Kumar 2019.

Methods	Design: parallel, equivalence RCT (single site) Setting: ICU of a tertiary neurosurgical centre in Bangalore, India Recruitment period: January 2012‐June 2014 Maximum follow‐up: 6 months
Participants	50 children were assessed; 30 met eligibility criteria (20 were excluded with reasons given). All children underwent initial resuscitation, evaluation and treatment for TBI as required; CT scan studies on admission were analysed using Marshall criteria (Marshall 1992) Inclusion criteria: children in age group 1‐16 years with severe TBI, defined as post‐resuscitation Pediatric GCS of ≤ 8, and presenting within 24 h of trauma Exclusion criteria: having a GCS of 3; absent brain stem reflexes; systemic injuries requiring immediate treatment; clinical evidence of significant spinal cord injuries; or presenting > 24 h after injury Baseline demographics Age: a mean for age is not reported. Trial authors report no significant difference between groups. There is a typographical error in the total number for the ranges of age groups, but the disaggregated totals add up to 30, and are reported as follows: of eligible children admitted to the trial, 14 children were aged 1‐5 years; 7 children age 6‐10; 9 children aged 11‐16 Gender: 12 girls, 18 boys; 6:14 in the HTS group; 6:16 in the MTL group Severity: post resuscitation GCS means and SDs were given by group. These were 7.4 (SD = 0.9) in the HTS group and 6.6 (SD=1.1) in the MTL group. Data on manner of injury (road accidents, falls); pupillary reaction to light; head CT findings; interval between injury and insertion of EVD, duration of monitoring, duration of ventilation, duration of ICU stay, and duration of hospital stay, were also collected N randomized: 30 (HTS, n = 14; MTL n = 16) N for whom data analysed for ICP/GOS data analysed: trial authors report no missing data; data available for the full sample (HTS, n = 14; MTL, n = 16)
Interventions	"The aim of the therapy [ICP monitor setup and treatment, protocol described] was to maintain the ICP below 15 mmHg in children between 1 and 10 years of age and 18mmHg in children age 11–16 years of age [7, 8]. When the ICP remained raised more than the cutoff value for more than 5 min in the absence of noxious stimuli like suction, positioning, etc., it qualified as an intracranial hypertensive (ICH) episode. For an ICH episode, the EVD was opened to drain CSF until it stopped flowing or up to 20 cc release of CSF whichever is first...After successful insertion of EVD and ICP monitoring, the patients were randomized to receive one of the interventional agents. ...[which]were administered if ICP remained persistently above the cutoff value for more than 5 min in spite of CSF drainage. The ICU staff informed each episode of raised ICP to one of the investigators (AK or DS), who was available at bedside before initiating treatment for reduction of ICP. The investigator personally documented ICP before initiation and after completion of treatment, and measured reduction in ICP for each dose of medication" (Kumar 2019 p 1000) Intervention: equiosmolar dose of 3% HTS (1027 mOsm/L) as a bolus of 2.5 mL/kg through the central venous line over a period of 5 min (n = 14 participants) Comparator: equiosmolar dose of 20% MTL (1098 mOsm/L) as a bolus of 0.5 g/kg (2.5 mL/kg; n = 16 participants) "If the ICP did not decrease even after two consecutive doses of the hyperosmolar agent, it was considered refractory to therapy" (Kumar 2019 p 1000) Monitoring of active treatment appears to have continued for 5‐6 days
Outcomes	Primary outcome: mean reduction of ICP ‐ defined as "the difference between ICP value before administering hyperosmolar agent and lowest ICP value after completion of bolus for each dose. The mean reduction in ICP was obtained by summing the difference in ICP values before and after treatment divided by number of doses during the entire period of ICP monitoring" (Kumar 2019, p 1001). Secondary outcome: neurological outcome (GOS) At 6 months. This was modified for children by replacing the outcome of 'work' with 'activity or scholastic performance' for age and pre‐injury status of child (Beers 2012).... The functional outcome was analyzed as death or survival in vegetative state versus survival with or without disability" (Kumar 2019 p 1001, emphasis added ‐ this is not the conventional 'cut‐off' for 'poor' or 'good' outcome) Other data collected included: ICP and ICP‐related data including mean ICP, MAP, and CPP for each day, number of episodes of raised ICP requiring CSF releases, volume of CSF drained, and number of doses of osmotic drugs required per day duration of monitoring, duration of ventilation, duration of ICU stay, and duration of hospital stay mean values of serum glucose, sodium, creatinine number of participants requiring isonotropes
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The randomization was done through computer‐generated random numbers" (Kumar 2019, p 1000). Appears adequate
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) participants	Low risk	Participants had brain injury and neither they nor parents/carers were likely to be aware of treatment allocation.
Blinding of participants and personnel (performance bias) Treating physicians	High risk	There is no indication that the treating physicians were blind to the treatments given, and the trial is described as "single blind".
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Trial authors report that "The person who assessed outcome was blinded for the interventional agent" (Kumar 2019, p 1001).
Incomplete outcome data (attrition bias) All outcomes	Low risk	For the outcome of ICP, there appear to be no data missing.
Selective reporting (reporting bias)	Unclear risk	Trial authors report that this trial was registered with Clinical Trials Registry of India (REF/2015/03/008696). The date suggests retrospective registration.
Other bias	Unclear risk	Authors note that "The limitation of our study was small sample size. ...For adequate power of study, a large multicenter study is warranted. The present underpowered study cannot generate a good class of evidence but demonstrates the feasibility of such studies at a larger scale. The second limitation was use of hyperosmolar therapy as second tier treatment after failure of CSF drainage to reduce the ICP. The EVD, when available, is recommended prior to hyperosmolar therapy.... The EVD as an initial treatment may dilute the effect of hyperosmolar therapy. It is not known, whether there will be any difference in ICP reduction between mannitol and hypertonic saline if any of these agents are administered as first‐line therapy. Many centers do not use EVD for ICP monitoring. When ICP monitoring is done using parenchymal sensor, option of CSF drainage is not available, and true effect of hyperosmolar therapy can be assessed. The third limitation was that we did not measure time to peak effect and duration of effect, cerebral blood flow, cerebral tissue oxygen, cerebral metabolism, cerebral injury biomarkers, cerebrospinal compliance, and pressure reactivity. The multimodal monitoring is labor intensive, and is not available in our set up" (Kumar 2019, pp 1004‐5).