Grading of subarachnoid haemorrhage severity

At one stage in the history of neurology and neurosurgery, subarachnoid haemorrhage scoring systems had proliferated wildly and at one stage there were apparently about 40 of them. Major criticisms of these older systems were subjective measures, retrospective methodology and poor inter-rater reliability. More recent systems (if you can call them that, from the 1980s) have been generally better received, but still there are none to describe as the "golden standard". All  feature some flaw or another. 

Historically, the college have asked about SAH numerous times. However, the theme has always been "how to find vasospasm" and "how to manage vasospasm" and "what is vasospasm", etc. Question 3 from the first paper of 2018 is the first question to focus on grading systems in any great detail. Specifically, attention was drawn to the WFNS and the Fisher scores; the candidates were asked to define them and describe their limitations.

If one had to limit their reading to one article only, one could do worse than the 2005 piece by Rozen and Macdonald from Neurocritical care. The authors soberly conclude that "because of the limitations of the current grading scale data, it is unreasonable to strongly advocate universal adoption of any of the available SAH grading scales".  

Older severity scoring systems:

Here are a few. There were apparently about 40 at one stage.

• Bramwell, 1933: "apoplectic vs. paralytic"
• Botterell (1956)
• Nishioka (1966)
• Cooperative Aneurysm Study system
• Hunt and Hess (1968) - a variant on Botterell's system; a measure of perioperative mortality risk for neurosurgeons, based on  (a) the intensity of meningeal inflammatory reaction, (b) the severity of neurological deficit, (c) the level of arousal, and (d) the presence of associated disease.

The Hunt and Hess scale of SAH severity

Grade	Criteria
Grade 1	Asymptomatic or minimal headache and slight nuchal rigidity
Grade 2	Moderate to severe headache, cranial nerve palsies, nuchal rigidity but no other neurological deficits
Grade 3	Drowsy, confused or mild focal deficit
Grade 4	Stupour, moderate-to-severe hemiparesis, early decerebrate rigidity and "vegetative disturbance" whatever that means
Grade 5	Deep coma, decerebrate rigidity, "moribund appearance".

This staging system came about 5 years before the Glasgow scale. It is unclear what neurosurgeons in the 60s understood by "vegetative disturbance" and "moribund appearance". Hunt and Hess themselves confessed in their introduction: “It is recognized that such classifications are arbitrary and that the margins between categories may be ill defined".

Features of this scale:

• Ambiguous definitions
• Grades 1, 2 and 3 have better outcomes than 4 and 5.
• Cohens kappa coefficient is 0.42, much greater than would be expected by chance
• Can be compressed into just 3 grades, which correlate with outcome:
  • Alert group (Grades 1 and 2) = 71% had good neurological outcome
  • Drowsy group (Grade 3) = 14% had good neurological outcome
  • Comatose group (Grade 4, 5) = none had good neurological outcome.

WFNS clinical scale

The World Federation of Neurological Surgeons grades SAH according to motor score and total GCS. Their grading scale gives us something of an idea as to what to expect in terms of survival.

the WFNS scale:

• Grade 1: GCS 15, no motor deficit.
• Grade 2: GCS 13-14 without deficit
• Grade 3: GCS 13-14 with focal neurological deficit
• Grade 4: GCS 7-12, with or without deficit.
• Grade 5: GCS <7 , with or without deficit.

Notice how in the most severe categories, it does not matter whether your motor deficit is present or not. Motor deficit really just distinguishes Grade 2 from Grade 3. This system relies on accurate GCS scoring in the absence of sedation. 

Advantages of the WFNS system

Keeping in mind that this scoring system dates back to the late 1980s, some major advantages need to be acknowledged.

• It is simple to calculate and apply
• It does not require imaging
• It relies on the GCS, which is a well-accepted and validated scoring system, and which predicts mortality in other intracranial catastrophes.
• By using the GCS, it depends on objective measures which have high inter-rater reliability.
• It acknowledges the need to include focal neurological deficit, which is a major predictor of morbidity (therefore it likely has an advantage over the GCS alone). 
• It is a well-validated prognostic tool (Rosen et al, 2005) and the grades have clear outcome differences, i.e. there is a stepwise increase in the likelihood of a poor outcome with increasing grade

Limitations of the WFNS system

• It relies on the accurate application of the GCS
• It does not incorporate imaging data
• It is unclear whether adding the additional focal neurodeficit dimension has any added benefit to the prognostic power of the scoring system (It may well be that admission GCS is the single best predictor of neurological outcome)
• the expert committee behind the WFNS did not explain the reasoning behind the specific GCS breakpoints used for their definition
• Grade 4, which represents a range of GCS scores between 7 and 12, includes a group of patients who may have widely different outcomes.
• There is a significant step in the likelihood of poor outcome between Grade 2 and Grade 3 (0.61 vs 1.78)
• There are very few Grade 3 patients in all studies of WFNS (usually, ~ 3%)

Fisher radiological scale

Another method of scoring SAH severity (with implications for vasospasm) is the Fisher scale, which was introduced in 1980. 

• Grade 1 - no haemorrhage
• Grade 2 - SAH less than 1mm thick, diffuse
• Grade 3 - SAH more than 1mm thick, with localised clots
• Grade 4 - intraventricular or parenchymal extension, with clots

Essentially, this scale relates the amount of blood to the risk of vasospasm. Grades 3 and 4 almost always develop it, and the other grades seem to be spared.

Advantages of the Fisher score

• Convenient: a CT brain is an investigation which the SAH patient is guaranteed to have
• Well-validated
• Unlike strictly clinically based systems, it can predict vasospasm
• Inter-rater reliability is high: Ogilvy et al (1998) reported a kappa value of 0.90 (i.e. close to perfect agreement).

Limitations of the Fisher score

• It was developed when the imaging resolution was approximately 1\20th of what is currently available 
• It may be difficult to apply for staff who are unfamiliar with CT imaging
• It was validated in a small series of patients (by Kistler et al, 1981)
• Grade 1 and 2 are very uncommon
• Clot density and clot clearance rate are important factors that infuence the development of vasospasm, but which are not included in the scale. The same score (4) is given to the patient with a tiny speck of blood in the ventricle, as well as to the patient whose ventricles are full of thick clot. 
• Unlike clinical scoring systems, it does not correlate very well with clinical outcome- only with vasospasm

Modified Fisher scale

Answering the concern that the old Fisher scale did not adjust its predictions on the basis of thick cisternal or ventricular blood, the scale was revised to include these findings (Frontera et al, 2006):

Modified Fisher scale for SAH

Grade	Subarachnoid blood	Intraventricular haemorrhage	Vasospasm incidence
0	None	None	0%
1	Focal or diffuse; thin	None	24%
2	Focal thin; or diffuse	Present	33%
3	Thick	None	33%
4	Thick	Present	40%

As you can see, the main difference is the introduction of intraventricular haemorrhage. If they have any blood in their ventricles, they automatically score a 2 (whereas before they would have scored a 0, which would markedly underestimate their risk of vasospasm)

It appears that the main criticisms of this scale system is the insufficient separation of risk categories (Grade 2 and 3 have the same vasospasm incidence) and the difficulty interpreters seem to have with deciding whether SAH is thick or thin (which degrades the interrater reliability somewhat).

Ogilvy and Carter combination scale.

"Factors that were strongly associated with outcome were used to develop a comprehensive grading system", says their abstract.  Points in this system are awarded to the patient. One point each  for Hunt and Hess grade 4 or 5, Fisher grade 3 or 4, aneurysm size (greater than 10mm), age over 50 years, and (sensibly) whether or not the lesion is a "giant posterior circulation lesion" (i.e. aneurysm greater than 25mm in diameter). By adding the points, a 5-point grading system is developed.

One scale to rule them all?

Which is the best? Hard to say. It may well be that admission GCS is the single best predictor of neurological outcome.