Arguments against ICU admission for comatose stroke patients

This issue has never come up in the CICM exam, but it certainly appears often in clinical reality.

Paul Marik co-authored a paper in 2014 which focused on this exact question. It, and the references it provides, have been the major sources of information for the summary presented below.

In brief:

"Stroke ICU"

In 1973, someone in Memphis tried to model the success of the "coronary care unit" model and put together a four-bed ICU specialised in the management of acute stroke, comparing their mortality to that of the adjacent neurology ward. But the mortality benefit never materialised. The authors lamented: "The conclusion seems inescapable that Stroke ICUs are not productive in reducing mortality in the acute stroke patient, at least in a general hospital setting."

Subsequent data overturned this conclusion. However, we must be conscious of what is being meant by "stroke ICU". In Australia, these facilities are better known as "stroke units". They are staffed by neurologists rather than intensivists, and offer only non-invasive monitoring. Generally, mechanical ventilation of any sort is not available. Rather than "intensive care", these units offer "organised inpatient care", representing an escalation in focused expertise and nurse-patient ratio from that of the regular wastebasket neurology ward.

Stroke units

A 1997 meta-analysis of then-available data had revealed that stroke patients perform better in stroke units across a number of examined domains.

Specifically:

• Stroke units offer a reduction in all-cause mortality.
• The most profoundly affected cause of mortality was mortality due to complications of immobility; stroke unit nurses seemed to turn their patients more frequently, and paid greater attention to pressure area care.
• More patients returned home with functional independence (though this was not statistically significant, it is still an encouraging finding). Weirdly, this slight improvement was sustained across a wide range of stroke severity.

In summary: admission to some sort of dedicated stroke-savvy high dependency environment is better for stroke patients than being admitted to some random stroke-ignorant ward environment, where they end up decomposing quietly in a state of neglect.

That much being established, let us focus on what (if anything) the "Proper ICU" can offer to the severe stroke patient; specifically, mechanical ventilation, invasive monitoring and and intracranial pressure management.

Mechanical ventilation in stroke patients

One typically ends up intubating these people for several mundane reasons:

• They have poor cough and gag, and end up with aspiration pneumonia.
• The immediate post-stroke period is characterised by a decreased level of consciousness.
• There are stroke-associated seizures.
• The patient is not extubated following some neurosurgical procedure.
• The CT brain requires the confused patient to remain still.

However, it is better to avoid intubation in these patients.

In-hospital mortality for intubated stroke patients

Probably the most recent entry into this arena was a study by Lahiri et al from the December 2014 issue of Neurocritical Care. This retrospective audit analuysed data from 798,255 admission for stroke, of whom 12.5% required mechanical ventilation. The authors made some attempt to discriminate between ischaemic stroke, intracranial haemorrhage, or subarachnoid haemorrhage. The mortality for mechanically ventilated patients with these conditions were as follows:

Long term mortality and quality of life in intubated stroke patients

An analysis of mortality in reasonably young patients (over 65) has revealed that being intubated in ICU after a stroke is a terrible prognostic indicator. Estimates of how terrible are fairly variable. For instance, Burtin et al in 1994 found that about 72% of their patients died during their ICU admission, and 91.8% over the first year - with only 13 survivors out of 170; whereas Thorsten et al in 1997 had a 66.9% 1-year mortality rate for patients treated in their specialised "Stroke ICU", with survivors having "only slight or no long-term disability". A Stroke article from 2000 summarised the reported 1-year mortality rate as 49% to 93%, with most survivors severely disabled, questioning the cost-effectiveness of ICU admission for comatose stroke patients.

In any case, about 50-60% of them die in hospital. Of the remaining 40%, in total only about 10% end up having a decent quality of life afterwards, with minimal deficits. For whatever reason, these people seem to be at an increased risk of ventilator-associated pneumonia- I suppose its that absence of cough. In one old study, two thirds of mechanically ventilated stroke patients died during their hospitalization, and most survivors were severely disabled. In a more recent study, mechanical ventilation was associated with a five-fold increase in the mortality hazard.

 

"Reversible pathology"

With all these negative things said about ventilating comatose stroke patients, it is important to make a distinction. Some stroke patients will benefit from mechanical ventilation: when it is being used to treat a reversible pulmonary pathology, such as aspiration pneumonia, rather than an "irreversible" neurological pathology (such as profound unconsciousness).

The aspiration pneumonia will resolve eventually. One could argue that it is a part of the natural history of stroke; "this is what they die from", etc. One would be reasonably correct. Pneumonia is very common killer after stroke, particularly the kind of stroke that actually presents with aspiration. Additionally, dysphagia following stroke is associated with a poorer functional outcome and increased mortality. However, this risk is not a life-long problem in all cases. Some patients - particularly those with subcortical stroke - tend to only have a transient risk of aspiration; in contrast, those with infarcts of the frontal operculum and insular cortex tend to have ongoing problems with swallowing. A well-aged 1997 article from Dysphagia traced the natural history of stroke-associated aspiration and found that after 7 days, 73% of the patients had recovered the ability to coordinate their swallowing, and that at 1 month only 15% were still aspirating.

In summary, it is fair to offer mechanical ventilation to stroke patients who are not comatose, and who have suffered aspiration in the course of their acute stroke presentation. Many of them will go on to have a satisfactory recovery.

Invasive haemodynamic monitoring

It is known that wild blood pressure fluctuations in stroke seem to be associated with an increased risk of death. A study from the same year (2006) had confirmed that stroke patients have a deranged blood pressure pattern, and a failure of normal circadian variation. A more careful control of post-stroke blood pressure may actualy be helpful, at least among patients with intracranial haemorrhage. One might make a connection on the basis of this - that careful invasive monitoring of arterial blood pressure might translate into more responsive antihypertensive management, and thus every stroke patient should have an arterial line.

However, one might counter with several points.

• Firstly, all the studies mentioned above are studies which detected this blood pressure variability using old-fashioned oscillometric manometers. Ergo, it is possible to detect such variability without puncturing any arteries.
• Secondly, the evidence for strict blood pressure control is not very strong. The recent CATIS trial (2014) did not find any difference in the risk of death or major disability, whereas the recent (2015) INTERACT2 trial found one merely needs to make a large SBP reduction within the first hour.
• Thirdly, even if you did assume that blood pressure management was of critical importance, there is no need to use agents which require beat-by-beat blood pressure titration. For instance, the ENOS trial (2015) discovered that GTN lowered blood pressure effectively, but still had no effect on functional outcome. Thus, one can use gentle slow-acting antihypertensives.

Thus, in general, it is impossible to argue that stroke patients require invasive haemodynamic monitoring. Non-invasive continuous monitoring options are also becoming available - for instance, continuous fingertip monitoring using the Nexfin device has recently enjoyed a small-scale trial. Apparently, its "feasible". Good luck, Edwards Life Sciences.

Intracranial Pressure Management

Another argument which can be used to push moribund stroke patients into the path of non-beneficial intensive care is the apparent need for intracranial pressure monitoring, and neurosurgical management in general. However, there are only a few indications for neurosurgical intervention in stroke:

They are as follows:

• Decompressive craniectomy for malignant MCA infarction
• Decompressive craniectomy for posterior fossa stroke
• EVD to manage hydrocephalus due to posterior fossa stroke

Yes, the infarcted brain is going to swell. Yes, this has some pressure effects. No, this does not mean that you need to measure their intracranial pressure. The 2014 AHA/ASA guidelines make a statement to that effect, qualifying it with the caveat that patients who continue to deteriorate neurologically should progress to decompressive craniectomy. So, according to them, we should clinically assess the patients and react to clinical findings suggestive of increased intracranial pressure. This closely resembles the evolving debate regarding the utility of ICP management intraumatic brain injury; even in that "classical" stronghold of neurosurgery, it is unclear whether the practice of invasive ICP monitoring influences mortality.

All this said, the strength of this recommendation is not great. The AHA/ASA attribute to it a Class III, Level C; which suggests that the support for this guideline is "general agreement" and "consensus opinion of experts", rather than actual evidence. So what evidence is there?

Well. Digging around, one can come up with a couple of papers to support the non-utility of ICP monitoring early in the progress of acute stroke.

• Bardutzky and Schwab (2007): even in henispheric infarction, oedema is slow to develop, and the efficacy of ICP management therapies is "disappointing". Again, not a randomised controlled trial, but rather a review article offering the opinion of experts.
• Poca et al (2010): a prospective cohort of 25 malignant MCA infarct patients. Weirdly, even in patients with a midline shift in excess of 7mm, the ICP generally did not exceed 20mmHg. These patients were comatose and with unequal pupils. What was the point of monitoring their ICP?

Thus, there may be no role for ICP monitoring in stroke. There may still be some role for a draining ventriculostomy. Cerebellar infarction may cause so much swelling in the posterior fossa that CSF drainage becomes affected and hydrocephalus develops. Raco et al (2003) reported on their experience with this problem; their trigger for EVD insertion or decompression were neurological deterioration and radiological evidence of ventricular enlargement. The outcomes were unsurprisingly good; of the 17 patients 10 had only mild deficits.