Table 1

Papers cited in the NICE guidelines about neurosurgical transfer

Paper	Study design	Source of data	No. of patients	Comparison	Outcomes
Hannan17 (2005)	Retrospective observational cohort study	New York State Trauma Registry, USA (1996–1998)	2763 head injured patients (GCS <14)	By ambulance destination: 2272 trauma centre 491 non-trauma centre	Odds of mortality (1) vs (2): 0.67 (0.53 to 0.85)
Di Russo18 (2005)	Retrospective observational cohort study	National Pediatric Trauma Registry, USA (1994–2002)	5460 intubated patients (age <20, primary diagnosis ‘injury’)	By site of intubation: 1939 in the field 1874 non-trauma centre 1647 trauma centre	Mortality stratified by RHISS: higher for (2) vs (3) at all severities
Stevenson19 (2001)	Simulation model to compare triage strategies	N. Staffordshire RNU Local data, publications and expert opinion	10 000 simulated head injuries	11 triage strategies	No superior strategy, but current policy (take to nearest DGH)→delayed intervention
Patel20 (2005)	Prospective observational cohort study	UK Trauma Audit and Research Network database (1996–2003)	6921 blunt head injuries admitted to TARN hospitals (any age)	By treatment location 2305 non-neurosurgery 2677 neurosurgery unit	Odds of mortality (1) vs (2): 2.15 (1.77 to 2.60)
Poon21 (1991)	Prospective observational cohort study	Single RNU in Hong Kong (1985–1989)	104 patients requiring surgery for EDH	By transfer: 71 direct to RNU 33 transferred DGH	Mortality rate: 4% (1) vs 24% (2)
Härtl22 (2006)	Prospective observational cohort study	Data from 24 trauma centres in New York State, USA (2000–2004)	1123 head injury patients treated at a trauma centre (GCS <9)	By transfer: 254 transferred from a non-trauma centre 864 direct to trauma centre	Odds of mortality (1) vs (2): 1.48 (1.03 to 2.12)

DGH, district general hospital; EDH, extradural haematoma; GCS, Glasgow Coma Scale; RHISS, relative head injury severity scale; RNU, regional neurosciences unit; TARN, Trauma Audit and Research Network.