Harutjunyan 2005.

Methods	Design: RCT Setting: ICU setting in a university hospital in Halle, Germany Recruitment period: February 2003‐August 2004 Maximum follow‐up: end of stay in the ICU
Participants	40 "neurosurgical patients at risk of increased ICP" were recruited; 8 did not receive medication as ICP threshold of 20 mmHg was not met; of the remaining 32, only a minority (n = 10) had TBI. Inclusion criteria:"Age >18 years, severe brain damage (Glasgow Coma Score <8) with cerebral edema – visualized by CT scan and continuous monitoring of ICP". Exclusion criteria:"elevated ICP due to space‐occupying lesions with indication for neurosurgical intervention (e.g. bleeding, hydrocephalus), severe renal failure, metabolic disorders, initial serum sodium >150 mmol/l and initial serum osmolarity >320 mosm/kg" (Harutjunyan 2005, p R531). Baseline demographics Age: mean of 47 years (SD = 16 years) across both groups Gender: 17 men, 15 women (8/9 F/M in HTS/HES Group; 7/8 in MTL group) Severity: GCS at baseline 6 ± 1.3 HTS/HES group, 5.8 ± 1.4 MTL group N randomized: 32 (HTS/HES = 17; MTL = 15). Of these, 6 in HTS/HES group had TBI and 4 in MTL group) N for whom ICP data analysed: 32 (HTS/HES = 17; MTL = 15)
Interventions	A standard treatment protocol was followed (details of which appear Harutjunyan 2005, p R531). "All patients were intubated and received pressure‐controlled mechanical ventilation .... Care was taken to keep the arterial partial oxygen pressure above 15 kPa, the hemoglobin concentration above 5.5 mmol/l and the CPP above 70 mmHg. If necessary, blood pressure was supported with vasopressor therapy. Blood glucose was adjusted to values between 6–8 mmol/l by continuous application of human insulin. Patients' core temperature was measured via the bladder, with a target temperature of 36.0–37.0°C. ....Analgosedation and continuous patient monitoring were managed ....Analgosedation at days 1–4 was performed using propofol and sufentanil or remifentanil. Thereafter, midazolam and sufentanil were administered. The standard monitoring ... An increase in ICP was treated first by deepening the sedation and analgesia by titrating the medication and adjusting to adequate ventilator settings. If ICP exceeded the 20 mmHg threshold for more than 5 min, the study medication ... was infused via the central venous line using an automated infusion system at a defined infusion rate. The infusion was stopped when ICP was reduced to <15 mmHg, defined as the treatment goal." Intervention: 7.2% HTS/HES (200/0.5) 6% by infusion Comparator: MTL 15%, by infusion "However, in the case of sustained ICP problems (ICP >15 mmHg or CPP <70 mmHg) after these measures, bolus applications of thiopentone (maximum single bolus: 5 mg/kg) were allowed. In these patients, the possibility of a space‐occupying lesion was excluded by CT scan."
Outcomes	The following were continuously measured: MABP heart rate, ICP calculated CPP SpO2 hematocrit At the end of stay in the ICU, the following were measured: duration of time in ICU mortality
Notes	Trial authors note their focus differs from others, i.e.: "The substantial difference in the design of the present and a comparable study is the fact that we did not administer a fixed total dose, but infused the study medication at a defined infusion rate until ICP decreased to <15 mmHg, the primary goal of our treatment. No clinical study has so far identified an exact dose‐effect relationship for hypertonic saline." (Harutjunyan 2005, R537)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The random code for group assignment was generated by computer" (Harutjunyan 2005, p R532)
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding of participants and personnel (performance bias) participants	Low risk	All participants had GCS score of < 8 and therefore had reduced cognitive function at the time of admission; outcome assessment was not long‐term, so participants had no chance of learning their assignment.
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, as the time of treatment and doses given were different between trial groups.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Objective outcomes such as laboratory measures or death are unlikely to have been much affected by problems of blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	A total of 40 neurosurgical patients were recruited and randomised to receive either 7.2% HTS/HES 200/0.5 or MTL 15%. Only 32 participants were evaluated since in 8 participants, ICP did not exceed 20 mmHg,therefore no trial medication was administered. No relevant clinical characteristics were revealed in the 8 participants not undergoing osmotic therapy. Full data were available for the 32 eligible by reason of exceeding 20 mmHg.
Selective reporting (reporting bias)	Unclear risk	This trial was retrospectively registered in 2005, the year of its publication (ISRCTN62699180). Although expected outcomes are present, in the absence of a published protocol or prospective trial registration, assessment of selective outcome reporting must remain as "unclear."
Other bias	High risk	For the purposes of this review question, this trial's data must be considered at high risk of bias because only a minority of participants had suffered a TBI; a situation further complicated by acknowledged baseline imbalances (trial authors note that "the clinical values in both groups were not normally distributed [at baseline]" (Francony 2008, p R532)