Late erythropoiesis‐stimulating agents to prevent red blood cell transfusion in preterm or low birth weight infants

Sanjay M Aher; Arne Ohlsson<sup>a</sup>

doi:10.1002/14651858.CD004868.pub6

Agentes estimulantes de la eritropoyesis tardía para prevenir la transfusión de glóbulos rojos en recién nacidos prematuros o de bajo peso al nacer

Declaraciones de intereses de los autores

Versión publicada: 28 enero 2020 Historial de versiones

https://doi.org/10.1002/14651858.CD004868.pub6

Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

Los lactantes prematuros tienen niveles plasmáticos bajos de eritropoyetina (EPO), lo que justifica el uso de agentes estimulantes de la eritropoyesis (AEE) para prevenir o tratar la anemia. La darbepoetina (Darbe) y la EPO son AEE actualmente disponibles.

Objetivos

Evaluar la efectividad y la seguridad de la iniciación tardía de los AEE, entre los ocho y los 28 días después del nacimiento, para reducir el uso de las transfusiones de glóbulos rojos (GR) en los lactantes prematuros o de bajo peso al nacer.

Métodos de búsqueda

Se utilizó la estrategia de búsqueda estándar del Grupo Cochrane de Neonatología en el Registro Cochrane Central de Ensayos Controlados (CENTRAL 2018, número 5), MEDLINE vía PubMed (1966 hasta 5 de junio de 2018), Embase (1980 hasta 5 de junio de 2018) y en CINAHL (1982 hasta 5 de junio de 2018). También se buscaron ensayos controlados aleatorizados y ensayos cuasialeatorizados en bases de datos de ensayos clínicos, actas de congresos y listas de referencias de los artículos recuperados.

Criterios de selección

Ensayos controlados aleatorizados o cuasialeatorizados de iniciación tardía del tratamiento con EPO (iniciado a los ocho días o más de vida) versus placebo o ninguna intervención en neonatos prematuros (< 37 semanas) o de bajo peso al nacer (< 2500 g).

Obtención y análisis de los datos

La recopilación y el análisis de los datos se realizaron según los métodos del Grupo de Revisión Cochrane de Neonatología (Cochrane Neonatal Review Group). Se utilizó el enfoque GRADE para evaluar la calidad de la evidencia.

Resultados principales

Se incluyeron 31 estudios (32 comparaciones) que asignaron al azar a 1651 lactantes prematuros. Las búsquedas en la literatura en 2018 identificaron un nuevo estudio para su inclusión. No se identificaron nuevos ensayos en curso y ningún estudio utilizó darbepoetina.

La mayoría de los ensayos incluidos tuvo un tamaño de muestra pequeño. El metanálisis mostró un efecto significativo sobre el uso de una o más transfusiones de glóbulos rojos (21 estudios (n = 1202); riesgo relativo típico (RR) 0,72, intervalo de confianza (IC) del 95%: 0,65 a 0,79; diferencia de riesgos típica (DR) ‐0,17, IC del 95%: ‐0,22 a ‐0,12; número típico necesario a tratar para un resultado beneficioso adicional (NNTB) 6, IC del 95%: 5 a 8). Hubo una heterogeneidad moderada para este resultado (RR I² = 66%; RD I² = 58%). La calidad de la evidencia fue muy baja. Se obtuvieron resultados similares en los análisis secundarios basados en diferentes combinaciones de dosis altas o bajas de EPO y la administración de suplementos de hierro. No hubo una reducción significativa del volumen total (mL/kg) de sangre transfundida por lactante (diferencia media típica (DM) ‐1,6 mL/kg, IC del 95%: ‐5,8 a 2,6); cinco estudios, 197 lactantes). Hubo heterogeneidad alta en este resultado (I² = 92%). Hubo una reducción significativa en el número de transfusiones por lactante (11 estudios que reclutaron a 817 lactantes; DM típica ‐0,22; IC del 95%: ‐0,38 a ‐0,06). Hubo heterogeneidad alta en este resultado (I² = 94%).

Tres estudios con 404 lactantes informaron retinopatía del prematuro (RP) (todos los estadios), con un RR típico de 1,27 (IC del 95%: 0,99; 1,64) y una DR típica de 0,09 (IC del 95%: ‐0,00; 0,18). Hubo heterogeneidad alta para en ambos resultados RR (I² = 83%) y DR (I² = 82%). La calidad de la evidencia fue muy baja. Tres ensayos que reclutaron a 442 lactantes informaron sobre la RP (etapa ≥ 3). El RR típico fue 1,73 (IC del 95%: 0,92 a 3,24) y la DR típica fue 0,05 (IC del 95%: ‐0,01 a 0,10). No hubo heterogeneidad para este resultado para el RR (I² = 18%) pero sí una alta heterogeneidad para el RD (I² = 79%). La calidad de la evidencia fue muy baja.No hubo diferencias significativas en otros resultados clínicos, como la mortalidad y la enterocolitis necrotizante. En cuanto a los resultados de la mortalidad y la enterocolitis necrotizante, la calidad de la evidencia fue moderada. No se reportaron resultados a largo plazo en cuanto a neurodesarrollo.

Conclusiones de los autores

La administración tardía de EPO reduce el uso de una o más transfusiones de eritrocitos y el número de transfusiones de eritrocitos por lactante (< una transfusión por lactante), pero no el volumen total (ml/kg) de eritrocitos transfundidos por lactante. Es probable que no se evite la exposición a los donantes, ya que la mayoría de los estudios incluían a neonatos que habían recibido transfusiones de eritrocitos antes del ingreso al ensayo. La EPO tardía no reduce ni aumenta significativamente ningún resultado adverso clínicamente importante salvo una tendencia hacia un aumento del riesgo de RP. No se indica la investigación adicional del uso del tratamiento tardío de EPO para prevenir la exposición a donantes. Los esfuerzos de investigación deben centrarse en la limitación de la exposición a los donantes durante los primeros días de vida en los neonatos enfermos, cuando los requisitos de eritrocitos son más probables y no puede evitarse mediante el tratamiento tardío con EPO. La administración de paquetes satélite (que dividen una unidad de sangre de donante en muchas alícuotas más pequeñas) puede reducir la exposición a donantes.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Resumen en términos sencillos

disponible en

Agentes estimulantes de la eritropoyesis tardía para prevenir la transfusión de glóbulos rojos en recién nacidos prematuros o de bajo peso al nacer

Pregunta de la revisión

Se revisó la evidencia sobre la efectividad y la seguridad del inicio tardío del tratamiento con agentes estimulantes de la eritropoyesis entre los ocho y los 28 días después del nacimiento, para reducir el uso de las transfusiones de glóbulos rojos (GR) en los recién nacidos prematuros o de bajo peso al nacer.

Antecedentes

La cantidad de eritrocitos circulantes (hematocrito) desciende después del nacimiento en todos los neonatos. Esto es particularmente cierto en neonatos prematuros debido a su respuesta deficiente a la anemia y a la cantidad de sangre que debe extraerse para la realización de pruebas. Los bajos niveles plasmáticos de eritropoyetina (EPO), una sustancia en la sangre que estimula la producción de glóbulos rojos en los lactantes prematuros, justifican el uso de agentes estimulantes de la eritropoyesis (EPO y darbepoetina) para prevenir o tratar la anemia.

Fecha de la búsqueda

La evidencia está actualizada hasta el 5 de junio de 2018.

Características de los estudios

Hasta la fecha, 1651 lactantes (entre ocho y 28 días de edad) prematuros se han reclutado en 31 estudios de administración tardía de EPO para reducir el uso de las transfusiones de eritrocitos y evitar la exposición a donantes.

No hubo estudios que usaran darbepoetina.

Fuentes de financiación de los estudios

No se ha recibido ninguna financiación para esta revisión y no existe ningún conflicto de intereses que declarar.

Resultados clave

El riesgo de recibir una transfusión de eritrocitos se reduce después del inicio del tratamiento con EPO. Sin embargo, el beneficio general de la EPO se reduce, ya que muchos de estos neonatos ya habían estado expuestos a transfusiones de sangre antes del ingreso a los ensayos. El tratamiento con EPO tardía no tuvo efectos importantes sobre la muerte o las complicaciones frecuentes del parto prematuro, excepto las tendencias hacia un aumento del riesgo de retinopatía del prematuro. La retinopatía del prematuro es una enfermedad del ojo que afecta a los recién nacidos prematuros. Se considera que la causa el crecimiento desorganizado de los vasos sanguíneos retinianos, lo que puede dar lugar a cicatrización y desprendimiento de la retina. La retinopatía del prematuro puede ser leve y resolverse espontáneamente, pero puede dar lugar a ceguera en los casos graves.

Calidad de la evidencia

La calidad de los estudios fue variable y a menudo faltó información importante con respecto a la ocultación de la asignación. Los tamaños de la muestra fueron pequeños y no se informaron resultados a largo plazo (18 a 24 meses de edad corregida). La calidad de la evidencia fue muy baja en el caso de los resultados del "uso de una o más transfusiones de glóbulos rojos", la "retinopatía del prematuro (no se han informado todas las etapas o estadio)" y la "retinopatía del prematuro (estadio ≥ 3)". En cuanto a los resultados de "enterocolitis necrotizante" y "mortalidad", la calidad de la evidencia fue moderada.

Authors' conclusions

Implications for practice

Late EPO administration results in a reduction in the use of one or more red blood cell transfusions following initiation of therapy. It minimally reduces the number of red blood cell transfusions per infant. It is not associated with reductions in mortality or other neonatal morbidities. The use of late EPO is not associated with any short‐term serious side effects except for a possible association with retinopathy of prematurity (ROP) stage 3 or higher. A large proportion of extremely low BW/preterm neonates require red blood cell transfusions during the first few days of life, when neither early nor late EPO administration could possibly have an impact. The decision to use late EPO will depend on the baseline rate of red blood cell transfusions in this population in a specific neonatal intensive care unit, the costs, the associated pain, and the values assigned to the clinical outcomes. Other means of reducing the need for red blood cell transfusions should be considered, including reduced blood sampling, and the use of 'satellite packs' from directed or universal donors.

Implications for research

There is no need for further research to assess the effectiveness of the late use of EPO in reducing red blood cell transfusions. Its effectiveness has been established in populations that were exposed to donor blood prior to study entry, minimising the clinical importance of this effect. Future research should focus on strategies to minimise red blood cell donor exposure (using multiple aliquots from a properly tested single donor) during the first week of life, when the likelihood of need for red blood cell transfusions is at its peak. Such strategies, in combination with late EPO treatment, may reduce further donor exposure in early infancy. All ongoing and planned studies should monitor the incidence of retinopathy of prematurity.

Summary of findings

Open in table viewer

Summary of findings for the main comparison.

Late EPO compared with placebo or no intervention for complications of preterm birth ‐ primary and secondary outcomes
Patient or population: preterm infants with low birth weight Settings: NICU Intervention: Late EPO Comparison: Placebo or no intervention
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 intervention	EPO
Use of one or more red blood cell transfusions (low and high dose of EPO) during initial hospital stay	High risk population		RR 0.72 (0.65 to 0.79)	1202 (21)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in 16 of the studies, about allocation concealment in 17 of the studies and about blinding (performance bias and detection bias) in 11 of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 66% and for the typical RD 58% (both moderate heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1202), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot was asymmetrical ‐ we downgraded the quality of the evidence by one step.
	603 per 1000	434 per 1000 (392 to 477)
Necrotising Enterocolitis ≥ Bell's stage 2 during initial hospital stay	High risk population		RR 0.88 (0.45 to 1.70)	656 (6)	⊕⊕⊕⊝ moderate	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in three of the studies and about blinding (performance bias and detection bias) in one of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for both the typical RR and for the typical RD was 0% (both no heterogeneity). Directness of evidence: Studies were conducted in the target population. Precision: Because of the sample size (n = 656), the point estimate was quite precise with a narrow 95% CI. Presence of publication bias: As only six studies were included in the analyses we did not perform a funnel plot.
	52 per 1000	46 per 1000 (23 to 88)
Mortality during initial hospital stay (all causes)	High risk population		RR 0.82 (0.49 to 1.39)	767 (14)	⊕⊕⊕⊝ moderate	Bias: We had concerns about random sequence generation in 11 of the studies, about allocation concealment in 7 of the studies and about blinding (performance bias and detection bias) in six of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR and for the RD was 0% (both no heterogeneity). Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1202), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot was symmetrical.
	63 per 1000	52 per 1000 (31 to 88)
Retinopathy of prematurity (all stages or stage not reported) during initial hospital stay	High risk population		RR 1.27 (0.99 to 1.64)	404 (3)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in none of the studies, and about blinding (performance bias and detection bias) in two of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 83% and for the typical RD 82% (both high heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the small sample size (n = 404 ), the point estimate was not precise. We downgraded the quality of the evidence by one step. Presence of publication bias: As there was only three studies included we did not perform a funnel plot.
	328 per 1000	417 per 1000 (325 to 538)
Retinopathy of prematurity (stage ≥ 3) during initial hospital stay	High risk population		RR 1.73 (0.92 to 3.24)	442 (3)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in none of the studies, and about blinding (performance bias and detection bias) in two of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 18% (no heterogeneity) and for the typical RD 79% (high heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the small sample size (n = 442 ), the point estimate was not precise. We downgraded the quality of the evidence by one step. Presence of publication bias: As there was only three studies included we did not perform a funnel plot.
	63 per 1000	109 per 1000 (58 to 203)
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; I² = I‐squared; NICU: Neonatal Intensive Care Unit; RR: Risk Ratio;
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.

Background

Description of the condition

After birth, the haemoglobin (a hemoprotein composed of globin and heme that gives red blood cells heir characteristic colour; function primarily to transport oxygen from the lungs to the body tissues) concentration of newborn infants falls to minimal levels of 11 g/dL in term infants by eight to 12 weeks of age, and to 7.0 to 10.0 g/dL in preterm infants by six weeks of age (Stockman 1978). This process is called physiologic anaemia of infancy (Strauss 1986). In very low birth weight (VLBW) infants, the haematocrit (the ratio of the volume of red blood cells to the total volume of blood) falls to approximately 24% in infants weighing 1.0 to 1.5 kg and to 21% in infants weighing less than 1.0 kg at birth (Stockman 1986). In extremely low birth weight (ELBW) infants, this decline in haematocrit is not 'physiologic', as it may be associated with clinical findings that prompt red blood cell transfusions. To our knowledge, the diagnostic accuracy of different clinical signs and laboratory findings has not been studied (Cohen 1998). It is still unknown how low haematocrit levels can fall before clinical signs of anaemia occur and what the minimum haematocrit level is acceptable in infants requiring supplemental oxygen (Ohls 2002). Nevertheless, 'top‐up' transfusions to treat low haemoglobin or low haematocrit levels are frequently used. As many as 80% of VLBW infants and 95% of ELBW infants receive blood transfusions during their hospitalisations (Widness 1996).

Description of the intervention

Erythropoiesis‐stimulating agents (ESAs) currently include erythropoietin (EPO) and darbepoetin (Darbe). Plasma erythropoietin levels in neonates are lower than those of older children and adults. Brown 1983 reported that, between two and 30 days of life, the mean EPO concentration was 10 mU/mL, as compared to 15 mU/mL in concurrently studied adults. A low plasma EPO level is a key reason that nadir haematocrit values of preterm infants are lower than those of term infants (Dallman 1981; Stockman 1986). Low plasma EPO levels provide a rationale for the use of EPO in the prevention or treatment of anaemia of prematurity. Studies in newborn monkeys and sheep have demonstrated that neonates have a larger volume of distribution and more rapid elimination of EPO, necessitating the use of higher doses than are required for adults (Ohls 2000). A systematic review of EPO administration noted a wide range of doses used, from 90 to 1400 IU/kg/week (Kotto‐Kome 2004). Side effects reported in adults include hypertension, bone pain, rash, and rarely, seizures. Only transient neutropenia (low concentration of neutrophils, a type of white blood cell) has been reported in neonates (Ohls 2000).

How the intervention might work

The primary goal of ESAs therapy is to reduce transfusions. Most transfusions are given during the first three to four weeks of life. The larger or stable preterm infants who respond best to EPO receive few transfusions. ELBW infants, who are sick and have the greatest need for red blood cell (RBC) transfusions shortly after birth, have not consistently responded to EPO. This suggests that EPO is a more effective erythropoietic stimulator in more mature neonates. ELBW neonates are more likely to need transfusions even if their erythropoiesis is stimulated (Kotto‐Kome 2004). In addition, ELBW neonates have a smaller blood volume and the relatively larger phlebotomy (phlebotomy ‐ the drawing of blood) volumes that are required during hospital stay often necessitate 'early' transfusions. In contrast, 'late' transfusions are more often given because of anaemia of prematurity (Garcia 2002). Most preterm infants who require blood transfusions will receive their first transfusion in the first two weeks of life (Zipursky 2000). Reducing the number of RBC transfusions reduces the risk of transmission of viral infections and may reduce costs. Frequent RBC transfusions may be associated with retinopathy of prematurity (ROP) (abnormal blood vessel development in the retina of the eye) (Hesse 1997) and bronchopulmonary dysplasia (a chronic lung disease that affects newborns and infants) (BPD).

In preterm infants, iron is needed for erythropoiesis. As neonatal blood volume expands with rapid growth, infants produce large amounts of haemoglobin. Several studies have observed a decrease in serum ferritin concentration (an indication of iron deficiency (Finch 1982) during erythropoietin treatment). The use of higher, more effective doses of erythropoietin might be expected to increase iron demand and the risk of iron deficiency. Iron supplementation during erythropoietin treatment has been observed to reduce the risk of the development of iron deficiency (Shannon 1995). The range of iron doses used in EPO‐treated infants is between 1 mg/kg/day to 10 mg/kg/day (Kotto‐Kome 2004).

EPO has been found to have important non‐haematopoietic (formation of blood cellular components) functions in the brain and other organs during development (Juul 2002). Administration of EPO could potentially have a neuro‐protective effect in preterm infants, especially in perinatal asphyxia (Dame 2001; Juul 2002). This aspect of EPO use in neonates will be systematically reviewed separately (Yu 2010).

A slightly modified, long‐acting version of EPO, darbepoetin alfa (Darbe), has been introduced (Egrie 2001). Darbepoetin was created by modifying five amino acids of the original EPO protein to generate two additional carbohydrate‐binding sites, thereby significantly increasing circulating half‐life and effectiveness. Compared with EPO, darbepoetin has an approximately three‐fold longer serum half‐life and greater in vivo potency. It can be administered less frequently to obtain the same biological response. A single subcutaneous dose of Darbe has been shown to accelerate erythropoiesis in preterm infants (Warwood 2005).

It is likely that additional studies of ESAs in preterm or LBW infants have been published since the reviews noted above, which included reports up to October 2002. We have performed a series of Cochrane reviews on the use of EPO in preterm infants, including: Early erythropoiesis‐stimulating agents in preterm or low birth weight infants (Ohlsson 2017), this review (late erythropoietin; Aher 2012a; Aher 2014), and Early versus late erythropoietin to prevent red blood cell transfusion in preterm and/or low birth weight infants (Aher 2012b). The cut‐off of seven days of age or less for 'early' and more than seven days for 'late' treatment with EPO, although somewhat arbitrary, was chosen based on previously published meta‐analyses (Garcia 2002; Kotto‐Kome 2004), and allowed us to compare the results between our reviews and previously published reviews. It is of note that in the latest update of the Early erythropoiesis‐stimulating agents in preterm or low birth weight infants (Ohlsson 2017) review, there was no significant difference in the incidence of ROP with the treatment with ESAs compared to controls.

Why it is important to do this review

This review concerns late administration of ESAs (starting in infants between eight and 28 days of age, after reaching 40 weeks' postmenstrual age (PMA)). The main rationale for such EPO therapy is to avoid exposure of neonates to multiple blood donors and the risks associated with this exposure. It is likely that many sick neonates would have received a transfusion prior to entry into trials enrolling infants more than seven days of age. We carried out a systematic review to evaluate all available studies where ESAs were begun after seven days of life, to assess the effect on the use of one or more red blood cell transfusions in preterm or very low birth weight infants.

Objectives

Primary objective

To assess the effectiveness and safety of late initiation of ESAs between eight and 28 days after birth in reducing red blood cell (RBC) transfusions in preterm or low birth weight infants.

Secondary objective

To distinguish between the effectiveness and safety of the intervention by subgroup analyses of low (≤ 500 IU/kg/week) and high (> 500 IU/kg/week) doses of EPO, and low (≤ 5 mg/kg/day) and high (> 5 mg/kg/day) doses of supplemental iron in reducing red blood cell transfusions in preterm or low birth weight infants.

Métodos

disponible en

Obtención y análisis de los datos

Evaluación del riesgo de sesgo de los estudios incluidos

Results

Description of studies

For the previous feedback‐initiated update, the study by Romagnoli (Romagnoli 2000) was moved from the early EPO review (Ohlsson 2012) to this late EPO review as the infants were of a mean (SD) age of 10 (±1) days at the initiation of EPO treatment. The study enrolled 230 infants. The 2012 update identified one new study for inclusion as did this update in 2018. As a result, 31 studies (32 comparisons) including 1651 preterm or low birth weight infants met the inclusion criteria. These studies were performed in 22 countries (Argentina, Australia, Austria, Belgium, Brazil, Canada, Finland, France, Germany, Greece, Iran, Israel, Italy, Japan, Norway, South Africa, Spain, Switzerland, Taiwan (Republic of China), the UK, Turkey, the USA). We decided to include studies that enrolled some infants who were less than eight days at study entry if most of the infants enrolled were older at entry. We made one further deviation from our protocol, as we included studies that enrolled infants beyond 28 days of age. Most of these studies enrolled some infants who were less than 28 days old but the inclusion criteria did not have 28 days of postnatal age as an upper limit. We could not separate out data for infants that were under 28 days at enrolment. The inclusion of these studies makes our review more comprehensive. The age at enrolment is stated for each study in the table Characteristics of included studies.

Results of the search

The study flow diagram shows the results of the searches (Figure 1). Only one new study (Basiri 2015) was identified for this update. No new ongoing trial was identified. No study investigated Darbe.

Figure 1

Study flow diagram: review update

Included studies

We included 31 studies (32 comparisons) in this review. We treated the report by Giannakopoulou as two separate studies; under Giannakopoulou 1998a, we reported on the 32 infants weighing < 1000 grams and under Giannakopoulou 1998b, we reported on the 36 infants weighing 1000 to 1300 grams. The studies are detailed in the table Characteristics of included studies and are briefly discussed below.

The detailed guidelines, used for transfusions, are outlined in the Additional Table (Table 1: Transfusion Guidelines).

Open in table viewer

Table 1. Transfusion guidelines

Reference	Indications
Akisu 2001	Guidelines for transfusions were not presented.
Al‐Kharfy 1996	The indications for transfusion were 1) shock, 2) cumulative loss of ≥ 10% of the blood volume in 72 hours or less when further blood sampling is expected, 3) Hb < 130 g/L in acutely ill neonates with cardiorespiratory disease, and 4) Hb < 80 to 100 g/L with clinical signs of anaemia. A volume of 15 mL/kg was recommended for each transfusion.
Atasay 2002	Criteria for blood transfusion (10 mL/kg packed red cells) were as follows: a Hct < 30% when signs and symptoms attributed to anaemia including persistent tachycardia (180 beats/min for 24 hours), frequent apnoea with bradycardia and daily weight gain < 10 g/kg despite optimal protein and caloric intake (3.5 g/kg, 100 kcal/kg/day). Infants were transfused with a Hct of 35% to 40% if they received more than 40% oxygen or ventilation therapy.
Bader 1996	Criteria for blood transfusion (10 mL/kg of red cells) were as follows: a) a Hct < 25%, b) an increased frequency of apnoeic events which required either stimulation or aminophylline therapy, c) changes in heart rate patterns, e.g. an increase in frequency of bradycardia (< 80 beats/min) or tachycardia (> 180 beats/min, d) failure of weight gain of > 10 g/kg/day despite an optimal caloric intake of > 120 kcal/day and e) lethargy without evidence of sepsis.
Basiri 2015	Transfusion guidelines for preterm infants included: (1) asymptomatic infants with Hct < 21% and reticulocytes < 2%; (2) infants with haematocrit < 31% and hood O₂ < 36% or mean airway pressure < 6 cm H₂O by continuous positive airway pressure (CPAP) or intermittent mandatory ventilation (IMV) or > 9 apnoeic and bradycardic episodes per 12 hours or 2/24 hours requiring bag‑and‑mask ventilation while on adequate methylxanthine therapy or heart rate > 18/min or respiratory rate > 80/min sustained for 24 hours or weight gain of < 10 g/day for 4 days on 100 kcal/kg/day or having surgery; (3) infants with haematocrit < 36% and requiring > 35% O₂ or mean airway pressure of 6 to 8 cm H₂O by CPAP or IMV.
Bechensteen 1993	Indications of blood transfusions were: 1) Hb < 80 g/L or 2) otherwise at the discretion of the clinician caring for the infant according to symptoms and signs.
Bierer 2009	Very specific transfusion criteria were in place for infants on mechanical ventilation, for infants who required supplemental oxygen and for infants on room air.
Chen 1995	Transfusions were given because of frequent and prolonged apnoea.
Corona 1998	Transfusions were considered based on the clinical condition (pallor, tachycardia, tachypnoea, apnoea with or without bradycardia, poor weight gain, difficulties with sucking) and haematological parameters (Hb < 70 g/L, Hct < 26%, with low reticulocyte counts).
Donato 1996	During the first week of life, infants were given transfusions of packed red blood cells for replacement when blood drawn for analysis was in excess of 8 mL/kg of body weight; fresh whole blood was given if signs attributable to hypovolaemia or anaemia developed. Subsequently, infants with heart rate > 180 beats/min, severe apnoea/bradycardia or poor weight gain (< 10 grams/day in spite of a 100 calories/day intake during 5 consecutive days) were transfused if the Hct was < 25% (or < 30% if oxygen or mechanical ventilation was required); asymptomatic infants were transfused only when a central Hct < 23% was reached.
Emmerson 1993	The decision to give a blood transfusion to a study infant was made by the medical staff of the neonatal unit who were blinded to the randomisation. The unit policy at the time of the study was to transfuse a preterm infant who had a Hb < 100 g/L and who had symptoms consistent with those caused by anaemia. The symptoms and signs of anaemia included poor feeding, tachycardia, tachypnoea, apnoea, and pallor. Infants with a Hb < 80 g/L were transfused even if asymptomatic.
Giannakopoulou 1998a; Giannakopoulou 1998b	Indications for blood transfusion were a Hb < 80 g/L or otherwise at the discretion of the physician treating the infants according to symptoms and signs.
Griffiths 1997	Infants were transfused if they were ventilated and/or oxygen‐dependent with a Hb of < 120 g/L, had clinically symptomatic anaemia, or were asymptomatic with a Hb of < 70 g/L. Infants were transfused if they were ventilated and/or oxygen‐dependent with a Hb of < 120 g/L, had clinically symptomatic anaemia, or were asymptomatic with a haemoglobin of < 70 g/L.
Javier Manchon 1997	Transfusion guidelines were similar in all three centres (details not provided).
Juul 2003	By NICU policy, on admission, infants weighing < 1000 grams at birth were assigned 1 unit of packed red blood cells divided into 8 aliquots. These aliquots were used for transfusions during the first month of life. The following transfusion guideline was used for infants of all birth weights: Transfusion is recommended for a Hct < 35% if the infant requires positive pressure with a mean airway pressure > 6 cm H₂O and requires > 35% oxygen. Transfusion is recommended for Hct < 30% if the infant requires oxygen (< 35% FIO₂), is receiving continuous positive airway pressure or intubated with mean airway pressure < 6 cm H₂O, if an infant has significant apnoea and bradycardia while receiving methylxanthines (> 9 episodes in 12 hours or 2 episodes in 24 hours requiring mask‐and‐bag ventilation), if the heart rate is > 180 beats/min or respiratory rate > 80/min and persists for 24 hours, if weight gain < 10 grams/day over 4 days despite adequate calories, or in the presence of sepsis. If the Hct is < 20%, no symptoms are necessary for transfusion. Transfusion volumes were standardised as follows: infants received an initial transfusion of 15 mL/kg over a period of 3 to 4 hours. A follow‐up Hct was checked after 4 hours. If the Hct was < 30%, a second aliquot of 10 mL/kg was given If the Hct was between 30% and 35%, an additional 5 mL/kg was given.
Kivivuori 1999	The Hct values were maintained at > 30% by red blood cell transfusions (10 mL/kg per time) in asymptomatic infants. In infants who had symptoms or signs of anaemia, red blood cells were transfused if the Hct value was < 40%. The transfusion policies were the same in all study hospitals.
Kumar 1998	The need for erythrocyte transfusion was assessed by the clinicians caring for each infant and the decision to transfuse was made without consulting the study investigators. According to the practice in the NICU, infants received transfusion if a Hct level of < 27% was associated with one of the following signs and symptoms of anaemia: 1) frequent apnoea and bradycardia, defined as > 6 episodes in 12 hours or any episode requiring bag‐and‐mask ventilation, in an infant with therapeutic serum levels of theophylline; 2) persistent tachycardia, defined as > 180 beats/min for more than 12 hours; 3) poor weight gain (< 10 grams/day averaged over a 7‐day period) despite adequate caloric intake; and 4) increasing oxygen requirement in infants with chronic lung disease despite optimum diuretic and bronchodilator therapy.
Maier 2002	Infants with artificial ventilation or in > 40% of inspired oxygen were not transfused unless Hct dropped to < 40%. Spontaneously breathing infants were not transfused unless Hct dropped to < 35% during the first 2 weeks of life, 30% during the 3rd to 4th weeks, and 25% thereafter. Transfusion was allowed when life‐threatening anaemia or hypovolaemia was assumed by the treating neonatologist, or surgery was planned. Twelve of the 14 centres used satellite packs of the original red cell pack to reduce donor exposure.
Meyer 1994	Direct quote: "The need for blood transfusion was assessed by the attending neonatal physician and decisions were made independently of the investigators. The following guidelines were developed, based on existing literature and nursery practices: a. Hct < 30% and 1) Weight gain of < 10 grams/day averaged over 1‐week period (infant tolerating full oral feeds and receiving adequate calories). 2) Three or more episodes of apnoea (respirations absent for 20 seconds) or bradycardia (heart rate of < 100 beats per minute) in a 24‐hour period not due to other causes and not responsive to methylxanthine treatment. 3) Tachycardia (> 170 beats/min) or tachypnoea (> 70 breaths/min) sustained over a 24‐hour period or associated with acute cardiac decompression. 4) A requirement for surgery. b. Development of a clinically significant patent ductus arteriosus (i.e. at least three of the following features: heart rate > 160 beats/min, brisk brachial and/or dorsalis pedis pulses, palpable precordial pulsation, systolic murmur, cardiomegaly on chest radiograph). c. Pulmonary disease and fractional inspired oxygen concentrations increasing by > 10% per week. d. Systemic infection (either clinically suspected or proven on blood culture) associated with a sudden decrease in hematocrit. e. Hematocrit of 22% or less and an absolute reticulocyte count of < 100 000 x 10⁹/L".
Pollak 2001	Standard NICU transfusion criteria were used (authors refer to Shannon 1995; see below).
Reiter 2005	Conservative transfusion guidelines were in place and followed. Criteria for red blood cell transfusion in the acutely ill infant requiring mechanical ventilation or nasal continuous positive airway pressure included: phlebotomy loss of > 15% of blood volume associated with hypotension, or Hct < 30%. Criteria for red blood cell transfusion in the convalescent infant requiring no more than supplemental oxygen included: Hct < 28% with symptomatic anaemia (tachycardia, poor somatic growth, or metabolic acidosis) or Hct < 20%.
Rocha 2001	The decision for blood transfusion within the whole study population was made by the assistant doctor of each newborn, and was based on the following criteria: Hct ≤ 20%, inadequate weight gain, 3 or more apnoea or bradycardia episodes within 24 hours, presurgical procedure requirement, disease associated with sudden Hct decline, restoration of the blood collected for lab exams, maintenance of Hct up to 30% associated with minimal ventilatory support requirement, and Hct up to 35% when ventilation requirements are greater. The assistant doctor who recommended blood transfusion did not know to which group the infant belonged.
Romagnoli 2000	Infants on mechanical ventilation and/or on more than 30% of inspired oxygen received packed erythrocytes when their Hct levels dropped below 40%. Otherwise the transfusion was performed when the Hct fell below 35% from the 2nd to the 4th week of life and below 23% thereafter.
Ronnestad 1995	Transfusions were given on the orders of the attending physician if Hb was < 90 g/L or otherwise as necessary according to signs and symptoms.
Samanci 1996	The need for packed erythrocyte transfusions was judged by the attending neonatologist. Guidelines for erythrocyte transfusions were developed as follows: 1) Hct of ≤ 22% and an absolute reticulocyte count of 100 000/microlitre; 2) Hct of ≤ 30% and a) tachycardia (> 180 beats/min) and tachypnoea (> 70 breaths/min) persisting for 24 hours; or b) 3 or more episodes of apnoea or bradycardia in 24 hours, not due to other causes and not responsive to methylxanthine treatment; or c) average weight gain of < 10 grams/day over a 1‐week period (infant tolerating full oral feed and receiving adequate calories); or d) undergoing surgery; 3) systemic infection associated with a sudden decrease in Hct.
Shannon 1991	Transfusions were ordered by the clinicians caring for each infant without consulting the investigators. Written guidelines for erythrocyte transfusions were developed for the nursery 1 year before the start of the study. A copy of these guidelines was taped to the bed of each study infant. In general, babies who were otherwise well received transfusions only if they had a Hct < 25% and signs referable to their anaemia, such as slowing in rate of growth, persistent severe tachycardia and tachypnoea, or worsening of episodes of apnoea and bradycardia.
Shannon 1992	See Shannon 1991 above.
Shannon 1995	Transfuse infants at Hct ≤ 20%: a) if asymptomatic with reticulocytes < 100 000/microlitre. Transfuse infants at Hct ≤ 30%: a) if receiving < 35% supplemental hood oxygen, b) if on CPAP or mechanical ventilation with mean airway pressure < 6 cm H₂O, c) if significant apnoea and bradycardia are noted (9 episodes in 12 hours or 2 episodes in 24 hours requiring bag‐and‐mask ventilation) while receiving therapeutic doses of methylxanthines, d) if heart rate > 180 beats/min or respiratory rate > 80 breaths /min persists for 24 hours, e) if weight gain < 10 grams/day is observed over 4 days while receiving ≥ 100 kcal/kg/day, f) if undergoing surgery. Transfuse for Hct ≤ 35%: a) if receiving > 35% supplemental hood oxygen, b) if intubated on CPAP or mechanical ventilation with mean airway pressure ≥ 6 to 8 cm H₂O. Do not transfuse: a) to replace blood removed for laboratory tests alone, b) for low Hct alone.
Whitehall 1999	Guidelines for red cell transfusions for anaemia of prematurity were based on the existing policy in the nursery, generally adopted by the neonatologists. They were as follows: I. Transfuse infants at Hb 80 g/L, (a) If reticulocyte count is < 4% and (b) If receiving supplemental oxygen > 30% or (c) If unexplained recurrent apnoea/bradycardia is noted (> 1 to 2/hour) or (d) If persistent tachycardia (heart rate > 170/min) or tachypnoea (respiratory rate > 60/min) is noted, or (e) If there is failure to gain weight or successive weight loss on weekly recordings for 3 consecutive weeks. In absence of a clear evidence in the literature justifying red cell transfusions at a Hb of 80 g/L in otherwise asymptomatic neonates who are failing to thrive, it was decided that failure to gain weight or successive weight loss on weekly recordings for 3 consecutive weeks was a fair and substantial clinical indicator of the need to transfuse. II. Transfuse infants at Hb 100 g/L, (a) If receiving supplemental oxygen 30% and (b) needing intermittent mandatory ventilation or continuous positive airway pressure by nasal prongs for recurrent (> 1 to 2 per hour), apnoea/bradycardia with saturations < 90% on the pulse oximeter. III. Transfuse Infants at Hb 120 g/L, (a) If receiving mechanical ventilatory support with mean airway pressure 10 cm H₂O and supplemental oxygen 30% during the acute phase of illness after birth.
Yamada 1999a	Conservative red blood cell transfusion guidelines were followed (details not presented, as we were unable to translate the information).
Yamada 1999b	Conservative red blood cell transfusion guidelines were followed (details not presented, as we were unable to translate the information).

^{CPAP: continuous positive airways pressure
FIO2: fraction of inspired oxygen
H2O: water
Hb: haemoglobin
Hct: haematocrit}

^{IMV: intermittent mandatory ventilation
NICU: neonatal intensive care unit
O2: oxygen}

Akisu 2001 was a single centre study performed at University of Ege, Izmir, Turkey.

Objective: To evaluate the effect of EPO on lipid peroxidation and the activities of erythrocyte antioxidant enzymes in very low birth weight (VLBW) infants.
Population: Appropriately grown preterm infants with post menstrual age (PMA) < 33 weeks' gestational age and birth weight < 1500 grams.
Intervention: The EPO group received high doses of EPO from day 10 of life, totaling 750 IU/kg/week (high dose). Infants in the control group received no placebo. All infants received 3 mg/kg/day (low dose) of elemental iron.
Outcomes assessed: Use of one or more red blood cell transfusions.

Al‐Kharfy 1996 was a single centre study performed in Canada.

Objective: To determine whether treatment with EPO reduces transfusion requirements in preterm neonates at risk of having bronchopulmonary dysplasia (BDP) and requiring multiple transfusions.
Population: Appropriately grown preterm infants with birth weight (BW) < 1250 grams and having a 75% probability of having BPD determined on day 10 of life and postnatal age 10 to 17 days.
Intervention: The EPO group received EPO 200 IU/kg body weight, by subcutaneous (sc) injection, on Monday, Wednesday and Friday for six weeks (600 IU/kg/week; high dose). The control group received sham injections. Oral ferrous sulphate solution was administered to the EPO group at 6 mg of elemental iron/kg/day (high dose) and the control group received 2 mg of elemental iron/kg/day (low dose).
Outcomes assessed: Number of transfusions per infant, mortality, sepsis, retinopathy of prematurity (ROP) (stage ≥ 3), hypertension, BPD at 28 days of age.

Atasay 2002 was a single centre study performed in Turkey.

Objective: To investigate the effect of early EPO treatment on induction of erythropoiesis and the need for transfusion in VLBW infants with acute neonatal problems.
Population: Infants with BW < 1500 grams and PMA < 32 weeks' gestational age.
Intervention: The EPO group received EPO 600 IU/kg/week (high dose) sc, at seven to ten days and continued over seven to eight weeks. The control group received no EPO, placebo, or iron. The EPO group was supplemented with oral iron (ferro glycine sulphate) at the dose of 3 mg/kg/day (low dose).
Outcomes assessed: Use of one or more red blood cell transfusion(s).

Bader 1996 was a two‐centre study performed in Israel.

Objective: To assess whether an iron dose of 6 mg/kg/day is sufficient to maintain serum ferritin at adequate levels (as per authors).
Population: Preterm infants with PMA < 34 weeks' gestational age and BW < 1750 grams and postnatal age three to five weeks.
Intervention: The EPO group received EPO 300 IU/kg/day sc three times a week (900 IU/kg/week; high dose), for a total duration of four weeks. The control group received no placebo or other intervention. Two weeks into the study, elemental iron supplementation was begun in both groups at a dose of 6 mg/kg/day (high dose).
Outcomes assessed: Use of one or more red blood cell transfusions, side effects, sudden infant death syndrome (SIDS).

Basiri 2015 was a single centre study performed in Iran.

Objective: To evaluate the effectiveness of EPO treatment in preventing anaemia of prematurity.
Population: Preterm infants of 24 to 31 weeks' postmenstrual age.
Intervention: The study group 250 U rhEpo/kg BW (0.1 ml solution/kg BW) given sc 3 days a week every other day on Monday, Wednesday and Friday. (= 3 x 250 U/week = 750 IU/week; high EPO). The control group received equal volume normal saline 0.1 ml solution/kg BW. Infants in both groups were given oral ferrous sulphate, 4 mg/kg/day during the entire treatment period (4 mg/kg/day; low iron (< 5 mg/kg/day).
Outcomes assessed: Use of one or more red blood cell transfusion(s), unspecified side effects.

Bechensteen 1993 was a four‐centre study performed in Norway.

Objective: To determine whether VLBW infants respond to EPO with increased erythropoiesis.
Population: Preterm infants with BW 900 grams to 1400 grams at three weeks of age.
Intervention: The EPO group received EPO 100 IU/kg three times a week (300 IU/kg/week; low dose) sc from three to seven weeks of age. The control group received neither EPO nor placebo. Oral iron 18 mg/day (high dose) regardless of weight, began at the start of the study (three weeks). If serum iron concentration fell below 16 micromol/L, the dose was increased to 36 mg/day.
Outcomes assessed: Use of one or more red blood cell transfusions, mortality, hypertension, neutropenia, side effects.

Bierer 2009 was a single centre study performed in the USA.

Objective: To determine if EPO administration to neonates requiring surgery would stimulate erythropoiesis.
Population: Neonates requiring surgery. Post‐natal age < 28 days. The majority of infants were preterm and/or LBW.
Intervention: The EPO group received EPO 200 IU/kg per day or 400 IU/kg sc three times weekly (high dose). The placebo group received an equal volume of normal saline if the infant was on total parenteral nutrition and, if not, a sham dose was given sc and a BandAid was placed over the sham injection site. Infants received oral iron supplementation when enteral feeds reached 60 mL/kg/day.
Outcomes assessed: Volume transfused during hospitalisation (mL/kg), number of donors the infants was exposed to.

Chen 1995 was a single centre study performed in Taiwan, Republic of China.

Objective: To evaluate the safety and efficacy of EPO for the treatment of anaemia of prematurity.
Population: Preterm infants with PMA ≤ 33 weeks' gestational age and BW ≤ 1750 grams, haemoglobin (Hb) < 10 g/dL and haematocrit (Hct) < 30%.
Intervention: The EPO group (A) received 150 IU/kg iv twice a week (300 IU/kg/week; low dose); Group B received packed washed erythrocyte transfusion, when their Hb levels were < 10 g/dL with signs and symptoms attributed to anaemia or who had a Hb level < 8 g/dL even if asymptomatic; group C did not received EPO or erythrocyte transfusions (three infants excluded from total 19, as they received erythrocyte transfusion later because of frequent episodes of apnoea). All infants received oral elemental iron 3 mg/kg/day (low dose). Group A and C were included in this review.
Outcomes assessed: Mortality, adverse effects.

Corona 1998 was a single centre study performed in Italy.

Objective: To evaluate the efficacy of EPO, establish the optimal dose, the age at which to start, the duration of the treatment, any adverse effects, and the reduction in red blood cell transfusions.
Population: Preterm infants (BW < 1500 grams and < 33 weeks' postmenstrual age).
Intervention: EPO group A received EPO 150 IU/kg/week sc low dose ; EPO group B received 300 IU/kg/week sc low dose; the control group (group C) received no treatment. All groups received oral iron 4 mg/kg/day (low dose).
Outcomes assessed: Use of one or more red blood cell transfusions, total volume (mL/kg) of blood transfused per infant (means but no standard deviation (SD) provided), side effects.

Donato 1996 was a single centre study performed in Argentina.

Objective: To assess the efficacy of three different doses of recombinant human EPO (rHuEPO) to reduce the need for transfusion in preterm infants with BW < 1500 grams.
Population: Preterm infants with PMA < 34 weeks' gestational age and BW < 1500 grams.
Intervention: The placebo group (A) received human seroalbumin. The three EPO groups: Group B received EPO 50 IU/kg (150 IU/kg/week; low dose), Group C received EPO 100 IU/kg (300 IU/kg/week; low dose) and Group D received EPO 250 IU/kg (750 IU/kg/week; high dose) sc during eight consecutive weeks. All participants were given oral iron 6 mg/kg/day (high dose) and folic acid (2 mg/day) supplements, starting on day 15 of age and continuing during whole treatment period.
Outcomes assessed: Use of one or more red blood cell transfusions, average number of transfusions per infant during treatment, mortality, mean length of hospitalisation time, side effects, hypertension, SIDS.

Emmerson 1993 was a single centre study performed in the UK.

Objective: To investigate the safety and efficacy of EPO for the prevention of anaemia of prematurity.
Population: Infants between 27 and 33 weeks' postmenstrual age.
Intervention: The EPO group received low dose EPO (between 50 and 150 IU) twice a week from seven days of age and the placebo group received 4% albumin from seven days of age until discharge home. All infants received iron (6.25 mg) in the form of ferrous glycine sulphate from four weeks of age (high dose).
Outcomes assessed: Use of one or more red blood cell transfusions, volume transfused (mL/kg), mortality, hospital stay, SIDS, neutropenia.

Giannakopoulou 1998a and Giannakopoulou 1998b was a single centre study performed in Greece.

Objective: To stimulate erythrocyte production by the use of EPO and thereby decrease the requirement for red blood cell transfusions.
Population: Preterm infants with BWs ≤ 1300 grams, postnatal age > 20 days.
Intervention: The EPO group received EPO 300 IU/kg body weight, three times a week (900 IU/kg/week; high dose) from 20 days of age for six to eight weeks. The control group did not received any placebo. All infants received oral elemental iron 10 mg/kg/day (high dose).
Outcomes assessed: Mortality, side effects, hypertension, neutropenia.
We treated this report as two separate studies; Under Giannakopoulou 1998a we reported on the 32 infants weighing < 1000 g and under Giannakopoulou 1998b we reported on the 36 infants weighing 1000 to 1300 grams.

Griffiths 1997 was a study conducted in four neonatal intensive care units (NICU) in Yorkshire, England.

Objective: To evaluate the role of EPO in reducing iron supplementation, which may exacerbate free radical change, leading to lung disease.
Population: Preterm infants with postmenstrual age ≤ 32 weeks and/or BW ≤ 1500 grams, requirement for mechanical ventilation and/or supplemental oxygen at birth. Postnatal age seven to 14 days.
Intervention: The EPO group received 480 IU/kg/week (low dose) and the control group received placebo (4% human albumin) starting at seven to 14 days of age. All infants received oral iron (3.0 mg/kg/day) (low dose) from four weeks after birth.
Outcomes assessed: Mortality, BPD (at 36 weeks' postmenstrual age), number of blood transfusions per infant (medians provided), SIDS.

Javier Manchon 1997 was a multi‐centre study involving three centres in Barcelona, Spain.

Objective: To test the therapeutic effect of EPO on anaemia of prematurity.
Population: Preterm infants < 34 weeks' postmenstrual age, who at 28 days after birth had Hb levels < 10.5 g/dL.
Intervention: The EPO group received high dose EPO (200 IU/kg/day) three days a week for four weeks and ferrous sulphate 4 mg/kg/day (low dose). The control infants did not receive placebo, EPO, or iron.
Outcomes assessed: Use of one or more red blood cell transfusions between 30 and 60 days of age.

Juul 2003 was a single centre study performed in the USA.

Objective: To determine whether enterally dosed EPO stimulates erythropoiesis in preterm infants.
Population: Preterm infants with BW between 700 to 1500 grams and receiving at least 30 mL/kg per day of enteral feeding.
Intervention: The EPO group received 1000 IU/kg (enterally) per day divided into two daily feedings (7000 IU/kg/week; high dose), for 14 days. The placebo group received 5% dextrose in water for 14 days. All participants received supplemental iron (iron dextran, 1.0 mg/kg/day, or enteral ferrous sulphate 6 mg/kg/day (high dose)).
Outcomes assessed: Phlebotomy loss (mL) and packed red blood cell transfusion volume (mL). Evidence of feeding intolerance and other adverse effects.

Kivivuori 1999 was a four‐centre study performed in Helsinki and Espoo, Finland.

Objective: To compare oral and intramuscular routes of administration of iron in EPO treated VLBW infants.
Population: VLBW infants (birth weight ranged from 625 to 1470 grams).
Intervention: One EPO group received EPO 300 IU/kg sc three times/week, 900 IU/kg/week (high dose) sc and oral iron 6 mg/kg/day (high dose). Another EPO group received EPO 900 IU/kg/week and weekly intramuscular (im) iron 12 mg/kg (high dose). The control group received im iron 12 mg/kg/week but no EPO.
Outcomes assessed: Use of one or more red blood cell transfusions, adverse effects.

Kumar 1998 was a single centre study performed in the USA.

Objective: To evaluate the efficacy and safety of EPO in VLBW infants with anaemia of prematurity.
Population: Preterm infants (PMA < 32 weeks, BW < 1250 grams) with anaemia of prematurity.
Intervention: The EPO group received 300 IU/kg/dose of EPO sc twice a week (600 IU/kg/week; high dose) for six weeks. The control group received an identical volume of placebo suspension (normal saline). All infants received elemental iron 6 mg/kg/day (high dose).
Outcomes assessed: Use of one or more red blood cell transfusions, number of erythrocyte transfusions (per infant), entry to discharge duration (days), side effects.

Maier 2002 was a multicentre study performed in 14 centres in four European countries (Belgium, France, Germany, Switzerland).

Objective: To investigate whether EPO reduces the need for transfusions in extremely low birth weight (ELBW) infants and to determine the optimal dose of treatment.
Population: Preterm infants with BW 500 to 999 grams.
Intervention: The EPO group received EPO 200 IU/kg/dose three times a week sc (600 IU/kg/week, high dose). The volume was increased by the equivalent of 50 IU/kg per dose if the haematocrit declined by 6% during any two‐week period during the trial, but was withheld if the haematocrit was > 45%. The control group received an identical volume of placebo. Enteral iron 3 mg/kg/day (low dose) was given to all infants from days three to five and was increased at days 12 to 14 to 6 mg/kg/day (high dose) and to 9 mg/kg/day (high dose) at days 24 to 26 of life.
Outcomes assessed: Use of one or more red blood cell transfusions, donor exposure, mortality during hospital stay, necrotising enterocolitis (NEC), IVH, periventricular leukomalacia (PVL), ROP, days in oxygen, days in NICU, days in hospital.

Meyer 1994 was a single centre study performed in South Africa.

Objective: To assess the efficacy of EPO in the treatment of anaemia of prematurity.
Population: Preterm infants (≤ 32 weeks' postmenstrual age, weight ≤ 1500 grams, postnatal age two to eight weeks, central haematocrit Hct ≤ 35%).
Intervention: The EPO group received high dose EPO (200 IU/kg three times a week; 600 IU/kg/week). The control group received an identical volume of placebo. All infants received 3 mg/kg/day of iron (low dose).
Outcomes assessed: Use of one or more red blood cell transfusions, sepsis, NEC, SIDS.

Pollak 2001 was a single centre study conducted in Vienna, Austria.

Objective: To test the efficacy and safety of combining intravenous iron in amounts approximately the in‐utero accretion rate and the postnatal iron loss with EPO in VLBW infants.
Population: Preterm infants < 31 weeks' postmenstrual age and < 1300 grams birth weight not treated with red blood cell transfusions during the study period.
Intervention: During a three‐day run‐in baseline period, 9 mg/kg/day of iron poly maltose complex (high dose) was administered to all participants in all groups. This was followed by a treatment period during which participants received: 1) the same oral iron supplementation dose alone (oral iron group ‐ control group); 2) 300 IU/kg/day of EPO iv at three‐day intervals (600 IU/kg/week, high dose) along with the same oral iron supplement as the oral iron group (EPO + oral iron group); or 3) 2 mg of intravenous iron sucrose/kg/day + EPO as in group two (iv iron + EPO group). To maintain comparability of iron intake among the three groups, this last group also received EPO and oral iron in an identical manner to the EPO + oral iron group.
Outcomes assessed: Mortality, sepsis, ROP, BPD (oxygen dependency at 36 weeks' postmenstrual age).

Reiter 2005 was a single centre study performed in the USA.

Objective: To determine the effectiveness of a 10‐day EPO course in preterm infants.
Population: Preterm infants < 32 weeks' postmenstrual age, haematocrit ≤ 28%, post‐conceptual age of < 48 weeks or five months chronological age.
Intervention: The EPO group received 300 IU/kg/day (high dose) and 6 mg/kg/day of enteral iron (high dose) versus iron only. Both groups received the intervention for 10 days.
Outcomes assessed: Use of one or more red blood cell transfusions, volume of red blood cells required (mL/kg).

Rocha 2001 was a single centre study performed in Brazil.

Objective: To assess the efficacy of EPO for the prevention and treatment of anaemia of prematurity.
Population: Preterm infants with PMA up to 33 weeks, BW up to 1550 grams and postnatal age between 10 and 35 days.
Intervention: The two EPO groups received either daily doses of 100 IU/kg of EPO or twice weekly doses of 350 IU/kg (700 IU/kg/week, high dose). The EPO groups were given iron (ferrous sulphate) 3 mg/kg/day enterally, and increased to 6 mg/kg/day in the second week of treatment. In the control group, iron supplementation was initiated around the 30th day of life. The control group did not receive any placebo.
Outcomes assessed: Mean number of blood transfusions per participant (no SD provided), two or more blood transfusions per participant.

Romagnoli 2000 was a single centre study performed in Italy.

Objective: To assess the effectiveness of rhEPO.
Population: Infants with PMA ≤ 30 weeks and those infants of 31 to 34 weeks' PMA suffering from respiratory distress syndrome and requiring mechanical ventilation.
Intervention: Treated infants received rhEPO doses of 300 IU/kg body weight sc on each Monday, Wednesday and Friday from the 2nd to the 7th week of life (high dose). Iron supplementation (Intrafer‐Geymonat) was given at a dose of 1 mg/kg per day iv from the 2nd to the 4th week (low dose); thereafter, 12 mg/kg/day orally until the 7th week of life.
Outcomes assessed: Retinopathy of prematurity (stage 1 to 2 and stage 3 to 4), IVH (> grade 2), sepsis, NEC, number of infants transfused, number of transfusions per infant.

Ronnestad 1995 was a single centre study performed in Norway.

Objective: To investigate whether EPO given to infants < 32 weeks' gestational age, fed with their own mother's milk supplemented with a bovine milk protein and electrolyte fortifier together with moderate iron supplementation, would ameliorate the anaemia and thus reduce the need for bred blood cell transfusions after the second week of life.
Population: Preterm infants, PMA < 32 weeks. Age 14 to 22 days.
Intervention: The EPO group received EPO 150 IU/kg three times per week (450 IU/kg/week; low dose) or placebo. All infants received 4 mg/kg/day of iron (low dose) as ferrous fumarase.
Outcomes assessed: Neutropenia.

Samanci 1996 was a single centre study performed in Turkey.

Objective: To determine whether EPO would prevent anaemia of prematurity and reduce the need for transfusion in infants with VLBW.
Population: Preterm infants with PMA ≤ 32 weeks, BW of ≤ 1250 grams. Postnatal age at the first dose was two to four weeks.
Intervention: The EPO group received 200 IU/kg sc three times weekly (600 IU/kg/week, high dose), for four weeks. The control group received an equivalent volume of placebo sc, three times weekly, for four weeks. All infants received oral supplements of elemental iron (3 mg/kg/day) (low dose) during the study period.
Outcomes assessed: Use of one or more red blood cell transfusions, number of blood transfusions per infant, NEC, IVH, major adverse events.

Shannon 1991 was a three‐centre study performed in the USA.

Objective: Not stated.
Population: Preterm infants with BWs ≤ 1250 grams.
Intervention: The EPO group received EPO 100 IU/kg, twice each week (200 IU/kg/week; low dose) for six weeks. The control group received intravenous injections of an identical volume of placebo twice each week for six weeks. All infants received 3 mg/kg/day of oral iron (low dose) and continued at the discretion of the attending physician.
Outcomes assessed: Use of one or more red blood cell transfusions, mortality, NEC, hypertension, neutropenia, side effects.

Shannon 1992 was a single centre study performed in the USA.

Objective: Not stated.
Population: Preterm infants with PMA < 33 weeks and BW < 1250 grams.
Intervention: The EPO group received EPO 100 IU/kg, 5 times a week (500 IU/kg/week; high dose). The control group received an identical volume of placebo suspension, five times a week. Oral iron was started in all infants at 3 mg/kg/day (low dose), divided into three doses and given between feedings. The iron dose was increased to 6 mg/kg/day (high dose) for infants who were tolerating full caloric feedings.
Outcomes assessed: Use of one or more red blood cell transfusions, major adverse events.

Shannon 1995 was a multi‐centre study at 11 centres in the USA.

Objective: To assess whether treatment with EPO would stimulate erythropoiesis and thereby reduce the need for erythrocyte transfusions in preterm infants.
Population: Preterm infants with PMA < 31 weeks with BW of ≤ 1250 grams.
Intervention: The EPO group received 100 IU/kg/day (from Monday through Friday (500 IU/kg/week; high dose)) for six weeks or until the infants were ready to be discharged home. Doses of EPO (or placebo) were adjusted weekly according to changes in body weight. The control group received an identical volume of placebo suspension. Participants received oral iron supplements at study entry to achieve a total enteral intake of 3 mg/kg/day of elemental iron (low dose). Total iron intake was increased to 6 mg/kg/day (high dose) when the infants tolerated full caloric feeding enterally.
Outcomes assessed: Use of one or more red blood cell transfusions, mean number of erythrocyte transfusions per infant, mortality, sepsis, NEC, ROP, hypertension, SIDS, side effects.

Whitehall 1999 was a single centre study conducted in Australia.

Objective: To evaluate safety and efficacy of EPO in reducing the need for red cell transfusions in anaemia of prematurity.
Population: Infants with PMA ≤ 32 weeks.
Intervention: Infants in the EPO group received 400 IU/kg every second day x 10 doses (high dose). Infants in the control group received no placebo. Both groups received 3 mg/kg/day of iron (low dose) increased to 6 mg/kg/day (high dose), as tolerated.
Outcomes assessed: Total volume (mL/kg) of blood transfused, number of transfusions per infant, mortality during hospital stay.

Yamada 1999a was a single centre study conducted in Japan.

Objective: To assess the efficacy of EPO in the treatment of anaemia of prematurity.
Population: Infants with BW 1000 to 1499 grams, PMA < 33 weeks, haemoglobin < 12 g/dL and oral milk intake > 50 mL/kg/day.
Intervention: The EPO group received EPO (200 IU/kg twice a week, low dose) for eight weeks and the control group received no study drug or placebo. All infants received 3 mg/kg/day of oral iron (low dose).
Outcomes assessed: Use of one or more red blood transfusions, total volume of blood transfused (mL/infant), number of transfusions per infant, side effects.

Yamada 1999b was a single centre study conducted in Japan.

Objective: To assess the efficacy of EPO in the treatment of anaemia of prematurity.
Population: Infants with BW 500 to 999 grams, PMA < 33 weeks, haemoglobin < 13 g/dL and oral milk intake > 50 mL/kg/day.
Intervention: The EPO group received low dose EPO (200 IU/kg twice a week) for eight weeks and the control group received no study drug or placebo. All infants received 3 mg/kg/day of oral iron (low dose).
Outcomes assessed: Use of one or more red blood transfusions, total volume of blood transfused (mL/infant), number of transfusions per infant, side effects.

Three different routes of administration were used; subcutaneous (sc) (Shannon 1992; Bechensteen 1993; Emmerson 1993; Meyer 1994; Yamada 1999a; Yamada 1999b; Ronnestad 1995; Shannon 1995; Al‐Kharfy 1996; Bader 1996; Donato 1996; Samanci 1996; Griffiths 1997; Javier Manchon 1997; Corona 1998; Giannakopoulou 1998a; Giannakopoulou 1998b; Kumar 1998; Kivivuori 1999; Whitehall 1999; Romagnoli 2000; Akisu 2001; Rocha 2001; Atasay 2002; Reiter 2005; Bierer 2009; Basiri 2015), intravenous (iv) (Pollak 2001; Shannon 1991; Chen 1995;), oral (Juul 2003), or iv or sc (Maier 2002). The dose of EPO varied from 150 IU/kg/week (Donato 1996; Corona 1998) to 2100 IU/kg/week (Reiter 2005) when given subcutaneously. When given intravenously, the dose varied from 200 IU/kg/week (Shannon 1991) to 300 IU/kg/week (Chen 1995). Juul 2003 provided 7000 IU/kg/week enterally.

Different EPO preparations were used; Recormon (Boehringer‐Mannheim, Germany) (Akisu 2001); Eprex (provided by Cilag Zug, Switzerland, Ortho Pharmaceutical Canada Ltd., Janssen‐Cilag or Guler Pharmaceutical Corp, Istanbul, Turkey) (Bechensteen 1993; Emmerson 1993; Meyer 1994; Chen 1995; Ronnestad 1995; Al‐Kharfy 1996; Bader 1996; Samanci 1996; Griffiths 1997; Giannakopoulou 1998a; Giannakopoulou 1998b; Kivivuori 1999; Whitehall 1999; Atasay 2002 ); Amgen (Shannon 1991), Eogen alpha (Amgen, Inc. Thousand Oaks, CA, USA) (Reiter 2005); unnamed products (Shannon 1992; Shannon 1995; Javier Manchon 1997; Corona 1998; Kumar 1998; Yamada 1999a; Yamada 1999b; Romagnoli 2000; Rocha 2001; Juul 2003; Bierer 2009), NeoRecormon (F. Hofman‐La Roche, Basel, Switzerland) (Maier 2002), Erypo (Janssen‐Cilag Pharma, Vienna, Austria) (Pollak 2001), Hemax (Bio Sidus, S. A.) (Donato 1996) and PDpoietin (Pooyesh Darou Inc, Iran) (Basiri 2015).

Three studies did not state that guidelines for red blood cell transfusions were in place (Shannon 1991; Chen 1995; Akisu 2001). In only one study was it explicit that infants who had received erythrocyte transfusions prior to study entry were excluded (Samanci 1996). For transfusion guidelines, see Additional Table (Table 1: Transfusion Guidelines).

Excluded studies

One study (Ohls 1991) was excluded as it compared an EPO‐treated group with a group receiving blood transfusions. Two studies (Messer 1993; Testa 1998) were excluded as they were not randomised controlled trials. Two studies reported only as abstracts were excluded as one study from Saudi Arabia lacked information to ascertain whether the study was a randomised controlled trial or not (Amin 2004) and the other study conducted in Iran did not provide the age of the infants at the time of enrolment (Ahmadpour Kacho 2004). We were unable to contact the authors for additional information. Warwood 2005 was a dose‐finding study of darbepoetin (longer‐acting and more potent than EPO). Infants were randomised to receive either one or four microgram/kg of a single dose of darbepoetin, without an untreated control group. Meyer 1996, Bechensteen 1997 and Pathak 2003 lacked an untreated control group. There were no outcomes of interest in the study by Widness 2006. In the searches in February 2010, we identified two additional studies (Badiee 2006; Pasha 2008). However, as they did not report on any of our prespecified outcomes we excluded them. In the search conducted in 2012, we identified and excluded three additional studies. In Mohammadzadeh 2005, both groups received EPO. Warwood 2011 reported that "only one of the 20 study subjects qualified for and received a subsequent rbc transfusion during the hospitalisation". The authors did not state if the infant who received the transfusion belonged to the darbepoetin or the control group. Ohls 2012 compared two dosing schedules; once a week versus three times a week of EPO in infants aged seven days or more.

Risk of bias in included studies

The assessment of individual studies is presented in the table 'Characteristics of included studies' and summarised in the 'Risk of Bias' graph (Figure 2) and in the 'Risk of bias' summary (Figure 3). All studies were reported as randomised controlled trials. Information on which to base our judgements on whether a study used concealed allocation or not was often not clearly reported. In general, the studies were of small sample size, ranging from eight (Shannon 1992) to 230 infants (Romagnoli 2000). The studies often lacked a sample size calculation.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

We performed two post hoc secondary analyses for the primary outcome 'Use of one or more red blood cell transfusions'. In the first, we compared those studies that used concealed allocation (a placebo or sham‐injection to blind the intervention) and in which there was blinding of outcome measure assessment to those studies in which this was not evident from the published report. In the second post hoc analysis, we compared the studies that used well defined criteria for red blood cell transfusions with those that used no or less well defined criteria.

Random sequence generation

There was low risk of bias in seven studies (22% of the studies; Meyer 1994; Shannon 1995; Samanci 1996; Whitehall 1999; Romagnoli 2000; Bierer 2009; Basiri 2015 ). In the remaining 78% of the studies, there was an unclear risk of bias.

Allocation

We considered the concealment of allocation to be appropriate in nine studies (28% of the studies; Bechensteen 1993; Shannon 1995; Al‐Kharfy 1996; Griffiths 1997; Whitehall 1999; Romagnoli 2000; Pollak 2001; Maier 2002; Bierer 2009). In the remaining 23 studies (72% of the studies), there was an unclear risk of bias.

Blinding

We considered there to be a low risk of performance bias and detection bias in 15 studies (47% of the studies; Shannon 1991; Shannon 1992; Emmerson 1993; Meyer 1994; Ronnestad 1995; Shannon 1995; Al‐Kharfy 1996; Donato 1996; Samanci 1996; Griffiths 1997; Kumar 1998; Maier 2002; Juul 2003; Bierer 2009; Basiri 2015). Many studies did not use a placebo or sham injection, precluding blinding of the intervention and the outcome measure assessment. There was a high risk of bias in 16 studies (50% of the studies; Bechensteen 1993; Chen 1995; Bader 1996; Javier Manchon 1997; Corona 1998; Giannakopoulou 1998a; Giannakopoulou 1998b; Kivivuori 1999; Whitehall 1999; Yamada 1999a; Yamada 1999b; Romagnoli 2000 Akisu 2001; Pollak 2001; Rocha 2001; Reiter 2005). In one study (3% of the studies; Atasay 2002), there was an unclear risk of bias.

Incomplete outcome data

There was a low risk of bias for 30 studies (94% of the studies) and a high risk of bias in two studies (6% of the studies; Juul 2003; Kivivuori 1999).

Selective reporting

The risk of bias was unclear to us for all the studies (100%) as the study protocols were not available for us to read.

Other potential sources of bias

We had serious concerns about possible 'other bias' in one study (Rocha 2001). The study Romagnoli 2000 was reported in a 'Research letter' format allowing only few details. We obtained information from the Editorial office of the European Journal of Pediatrics, that 'Research letters' are peer‐reviewed. For details regarding these two studies, please see the notes under Characteristics of included studies.

Effects of interventions

See: Summary of findings for the main comparison

For this update, one new study was identified (Basiri 2015). This added 60 infants to the review. The review now includes 31 studies (32 comparisons) that randomised 1651 infants. For details of results, see Data and analyses.

All studies compared late initiation of erythropoietin (EPO) (8 ‐ 28 days) versus placebo or no intervention.

Primary Outcome:

The use of one or more red blood cell transfusions (Outcome 1.1)

See Figure 4.

Figure 4

Forest plot of comparison: 1 Late initiation of EPO (8 ‐ 28 days) vs placebo or no intervention, outcome: 1.1 Use of one or more red blood cell transfusions (low and high dose of EPO).

A total of 21 studies including 1202 infants reported on the use of one or more red blood cell transfusions following the use of either low or high dose of EPO. There was a significant reduction in the use of one or more red blood cell transfusions (typical risk ratio (RR) 0.72, 95% confidence interval (CI) 0.65 to 0.79; typical risk difference (RD); ‐0.17, 95% CI ‐0.22 to ‐0.12; number needed to treat for an additional beneficial outcome (NNTB); 6, 95% CI 5 to 8). There was moderate heterogeneity for this outcome (for RR; I² = 66.0%; for RD; I² = 58.0%). The quality of the evidence was very low.

Subgroup analyses:

We conducted further analyses including studies that used a high dose of EPO (> 500 IU/kg/week) or a low dose of EPO (≤ 500 IU/kg/week).

High dose of EPO (Outcome 1.2)

The summary estimates for 15 studies including 972 infants testing a high dose of EPO (Outcome 1.2) were statistically significant with a typical RR of 0.77 (95% CI 0.68 to 0.86), a typical RD of ‐0.14 (95% CI ‐0.19 to ‐0.08) and a NNTB of 7 (95% CI 5 to 13). There was moderate heterogeneity for this outcome for RR (I² = 63%) and for RD (I² = 62%).

We conducted a subgroup analysis for high dose of EPO in combination with high dose of iron (Outcome 1.2.1). Six studies (n = 318) showed a typical RR of 0.74 (95% CI 0.62 to 0.88), a typical RD of ‐0.16 (95% CI ‐0.24 to ‐0.08) and NNTB of 6 (95% CI 4 to 13). There was high heterogeneity for both RR (I² = 79%) and for RD (I² = 78%).

Nine studies of high EPO and low dose of iron (Outcome 1.2.2) (n = 654) showed a typical RR of 0.78 (95% CI 0.67 to 0.90), a typical RD of ‐0.13 (95% CI ‐0.20 to ‐0.06) and NNTB of 8 (95% CI 5 to 20). There was moderate heterogeneity for RR (I² = 52%) and low heterogeneity for RD (I² = 29%).

Low dose of EPO (Outcome 1.3)

The summary estimates for seven studies including 239 infants testing a low dose of EPO (Outcome 1.3) were statistically significant with a typical RR of 0.53 (95% CI 0.42 to 0.67), a typical RD of ‐0.34 (95% CI ‐0.45 to ‐0.23) and a NNTB of 3 (95% CI 2 to 4). There was moderate heterogeneity for RR (I² = 59%) and no heterogeneity for RD (I² = 14%).

We conducted a subgroup analysis for low dose of EPO in combination with high dose of iron (Outcome 1.3.1). Three studies (n = 77) showed a typical RR of 0.50 (95% CI 0.31 to 0.79), a typical RD of ‐0.31 (95% CI ‐0.49 to ‐0.13) and a NNTB of 3 (95% CI 2 to 8). There was no significant heterogeneity for this outcome for RR (I² = 0%) and RD (I² = 0%).

Four studies (n = 162) evaluated the effectiveness of low dose of EPO in combination with low dose of iron (Outcome 1.3.2). The typical RR was 0.54 (95% CI 0.41 to 0.71), the typical RD was ‐0.36 (95% CI ‐0.49 to ‐ 0.22) and the NNTB was 3 (95% CI 2 to 5). There was high heterogeneity (I² = 76%) for RR and moderate heterogeneity for RD (I² = 53%).

Secondary outcomes:

The total volume (mL/kg) of blood transfused per infant (Outcome 1.4)

Five studies including 197 infants reported on the total volume of blood transfused per infant. The typical mean difference (MD) between the groups was not statistically significant, with a typical MD of ‐1.61 mL/kg (95% CI ‐5.78 to 2.57) transfused per infant. There was high heterogeneity (I² = 92%). The very high heterogeneity may indicate that the data are skewed and it brings uncertainty to the estimate. Corona 1998 (n = 60) reported on this outcome but provided only the means with no standard deviation (SD). In the two EPO groups combined, the mean was 20 mL/kg and in the control group it was 32 mL/kg (P < 0.01, according to the authors).

Number of red blood cell transfusions per infant (Outcome 1.5)

The number of red blood cell transfusions per infant was reported in 11 studies enrolling 817 infants. The significant typical MD was ‐0.22 (95% CI ‐0.38 to ‐0.06) favouring the EPO group. There was high heterogeneity (I² = 94%). In Griffiths 1997 (n = 42), the median number of blood transfusions was lower for the infants in the EPO group (difference in medians ‐2 (95% CI ‐4 to 0)).

Number of donors to whom the infant was exposed (Outcome 1.6)

Two studies reported on donor exposure in 165 enrolled infants. The significant MD was 0.45 (95% CI 0.20 to 0.69) indicating a higher donor exposure number in the EPO group. There was high heterogeneity for this outcome (I² = 93%).

Mortality during initial hospital stay (all causes of mortality) (Outcome 1.7)

Thirteen studies (14 comparisons) including 767 infants reported on mortality during initial hospital stay. The non‐significant typical RR was 0.82 (95% CI 0.49 to 1.39) and the typical RD was ‐0.01 (95% CI ‐0.05 to 0.02). There was no statistically significant heterogeneity for this outcome for either RR (I² = 0%) or RD (I² = 0%). The quality of the evidence was moderate.

Retinopathy of prematurity (ROP) (all stages or stage not reported) (Outcome 1.8)

Three studies including 404 infants reported on ROP (all stages), with a non‐significant typical RR 1.27 (95% CI 0.99 to 1.64) and a typical RD of 0.09 (95% CI ‐0.00 to 0.18). There was high heterogeneity for this outcome for both RR (I² = 83%) and RD (I² = 82%). The quality of the evidence was very low.

Retinopathy of prematurity (ROP) stage ≥ 3 (Outcome 1.9)

Three trials enrolling 442 infants reported on severe ROP (stage ≥ 3). The typical RR was 1.73 (95% CI 0.92 to 3.24) and the typical RD was 0.05 (95% CI ‐0.01 to 0.10); neither was statistically significant. There was unimportant heterogeneity for this outcome for RR (I² = 18%) but high heterogeneity for RD (I² = 79%). The quality of the evidence was very low.

Proven sepsis (clinical symptoms and signs of sepsis and positive blood culture) (Outcome 1.10)

Five studies including 551 infants reported on this outcome. The typical RR was 0.75 (95% CI 0.52 to 1.09) and the typical RD was ‐0.05 (95% CI ‐0.11 to 0.01), neither statistically significant. There was no heterogeneity for this outcome for either RR (I² = 0%) or RD (I² = 0%).

Necrotising enterocolitis (NEC) (Bell's stage II or higher) (Outcome 1.11)

Six studies including 656 infants reported on NEC. In some studies, the stage was not reported but the results were included in the meta‐analyses. The typical RR was 0.88 (95% CI 0.45 to 1.70) and the typical RD ‐0.01 (95% CI ‐0.04 to 0.02).Neither estimate was statistically significant. There was no heterogeneity for this outcome for either RR (I² = 0%) or RD (I² = 0%). The quality of the evidence was moderate.

Intraventricular haemorrhage (IVH); all grades (Outcome 1.12)

Four studies including 454 infants reported on intraventricular haemorrhage (all grades). In Samanci 1996, there were no events in either group and that study was therefore disregarded for the RR. The non‐significant typical RR was 0.87 (95% CI 0.53 to 1.42) and typical RD was ‐0.02 (95% CI ‐0.07 to 0.04). There was no heterogeneity for either RR (I² = 0%) or RD (I² = 0%). IVH was probably not a relevant outcome in this review as most haemorrhages occur during the first few days of life and infants were enrolled later in these studies.

Periventricular leukomalacia (PVL); cystic changes in the periventricular areas (Outcome 1.13)

One study enrolling 145 infants reported on PVL. The non‐significant RR was 4.80 (95% CI 0.57 to 40.05) and the RD was 0.05 (95% CI ‐0.01 to 0.12). The test for heterogeneity was not applicable.

Bronchopulmonary dysplasia (BPD) (supplemental oxygen at 28 days of age)(Outcome 1.14)

Two studies (n = 285) reported on BPD at 28 days. In Al‐Kharfy 1996, all infants in both groups had BPD. The typical RR was 1.25 (95% CI; 1.00 to 1.55) and the RD was 0.10 (95% CI 0.00 to 0.20). There was very high heterogeneity for both RR (I² = 97%) and RD (I² = 88%). Both results were of borderline statistical significance (test for overall effect: Z = 1.99 (P = 0.05)). The extremely high heterogeneity for this analysis may relate to the fact that in one study all the infants had BPD at 28 days of age. The results should be interpreted with caution.

Bronchopulmonary dysplasia (BPD) (supplemental oxygen at 36 weeks' postmenstrual age (PMA)) (Outcome 1.15)

Three studies enrolling 216 infants reported on the use of supplemental oxygen at 36 weeks' PMA. The typical RR was 0.89 (95% CI 0.59 to 1.35) and the typical RD was ‐0.03 (95% CI‐0.15 to 0.08); neither was statistically significant. There was moderate heterogeneity for RR (I² = 56%) and for RD (I² = 59%).

Sudden infant death (SIDS) after discharge (Outcome 1.16)

Six studies including 363 infants reported on SIDS. The typical RR was 1.06 (95% CI 0.25 to 4.52) and the typical RD was 0.00 (95% CI ‐0.03 to 0.04). Neither was statistically significant. There was no heterogeneity for either RR (I² = 0%) or RD (I² = 0%).

Neutropenia (Outcome 1.17)

Five studies (six comparisons) enrolling 164 infants reported on neutropenia. The typical RR was 0.28 (95% CI 0.05 to 1.54) (in only two studies did the outcome of interest occur), and the typical RD was ‐0.04 (‐0.11 to 0.03); neither was statistically significant. There was no heterogeneity for RR (I² = 0%) and for RD (I² = 0%).

Hypertension (Outcome 1.18)

Seven studies including 361 infants reported on hypertension. The RR was 1.20 (95% CI 0.46 to 3.14) and the RD 0.01 (95% CI ‐0.04 to 0.05); neither was statistically significant. There was no statistically significant heterogeneity for either RR (I² = 21%) or RD (I² = 0%).

Length of hospital stay (days) (Outcome 1.19)

Length of hospital stay was reported in two studies enrolling 55 infants. There was no significant difference between the groups with a typical MD of ‐0.35 days (95% CI ‐12.83 to 12.13). There was no heterogeneity (I² = 0%).

Long‐term outcomes assessed at any age beyond one year of age by a validated cognitive, motor, language, or behavioural/school/social interaction/adaptation test

Long‐term neurodevelopmental outcomes were not reported in any study.

Any side effects reported in the trials

There were no serious side effects reported in most of the trials that specifically reported on adverse events (Shannon 1991; Shannon 1992; Bechensteen 1993; Yamada 1999a; Yamada 1999b; Chen 1995; Shannon 1995; Bader 1996; Donato 1996; Samanci 1996; Corona 1998; Giannakopoulou 1998a; Giannakopoulou 1998b; Kumar 1998; Kivivuori 1999; Rocha 2001; Juul 2003). Griffiths 1997 reported a total of 41 different types of adverse events, with infection (positive blood cultures), pneumonia, and patent ductus arteriosus being the most common.

Secondary (post hoc) analyses:

In an attempt to further explore the heterogeneity observed in the primary outcome and subgroup analyses, we performed a post hoc analysis comparing the results of studies that we judged as high quality with those that we identified as of lower quality or could not precisely define their quality because of lack of information. We also compared the results of studies that used strict criteria for red blood cell transfusions to those that used no criteria or less strict criteria.

Use of one or more blood transfusions (secondary analysis based on quality) (Outcome 1.20)

For six high‐quality studies enrolling 417 infants, the typical RR was 0.83 (95% CI 0.73 to 0.95); the typical RD was ‐0.12 (95% CI ‐0.21 to ‐0.04). For 15 studies of uncertain quality enrolling 785 infants, the typical RR was 0.63 (95% CI 0.54 to 0.73) and the typical RD was ‐0.20 (‐0.26 to ‐0.14). The summary effect size was larger in the studies of poor quality. There was low heterogeneity for the high‐quality studies (I² = 49% for RR and 47% for RD), as well as for the studies of uncertain quality which had moderate heterogeneity (I² = 71% for RR and 61% for RD).

Use of one or more blood transfusions (secondary analysis based on criteria for red blood cell transfusions) (Outcome 1.21; Outcome 1.21.1; Outcome 1.21.2)

We considered 16 studies enrolling 1023 infants to have used strict (although variable) guidelines for red blood cell transfusions, and three studies enrolling 97 infants to have used no criteria or less strict criteria. We excluded two studies for which we were unable to translate the text regarding possible transfusion guidelines (Yamada 1999a; Yamada 1999b). For the 16 studies using strict red blood cell transfusion guidelines, the typical RR was 0.76 (95% CI 0.69 to 0.85) and the typical RD was ‐0.15 (95% CI ‐0.20 to ‐0.09); NNTB 7 (95% CI 5 to 11). There was moderate heterogeneity for RR (I² = 61%) and RD (I² = 53%). For the studies using no criteria or less strict criteria, the typical RR was 0.25 (95% CI 0.08 to 0.77) and the typical RD was ‐0.21 (95% CI ‐0.36 to ‐0.07); NNTB 5 (95% CI 3 to 14). There was no heterogeneity for the studies using no criteria or less strict criteria for RR (I² = 0%) and RD (I² = 0%). The summary effect size was larger for the studies that did not use strict guidelines for red blood cell transfusions compared to those that did.

Funnel plot

A funnel plot for the primary outcome 'Use of one or more red blood cell transfusions' was asymmetric, with a relative absence of smaller studies not having a protective effect (see Figure 5).

Figure 5

Funnel plot of comparison: 1 Late initiation of EPO (8 ‐ 28 days) vs placebo or no intervention, outcome: 1.1 Use of one or more red blood cell transfusions (low and high dose of EPO).

Enterally‐dosed EPO

One study (Juul 2003) using enterally‐dosed EPO found that the intervention did not significantly influence erythropoiesis or iron utilisation when given for a two‐week period, nor did it elevate the serum EPO concentration in preterm or term infants. The authors concluded that enterally‐dosed EPO is not an effective substitute for parenteral administration (Juul 2003).

Discussion

Summary of main results

The current review included 31 studied (32 comparisons) meeting our inclusion criteria. These studies included a total of 1651 preterm or low birth weight infants and reported on at least one of the outcomes of interest for this systematic review.

The results showed that late administration of erythropoietin reduces the use of one or more blood transfusions following study entry. These results were quite consistent (overlapping confidence intervals) when including studies that used both low and high doses of EPO in combination with low and high doses of iron.

With the inclusion of Romagnoli 2000, in which there was a statistically significant increase in retinopathy of prematurity (ROP) at all stages and at stage 3 or higher, the overall estimates in the meta‐analyses now showed a trend towards increase for both these outcomes. Three studies including 404 infants reported on ROP (all stages), with a typical RR 1.27 (95% CI 0.99 to 1.64) and a typical RD of 0.09 (95% CI ‐0.00 to 0.18). This outcome was not statistically significantly different between the groups. There was high heterogeneity for this outcome for both RR (I² = 83%) and RD (I² = 82%). Three trials enrolling 442 infants reported on severe ROP (stage 3 or higher). The typical RR was 1.73 (95% CI 0.92 to 3.24) and the typical RD was 0.05 (95% CI ‐0.01 to 0.10); neither was statistically significant. There was minimal heterogeneity for this outcome for RR (I² = 18%) but high heterogeneity for RD (I² = 79%). As the cut‐off for early EPO treatment versus late EPO treatment was set arbitrarily at eight completed days of age of initiation of treatment, we thought it justified to include all studies that have reported on ROP in secondary analyses. Details of these analyses were presented in the 2014 update of the early EPO review (Ohlsson 2012). We combined all studies that reported on ROP at stage 3 or higher regardless of at what age the EPO treatment was initiated. We included seven studies from the early EPO review and three studies from the late EPO review that reported on this outcome. A total of 10 studies enrolling 1303 infants reported on this outcome. There was a significantly increased risk of ROP stage 3 or higher (typical RR 1.48 (95% CI 1.02 to 2.13); typical RD 0.03 (95% CI 0.00 to 0.06) (P = 0.04 for RR and 0.03 for RD); number needed to treat for an additional harmful outcome (NNTH) 33 (95% CI 17 to infinity)). There was no heterogeneity for RR (1² = 0%) and moderate heterogeneity (I² = 50%) for RD.

Overall completeness and applicability of evidence

The number needed to treat for an additional beneficial outcome (NNTB) to avoid one red blood cell transfusion was low (range 3 to 8, for different combinations of EPO and iron). The clinical importance of this finding is lessened by the fact that any donor exposure was not avoided, as many infants required red blood cell transfusions prior to study entry. Only two studies reported on donor exposure. In the study by Bierer 2009, there was a statistically significant increase in the number of donors the infants were exposed to but not in the study by Maier 2002. In the meta analysis of the two studies, a significant difference in the typical mean difference for number of transfusions was noted indicating a higher donor exposure number in the EPO group. There was very high heterogeneity for this outcome (I² = 93%), reducing the importance of the finding. In addition, in the 2012 and 2013 updates there was no statistically significant reduction in the total volume (mL/kg) of blood transfused per infant. There was a small reduction in the mean number of transfusions (0.2) per infant.

The need for intravenous, intramuscular, or subcutaneous injections with EPO/iron treatment in the neonatal period is associated with repeated painful stimuli and could potentially have adverse long‐term effects. This has not been addressed in any study.

Quality of the evidence

The study quality varied and important information regarding the random sequence generation and whether the allocation was concealed or not was often missing. No study was reported according to the 'Consort' statement (CONSORT 2012). Sample sizes were small and long‐term outcomes (18 to 24 months corrected age) were not reported. In only one study (Samanci 1996), did the authors state that infants were not eligible to enter the study if they had previously received a red blood cell transfusion. Most studies followed guidelines for red blood cell transfusions, although these varied between the studies.

For the primary outcome of 'use of one or more blood transfusions', the typical risk ratio (RR) for six high‐quality studies was 0.83 (95% CI 0.73 to 0.95). For 15 studies of uncertain quality, the typical RR was 0.63 (95% CI 0.54 to 0.73). The CIs for these analyses were not overlapping, indicating that there were statistically significant differences in the effect sizes between studies that could be ascertained as being of high quality and studies of uncertain quality. There was a reduction in heterogeneity when the high‐quality studies were analysed separately. Studies of higher quality often show lower effect sizes (Schulz 1995). Judging the quality of a study depends to a large extent on the information published, and obtaining additional information from the authors may change the evaluations. The typical effect size for studies that used strict red blood cell transfusion guidelines was smaller (RR 0.76) than for studies that used no or less strict criteria (RR 0.25).

Of concern is the finding of moderate heterogeneity for the primary outcome, including all combinations of low and high EPO and low and high iron treatment. The heterogeneity remained for individual combinations of EPO and iron. The heterogeneity could possibly be explained by the fact that the studies were conducted in 22 countries, with presumably different care practices. Of note, the control rates for red blood cell transfusions varied markedly between studies. As noted in the Additional Table 1, there was large variation in the guidelines for red blood cell transfusions. The results of these post hoc analyses should therefore be interpreted with caution.

Potential biases in the review process

As outlined in Figure 5 (Funnel plot of comparison: 1 Late initiation of EPO (8 ‐ 28 days) versus placebo or no intervention, outcome: 1.1 Use of one or more red blood cell transfusions (low and high dose of EPO)), there was a paucity of smaller studies not having a protective effect. It is possible that such studies have been conducted and that the results have not been published.

Several studies were supported by pharmaceutical companies and some authors were employed by such companies.

We are not aware of any potential biases in our own review process.

Agreements and disagreements with other studies or reviews

Systematic reviews of the efficacy of EPO in anaemia of prematurity have been published (Vamvakas 2001; Garcia 2002; Kotto‐Kome 2004). Vamvakas 2001 concluded that there is extreme variation in the results, and, until this variation is better understood, it is too early to recommend EPO as standard treatment for the anaemia of prematurity. Garcia 2002 concluded that administering EPO to VLBW neonates can result in a modest reduction in late erythrocyte transfusions and that this effect is dependent on the dose of EPO used. Kotto‐Kome 2004 concluded that if EPO is begun in the first week of life, a moderate reduction can be expected in the proportion of VLBW neonates transfused. The reduction was less significant for early transfusion than for late transfusion. Direct comparisons regarding the results of this systematic overview and previous reviews (Vamvakas 2001; Garcia 2002; Kotto‐Kome 2004) are not appropriate as our review included a much larger sample of studies. In 2015, the French Society of Neonatology published "Recombinant human erythropoietin in neonates: guidelines for clinical practice". The publication is labelled as a systematic evidence review. They reported that "Early EPO reduced the risk of RBC transfusions, donor exposure, and the number of transfusions in very preterm infants (LE2) (LE = level of evidence). Late EPO reduced the risk of RBC transfusions and the number of transfusions in very preterm infants (LE2). There is no difference between the effectiveness of early and late EPO (LE2). There is no difference between high‐dose and low‐dose EPO (LE2). The level of evidence is too low to recommend the intravenous route. EPO has no impact on the rate of bronchopulmonary dysplasia, necrotising enterocolitis (LE3), and retinopathy of prematurity (LE2). The level of evidence is too low to recommend EPO for neuro protection in very preterm or term infants". They concluded; "EPO to reduce RBC transfusion in very preterm infants is recommended (Level A). The optimal time to start therapy is unknown (Level B). The recommended dose is 750 IU/kg/week via three subcutaneous injections for 6 weeks (Level B)". To our knowledge, there is no other recent systematic review for late EPO (Lopez 2015).

Figure 1

Study flow diagram: review update

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 4

Forest plot of comparison: 1 Late initiation of EPO (8 ‐ 28 days) vs placebo or no intervention, outcome: 1.1 Use of one or more red blood cell transfusions (low and high dose of EPO).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 5

Funnel plot of comparison: 1 Late initiation of EPO (8 ‐ 28 days) vs placebo or no intervention, outcome: 1.1 Use of one or more red blood cell transfusions (low and high dose of EPO).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.1

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 1 Use of one or more red blood cell transfusions (low and high dose of EPO).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.2

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 2 Use of one or more red blood cell transfusions (high dose of EPO).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.3

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 3 Use of one or more red blood cell transfusions (low dose of EPO).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.4

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 4 Total volume (mL/kg) of red blood cells transfused per infant.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.5

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 5 Number of red blood cell transfusions per infant.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.6

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 6 Number of donors the infant was exposed to.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.7

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 7 Mortality during initial hospital stay (all causes).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.8

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 8 Retinopathy of prematurity (all stages or stage not reported).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.9

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 9 Retinopathy of prematurity (stage ≥ 3).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.10

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 10 Proven sepsis.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.11

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 11 Necrotising Enterocolitis ≥ Bell's stage 2.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.12

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 12 Intraventricular haemorrhage all grades (or grade not specified).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.13

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 13 Periventricular leukomalacia.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.14

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 14 Bronchopulmonary dysplasia (supplementary oxygen at 28 days).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.15

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 15 Bronchopulmonary dysplasia (supplementary oxygen at 36 weeks' postmenstrual age).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.16

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 16 SIDS.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.17

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 17 Neutropenia.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.18

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 18 Hypertension.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.19

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 19 Length of hospital stay (days).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.20

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 20 Use of one or more red blood cell transfusions (secondary analysis based on study quality).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.21

Comparison 1 Late initiation of EPO (8‐28 days) vs. placebo or no intervention, Outcome 21 Use of one or more red blood cell transfusions (secondary analysis based on RBC transfusion guidelines).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Late EPO compared with placebo or no intervention for complications of preterm birth ‐ primary and secondary outcomes
Patient or population: preterm infants with low birth weight Settings: NICU Intervention: Late EPO Comparison: Placebo or no intervention
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 intervention	EPO
Use of one or more red blood cell transfusions (low and high dose of EPO) during initial hospital stay	High risk population		RR 0.72 (0.65 to 0.79)	1202 (21)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in 16 of the studies, about allocation concealment in 17 of the studies and about blinding (performance bias and detection bias) in 11 of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 66% and for the typical RD 58% (both moderate heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1202), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot was asymmetrical ‐ we downgraded the quality of the evidence by one step.
	603 per 1000	434 per 1000 (392 to 477)
Necrotising Enterocolitis ≥ Bell's stage 2 during initial hospital stay	High risk population		RR 0.88 (0.45 to 1.70)	656 (6)	⊕⊕⊕⊝ moderate	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in three of the studies and about blinding (performance bias and detection bias) in one of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for both the typical RR and for the typical RD was 0% (both no heterogeneity). Directness of evidence: Studies were conducted in the target population. Precision: Because of the sample size (n = 656), the point estimate was quite precise with a narrow 95% CI. Presence of publication bias: As only six studies were included in the analyses we did not perform a funnel plot.
	52 per 1000	46 per 1000 (23 to 88)
Mortality during initial hospital stay (all causes)	High risk population		RR 0.82 (0.49 to 1.39)	767 (14)	⊕⊕⊕⊝ moderate	Bias: We had concerns about random sequence generation in 11 of the studies, about allocation concealment in 7 of the studies and about blinding (performance bias and detection bias) in six of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR and for the RD was 0% (both no heterogeneity). Directness of evidence: Studies were conducted in the target population. Precision: Because of the large sample size (n = 1202), the point estimate was precise with a narrow 95% CI. Presence of publication bias: The funnel plot was symmetrical.
	63 per 1000	52 per 1000 (31 to 88)
Retinopathy of prematurity (all stages or stage not reported) during initial hospital stay	High risk population		RR 1.27 (0.99 to 1.64)	404 (3)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in none of the studies, and about blinding (performance bias and detection bias) in two of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 83% and for the typical RD 82% (both high heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the small sample size (n = 404 ), the point estimate was not precise. We downgraded the quality of the evidence by one step. Presence of publication bias: As there was only three studies included we did not perform a funnel plot.
	328 per 1000	417 per 1000 (325 to 538)
Retinopathy of prematurity (stage ≥ 3) during initial hospital stay	High risk population		RR 1.73 (0.92 to 3.24)	442 (3)	⊕⊝⊝⊝ very low	Bias: We had concerns about random sequence generation in two of the studies, about allocation concealment in none of the studies, and about blinding (performance bias and detection bias) in two of the studies. We downgraded the quality of the evidence by one step. Heterogeneity/Consistency: I² for the typical RR was 18% (no heterogeneity) and for the typical RD 79% (high heterogeneity). We downgraded the quality of the evidence by one step. Directness of evidence: Studies were conducted in the target population. Precision: Because of the small sample size (n = 442 ), the point estimate was not precise. We downgraded the quality of the evidence by one step. Presence of publication bias: As there was only three studies included we did not perform a funnel plot.
	63 per 1000	109 per 1000 (58 to 203)
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; I² = I‐squared; NICU: Neonatal Intensive Care Unit; RR: Risk Ratio;
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.

Ir a la tabla de la revisión

Table 1. Transfusion guidelines

Reference	Indications
Akisu 2001	Guidelines for transfusions were not presented.
Al‐Kharfy 1996	The indications for transfusion were 1) shock, 2) cumulative loss of ≥ 10% of the blood volume in 72 hours or less when further blood sampling is expected, 3) Hb < 130 g/L in acutely ill neonates with cardiorespiratory disease, and 4) Hb < 80 to 100 g/L with clinical signs of anaemia. A volume of 15 mL/kg was recommended for each transfusion.
Atasay 2002	Criteria for blood transfusion (10 mL/kg packed red cells) were as follows: a Hct < 30% when signs and symptoms attributed to anaemia including persistent tachycardia (180 beats/min for 24 hours), frequent apnoea with bradycardia and daily weight gain < 10 g/kg despite optimal protein and caloric intake (3.5 g/kg, 100 kcal/kg/day). Infants were transfused with a Hct of 35% to 40% if they received more than 40% oxygen or ventilation therapy.
Bader 1996	Criteria for blood transfusion (10 mL/kg of red cells) were as follows: a) a Hct < 25%, b) an increased frequency of apnoeic events which required either stimulation or aminophylline therapy, c) changes in heart rate patterns, e.g. an increase in frequency of bradycardia (< 80 beats/min) or tachycardia (> 180 beats/min, d) failure of weight gain of > 10 g/kg/day despite an optimal caloric intake of > 120 kcal/day and e) lethargy without evidence of sepsis.
Basiri 2015	Transfusion guidelines for preterm infants included: (1) asymptomatic infants with Hct < 21% and reticulocytes < 2%; (2) infants with haematocrit < 31% and hood O₂ < 36% or mean airway pressure < 6 cm H₂O by continuous positive airway pressure (CPAP) or intermittent mandatory ventilation (IMV) or > 9 apnoeic and bradycardic episodes per 12 hours or 2/24 hours requiring bag‑and‑mask ventilation while on adequate methylxanthine therapy or heart rate > 18/min or respiratory rate > 80/min sustained for 24 hours or weight gain of < 10 g/day for 4 days on 100 kcal/kg/day or having surgery; (3) infants with haematocrit < 36% and requiring > 35% O₂ or mean airway pressure of 6 to 8 cm H₂O by CPAP or IMV.
Bechensteen 1993	Indications of blood transfusions were: 1) Hb < 80 g/L or 2) otherwise at the discretion of the clinician caring for the infant according to symptoms and signs.
Bierer 2009	Very specific transfusion criteria were in place for infants on mechanical ventilation, for infants who required supplemental oxygen and for infants on room air.
Chen 1995	Transfusions were given because of frequent and prolonged apnoea.
Corona 1998	Transfusions were considered based on the clinical condition (pallor, tachycardia, tachypnoea, apnoea with or without bradycardia, poor weight gain, difficulties with sucking) and haematological parameters (Hb < 70 g/L, Hct < 26%, with low reticulocyte counts).
Donato 1996	During the first week of life, infants were given transfusions of packed red blood cells for replacement when blood drawn for analysis was in excess of 8 mL/kg of body weight; fresh whole blood was given if signs attributable to hypovolaemia or anaemia developed. Subsequently, infants with heart rate > 180 beats/min, severe apnoea/bradycardia or poor weight gain (< 10 grams/day in spite of a 100 calories/day intake during 5 consecutive days) were transfused if the Hct was < 25% (or < 30% if oxygen or mechanical ventilation was required); asymptomatic infants were transfused only when a central Hct < 23% was reached.
Emmerson 1993	The decision to give a blood transfusion to a study infant was made by the medical staff of the neonatal unit who were blinded to the randomisation. The unit policy at the time of the study was to transfuse a preterm infant who had a Hb < 100 g/L and who had symptoms consistent with those caused by anaemia. The symptoms and signs of anaemia included poor feeding, tachycardia, tachypnoea, apnoea, and pallor. Infants with a Hb < 80 g/L were transfused even if asymptomatic.
Giannakopoulou 1998a; Giannakopoulou 1998b	Indications for blood transfusion were a Hb < 80 g/L or otherwise at the discretion of the physician treating the infants according to symptoms and signs.
Griffiths 1997	Infants were transfused if they were ventilated and/or oxygen‐dependent with a Hb of < 120 g/L, had clinically symptomatic anaemia, or were asymptomatic with a Hb of < 70 g/L. Infants were transfused if they were ventilated and/or oxygen‐dependent with a Hb of < 120 g/L, had clinically symptomatic anaemia, or were asymptomatic with a haemoglobin of < 70 g/L.
Javier Manchon 1997	Transfusion guidelines were similar in all three centres (details not provided).
Juul 2003	By NICU policy, on admission, infants weighing < 1000 grams at birth were assigned 1 unit of packed red blood cells divided into 8 aliquots. These aliquots were used for transfusions during the first month of life. The following transfusion guideline was used for infants of all birth weights: Transfusion is recommended for a Hct < 35% if the infant requires positive pressure with a mean airway pressure > 6 cm H₂O and requires > 35% oxygen. Transfusion is recommended for Hct < 30% if the infant requires oxygen (< 35% FIO₂), is receiving continuous positive airway pressure or intubated with mean airway pressure < 6 cm H₂O, if an infant has significant apnoea and bradycardia while receiving methylxanthines (> 9 episodes in 12 hours or 2 episodes in 24 hours requiring mask‐and‐bag ventilation), if the heart rate is > 180 beats/min or respiratory rate > 80/min and persists for 24 hours, if weight gain < 10 grams/day over 4 days despite adequate calories, or in the presence of sepsis. If the Hct is < 20%, no symptoms are necessary for transfusion. Transfusion volumes were standardised as follows: infants received an initial transfusion of 15 mL/kg over a period of 3 to 4 hours. A follow‐up Hct was checked after 4 hours. If the Hct was < 30%, a second aliquot of 10 mL/kg was given If the Hct was between 30% and 35%, an additional 5 mL/kg was given.
Kivivuori 1999	The Hct values were maintained at > 30% by red blood cell transfusions (10 mL/kg per time) in asymptomatic infants. In infants who had symptoms or signs of anaemia, red blood cells were transfused if the Hct value was < 40%. The transfusion policies were the same in all study hospitals.
Kumar 1998	The need for erythrocyte transfusion was assessed by the clinicians caring for each infant and the decision to transfuse was made without consulting the study investigators. According to the practice in the NICU, infants received transfusion if a Hct level of < 27% was associated with one of the following signs and symptoms of anaemia: 1) frequent apnoea and bradycardia, defined as > 6 episodes in 12 hours or any episode requiring bag‐and‐mask ventilation, in an infant with therapeutic serum levels of theophylline; 2) persistent tachycardia, defined as > 180 beats/min for more than 12 hours; 3) poor weight gain (< 10 grams/day averaged over a 7‐day period) despite adequate caloric intake; and 4) increasing oxygen requirement in infants with chronic lung disease despite optimum diuretic and bronchodilator therapy.
Maier 2002	Infants with artificial ventilation or in > 40% of inspired oxygen were not transfused unless Hct dropped to < 40%. Spontaneously breathing infants were not transfused unless Hct dropped to < 35% during the first 2 weeks of life, 30% during the 3rd to 4th weeks, and 25% thereafter. Transfusion was allowed when life‐threatening anaemia or hypovolaemia was assumed by the treating neonatologist, or surgery was planned. Twelve of the 14 centres used satellite packs of the original red cell pack to reduce donor exposure.
Meyer 1994	Direct quote: "The need for blood transfusion was assessed by the attending neonatal physician and decisions were made independently of the investigators. The following guidelines were developed, based on existing literature and nursery practices: a. Hct < 30% and 1) Weight gain of < 10 grams/day averaged over 1‐week period (infant tolerating full oral feeds and receiving adequate calories). 2) Three or more episodes of apnoea (respirations absent for 20 seconds) or bradycardia (heart rate of < 100 beats per minute) in a 24‐hour period not due to other causes and not responsive to methylxanthine treatment. 3) Tachycardia (> 170 beats/min) or tachypnoea (> 70 breaths/min) sustained over a 24‐hour period or associated with acute cardiac decompression. 4) A requirement for surgery. b. Development of a clinically significant patent ductus arteriosus (i.e. at least three of the following features: heart rate > 160 beats/min, brisk brachial and/or dorsalis pedis pulses, palpable precordial pulsation, systolic murmur, cardiomegaly on chest radiograph). c. Pulmonary disease and fractional inspired oxygen concentrations increasing by > 10% per week. d. Systemic infection (either clinically suspected or proven on blood culture) associated with a sudden decrease in hematocrit. e. Hematocrit of 22% or less and an absolute reticulocyte count of < 100 000 x 10⁹/L".
Pollak 2001	Standard NICU transfusion criteria were used (authors refer to Shannon 1995; see below).
Reiter 2005	Conservative transfusion guidelines were in place and followed. Criteria for red blood cell transfusion in the acutely ill infant requiring mechanical ventilation or nasal continuous positive airway pressure included: phlebotomy loss of > 15% of blood volume associated with hypotension, or Hct < 30%. Criteria for red blood cell transfusion in the convalescent infant requiring no more than supplemental oxygen included: Hct < 28% with symptomatic anaemia (tachycardia, poor somatic growth, or metabolic acidosis) or Hct < 20%.
Rocha 2001	The decision for blood transfusion within the whole study population was made by the assistant doctor of each newborn, and was based on the following criteria: Hct ≤ 20%, inadequate weight gain, 3 or more apnoea or bradycardia episodes within 24 hours, presurgical procedure requirement, disease associated with sudden Hct decline, restoration of the blood collected for lab exams, maintenance of Hct up to 30% associated with minimal ventilatory support requirement, and Hct up to 35% when ventilation requirements are greater. The assistant doctor who recommended blood transfusion did not know to which group the infant belonged.
Romagnoli 2000	Infants on mechanical ventilation and/or on more than 30% of inspired oxygen received packed erythrocytes when their Hct levels dropped below 40%. Otherwise the transfusion was performed when the Hct fell below 35% from the 2nd to the 4th week of life and below 23% thereafter.
Ronnestad 1995	Transfusions were given on the orders of the attending physician if Hb was < 90 g/L or otherwise as necessary according to signs and symptoms.
Samanci 1996	The need for packed erythrocyte transfusions was judged by the attending neonatologist. Guidelines for erythrocyte transfusions were developed as follows: 1) Hct of ≤ 22% and an absolute reticulocyte count of 100 000/microlitre; 2) Hct of ≤ 30% and a) tachycardia (> 180 beats/min) and tachypnoea (> 70 breaths/min) persisting for 24 hours; or b) 3 or more episodes of apnoea or bradycardia in 24 hours, not due to other causes and not responsive to methylxanthine treatment; or c) average weight gain of < 10 grams/day over a 1‐week period (infant tolerating full oral feed and receiving adequate calories); or d) undergoing surgery; 3) systemic infection associated with a sudden decrease in Hct.
Shannon 1991	Transfusions were ordered by the clinicians caring for each infant without consulting the investigators. Written guidelines for erythrocyte transfusions were developed for the nursery 1 year before the start of the study. A copy of these guidelines was taped to the bed of each study infant. In general, babies who were otherwise well received transfusions only if they had a Hct < 25% and signs referable to their anaemia, such as slowing in rate of growth, persistent severe tachycardia and tachypnoea, or worsening of episodes of apnoea and bradycardia.
Shannon 1992	See Shannon 1991 above.
Shannon 1995	Transfuse infants at Hct ≤ 20%: a) if asymptomatic with reticulocytes < 100 000/microlitre. Transfuse infants at Hct ≤ 30%: a) if receiving < 35% supplemental hood oxygen, b) if on CPAP or mechanical ventilation with mean airway pressure < 6 cm H₂O, c) if significant apnoea and bradycardia are noted (9 episodes in 12 hours or 2 episodes in 24 hours requiring bag‐and‐mask ventilation) while receiving therapeutic doses of methylxanthines, d) if heart rate > 180 beats/min or respiratory rate > 80 breaths /min persists for 24 hours, e) if weight gain < 10 grams/day is observed over 4 days while receiving ≥ 100 kcal/kg/day, f) if undergoing surgery. Transfuse for Hct ≤ 35%: a) if receiving > 35% supplemental hood oxygen, b) if intubated on CPAP or mechanical ventilation with mean airway pressure ≥ 6 to 8 cm H₂O. Do not transfuse: a) to replace blood removed for laboratory tests alone, b) for low Hct alone.
Whitehall 1999	Guidelines for red cell transfusions for anaemia of prematurity were based on the existing policy in the nursery, generally adopted by the neonatologists. They were as follows: I. Transfuse infants at Hb 80 g/L, (a) If reticulocyte count is < 4% and (b) If receiving supplemental oxygen > 30% or (c) If unexplained recurrent apnoea/bradycardia is noted (> 1 to 2/hour) or (d) If persistent tachycardia (heart rate > 170/min) or tachypnoea (respiratory rate > 60/min) is noted, or (e) If there is failure to gain weight or successive weight loss on weekly recordings for 3 consecutive weeks. In absence of a clear evidence in the literature justifying red cell transfusions at a Hb of 80 g/L in otherwise asymptomatic neonates who are failing to thrive, it was decided that failure to gain weight or successive weight loss on weekly recordings for 3 consecutive weeks was a fair and substantial clinical indicator of the need to transfuse. II. Transfuse infants at Hb 100 g/L, (a) If receiving supplemental oxygen 30% and (b) needing intermittent mandatory ventilation or continuous positive airway pressure by nasal prongs for recurrent (> 1 to 2 per hour), apnoea/bradycardia with saturations < 90% on the pulse oximeter. III. Transfuse Infants at Hb 120 g/L, (a) If receiving mechanical ventilatory support with mean airway pressure 10 cm H₂O and supplemental oxygen 30% during the acute phase of illness after birth.
Yamada 1999a	Conservative red blood cell transfusion guidelines were followed (details not presented, as we were unable to translate the information).
Yamada 1999b	Conservative red blood cell transfusion guidelines were followed (details not presented, as we were unable to translate the information).
^{CPAP: continuous positive airways pressure FIO2: fraction of inspired oxygen H2O: water Hb: haemoglobin Hct: haematocrit} ^{IMV: intermittent mandatory ventilation NICU: neonatal intensive care unit O2: oxygen}

Table 1. Transfusion guidelines

Ir a la tabla de la revisión

Comparison 1. Late initiation of EPO (8‐28 days) vs. placebo or no intervention

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Use of one or more red blood cell transfusions (low and high dose of EPO) Show forest plot	21	1202	Risk Ratio (M‐H, Fixed, 95% CI)	0.72 [0.65, 0.79]

2 Use of one or more red blood cell transfusions (high dose of EPO) Show forest plot	15	972	Risk Ratio (M‐H, Fixed, 95% CI)	0.77 [0.68, 0.86]

2.1 High dose iron	6	318	Risk Ratio (M‐H, Fixed, 95% CI)	0.74 [0.62, 0.88]
2.2 Low dose iron	9	654	Risk Ratio (M‐H, Fixed, 95% CI)	0.78 [0.67, 0.90]
3 Use of one or more red blood cell transfusions (low dose of EPO) Show forest plot	7	239	Risk Ratio (M‐H, Fixed, 95% CI)	0.53 [0.42, 0.67]

3.1 High dose of iron	3	77	Risk Ratio (M‐H, Fixed, 95% CI)	0.50 [0.31, 0.79]
3.2 Low dose of iron	4	162	Risk Ratio (M‐H, Fixed, 95% CI)	0.54 [0.41, 0.71]
4 Total volume (mL/kg) of red blood cells transfused per infant Show forest plot	5	197	Mean Difference (IV, Fixed, 95% CI)	‐1.61 [‐5.78, 2.57]

5 Number of red blood cell transfusions per infant Show forest plot	11	817	Mean Difference (IV, Fixed, 95% CI)	‐0.22 [‐0.38, ‐0.06]

6 Number of donors the infant was exposed to Show forest plot	2	165	Mean Difference (IV, Fixed, 95% CI)	0.45 [0.20, 0.69]

7 Mortality during initial hospital stay (all causes) Show forest plot	14	767	Risk Ratio (M‐H, Fixed, 95% CI)	0.82 [0.49, 1.39]

8 Retinopathy of prematurity (all stages or stage not reported) Show forest plot	3	404	Risk Ratio (M‐H, Fixed, 95% CI)	1.27 [0.99, 1.64]

9 Retinopathy of prematurity (stage ≥ 3) Show forest plot	3	442	Risk Ratio (M‐H, Fixed, 95% CI)	1.73 [0.92, 3.24]

10 Proven sepsis Show forest plot	5	551	Risk Ratio (M‐H, Fixed, 95% CI)	0.75 [0.52, 1.09]

11 Necrotising Enterocolitis ≥ Bell's stage 2 Show forest plot	6	656	Risk Ratio (M‐H, Random, 95% CI)	0.88 [0.45, 1.70]

12 Intraventricular haemorrhage all grades (or grade not specified) Show forest plot	4	454	Risk Ratio (M‐H, Fixed, 95% CI)	0.87 [0.53, 1.42]

13 Periventricular leukomalacia Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only

14 Bronchopulmonary dysplasia (supplementary oxygen at 28 days) Show forest plot	2	285	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [1.00, 1.55]

15 Bronchopulmonary dysplasia (supplementary oxygen at 36 weeks' postmenstrual age) Show forest plot	3	216	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.59, 1.35]

16 SIDS Show forest plot	6	363	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.25, 4.52]

17 Neutropenia Show forest plot	6	164	Risk Ratio (M‐H, Fixed, 95% CI)	0.28 [0.05, 1.54]

18 Hypertension Show forest plot	7	361	Risk Ratio (M‐H, Fixed, 95% CI)	1.20 [0.46, 3.14]

19 Length of hospital stay (days) Show forest plot	2	55	Mean Difference (IV, Fixed, 95% CI)	‐0.35 [‐12.83, 12.13]

20 Use of one or more red blood cell transfusions (secondary analysis based on study quality) Show forest plot	21	1202	Risk Ratio (M‐H, Fixed, 95% CI)	0.72 [0.65, 0.79]

20.1 HIgh quality studies	6	417	Risk Ratio (M‐H, Fixed, 95% CI)	0.83 [0.73, 0.95]
20.2 Studies of uncertain quality	15	785	Risk Ratio (M‐H, Fixed, 95% CI)	0.63 [0.54, 0.73]
21 Use of one or more red blood cell transfusions (secondary analysis based on RBC transfusion guidelines) Show forest plot	19	1120	Risk Ratio (M‐H, Fixed, 95% CI)	0.74 [0.67, 0.82]

21.1 Strict RBC transfusion guidelines	16	1023	Risk Ratio (M‐H, Fixed, 95% CI)	0.76 [0.69, 0.85]
21.2 No or less strict RBC guidelines	3	97	Risk Ratio (M‐H, Fixed, 95% CI)	0.25 [0.08, 0.77]

Comparison 1. Late initiation of EPO (8‐28 days) vs. placebo or no intervention

Ir a la tabla de la revisión

	Idioma de la Revisión Cochrane Escoja su idioma de preferencia para las revisiones Cochrane y otros contenidos. Las secciones sin una traducción aparecerán en inglés.

	Idioma de la web Escoja su idioma de preferencia para la web de la Biblioteca Cochrane.

Idioma de la Revisión Cochrane

Idioma de la web

Resumen

Antecedentes

Objetivos

Métodos de búsqueda

Criterios de selección

Obtención y análisis de los datos

Resultados principales

Conclusiones de los autores

PICO

PICO

Population

Intervention

Comparison

Outcome

Resumen en términos sencillos

Agentes estimulantes de la eritropoyesis tardía para prevenir la transfusión de glóbulos rojos en recién nacidos prematuros o de bajo peso al nacer

Resumen visual

Authors' conclusions

Implications for practice

Implications for research

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Primary objective

Secondary objective

Métodos

Obtención y análisis de los datos

Evaluación del riesgo de sesgo de los estudios incluidos

Results

Description of studies

Results of the search

Included studies

Excluded studies

Risk of bias in included studies

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Primary Outcome:

The use of one or more red blood cell transfusions (Outcome 1.1)

Subgroup analyses:

High dose of EPO (Outcome 1.2)

Low dose of EPO (Outcome 1.3)

Secondary outcomes:

The total volume (mL/kg) of blood transfused per infant (Outcome 1.4)

Number of red blood cell transfusions per infant (Outcome 1.5)

Number of donors to whom the infant was exposed (Outcome 1.6)

Mortality during initial hospital stay (all causes of mortality) (Outcome 1.7)

Retinopathy of prematurity (ROP) (all stages or stage not reported) (Outcome 1.8)

Retinopathy of prematurity (ROP) stage ≥ 3 (Outcome 1.9)

Proven sepsis (clinical symptoms and signs of sepsis and positive blood culture) (Outcome 1.10)

Necrotising enterocolitis (NEC) (Bell's stage II or higher) (Outcome 1.11)

Intraventricular haemorrhage (IVH); all grades (Outcome 1.12)

Periventricular leukomalacia (PVL); cystic changes in the periventricular areas (Outcome 1.13)

Bronchopulmonary dysplasia (BPD) (supplemental oxygen at 28 days of age)(Outcome 1.14)

Bronchopulmonary dysplasia (BPD) (supplemental oxygen at 36 weeks' postmenstrual age (PMA)) (Outcome 1.15)

Sudden infant death (SIDS) after discharge (Outcome 1.16)

Neutropenia (Outcome 1.17)

Hypertension (Outcome 1.18)

Length of hospital stay (days) (Outcome 1.19)

Long‐term outcomes assessed at any age beyond one year of age by a validated cognitive, motor, language, or behavioural/school/social interaction/adaptation test

Any side effects reported in the trials

Secondary (post hoc) analyses:

Use of one or more blood transfusions (secondary analysis based on quality) (Outcome 1.20)

Use of one or more blood transfusions (secondary analysis based on criteria for red blood cell transfusions) (Outcome 1.21; Outcome 1.21.1; Outcome 1.21.2)

Funnel plot

Enterally‐dosed EPO

Discussion

Summary of main results

Overall completeness and applicability of evidence

Quality of the evidence

Potential biases in the review process