Zhou 2006.

Methods	Randomised.
Participants	112 puerperants with PPH due to uterine atony who received routine management for uterine atony. Exclusions were as follows: younger than 18 years of age; any pre‐existing heart condition; high blood pressure for which they had received medication in the previous 2 years; any pre‐existing blood condition, whether from birth or contracted later in life, such as haemophilia; history of suffering from or exhibiting symptoms of progressive hepatitis or endocrinosis; having undergone traditional caesarean; having undergone general anaesthetic in case of placenta previa, or if the cervical muscles had undergone surgery. 52 assigned to test group, 60 to control group.
Interventions	4 mg estradiol benzoate injected intramuscularly with routine management when bleeding exceeded 500 mL versus routine management only for the control group. Routine management of the control group was described as 'uterine massage and uterotonics administration' and included '20 U cervical muscle injection to contract the uterus; 20 U intravenous drip to contract the uterus. In case of the cervical muscles not restoring, injection or intravenous drip did not exceed 80 U. Where rate of blood loss exceeded 2000 ml, hysterectomy was performed'.
Outcomes	Rate of blood loss at 2 hours and 2 to 24 hours and any reported instances of hysterectomy up to 24 hours.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	High risk	Authors reported data only for blood loss and hysterectomy.
Selective reporting (reporting bias)	High risk	Authors reported data only for blood loss and hysterectomy. No prior public registration of protocol.
Other bias	High risk	Unclear how blood loss was measured. No sample size calculation. We were unable to identify management before randomisation.