Removing some of the skull is a neat way of cheating the Monro-Kellie doctrine. Pity it does you no favours.   LITFL have an excellent page devoted to this practice. Additionally, an article directed at neurosurgeons is available and it highlights all the important features. 

Decompressive craniectomy (not specifically in TBI) has appeared several times in the CICM Part II past papers. The most detailed SAQ had been Question 25 from the second paper of 2018,  which had asked the candidates to critically evaluate the use of decompressive craniectomy in traumatic brain injury. It was also the subject of in Question 9 from the first paper of 2009, where the focus was more on indications complications and outcomes. Its use in malignant MCA infarction appeared in  Question 2 from the second paper of 2014. To the latter, an entire chapter is dedicated ("Decompressive craniectomy for malignant MCA infarction").

As most indications for decompression are neurological and nuerosurgical, the best article to read about this topic would probably be the 2016 paper by Crudele et al, provided one can get a hold of a free copy. It goes through TBI, SAH, malignant MCA infarction and parenchymal intracerebral haemorrhage.  The other good articles for this topic are locked up behind paywalls. However, if one were ever going to throw money into that bottomless hole that is Springer-Verlag, one could do worse than this article by S.Y Chu, from Current Treatment Options in Neurology (2015). Its major attraction is a nice big table discussing indications for decompressive craniectomy. Owing to his poverty, the author of this manuscript could not throw money into that particular hole, and was forced to compile his own list of indications.

In brief:

In greater detail:

Rationale for decompressive craniectomy

To put simply, borrowing words from the abstract to an article by Quinn et al (2011),

"the rationale for this procedure is based on the Monro-Kellie Doctrine; expanding the physical space confining edematous brain tissue after traumatic brain injury will reduce intracranial pressure".

If one were forced to spell things out in a viva or a CICM SAQ, it would benefit from a platform approach, as if explaining the concept to a toddler - as this would be much easier to mark than prose. Thus:

• Raised intracranial pressure contributes to secondary brain injury and morbidity/mortality from TBI
• Intracranial pressure is governed by the Monro-Kellie Doctrine, where intracranial content is confined within a finite space and any expansion of one component produces increased pressure on other components
• Decompressive craniectomy increases the volume available to the intracranial contents
• The consequence should be decreased ICP and decreased secondary brain injury.

The surgical approach

There are many pages written on this topic, and not being a surgeon, I will refrain from meaninglessly gibbering about it. A few features, however, are worth knowing:

• A bilateral craniectomy is indicated if there is diffuse generalised swelling, as supported by a lack of midline shift. Unilateral procedures are performed for unilateral problems.
• How much skull to remove? Apparently, a minimum diameter of 10cm allows an additional volume of 50ml.
• Don't be stingy. A small craniectomy tends to cause venous congestion around the reflected dura; additionally, the brain will herniate through the opening like toothpaste, infarcting and lacerating itself on the sharp edges.

Indications for decompressive craniectomy

There are several indications for decompressive cranienctomy, of which some are vigorously debated, and others well accepted. Frequently, the debate is not about the intracranial pressure benefits of removing the bit of skull, but more about the enhancement of survival for the severely disabled patients who would otherwise have died.

Malignant MCA infarction

This is the only agreed-upon firm indication for decompressive craniectomy. It is supported by the findings of three European trials, and enjoys a thorough discussion elsewhere.

In brief:

• DECIMAL trial: 38 young patients (18-55 years old) were recruited. Craniectomy decreased mortality by 50%.
• HAMLET trial: 64 patients were enrolled; craniectomy also decreased mortality by about 50%.
• DESTINY trial: 32 patients were enrolled; again a decrease in mortality of roughly 50%.

It is worth pointing out that in the DESTINY trial, the numbers were small because statistical significance was achieved with only 32 patients. Conversely, the research board terminated the DECIMAL trial because they had trouble recruiting. Anyway. The bottom line is a decrease in mortality. Additionally, it seems in the craniectomised survivors the hemiplegia is less severe; this may be viewed as a positive step, even though the therapy still generates a stream of disabled hemiplegic people.

Traumatic brain injury

This was the subject of Question 25 from the second paper of 2018. In summary, the evidence for decompressive craniectomy was explored in several trials which need to be well known to the ICU trainee:

• The DECRA trial enrolled relatively few patients (n=155) and, when they sustained an ICP of over 20 for 15 minutes (which was refractory to stage 1 interventions) randomised them to either receive standard care (i.e. some Stage 2 ICP lowering measures) or decompressed them immediately. Even though this controlled the ICP (from 20 down to 10-15) there was no strong evidence for any mortality benefit (19% vs 18%). There was, however,  a significant difference in the neurological outcome: the practice produced a population of severely brain-injured patients who would have otherwise died, and who now live on in a severely disabled or persistently unconscious state. Apart from allowing the survival of severely brain-injured patients (who would have otherwise died), axonal stretch and venous infarction were blamed by the investigators. This trial was criticised for many things; Torrers (2012) has an excellent breakdown. Most of the critics were outraged by the low threshold for decompression. They complained that under most circumstance, reasonable people would try a few other things first, and wait longer before moving to surgery. Other problems with it included:
  • Patients with "mass lesions" were excluded, which is a description of anything ranging from a brain tumour to an extradural haematoma (thus, a large proportion of routine TBI cases was excluded).
  • The groups were heterogeneous (the decompressive craniectomy group probably had patients with more severe brain injuries, as demonstrated by the increased prevalence of unreactive pupils among them).
  • The bifrontal procedure which was mandatory for the surgical group is actually suboptimal for controlling ICP (without falcine sectioning you don't get the same benefit) but this argument falls somewhat flat considering how good the ICP control was in the decompression group
  • The 155 patients for this trial took 8 years to enrol across 15 tertiary units; the timeframe (2002-2008) saw at least one change of the BTF guidelines as well as other changes in ICU management, which may serve to confound the result.
  •  Prior to the randomisation, both groups had relatively normal-ish ICPs (upper limit of normal, approaching 20 mmHg). 
• The RESCUEicp trial enrolled 408 patients and had a slightly higher thrreshold for craniotomy (sustained ICP of > 25mmHg, and sustained for longer, in spite of a good trial of Stage 2 therapies). Again, decompression saved lives (mortality 26.9% vs 48.9%) but produced many more patients who could be broadly described as "severely disabled". Several criticisms exist:
  • Of the patients randomised to medical management, 37% underwent decompressive craniectomy anyway. 
  • As with DECRA, recruitment of all 408 patients took 10 years. Half the centres recruited only 3 patients, or fewer.
  • Therapeutic hypothermia was one of the optional Stage 2 rescue therapies prior to reaching for the bone saw, which - we now realise- may confer a survival disadvantage.

The new (2016) BTF Guidelines are careful to make clear that improving outcome is not what this technique is for. A large bifrontal decompression (no less than 12 × 15 cm) is recommended if you are going to go down that road.

Advantages of decompression in TBI include

• Maybe some sort of mortality benefit (cohort studies such as Sonuca et al, 2010; as well as RESCUIicp)
• Shorter ICU stay
• Less ICP-targeting interventions
• Lower ICP

Disadvantages include:

• Conflicting evidence for mortality benefit (19% vs 18% in the DECRA trial, versus 26.9% vs 48.9% in RESCUIicp)
• Worse neurological outcome in survivors
• Multiple complications associated with decompressive craniectomy:
  • Herniation through the defect
  • Delayed paradoxical herniation
  • Subdural hygroma
  • Infection
  • Bleeding
  • Post-traumatic hydrocephalus
  • "Sinking Flap Syndrome"
  • Bone resorption

Subarachnoid haemorrhage with vasospasm-associated oedema

Case series report good outcomes. However, even people who publish these case series tend to mention that there is a lack of definitive evidence to support this practice. In general, the studies which report good outcomes also report that one is better off performing a craniectomy within 48 hours of the haemorrhage. Patients with progressive brain oedema but no radiological signs of infarction
and those with obvious hematoma will benefit the most. Overall, though large matched group analysis failed to identify any real survival benefit (or improvement in disability), a small subset of patients (about 25%) achieved a good long-term outcome, suggesting that careful patient selection is the key to success.

Non-traumatic intracranial haemorrhage

Again, this practice is built on shaky foundations. Small case series report good outcomes and refer to decompression as a "life saving" measure. A more recent (2015) local case series of ICH in patients with arteriovenous malformations produced outcomes which the authors described as "good", even though of their twelve patients seven ended up with severe disability and one was described as "vegetative". The same author, writing about hypertensive ICH in 2013, noted a 71% incidence of a poor outcome in a case series of 21 patients, with a mortality rate of 10% overall.

Encephalitis

A literature review from 2008 identified papers with a total of 13 cases. In the majority of these case reports, the authors waited until there were features of brainstem herneation before they went ahead with the procedure. The reviewers were forced to conclude that if you are about to cone, decompressive craniectomy begins to look very attractive.

Dural sinus thrombosis

Small scale retrospective studies have found that in the patients with the most severe cereral venous thrombosis, decompression leads to more favourable outcomes. Most importantly, it seems that full systemic anticoagulation within 24 hours of surgery seems to be safe. A 2009 review from Stroke courageously asserted that the therapy is "lifesaving" and that the outcome can be "excellent", on the basis of three cases. A larger and more recent case series from Pakistan also reported excellent outcomes from a series of seven patients, although worryingly "1 of 7 left against medical advice (in a comatose state) and was lost to follow-up."

Subdural haematoma

Decompressive craniectomy is an option for this condition, even though surveys indicate that the majority of neurosurgeons use it in less than 25% of cases. A 2012 retrospective review found that the patients who were decompressed were likely to have a more severe pathology and be more comatose, which likely influenced the somewhat discouraging mortality data for that group.

Complications of decompressive craniectomy

The complications of decompressive craniectomy are well summarised in an article by Margules and Jallo. I will present them as a list:

• Brain hernation though the opening: This wil result in localised pressure necrosis around the edges, as well as venous "pinching" and congestion of the whole hernated mass. The smaller the craniectomy, the greater the risk of this occurring.
• Delayed paradoxical herneation: This is a weird complication of lumbar puncture in a pre-cranioplasty patient. Essentially, the opening of a lumbar drain allows the CSF to leak freely out of the spinal canal; however because the cranium is open, there is no negative pressure fastening the brainstem into the skull, and it readily herneates through the foramen magnum, killing the patient.
• Subdural hygroma: This is a collection of CSF which forms under the operative site after a craniectomy is finalised. It seems about half of all patients end up with this, to some degree. Most disappear harmlessly; some persist and cause a mass effect.

• Infection: The risk of this is abut 3-7%, with perioperative antibiotics. Bone flap osteomyelitis is more rare, but much more disturbing.
• Bleeding: A non-specific surgical complication.
• Post-traumatic hydrocephalus: In general, people who have decompressive craniectomy tend to develop this about 4 times as often as people who are managed non-surgicaly. However, people who end up needing decompression probably also had a more severe head injury. Potentially, some sort of arachnoid granulation injury could be the aetiological cause of this complication.
• Syndrome of the Trephined, or "Sinking Flap Syndrome": This is a long-term complication, associated with poor cognitive performance, memory loss, irritability, headaches and dizzyness. Nobody actually knows what causes it, but people tend to blame the fact that the brain is exposed to atmospheric pressure, and is thus subject to unusual changes of surface perfusion. These problems tend to resolve after cranioplasty.
• Bone resorption: After you have carefully positioned your bone flap and closed the patient's scalp, you may find to your dismay that within months the flap is gone. Where did it go? It was gobbled up by osteoclasts, as any stray piece of dead bone would be. Synthetic cranioplasty may be the answer to this problem, especially if the flap is in many pieces.