Tuesday, November 11, 2014

What Really Happened At Yorkville Endoscopy?

The release of the Health Department's inquiry into the death of Joan Rivers has certainly shed some light on the event last August at Yorkville Endoscopy clinic. While the following is merely speculative, and we may never know exactly what occurred in that procedure room, we can certainly use some analysis of published data to try to draw a clearer timeline of what led up to her demise.

For months, most anesthesiologists, including myself, thought for sure that Ms. Rivers went into laryngospasm during the case. This is because most of us have seen it happen during a routine endoscopy. Then word came out that there was an ENT surgeon involved who may have been examining her vocal cords without prior consent to look for a source of her hoarse voice. This really put the laryngospasm theory into overdrive. We all slapped our foreheads with our palms and went "Duh! What do you expect will happen when you touch the cords during a sedation case?"

The official report from the Health Department doesn't paint such a clear cut picture. First of all, as it turns out, an anesthesiologist was present during the procedure. I guess Dr. Cohen will use an anesthesiologist for his VIP patients after all. The report also throws the whole laryngospasm theory in doubt. In the document, it helpfully lists Ms. Rivers' vital signs from preop up to when she went into cardiac arrest at five minute intervals. After looking at the data closely, it seems more likely she did not go into respiratory arrest at all before the cardiac arrest.

We know that she had perfectly normal vital signs in preop at 8:44 AM: BP 118/80, HR 62, O2 sat 100%. The patient went into the operating room at 9:00 AM. Then we get to see the vitals as they occurred during the procedure. 

9:12 AM: BP 117/60, HR 71, O2 sat 92%
9:16 AM: BP 92/54, HR 56, O2 sat 94%
9:21 AM: BP 89/44, HR 54, O2 sat 97%
9:26 AM: BP 84/40, HR 47, O2 sat 92%

The report does not indicate what time the procedure started. However it does say that Dr. Korovin started first with a nasolaryngoscopy before handing the patient over to Dr. Cohen for the EGD. The endoscope was removed from the patient at 9:28 at which time Dr. Korovin attempted to perform a second nasolaryngoscopy. At about 9:30 was when Code Blue was called.

So let's assume that the first VS was the baseline before the patient entered the room. Judging by the drop in blood pressure and oxygen saturation at 9:16, she had probably already started receiving some sedation with propofol. The electronic anesthesia record noted that propofol was given at the following times: 100 mg @ 9:21.45, 100 mg @ 9:21.46, 50 mg @ 9:21.48, and 50 mg @ 9:21.50. This is a total of 300 mg of propofol pushed in five seconds.

The anesthesiologist claims that those quantities are mistakes. The patient had only received 120 mg of propofol for the whole case, which she wrote in an addendum to the record later that day. She told the investigators that she accidentally double clicked the computer for the 100 mg boluses and really meant to click on 20 mg bolus instead of double clicking 50 mg boluses.

One thing about electronic medical records--just because you click on an event doesn't mean it actually happened at that time. You may have been too busy with your responsibilities before realizing you forgot to enter it on the computer. You then finally document it even though it happened a few minutes prior. Even though the anesthesiologist may appear to have first bolused the propofol at 9:21.45, it is conceivable that this was a slow bolus given over several minutes but not charted until that time. If we presume that the hypotension seen at 9:16 was already the beginning of the sedation, then the drug was not entered into the computer for five minutes. At that point the anesthesiologist documented 100 mg at 9:21 even though it may have been given over several prior minutes.

Why was her O2 sat only 92% at 9:12 even though the rest of her vitals were normal? This maybe an instance of bad oximetry data. There are many reasons for a sat monitor to pick up an erroneous reading. Perhaps her fingers were cold. Maybe she was shivering. Or maybe somebody was moving her arm. The computer will record whatever number it sees at a particular time without regard to its accuracy. It is up to the user to write a note explaining the false data in the record.

As the oxygen saturation started to drop during the procedure, the anesthesiologist says she turned up the oxygen flow and applied jaw thrust to open up the airways. The anesthesiologist's statement that she kept the O2 sat greater than 90% is true based on the recorded data. But unfortunately that was not Ms. Rivers' problem.

I've seen many laryngospasms during an endoscopy. On nearly all of them, the O2 saturation drops first because the patient can't pass any air. Then the body starts fighting like mad to open up the airways. This results in a severe tachycardia that doesn't slow down until the cords open up or the O2 sat drops so low that the heart becomes ischemic, leading to bradycardia, a most ominous sign.

But Ms. Rivers' problem doesn't seem to be hypoxia. Her sat never goes below 90. Instead we see hypotension and bradycardia first. This is most likely due to a depressed cardiac output. If the CO drops, the O2 saturation eventually falls too as the heart can't pump enough blood for adequate oxygen exchange in the lungs.

What could have caused her heart to fail? Right at the top of the list is propofol. That is why propofol is such a dangerous drug. If not used judiciously, it can cause severe cardiac depression and hypotension, sometimes fatally. In an elderly patient, the impaired heart function can be quite profound. While I want to give the anesthesiologist the benefit of the doubt and agree that she didn't give 300 mg of propofol to an 81 year old in a span of five seconds, which would no doubt lead to a catastrophic bottoming out of her blood pressure, even 100 mg of propofol if pushed too quickly can cause a severe decreased cardiac output.

Another possibility is a strong vasovagal reaction. This is especially true with something as stimulating as a laryngoscopy. If the patient is not sedated adequately, the stimulation from the procedure can cause the BP and HR to drop in some people. This usually goes away once the stimulant is removed by withdrawing the scope. Respiration again is not impaired but the sat may fall because of the decreased blood flow to the lungs and tissue.

Could she have suffered a pulmonary embolus? A PE can cause acute hypotension and bradycardia as the right heart workload suddenly increases. But a PE also usually causes severe hypoxemia, especially for one that leads to death like a saddle embolus. The medical examiner's report also failed to mention a PE as the cause of her death.

An acute myocardial infarction? Maybe. But it would be awfully coincidental for her to suffer an MI right at the time she was getting a procedure done, though it wouldn't be outside the realm of possibility.

If laryngospasm is no longer the likely cause of Ms. Rivers' death, it also explains why she had a hypoxic brain injury. It is easy enough to treat laryngospasm, especially with an anesthesiologist and an ENT surgeon at the bedside already. If it can't be broken with positive pressure ventilation, then give a muscle relaxant to open up the cords and intubate the patient. None of that should take so long as to cause brain ischemia. However if she had heart failure, the low cardiac output would starve the brain of oxygen leading to its death. She sent into cardiac arrest at 9:30 and was not fully resuscitated until 10:00. That is a full thirty minutes of suboptimal blood flow and oxygen exchange to the delicate brain tissue. Thus the brain dies.

Therefore the most likely reason that Ms. Rivers died that day was her heart failed. Either she went into such a severe vasovagal reaction during the laryngoscopy that they couldn't revive her heart or the anesthesiologist felt she needed to give so much sedation to perform the laryngoscopy (Anesthesia! The patient is moving!) that it irreversibly impaired her heart function.

That's just my two cents.


  1. That is one very real possibility. Except that I don't think there was any direct laryngoscopy here. It seems that the laryngoscopy was with a nasal fiberoptic. In which case not much sedation is needed.

    There is some untreated hypotension there, but the way things evolved between 9:12 and 9:26, I doubt that the initial problem was propofol overdose. By 9:26, the anesthesiologist would have known the patient's propofol tolerance, after already having completed the first nasolaryngoscopy and an endoscopy successfully.

    Yes, there is some untreated hypotension, but again chances are very small that the anesthesiologist would have given a ton of propofol when the case was almost done. Either there was a cardiac arrhythmia (there are mentions of both VTach and asystole in the report), or there was indeed an unexpected laryngospasm. Now what's the first thing they would do during a laryngospasm, in the absence of sux? Push propofol. Still laryngospam. Push more propofol. That might have been what lead to the asystole. And if asystole, why only 1 mg of epi given in 30 minutes of CPR?

    There is something missing here that we will all find out only in the court.

  2. Four sets of VS over 30 minutes tells us very little about what happened throughout the procedure. You are making a lot of unsupported assumptions based on scant empirical information. This patient had no significant cardiac history. It's far more likely that she experienced one or more episodes of significant hypoxia unrecorded by the anesthesiologist. An End-Tidal CO2 monitor should be a requirement for Propofol sedation procedures. Was one utilized at Yorkville? Was Ms. Rivers appropriately resuscitated? More questions than answers here.