Methods	Prospective RCT.
Participants	40 women with diagnosis of HELLP in postpartum period. (HELLP defined using as presence of clinical signs and symptoms consistent with diagnosis of pre-eclampsia and lab evidence of hemolysis, hepatic dysfunction and thrombocytopenia. Mississippi triple class system: women with class 1 or 2 HELLP, i.e. platelets < 100,000 cells/mm3) Exclusions: women with clinical evidence of chorioamnionitis at delivery
Interventions	Intervention group: (20 women) immediately after delivery, 10 mg IV dexamethasone, then 10 mg 12 hrs later, 5 mg at 24 and 36 hrs. Total of 30 mg over 36 hrs. Control group: (20 women) routine care, no corticosteroids. Both groups were intensively monitored, MAP, urinary output every two hrs and lab tests every 6-12hrs. Both groups had magnesium sulfate IV 2 gm/hr
Outcomes	Symptom resolution and laboratory values returning to normal (adequate urine output, change in MAP, BP < 160/100, platelet trend upwards, LDH and liver enzymes downwards) Postpartum infectious morbidity.
Notes
Risk of bias
Item	Authors’ judgement	Description
Adequate sequence generation?	Yes	Computer generated card.
Allocation concealment?	Yes	“computer generated card enclosed in an opaque envelope.”
Blinding? All outcomes	Unclear	1 group received no treatment, not clear whether partial blinding was attempted
Incomplete outcome data addressed? All outcomes	No	While there was no mention of loss to follow up there were considerable missing data for some outcomes (e.g. results for 25/40 for urinary output)
Free of other bias?	Unclear	No baseline imbalance was reported although the graphs in the results section seem to suggest a different distribution of lab values in the 2 groups. This may not be meaningful as the numbers are small and the distribution may not be normal. Results are presented in graphs that are difficult to interpret and data are not in a form that allows us to analyse using Review Manager Software (RevMan 2008).
Overall risk of bias?	Unclear	Risk of performance bias as blinding not indicated. High levels of missing data