Implanted vagus nerve stimulation held its benefit into a second year in severe depression

Vagus nerve stimulation for depression has an unusual regulatory and evidentiary history: a device approved in the United States more than two decades ago, with reimbursement long withheld, and a large confirmatory trial — RECOVER — conducted years afterwards. The population it targets is narrow and difficult: people whose depression has not responded to multiple adequate treatment attempts.

A report from that trial published on January 13 addresses the question that matters most for a surgically implanted device: does the benefit last [s1]?

What the report covers

RECOVER randomised participants with moderate to severe major depression who had failed at least four antidepressant trials in the current episode to blinded, adjunctive vagus nerve stimulation for 12 months, followed by 12 months of open-label adjunctive stimulation [s1]. The study ran from September 2019 to April 2025 [s1].

This report examines 214 participants from among those randomised to active stimulation, followed through the second year [s1].

The 24-month numbers

Response, defined as a reduction of 50% or more, was reached by 34.3% of participants on the Montgomery-Åsberg Depression Rating Scale, 40.9% on the clinician-rated QIDS, and 41.4% on the self-rated QIDS [s1].

Remission rates were 21.5% on MADRS, 24.3% on QIDS-C, and 23.8% on QIDS-SR [s1].

Two further findings speak to durability specifically. Over 80% of participants who had clinically meaningful benefit at 12 months maintained it at both 18 and 24 months [s1]. And among those without meaningful benefit at 12 months, a median of 37.8% achieved benefit by 24 months [s1].

Function and quality of life moved as well: 65.7% achieved meaningful benefit on the quality-of-life measure and 61.1% on a daily-function item [s1].

The design limits what the second year can prove

The blinded, randomised comparison ran for the first 12 months. The second year was open-label, single-arm, and prospective [s1]. Everyone in this report was receiving active stimulation and knew it.

The authors state the limitation directly: the extension was observational and did not demonstrate that vagus nerve stimulation was necessary, absent a randomised discontinuation design [s1].

That matters more than it might for a drug trial, for two reasons. Depression symptom scales are subject to expectancy effects, and a visible, surgically implanted device is about as strong an expectancy cue as clinical research produces. And depression fluctuates: some proportion of any severely ill cohort improves over two years regardless of what is done, though the proportion is low in populations selected for this degree of treatment resistance.

The late-response finding — that a substantial share of 12-month non-responders benefited by 24 months [s1] — is the one most affected. It is consistent with a slow-acting treatment effect. It is also consistent with regression to the mean and natural fluctuation in a cohort measured repeatedly. Only a randomised discontinuation at 12 months could separate them.

Adverse events were not detailed in the material available for this report.

What the numbers mean in context

A 21.5% remission rate would be unremarkable for a first-line antidepressant [s1]. In a population that has failed four or more antidepressant trials in the current episode [s1], it is a different proposition — this is a group for whom the expected remission rate on further conventional medication trials is low.

The population selection is therefore load-bearing in both directions. It is why a modest-looking number is clinically interesting, and it is why the result cannot be extrapolated to less treatment-resistant depression, where the risk-benefit calculation of an implanted device is entirely different.

What the report does establish

Within its design, it establishes that benefit observed at 12 months largely persisted at 18 and 24 months in this cohort [s1]. Persistence is not the same as attribution, but it is the necessary first question: a treatment whose effect faded during the second year would be a poor fit for a chronic condition and a permanent implant.

It also establishes a timescale for assessing the treatment. A 12-month non-response was not, in this cohort, a reliable indication that the device would not help [s1] — which has implications for how long a trial of the device should run before it is judged.

What to watch

Whether coverage and reimbursement decisions cite this durability evidence; whether any randomised discontinuation study is undertaken; and whether adverse event and explantation data across the full 24 months are reported separately.

This article describes results from a clinical trial of an implanted device and is informational only. It is not medical advice and does not recommend any device, procedure, or treatment.

Sources

• [s1] Durability of the benefit of vagus nerve stimulation in markedly treatment-resistant major depression: a RECOVER trial report, International Journal of Neuropsychopharmacology, 2026;29(1), published online 2026-01-13.