Summary of findings 2.

Erythropoietin compared with placebo or no treatment for complications of preterm birth
Patient or population: preterm infants with low birth weight Settings: NICU Intervention: EPO Comparison: placebo or no treatment
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Placebo or no treatment	EPO
Mortality during initial hospital stay (all causes of mortality)	High‐risk population		RR: 0.89 (95% CI 0.68 to 1.16)	2212 (20)	⊕⊕⊕⊕ high	Bias: We had concerns about bias (lack of blinding) in 10 of the included studies, but the outcome of mortality is not likely to be affected by researchers knowing the treatment assignment. We did not downgrade the quality of evidence on this item. Heterogeneity/Consistency: We noted no heterogeneity (I2 = 0%). Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (2212), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot was symmetrical.
	92 per 1000	82 per 1000 (0 to 172)
Retinopathy of prematurity (stage ≥ 3)	High‐risk population		RR: 1.24 (95% CI 0.81 to 1.90)	1283 (8)	⊕⊕⊕⊕ high	Bias: We found no risk of bias in any of the studies, except in the smallest study that enrolled 40 neonates. We did not downgrade the quality of evidence. Heterogeneity/Consistency: We noted no heterogeneity for RR (I2 = 0%) and low (I2 = 34%) heterogeneity for RD. Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1283), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot, which included 8 studies, was symmetrical.
	53 per 1000	65 per 1000 (0 to 195)
Intraventricular haemorrhage (grades III and IV)	High‐risk population		RR: 0.60 (95% CI 0.43 to 0.85)	1460 (8)	⊕⊕⊕⊝ moderate	Bias:The intervention was not blinded in the largest study, Song 2016 (n= 743). That study carried a weight of 72.8% in the analysis and was the only individual study that showed a significant reduction in IVH (grades III and IV). We downgraded the quality of the evidence by 1 step. Heterogeneity/Consistency: Heterogeneity was low (I2 = 45%). Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1460), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot, which included 8 studies, was symmetrical.
	111 per 1000	67 per 1000 (0 to 126)
Periventricular leukomalacia	High‐risk population		RR: 0.66 (95% CI 0.48 to 0.92)	1469 (6)	⊕⊕⊕⊝ moderate	Bias: The intervention was not blinded in the largest study, Song 2016 (n = 743). That study carried a weight of 89.2% in the analysis and was the only individual study that showed a significant reduction in PVL. We downgraded the quality of the evidence by 1 step. Heterogeneity/Consistency: We noted no heterogeneity (I2 = 5%). Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1469), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot, which included 6 studies, was symmetrical.
	111 per 1000	71 per 1000 (0 to 150)
Survivors at discharge from hospital without severe IVH, PVL, ROP	High‐risk population		RR: 1.00 (95% CI 0.93 to 1.08)	443 (1)	⊕⊕⊕⊕ high	Bias: We noted low risk of bias. Heterogeneity/Consistency: N/A, as only 1 study. Directness of evidence: The study was conducted in the target population. Precision: Because of the relatively large sample size (n = 443), the point estimate was precise with a narrow 95% CI. Presence of publication bias: As only 1 study was included, we did not develop a funnel plot.
	855 per 1000	856 per 1000
Time to achieve full enteral feeding (days)	Mean time to achieve full enteral feeding was 16.3 days (SD 5.3) in the control group.	Mean time to achieve full enteral feeding in the intervention groups was 2.90 days shorter.	MD: ‐2.90 (95% CI ‐5.77 to ‐0.03)	50 (1)	⊕⊕⊝⊝ low	Bias: We had concerns about blinding of the intervention and outcome assessments. We downgraded the quality of evidence by 1 step. Heterogeneity/Consistency: N/A, as only 1 study. Directness of evidence: The study was conducted in the target population. Precision: Because of the small sample size (n = 50), the 95% CI around the point estimate was wide. Presence of publication bias: As only 1 study was included, we did not prepare a funnel plot.
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; EPO: erythropoietin; IVH: intraventricular haemorrhage; MD: mean difference; N/A: not applicable; NICU: neonatal intensive care unit; PVL: periventricular leukomalacia; RD: risk difference; ROP: retinopathy of prematurity; RR: risk ratio; SD: standard deviation.
GRADE Working Group grades of evidence. High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.