Workhorse HiRes 90K implant circuit recalled as it can “Fail Ugly”
–>Did the surprising lack of burn-in cause this recall?
January 7, 2011; Version 1.2
Please be sure to read the follow-up article Failing Ugly II: More on the Advanced Bionics Recall Although we’re not investigating earlier failures per se, we are interested in how AB handled them.
The Hearing Blog is opening an investigation into the two confirmed, and nine (or more) possible catastrophic failures of the Advanced Bionics HiRes 90K cochlear implant (CI) electronics module in the last several weeks, resulting in extreme overstimulation, causing extreme discomfort to the point of pain and headaches, with the possibility of permanent damage to the cochlea and auditory nerve. In general, when there is a problem with a CI, it will “fail gracefully” by shutting down. However, this particular recall involves the implant “failing hot,” which we call Failing Ugly.
We chose Failing Ugly on purpose: When a CI fails, it should fail gracefully, quietly shutting down until restarted or serviced as needed. Think of driving down the highway and getting a flat tire, where you let off the gas, steer to the shoulder, and coast to a stop: No Big Deal, as Stuff Happens. Now picture this same car with a throttle stuck wide open: This is Failing Ugly, and the end result can be messy, sometimes ending in a wreck. The same is what usually happens when a CI processor or implant fails: It is supposed to fail gracefully, but as we will show in this article this is .NOT. what is happening.
There have been two separate documents circulating about the Advanced Bionics HiRes 90k implant circuit failures: The first was the announcement Tuesday morning (11/23/2010) from Sonova’s Invester Relations department, which was picked up by the Wall Street Journal later that morning in an article titled Sonova Shares Tumble on Hearing Implant Recall. Late Tuesday evening, the Sonova announcement was placed on the Advanced Bionics website. There is now an abridged recall notice here on the FDA website. We also note here that any and all announcements issued by AB over this recall must be approved in advance by the US Food & Drug Administration (FDA).
Beyond that, there was a second document: The official letter sent to AB CI recipients who have devices from the affected batche(s)/lot(s).
UPDATE: The 5 page third PDF document, sent to UK and EU CI centers is in the second article in this series on the AB recall.
The Hearing Blog received the following copy of the Recall letter at 9:30PM EST on Tuesday night:
This is the email sent out by Advanced Bionics on Tuesday 11/23/2010 at 9:30PM EST (Wednesday 2:30AM UK). This is more serious than in the press release from Sonova’s investor relations department from earlier in the day.
—————**** Recall Notification****—————
Advanced Bionics HiRes 90K Cochlear Implant
Dear Cochlear Implant Recipient or Parent,
Our mission at Advanced Bionics is to improve the lives of the hearing impaired, and the safety and well-being of our recipients is our first priority. Because we are committed to ensuring that our products are as safe as possible, we are voluntarily informing you that we have become aware of an issue with the HiRes 90K cochlear implant. The issue can result in pain, overly loud sounds, and/or sudden shock sensation in the implanted ear while the implant is receiving power.
Thus far, our investigation shows that of the more than 28,000 implanted HiRes 90K devices, only two explanted devices have been confirmed to have this issue. There are 9 other patients with similar symptoms which our investigations have not yet ruled out as related to this issue. For the two confirmed cases, there were no symptoms upon initial activation of device. However, the patients experienced symptoms after 8-10 days of use. Both recipients were re-implanted with HiRes 90K devices and their clinicians report that they are progressing well.
We are conducting an extensive investigation of this issue using an independent scientific research organization. Their current analysis suggests that, if present, the issue will first occur within 90 days of device use. However, the onset of symptoms may be delayed after initial activation, and they may continue to occur intermittently. In the unlikely event that this symptom occurs, continued device use may lead to damage to the inner ear and/or the auditory nerve. If you or your child experience pain related to implant use, remove the external equipment immediately and contact your cochlear implant clinician to schedule an appointment. If the evaluation of your device identifies this problem, device replacement is advised.
It is important to note that it is not unusual for cochlear implant recipients to experience overly loud sounds. Most of these cases can be resolved with standard troubleshooting, such as the replacement of external equipment. The vast majority of these cases are not signs of the issue described above and do not require explant of the device. We will notify you again as more information from our investigation becomes available.
We sincerely regret any concern this notification may cause. Advanced Bionics assures you that we will correct this issue and continue to improve our product reliability. If you have any questions regarding this letter, please contact an Advanced Bionics representative at 877-577-4628 (telephone) Monday – Friday 5:00 a.m. – 7:00 p.m. PST or http://www.advancedbionics.com (live chat) Monday – Friday 5:00 a.m. – 7:00 p.m. PST.
In order to assure the effectiveness of this communication, please complete the enclosed acknowledgment form and return it to us at your earliest convenience by using one of the following options:
Mail: Self-Addressed stamped envelope
Unfortunately, there is a 30 day grace period for the FDA’s MAUDE Adverse Event Report database. Speculation on the Internet is rampant, with this 30 day delay until submissions are made public only stoking fears, of which we will not participate. To date, over 200,000 hearing impaired people have received cochlear implants, from as young as 5 months all the way into their 90’s. Cochlear implant technology is safe and effective for all ages with very few exceptions: This underscores why our independent investigation is important, to ensure the excellent reputation CI’s have achieved over the last two decades is maintained. It is our belief that sunshine is the best disinfectant, so once the cause of these failures is pinpointed and rectified, the Fear, Uncertainty and Doubt (FUD) can be refuted, putting both affected recipients and candidates’ minds at ease.
We need to remind our readers that The Hearing Blog is a privately hosted, independent publication that accepts no advertising, no samples, and no manfacturer or distributor funding. We are modeled on the same decades-long policy Consumer Reports‘ uses, to assure readers the content is unbiased.
We also need to remind our readers that hearing is important: Just look at the 200 thousand people who have undergone general surgery to get it restored, including many of our friends like Kristin Fleig, who was dying to hear again… And (almost) did.
This article is tentatively divided into three subsequent sections:
Engineering Analysis: The “who, when, where, why and how.”
First, what is happening is that there appears to be a component-level failure in the implant circuit, one which overrides the many fail-safe mechanisms built in since this circuit was introduced in 2003. This failure causes an overvoltage — an overcurrent — to be applied to one or more electrode contacts, powerful enough to cause a loud, painful jolt. In addition, there is the distinct possibility this would induce current biasing if (and we stress the word IF) the failure includes a shorted output tantalum electrolytic capacitor, which would cause a(n electro)plating effect, ripping metallic crystals off of the electrode button contacts, destroying it (them) over time… And causing consequent damage in the tight spaces of the cochlea, as such electro-chemical reactions can produce bio-toxic byproducts as well as introduce potentially damaging changes in the local pH. Is this a possibility? Let’s look at the carefully worded recall notice, which was approved by the FDA:
In the unlikely event that this symptom occurs, continued device use may lead to damage to the inner ear and/or the auditory nerve.
This statement would not be present in the recall notice if the failure would not cause damage from bio-toxic byproducts &/or local pH changes. This “failing hot” is also why we use the term “Failing Ugly.”
Let’s use an easy-to-understand analogy of one possible failure mode: Picture a Class B (push-pull) audio amplifier with +/- X volt power supply rails coupled to a loudspeaker voice coil through a tantalum electrolytic capacitor. Now, short out one of the two transistors, connecting that supply bus directly to the capacitor: The voice coil is still protected, because the capacitor blocks the DC bias — It acts as a fail-safe, limiting damage. This is also why we qualified the failure above with the word IF. Now, IF we have a second failure where the capacitor shorts out — which is the failure mode of tantalum electrolytic capacitors — now the power supply bus is directly coupled to the loudspeaker voice coil, which will cause the speaker driver to travel to one or the other limit, caused by the DC bias across the terminals. In the case of the implant (which uses lower voltages than an audio amplifier), this will cause the electroplating effect, and the consequent electro-chemical reactions described above. So important is the prevention of plating, the FDA requires all CI stimulations to pass a rigid mathmatical and electrical test where the integral (summation) of the total charge over any 10 minute span must be zero.
One failure mode theory can be pretty much set aside: Dendritic crystalline growth on the circuit chip, which could indeed happen in a matter of days. For this to occur, it would require moisture to be present, which is due to a failure of the hermetic seals in the implant case. One longtime implant engineer told me that dendrites will grow with just a few monolayers of water present and a small dc voltage. The reason we are setting this aside is that if indeed it were true, there would have been dozens, probably hundreds of similar overstimulation failures in the last 6-7 years from the infamous Astro-Seal (“Vendor B”) episode.
Another tantalising possibility is a software glitch in the new SoundWave 2.0 software. The stim software that is downloaded from SoundWave to the processor is supposed to be identical between SW V1.6x and V2.0; but this is one area where bit-for-bit comparisons can be easily verified and ruled out.[This is not to be confused with the boot-up of the implant electronics package itself: What sets Advanced Bionics' method apart from the others is that when the implant circuit powers up, large data tables are downloaded from the processor into the implant memory itself as a BIN file, taking 1.5 to 2 seconds before stimulation starts. These data tables contain various combinations of complex electrode firing patterns, which allows for the 88,000 forward data updates per second to take place. There are many software fail-safes built in that cause a graceful shutdown, so this too can be quickly ruled out.]
So, we are back to a component-level failure that causes the output to Fail Ugly. If the design engineers did their job properly, there will only be a minimum of any singular component failure that will cause this nasty overall failure; and more importantly from a reliability standpoint, adding the complexity of additional monitoring circuitry will greatly decrease the overall MTBF (mean time between failures, the reciprocal of the failure rate per million hours).
The open questions Advanced Bionics needs to answer are:
- Which component failed — Semiconductor, or tantalum capacitor?
- Was it an open or short failure?
- Have these failed circuits been attached to 1j or Helix electrode arrays, or to both?
- Where in the circuit is this failed component located?
- What is the proximate cause of the component failure?
- What are the destructive and non-destructive test results from this batch of components?
- What have been any recent component changes?
- Have any components been purchased from a new or different vendor?
- What have been any recent QA/QC changes?
- Since these are “infant mortality” failures occurring at 8 & 10 days (and perhaps the one at 11 days along with 8 others), why weren’t these detected during incoming inspection or during burn-in?
- What is/are the manufacturing date(s) &/or serial numbers of the implant batch(es) containing the offending component batch?
- Is the batch of bad components confined to one manufacturing batch? To one reel?
- Can this failure be detected with the SoundWave 1.6x or 2.0 tools as released to the CI centers? To the tools issued to the clinicians?
- Can Neural Response Imaging (NRI) be used as issued to detect and/or confirm this failure? If not, is there a software update to NRI in development?
I just received my AB implant within the last year: Don’t panic, as your implant is probably .NOT. in the batch subject to this particular Ugly Failure. Unless you show symptoms of the painfully loud overstimulation, you’re OK for the moment.
Just to clarify, the implant electronics from the three manufacturers licensed in the United States — AB, MedEl, and Cochlear — all have similar failure rates where they fail gracefully. This recall is totally different, as it involves Failing Ugly.
I’m considering an AB implant: Do I fish, or cut bait?
Once you decide to get a cochlear implant, there are many factors that will go into your choice, as well as the influence (both good and bad) of your CI audiologist and surgeon. If you have not yet read Tina Lannin’s excellent blog articles on CI’s in general, and more specifically on choosing a CI brand, please do so, as it contains very good information.
1. Ask your surgeon if s/he implants AB’s 1j semi-curved (lateral) or Helix (perimodional) electrode [But note this one surgeon's response when he was asked]:
- If he implants the 1j electrode, then a very good substitute is the MedEl Sonata Ti100 implant with the 31.6mm FlexSoft electrode and the Opus 2 speech processor (either with or without the optional remote — makes no difference). It is our belief that the speed and current steering of the MedEl implant when combined with their FineHearing stimulation software is more than adequate to the task; and since you would be getting a lateral placement with either the 1j or FlexSoft, the longer MedEl electrode combined with the tonotopic matching provided in Maestro 3.x MAP (programming) software (which is .NOT. in AB’s SoundWave 2.0) will provide a better overall experience, especially if a hearing aid is used in the other ear;
- If he implants the Helix electrode, it’s a tougher call: Even though Soundwave 2.0 does not provide for tonotopic matching (which we consider a major limitation of the AB system), the perimodional (tight spiral) placement does indeed provide a cleaner (purer) stimulation. Unlike brokerage Helvea analyst Daniel Jelovcan’s prediction of the AB implant being off the market for six months, we here at the Hearing Blog believe it will be closer to six weeks .IF. — And only If — the failed component has been identified and has been scrubbed from the implant production stream;
2. If you are the parents of a pre-lingually deafened infant scheduled to be implanted before the age of 18 months, we recommend .NOT. delaying the scheduled implant surgery date, as every day that goes by is that much less auditory and speech development will occur in your child’s brain as it is developing. For this group, The Hearing Blog recommends either MedEl’s Maestro system with the Sonata Ti100 implant connected to the 31.6mm FlexSoft electrode using the Opus 2 speech processor and the FineHearing stimulation software; or Cochlear’s Nucleus 5 system with the 25.4mm Contour (perimodiolar) array and the two-way Remote Assistant for its diagnostics.
We here at The Hearing Blog take no comfort in publishing this article, because cochlear implants are safe, effect… And life-changing to over 200,000 people so far.~