it at 18 months and 81.3% at 24 months across seven symptom, function and quality-of-life measures. Roughly 30% of those with no meaningful benefit at 12 months developed one during the second year. The authors reported that these results were not accounted for by changes in psychotropic medications or by ECT, TMS or ketamine use in the second year (Conway 2026, Int J Neuropsychopharmacol). The obvious caveat: the sham arm was not followed in the same way, so this is single-arm durability data, and the relapse rate one would expect in this population without VNS has to be inferred from elsewhere — for instance, the 71% one-year relapse rate after STAR*D's fourth step.

Where the evidence and coverage stand now

On June 4, 2025, LivaNova announced that it had begun the formal CMS process to seek reconsideration of national Medicare coverage for VNS in unipolar treatment-resistant depression, citing the published RECOVER analyses and the 24-month durability data (LivaNova press release, 2025). A further RECOVER analysis has identified baseline predictors of response (Aaronson 2025), a manufacturer-funded international consensus statement on patient selection and management was published in March 2026 (McAllister-Williams 2026), and a formal review of the trial's lessons appeared in the American Journal of Psychiatry in 2026 (Conway 2026). As of this page's last update (September 2026) we are not aware of a published CMS decision on that request; the current status is on the CMS Medicare Coverage Database under "Vagus Nerve Stimulation for Treatment Resistant Depression." Whether or not CMS moves from coverage-with-evidence-development to standard coverage will substantially affect access to VNS for the patients who would benefit most.

Practical considerations

The implantation procedure

VNS implantation is usually an outpatient surgical procedure performed by a neurosurgeon, general, vascular or ENT surgeon trained in the procedure (StatPearls 2023). Our program does not implant VNS devices; we evaluate candidates, refer to an implanting center (see Resources) and manage devices afterward. Almost all patients go home the same day; some surgeons keep patients overnight. Details:

• Duration: Typically 1–2 hours under general anesthesia
• Incisions: Two small incisions — one below the left collarbone for the pulse generator, one in the left side of the neck for the lead attachment to the vagus nerve
• Recovery: Mild incision pain and restricted activity for 1–2 weeks; most patients return to normal activity quickly
• Surgical complication rate: Low; risks include infection, bleeding, injury to surrounding nerves or vessels. Serious complications are rare.

Activation and titration

The device is typically activated 2–4 weeks post-op to allow incision healing. At the first activation visit, initial parameters are set at a low level and gradually increased over subsequent visits based on tolerability. Most patients notice a mild sensation in the throat during stimulation (throat tightness, slight voice change, coughing). These sensations usually become less noticeable with time as the patient adapts, and parameters can be adjusted if they remain bothersome.

Device management and follow-up

Once activated, the device requires periodic check-ups by the psychiatrist or neurologist managing the patient:

• Every 3–6 months during the first 1–2 years, during which parameters may be adjusted to optimize response
• Less frequent (annually) once stable
• Visits involve interrogation of the device via an external programmer (painless, takes a few minutes), review of response and side effects, and any parameter adjustments

Side effects

The most common VNS side effects occur during stimulation and typically diminish with time or with parameter adjustment:

• Voice changes (most common). Stimulation spreads to the recurrent laryngeal branch of the vagus and contracts the left vocal fold during each pulse train, which is why the voice sounds different while the device is on (Ardesch 2010)
• Cough
• Throat tightness or discomfort
• Shortness of breath with exertion during stimulation (dyspnea was the one adverse event that was significantly more frequent with active than sham VNS in RECOVER, p=0.035; no new safety signals were identified) (Conway 2025)
• Sore throat, neck pain, or swallowing difficulty during stimulation (listed in the manufacturer's labeling)
• Infection at the implant site (uncommon, mostly early post-op); a serious infection can require explantation
• Surgical risks: bleeding, injury to nearby nerves or vessels, and rarely injury to the recurrent laryngeal nerve with longer-lasting vocal-cord weakness (vocal cord paralysis is listed in the manufacturer's labeling)

Most side effects are mild and don't require discontinuation. Severe or persistent side effects can usually be managed with stimulation parameter adjustment (decreasing current, pulse width, or duty cycle). The manufacturer's labeling asks that patients be monitored closely for clinical worsening and suicidality, particularly around medication changes and stimulation-parameter changes, and notes that safety and effectiveness have not been established in acute suicidality (LivaNova patient safety information, February 2025).

Battery and device replacement

The battery lasts several years; published estimates range from about five to more than ten depending on the model and stimulation parameters, and the manufacturer's current labeling should be consulted for the implanted model. When the battery depletes, the pulse generator is replaced with an outpatient procedure — a smaller surgery than the original implantation, since the lead remains in place. Some patients may have multiple generator replacements over the course of decades of treatment.

Cost and insurance coverage

Published estimates for device plus implantation range from about $25,000 (CNN, December 2024) to about $50,000 (NBC News, August 2018), before surgeon, anesthesia and follow-up charges; confirm current pricing with the hospital. Insurance coverage for VNS in depression is currently limited:

• Medicare: Covered only through Coverage with Evidence Development inside a CMS-approved trial (NCD 160.18, as revised February 15, 2019); generator replacement for already-implanted patients is covered. A reconsideration request has been pending since June 2025.
• Private insurance: Variable; some insurers cover, most don't; prior authorization is almost always required
• Manufacturer programs: LivaNova runs a Patient Access Program that supports prior-authorization and medical-necessity documentation for commercially insured and Tricare patients; it does not advertise financial assistance with the cost of the device. Verify current terms with the device representative.

MRI compatibility

Modern VNS devices are MR-conditional, meaning MRIs can be safely performed with specific scanning protocols (typically head MRI at certain strengths and with certain coil types). Older VNS devices may be MR-unsafe or have more restrictive MR conditions. Stimulation is programmed off for the scan and restored afterward, and the patient's hand-held magnet must not enter the scanner room. Always confirm the specific device model against the manufacturer's current MRI guidelines before imaging.

Where VNS fits in the neuromodulation landscape

For most patients with treatment-resistant depression, the typical sequence is:

1. Antidepressant trials across drug classes with augmentation strategies
2. Psychotherapy (typically CBT or IPT)
3. TMS (noninvasive, FDA-cleared for TRD, reasonable first neuromodulation option)
4. ECT (highest efficacy, best for severe or urgent cases) or ketamine/esketamine (rapid-acting)
5. VNS (for patients with chronic illness who've been through the above with partial or inadequate benefit)

This isn't a rigid sequence — clinical urgency, patient preference, side effect profiles, and practical considerations (insurance, geography, logistics) all shift the ordering. But the general principle holds, and it maps onto the staging ladder above: VNS tends to earn its place after other options have been tried, not as a first or second-line intervention.

VNS vs. maintenance ECT

A meaningful comparison for some patients is VNS versus indefinite maintenance ECT. Both are appropriate for chronic treatment-resistant depression with demonstrated ECT responsiveness. Tradeoffs:

• Maintenance ECT: Known response, faster time to effect, but requires ongoing procedures under general anesthesia, has cumulative cognitive costs, and is logistically burdensome
• VNS: Single surgical procedure then background operation, no ongoing general anesthesia, minimal cognitive burden — but slow onset and uncertain response

Some patients do well transitioning from maintenance ECT to VNS; others need both strategies concurrently; others do better continuing with maintenance ECT. The right choice depends on the patient's specific history and preferences. As discussed under Who VNS is for, the Aaronson 2017 registry showed the highest 5-year cumulative response rate of any reported subgroup in patients with prior ECT response — this is the patient profile where the case for considering VNS is strongest.

Ongoing management of existing VNS patients

Patients with existing VNS devices are typically managed by their outpatient psychiatrist, with parameter adjustments and troubleshooting provided in collaboration with the VNS device manufacturer's clinical support team (LivaNova field clinical engineers). This is a routine part of the psychiatric care of VNS patients and doesn't typically require specialized neurosurgical follow-up unless there's a device complication.

If you or your patient have a VNS device and need ongoing management — including parameter adjustments, response evaluation, side effect management, or coordination of care — our program welcomes these patients. Use the general Prisma Health Neuromodulation Program contact on the home page.

Resources: programs, studies, patient stories and further reading

Every link below was opened and checked on September 18, 2026. Each entry says who publishes it and, where the page shows one, its date. Manufacturer pages are labeled as such. None of these organizations pays us to be listed, and we receive nothing for these links. Our program does not implant VNS devices; for patients who have one, we manage the device with the manufacturer's field clinical engineers for programming support, as every program does.

References

Primary sources for every figure and claim on this page, organized thematically. PMIDs and DOIs are included where they exist. Each citation was checked against PubMed, ClinicalTrials.gov or the issuing agency on September 18, 2026.

Staging treatment resistance and STAR*D

1. Rush AJ, Trivedi MH, Wisniewski SR, Nierenberg AA, Stewart JW, Warden D, et al. Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. Am J Psychiatry. 2006;163(11):1905–1917. doi:10.1176/ajp.2006.163.11.1905. PMID: 17074942. (Remission rates by step 36.8%, 30.6%, 13.7%, 13.0% on the QIDS-SR₁₆; cumulative remission 67%; relapse rates during follow-up rising by step.)
2. Pigott HE, Kim T, Xu C, Kirsch I, Amsterdam J. What are the treatment remission, response and extent of improvement rates after up to four trials of antidepressant therapies in real-world depressed patients? A reanalysis of the STAR*D study's patient-level data with fidelity to the original research protocol. BMJ Open. 2023;13(7):e063095. doi:10.1136/bmjopen-2022-063095. PMID: 37491091. (Cumulative remission 35.0% using the protocol-specified HRSD outcome.)
3. Rush AJ, Trivedi M, Fava M, Thase M, Wisniewski S. The STAR*D data remain strong: reply to Pigott et al. Am J Psychiatry. 2023;180(12):919–920. doi:10.1176/appi.ajp.20230869. PMID: 38037409.
4. Conway CR, George MS, Sackeim HA. Toward an evidence-based, operational definition of treatment-resistant depression: when enough is enough. JAMA Psychiatry. 2017;74(1):9–10. doi:10.1001/jamapsychiatry.2016.2586. PMID: 27784055. (Proposes TRD = failure of two adequate antidepressant trials, with staging into two versus three or more failures.)
5. Sackeim HA, Aaronson ST, Bunker MT, Conway CR, Demitrack MA, George MS, et al. The assessment of resistance to antidepressant treatment: rationale for the Antidepressant Treatment History Form: Short Form (ATHF-SF). J Psychiatr Res. 2019;113:125–136. doi:10.1016/j.jpsychires.2019.03.021. PMID: 30974339.
6. Fava M. Diagnosis and definition of treatment-resistant depression. Biol Psychiatry. 2003;53(8):649–659. doi:10.1016/s0006-3223(03)00231-2. PMID: 12706951. (Reviews the Thase–Rush and Massachusetts General Hospital staging models.)
7. Fekadu A, Wooderson S, Donaldson C, Markopoulou K, Masterson B, Poon L, Cleare AJ. A multidimensional tool to quantify treatment resistance in depression: the Maudsley staging method. J Clin Psychiatry. 2009;70(2):177–184. doi:10.4088/jcp.08m04309. PMID: 19192471.
8. Rush AJ, Aaronson ST, Demyttenaere K. Difficult-to-treat depression: a clinical and research roadmap for when remission is elusive. Aust N Z J Psychiatry. 2019;53(2):109–118. doi:10.1177/0004867418808585. PMID: 30378447.

Original approval era

1. Rush AJ, George MS, Sackeim HA, et al. Vagus nerve stimulation (VNS) for treatment-resistant depressions: a multicenter study. Biol Psychiatry. 2000;47(4):276–286. doi:10.1016/s0006-3223(99)00304-2. PMID: 10686262.
2. Sackeim HA, Rush AJ, George MS, et al. Vagus nerve stimulation (VNS) for treatment-resistant depression: efficacy, side effects, and predictors of outcome. Neuropsychopharmacology. 2001;25(5):713–728. doi:10.1016/S0893-133X(01)00271-8. PMID: 11682255.
3. Marangell LB, Rush AJ, George MS, et al. Vagus nerve stimulation (VNS) for major depressive episodes: one year outcomes. Biol Psychiatry. 2002;51(4):280–287. doi:10.1016/s0006-3223(01)01343-9. PMID: 11958778.
4. Rush AJ, Marangell LB, Sackeim HA, et al. Vagus nerve stimulation for treatment-resistant depression: a randomized, controlled acute phase trial. Biol Psychiatry. 2005;58(5):347–354. doi:10.1016/j.biopsych.2005.05.025. PMID: 16139580. (The 10-week sham-controlled RCT that did not separate active from sham on its primary endpoint.)
5. Rush AJ, Sackeim HA, Marangell LB, et al. Effects of 12 months of vagus nerve stimulation in treatment-resistant depression: a naturalistic study. Biol Psychiatry. 2005;58(5):355–363. doi:10.1016/j.biopsych.2005.05.024. PMID: 16139581. (The 12-month naturalistic comparator study, published in the same issue, that informed FDA approval.)
6. U.S. Food and Drug Administration. Premarket approval P970003/S050, VNS Therapy System, supplemental indication for treatment-resistant depression, approved July 15, 2005. FDA PMA database (accessdata.fda.gov).

Long-term outcomes — the 5-year registry

1. Aaronson ST, Sears P, Ruvuna F, Bunker M, Conway CR, Dougherty DD, Reimherr FW, Schwartz TL, Zajecka JM. A 5-year observational study of patients with treatment-resistant depression treated with vagus nerve stimulation or treatment as usual: comparison of response, remission, and suicidality. Am J Psychiatry. 2017;174(7):640–648. doi:10.1176/appi.ajp.2017.16010034. PMID: 28359201. (n=795, 61 sites, prospective open-label observational. Cumulative 5-year response 67.6% VNS+TAU vs. 40.9% TAU; remission 43.3% vs. 25.7%; ECT-responder subgroup 71.3% vs. 56.9%; ECT non-responders 59.6% vs. 34.1%. An erratum was published in Am J Psychiatry 2017;174(9):907.)
2. Cyberonics, Inc. A long-term, prospective, observational, multi-center patient outcome registry to collect data in patients with treatment-resistant depression (TRD) who are currently in a major depressive episode. ClinicalTrials.gov identifier NCT00320372; results first posted December 2015. clinicaltrials.gov/study/NCT00320372. (Source of the visit-by-visit response, remission, Q-LES-Q-SF and mortality data plotted on this page.)
3. Conway CR, Kumar A, Xiong W, Bunker M, Aaronson ST, Rush AJ. Chronic vagus nerve stimulation significantly improves quality of life in treatment-resistant major depression. J Clin Psychiatry. 2018;79(5):18m12178. doi:10.4088/JCP.18m12178. PMID: 30152645. (Quality-of-life analysis of the same registry cohort.)
4. McAllister-Williams RH, Sousa S, Kumar A, Greco T, Bunker MT, Aaronson ST, Conway CR, Rush AJ. The effects of vagus nerve stimulation on the course and outcomes of patients with bipolar disorder in a treatment-resistant depressive episode: a 5-year prospective registry. Int J Bipolar Disord. 2020;8(1):13. doi:10.1186/s40345-020-0178-4. PMID: 32358769. (Bipolar subgroup analysis from the 5-year registry.)

VNS combined with or replacing maintenance ECT

1. Aaronson ST, Goldwaser EL, Kutzer DJ, McAllister-Williams RH, Sackeim HA. Vagus nerve stimulation in patients receiving maintenance therapy with electroconvulsive therapy: a series of 10 cases. J ECT. 2021;37(2):84–87. doi:10.1097/YCT.0000000000000724. PMID: 34029305.
2. Aboubakr O, et al. Vagus nerve stimulation allows to cease maintenance electroconvulsive therapy in treatment-resistant depression: a retrospective monocentric case series. Front Psychiatry. 2023;14:1305603. doi:10.3389/fpsyt.2023.1305603. PMID: 38352166. (7 patients; 4 of 7 stopped maintenance ECT after VNS implantation.)

The RECOVER trial and analyses (2024–2026)

1. LivaNova. A prospective, multi-center, randomized controlled blinded trial demonstrating the safety and effectiveness of VNS Therapy System as adjunctive therapy versus a no stimulation control in subjects with treatment-resistant depression (RECOVER). ClinicalTrials.gov identifier NCT03887715; study start September 26, 2019. clinicaltrials.gov/study/NCT03887715.
2. Conway CR, Aaronson ST, Sackeim HA, Duffy W, Stedman M, Quevedo J, et al. Clinical characteristics and treatment exposure of patients with marked treatment-resistant unipolar major depressive disorder: a RECOVER trial report. Brain Stimul. 2024;17(2):448–459. doi:10.1016/j.brs.2024.03.016. PMID: 38574853. (Baseline characterization: mean MADRS 34.3; mean 13.3 / median 11 lifetime failed treatments; mean current-episode duration 17.8 years; 70.8% with prior ECT, TMS or esketamine in the current episode; 24.9% employed; 77% with suicidal ideation.)
3. Conway CR, Aaronson ST, Sackeim HA, George MS, Zajecka J, Bunker MT, et al. Vagus nerve stimulation in treatment-resistant depression: a one-year, randomized, sham-controlled trial. Brain Stimul. 2025;18(3):676–689. doi:10.1016/j.brs.2024.12.1191. PMID: 39706521. (Main RECOVER results: n=493; primary MADRS percent-time-in-response endpoint not met; significant active-VNS advantages on CGI-I response and partial response, QIDS-SR response, QIDS-C partial response; dyspnea more frequent with active VNS, p=0.035; 88.4% completion.)
4. Rush AJ, Conway CR, Aaronson ST, George MS, Riva-Posse P, Dunner DL, et al. Effects of vagus nerve stimulation on daily function and quality of life in markedly treatment-resistant major depression: findings from a one-year, randomized, sham-controlled trial. Brain Stimul. 2025;18(3):690–700. doi:10.1016/j.brs.2024.12.1187. PMID: 39701918. (Mini-Q-LES-Q p=0.050; WPAI item 6 p=0.050; Q-LES-Q p=0.061; WHODAS 2.0 p=0.304; EQ-5D VAS p=0.125; time in meaningful benefit favored active VNS on Q-LES-Q, Mini-Q-LES-Q and WPAI item 6.)
5. Sackeim HA, Aaronson ST, Conway CR, et al. Characterizing the effects of vagus nerve stimulation on symptom improvement in markedly treatment-resistant major depressive disorder: a RECOVER trial report. J Affect Disord. 2025;380:135–145. doi:10.1016/j.jad.2025.03.124. PMID: 40127770.
6. Philip NS, Brunoni AR. Beyond the primary outcome: why the RECOVER trial for vagus nerve stimulation matters in treatment-resistant depression. Brain Stimul. 2025;18(3):619–621. doi:10.1016/j.brs.2025.03.001. PMID: 40054575. PMCID: PMC12834334. (Editorial; source for the 18.9% vs. 16.3% MADRS percent-time figures quoted on this page.)
7. Aaronson ST, et al. Prognostic and prescriptive predictors of treatment response to adjunctive VNS therapy in major depressive disorder: a RECOVER trial report. J Clin Psychiatry. 2025;86(3):25m15850. doi:10.4088/JCP.25m15850. PMID: 40673907.
8. Conway CR, Rush AJ, Aaronson ST, Bunker MT, Gordon C, George MS, et al. Durability of the benefit of vagus nerve stimulation in markedly treatment-resistant major depression: a RECOVER trial report. Int J Neuropsychopharmacol. 2026;29(1):pyaf080. doi:10.1093/ijnp/pyaf080. PMID: 41529263. PMCID: PMC12799215. (n=214; median 83.1% and 81.3% of participants with meaningful benefit at 12 months retained it at 18 and 24 months; about 30% of non-benefiters at 12 months gained benefit in year 2.)
9. Conway CR, Sackeim HA, Rush AJ, Bunker M, Gordon C, Kriedt CL, Aaronson ST. The RECOVER trial of vagus nerve stimulation in markedly treatment-resistant depression: critical findings, lessons learned, and future directions. Am J Psychiatry. 2026; published online August 26, 2026. doi:10.1176/appi.ajp.20251339. PMID: 42642808. (Source for the CGI-I 53.8% vs. 39.8%, QIDS-C 39.6% vs. 30.7%, WPAI item 6 43.3% vs. 37.5% and quality-of-life 45.4% vs. 37.1% figures.)
10. McAllister-Williams RH, Alva G, Banov M, Conway CR, Cusin C, De Beer JC, et al. The use of vagus nerve stimulation (VNS) in the management of patients with difficult-to-treat major depressive disorder (MDD): an expert consensus statement. Neuropsychiatr Dis Treat. 2026;22:554007. doi:10.2147/NDT.S554007. PMID: 41884079. (32 experts, modified Delphi, consensus on 75% of 55 statements. Funding statement: "This research was funded by LivaNova (Zaventem, Belgium). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript." Many panelists disclose LivaNova consulting, research or speaking relationships.)
11. Washington University School of Medicine in St. Louis. Vagus nerve stimulation relieves severe depression (news release), December 2024. medicine.washu.edu. (Source for "84 sites," "three quarters unable to work," and the timing of improvement in the final three months.)

Coverage history

1. Centers for Medicare & Medicaid Services. Decision memo for vagus nerve stimulation for treatment of resistant depression (TRD) (CAG-00313R), May 4, 2007. National non-coverage. CMS Medicare Coverage Database.
2. Centers for Medicare & Medicaid Services. Decision memo for vagus nerve stimulation (VNS) for treatment resistant depression (TRD) (CAG-00313R2), February 15, 2019. Coverage with Evidence Development in CMS-approved double-blind, randomized, placebo-controlled trials with ≥1 year follow-up; patient criteria include a major depressive episode of ≥2 years or ≥4 episodes and ≥4 prior failed treatments of adequate dose and duration. NCD 160.18. cms.gov.
3. LivaNova PLC. LivaNova recognizes final decision from the U.S. Centers for Medicare & Medicaid Services regarding national coverage determination for VNS Therapy for treatment-resistant depression (press release), February 15, 2019.
4. LivaNova PLC. LivaNova initiates process with U.S. Centers for Medicare and Medicaid Services for reconsideration of national coverage for VNS Therapy for treatment-resistant depression (press release), June 4, 2025. Business Wire.

Mechanism, anatomy and reviews

1. Berthoud HR, Neuhuber WL. Functional and chemical anatomy of the afferent vagal system. Auton Neurosci. 2000;85(1-3):1–17. doi:10.1016/S1566-0702(00)00215-0. PMID: 11189015. (Glutamate at the primary afferent synapse in the solitary nucleus; central projections to the amygdala and limbic system.)
2. Henry TR. Therapeutic mechanisms of vagus nerve stimulation. Neurology. 2002;59(6 Suppl 4):S3–S14. doi:10.1212/wnl.59.6_suppl_4.s3. PMID: 12270962. (Vagal afferent effects on the reticular activating system, central autonomic network, limbic system and the diffuse noradrenergic projection system.)
3. Ruffoli R, Giorgi FS, Pizzanelli C, Murri L, Paparelli A, Fornai F. The chemical neuroanatomy of vagus nerve stimulation. J Chem Neuroanat. 2011;42(4):288–296. doi:10.1016/j.jchemneu.2010.12.002. PMID: 21167932. (Thalamic projection of central vagal afferents; pivotal role of the serotonin and norepinephrine systems.)
4. Bonaz B, Bazin T, Pellissier S. The vagus nerve at the interface of the microbiota-gut-brain axis. Front Neurosci. 2018;12:49. doi:10.3389/fnins.2018.00049. PMID: 29467611. PMCID: PMC5808284. (80% afferent / 20% efferent composition; NTS projections to the parabrachial nucleus, paraventricular hypothalamus and amygdala.)
5. Carreno FR, Frazer A. Vagal nerve stimulation for treatment-resistant depression. Neurotherapeutics. 2017;14(3):716–727. doi:10.1007/s13311-017-0537-8. PMID: 28585221. PMCID: PMC5509631.
6. Müller HHO, Moeller S, Lücke C, Lam AP, Braun N, Philipsen A. Vagus nerve stimulation (VNS) and other augmentation strategies for therapy-resistant depression (TRD): review of the evidence and clinical advice for use. Front Neurosci. 2018;12:239. doi:10.3389/fnins.2018.00239. PMID: 29692707.
7. Austelle CW, O'Leary GH, Thompson S, Gruber E, Kahn A, Manett AJ, Short B, Badran BW. A comprehensive review of vagus nerve stimulation for depression. Neuromodulation. 2022;25(3):309–315. doi:10.1111/ner.13528. PMID: 35396067. PMCID: PMC8898319. (NTS projections to the dorsal raphe and locus coeruleus; CSF 5-HIAA and neurotrophic findings.)
8. Ardesch JJ, Sikken JR, Veltink PH, van der Aa HE, Hageman G, Buschman HP. Vagus nerve stimulation for epilepsy activates the vocal folds maximally at therapeutic levels. Epilepsy Res. 2010;89(2-3):227–231. doi:10.1016/j.eplepsyres.2010.01.005. PMID: 20129758. (Intra-operative vocal-fold EMG in 8 patients: hoarseness with VNS results from stimulation of the recurrent laryngeal nerve; ipsilateral vocal-fold contraction at therapeutic currents.)
9. Mandalaneni K, Rayi A. Vagus Nerve Stimulator. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; updated August 7, 2023. ncbi.nlm.nih.gov/books/NBK562175. (Approximately 80%/20% afferent/efferent; NTS projections to locus coeruleus, raphe and hypothalamus; left-sided implantation because the right vagus supplies the sinoatrial node.)

Rules cited in the note on testimonials

1. Federal Trade Commission. Guides Concerning the Use of Endorsements and Testimonials in Advertising, 16 CFR Part 255 (revised 2023; 88 FR 48092, effective July 26, 2023), §255.2 Consumer endorsements. law.cornell.edu/cfr/text/16/255.2.
2. U.S. Department of Health and Human Services. HIPAA Privacy Rule, 45 CFR §164.508(a)(3): authorization required for marketing uses and disclosures of protected health information. law.cornell.edu/cfr/text/45/164.508.
3. South Carolina Code of Laws §40-47-110(B) (Medical Practice Act, grounds for misconduct) and S.C. Code of Regulations 81-60 (Principles of Medical Ethics adopted by the State Board of Medical Examiners). scstatehouse.gov.
4. American Medical Association. Code of Medical Ethics Opinion 9.6.1, Advertising and Publicity. code-medical-ethics.ama-assn.org.

Sources for clinical commentary and images used on this page

1. Sheppard Pratt Health System. New study finds promising treatment option for treatment-resistant depression (press release), March 31, 2017. sheppardpratt.org. (Source of the Aaronson quotation.)
2. Manu5 / Scientific Animations. "Vagus nerve stimulation" (illustration), 2018. Wikimedia Commons, CC BY-SA 4.0.
3. Hellerhoff. "Vagal nerve stimulator VNS bei Epilepsie 56W – CR – 001" (radiograph), 2024. Wikimedia Commons, CC BY-SA 4.0.