Duke 2002.

Methods	Setting: Hospital Inpatient department Study design: Randomised, parallel group, multi‐centre, controlled trial Location: Papua New Guinea Timing and duration: September 1997 to October 2000 Number of centres: 3 Source of funding: Roche, World Health Organization, and Royal Australasian College of Physicians
Participants	Children with clinical signs of meningitis, cloudy or turbid CSF with moderate or large amounts of leucocytes and protein on dipstick testing (Multistix 10 SG) were eligible for inclusion. Children with renal failure, congenital heart disease, who had received parenteral antibiotics for 48 hours or more in the week prior to presentation or who were septic or in hypovolaemic shock were excluded from enrolment Age: > 1 month to < 12 years
Interventions	Nasogastric tube fluids at 60% of maintenance fluids, (maintenance fluids defined by "100 ml/kg/day for the first 10 kg of body weight, 50 ml/kg for the second 10 kg, and 20 ml/kg for over 20 kg") as expressed breast milk or other milk feed, divided into feeds given every 3 hours versus 100% of normal maintenance fluids (defined as above) administered intravenously (given nasogastrically in 7 children because an intravenous cannula could not be inserted) given as a solution containing 0.45% sodium chloride and 5% dextrose plus 10 mmol/L of potassium chloride per litre Duration: 48 hours
Outcomes	Death Neurological sequelae Oedema (including cerebral) Serum and urinary sodium Seizures
Notes	260 of the 357 children had confirmed bacterial meningitis. The paper states that although no bacteria were isolated in the other children the diagnosis was "definitely meningitis". Numbers of children without isolated bacteria was similar between groups Severe sequelae were considered to be present if 14 days after commencing treatment there was a severe motor deficit (marked spasticity, hemiplegia, severe hypotonia) and at least one of the following: a major sensory deficit (inability to fix and follow in an age‐appropriate way or no response to sound), persistent convulsions or coma All children received phenobarbitone, and received oxygen for the first 48 hours. The 1st 150 children received chloramphenicol, the rest ceftriaxone. Mechanical ventilation was not available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate, with comparable treatment and control groups at entry
Allocation concealment (selection bias)	Low risk	Adequate, using sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Clearly not
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Low risk for outcomes of death and acute severe neurological sequelae
Incomplete outcome data (attrition bias) All outcomes	Low risk	Adequate overall with 11 of 357 excluded post‐randomisation as found not to have meningitis. However, over 10% of participants lost to follow‐up at 3 months
Selective reporting (reporting bias)	Low risk	Adequate, with a good range of appropriate outcomes reported. 14 days is somewhat early to judge whether severe sequelae were present
Other bias	Low risk	A large and well‐described study