Brodke 2003.

Methods	Method of randomisation: based on the day of admission Assessor blinding: unclear, but probably not Loss to follow‐up or excluded from analysis: Anterior group: 2/22 (2 died) Posterior group: 3/30 (3 did not complete 6 months of follow‐up)
Participants	Country: USA Period of study recruitment: 01/01/1991 to 31/12/1993 Total number: 52 participants Inclusion criteria Unstable cervical injuries between C3 and C7 Complete or incomplete spinal cord injuries Minimum six months of follow‐up (post‐randomisation exclusion) Exclusion criteria Patients requiring a specific approach for reduction or decompression Patients with radiculopathies only or neurologically intact Age (of 47 followed up) Anterior group: 38 years Posterior group: 33 years Gender Male: 37 Female: 10
Interventions	Immediate reduction in the Emergency Department. Surgical stabilisation a few days later. 1. Anterior procedure: discectomy or corpectomy + autologous tricortical iliac crest graft harvested for fusion + fixation with an anterior cervical locking plate 2. Posterior procedure: posterior fusion with iliac crest cancellous autograft + fixation with lateral mass screws and plates Postoperative care: Minerva braces or Miami J collars for 8 to 10 weeks. Mobilisation as soon as tolerated and transferred for spinal cord rehabilitation as the patient's other medical conditions allowed
Outcomes	Length of follow‐up: minimum six months (anterior: mean 17 months; posterior: mean 14 months) Neurological outcomes: ASIA (American Spinal Injury Association) motor score and Frankel score Radiographical outcomes: fusion and sagittal alignment Clinical outcomes: postoperative pain Complications
Notes	The authors of the study did not respond to our questions about missing information on trial methodology The population of the study included the following types of injuries: burst fracture (anterior versus posterior): 4 versus 3 pure facet dislocations: 6 versus 18 burst fracture plus facet dislocation: 8 versus 4 flexion‐compression injury plus facet dislocation: 1 versus 2 extension/distraction injury: 1 versus 0
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Randomisation was based on the day of admission
Allocation concealment (selection bias)	High risk	Since the allocation sequence was deemed inadequate, it could not be properly concealed
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding is not feasible for participants nor treatment providers in this sort of trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding of outcome assessors was not mentioned. There was reference to independent review by the lead author
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	All participants lost to follow‐up or excluded were reported. However, their data were not presented. Five participants were not included for final pain analysis: anterior group: 2 participants ‐ died of other injuries in the early postoperative period (9%) posterior group: 3 participants ‐ did not complete six months of follow‐up (10%)
Selective reporting (reporting bias)	Unclear risk	All the outcomes described in the methods section were presented in the results. However, the scoring system for neurological status was not specified, not fully reported and is inappropriate (it was not used in this review)
Other bias	High risk	There was imbalance between the two groups in the type of fracture and extent of cord injury (Frankel level) There was a three month between‐group difference in the mean time to follow‐up, and the minimum follow‐up time of six months did not allow evaluation of long‐term clinical and radiographical outcomes Difference in timing of surgery. The anterior group was operated on an average 10 days from injury; the posterior group an average five days from injury