UK 2009.

Methods	Allocation concealment: centralised automated interactive voice response system. Two arms
Participants	39 singleton pregnant women with PE, defined as DBP ≥ 90 mmHg on 2 separate occasions, at least 4 hrs apart occurring after 20 weeks' gestation and associated with significant proteinuria (≥ 2 + protein on urinalysis), and women who developed superimposed PE on background chronic HT if they developed new significant proteinuria. Excluded: delivery indicated within 24 hrs, uncontrolled HT (BP > 170/110 mmHg), renal or collagen vascular disease, relative contraindications to sildenafil or taking medications that interact with sildenafil.
Interventions	Exp: oral sildenafil 20 mg three times daily. (up to 80 mg a day). (19 women) Control: oral placebo of identical appearance three times daily (20 women)
Outcomes	Women: prolongation of pregnancy from randomisation to delivery (days),umbilical artery pulsatility index, maternal serum urate, maternal BP, adverse events, dosing and pharmacokinetics of sildenafil in women with PE. Babies: Apgar scores, umbilical arterial and venous cord gases at delivery and placental weight, fetal adverse events.
Notes	Methods for measuring BP not mentioned. Funding: funded by industry. Conflict of interests declared.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	1:1 ratio using a centralised automated interactive voice response system. Randomisation stratified by centre, gestational age (24 to 28 weeks, ≥ 28 to 34 weeks) and disease status (PE alone or PE with IUGR).
Allocation concealment (selection bias)	Low risk	Centralised automated interactive voice response system.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind. Placebo of identical appearance.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Yes.
Incomplete outcome data (attrition bias) All outcomes	High risk	4 women (10.3%) withdrew consent after randomisation (2 in each group).
Selective reporting (reporting bias)	Low risk	Protocol registered in advance. ClinicalTrials.gov Identifier: NCT00141310.
Other bias	High risk	Small numbers. Funded by industry. Conflict of interests declared.