I'll just quote from Public Citizen's review of the VNS literature...
"The VNS device is implanted beneath the left clavicle in an outpatient procedure that typically costs $25,000 (including the device). A lead runs to the vagus nerve and generates 30-second electrical pulses every five minutes. The device was approved in 1997 for refractory epilepsy, but Cyberonics pursued the additional indication of TRD, perhaps because the company estimated that the market for the latter was 4.4 million people in the U.S. alone. In a May 11, 2005, letter to the FDA urging the agency to not approve the device, we described the inadequacies in the data supporting the efficacy of VNS for TRD. Here, we summarize those data and incorporate the entire letter by reference.
Expecting to demonstrate the efficacy of VNS over the short-term, Cyberonics conducted an appropriately designed randomized, controlled trial (Study D02 Acute Phase) of three months duration in which all TRD patients were implanted with the device, but only half had the device turned on. The other half did not have the device turned on, and so received sham therapy. The study was a failure. On the primary outcome measure (the Hamilton Rating Scale for Depression, or HRSD), VNS showed no efficacy compared to sham therapy; the same was true for nine of ten secondary analyses. This remains the best-designed study of VNS to date.
Following Study D02, the company offered sham therapy patients the chance to have their device turned on (Study D02 Long-term Phase). Predictably, the patients improved over time. This is a near-ubiquitous finding in studies of depression patients due to both the placebo effect and the tendency of patients enrolled into studies of relapsing conditions to improve over time (regression to the mean) because their condition is typically worse at the time of enrollment than at other times. In this and the follow-up to Study D02, patients were unblinded and, unlike in the acute phase of Study D02, were permitted to change concomitant therapies including other antidepressants and even electroshock therapy.
Facing rejection of their application by the FDA, the company opted to add a comparison group for Study D04, which merely compared the Study D02 Long-term Phase patients to this hastily assembled comparison group. Like Study D02 Long-term Phase, there was no blinding, concomitant therapies were permitted to change over time and the comparison patients were recruited from overlapping, but different, sites. Indeed, the authors of a published report on VNS acknowledge that the comparison arm had not originally been intended to serve as the [control]; it was intended to describe health care costs . A modest benefit was reported by the researchers, but only after the only (secondary) outcome that was positive in Study D02 was hand-picked to be the primary outcome for Study D04. Moreover, in an analysis mandated by the FDA, adjusting for overlapping sites and concomitant treatment produced no statistically significant finding on any primary or secondary outcome.
A weaker package of studies is difficult to imagine. Yet, inconceivably, the FDA issued an approvable letter on February 2, 2005, and approved the device on July 15, 2005, overruling an August 12, 2004, non-approvable letter the FDA had sent to the company."
There's a lot more in the full document, which you can peruse here. Suffice to say it's very shady. Of course, all of this didn't stop star academics like Charles Nemeroff from signing off on VNS therapy's alleged wonders! Though having several hundred publications under his belt, that has not stopped (perhaps it's actually encouraged) him to get involved with some chicanery related to VNS and other controversial treatments. This does not make academic psychiatry look good when "respected" big name psychiatrists pimp treatments that clearly provide little to no benefit.