Results: observational study

Study flow chart

Please see Report Supplementary Material 2.

Recruitment

A total of 386 patients from 37 sites in 10 countries were recruited to the observation data set between 8 September 2014 and 29 April 2019 (UK 22, India 4, Canada 3, USA 2, Spain 1 Malaysia 1, Greece 1, Germany 1, Australia 1, Hungary 1).

Numbers analysed

A total of 3566 patients were screened for eligibility, with 386 enrolled in the observed arm. There were 141 in the craniotomy arm and 245 in the DC arm.

Baseline data

Behavioural history, demography and coagulation history

Patients having DC were 17 years younger on average, less likely to have insurance, less likely to have been on anticoagulants or aspirin and more likely to have had a GCS between 3 and 8 (Table 11).

Accident details and pre-hospital events

Almost all injuries were closed-head injuries (96%). Patients undergoing DC were more likely to have extracranial injury requiring admission. They were more likely to have been injured as a motorcyclist and less likely to have been injured in an incidental fall. Their experience of pre-hospital events was similar. (Combining suicide, other non-intentional injuries with other gives p < 0.0001 Table 12.)

Medical history

Patients treated by craniotomy had more medical history (Table 13).

Please see Appendix 10 for admission vital signs table.

Glasgow Outcome Scale-Extended outcomes

Six-month Glasgow Outcome Scale-Extended scores

Twelve-month Glasgow Outcome Scale-Extended scores

Case report form stage 1 results

Treatment received

Of the craniotomy patients, three mention use of floating brain flap, and one of the DC patients mention bifrontal DC. There were seven ‘Other’ reasons for not randomising in the craniotomy group: age (1), system failure (2), family refuse randomisation (2), recovered from stroke (1), infection of scalp (1) (Table 16).

There were three ‘Other’ reasons for not randomising in those that had DC: skull fracture (2), not expected to survive (1).

Computed tomography scan results

Patients with parenchymal lesions or traumatic subarachnoid haemorrhage (SAH) were more likely to have DC. Please see Appendix 10 for full table.

Pre-intubation Glasgow Coma Scale and pupil reactivity

Patients receiving DC have lower GCS scores (Mann–Whitney U-test p < 0.0001) and are more likely to have neither pupil reactive (than at least one reactive) (p = 0.009). Please see Appendix 10 for full table.

Surgery details and postoperative destination

There was no difference in the location of the ASDH between the treatment groups (p = 0.630).

There were no differences between the treatment groups in terms of the type of skin incision, with 89% reverse question mark (p = 0.695), or in the size of the bone flap [with a mean of cm (± cm)].

A drain was used in 78% of patients and was usually subgaleal (55%) rather than subdural (23%) [no difference between treatments (p = 0.718)]. Intracranial monitoring devices were used in 67% of patients, and most (78%) being located ipsilateral to the ASDH. They were more likely to be ipsilateral in craniotomy (p = 0.004). For the patients who underwent DC, the bone flap was either stored inside the body (34%) or disposed of (35%) (Table 17).

There were more likely to be intraoperative complications in DC (χ² p = 0.010). There were no differences between treatment groups in the destination from the operating theatre (χ² p = 0.984).

Stage 2 case report form results

Glasgow Coma Scale and discharge details

Stage 2 GCS and discharge details for first stay of any patient in ITU. Table 18 includes all those who were admitted to neurocritical care unit (NCCU) from the operating theatre.

Stage 3 case report form results

Neurosurgical interventions within 14 days of index surgery

These are recorded on the discharge from NSU CRF. The information is not recorded for 84 patients (24 craniotomy and 60 DC) (including most of the patients who died in ICU or were discharged from there to a different hospital) (Table 19).

Length of stay of all patients discharged from neurosurgical unit or intensive care unit

Among patients who were discharged alive, those who had had DC as initial surgery had a longer stay (Mann–Whitney U-test p = 0.007). There was no difference in time to death for those who died in hospital (Mann–Whitney U-test p = 0.449) (Table 20).

Status and destination at discharge from any part of the neurosurgical unit

Discharge destination (combining nursing home, unknown and other) (p = 0.014). GCS total χ² = 0.056 (3 category) (Mann–Whitney U-test on full score p = 0.048). Please see Appendix 10 for full table.

Stage 4 case report form results

Stage 4 cranioplasty and shunt use

Details about cranioplasty are recorded at 12 months.

If a patients had a DC, then there was no difference between the allocated groups in whether they had a cranioplasty by 12 months (χ² p = 0.466) (Table 21).

There was no difference in the rate of hydrocephalus between the two patient treatment groups (χ² p = 0.145) or in the use of a shunt if they did develop hydrocephalus (Fisher’s exact p = 0.247).

Adverse events: observational

One hundred and twenty-three AEs are reported by 76 (54%) of the 141 patients having craniotomy as initial treatment. The maximum number of events reported for a single patient was 7, and the average number of events reported by those randomised to craniotomy was 0.87. One hundred and sixty-seven AEs were reported by 105 (43%) of the 245 patients having DC as initial treatment. The maximum number of events reported for a single patient in this group was 4, and the average number of events reported by a patient randomised to DC was 0.68. There are no differences in the types of AEs experienced between the two treatment groups (Table 22).

Procedure-related adverse events: observational

The maximum number of events reported for a single patient was five. Please see Appendix 10 for full table.