Wills 2005.

Methods	RCT Parallel design Single centre
Participants	Total number of randomised participants: 383 Inclusion criteria: 2‐15 years of age; presenting directly to the hospital with clinical DSS; parent or guardian provided consent Exclusion criteria: not reported Participant condition: DSS Baseline characteristics Colloids group (dextran) Age, median (range): 10 (6‐14) years Gender, M:F: 57:69 Weight (median): 25 (15‐43) kg BP, median (range): SBP: 90 (75‐110); DBP: 75 (57‐90) mmHg Colloids group (HES) Age, median (range): 10 (4.5‐14) years Gender, M:F: 70:59 Weight (median): 25 (14‐40) kg BP, median (range): SBP: 90 (80‐110); DBP: 75 (60‐90) mmHg Crystalloids group Age, median (range): 10 (5‐14) years Gender, M:F: 66:62 Weight (median): 25 (15‐42) kg BP, median (SD): SBP: 90 (72‐113); DBP: 75 (55‐95) mmHg Country: Vietnam Setting: paediatric ICU
Interventions	Colloids group (dextran) Participants: n = 126; losses = 0; analysed = 126 Details: 6% dextran, described as an isotonic colloid Additional details: each participant received 15 mL/kg of body weight of allocated fluid over 1‐h period followed by 10 mL/kg over the second hour; after infusion of study fluid participants received a standard schedule of RL Colloids group (HES) Participants: n = 129; losses = 0; analysed = 129 Details: 6% HES, described as an isotonic colloid Additional details: same as dextran group Crystalloids group Participants: n = 128; losses = 0; analysed = 128 Details: RL Additional details: same as dextran group
Outcomes	Outcomes measured/reported: requirement for supplemental intervention with rescue colloid; time taken to achieve initial and sustained cardiovascular stability; pattern of change in haematocrit; days in hospital; adverse effects (including need for blood transfusion, rashes), mortality Outcomes relevant to the review: need for transfusion of a blood product; mortality (time point not reported); adverse events (rashes)
Notes	Funding/declarations of interest: supported by the Wellcome Trust Study dates: August 1999‐March 2004 Note: this study included a separate arm comparing two colloids for participants with severe shock (pulse pressure, ≤ 10 mm Hg); we did not include these participants because colloids were not compared with a crystalloid
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Use of computer‐generated random numbers completed by independent research staff
Allocation concealment (selection bias)	Low risk	Allocation concealed through treatment packs of fluid prepared in advance, in cardboard containers, and only identifiable by a study number
Blinding of participants and personnel (performance bias): mortality	Low risk	Treatment packs of fluid were prepared in advance, in cardboard containers, and only identifiable by a study number
Blinding of participants and personnel (performance bias): transfusion/renal replacement therapy/adverse events	Low risk	Treatment packs of fluid were prepared in advance, in cardboard containers, and only identifiable by a study number
Blinding of outcome assessment (detection bias): mortality	Low risk	No details of blinding for assessment of mortality; lack of blinding unlikely to influence data for this outcome
Blinding of outcome assessment (detection bias): transfusion/renal replacement therapy/adverse events	Low risk	Blinding reported for assessment of other outcomes, and we assumed that assessment of transfusion data was also blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	No losses
Selective reporting (reporting bias)	Unclear risk	No details of clinical trials registration or prepublished protocol; not feasible to assess risk of selective outcome reporting bias
Baseline characteristics	Low risk	Baseline characteristics appear comparable
Other bias	High risk	31% participants in the crystalloid group were also given colloids and this may have influenced study results