For which stroke patients, if any, do you recommend implantable loop recorder for long-term cardiac monitoring and why?

Fantastic and pertinent question! I won't pretend that I have an answer, but do have a few thoughts that may help frame further discussion:

1. We derive our evidence for the efficacy of anticoagulation in stroke prevention from older trials designed to answer that specific question (SPAF, etc.).
2. In the older trials, AF was historical (paroxysmal or persistent) or diagnosed on ECG, Holter, etc. There are major differences between those patients and the average patient in the CRYSTAL AF study or LOOP study (beyond the scope of this discussion).
3. Modern trials have focused on related points (DOAC non-inferiority, LAA closure in subgroups, etc.), but do not replicate the primary ASA vs anticoagulation ($$$ to do).
4. In the US, there was no interest in funding the type of trial that would answer the relevant questions here, like "What are the risks/benefits of anticoagulation in stroke patients with AF detected on ILR at ~8 months from stroke onset?".
5. Luckily, our friends in Europe (who are establishing themselves as the leaders in many stroke and stroke-adjacent clinical trial spaces) did LOOP. We do see an absolute risk reduction of 1.1% in LOOP for the primary outcome in the ILR group. But that doesn't even come close to the absolute risk reduction we would expect from SPAF, etc.

So, to give my personal answer to your question, in light of the points above, I have and will continue to recommend relatively few ILRs in my practice. Mostly, I recommend them in patients with clear embolic stroke and a negative 30-day MCT.

Perhaps, the ARCADIA trial will give us more concrete answers on ESUS and this will become less complex. For now, though, I think LOOP is not the end of the discussion (lots to come on duration, type of AF, and risk), but gives more evidence for providers like me to continue their current practice.

The term embolic stroke of unknown source, or ESUS, implies an embolic type of stroke, i.e., sudden onset, cortical subcortical distribution, not lacunar, and with no demonstrated atherosclerosis either in the carotids or intracranially. Since we suspect a source of embolism, further evaluation with an echocardiogram, even a TEE, a workup for hypercoagulable states, and prolonged cardiac monitoring are essential. We are looking for paroxysmal atrial fibrillation, which would be an indication of anticoagulation.

The question was framed as ESUS; i.e., embolic stroke. Typically, neurologists adjudicate whether a stroke is "embolic" or not but my experience is that younger neurologists rarely use the term "lacunar". Thus, if the stroke is clearly NOT due to atherosclerosis of the major intracranial vessels, then "embolic" is invoked. If in fact, it's true that 20% of strokes are "lacunar", then some strokes labeled as "embolic", leading to prolonged cardiac monitoring may lead to "unnecessary" anticoagulation rather than continuing to focus on the basic risk factors that lead to "lacunes".

Ultimately, even with 100 more studies, it may be true that with "perfect" control of risk factors for "lacunes" adding anticoagulation (or substituting for the antiplatelet agents rather than dual therapy), it will have some "absolute risk reduction" and the discussion will continue.

I usually consider ILR in patients stratified to high risk for PAF with suspected embolic stroke with so far negative workup for embolic causes including atherosclerotic disease (e.g., SYNC, aortic arch evaluation, athero embolism from mild to moderate intracranial stenosis), structural and culprit functional cardioembolic causes, hyper-coagulable state work up, and negative 30-day external cardiac monitoring.

Depending on the institution and type of external monitoring available, some monitors require patients to charge batteries, which if they are going to acute rehab or SNF and they have upper extremity plegia, may be difficult for them. So in these patients, if other causes of embolism (based on age and risk factors as well) have been ruled out and there is a high suspicion for PAF, ILR can be considered as they would have difficulty managing the external monitor.