Chapter 2: Shared Decision Making (When It’s Really Too Hard to Decide)

Last week, I told a story of a case of an 83 year-old man with severe symptomatic aortic stenosis who has had several episodes of syncope. He is healthy aside from what is termed by his neurologist as “Mild Alzheimer’s Disease”. In the previous post, I explained that there is consensus that he should have his aortic valve replaced and that without such replacement, his mortality would be expected to be extremely high over the next few years.

Until recently, there was only one available way to treat patients with severe symptomatic aortic stenosis. That involved open cardiac surgery where the patient’s heart is stopped, he or she is placed on cardiopulmonary bypass, the aorta is cross-clamped, the old valve is excised, and a new one is sewn in. In 2011, a new device was approved that would allow teams of interventional cardiologists and cardiac surgeons to replace the native aortic valve using a trans-cutaneous (through the skin) approach on a beating heart. This meant that the heart would not have to be stopped and generally provided a means to perform aortic valve replacement (AVR) on very sick patients who were either not candidates for surgical AVR (SAVR) or were deemed very high risk for SAVR. As an aside, the assessment of risk is determined by the use of calculators that input variables shown to predict bad outcomes in surgery. These include things like your kidney or lung function and your general health status. These are estimates and are imperfect.

FDA approval of TAVR devices came on the heels of a series of clinical trials showing that TAVR was at least equivalent to (and likely better) than SAVR in high-risk patients and was clearly superior to medical management in patients not deemed to be a candidate for surgery at all. Since then, the practice has been that patients are referred for consideration of AVR and a team of surgeons and interventional cardiologists will weigh in and decide which procedure is better for that patient. Practically, patients are referred for TAVR. Those patients considered appropriate for TAVR by the interventional cardiologist are then required to see a cardiac surgeon who evaluates surgical risk. If the risk is deemed high, the patients can have TAVR. If the risk is not sufficiently high, the surgeon will reject the request for TAVR and offer SAVR. There is a discussion of the general approach in the latest guidelines from the American Heart Association and the American College of Cardiology.

Our patient is complex in that his surgical risk is not high because aside from his dementia and his age, he is very healthy. He would not have qualified for either of the main trials conducted on TAVR pre-approval. On that basis, he was rejected for TAVR and referred for SAVR. However, there is some evidence that patients with cognitive impairment do poorly after cardiac surgery. The hypothesis is that the process of going on cardiopulmonary bypass and clamping the aorta deprives the brain of oxygen, injures it and might accelerate cognitive decline. There is also some evidence that patients with even mild cognitive impairment are at greater risk for post-operative delirium which can lead to other post-operative complications, prolong hospital stays, and worsen outcomes.

There are scant (none really) robust data in this area, and no prospective studies comparing the outcomes of patients with cognitive impairment getting SAVR vs. TAVR. There are a few observational studies showing that the risk of post-operative delirium is greater in SAVR vs. TAVR. One example is here.

So what to do with this patient?

It is clear that neither he nor his wife cares greatly about TAVR vs. SAVR. In fact, they really had never heard of TAVR before it was mentioned to them by their doctor. It is also clear, that by the books, this patient should have SAVR. TAVR has not been studied in patients with cognitive impairment. It very well could provide better outcomes vis-à-vis cognitive decline, but we just don’t know. We also know that he will likely have a good medical outcome with either procedure. There are some differences in the results, but they are comparable, and in the hands of an experienced interventional cardiologist, TAVR is generally safe. Remember that this is an 83 year-old man, and he is a man with cognitive impairment that will get worse.

So what do we do with the plausible but untested hypothesis that he will do better with TAVR vs. SAVR? The arguments are as follows:

1. TAVR has not been studied in low-risk people and has never been demonstrated to improve outcomes (vs. SAVR) in terms of the rate of cognitive decline. It is also not approved for this indication, so he should have SAVR.
2. While there are no data to support better outcomes, the observational data suggest that he will do better with TAVR. At the very least, the outcomes are probably equivalent, especially given his age. So on the chance that TAVR will reduce his risk of post-operative delirium, or the smaller chance that it will cause less brain injury and thus result in less rapid cognitive decline, we should offer him TAVR.
3. There is a 3^rd option: with his age and cognitive function, he should have neither. In fact, one interesting exercise would be to pretend there is no TAVR available. Would we refer him for SAVR knowing full well that his risk of dying is at least 80% over the next 3-5 years without surgery?

I am still thinking hard about this case. I genuinely do not know the right answer. The patient has no preference. The surgeons support argument 1. The primary care doctor and the interventional cardiologist argue for #2. So the decision, in this case, really does fall on me – assuming I could convince the surgeons to agree. The reflex answer is that it is not approved and not proven, so go with SAVR. Yet, there is likely little harm to the patient to go with TAVR, and there might be real benefit. So when folks suggest that doctors will someday be replaced by computers, I invite them to imagine how a computer might help me make this decision. Until then, I will struggle with finding an answer that does not exist. As always, I welcome any thoughts or comments.