Erythropoietin or darbepoetin for patients with cancer

Abstract

Background

Anaemia associated with cancer and cancer therapy is an important clinical factor in the treatment of malignant diseases. Therapeutic alternatives are recombinant human erythropoiesis stimulating agents (ESAs) and red blood cell transfusions.

Objectives

To assess the effects of ESAs to either prevent or treat anaemia in cancer patients.

Search methods

This is an update of a Cochrane review first published in 2004. We searched the Central Register of Controlled Trials (CENTRAL), MEDLINE and EMBASE and other databases. Searches were done for the periods 01/1985 to 12/2001 for the first review, 1/2002 to 04/2005 for the first update and to November 2011 for the current update. We also contacted experts in the field and pharmaceutical companies.

Selection criteria

Randomised controlled trials on managing anaemia in cancer patients receiving or not receiving anti‐cancer therapy that compared the use of ESAs (plus transfusion if needed).

Data collection and analysis

Several review authors assessed trial quality and extracted data. One review author assessed quality assessment and extracted data, a second review author checked for correctness.

Main results

This update of the systematic review includes a total of 91 trials with 20,102 participants. Use of ESAs significantly reduced the relative risk of red blood cell transfusions (risk ratio (RR) 0.65; 95% confidence interval (CI) 0.62 to 0.68, 70 trials, N = 16,093). On average, participants in the ESAs group received one unit of blood less than the control group (mean difference (MD) ‐0.98; 95% CI ‐1.17 to ‐0.78, 19 trials, N = 4,715). Haematological response was observed more often in participants receiving ESAs (RR 3.93; 95% CI 3.10 to 3.71, 31 trials, N = 6,413). There was suggestive evidence that ESAs may improve Quality of Life (QoL). There was strong evidence that ESAs increase mortality during active study period (hazard ratio (HR) 1.17; 95% CI 1.06 to 1.29, 70 trials, N = 15,935) and some evidence that ESAs decrease overall survival (HR 1.05; 95% CI 1.00 to 1.11, 78 trials, N = 19,003). The risk ratio for thromboembolic complications was increased in patients receiving ESAs compared to controls (RR 1.52, 95% CI 1.34 to 1.74; 57 trials, N = 15,498). ESAs may also increase the risk for hypertension (fixed‐effect model: RR 1.30; 95% CI 1.08 to 1.56; random‐effects model: RR 1.12; 95% CI 0.94 to 1.33, 31 trials, N = 7,228) and thrombocytopenia/haemorrhage (RR 1.21; 95% CI 1.04 to 1.42; 21 trials, N = 4,507). There was insufficient evidence to support an effect of ESA on tumour response (fixed‐effect RR 1.02; 95% CI 0.98 to 1.06, 15 trials, N = 5,012).

Authors' conclusions

ESAs reduce the need for red blood cell transfusions but increase the risk for thromboembolic events and deaths. There is suggestive evidence that ESAs may improve QoL. Whether and how ESAs affects tumour control remains uncertain. The increased risk of death and thromboembolic events should be balanced against the potential benefits of ESA treatment taking into account each patient’s clinical circumstances and preferences. More data are needed for the effect of these drugs on quality of life and tumour progression. Further research is needed to clarify cellular and molecular mechanisms and pathways of the effects of ESAs on thrombogenesis and their potential effects on tumour growth.

Plain language summary

Erythropoietin or darbepoetin for patients with cancer

Introduction

Researchers in The Cochrane Collaboration conducted a review of the effect of epoetin and darbepoetin for people with cancer. After searching for all relevant studies, they found 91 studies with up to 20,102 people. Their findings are summarized below:

What the research says:

In people with cancer‐related anaemia:

‐ Epoetin and darbepoetin decrease the need for red blood cell transfusions; however, they also increase the risk for hypertension, thromboembolic events and deaths

‐ It is not clear whether epoetin and darbepoetin improve quality of life, by making you feel less tired

When you have cancer, you often also have anaemia. Anemia means having lower than normal red blood cells in your blood. This might become even worse with cancer treatment such as chemotherapy and it is measured by the amount of haemoglobin in your red blood cells. As haemoglobin is responsible for carrying oxygen throughout your body, when you have anaemia you might experience symptoms such as extreme tiredness, shortness of breath, dizziness and chest pain. In order to treat anaemia, doctors often use red blood cell transfusions. Transfusions improve the symptoms of anaemia very quickly; however, they can have some infrequent complications like allergic reactions or transmission of infectious diseases.

Epoetin and darbepoetin belong to a group of medications called 'Erythropoiesis Stimulating Agents'. Erythropoietin is the name of a hormone produced mainly in the kidney, which takes part in the production of red blood cells. Epoetin and darbepoetin work in a similar way to this hormone to increase the number of red blood cells and treat anaemia. Epoetin and darbepoetin are not used as anti‐cancer therapy but as supportive treatment to treat anaemia caused by cancer or anticancer therapy. These drugs are marketed as Epogen®, Procrit®, (Eprex®), Recormon®, and Aranesp® and are given subcutaneously.

What happens to people with cancer related anaemia who take epoetin or darbepoetin:

‐ Twenty‐five out of 100 persons receiving epoetin or darbepoetin had to undergo red blood cell transfusions, compared to 39 out of 100 persons not receiving epoetin or darbepoetin.

‐ More people who received epoetin or darbepoetin died during and up to 30 days after the end of study compared with people who took placebo or underwent standard treatment. The increased risk for people taking epoetin or darbepoetin was 17%. One hundred and fourteen out of 1,000 persons receiving epoetin or darbepoetin died, compared with 98 out of 1,000 persons not receiving epoetin or darbepoetin. We could not identify particular characteristics of people or treatment strategies that increased or decreased the risk for dying.

‐ Concerning long‐term survival people taking epoetin or darbepoetin were 5% more at risk for dying than people taking placebo or receiving standard treatment.

‐ People receiving epoetin or darbepoetin rated their fatigue symptoms to be an average of 2.08 points improved on a scale of 0‐52 points after 3‐4 months, compared with people taking placebo or having standard treatment. This improvement, however, is less than the 3.0 point increase which is considered to be the minimum required for the patient to feel a difference in his experience of fatigue‐related symptoms using this scale.

‐ People taking epoetin or darbepoetin rated their fatigue and anaemia symptoms had to be an average of 6.14 points improved after three to four months, on a scale of 0‐80 points. This improvement is considered to reflect a positive change in the way patients experience their fatigue and anaemia related symptoms, as it is more than four to five points of increase which is the minimum required for this scale.

‐ Seven people out of 100 who took epoetin or darbepoetin suffered a thromboembolic event such as stroke and myocardial infarction compared with five people out of 100 who did not receive epoetin or darbepoetin.

‐ Six out of 100 people receiving epoetin or darbepoetin developed high blood pressure compared with four out of 100 people who took placebo or had standard care.

Summary of findings

Summary of findings for the main comparison. Erythropoietin or Darbepoetin for patients with cancer.

Erythropoietin or Darbepoetin for patients with cancer
Patient or population: patients with cancer Settings: Intervention: Erythropoietin or Darbepoetin
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Erythropoietin or Darbepoetin
Overall survival	Low		HR 1.05 (1 to 1.11)	19003 (78 studies)	⊕⊕⊕⊕ high
	50 per 1000	52 per 1000 (50 to 55)
	Moderate
	142 per 1000	149 per 1000 (142 to 156)
	High
	250 per 1000	261 per 1000 (250 to 273)
On‐study mortality Death occurring up to 30 days after active study protocol	Low		HR 1.17 (1.06 to 1.29)	15935 (70 studies)	⊕⊕⊕⊕ high
	10 per 1000	12 per 1000 (11 to 13)
	Moderate
	59 per 1000	69 per 1000 (62 to 75)
	High
	98 per 1000	114 per 1000 (104 to 125)
Thrombotic events	Low		RR 1.52 (1.33 to 1.73)	15278 (57 studies)	⊕⊕⊕⊝ moderate1
	20 per 1000	30 per 1000 (27 to 35)
	Moderate
	46 per 1000	70 per 1000 (61 to 80)
	High
	100 per 1000	152 per 1000 (133 to 173)
Quality of Life ‐ Change in FACT‐Fatigue (13 items) FACT‐F 13 sub‐scale		The mean quality of life ‐ change in fact‐fatigue (13 items) in the intervention groups was 2.08 higher (1.43 to 2.72 higher)		4965 (18 studies)	⊕⊕⊝⊝ low2,3
Quality of Life ‐ Change in FACT‐Anaemia (20 items) FACT‐An 20		The mean quality of life ‐ change in fact‐an (20 items) in the intervention groups was 6.14 higher (4.55 to 7.73 higher)		1085 (6 studies)	⊕⊕⊝⊝ low4,5
Participants receiving red blood cell transfusions	Low		RR 0.65 (0.62 to 0.68)	15877 (70 studies)	⊕⊕⊕⊝ moderate6
	300 per 1000	195 per 1000 (186 to 204)
	Moderate
	389 per 1000	253 per 1000 (241 to 265)
	High
	700 per 1000	455 per 1000 (434 to 476)
*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; RR: Risk ratio; HR: Hazard 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.

¹ A funnel plot analysis revealed a significant asymmetry, suggesting that negative results (in this case no thrombotic event) have been underreported 
 ² Only 18 out of a total of 91 studies assessing FACT‐13 reported this endpoint, which suggests some reporting bias, even if the funnel plot did not show an asymmetry. 13 more studies (i.e. studies that stated they either used FACT‐F or that they used FACT‐An 47 but do not report separately for FACT‐F or report it in a way that we cannot use. 
 ³ Overall, it seems that there is an effect to fatigue‐related symptoms for patients treated with ESAs compared to controls; this effect, however, did not reach the threshold for a clinically important difference defined as 3.0. 
 ⁴ Only 6 out of a total of 91 studies assessing FACT‐20 reported this endpoint, which suggests some reporting bias, even if the funnel plot did not show an asymmetry. Data from 19 studies addressing QoL can not used for review in cause of missing data (i.e.16 studies that stated they either used fact‐an or that they used fact‐an 47 but do not report separately for fact‐an). Out of these 4 studies which refer to “FACT‐An”/ to “fatigue sub‐scale”, were not useable because we cannot safely say to which instrument they are refer to. 
 ⁵ Overall, the effects of ESAs to fatigue and anaemia related symptoms seem to be beneficial and the difference between groups reaches both statistical and clinical significance (clinically important difference defined as 4‐5). 
 ⁶ A funnel plot analysis showed significant asymmetry between the studies, suggesting that negative results were under reported.

Background

Description of the condition

Anaemia, defined as a deficiency in the concentration of haemoglobin‐containing red blood cells, is a widely prevalent complication among cancer patients (Knight 2004). The prevalence of anaemia varies according to the type of neoplasia (Monnerat 1999). Patients with haematological malignancies frequently experience anaemia. At the time of diagnosis, 30% to 40% of patients with non‐Hodgkin's lymphoma (NHL) or Hodgkin's lymphoma (HD) and up to 70% of patients with multiple myeloma are anaemic; the figures are even higher in myelodysplastic syndromes (Garton 1995; Greenberg 1994). The extent of anaemia is also influenced by the type of cytostatic treatment. It is known that the proportion of anaemic patients with solid cancers rises up to 50% after chemotherapy or combined radiochemotherapy (Dalton 1998; Harrison 2001; Ludwig 1998; Reed 1994).

The National Cancer Institute and others have agreed to use the following classification for anaemia based on haemoglobin (Hb) values (Groopman 1999):

• Grade 0, within normal limits, Hb values are 12.0 to 16.0 g/dL for women and 14.0 to 18.0 g/dL for men

• Grade 1, mild (Hb 10 g/dL to normal limits)

• Grade 2, moderate (Hb 8.0 to 10.0 g/dL)

• Grade 3, serious/severe (Hb 6.5 to 7.9 g/dL)

• Grade 4, life threatening (Hb less than 6.5 g/dL).

The pathophysiology of tumour anaemia is multi‐factorial (Mercadante 2000). In advanced stages of haematological malignancies, bone marrow involvement with malignant cells often leads to progressive anaemia. After exclusion of other causes, e.g. iron or vitamin deficiencies, occult bleeding, autoimmune haemolysis or pure red blood cell aplasia, anaemia can be related to "anaemia of chronic disorders". It is characterised by a close interaction between the tumour cell population and the immune system, leading to the activation of macrophages and increased expression of various cytokines, especially Interferon‐g, Interleukin‐1, Interleukin‐6 and tumour necrosis factor. This is followed by insufficient endogenous erythropoietin synthesis, suppressed differentiation of erythroid precursor cells in the bone marrow and alterations of iron metabolism (Johnson 1990; Ludwig 1998; Nowrousian 2002a; Spivak 1994a). For example, the elevation of interleukin 6 (Il‐6) leads to an up‐regulation of Hepcidin and a diminished erythropoietin‐synthesis. Hepcidin is one of the mediators that are released in inflammatory processes. It causes the destruction of another protein called ferroportin, which is important for the transport of Fe2+ from the enterocyte and the macrophage into the blood (Ganz 2011). The anaemia of chronic disorders, or chronic tumour anaemia is the most common type in patients with malignant disease, although it is often aggravated by chemo‐ or radiotherapy. In particular, platinum‐based chemotherapy regimens may diminish endogenous erythropoietin production by damaging renal tubular cells (Wood 1995).

Manifestation and severity of anaemia vary considerably among individual patients. Mild‐to‐moderate anaemia can cause typical symptoms including headache, palpitations, tachycardia and shortness of breath. Chronic anaemia may result in severe organ damage affecting the cardiovascular system, immune system, lungs, kidneys, muscles and the central nervous system (Ludwig 2001; Nissenson 1992). In addition to physical symptoms, the subjective impact of cancer‐related anaemia on quality of life (QoL), mental health and social activities may be substantial. Clinical studies have reported correlations between Hb levels and quality of life domains, for example mood, appetite (Leitgeb 1994), and the ability to work (Cella 1998; Thomas 1998).

Another aspect of anaemia in patients with malignant disease is the effect on the tumour itself. For malignant diseases such as Hodgkin's Disease (HD), chronic lymphocytic leukaemia (CLL), cervical carcinoma and cancer of the head and neck, anaemia has been reported to be an independent prognostic factor (Caro 2001; Hasenclever 1998; Nowrousian 2002b; Van Belle 2003). There is evidence that anaemia, with the consequence of increased tumour hypoxia, might result in a poorer response to radio‐ or chemotherapy (Hockel 1993; Nordsmark 1996; Van Belle 2003; Vaupel 1989; Vaupel 2000). Severe symptoms of anaemia may also necessitate dose reduction or delay of chemotherapy. All these factors may lead to a higher tumour burden and a decreased overall survival (Glaser 2001; Grau 2000; Knocke 1999). These observations have generated the hypothesis that strategies to diminish cancer‐related anaemia might alleviate not only anaemia‐related symptoms and improve quality of life, but also might improve tumour response and extend overall survival time. However, randomised controlled trials testing this hypothesis have generated conflicting evidence (Antonadou 2001; Henke 2003; Leyland‐Jones 2005).

Historically, blood transfusion was the conventional treatment of choice for severe cancer‐related anaemia. The literature reports a critical degree of anaemia as a Hb level below 8 g/dL, while mild‐to‐moderate anaemia (Hb level 8‐10 g/dL) usually has been left untreated (Carson 2012; Cella 1999; Glaspy 1997a; Henry 1992; Koeller 1998). Although homologous blood transfusion is the fastest method to alleviate symptoms, short‐ and long‐term risks exist (Engert 2000). Potential complications associated with blood transfusion are transmission of infectious diseases, transfusion reactions, allo‐immunisation, over‐transfusion and immune modulation with possible adverse effects on tumour growth (Landers 1996). The risk of severe infectious complications of blood transfusions are 1: 30.000 to 1:250.000 units of blood transfused for Hepatitis B, 1:30.000 to 1:150.000 for Hepatitis C and 1:250.000 to 1:1.000.000 for HIV (Goodnough 1999). Other complications such as allergic reactions and death due to major incompatibilities are infrequent but may be of concern (Williamson 1999). The development of intensified anti‐neoplastic therapies has increased the risk for blood transfusion, prompting oncologists to weigh the advantages and disadvantages of this treatment.

Description of the intervention

Recombinant human erythropoietin is a treatment option for cancer‐related anaemia. Human erythropoietin is an acidic glycoprotein hormone. Approximately 90% of the hormone is synthesised in the kidney and 10% in the liver (Koury 1988; Koury 1991). Basal production maintains a relatively constant plasma concentration of erythropoietin in individuals, within a range from 9 to 26 mU/mL. Tissue hypoxia is the most important trigger for increased synthesis. The effects of erythropoietin in the bone marrow are mediated by a specific surface receptor located mainly on erythroid progenitor and precursor cells (D´Andrea 1989; Spivak 1994b). Two major functions of erythropoietin are described: stimulating progenitor cell proliferation and maintaining their viability (Koury 1990).

Several short‐ and long‐lasting forms of recombinant human erythropoiesis‐stimulating agents (ESAs) are available, including Epoetin‐a and Epoetin‐ß and darbepoetin‐a (Darbepo) (Glaspy 2003; Halstenson 1991; Hedenus 2002; Joy 2002; Storring 1998; Vansteenkiste 2002). Recently, novel ESA molecules, such as continuous erythropoietin receptor activator (CERA) (Gascon 2010b), and biosimilars (epoetin theta, epoetin delta) have been developed (Jelkmann 2010). Clinical trials directly comparing Epo and Darbepo have been published and suggest that Epo and Darbepo are similarly effective with regard to Hb response and proportion of patients transfused (Alexopoulos 2004; Glaspy 2003; Schwartzberg 2004; Waltzman 2004). Based on these data, it seemed justified to combine both short‐ and long‐lasting erythropoietin formulations in one meta‐analysis.

How the intervention might work

Erythropoietin was first approved for the treatment of anaemia in chronic kidney failure. In 1990, erythropoietin was introduced in cancer therapy regimens for patients with multiple myeloma. A pilot study showed haematological response rates of 85% and an improved performance status (Ludwig 1990). Adverse effects such as hypertension, headaches and thrombotic events conclusively attributable to erythropoietin treatment were reported in very few patients (Beguin 1998). However, several randomised controlled trials reported increased incidences of thrombotic events, tumour progression and deaths (Hedenus 2003; Henke 2003; Leyland‐Jones 2005; Overgaard 2009; Smith 2008; Thomas 2008; Untch 2011_1; Wright 2007).

Why it is important to do this review

Since ESAs were licensed for the treatment of anaemia in cancer patients, more than 20 systematic reviews and meta‐analyses have been published. While there is clear evidence that ESAs reduce the need for red blood cell transfusions, increase Hb levels and also increase the risk for thromboembolic complications, there is ongoing debate on the effects of ESAs on QoL, tumour progression and mortality (Aapro 2006; Aapro 2008; Aapro 2009; Bennett 2008; Bottomley 2002; Cella 2004; Clark 2002; Devon 2009; Glaspy 2010; Hedenus 2005; Hellström‐L 1995; Jones 2004; Kimel 2008; Lambin 2009; Ludwig 2009; Minton 2008; Minton 2010; Quirt 2003; Quirt 2005; Ross 2003; Ross 2006; Ross 2007; Seidenfeld 2001b; Seidenfeld 2001a; Seidenfeld 2006; Tonelli 2009; Wilson 2007).

The first Cochrane review (published in 2004 in The Cochrane Library and 2005 as print publication) on this subject included 27 randomised controlled studies on erythropoietin with 3,287 adults, published between 1985 and April 2002 (Bohlius 2005) in collaboration with authors from a previous review conducted for AHRQ (Seidenfeld 2001a;Seidenfeld 2001b). For the first update, we included trials published between 2002 and 2005 in collaboration with an independent review team from the Department of Public Health and Epidemiology at the University of Birmingham and the support of the UK Department of Health (Bohlius 2006a). In 2009, we conducted a separate meta‐analysis based on individual patient data (IPD), which focused on on‐study mortality and overall survival in all cancer patients and in patients receiving chemotherapy (Bohlius 2009a; Bohlius 2009b).Given that this analysis was restricted to survival outcomes, we present here the second update of the Cochane Review, including all relevant outcomes and integrating results generated for the IPD analysis as well as recently published trials.

Objectives

To systematically review evidence on the outcomes of using recombinant human erythropoietin or darbepoetin to prevent or alleviate anaemia in patients with malignant disease, with respect to haematological response, red blood cell transfusion need, changes in quality of life, tumour response, on‐study mortality, overall survival and adverse events.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials using recombinant human erythropoietin or darbepoetin to treat or prevent anaemia in patients with malignant disease. Placebo control, as opposed to "no treatment", was not required for inclusion, but was considered in evaluating study quality. We excluded trials in which patients were allocated by a quasi‐random method, for example date of birth or day of month, as we considered this study design to be of poor quality leading to unreliable results. We included only trials with more than 10 documented participants in each study arm or relevant stratum. We excluded interim analyses of ongoing studies. We included studies that were stopped or suspended prematurely. The searches did not include language restrictions.

Types of participants

We included only participants diagnosed with malignant disease, using clinical and histological/cytological criteria, regardless of type or stage of the disease or previous therapy. All study participants had to be anaemic or at risk for anaemia from chemotherapy, radiotherapy or combination therapy, or the underlying malignant disease. Other causes of anaemia, such as haemolysis, iron deficiency and occult bleeding, had to have been excluded. We included studies with patients of every age. We excluded trials if more than 80% of participants were diagnosed with an acute leukaemia.

Types of interventions

We included studies evaluating the use of recombinant human erythropoiesis‐stimulating agents (ESAs) to prevent or reduce anaemia in cancer patients, given singly or concomitantly with chemotherapy, radiotherapy or combination therapy. ESAs could be administered subcutaneously or intravenously. In previous versions of this review, we required dosages of at least 300 U/kg body weight per week (epoetin‐a and beta) given for at least four weeks. For the current update we removed this criterion and we included studies or study arms with low dosages as well. We allowed dose adaptation of ESAs depending on haematological response. Concomitant supportive treatments, e.g. granulocyte colony‐stimulating factors (G‐CSF), had to be given equally in all study arms. In previous versions of this review, this criterion applied to iron supplementation as well. However, for the current update, we have changed this criterion and included trials using iron supplementation in the experimental but not in the control arm as well. We excluded trials on high‐dose myeloablative chemotherapy regimens followed by bone marrow or peripheral blood stem cell transplantation, as well as trials using erythropoietin for short‐term preoperative treatment to correct anaemia or to support collection of autologous blood prior to cancer surgery.

Included trials addressed one or more of the following comparisons of interest:

1. ESAs versus placebo or no treatment.

2. ESAs and red blood cell (RBC) transfusion as necessary versus observation and RBC transfusion as necessary, alone or with placebo.

3. ESAs plus conventional‐dose cancer therapy (non‐myeloablative chemotherapy and/or radiotherapy) versus identical therapy alone or with placebo.

4. ESAs and RBC transfusion as necessary plus conventional‐dose cancer therapy versus observation and RBC transfusion as necessary plus identical therapy, alone or with placebo.

Types of outcome measures

Primary outcomes

• Haematological response

Measured as a binary outcome (proportion of patients with an increase in Hb level of 2 g/dL or more, or an increase in haematocrit of 6% points or more; measured as continuous data (change in Hb level from baseline until end of study). Of note, in the previous review we had restricted this outcome to studies with baseline Hb levels < 12 g/dL; in the current update we have removed this restriction.

• Patients receiving RBC transfusions

• Number of RBC units transfused per patient

• Overall survival

• On‐study mortality. Of note, we added on‐study mortality as a new outcome to the current update of the review.

Secondary outcomes

• Tumour response (complete response)

• Changes in quality of life including cancer‐related fatigue and anaemia symptoms, measured with the Functional Assessment of Cancer Therapy Anaemia (FACT‐An) Total scale (47 items), Anaemia sub‐scale (20 items), or Fatigue sub‐scale (13 items). We have chosen this instrument and its sub‐scales because they have been widely used in ESA trials and have good responsiveness to change and good convergent and discriminant validity (Cella 1997; Cella 2002; Cella 2007; Yellen 1997).

• Adverse events (thromboembolic events, hypertension, haemorrhage/thrombocytopenia, rash/irritation/pruritus, seizures)

Search methods for identification of studies

We used the following sources to identify the studies for this updated review:

1. Previous Cochrane reviews on the effects of ESAs on cancer patients, mainly Bohlius 2006a and Bohlius 2009a; Bohlius 2009b.

2. Electronic search in bibliographic databases.

3. Conference Proceedings.

4. ODAC Documents 2004, 2007 and 2008 (see next section).

5. Reference lists of other systematic reviews and meta‐analyses.

Electronic searches  Search strategies have been adapted from those suggested in the Cochrane Handbook for Systemtatic Reviews of Interventions (Lefebvre 2011). For detailed search strategies for each database, please refer to Appendix 1; Appendix 2; Appendix 3.For previous searches for this review see previous version of the review (Bohlius 2006a). For the current version, we identified relevant trials in any language through electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and EMBASE. We conducted the initial search for the period 2004 to September 2009. We updated the search using the same search strategy in January 2011 and in November 2011. We did not use language restrictions. The full search strategy is on file.

Electronic searches

Search strategies have been adapted from those suggested in the Cochrane Handbook for Systemtatic Reviews of Interventions (Lefebvre 2011). For detailed search strategies for each database, please refer to Appendix 1; Appendix 2; Appendix 3.

For previous searches for this review, see previous version of the review (Bohlius 2006a). For the current version, we identified relevant trials in any language through electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and EMBASE. We conducted the initial search for the period 2004 to September 2009. We updated the search using the same search strategy in January 2011 and in November 2011. We did not use language restrictions. The full search strategy is on file.

Searching other resources

Conference Proceedings

We searched conference proceedings of the American Society of Clinical Oncology, American Society of Hematology and European Society of Medical Oncology from January 1989 to December 2001 for the original Cochrane review; from January 2000 to December 2004 for the first update; and from January 2005 to November 2011 for the present update. We conducted searches of conference proceedings online, with CD‐ROMs or by handsearching.

Ongoing trials

For the previous versions of this review, we searched various sources for ongoing studies. For the current update we did not search specifically for ongoing trials in online registries.

Contact with authors

For previous versions of this review, we contacted groups or individuals as well as pharmaceutical companies who conducted randomised trials on recombinant human erythropoietin in cancer patients. For the current review, we contacted authors only to clarify eligibility of studies, if necessary.

Reference lists 
 We checked reference lists of identified guidelines, systematic reviews and clinical trials for additional information.

Oncology Drug Advisory Committee Documents

For the current update, we evaluated documents presented at the Oncology Drug Advisory Committee (ODAC) hearing at the USA Food and Drug Administration (FDA), held in May 2004, May 2007 and March 2008. These documents include briefing documents plus additional power point presentations prepared by medical review authors of the FDA (FDA ODAC 2007; FDA ODAC 2008; Luksenburg 2004), as well as documents and additional power point presentations prepared by the companies Roche, Johnson & Johnson and Amgen (Amgen ODAC 2004; Amgen ODAC 2007; Amgen ODAC 2008; Johnson & Johnson ODAC 2004; Roche ODAC 2004). All of these documents are publicly available at FDA web pages.

Data collection and analysis

Selection of studies

Several review authors screened titles and abstracts of studies identified from the above sources according to the eligibility criteria stated previously. We did this step in duplicate. For the first Cochrane review this was undertaken by Simon Langensiepen and Julia Bohlius, for the first update this was undertaken by Jayne Wilson, Sunsan Brunskill, Julia Bohlius, Olaf Weingart and Sven Trelle. For the current update this was done by Annette Mettler, Julia Bohlius, Nadège Robert and Thomy Tonia. If eligibility could not be adequately assessed by screening the title and abstract, we obtained a full text version of the study for assessment. Studies that appeared to meet the inclusion criteria in the initial screening were further assessed for eligibility using a form containing the following questions.

1. Is the study described as randomised?

2. Did the participants in the study have a previously treated or untreated malignant disease?

3. Were the participants anaemic or at risk for anaemia from chemotherapy and/or radiotherapy or their malignant disease?

4. Was one group given Epoetin‐a or Epoetin‐ß or Darbepoetin‐alfa or any other ESA subcutaneously or intravenously?

5. Did the control group receive the same care (e.g. chemotherapy and supportive therapies) with or without placebo?

6. Did the study document relevant outcome measures?

To be eligible, studies had to meet all of the criteria stated above. If there was insufficient information to judge eligibility, we contacted the first author of the report for clarification. We resolved any disagreements between the review authors by discussion. We identified duplicate reports.

Data extraction and management

Data extraction was performed by one review author and facts were checked by a second review author. For the original Cochrane review this was done by Simon Langensiepen and Julia Bohlius. For the first update all publications until September 2004 were extracted by Julia Bohlius, Jayne Wilson and Susan Brunskill. For the current update, data were extracted by Julia Bohlius, Olaf Weingart, Annette Mettler, Nadège Robert and Thomy Tonia. For the previous version QoL data were extracted by Jayne Wilson, Susan Brunskill and Chris Hyde. For the current version, QoL data were extracted by Thomy Tonia, Annette Mettler, Nadège Robert and Julia Bohlius. For data extractions we used a standardised data extraction form. This form included the following items.

• General information: title, authors, source, contact address of corresponding author, year of publication, any duplicate publications, trial setting, recruitment dates, funding.

• Trial characteristics: design, method of randomisation, concealment of allocation, blinding of patients and clinicians.

• Patients: sampling, inclusion and exclusion criteria, sample size, baseline characteristics, similarity of groups at baseline, diagnostic criteria, withdrawals, losses to follow‐up.

• Interventions: placebo use, dose, dosing regimen, duration, route of administration, RBC transfusion trigger, co‐medications with dose, route and timing

• Outcomes: outcomes as specified above.

We resolved disagreements arising at any stage by discussion and consensus.

Referencing of studies 
 If we identified several publications for one study, we extracted the data from the most recent publication and amended this with information from other publications. We used the following suffixes to indicate the source of data.

• We indicated data that were taken from either full text or abstract publication or based on personal communication with author name and year of publication

• We cited data that were taken from the individual patient data meta‐analysis by Bohlius 2009a;Bohlius 2009b using the reference for the original study and adding the suffix “IPD”. For example, we cited survival data for the Hedenus 2003 which we had taken from the individual patient data meta‐analysis by Bohlius 2009a; Bohlius 2009b as "Hedenus 2003 IPD"

• We assigned the following suffixes to data taken from an ODAC presentation:

• • J&J 2004 if we took the data from a J&J presentation for the ODAC 2004 hearing

  • J&J 2007 if we took the data from a J&J presentation for the ODAC 2007 hearing

  • Roche 2004 if we took the data from a Roche presentation for the ODAC 2004 hearing

  • FDA 2004 if we took the data from a FDA presentation for the ODAC 2004 hearing

  • FDA 2007 if we took the data from a FDA presentation for the ODAC 2007 hearing

  • Amgen 2007 if we took the data from an Amgen presentation for the ODAC 2007 hearing

• For each unique study we assigned a five digit random number, entered for each reference in the "other" field. These numbers are identical to the numbers assigned to the studies in the IPD review by Bohlius 2009a; Bohlius 2009b. We assigned new five digit numbers for studies which we had not included in the IPD review.

• For studies comparing more than one experimental arm to one control arm, we assigned a separate reference for each study arm. We labelled these study arms with the author and publication year of the main publication and added the suffixes a; b; c etc. For example, the study by Cazzola 1995 compared four different experimental study arms with one control group. The four different study arms are listed in the included studies and the data analyses as Cazzola 1995a; Cazzola 1995b; Cazzola 1995c and Cazzola 1995d.

Due to this referencing system a given study may appear more than once in the list of included studies. The total number of unique studies included is 91. For these 91 studies, a total of 198 study identifiers are listed in the sections Included studies and Characteristics of included studies.

Methods for QoL

In the previous review, we included any validated QoL instrument. For the current update, we restricted the analysis to studies meeting the inclusion criteria and also reporting QoL data using one of the following instruments.

a) FACT‐F 13: Functional Assessment of Cancer Therapy Fatigue sub‐scale, consisting of 13 fatigue‐specific questions

b) FACT‐An 20: Functional Assessment of Cancer Therapy Anaemia sub‐scale, consisting of 20 questions (13 from FACT‐F, plus 7 anaemia specific questions)

c) FACT‐An 47: Functional Assessment of Cancer Therapy Anaemia total score, consisting of 47 questions (20 from the FACT‐An sub‐scale plus 27 from the FACT‐General scale)

In the previous update, we excluded studies that did not clearly state the number of participants contributing to QoL data. For the current update, we decided to also include studies that did not report the exact number of participants contributing QoL data, in order to get as much information as possible. We included all studies reporting change between baseline and end of treatment per group. We limited the included QoL data to scores of the full instruments or sub‐scales that have been validated as stand‐alone instruments.

Rather than focusing on statistical significance, meaningful interpretation of QoL results emphasizes the degree to which differences between treatment and control arms exceed a Clinically Important Difference (CID) previously established for the QoL instruments utilized. For FACT‐Fatigue, this has been estimated to be three (Cella 2002) and for FACT‐An 20 four to five (Cella, personal communication, March 2010). For FACT‐An 47 a CID has not yet been established.

Assessment of risk of bias in included studies

Quality Assessment 
 Two review authors independently assessed the full text articles of the eligible studies for quality. For the original Cochrane review this was undertaken by Simon Langensiepen and Julia Bohlius. For the first update this was done by Julia Bohlius and either Jayne Wilson or Susan Brunskill for the trials published until September 2004, and Julia Bohlius and Sven Trelle for trials published between September 2004 and April 2005. For the current update, this assessment was conducted by Annette Mettler, Nadège Robert and Thomy Tonia. For the original Cochrane review, we contacted all first authors or sponsoring pharmaceutical companies of the included trials in order to obtain detailed information on the study design. For the previous and current update (years 2002 to November 2011) this was not undertaken due to time limitations. However, for the IPD review (Bohlius 2009a; Bohlius 2009b) we had access to the clinical study reports of the included studies and used these to assess the quality of studies. We integrated these assessments in the current review update. For all other studies, we took this information from publications.

To assess the methodological quality and the risk of bias, we designed a quality assessment form specifically for the topic of this review, according to the recommendations in Chapter Eight of the Cochrane Handbook for Systematic Reviews ( Higgins 2011a). The sources we used for designing this form are: Alderson 2005; Jüni 2001; Moher 2001; Verhagen 1998) and the form contains the following questions.

1. Was allocation truly random?

2. Was the treatment allocation concealed?

3. Were study participants blinded (masked) to the treatment they received?

4. Were study clinicians blinded (masked) to the treatment received by individual study participants?

5. Did the analysis include an intention‐to‐treat (ITT) analysis?

We excluded studies from the analysis if they were not truly randomised or had inadequately concealed treatment allocation, e.g. if participants were assigned to treatments in alternate order, or according to their birth dates, or the day of the week they arrived at the treatment centre. We tested the effect of individual quality variables (allocation, blinding, ITT) in subgroup analyses. Because of the problematic use of quality summary scores, we did not use summary scores (Jüni 1999; Schulz 1995). We assessed baseline participant characteristics to see if the groups were balanced at baseline. We also assessed whether the number of patient withdrawals, dropouts and lost to follow‐up was reported for each study group; however, we did not use this information for the analyses.

For the present update we applied additional quality criteria to assess the quality of studies reporting data on tumour control. These criteria were as follows.

1. The study population had to be homogenous, i.e. all participants had to have the same tumour type and ‐ if relevant ‐ the same tumour stage. Alternatively, the study had to be stratified by tumour type or tumour stage. 

2. The participants of the study had to receive a predefined, identical anticancer therapy. As above, we considered a study to meet this criterion if the study was stratified by treatment.

3. The study had to be designed to assess tumour response or tumour control prospectively or tumour control/response had to be the primary or secondary study outcome.

Dealing with missing data

Incomplete reporting of data

As suggested in Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), we took the following steps in addressing missing data.

For studies that did not report the number of patients evaluated for a given outcome, we used the number of patients randomised per study arm as denominator. In studies where only the total number of population was reported (and not the number of patients per arm) and where the randomisation was 1:1, we assumed that randomisation was 1:1 for a given outcome as well, thus assuming that the outcome was calculated using ITT analysis.

Binary data: if only percentages but not absolute number of events were reported, we used percentages to calculate numerators.

Continous data: If estimates for mean and standard deviations (SD) were not reported, we used the methods published by Hozo et al to convert median and range estimates of outcomes such as Hb change into mean and SD (Hozo 2005). If mean change was not reported, we calculated this as the difference between end of treatment and baseline value. If SDs were not reported, we estimated them from standard errors, confidence intervals or ranges, whenever reported. If numerical data were not reported, we estimated means and/ or SDs from graphs or figures. If SDs and means were reported for subgroups only, we calculated a pooled SD by pooling the SDs of the two different subgroups. For some studies it was not clear whether the reported numbers were referring to SDs or SEs. To maximise the likelihood that our assumptions were valid, we assumed that the numbers were SDs and converted them to SEs and vice versa. We then assumed that the numbers were referring to SDs or SEs depending on which one seemed more likely for this outcome, as suggested in Deeks 1997. If mean change between baseline and end of treatment and corresponding SD was not reported, they were calculated by using a correlation coefficient. In addition, we used secondary sources to identify data that were not reported in the original publication. These secondary sources included ODAC briefing documents and other meta‐analyses. For example for QoL we retrieved data from secondary publications, that had attempted to retrieve missing information from authors (Minton 2008; Minton 2010) or imputed data (Tonelli 2009). To account for all these calculations, we conducted a subgroup analysis taking into account the source of data, as well as a sensitivity analysis differentiating between the studies for which we had to impute data in some way and those for which we did not impute any data.

Handling of discrepant data: If a study was published in several different publications, reports and presentations, we extracted the most recent or most comprehensive data. We compared the data of one study taken from different sources. If the data from different sources were discrepant, we applied the following rules to decide which data to use for analysis.

• Most complete data set, e.g. where the sample size is largest OR data with consistent outcome definitions across trials were chosen for analysis.

• If for one study outcome two different analyses were available (e.g. adjusted versus unadjusted analysis), we chose the unadjusted data for analysis.

Age: if age was not reported but the cancer was typical for adults, e.g. breast cancer or non‐small cell lung cancer (NSCLC), we classified the study as conducted in an “adult" population.

Assessment of heterogeneity

As suggested in Chapter 9 of the Cochrane Handbook of Systematic Reviews on Interventions (Deeks 2011), we explored potential causes of heterogeneity by performing sensitivity and subgroup analyses for primary and selected secondary outcome measures (see below). We used the P value of the homogeneity test and the I² statistic only to describe the extent of heterogeneity inherent in a meta‐analysis.

Assessment of reporting biases

In meta‐analyses with at least 10 trials, we generated a funnel plot and performed a linear regression test (Egger 1997) to examine the potential presence of bias. We considered a P value of less than 0.1 as significant for the linear regression test (Sterne 2011).

Data synthesis

We performed analyses according to the recommendations of Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011).

We assumed a fixed‐effect model for all meta‐analyses. For binary data, we used the risk ratio as a measure of treatment effect and we used the Mantel‐Haenszel method for pooling. We used the estimated overall RR and a range of plausible values for the baseline‐risk to estimate numbers needed to benefit (NNTB) and numbers needed to harm (NNTH). For continuous data, we calculated the mean differences (MD) if the outcome was measured on the same scale in all trials. For QoL we combined only identical scales and sub‐scales in a given meta‐analysis. For time to event data, i.e. overall survival, we calculated hazard ratios (HR) based on individual patient data (IPD). If IPD were not available, we calculated the HR from published reports including secondary analyses (Bohlius 2009a; Bohlius 2009b), using methods described in Parmar et al (Parmar 1998) or binary mortality data. We performed all analyses using Review Manager (RevMan) 5.1; we used the statistical software package R (Ihaka 1996) for additional analyses that could not be done with RevMan 5.1.

In addition to subgroup analyses, we conducted random‐effects meta‐regression (see Section 9.6.4, Cochrane Handbook) for the following outcomes: Hb response, Hb change, participants receiving red blood cell transfusions, FACT‐Fatigue, FACT‐An 47 using R function rma.uni in R package metafor (Viechtbauer 2010). We used the DerSimonian‐Laird method to estimate the between‐study variance in meta‐regression. In model selection, we considered all covariates showing a significant effect (P < 0.05) in a univariate analysis. We restricted the analysis to studies providing information on all variables that were statistically significant in univariate analyses. For model selection, we used a back‐wise selection method; we removed consecutively the covariate with the largest P value as long as the P value was larger than 0.05.

In several studies different ESAs, dosages, and routes or schedules of administration were compared with one control group (Cazzola 1995; Henke 1999; Kotasek 2003; Kunikane 2001; Osterborg 1996; Smith 2003; Suzuki 2008; Ten Bokkel 1998; Thatcher 1999). For each multi‐arm study, we divided and randomly assigned control patients to the corresponding number of separate treatment groups for entry into RevMan (base model). As this might influence the weighting of the studies and thus the pooled results, we alternatively merged the two (or more) active arms of one study into one single experimental arm and compared it to the entire control group. We compared and described results for each outcome.

Subgroup analysis and investigation of heterogeneity

We conducted subgroup analyses using the following factors, if appropriate.·        

• Hb at study entry (Hb level < 10 g/dL versus 10 to 12 g/dL versus >12 g/dL versus unclear/not reported)

• Solid tumours versus haematological malignancies (excluding myelodysplastic syndrome (MDS)) versus MDS versus mixed versus unclear versus not reported)

• Age (children versus adults)

• Age (only children versus adults > 18 years versus > 70% non‐elderly adults aged 18‐65 versus only non‐elderly adults versus > 70% elderly adults aged > 65 years versus only elderly adults aged > 65 years)

• Type of treatment given (chemotherapy versus radiotherapy/radiochemotherapy versus no therapy versus other)

• Type of treatment given (> 70% of patients receiving platinum‐based chemotherapy versus < 70% of patients receiving platinum‐based chemotherapy versus chemotherapy without platinum versus chemotherapy some (not quantified) patients receiving platinum versus chemotherapy no details given versus radiochemotherapy versus radiotherapy versus no therapy versus other). We categorised studies with less than 70% of patients receiving chemotherapy as “other” (P‐174 IPD; Rose 1994 IPD)

• Short‐lasting ESA versus long‐lasting ESA

• Duration of ESA medication (six to nine weeks versus 12 to 16 weeks versus more than 20 weeks)

• Iron supplementation (fixed iron supplementation versus iron as necessary versus no explicit statement/no iron versus explicit NO iron versus iron handled differently in study arm)

• Study quality parameters (concealment of allocation, masking, intention‐to‐treat analysis)

• Source of data (full text publications versus abstract publications versus unreported data versus data reported at FDA/ODAC hearing versus other)

Compared to the previous version of this review, we added new subgroups to differentiate with more detail different age groups and different anti‐cancer therapy groups. For the subgroup “iron supplementation”, we added the categories “iron given differently in both study arms”, because of the change in the inclusion criteria described above and “explicitly stated NO iron”. However, no study was included in the latter subgroup. For the subgroup “publication”, we added the category “other", containing clinical trial results from sources that did not fit any other category.  We dropped the following subgroup analyses as they did not seem to be relevant any longer or insufficient information was available: Number of drop outs documented; use of G‐CSF.

Sensitivity analysis

To test the robustness of the results, we conducted random‐effects meta‐analyses. We reported the estimates of the random‐effects only a) if they showed a difference to the fixed‐effect model or b) if they were necessary to allow for comparison with other meta‐analyses in the discussion section. We explored the influence of single large studies and the influence of different data sets, e.g. adjusted versus unadjusted data.

Results

Description of studies

Results of the search

Our initial literature search in March 2001 retrieved 1,592 references. For the first update of this review the Birmingham team identified and screened another 1,859 references. For the current update, we identified and screened a total of 2,207 references (999 at the first search on September 2009 and 1208 at the two searches conducted in 2011 (January and November)). For details, see the PRISMA flow diagram in Figure 1.

1.

Study flow diagram.

Included studies

The previous update of the Cochrane review (Bohlius 2006a) evaluated 57 studies with 9,353 participants. In the current update we include a total of 91 studies with 20,102 participants. All included trials were reported in English. Details are displayed in the Characteristics of included studies table. For those 91 studies, 198 references were included, see details as described in "Referencing of studies" in the Methods section.

Studies that were previously excluded and are now included

Four of the previously excluded studies are included in the current update, due to the change in the iron supplementation rule: for this update, studies are also included if they had different iron supplementation policies between the different study arms (Blohmer 2011; Moebus 2007; Rosen 2003; Sweeney 1998). We had previously excluded another study by mistake, we have now included this study (Gebbia 2003).

Studies that were previously included and are now excluded

We excluded two of the previously included studies (Henze 2002; Vadhan‐Raj 2004 ) because a substantial number of participants underwent major surgical procedures during ESA treatment.

Studies that were previously ongoing and are now included

We included four studies that were previously ongoing (Aapro 2008; Antonadou 2001; Charu 2007; Pronzato 2010).

Updated data for already included studies

If new publications supplemented or updated information from earlier sources, we added the new data and cited the new source as the study name. This led to the following changes: we replaced Coiffier 2001 with Boogaerts 2003, Janinis 2003 with Christodoulou 2009, EPO‐CAN‐15 with Goss 2005, N93 004 with Grote 2005, EPO‐GBR‐7 with Hoskin 2009, Machtay 2004 with Machtay 2007, Milroy 2003 with Milroy 2011, Savonije 2004 with Savonije 2005, GOG0109 with Thomas 2008, EPO‐CAN‐20 with Wright 2007, we amended Razzouk 2004 with Razzouk 2006, and Pronzato 2002 with Pronzato 2010.

Newly identified studies

Screening the references of other reviews on this subject, led to the identification and inclusion of three new studies (Gebbia 2003; ML17616 2006; ML17620 2006 ). We included three studies (EPO‐GER‐20 IPD; Kotasek 2002 IPD; OBE/EPO‐INT‐03 IPD) that were previously unpublished and retrieved for the IPD review (Bohlius 2009a; Bohlius 2009b). We identified 23 additional new studies that were published after the search for the previous review was conducted. We included these studies in the present update. From those, we had also included 11 in the IPD review (Aapro 2008; Debus 2006 J&J 2007; Gordon 2008; Hernandez 2009; Milroy 2011; Pirker 2008; Ray‐Coquard 2009; Smith 2008; Strauss 2008; Untch 2011_1; Wilkinson 2006) and we identified twelve from the literature search update (Engert 2010; Fujisaka 2011; Gupta 2009; Katakami 2008; Krzakowski 2008; Mystakidou 2005; Overgaard 2009; Suzuki 2008b; Tsuboi 2009; Tjulandin 2010; Tjulandin 2011; Winquist 2009).

Other changes

In the previous review (Bohlius 2006a), we had excluded studies and study arms with very low ESA dosages, in the current update we included all studies and study arms regardless of ESA dosage, for example see Cazzola 1995.

Study population

Hb level: We grouped studies by mean or median baseline Hb level at study entry. Thirty studies examined a study population with mean or median Hb below 10 g/dL at study entry, 38 had a baseline Hb between 10 and 12 g/dL and 18 studies sought to prevent anaemia, thus Hb at baseline was > 12 g/dL. Because of missing information, five studies could not be categorized Debus 2006 J&J 2004; EPO‐GER‐20 IPD; ML17620 2006; OBE/EPO‐INT‐03 IPD; P‐174 J&J 2004). Trials that directly compared the outcomes of initiating erythropoietin treatment at alternative Hb thresholds were not included in the present review.

Disease: Fifty‐nine studies analysed participants with solid tumours only, 12 studies included patients with haematological malignancies only, two trials included exclusively patients with MDS (Italian 1998; Thompson 2000) while 18 trials included patients with both solid tumours and haematological malignancies.

Treatment: The majority of participants received concomitant chemotherapy, which was given in 63 studies. In seven studies participants were treated with radiotherapy and in eight with radiochemotherapy. In nine trials no concomitant anticancer therapy was given. Finally, we categorized the type of anti‐cancer therapy administered as "unclear" in one study (Winquist 2009) and three studies with less than 70% of patients given chemotherapy as "other" (P‐174 J&J 2004; Rose 1994; Rosenzweig 2004).

Age: All studies but one (Razzouk 2006) evaluated adult participants.

Intervention: All trials compared erythropoiesis‐stimulating treatment initiated at study entry (plus RBC transfusion if necessary) with observation and transfusion of RBCs when the patient's Hb level fell below a defined threshold or at the discretion of the treating physician.

Study drug: Short‐lasting erythropoietins (including epoetin alpha, epoetin beta, epoetin theta) were administered in 76 studies and darbepoetin in 15 studies.

Duration: Duration of study medication was up to nine weeks in 17 studies, between 12 and 16 weeks in 50 studies and more than 17 weeks in 21 studies. In three studies the duration of study drug administration was unclear or not reported (EPO‐GER‐20 IPD; EPO‐INT‐1 J&J 2004; OBE/EPO‐INT‐03 IPD).

Route of administration: In all but three studies erythropoietin was administered subcutaneously. In two studies erythropoietin was given intravenously (Razzouk 2006; Wurnig 1996). Another study compared intravenous with subcutaneous administration (Henke 1999).

More details are provided in the table Characteristics of included studies.

Excluded studies

Overall, we excluded 227 trials, for reasons documented in the Characteristics of excluded studies. Thirty studies were excluded as ongoing trials (see next section) while one study was not retrievable.The largest group was excluded for not being randomised controlled trials (98 studies). We excluded 48 studies because participants of both study arms received ESAs, while we excluded 13 studies since they randomised less than 10 participants per study arm, our current threshold for inclusion. We excluded two previously included studies (Henze 2002; Vadhan‐Raj 2004) from this update, since most of their participants underwent surgery during the study; the same reason for exclusion was applied to another eight studies. Three studies were excluded due to the participants not having cancer, while three more due to the participants having acute leukaemia. Finally, 21 studies were excluded for other reasons, documented in the Characteristics of excluded studies.

Ongoing trials 
 Overall, we identified 30 studies that are currently ongoing. For the previous Cochrane update, 29 trials were identified as ongoing. Since then, five were completed and are included in the present meta‐analysis (Aapro 2008; Antonadou 2001; Charu 2007; Pronzato 2010; Thomas 2008). We identified a mistake in the ongoing trials of the previous version: namely CDR0000068669 and EORTC 22996‐24002 were in fact the same study, now identified as Lambin 2006.Three of the previously ongoing trials are now excluded (Elsaid 2001; Miller 2004; Steensma 2011). Additionally, note that we were uncertain whether H Thomas 1997 and R Thomas 2002 are the same studies as the included Thomas 2002. We did not contact the authors and we, therefore, refer to these three documents as separate studies, with the first two still considered as ongoing trials. The remaining 20 trials are still ongoing.

We newly identified another 10 trials ongoing. We had excluded one (Gamucci 1993) in the previous Cochrane update and we identified another six (Boehrer 2010; Delarue 2009; Gascon 2010; Ghavamzadeh 2010; Liang 2009; Yousseff 2011) by the 2011 literature search. We identified the remaining three (Nitz 2008; Park 1996 and Rexer 2006) by other sources.

Risk of bias in included studies

For risk of bias table see Figure 2.

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study. If for a given study more than one reference was included we reported our judgement only for the main reference and left the other references of the same study empty.

Allocation

Randomisation and concealment of allocation.

All included studies were described by the authors as randomised. For 27 studies we had some information on the methods used for randomisation, for 64 studies details were unavailable or the reporting was unclear. For 50 studies we judged the method for concealing allocation of treatment to be adequate. In 41 studies the method for allocation concealment could not be determined.

Blinding

Masking 
 Forty‐six trials used a placebo control, 45 trials did not.

Incomplete outcome data

Most studies included intention‐to‐treat analyses in their reports, or excluded less than 10% of the patients randomised in the study from the analysis. However, the number of participants evaluated in a study varied between the outcomes assessed. Therefore, for each of the outcomes, we made an assessment as to whether or not the analysis was based on intention‐to‐treat or excluded less than 10% of the participants initially included. This information is displayed in the specific outcomes section in RevMan.

Selective reporting

For the first version of the Cochrane review, we contacted first authors to obtain unreported data on study design, participant characteristics and selected outcome data. For 19 of the 27 trials included in the first Cochrane review additional unpublished data were provided by the authors or pharmaceutical companies (Abels 1993; Cascinu 1994; Case 1993; Cazzola 1995; Boogaerts Coiffier 2001; Dammacco 2001; Del Mastro 1997; Henry 1995; Italian 1998; Kurz 1997; Littlewood 2001; Oberhoff 1998; Osterborg 1996; Osterborg 2002; Rose 1994; Ten Bokkel 1998; Thatcher 1999; Thompson 2000; Throuvalas 2000). For the update of the present review, authors of published reports were not contacted to obtain missing information.

Publication bias 
 Funnel plot analyses were performed to investigate publication bias or other biases and are reported in the specific outcome sections.

Reporting bias 
 No single outcome was reported consistently by all studies (N = 91) included in this systematic review. Some outcomes were estimated based on large proportions of the 91 studies and 20,102 participants included:

95% for overall survival (19,003/20,102), 79% for on‐study mortality (15,935/20,102), 80% for risk of transfusions (16,093/20,102), 77% for risk of thromboembolic events (15,498/20,102) and 58% for change in Hb (11,609/20,102). For other outcomes, fewer participants were included in the meta‐analyses: 36% (7,228/20,102) for hypertension, 32% (6,413/20,102) for haematologic response, 25% (5,012/20,102) for complete tumour response, 25% (5,012/20,102) for number of units transfused and 22% (4,507/20,102) for thrombocytopenia/haemorrhage.

For some outcomes only small proportions of participants could be evaluated, which questions the validity of the results achieved: 14% (2,890/20.102) for seizure, 12% (2,485/20,102) for rash.

Statistically significant differences between data taken from full text publications, abstract publications and unpublished data were found in the subgroups analyses for Hb response, Hb change, risk for transfusions and number of units transfused.

Other potential sources of bias

For some studies more than one source of data was available (e.g. unpublished data from the authors, FDA reports, IPD review). As for intention‐to‐treat mentioned above, we recorded the source of data separately for each outcome.

Effects of interventions

See: Table 1

Primary outcomes

(1) Haematological response

We defined this binary outcome as the proportion of participants with an increase in haemoglobin (Hb) level of 2 g/dL or more, or increase in haematocrit of six percentage points or more, unrelated to transfusion.

Overall, 31 trials including 6,413 participants were analysed. Of those studies, 22 trials including 4,307 participants were included in the previous Cochrane reviews (1985 to 2006). Nine trials (N= 2,106) were newly identified and added to the analysis (Aapro 2008; Charu 2007; Milroy 2011; ML17616 2006; ML17620 2006; Razzouk 2006; Suzuki 2008; Tjulandin 2010; Tjulandin 2011). As some of the trials (Cazzola 1995; Hedenus 2002; Kotasek 2003; Osterborg 1996; Smith 2003; Suzuki 2008; Tjulandin 2010) were split into subsets for analysis purposes, the number of trials displayed is 46.

We observed haematological response in 2,050 out of 3,710 participants in the erythropoietin and darbepoetin groups compared with 434 of 2,703 in the control groups, corresponding to a risk ratio (RR) for haematological response under the treatment with erythropoietin or darbepoetin of (RR 3.39; 95% confidence interval (CI) 3.10 to 3.71). There was significant statistical heterogeneity between the trials (I²= 53%), indicating that variation between trials in the magnitude of ESA effects on haematological response was larger than what would be expected from chance alone. However, all studies indicated a beneficial effect of ESAs with regard to haematological response. The funnel plot analysis was asymmetric (P = 0.0015), suggesting that beneficial effects were over reported.

In seven studies (Cazzola 1995; Hedenus 2002; Kotasek 2003; Osterborg 1996; Smith 2003; Suzuki 2008; Tjulandin 2010), two or more epoetin dosages or preparations were compared with one control group. Merging the active arms into one experimental arm for each of those four studies did not influence the overall result markedly (RR 3.42; 95% CI 3.12 to 3.74). Compared with the previous review (RR= 3.43 (95% CI 3.07 to 3.84, 22 trials, N = 4,307, Bohlius 2006a) neither the point estimate, nor the confidence interval have changed substantially.

To identify the source(s) of heterogeneity, subgroup analyses were performed. Univariate analyses identified significant differences (P < 0.05) between subgroups for baseline Hb level, age, type of anti‐cancer therapy, iron supplementation, duration of treatment, placebo control and type of publication. The backward selection resulted in a multivariate model containing the covariates Hb at baseline, age and iron supplementation. The final model is presented in Additional Table 2. For each combination of Hb at baseline, age and type of iron supplementation the risk ratio can be calculated from Table 2. For example, the logarithm of the risk ratio for a trial including adults with Hb between 10 and 12 g/dL at baseline and receiving iron as necessary is Intercept + adults + Hb 10‐12 g/dL + iron given as necessary = 0.81 + 0.62 + 0.35 = 1.78. Accordingly, the risk ratio is 5.93.

1. Haematologic response: results of meta‐regression analysis.

Variable	log (effect size)	standard error	P value
Intercept	0.81	0.1189	<0.0001
Hb baseline 10‐12 g/dL	0.62	0.1430	<0.0001
Hb baseline > 12 g/dL	0.85	0.4694	0.0688
Children	‐0.68	0.1653	<0.0001
Iron given differently in both study arms	‐0.64	0.2348	0.0068
Iron supplementation as necessary	0.35	0.1315	0.0081

Hb: haemoglobin

We calculated numbers needed to benefit (NNTB) for several hypothetical baseline risks. In a population with an underlying risk [likelihood] of 6% to achieve haematological response, the NNTB would be 6.97 (95% CI 6.15 to 7.94); thus, about seven patients would need to be treated to achieve one additional Hb responder. In a population with an underlying risk of 16% the NNTB would be 2.62 (95% CI 2.31 to 2.98); thus, about three patients would need to be treated to achieve one additional Hb responder. In a population with a hypothetical baseline risk of 30% the NNTB would be 1.39 (95% CI 1.23 to 1.59); thus, one to two patients would need to be treated to achieve one additional Hb responder.

Overall, the analysis confirms prior evidence that ESAs yield haematological response in most but not all patients with malignant disease.

(2) Change in haemoglobin level from baseline until end of study

Fifty‐six studies including 11,609 participants reported the hb change from the start until the end of the study. Fifteen of those were included in the 2006 update (Bohlius 2006a). As some of the trials had multiple experimental arms (Cazzola 1995; Hedenus 2002; Henke 1999; Kotasek 2003; Krzakowski 2008; Kunikane 2001; Osterborg 1996; Smith 2003; Ten Bokkel 1998; Tjulandin 2010), the number of trials displayed in MetaView is 75. For this update, we included not only studies that reported the mean Hb change from baseline, but also studies that reported baseline and end of treatment (EOT) Hb values only, see Methods section.

The meta‐analysis showed that the mean difference was 1.57 (95% CI 1.51 to 1.62), showing a statistically significant difference in favour of treatment. There was, however, substantial heterogeneity between the trials (test for heterogeneity I² = 87%), although all but one study indicated a beneficial effect of ESAs with regard to Hb change. Funnel plot analysis did show some evidence for significant asymmetry (P = 0.037), suggesting that beneficial findings were over reported. Merging each multi‐arm trial into a single data set (i.e. a two‐arm trial) did not change the results (mean difference (MD) 1.56, 95% CI 1.51 to 1.62).

Examining the single studies, most of them reported a statistically significant change in Hb level for participants treated with erythropoietin or darbepoetin. A few studies, however, did not report a significant effect on Hb change (Cazzola 1995; Leyland‐Jones 2005; Razzouk 2006). Overall, the MD ranged from (MD ‐0.06; 95% CI ‐1.77 to 1.65, Cazzola 1995a) to (MD 3.30; 95% CI 1.13 to 5.47, Henke 1999c).

To identify the source(s) of heterogeneity, we performed subgroup analyses. Univariate analyses identified significant differences between subgroups for baseline Hb level, different malignancies, age, type of anti‐cancer therapy, short‐lasting versus long‐lasting ESA, duration of ESA treatment, iron supplementation, masking, intention‐to‐treat and type of publication. The backward selection resulted in a multivariate model containing the covariates type of ESA and age. The final model is presented in Additional Table 3 For each combination of type of ESA and age (adults or children) the mean difference can be calculated from Table 3. For example, the mean differences for a trial in adult patients receiving short‐lasting ESA is = Intercept + short lasting ESA = 1.15 + 0.56 = 1.71.

2. Hb change: results of meta‐regression analysis.

Variable	mean difference	standard error	P value
Intercept	1.15	0.1792	<0.0001
Children	‐1.41	0.6911	0.0414
Short acting ESA	0.56	0.2041	0.0060

ESA: erythropoiesis stimulating agent

Overall, there is a statistically significant effect of ESAs on Hb change; compared with controls patients receiving ESAs achieve on average an increase of Hb levels of 1.57 g/dL from baseline to end of treatment (between 1.51 and 1.62 g/dL in 95% of patients).

(3) Patients receiving RBC transfusions

Overall, the updated analysis included 70 trials with 16,093 participants. Of those, 42 trials with 6,510 participants were included in the previous Cochrane review (Bohlius 2006a). Thirty trials were newly identified and added to the analysis. (Aapro 2008; Blohmer 2011; Charu 2007; Engert 2010; EPO‐INT‐3 J&J 2004; Fujisaka 2011; Gebbia 2003; Gordon 2008; Goss 2005; Grote 2005; Gupta 2009; Hernandez 2009; Katakami 2008; Krzakowski 2008; Leyland‐Jones 2005; Milroy 2011; Moebus 2007; O'Shaughnessy 2005; Pronzato 2010; Pirker 2008; Ray‐Coquard 2009; Rosen 2003; Smith 2008; Strauss 2008; Tjulandin 2010; Tjulandin 2011; Tsuboi 2009; Untch 2011_1; Wilkinson 2006; Wright 2007). Two studies that were included in the previous review (Henze 2002; Vadhan‐Raj 2004 ) were excluded from the current version because study participants received major surgery during ESA treatment. As some trials with multiple experimental arms were split into subsets (Cazzola 1995; Hedenus 2002; Kotasek 2003; Krzakowski 2008; Kunikane 2001; Osterborg 1996; Smith 2003;Ten Bokkel 1998; Thatcher 1999; Tjulandin 2010) the number of trials displayed is 88.

The risk ratio to receive red blood cell transfusions was statistically significantly reduced in the study groups receiving ESAs by 35% (RR 0.65, 95% CI 0.62 to 0.68). There was statistically significant heterogeneity between the trials (I² = 60%) indicating that variation in the effect of ESAs between trials was larger than would have resulted from chance alone. However, the majority of studies indicated a beneficial effect of ESAs with regard to the need of red blood cell transfusions. A funnel plot analysis showed significant asymmetry between the studies (P < 0.00001), suggesting that beneficial findings were over reported. Ten studies compared two or more ESA dosages/formulations with one control group. Merging the active arms of each study into a single experimental arm and comparing it to the study's entire control group did not substantially change the overall result (RR 0.65; 95% CI 0.62 to 0.68). Compared with the previous version of this review, the results did not change markedly (Bohlius 2006a: RR 0.64, 95% CI 0.60 to 0.68, 42 trials, N = 6,510).

To identify the source(s) of heterogeneity, we performed subgroup analyses. Univariate analyses identified significant differences (P < 0.05) between subgroups for baseline Hb level, different malignancies, age, type of anti‐cancer therapy, iron supplementation, concealment of allocation, placebo control, intention‐to‐treat analysis and type of publication. The backward selection resulted in a multivariate model containing the covariates Hb at baseline and underlying malignancy. The final model is presented in Additional Table 4. For each combination of Hb level and type of malignancy, the risk ratio can be calculated from Table 4. For example, the logarithm of the risk ratio for a trial in patients with solid tumours and baseline Hb level 10‐12 g/dL is = Intercept + solid + Hb 10 to 12 g/dL = ‐0.22 + ‐0.39 + ‐0.15 = ‐0.76. Accordingly, the risk ratio is 0.47.

3. Participants receiving RBC transfusions: results of meta‐regression analysis.

Variable	log(effect size)	Standard error	P value
Intercept	‐0.22	0.0506	<0.0001
Hb 10 ‐ 12 g/dL	‐0.15	0.0650	0.0254
Hb > 12 g/dL	0.07	0.0601	0.2774
MDS	0.01	0.1004	0.8967
Solid and haematological tumours	‐0.03	0.0648	0.6726
Solid tumours	‐0.39	0.0637	<0.0001

MDS: myelodysplastic syndrome 
 RBC: red blood cell

To estimate the absolute effectiveness of erythropoietin, we applied the overall risk ratio of (RR 0.65; 95% CI 0.62 to 0.68) to a range of plausible values for the baseline‐risk. In a hypothesised population with an estimated risk of 30% to require RBC transfusions the NNTB is 9.52 (95% CI 8.77 to 10.42): about nine to 10 patients would need to receive ESAs to spare one patient from RBC transfusion. In a hypothesised population with an estimated risk of 50% to require RBC transfusions, the NNTB is 5.71 (95% CI 5.26 to 6.25): about five to six patients would need to receive erythropoietin to spare one patient from RBC transfusion. In a hypothesised population with an estimated risk of 70% to require RBC transfusions the NNTB is 4.08 (95% CI 3.76 to 4.46). In this setting about four patients would need to receive erythropoietin to spare one patient from RBC transfusion.

Overall, the data confirm results from prior analyses that ESAs reduce the risk ratio to receive RBC transfusions in patients with malignant disease. The effect size might be influenced by the underlying disease.

(4) Number of red blood cell units transfused

Overall, 19 studies evaluating a total of 4,715 patients are included in this update. Of those, 14 studies with 2,353 patients were included in the previous review (Bohlius 2006a). Five trials with 2,362 participants were newly identified and added to the analysis (Engert 2009; Grote 2005; Hernandez 2009; Savonije 2005; Thatcher 1999). As four multi‐arm studies were split into subsets (Ten Bokkel 1998; Thatcher 1999; Osterborg 1996; Cazzola 1995), the number of studies displayed in Meta‐View is 25.

The overall mean difference showed a statistically significant benefit for participants receiving ESAs (MD ‐0.98; 95% CI ‐1.17 to ‐0.78): the ESA group received on average 0.98 units of blood less per participant than the control group, who received an average of 3.65 units. Overall, there was moderate statistical heterogeneity between the trials (I² = 30%). All but one study indicated a reduced need of red blood cell units in patients receiving ESAs compared to controls. A funnel plot analysis did not show statistically significant asymmetry (P = 0.558). In four studies, two or more different ESA dosages were compared with one control group (Cazzola 1995; Osterborg 1996; Ten Bokkel 1998; Thatcher 1999). Merging the active arms of each study into a single experimental arm and comparing each to that study's entire control group did not substantially change the overall result (MD ‐0.98; 95% CI ‐1.17 to ‐0.78). Compared with the previous review (Bohlius 2006a: MD ‐1.05; 95% CI ‐1.32 to ‐0.78, 14 trials, N = 2,353), the overall result did not change markedly.

Subgroup analyses were conducted for all the comparisons and were found to be statistically significant for different age groups (test between subgroups P = 0.02), duration of ESA therapy (test between subgroups P = 0.03) and type of publication (test between subgroups P = 0.04). However, the absolute differences between subgroups were small.

Overall, the analysis suggests that ESAs modestly but statistically significantly reduces the number of RBC units transfused per patient.

(5) Overall survival

We defined overall survival as longest follow‐up available. Overall survival data were available from a total of 78 trials including 19,003 participants. Of those, 42 trials including 8,167 participants were included in the 2006 Cochrane review (Bohlius 2006a). Of these, we excluded one study which we had included in the previous review Vadhan‐Raj 2004 from the current version because study participants received major surgery during ESA treatment. For 30 previously included studies data were updated with newly available information from the individual patient meta‐analysis (Bohlius 2009a; Bohlius 2009b). Thirty‐seven trials were newly identified and added to the analysis Aapro 2008; Antonadou 2001; Blohmer 2011; Charu 2007 IPD; Christodoulou 2009; Debus 2006 IPD; Engert 2010; EPO‐GER‐20 IPD; Fujisaka 2011; Gordon 2008 IPD; Gupta 2009; Hernandez 2009; Huddart 2002 IPD; Kotasek 2002 IPD; Krzakowski 2008; Milroy 2003 IPD; ML17616 2006; ML17620 2006; Moebus 2007 IPD; Mystakidou 2005; OBE/EPO‐INT‐03 IPD; Overgaard 2009; Pirker 2008 IPD; Pronzato 2010 IPD; Quirt 1996 IPD; Ray‐Coquard 2009 IPD; Rosen 2003; Strauss 2008 IPD; Strauss 2008 IPD; Sweeney 1998; Thomas 2002 IPD; Tjulandin 2010; Tjulandin 2011; Tsuboi 2009; Untch 2011_2; Wilkinson 2006 IPD; Winquist 2009. Two studies (Krzakowski 2008; Tjulandin 2010) were split into subsets and four studies (Cascinu 1994; Hedenus 2002; Kurz 1997; Sweeney 1998) reported zero events; as a result the overall number of studies displayed in Meta‐View is 80.

The overall estimate is a hazard ratio (HR) of 1.05 (95% CI 1.00 to 1.11) in favour of placebo/no treatment. The heterogeneity between the trials was low, with an I² of 21%. Funnel plot analysis did not suggest asymmetry (P = 0.92).

In two studies (Krzakowski 2008; Tjulandin 2010), two different ESA dosages or formulations were compared with one control group. Merging the active arms into one single experimental arm compared with the entire control group did not influence the overall result (HR 1.05; 95% CI 1.00 to 1.11).

Although no statistically significant heterogeneity was apparent, we conducted subgroup analysis to explore the underlying clinical heterogeneity to assess the influence of clinical differences between the studies. Tests for heterogeneity between subgroups showed statistically significant differences for the analyses of baseline Hb levels (P = 0.02), iron supplementation (P = 0.005) and intention‐to‐treat analysis (P = 0.02). However, when excluding studies with unclear values for the subgroup analyses conducted, none of the analyses remained statistically significant. Statistically significant differences were not detected for any of the other subgroup analyses conducted.

We further investigated how single large studies influenced the overall results. There was no single study that contributed more than 10% weight to the overall analysis. There were four studies which each contributed more than 5% weight to the analysis (Aapro 2008 IPD; Debus 2006 IPD; Pirker 2008 IPD; Smith 2008 IPD). Taken together these four studies contributed 27% weight to the overall analysis. Of those, two suggested a negative impact on survival (Aapro 2008 IPD; Smith 2008 IPD) and two suggested a beneficial impact on overall survival (Debus 2006 IPD; Pirker 2008 IPD). Exclusion of the two unfavourable studies resulted in an HR of 1.04 (95% CI 0.98 to 1.10). Exclusion of the two favourable studies resulted in an HR of 1.08 (95% CI 1.02 to 1.14). We also investigated the effects of adjusted and unadjusted data where different results were reported (Henke 2003; Littlewood 2001). Using adjusted instead of unadjusted data for the Henke 2003 and Littlewood 2001 studies did not change the overall result (HR 1.05, 95% CI 1.00 to 1.11; adjusted data).

In summary, there is no evidence that erythropoietin or darbepoetin improves overall survival. Based on the data available it was not possible to clearly identify a subgroup of participants that was at higher or lower risk to experience detrimental effects from ESAs.

(6) On‐study mortality

We defined on‐study mortality as deaths occurring up to 30 days after the active study period. Such data were available for 78 studies including 15,935 patients. Two studies (Krzakowski 2008; Tjulandin 2010) were split into subsets and eight studies (Cascinu 1994; Del Mastro 1997; Hedenus 2002; Kurz 1997; Moebus 2007 IPD; Strauss 2008 IPD; Sweeney 1998; Untch 2008 IPD) reported zero events; as a result, the overall number of studies displayed in Meta‐View is 72.

The overall estimate showed an HR of 1.17 (HR 1.17; 95% CI 1.06 to 1.29) in favour of control. There was no heterogeneity between the studies (I² = 0%). A funnel plot analysis did not show evidence for significant asymmetry (P = 0.693).

Although there was no evidence for any statistical heterogeneity between the studies (I² = 0%), we conducted predefined subgroup analyses. Univariate analysis identified statistically significant differences only for intention‐to‐treat analyses (P = 0.04). No robust statistically significant differences were identified for any of the other subgroups of interest (i.e. baseline Hb level, type of malignancy, duration of treatment, type of anti‐cancer therapy, age, iron supplementation, type of publication, epoetin versus darbepoetin, type of data, concealment of allocation and masking).

We conducted a sensitivity analysis including eight additional studies. For seven of those (Antonadou 2001; Blohmer 2011; Christodoulou 2009; Engert 2009; Gupta 2009; Overgaard 2009; Winquist 2009), we calculated on‐study mortality from Kaplan Meier curves for overall survival and for the remaining study (Tsuboi 2009), we used long‐term mortality data. This sensitivity analysis yielded similar results; with an HR of 1.16 (95% CI 1.05 to 1.27, 78 trials, N=19,018).

We further investigated how single large studies influenced the overall results. There were two studies (Leyland‐Jones 2005 IPD; Smith 2008 IPD) which each contributed more than 10% weight to the analysis. Taken together these four studies contributed 28% weight to the overall analysis. Both studies suggested a negative impact on mortality. Exclusion of these two unfavourable studies resulted in an HR of 1.09 (95% CI 0.97 to 1.23).

Overall, there is evidence that ESA treatment increases mortality in cancer patients during active study period compared with controls.

Secondary outcomes

(7) Tumour response (complete response)

Tumour response (complete response) data were available from a total of 15 trials including 5,012 participants. Of these, 11 trials were included in the previous Cochrane review (Bohlius 2006a). Two previously included studies (Bamias 2003; Vadhan‐Raj 2004) were excluded from the current update since they did not explicitly state that they evaluated complete tumour response. We replaced previous data with updated study results where available (Grote 2005; Hoskin 2009; Machtay 2007). We included four additional studies published since 2006 (Engert 2009; Strauss 2008; Untch 2011_1; Wilkinson 2006). Since two multi‐arm trials were split into subsets (Cazzola 1995; Ten Bokkel 1998) the number of trials displayed in MetaView is 19.

The overall estimate shows a risk ratio of 1.02 (RR 1.02; 95% CI 0.98 to 1.06) that was not statistically significant. There was no significant heterogeneity between the trials (I² = 0%). Funnel plot analysis did not show strong evidence for asymmetry (P = 0.149). Merging the multi‐arm trial into one data set did not change the overall result (RR 1.02; 95% CI 0.98 to 1.06, 15 trials). Compared to the previous report (RR fixed‐effect: 1.12; 95% CI 1.01 to 1.23, random‐effect: 1.09, 95% CI 0.94 to 1.26, 13 trials , N = 2833), the updated review shows more conservative results. Only five of the included studies (Engert 2009; Hoskin 2009 GBR‐7 FDA 04; Machtay 2007; Strauss 2008; Untch 2011_1) met our specific quality criteria for assessment of tumour response (see Methods section). For the five studies meeting these criteria, the risk ratio was 0.99 (95% CI 0.95 to 1.02, five studies, N=2,476). For the remaining 10 studies with low quality the point estimate for RR suggested there might be a benefit for patients receiving ESAs, but the confidence interval did not reach statistical significance (RR 1.15, 95% CI 0.98 to 1.36, 10 trials, N = 2,536), with some evidence for a difference between subgroups tested (P = 0.07).

In conclusion, the data available suggest that ESAs do not have a beneficial effect on tumour control. The data are insufficient to exclude detrimental effects.

(8) Changes of health‐related quality of life

Twenty‐three studies including 5,584 patients reported results on QoL, as measured with FACT‐F, FACT‐An 20 or FACT‐An 47. In order to perform a meta‐analysis for the FACT measures, we had to extract means and SDs. In cases where no numerical data were given, we calculated means and/ or SDs from graphs or figures (Boogaerts 2003; Charu 2007; Gordon 2008; Hedenus 2003). In one trial (Christodoulou 2009) where SDs were reported separately for platinum and non platinum chemotherapy, we calculated a pooled SD by pooling the SDs of the two different subgroups. For the same study we also imputed the mean and the SD for changes between baseline and end of treatment, by using a correlation coefficient. For one study with missing SDs (Littlewood 2001) and no other way of imputing them, we used the SDs that have been already published by another meta‐analysis (Tonelli 2009). According to the authors of this meta‐analysis, the SDs were calculated as follows: for FACT‐fatigue they imputed the SD using the baseline SD reported in another publication for the same study (Fairclough 2003) and for the FACT‐An the SD was imputed using the average of the SDs from other studies who reported a SD for FACT‐An (personal communication with Dr. Lloyd, February 2010). For two further studies (Kotasek 2003; Vansteenkiste 2002), we used the mean and SDs reported in a meta‐analysis by Minton and colleagues (Minton 2008). These data were not available in the publications and were obtained by the authors of the meta‐analysis from the authors of the original studies or the pharmaceutical companies (personal communication with Dr. Minton, February 2010). To account for all these calculations, we conducted a sensitivity analysis, differentiating between the studies for which we had to impute data in some way and those for which we did not impute any data. One study (Christodoulou 2009) reported results separately for participants treated with platinum‐based chemotherapy and those treated with non‐platinum‐based chemotherapy. It is therefore included in both the platinum and non‐platinum chemotherapy subgroups for all the QoL outcomes below.

a) FACT‐F 13 sub‐scale

A total of 18 studies (4,965 patients) reported data for this outcome. The MD was 2.08 (95% CI 1.43 to 2.72). Heterogeneity between the included studies was moderate (I² = 53%). A funnel plot analysis showed significant asymmetry (P = 0.02772) with over reporting of studies that showed beneficial effects of ESAs.

The beneficial effect of ESAs on QoL measured with FACT‐F 13 was significantly larger in patients receiving chemotherapy than in those receiving radiotherapy or no anticancer therapy (P < 0.0001). Patients with Hb levels below 12 g/dL also had significantly (P = 0.0025) larger effects compared to patients with baseline Hb level > 12 g/dL, however, only one trial using radiotherapy had Hb levels > 12 g/dL at baseline (Hoskin 2009). The effect of epoetin was significantly (P < 0.0004) different from darbepoetin, however, the association is potentially confounded by three darbepoetin trials without anticancer therapy (Charu 2007; Gordon 2008; Smith 2008). The observed effect was larger in unblinded trials (MD 3.76, 955 CI 2.60 to 4.92) compared with double‐blind trials (MD 1.33, 95% CI 0.56 to 2.10, P = 0.0006). Significant differences were also observed for age group (adults versus >70% non‐elderly adults, P = 0.008). The backward selection resulted in a model containing the covariate type of ESA formulation, see Table 5. The MD for a trial in patients receiving short‐lasting ESAs is = Intercept + short lasting ESA = 1.09 + 2.20 = 3.29. However, in these analyses single studies were compared to groups of studies and thus results are not readily interpretable.

4. FACT‐F 13: results of meta‐regression analysis.

Variable	mean difference	standard error	P value
Intercept	1.09	0.6158	0.0779
Short acting ESA	2.20	0.8346	0.0083

ESA: erythropoiesis stimulating agent

One trial (Tsuboi 2009) reported two different sets of scores for the control group: in one set they substituted the missing data of two patients by the maximum decrease in score for all patients. The other set of results did not include data for these two patients. We used the scores without the substitution for the main analysis and we conducted a sensitivity analysis using the scores with the substituted data. The results did not change much, with the MD being 2.10 (95% CI 1.46 to 2.75, 18 studies, N = 4,967) and remained statistically significant (P < 0.00001).

Overall, it appears that there is an effect on fatigue‐related symptoms for patients treated with erythropoietin or darbepoetin compared with controls; this effect, however, did not reach the threshold for a clinically important difference defined as 3.0 (Cella 2002).

b) FACT‐An 20

Six studies were included for this outcome (Chang 2005; Christodoulou 2009; Littlewood 2001; O'Shaughnessy 2005; Savonije 2005; Wright 2007). The estimated MD is 6.14 (95% CI 4.55 to 7.73, N = 1,085). There was no evidence for statistical heterogeneity between the studies (I² = 0%). A funnel plot analysis was not done, because less than 10 studies were included in this analysis.

Subgroup analyses did not identify any significant differences in the magnitude of effect between the different subgroups. Of note, FACT‐An 20 improvements were seen across all Hb baseline subgroups (Hb < 10, 10 to 12 and > 12 g/dL).

Overall, the effects of ESAs on fatigue‐ and anaemia‐related symptoms appear to be beneficial and the difference between groups reaches both statistical and clinical significance (clinically important difference defined as four to five), however only six studies were included in this analysis.

c) FACT‐An Total 47

Nine studies, including 1,815 participants, reported data for this outcome. The estimated MD is 6.92 (95% CI 4.59 to 9.25). As one multi‐arm study (Krzakowski 2008) reported results separately for each arm, the number of studies appearing in Metaview is 10.Heterogeneity between the studies was quite high (I² = 85%). A funnel plot analysis was not done, because less than 10 trials were included in this analysis. Merging the two arms of the multi‐arm study mentioned above in one data set did not change the results (MD 6.92; 95% CI 4.59 to 9.25, nine studies, N = 1,815) or heterogeneity markedly (I² = 87%).

One study (Mystakidou 2005) reported an unusually high change for the treatment group (a mean change of 43.3 when the average mean change from all the other studies is 5.14). We, therefore, conducted a sensitivity analysis in which we excluded this study. The results are indeed different than the original analysis with a MD of 3.46 (95% CI 0.96 to 5.96, eight studies, N = 1,715). The statistical heterogeneity between the studies almost disappeared (I² = 0%). The results, however, remained statistically significant (P = 0.007).

We conducted sub‐group analysis that revealed the following subgroups as having significantly different magnitudes of ESA effect between each other: imputed versus non‐imputed data (P = 0.005), baseline Hb level (P < 0.0001), type of anti‐cancer therapy (P < 0.0001), duration of ESA treatment (P < 0.0001) and ITT analysis (P = 0.01). The backward selection resulted in a model containing the covariate type of anticancer therapy, see Additional Table 6. However, differences can also be explained by the study conducted by Mystakidou 2005 (see paragraph above), when removing this study no significant differences between subgroups were evident (data not shown).

5. FACT‐An 47: results of meta‐regression analysis.

Variable	mean difference	Standard error	P value
Intercept	6.10	2.3783	0.0103
Chemotherapy, < 70% of participants receiving platinum‐based chemotherapy	‐5.82	3.1929	0.0683
Chemotherapy, > 70% of participants receiving platinum‐based chemotherapy	0.81	3.7926	0.8303
No anticancer therapy	23.80	4.0571	<0.0001
Radiotherapy	‐4.20	3.9147	0.2836

Overall, there is a statistically significant difference between patients treated with ESAs and controls when combining QoL parameters and fatigue‐ and anaemia‐related symptoms, which is however, most likely not clinically important.

(9) Adverse events

(9.1) Thromboembolic events

Data from thromboembolic complications were available from a total of 57 trials, including 15,498 participants. As three multi‐arm trials (Osterborg 1996; Ten Bokkel 1998; Thatcher 1999) were split into subsets and four studies reported zero events (Cascinu 1994; Gupta 2009; P‐174 J&J 2004; Thatcher 1999a), the number of studies displayed is 60. Thirty‐five studies including 6,769 participants were included in the 2006 update of the Cochrane review (Bohlius 2006a), results for 23 studies were newly identified (Aapro 2008; Blohmer 2011; Charu 2007 Amgen 2007; Debus 2006 J&J 2007; Engert 2009; Fujisaka 2011; Gordon 2008; Gupta 2009; Hedenus 2003 FDA 2007; Hernandez 2009; Milroy 2011; Moebus 2007 J&J 2007; Overgaard 2009; Pirker 2008; Pronzato 2010; Ray‐Coquard 2009; Smith 2008; Strauss 2008; Tjulandin 2011; Tsuboi 2009; Untch 2011_1; Wilkinson 2006; Winquist 2009). We removed one study from the analysis because a substantial number of participants underwent major surgical procedures during ESA treatment (Vadhan‐Raj 2004).

The overall risk ratio to suffer thromboembolic complications was increased by 52% for patients receiving ESAs (RR 1.52; 95% CI 1.34 to 1.74). There was no significant statistical heterogeneity between the trials (I² = 0%). A funnel plot analysis revealed a significant asymmetry (P = 0.02137), suggesting that harmful events (in this case thrombotic event) have been over reported. Merging the multi‐arm trials into one data set, did not change the results (RR 1.53; 95% CI 1.34 to 1.74). Compared with the results from the previous Cochrane review (RR 1.67; 95% CI 1.35 to 2.06, 35 trials, N = 6,769, Bohlius 2006a), results are similar. Subgroup analyses for predefined variables did not show robust evidence for statistically significant differences in magnitude or direction of the ESA effect between any of the subgroups tested (e.g. baseline Hb level, type of malignancy, duration of treatment, type of anti‐cancer therapy, age, iron supplementation, type of publication, epoetin versus darbepoetin, type of data, concealment of allocation and masking).

We calculated numbers needed to harm (NNTH) for several hypothetical baseline risks. In a population with an underlying risk of 2% the NNTH would be 96 (95% CI 68 to 147), thus one thromboembolic complication would occur for about every 96 patients treated. In a population with an underlying risk of 5% the NNTH would be 38 (95% CI 27 to 59), thus for every 38 patients treated with ESAs one additional thromboembolic complication might happen. In a population with a hypothetical baseline risk of 10% the NNTH would be 19 (95% CI 14 to 29), thus for every 19 patients treated one additional thromboembolic complication may happen.

In conclusion, the data available for the present analysis confirm and strengthen conclusions from the prior versions of this review that treatment with ESAs increases the risk of thrombosis or related complications.

(9.2) Hypertension

Hypertension data were available from a total of 31 trials including 7,228 participants. Of these trials, 16 including 2,263 randomised participants were included in the updated Cochrane review of 2006 (Bohlius 2006a), for 15 studies Fujisaka 2011; Gordon 2008; Hernandez 2009; Hoskin 2009; Krzakowski 2008; Milroy 2011; Osterborg 2002; Pirker 2008; Razzouk 2006; Savonije 2005; Smith 2008; Tjulandin 2010; Tjulandin 2011; Tsuboi 2009; Wilkinson 2006 data were newly identified. As six of the trials were split into subsets (Krzakowski 2008; Kunikane 2001; Osterborg 1996; Ten Bokkel 1998; Thatcher 1999; Tjulandin 2010) and two trials reported zero events (Cascinu 1994; Iconomou 2003) the number of trials displayed in MetaView is 37.

The risk ratio to develop hypertension for erythropoietin‐treated participants was increased by 30% (fixed‐effect model RR 1.30; 95% CI 1.08 to 1.56), reaching statistical significance (P = 0.006). Using the random‐effects model, however, the result was not statistically significant (RR 1.12, 95% CI 0.94 to 1.33). There was no statistical heterogeneity between the trials (I² = 0%). A funnel plot analysis revealed significant asymmetry (P value < 0.001), suggesting that harmful events (in this case hypertension) have been over reported. Merging the different arms of the multi‐arm trials did not change the overall result (RR 1.31, 95% CI 1.09 to 1.58). The updated results are similar to the previous Cochrane review (RR 1.24; 95% CI 1.00 to 1.54, 16 trials, N = 2,263 Bohlius 2006a).

One single study (Rose 1994) contributed 40% weight to this analysis. Excluding this study, the effect of ESAs on hypertension increased: fixed‐effect: RR 1.52; 95% CI 1.18 to 1.97, random‐effects: RR 1.40; 95% CI 1.07 to 1.83. For another trial, both published and unpublished data were available (Dammacco 2001). In the published sources 3/69 participants in the erythropoietin group and 1/76 participants in the control group were reported to suffer from hypertension. In the unpublished study report 43/69 participants in the erythropoietin group and 36/76 in the control group had hypertension. Including these data instead of the published numbers did not change the overall results significantly, (fixed‐effect model: RR 1.29; 95% CI 1.10 to 1.52), random‐effects model: RR 1.16; 95% CI 1.00 to 1.35).

We calculated numbers needed to harm for several hypothetical baseline risks. In a population with an underlying risk of 2% the NNTH would be 167 (95% CI 89 to 625), thus one patient would be affected with hypertension for about every 167 patients treated. In a population with an underlying risk of 5% the NNTH would be 67 (95% CI 36 to 250), thus for every about 67 patients treated with ESAs one additional episode of hypertension might happen. In a population with a hypothetical baseline risk of 10% the NNTH would be 33 (95% CI 18 to 125), thus for every about 33 patients treated, one additional hypertension episode may happen.

Overall, the data available in this analysis provide suggestive but not robust evidence, that ESAs in the treatment of cancer patients may increase the risk ratio to suffer from hypertension.

(9.3) Haemorrhage/Thrombocytopenia

Haemorrhage/thrombocytopenia data were available from a total of 21 trials including 4,507 participants. Of these trials, 10 including 1,488 randomised participants were included in the update of the Cochrane review in 2006 (Bohlius 2006a). Eleven additional trials Fujisaka 2011; Gebbia 2003; Goss 2005; Gupta 2009; Milroy 2011; Pirker 2008; Savonije 2005; Strauss 2008; Tsuboi 2009; Untch 2011_1; Witzig 2005) including 3,019 participants were added for this update. As three of the trials were split into subsets (Kunikane 2001; Osterborg 1996; Thatcher 1999) and three trials report zero events (Cascinu 1994; Gupta 2009; Osterborg 1996b), the number of trials displayed in MetaView is 24.

The risk ratio to develop thrombocytopenia was increased for erythropoietin‐treated participants (RR 1.21; 95% CI 1.04 to 1.42), reaching statistical significance (P = 0.01). When using the random‐effects model, the results were also statistically significant (RR 1.18, 95% CI 1.02 to 1.36). There was no statistical heterogeneity between the trials (I² = 0%). The funnel plot analysis did not show a significant asymmetry (P = 0.698). Results are similar to the previous review (RR = 1.13, 95 % CI 0.08 to 1.60, 10 trials, N = 1,488 Bohlius 2006a), however, the larger number of included studies and patients increased statistical power and the updated result reached statistical significance. Merging the arms of the multi‐arm trials mentioned above in one data set did not substantially change the results (RR 1.21, 95% CI: 1.04 to 1.41). Subgroup analyses were not performed.

We calculated numbers needed to harm for several hypothetical baseline risks. In a population with an underlying risk of 2% the NNTH would be 238 (95% CI 122 to 1,250), thus one patient would develop thrombocytopenia or haemorrhage for about every 238 patients treated. In a population with an underlying risk of 5% the NNTH would be 95 (95% CI 49 to 500), thus for about every 95 patients treated with ESAs one additional case of thrombocytopenia or haemorrhage might happen. In a population with a hypothetical baseline risk of 10% the NNTH would be 48 (95% CI 24 to 250), thus for every 48 patients treated one additional case of thrombocytopenia or haemorrhage might occur.

Overall, there is some evidence, that ESAs may increase the risk for thrombocytopenia/haemorrhage.

(9.4) Rash, Irritation, Pruritus

Data were available from a total of 16 trials including 2,485 participants. Of those, eight trials with 675 participants were included in the 2006 update (Bohlius 2006a). Data from eight more trials including 1,810 participants were added for this update. As two of the trials were split into subsets (Osterborg 1996; Thatcher 1999) and two trials reported zero events (Gupta 2009; Kurz 1997) the number of trials in MetaView is 18.

Overall, 53 events of skin rash, irritation or pruritus were reported in the erythropoietin group (N = 1,359) and 27 cases in the control group (N = 1,126) , resulting in a risk ratio of 1.49 (95% CI 0.99 to 2.24). There was no significant heterogeneity between the studies (I² = 0%). Merging the different subsets of the multi‐arm trials into one did not change the results (RR1.50, 95% CI 1.00 to 2.27). The funnel plot analysis did not show a significant asymmetry (P = 0.745). Further sensitivity analyses were not done. Compared with the previous Cochrane review (Bohlius 2006a), results are similar (RR 1.17; 95% CI 0.63 to 2.18, eight trials, N = 675 Bohlius 2006a). Based on the data available there is insufficient evidence to conclude that erythropoietin increases the risk of skin reactions.

(9.5) Seizures

Data on seizures were available from eight trials including 2,890 participants (Cascinu 1994; Case 1993; Gordon 2008; Henry 1995; Hernandez 2009; Pirker 2008; Savonije 2005; Smith 2008). Three of those trials (Cascinu 1994; Case 1993; Henry 1995) including 389 participants were included in the 2006 version of the Cochrane review (Bohlius 2006a), data from five further trials including 2,501 participants have been added to this update (Gordon 2008; Hernandez 2009; Pirker 2008; Savonije 2005; Smith 2008. Overall, 19 events of seizure were reported in the erythropoietin group (N=1,583) and 21 events in the control group (N = 1,307), resulting in a risk ratio of 0.77 (95% CI 0.42 to 1.41). There was no significant statistical heterogeneity between the trials (I² = 3%). Overall, there was no evidence for significant differences between the treatment groups compared.

Discussion

Summary of main results

This systematic review analysed the effectiveness and safety of erythropoiesis‐stimulating agents (ESAs) for managing anaemia in cancer patients. The primary findings of this updated review are as follows: ESAs significantly reduce the need for red blood cell transfusions and increase haematological response in cancer patients. However, there is also strong evidence that ESAs increase mortality during study period and some evidence that ESAs reduce overall survival. In addition, there is strong evidence that ESAs increase the risk for thromboembolic complications and some evidence that ESAs increase the risk of hypertension and thrombocytopenia/haemorrhage. The available data are insufficient to evaluate the effect of ESAs on tumour response. Their effect on patients' quality of life (QoL) and specific anaemia‐ and fatigue‐related symptoms (FACT‐An) reaches statistical and clinical significance; however, while it reaches statistical significance, it fails to reach clinical significance for fatigue‐related symptoms only (FACT‐F).

Overall completeness and applicability of evidence

The main strength of the newly updated review is the large number of studies and patients included. All studies were systematically evaluated and outcomes from previous versions of this review updated.

Quality of the evidence

The main weakness of this review are reporting and publication biases. For the outcomes mortality during study period and overall survival this was less of a problem, since we used results that were generated in an individual patient data meta‐analysis (Bohlius 2009a; Bohlius 2009b), which had included the majority of studies in the field. However, besides survival and mortality, no other outcomes were assessed in that review and thus for the remaining outcomes we had to rely on the published evidence. For several of these outcomes only a few studies could be evaluated. For example, hypertension was reported in only 31 of 91 studies, thrombocytopenia in 21 and thromboembolic complications in 57 of 91 studies, haemoglobin (Hb) response in 31 and the number of red blood cell units transfused in 20 of 91 studies. Similarly, numerical QoL data for FACT‐An or FACT‐F were only reported in 23 out of 91 trials. For some but not all of these outcomes, the suspected publication bias was further supported by asymmetric funnel plots. The numbers indicated here underline the lack of complete outcome reporting for a major part of studies and outcomes. In the light of these apparent reporting biases, we estimate that both the beneficial (number of patients transfused, QoL (FACT‐F), Hb response, change in Hb) as well as the harmful effects of ESAs other than mortality (thromboembolic events, hypertension) have been overestimated in our present analysis.

Agreements and disagreements with other studies or reviews

This and previous analyses provide consistent evidence that ESAs reduce the risk for RBC transfusions by approximately 30% to 40%. The point estimated generated in the current update is in line with previous (Bohlius 2006a) and other systematic reviews and meta‐analyses (Tonelli 2009). fOur analysis also provides consistent evidence that ESAs reduce the average number of RBC units transfused. Participants on ESAs received on average one unit of RBCs less (mean difference (MD) ‐0.98, 95% confidence interval (CI) ‐1.17 to ‐0.78) compared with controls which is in line with a previous, independent meta‐analysis, reporting a weighted mean difference (WMD) of −0.80 units (95% CI −0.99 to −0.61) (Tonelli 2009 HTA).

Apart from improving physiologic parameters such as Hb and haematocrit, erythropoietin was also hypothesised to improve QoL and to alleviate fatigue. Improved QoL after ESA treatment was previously reported in community based, single‐arm studies (Demetri 1998; Gabrilove 2001; Glaspy 1997). However, these data were inconclusive because they lacked controls. Later, several randomised controlled studies were conducted to investigate this question. QoL trials require specific methodological standards, such as the use of validated instruments, double blinding, a prospective plan to minimise missing data, investigating the pattern of missing data, and addressing missing data in the analysis (Aaronson 1991; Brandberg 2000). Our updated analysis does not provide evidence for a clinically important improvement of fatigue in patients receiving ESAs compared to controls. This result is more conservative compared to previous meta‐analyses on the same topic (Minton 2010; Tonelli 2009). While previous analyses identified differences for FACT‐F just at (MD 3.00; 95% CI 1.36‐4.64; 10 RCTs, N = 3,169) (Tonelli 2009) or above (MD 3.72; 95% CI 2.38‐5.06; 12 RCTs, N = 2,671) (Minton 2010, personal communication) the threshold for a clinically important difference, our analysis showed a MD of 2.08 (95% CI 1.43‐2.72, 18 RCTs, N = 4,965) which is below the estimated threshold of 3.0 (Cella 2002) . For FACT‐An 20 there seems to be a clinically important improvement in patients receiving ESAs with a MD of 6.14 (95% CI 4.55 to 7.73, six studies, N = 1,085), which is above the CID for FACT‐An, defined as four to five (Cella, personal communication March 2010). However, only a small number of published RCTs on ESAs reported QoL outcomes and therefore more evidence is needed for definitive conclusions. For FACT‐An 47 there is no established CID; however, we expect it to be at least above five, as this is the threshold for FACT‐An 20, which includes less than half the questions of FACT‐An 47. In our analysis there was no robust evidence for a clinically important improvement of QoL measured with FACT‐An 47.

Besides the beneficial effects of ESAs indicated above, our review also identified harmful effects, including an increased risk for thromboembolic events, hypertension, thrombocytopenia, death during study period and potentially decreased overall survival. The increased risk for thromboembolic events in patients receiving ESAs has been observed and reported by single RCTs (Aapro 2008; Goss 2005 J&J 2004; Pirker 2008) and previous meta‐analyses on the same topic (Aapro 2008;Aapro 2009b; Bennett 2008; Bohlius 2006a; Glaspy 2010; Ludwig 2009; Ross 2006; Seidenfeld 2006; Tonelli 2009). The effect estimates generated are comparable across the different meta‐analyses reported. The impact of baseline Haemoglobin levels has not been clarified to date. While one individual patient data meta‐analysis restricted to studies using Darbepoetin for patients receiving chemotherapy suggested an increased risk for thromboembolic events with higher baseline haemoglobin levels (Ludwig 2009), this was not confirmed in a second IPD analysis restricted to RCTs comparing epoetin beta verus control (Aapro 2009b). In this analysis, the risk for thromboembolic events decreased with increasing baseline haemoglobin levels (Aapro 2009b). Besides, there are reports that ESAs are inherently thrombogenic irrespective of baseline or current Hb levels (Barbera 2010; Fuste 2002; Malyszko 1995; Stasko 2002; Stohlawetz 2000). Strategies to reduce the risk for thromboembolic events by using e.g. anti‐coagulating drugs have not yet been evaluated in randomised controlled trials (Aapro 2009a).

Patients receiving ESAs may also have an increased risk for hypertension. In patients with chronic renal failure, hypertension is a common adverse effect of ESAs (Palmer 2010). This increase in blood pressure can be partly explained by the elevated blood viscosity and the loss of hypoxia‐induced vasodilatation in association with the increased Hb level (Cirillo 1993). The present analysis shows some evidence that ESAs also increase the risk for hypertension in cancer patients by approximately 30%. However, the result was statistically significant only in the fixed‐effect model and not in the random‐effects model. In addition, only 31 of 91 studies reported for this outcome. A funnel plot analysis revealed a significant asymmetry (P value < 0.001), suggesting that negative results (in this case no hypertension) have been underreported. Thus, the effect of ESAs on hypertension might be overestimated in the present analysis. A previous meta‐analysis in cancer patients has also identified an increased risk (RR 1.41, 95% CI 0.94 to 2.12, 17 studies, 3,792 patients) which failed to reach conventional level of statistical significance (Tonelli 2009). We also found some evidence that ESAs increase the risk for thrombocytopenia and haemorrhage (RR 1.21, 95% CI, 1.04 to 1.42, 21 trials, 4,507 patients, no evidence for publication bias), which has not been reported before. While the increased risk for thromboembolic events has been established, hypertension and thrombocytopenia/haemorrhage require closer monitoring in future studies.

In the current update of the systematic review, we found no evidence for a beneficial effect of ESAs on tumour control. At the same time, uncertainties remain as important information on tumour grade, tumour stage, intensity of anti‐neoplastic treatment received, timing and method of tumour response assessment were not reported. Relevant studies such as the Henke 2003 or the Antonadou 2001 study could not be included in the meta‐analysis, as they did not report data for complete tumour response. Overall, the data were not sufficient either to exclude or to prove a tumour promoting effect of erythropoiesis‐stimulating factors.

Single randomised controlled trials and previous meta‐analyses based on individual patient data (Bohlius 2009a; Bohlius 2009b) have reported an increased risk for death during the active study period. In the current review we have differentiated the active study period, defined as mortality during ESA treatment plus a short follow‐up time of 30 days, versus overall survival defined as the longest follow‐up available. We integrated the data that were generated previously for the individual patient data meta‐analysis (Bohlius 2009a; Bohlius 2009b) as well as studies which were not included in the IPD review. The updated review confirms the previous IPD review (Bohlius 2009a; Bohlius 2009b) showing an increased risk for on‐study mortality including all cancer patients (HR present meta‐analysis: 1.17, 95% CI 1.06 to 1.29, 70 studies, 15,935 patients; HR IPD review: 1.17, 95% CI 1.06‐1.30, 53 studies, N = 13,933). These results are in line with other meta‐analyses of on‐study mortality regardless of underlying cancer therapy. Aapro 2009b reported an HR of 1.13 (95% CI 0.87 to 1.46, 12 studies, N = 2,297) in studies on epoetin beta and Ross 2006 reported an HR of 1.14 (95% CI 0.90 to 1.45, 17 studies). Both meta‐analyses failed to reach conventional levels of statistical significance which may be explained by the fact that fewer studies and patients were included in each of these analyses. There was some evidence that ESAs decreased overall survival defined as longest follow‐up available (HR 1.05, 95% CI 1.00 to 1.11). As in the previous review (Bohlius 2009a; Bohlius 2009b), the effect was small and of borderline statistical significance. It remains uncertain whether or not the risk is also increased in the subset of patients receiving chemotherapy while also receiving an ESA. In the current updated review the HR for on ‐study mortality in patients receiving chemotherapy is 1.10 (95% CI 0.98 to 1.24, 50 studies, 12,058 patients) which is identical to the HR reported in the previous IPD review (Bohlius 2009a; Bohlius 2009b). However, there were no statistically significant differences between this subgroup and the total patient sample in either the prior IPD meta‐analysis or the current update and an increased risk in cancer patients undergoing chemotherapy and receiving ESAs cannot be excluded. FDA (FDA 2010) has recommended restricting the use of ESAs to cancer patients receiving chemotherapy with palliative intent. ESAs should not be used in patients receiving radiotherapy or no antic‐cancer therapy. However, an increased risk for death in patients receiving chemotherapy cannot be excluded.

Several hypotheses have been proposed to explain the increased risk for death in patients receiving ESAs. One is that erythropoietin might directly influence tumour cell growth. In vitro studies have reported high levels of erythropoietin receptors in breast cancer cells and other malignancies (Acs 2001; Arcasoy 2002; Bennett 2010; Henke 2006; Jelkmann 2004; Jelkmann 2008; McKinney 2011;Yasuda 2003). Either endogenously produced or exogenously administered, erythropoietin may promote the proliferation and survival of cancer cells expressing erythropoietin receptor (Acs 2001; Acs 2002; Arcasoy 2002; Bennett 2010; Henke 2006; Jelkmann 2004; Jelkmann 2008; McKinney 2011; Yasuda 2003). However, conflicting evidence has been reported from pre‐clinical studies on the effects of rHu erythropoietin on cultured cell lines (Bennett 2010). Tumour stimulation through erythropoietin has been suggested in some studies, whereas other studies have not observed a relationship between rHuEPO and tumour cell growth (Bennett 2010; McKinney 2011). An alternative hypothesis suggests that the efficacy of malignancy treatments, both radiotherapy and oxygen‐dependent chemotherapy, can be enhanced by decreasing tumour hypoxia. Tumour tissue is often hypoxic and this hypoxia may increase if the patient is anaemic (Becker 2000; Henke 2000; Tatum 2006; Vaupel 2001; Vaupel 2008). This hypothesis suggests that tumour hypoxia diminishes the effectiveness of radiotherapy and oxygen‐dependent chemotherapy (Tatum 2006; Vaupel 2001; Vaupel 2008). Evidence for this hypothesis comes from reports that tumour control and overall survival are better in solid tumour patients with better tumour oxygenation (Hockel 1993; Knocke 1999). In addition, some authors have reported that the effectiveness of oxygen‐dependent radiotherapy is impaired in anaemic patients (Frommhold 1998; Grau 2000). Enhanced cytocidal efficacy in adequately oxygenated cells has been documented for a number of cytotoxic drugs, such as cyclophosphamide, carboplatin and doxorubicin (Teicher 1981; Teicher 1994). In animal models, cyclophosphamide (Thews 2001) and cisplatin (Silver 1999) have yielded better tumour control with improved tissue oxygenation. Given these observations, it seems plausible that increasing the Hb level with erythropoietin may improve tumour oxygenation and thus tumour control and eventually overall survival. This has been partly demonstrated in animal models (Kelleher 1998). However, other pre‐clinical studies have demonstrated that experimental tumour cells acclimatize rapidly to acute anaemia and return to normal radio sensitivity despite continuing anaemia (Hirst 1984). Chronic anaemia does not necessarily produce radio‐resistance of experimental tumours (Koong 1991) and the correction of anaemia by erythropoietin does not necessarily increase radio‐sensitivity (Joiner 1993).

Authors' conclusions

Implications for practice.

ESAs reduce the need for red blood cell transfusions but increase the risk for thromboembolic events and deaths. There is suggestive evidence that ESAs may improve QoL. Whether and how ESAs affects tumour control remains uncertain. The increased risk of death and thromboembolic events should be balanced against the potential benefits of ESA treatment taking into account each patient’s clinical circumstances and preferences.

Implications for research.

More data are needed for the effect of these drugs on quality of life, tumour progression and other adverse effects. Further research is needed to clarify cellular and molecular mechanisms and pathways of the effects of ESAs on thrombogenesis and their potential effects on tumour growth.

What's new

Date	Event	Description
11 May 2012	New citation required and conclusions have changed	Authors changed (new authors: Tonia T, Mettler A, Robert N) Substantive update, in the previous review the outcome haematological response was restricted to studies with baseline Hb levels < 12 g/dL; in the current update this restriction was removed. On‐study mortality was added as a new outcome to the current update of the review. Studies using iron supplements in one study arm only were included, in the previous version these studies were excluded. Studies using any dose of ESAs were included, in the previous review studies using very low dosages had been excluded. Any type of ESA was included in the review.
30 November 2011	New search has been performed	New search

History

Protocol first published: Issue 4, 2002
 Review first published: Issue 3, 2004

Date	Event	Description
15 August 2008	Amended	Converted to new review format.
24 May 2006	New citation required and conclusions have changed	Substantive amendment

Appendices

Appendix 1. EMBASE search strategy

Please note that the same search strategy was used for the other searches conducted on the dates mentioned in the main text.

No.	Query
#70	#68 AND [1‐8‐2009]/sd NOT [11‐1‐2011]/sd
#68	#38 AND #67
#67	#63 NOT #66
#66	#64 NOT #65
#65	'human'/exp OR human
#64	'animal'/exp OR animal
#63	#58 NOT #62
#62	#59 OR #60 OR #61
#61	'abstract report' OR letter
#60	'case report'
#59	'case study'
#58	#39 OR #40 OR #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 OR #52 OR #53 OR #54 OR #55 OR #56 OR #57
#57	'prospective study'
#56	placebo*
#55	'triple' NEAR/5 'blind'
#54	'treble' NEAR/5 'blind'
#53	double AND blind*
#52	single AND blind*
#51	'allocated' NEAR/2 'random'
#50	'allocated randomly'
#49	'randomly allocated'
#48	'random allocation'
#47	rct
#46	randomi?ed AND controlled AND trial?
#45	'placebo'/exp OR placebo
#44	'crossover procedure'/exp OR 'crossover procedure'
#43	'double blind procedure'/exp OR 'double blind procedure'
#42	'single blind procedure'/exp OR 'single blind procedure'
#41	'randomization'/exp OR 'randomization'
#40	'randomized controlled trial'/exp OR 'randomized controlled trial'
#39	'clinical trial'/exp OR 'clinical trial'
#38	#36 AND #37
#37	#21 AND #27
#36	#28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35
#35	carcinoma*
#34	tumo?r*
#33	chemotherapy
#32	myelodysplas*
#31	oncolog*
#30	cancer*
#29	malignan*
#28	'neoplasm'/exp OR neoplasm
#27	#22 OR #23 OR #24 OR #25 OR #26
#26	anemi*
#25	anaemi*
#24	anemia
#23	anaemia
#22	'anemia'/exp OR anemia
#21	#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20
#20	'erythropoietin receptor'/exp OR 'erythropoietin receptor'
#19	micer*
#18	mircer*
#17	methoxy AND polyethylene AND 'glycol epoetin' AND beta
#16	continuous AND erythropo?es* AND ('receptor'/exp OR receptor) AND activator
#15	cera*
#14	darbepo?eti*
#13	cepo*
#12	procit*
#11	aranesp*
#10	neorecormon*
#9	eprex*
#8	haemopo*etin*
#7	haematopo*etin*
#6	hemopo*etin*
#5	hematopo*etin*
#4	epo*
#3	eritropo*
#2	eryt*ropo*
#1	'erythropoietin' OR 'recombinant erythropoietin'/exp

Appendix 2. MEDLINE search strategy

Please note that the same search strategy was used for the other searches conducted on the dates mentioned in the main text.

MEDLINE /Ovid (February 2011 to November 2011)

#	Searches
1	exp ERYTHROPOIETIN/
2	exp ERYTHROPOIETIN, RECOMBINANT/
3	erythropoietin.mp.
4	erythropoiesis.mp.
5	exp EPOETIN ALFA/
6	epoetin.mp.
7	epo.mp.
8	epoetin alfa.mp.
9	epoetin beta.mp.
10	eprex.mp.
11	neorecormon.mp.
12	aranesp.mp.
13	procrit.mp.
14	recombinant erythropoietin.mp.
15	darbepoetin alfa.mp.
16	darbepoetin.mp.
17	RECEPTORS, ERYTHROPOIETIN/
18	CERA.mp.
19	or/1‐18
20	exp ANEMIA/dt, th [Drug Therapy, Therapy]
21	anaemia.mp.
22	anemia.mp.
23	(anemi$ adj3 cancer).mp.
24	(anaemi$ adj3 cancer).mp.
25	or/20‐24
26	exp Neoplasms/
27	malignan$.mp.
28	cancer$.mp.
29	oncolog$.tw.
30	myelodysplas$.tw.
31	chemotherapy.mp.
32	tumo?r$.mp.
33	carcinom$.mp.
34	or/26‐33
35	19 and 25
36	34 and 35
37	randomized controlled trial.pt.
38	controlled clinical trial.pt.
39	randomized.ab.
40	placebo.ab.
41	drug therapy.fs.
42	randomly.ab.
43	trial.ab.
44	groups.ab.
45	or/37‐44
46	humans.sh.
47	45 and 46
48	36 and 47
49	limit 48 to ed=20080101‐20081231
50	randomized controlled trial.pt.
51	controlled clinical trial.pt.
52	randomized controlled trials as topic/
53	random allocation/
54	double blind method/
55	single blind method/
56	or/50‐55
57	(ANIMALS not HUMANS).sh.
58	56 not 57
59	clinical trial.pt.
60	exp clinical trial as topic/
61	(clin$ adj25 trial$).ti,ab.
62	((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab.
63	placebos/
64	placebo$.ti,ab.
65	random$.ti,ab.
66	research design/
67	or/59‐66
68	67 not 57
69	68 not 58
70	comparative study/
71	exp evaluation studies/
72	follow up studies/
73	prospective studies/
74	(control$ or prospectiv$ or volunteer$).ti,ab.
75	or/70‐74
76	75 not 57
77	76 not (58 or 69)
78	58 or 69 or 77
79	36 and 78
80	48 or 79
81	limit 80 to ed=20090801‐20110201
82	limit 80 to ed=20110201‐20111201

Appendix 3. CENTRAL search strategy

Please note that the same search strategy was used for the other searches conducted on the dates mentioned in the main text.

Cochrane Central Register of Controlled Trials (Cochrane Library 2011, Issue 3)

ID	Search
#1	(erythropoietin)
#2	MeSH descriptor Erythropoietin explode all trees
#3	epoetin
#4	epo
#5	(epoetin next alfa)
#6	(epoetin next beta)
#7	(darbepoetin next alfa)
#8	eprex
#9	neorecormon
#10	aranesp
#11	procrit
#12	(recombinant near erythropoietin)
#13	"continuous erythropoietin receptor activation"
#14	"continuous erythropoietin receptor activator"
#15	CERA
#16	C.E.R.A.
#17	erythropoiesis
#18	darbepoetin
#19	(#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18)
#20	anemia
#21	anaemia
#22	MeSH descriptor Anemia explode all trees
#23	(anemi* near cancer)
#24	(anaemi* near cancer)
#25	(#20 OR #21 OR #22 OR #23 OR #24)
#26	(#19 AND #25)
#27	(#26), from 2007 to 2008
#28	(#26), from 2008 to 2009
#29	(#26), from 2009
#30	(#28 OR #29)
#31	(#26), from 2009 to 2011
#32	(#26), in 2011

Data and analyses

Comparison 1. Haematologic response.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Haematological response ‐ overall	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
2 Haematologic response ‐ baseline Hb	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
2.1 Hb <= 10 g/dL	36	4137	Risk Ratio (M‐H, Fixed, 95% CI)	3.02 [2.72, 3.35]
2.2 Hb 10 to 12 g/dL	8	1775	Risk Ratio (M‐H, Fixed, 95% CI)	4.60 [3.79, 5.58]
2.3 Hb > 12 g/dL	1	380	Risk Ratio (M‐H, Fixed, 95% CI)	7.48 [3.00, 18.62]
2.4 Hb category unclear	1	121	Risk Ratio (M‐H, Fixed, 95% CI)	2.04 [1.20, 3.46]
3 Haematologic response ‐ different malignancies	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
3.1 solid tumours	18	3089	Risk Ratio (M‐H, Fixed, 95% CI)	3.56 [3.12, 4.07]
3.2 haematological malignancies	15	1623	Risk Ratio (M‐H, Fixed, 95% CI)	3.12 [2.64, 3.69]
3.3 MDS	2	151	Risk Ratio (M‐H, Fixed, 95% CI)	4.27 [0.86, 21.19]
3.4 mixed	12	1550	Risk Ratio (M‐H, Fixed, 95% CI)	3.38 [2.80, 4.08]
3.5 not reported	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Haematological response‐ age	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
4.1 children	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	1.62 [1.20, 2.18]
4.2 adults	45	6191	Risk Ratio (M‐H, Fixed, 95% CI)	3.55 [3.23, 3.90]
5 Haematological response‐ age differentiated	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
5.1 only children <18 years	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	1.62 [1.20, 2.18]
5.2 adults ≥18 years	43	5623	Risk Ratio (M‐H, Fixed, 95% CI)	3.34 [3.03, 3.68]
5.3 >70% non‐elderly 18‐65 years	2	568	Risk Ratio (M‐H, Fixed, 95% CI)	8.26 [5.22, 13.06]
5.4 only non‐elderly adults	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.5 >70% elderly >65 years	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.6 only elderly adults	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Haematologic response ‐ different therapies	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
6.1 chemotherapy	38	5562	Risk Ratio (M‐H, Fixed, 95% CI)	3.32 [3.02, 3.64]
6.2 radiotherapy/radiochemotherapy	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.3 no therapy	7	630	Risk Ratio (M‐H, Fixed, 95% CI)	5.50 [3.25, 9.31]
6.4 other	1	221	Risk Ratio (M‐H, Fixed, 95% CI)	2.87 [1.69, 4.85]
7 Haematologic response ‐ different therapies differentiated	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
7.1 chemotherapy, >70% with platinum	7	1301	Risk Ratio (M‐H, Fixed, 95% CI)	3.12 [2.51, 3.87]
7.2 chemotherapy, <70% platinum containing	3	659	Risk Ratio (M‐H, Fixed, 95% CI)	2.71 [2.16, 3.40]
7.3 chemotherapy without platinum (all patients)	17	2614	Risk Ratio (M‐H, Fixed, 95% CI)	3.73 [3.26, 4.26]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	8	643	Risk Ratio (M‐H, Fixed, 95% CI)	4.05 [2.84, 5.77]
7.5 chemotherapy no details given	3	345	Risk Ratio (M‐H, Fixed, 95% CI)	2.13 [1.59, 2.86]
7.6 radiochemotherapy	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.7 radiotherapy	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.8 no therapy	7	630	Risk Ratio (M‐H, Fixed, 95% CI)	5.50 [3.25, 9.31]
7.9 other	1	221	Risk Ratio (M‐H, Fixed, 95% CI)	2.87 [1.69, 4.85]
8 Haematologic response ‐ epoetin versus darbepoetin	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
8.1 Epoetin	30	5270	Risk Ratio (M‐H, Fixed, 95% CI)	3.27 [2.97, 3.59]
8.2 Darbepoetin	16	1143	Risk Ratio (M‐H, Fixed, 95% CI)	4.22 [3.22, 5.55]
9 Haematologic response ‐ duration of ESA medication	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
9.1 6 to 9 weeks	6	349	Risk Ratio (M‐H, Fixed, 95% CI)	3.74 [1.94, 7.19]
9.2 12 to 16 weeks	34	4574	Risk Ratio (M‐H, Fixed, 95% CI)	3.17 [2.87, 3.51]
9.3 more than 17 weeks	6	1490	Risk Ratio (M‐H, Fixed, 95% CI)	4.41 [3.52, 5.52]
10 Haematologic response ‐ iron supplementation	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
10.1 fixed iron supplementation	2	441	Risk Ratio (M‐H, Fixed, 95% CI)	2.43 [1.92, 3.07]
10.2 iron supplementation as necessary	36	5265	Risk Ratio (M‐H, Fixed, 95% CI)	3.63 [3.27, 4.03]
10.3 no explicit statement on iron supplementation or no iron given	7	399	Risk Ratio (M‐H, Fixed, 95% CI)	4.82 [2.64, 8.81]
10.4 explicitly stated NO IRON	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 iron given differently in both study arms	1	308	Risk Ratio (M‐H, Fixed, 95% CI)	2.22 [1.63, 3.01]
11 Haematologic response ‐ allocation concealment	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
11.1 adequate	30	4721	Risk Ratio (M‐H, Fixed, 95% CI)	3.35 [3.02, 3.72]
11.2 unclear	16	1692	Risk Ratio (M‐H, Fixed, 95% CI)	3.50 [2.92, 4.19]
12 Haematologic response ‐ masking	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
12.1 double‐blind	29	3430	Risk Ratio (M‐H, Fixed, 95% CI)	3.02 [2.70, 3.37]
12.2 unblinded	17	2983	Risk Ratio (M‐H, Fixed, 95% CI)	4.07 [3.49, 4.74]
13 Haematologic response ‐ intention‐to treat	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
13.1 ITT or less than 10% of participants per study arm excluded	41	5657	Risk Ratio (M‐H, Fixed, 95% CI)	3.32 [3.03, 3.65]
13.2 more than 10% of participants per study arm excluded	2	512	Risk Ratio (M‐H, Fixed, 95% CI)	6.85 [3.76, 12.48]
13.3 unclear	3	244	Risk Ratio (M‐H, Fixed, 95% CI)	3.04 [1.92, 4.80]
14 Haematologic response ‐ publication	46	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.39 [3.10, 3.71]
14.1 full text publication	32	5229	Risk Ratio (M‐H, Fixed, 95% CI)	3.44 [3.12, 3.80]
14.2 abstract publication	2	123	Risk Ratio (M‐H, Fixed, 95% CI)	5.19 [2.23, 12.04]
14.3 unpublished data	10	880	Risk Ratio (M‐H, Fixed, 95% CI)	3.44 [2.45, 4.82]
14.4 FDA hearing	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.5 clinical trial result information	2	181	Risk Ratio (M‐H, Fixed, 95% CI)	1.98 [1.36, 2.89]
15 Haematological response ‐ merged experimental arms	31	6413	Risk Ratio (M‐H, Fixed, 95% CI)	3.42 [3.12, 3.74]

1.1. Analysis.

Comparison 1 Haematologic response, Outcome 1 Haematological response ‐ overall.

1.2. Analysis.

Comparison 1 Haematologic response, Outcome 2 Haematologic response ‐ baseline Hb.

1.3. Analysis.

Comparison 1 Haematologic response, Outcome 3 Haematologic response ‐ different malignancies.

1.4. Analysis.

Comparison 1 Haematologic response, Outcome 4 Haematological response‐ age.

1.5. Analysis.

Comparison 1 Haematologic response, Outcome 5 Haematological response‐ age differentiated.

1.6. Analysis.

Comparison 1 Haematologic response, Outcome 6 Haematologic response ‐ different therapies.

1.7. Analysis.

Comparison 1 Haematologic response, Outcome 7 Haematologic response ‐ different therapies differentiated.

1.8. Analysis.

Comparison 1 Haematologic response, Outcome 8 Haematologic response ‐ epoetin versus darbepoetin.

1.9. Analysis.

Comparison 1 Haematologic response, Outcome 9 Haematologic response ‐ duration of ESA medication.

1.10. Analysis.

Comparison 1 Haematologic response, Outcome 10 Haematologic response ‐ iron supplementation.

1.11. Analysis.

Comparison 1 Haematologic response, Outcome 11 Haematologic response ‐ allocation concealment.

1.12. Analysis.

Comparison 1 Haematologic response, Outcome 12 Haematologic response ‐ masking.

1.13. Analysis.

Comparison 1 Haematologic response, Outcome 13 Haematologic response ‐ intention‐to treat.

1.14. Analysis.

Comparison 1 Haematologic response, Outcome 14 Haematologic response ‐ publication.

1.15. Analysis.

Comparison 1 Haematologic response, Outcome 15 Haematological response ‐ merged experimental arms.

Comparison 2. Change of haemoglobin level.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in Hb values ‐ overall	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
2 Change in Hb values ‐ baseline Hb	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
2.1 Hb <=10g/dL	41	5092	Mean Difference (IV, Fixed, 95% CI)	1.53 [1.43, 1.62]
2.2 Hb 10 to 12 g/dL	23	3572	Mean Difference (IV, Fixed, 95% CI)	1.79 [1.71, 1.88]
2.3 Hb >12 g/dL	10	2824	Mean Difference (IV, Fixed, 95% CI)	1.12 [0.99, 1.25]
2.4 unclear	1	121	Mean Difference (IV, Fixed, 95% CI)	1.53 [0.91, 2.15]
3 Change in Hb values ‐ different malignancies	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
3.1 solid tumours	45	6262	Mean Difference (IV, Fixed, 95% CI)	1.76 [1.68, 1.83]
3.2 haematological malignancies	14	2391	Mean Difference (IV, Fixed, 95% CI)	1.05 [0.90, 1.19]
3.3 MDS	1	28	Mean Difference (IV, Fixed, 95% CI)	1.87 [0.89, 2.85]
3.4 mixed	15	2928	Mean Difference (IV, Fixed, 95% CI)	1.35 [1.21, 1.48]
3.5 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Change in Hb values ‐ age	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
4.1 children	1	222	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐0.27, 0.87]
4.2 adults	74	11387	Mean Difference (IV, Fixed, 95% CI)	1.58 [1.52, 1.64]
5 Change in Hb values ‐ age differentiated	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
5.1 only children <18 years	1	222	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐0.27, 0.87]
5.2 adults ≥18 years	62	8214	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.49, 1.64]
5.3 > 70% non‐elderly 18‐65 years	9	1136	Mean Difference (IV, Fixed, 95% CI)	1.99 [1.88, 2.10]
5.4 only non‐elderly adults	2	1992	Mean Difference (IV, Fixed, 95% CI)	0.76 [0.60, 0.91]
5.5 > 70% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.6 only elderly adults	1	45	Mean Difference (IV, Fixed, 95% CI)	2.70 [1.87, 3.53]
6 Change in Hb values ‐ different therapies	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
6.1 chemotherapy	58	8956	Mean Difference (IV, Fixed, 95% CI)	1.46 [1.39, 1.52]
6.2 radiotherapy/radiochemotherapy	8	974	Mean Difference (IV, Fixed, 95% CI)	2.41 [2.24, 2.58]
6.3 no therapy	9	1679	Mean Difference (IV, Fixed, 95% CI)	1.42 [1.25, 1.60]
7 Change in Hb values ‐ different therapies differentiated	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
7.1 chemotherapy, >70% with platinum	19	2126	Mean Difference (IV, Fixed, 95% CI)	1.68 [1.55, 1.82]
7.2 chemotherapy, <70% platinum containing	7	1315	Mean Difference (IV, Fixed, 95% CI)	1.84 [1.62, 2.05]
7.3 chemotherapy without platinum	22	4511	Mean Difference (IV, Fixed, 95% CI)	1.33 [1.25, 1.42]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	9	782	Mean Difference (IV, Fixed, 95% CI)	1.37 [1.13, 1.61]
7.5 chemotherapy, no details given	1	222	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐0.27, 0.87]
7.6 radiotherapy	5	696	Mean Difference (IV, Fixed, 95% CI)	2.32 [2.01, 2.63]
7.7 radiochemotherapy	3	278	Mean Difference (IV, Fixed, 95% CI)	2.45 [2.25, 2.65]
7.8 no therapy	9	1679	Mean Difference (IV, Fixed, 95% CI)	1.42 [1.25, 1.60]
8 Change in Hb values ‐ epoetin vs darbepoetin	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
8.1 Epoetin	57	8304	Mean Difference (IV, Fixed, 95% CI)	1.69 [1.62, 1.75]
8.2 Darbepoetin	18	3305	Mean Difference (IV, Fixed, 95% CI)	1.13 [1.00, 1.25]
9 Change in Hb values ‐ duration of ESA medication	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
9.1 6 to 9 weeks	17	1235	Mean Difference (IV, Fixed, 95% CI)	2.30 [2.16, 2.44]
9.2 12 to 16 weeks	43	6393	Mean Difference (IV, Fixed, 95% CI)	1.62 [1.54, 1.70]
9.3 more than 17 weeks	15	3981	Mean Difference (IV, Fixed, 95% CI)	1.05 [0.95, 1.16]
10 Change in Hb values ‐ iron supplementation	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
10.1 fixed iron supplementation	14	2483	Mean Difference (IV, Fixed, 95% CI)	1.94 [1.81, 2.07]
10.2 iron supplementation as necessary	53	8807	Mean Difference (IV, Fixed, 95% CI)	1.47 [1.40, 1.53]
10.3 no explicit statement on iron supplementation or no iron given	8	319	Mean Difference (IV, Fixed, 95% CI)	1.66 [1.34, 1.99]
10.4 explicitly stated no iron new	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 iron given differently in both study arms new ongoing	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11 Change in Hb values ‐ allocation concealment	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
11.1 adequate	45	6768	Mean Difference (IV, Fixed, 95% CI)	1.52 [1.45, 1.60]
11.2 unclear	30	4841	Mean Difference (IV, Fixed, 95% CI)	1.62 [1.53, 1.71]
12 Change in Hb values ‐ masking	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
12.1 double‐blind	42	7438	Mean Difference (IV, Fixed, 95% CI)	1.47 [1.40, 1.55]
12.2 unblinded	33	4171	Mean Difference (IV, Fixed, 95% CI)	1.67 [1.59, 1.75]
13 Change in Hb values ‐ intention‐to‐treat	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
13.1 ITT or less than 10% of participants per study arm excluded	57	9137	Mean Difference (IV, Fixed, 95% CI)	1.61 [1.55, 1.68]
13.2 more than 10% of participants per study arm excluded	18	2472	Mean Difference (IV, Fixed, 95% CI)	1.28 [1.13, 1.43]
13.3 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14 Change in Hb values ‐ publication	75	11609	Mean Difference (IV, Fixed, 95% CI)	1.57 [1.51, 1.62]
14.1 full text publication	59	10026	Mean Difference (IV, Fixed, 95% CI)	1.54 [1.47, 1.60]
14.2 abstract publication	3	252	Mean Difference (IV, Fixed, 95% CI)	1.70 [1.30, 2.09]
14.3 unpublished data	10	804	Mean Difference (IV, Fixed, 95% CI)	1.56 [1.36, 1.76]
14.4 ODAC documents	1	346	Mean Difference (IV, Fixed, 95% CI)	2.5 [2.14, 2.86]
14.5 Clinical trials results	2	181	Mean Difference (IV, Fixed, 95% CI)	1.48 [0.94, 2.02]
15 Change in Hb values ‐ experimental arms merged	56	11609	Mean Difference (IV, Fixed, 95% CI)	1.56 [1.51, 1.62]
16 Change in Hb values‐ sensitivity analysis	65	8685	Mean Difference (IV, Fixed, 95% CI)	1.66 [1.59, 1.72]
17 Change in Hb values ‐ publication sensitivity analysis excluding Henke 2003	74	11263	Mean Difference (IV, Fixed, 95% CI)	1.54 [1.48, 1.60]
17.1 full text publication	58	8723	Mean Difference (IV, Fixed, 95% CI)	1.63 [1.57, 1.70]
17.2 abstract publication	4	1555	Mean Difference (IV, Fixed, 95% CI)	0.87 [0.69, 1.04]
17.3 unpublished data	10	804	Mean Difference (IV, Fixed, 95% CI)	1.56 [1.36, 1.76]
17.4 ODAC documents	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.5 clinical study report	2	181	Mean Difference (IV, Fixed, 95% CI)	1.48 [0.94, 2.02]

2.1. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 1 Change in Hb values ‐ overall.

2.2. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 2 Change in Hb values ‐ baseline Hb.

2.3. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 3 Change in Hb values ‐ different malignancies.

2.4. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 4 Change in Hb values ‐ age.

2.5. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 5 Change in Hb values ‐ age differentiated.

2.6. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 6 Change in Hb values ‐ different therapies.

2.7. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 7 Change in Hb values ‐ different therapies differentiated.

2.8. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 8 Change in Hb values ‐ epoetin vs darbepoetin.

2.9. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 9 Change in Hb values ‐ duration of ESA medication.

2.10. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 10 Change in Hb values ‐ iron supplementation.

2.11. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 11 Change in Hb values ‐ allocation concealment.

2.12. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 12 Change in Hb values ‐ masking.

2.13. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 13 Change in Hb values ‐ intention‐to‐treat.

2.14. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 14 Change in Hb values ‐ publication.

2.15. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 15 Change in Hb values ‐ experimental arms merged.

2.16. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 16 Change in Hb values‐ sensitivity analysis.

2.17. Analysis.

Comparison 2 Change of haemoglobin level, Outcome 17 Change in Hb values ‐ publication sensitivity analysis excluding Henke 2003.

Comparison 3. Participants receiving red blood cell transfusions.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants receiving red blood cell transfusions ‐ overall	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
2 Participants receiving red blood cell transfusions ‐ baseline Hb	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
2.1 Hb <=10 g/dL	42	5605	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.67, 0.76]
2.2 Hb 10 to 12 g/dL	29	5669	Risk Ratio (M‐H, Fixed, 95% CI)	0.50 [0.46, 0.55]
2.3 Hb > 12 g/dL	17	4819	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.67, 0.78]
2.4 unclear	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Participants receiving red blood cell transfusions ‐ different malignancies	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
3.1 solid tumours	53	9305	Risk Ratio (M‐H, Fixed, 95% CI)	0.50 [0.46, 0.54]
3.2 haematological malignancies	15	2852	Risk Ratio (M‐H, Fixed, 95% CI)	0.80 [0.75, 0.86]
3.3 MDS	2	151	Risk Ratio (M‐H, Fixed, 95% CI)	0.80 [0.68, 0.96]
3.4 mixed	19	3785	Risk Ratio (M‐H, Fixed, 95% CI)	0.74 [0.68, 0.80]
3.5 not reported	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Participants receiving red blood cell transfusions ‐ age	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
4.1 children	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	0.84 [0.71, 0.99]
4.2 adults	87	15871	Risk Ratio (M‐H, Fixed, 95% CI)	0.64 [0.62, 0.67]
5 Participants receiving red blood cell transfusions ‐ age differentiated	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
5.1 only children <18 years	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	0.84 [0.71, 0.99]
5.2 adults ≥18 years	70	11556	Risk Ratio (M‐H, Fixed, 95% CI)	0.62 [0.59, 0.66]
5.3 >68% non‐elderly 18‐65 years	11	1343	Risk Ratio (M‐H, Fixed, 95% CI)	0.49 [0.41, 0.60]
5.4 only non‐elderly adults	5	2927	Risk Ratio (M‐H, Fixed, 95% CI)	0.78 [0.72, 0.84]
5.5 >68% elderly >65 years	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.6 only elderly adults	1	45	Risk Ratio (M‐H, Fixed, 95% CI)	0.40 [0.17, 0.94]
6 Participants receiving red blood cell transfusions ‐ different therapies	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
6.1 chemotherapy	71	13405	Risk Ratio (M‐H, Fixed, 95% CI)	0.64 [0.61, 0.67]
6.2 radio/radiochemotherapy	6	693	Risk Ratio (M‐H, Fixed, 95% CI)	0.45 [0.34, 0.58]
6.3 no therapy	10	1774	Risk Ratio (M‐H, Fixed, 95% CI)	0.81 [0.72, 0.91]
6.4 unclear/other	1	221	Risk Ratio (M‐H, Fixed, 95% CI)	0.77 [0.60, 0.99]
7 Participants receiving red blood cell transfusions ‐ different therapies differentiated	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
7.1 chemotherapy, > 70% with platinum	26	3592	Risk Ratio (M‐H, Fixed, 95% CI)	0.49 [0.44, 0.54]
7.2 chemotherapy, < 70% with platinum	10	2043	Risk Ratio (M‐H, Fixed, 95% CI)	0.63 [0.55, 0.73]
7.3 chemotherapy without platinum (all patients)	23	6509	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.69, 0.78]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	9	772	Risk Ratio (M‐H, Fixed, 95% CI)	0.53 [0.43, 0.65]
7.5 chemotherapy no details given	3	489	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.63, 0.86]
7.6 radiochemotherapy	6	693	Risk Ratio (M‐H, Fixed, 95% CI)	0.45 [0.34, 0.58]
7.7 radiotherapy	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.8 no therapy	10	1774	Risk Ratio (M‐H, Fixed, 95% CI)	0.81 [0.72, 0.91]
7.9 unclear/other	1	221	Risk Ratio (M‐H, Fixed, 95% CI)	0.77 [0.60, 0.99]
8 Participants receiving red blood cell transfusions ‐ epoetin versus darbepoetin	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
8.1 Epoetin	67	11786	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
8.2 Darbepoetin	21	4307	Risk Ratio (M‐H, Fixed, 95% CI)	0.66 [0.60, 0.72]
9 Participants receiving red blood cell transfusions ‐ duration of ESA medication	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
9.1 6 to 9 weeks	13	815	Risk Ratio (M‐H, Fixed, 95% CI)	0.60 [0.49, 0.74]
9.2 12 to 16 weeks	52	8413	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.61, 0.69]
9.3 more than 17 weeks	23	6865	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.61, 0.70]
10 Participants receiving red blood cell transfusions ‐ iron supplementation	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
10.1 fixed iron supplementation	10	2104	Risk Ratio (M‐H, Fixed, 95% CI)	0.58 [0.49, 0.70]
10.2 iron supplementation as necessary	63	12060	Risk Ratio (M‐H, Fixed, 95% CI)	0.68 [0.65, 0.71]
10.3 no explicit statement on iron supplementation or no iron given	11	645	Risk Ratio (M‐H, Fixed, 95% CI)	0.52 [0.42, 0.65]
10.4 explicitly stated NO IRON	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 iron given differently in both study arms new	4	1284	Risk Ratio (M‐H, Fixed, 95% CI)	0.51 [0.43, 0.61]
11 Participants receiving red blood cell transfusions ‐ allocation concealment	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
11.1 adequate	55	10898	Risk Ratio (M‐H, Fixed, 95% CI)	0.61 [0.58, 0.65]
11.2 unclear	33	5195	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.67, 0.76]
12 Participants receiving red blood cell transfusions ‐ masking	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
12.1 double‐blind	50	9677	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.68, 0.75]
12.2 unblinded	38	6416	Risk Ratio (M‐H, Fixed, 95% CI)	0.52 [0.47, 0.56]
13 Participants receiving red blood cell transfusions ‐ intention‐to treat	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
13.1 ITT or less than 10% of participants per study arm excluded	73	13772	Risk Ratio (M‐H, Fixed, 95% CI)	0.66 [0.63, 0.69]
13.2 more than 10% of participants per study arm excluded	11	1035	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.36, 0.63]
13.3 unclear	4	1286	Risk Ratio (M‐H, Fixed, 95% CI)	0.45 [0.33, 0.62]
14 Participants receiving red blood cell transfusions ‐ publication	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
14.1 fulltext publication	65	12678	Risk Ratio (M‐H, Fixed, 95% CI)	0.66 [0.63, 0.70]
14.2 abstract publication	7	1242	Risk Ratio (M‐H, Fixed, 95% CI)	0.40 [0.32, 0.49]
14.3 unpublished data	14	1658	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.67, 0.80]
14.4 FDA presented data	1	314	Risk Ratio (M‐H, Fixed, 95% CI)	0.60 [0.47, 0.78]
14.5 Other	1	201	Risk Ratio (M‐H, Fixed, 95% CI)	0.44 [0.26, 0.73]
15 Participants receiving red blood cell transfusions ‐ first 4 weeks are...	88	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]
15.1 included in the analysis	32	5319	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.61, 0.70]
15.2 excluded from the analysis	23	3288	Risk Ratio (M‐H, Fixed, 95% CI)	0.68 [0.61, 0.75]
15.3 unclear	33	7486	Risk Ratio (M‐H, Fixed, 95% CI)	0.64 [0.60, 0.68]
16 Participants receiving red blood cell transfusions ‐ experimental arms merged	70	16093	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.62, 0.68]

3.1. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 1 Participants receiving red blood cell transfusions ‐ overall.

3.2. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 2 Participants receiving red blood cell transfusions ‐ baseline Hb.

3.3. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 3 Participants receiving red blood cell transfusions ‐ different malignancies.

3.4. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 4 Participants receiving red blood cell transfusions ‐ age.

3.5. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 5 Participants receiving red blood cell transfusions ‐ age differentiated.

3.6. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 6 Participants receiving red blood cell transfusions ‐ different therapies.

3.7. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 7 Participants receiving red blood cell transfusions ‐ different therapies differentiated.

3.8. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 8 Participants receiving red blood cell transfusions ‐ epoetin versus darbepoetin.

3.9. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 9 Participants receiving red blood cell transfusions ‐ duration of ESA medication.

3.10. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 10 Participants receiving red blood cell transfusions ‐ iron supplementation.

3.11. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 11 Participants receiving red blood cell transfusions ‐ allocation concealment.

3.12. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 12 Participants receiving red blood cell transfusions ‐ masking.

3.13. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 13 Participants receiving red blood cell transfusions ‐ intention‐to treat.

3.14. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 14 Participants receiving red blood cell transfusions ‐ publication.

3.15. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 15 Participants receiving red blood cell transfusions ‐ first 4 weeks are....

3.16. Analysis.

Comparison 3 Participants receiving red blood cell transfusions, Outcome 16 Participants receiving red blood cell transfusions ‐ experimental arms merged.

Comparison 4. Number of red blood cell units transfused per patient.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of RBC units transfused ‐ overall	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
2 Number of RBC units transfused ‐ baseline Hb	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
2.1 Hb < 10g/dL	16	1996	Mean Difference (IV, Fixed, 95% CI)	‐1.02 [‐1.35, ‐0.68]
2.2 Hb 10 to 12g/dL	5	1096	Mean Difference (IV, Fixed, 95% CI)	‐0.95 [‐1.24, ‐0.65]
2.3 Hb > 12g/dL	4	1623	Mean Difference (IV, Fixed, 95% CI)	‐0.97 [‐1.41, ‐0.54]
3 Number of RBC units transfused ‐ age differentiated	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
3.1 only children <18 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 adults ≥18 years	21	3093	Mean Difference (IV, Fixed, 95% CI)	‐0.81 [‐1.04, ‐0.57]
3.3 >68% non‐elderly 18‐65 years	3	353	Mean Difference (IV, Fixed, 95% CI)	‐1.25 [‐1.72, ‐0.77]
3.4 only non‐elderly adults	1	1269	Mean Difference (IV, Fixed, 95% CI)	‐1.60 [‐2.18, ‐1.02]
3.5 >68% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.6 only elderly > 65 J	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Number of RBC units transfused ‐ age	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
4.1 adults	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
5 Number of RBC units transfused ‐ different malignancies	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
5.1 solid tumours	10	1437	Mean Difference (IV, Fixed, 95% CI)	‐1.07 [‐1.34, ‐0.80]
5.2 haematological malignancies	10	2254	Mean Difference (IV, Fixed, 95% CI)	‐1.19 [‐1.63, ‐0.76]
5.3 mixed	5	1024	Mean Difference (IV, Fixed, 95% CI)	‐0.62 [‐1.01, ‐0.24]
6 Number of RBC units transfused ‐ different therapies	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
6.1 chemotherapy	23	4376	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.18, ‐0.78]
6.2 radiotherapy/radiochemotherapy	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.3 no therapy	1	118	Mean Difference (IV, Fixed, 95% CI)	‐0.67 [‐1.81, 0.47]
6.4 unclear/ other	1	221	Mean Difference (IV, Fixed, 95% CI)	‐2.2 [‐6.44, 2.04]
7 Number of RBC units transfused ‐ different therapies differentiated	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
7.1 chemotherapy > 70% with platinum	10	1344	Mean Difference (IV, Fixed, 95% CI)	‐1.11 [‐1.40, ‐0.83]
7.2 chemotherapy, <70% with platinum	2	584	Mean Difference (IV, Fixed, 95% CI)	‐0.52 [‐0.93, ‐0.12]
7.3 chemotherapy without platinum	9	2054	Mean Difference (IV, Fixed, 95% CI)	‐1.23 [‐1.69, ‐0.78]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	2	394	Mean Difference (IV, Fixed, 95% CI)	‐1.01 [‐1.80, ‐0.21]
7.5 chemotherapy no details given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.6 radiochemotherapy	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.7 radiotherapy	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.8 no therapy	1	118	Mean Difference (IV, Fixed, 95% CI)	‐0.67 [‐1.81, 0.47]
7.9 unclear/other	1	221	Mean Difference (IV, Fixed, 95% CI)	‐2.2 [‐6.44, 2.04]
8 Number of RBC units transfused ‐ epoetin versus darbepoetin	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
8.1 Epoetin	23	4052	Mean Difference (IV, Fixed, 95% CI)	‐1.06 [‐1.29, ‐0.82]
8.2 Darbepoetin	2	663	Mean Difference (IV, Fixed, 95% CI)	‐0.78 [‐1.15, ‐0.42]
9 Number of RBC units transfused ‐ duration of ESA medication	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
9.1 6 to 9 weeks	6	364	Mean Difference (IV, Fixed, 95% CI)	‐1.25 [‐1.75, ‐0.75]
9.2 12 to 16 weeks	12	2688	Mean Difference (IV, Fixed, 95% CI)	‐0.78 [‐1.03, ‐0.54]
9.3 more than 17 weeks	7	1663	Mean Difference (IV, Fixed, 95% CI)	‐1.40 [‐1.84, ‐0.95]
10 Number of RBC units transfused ‐ iron supplementation	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
10.1 fixed iron supplementation	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.2 iron supplementation as necessary	24	4402	Mean Difference (IV, Fixed, 95% CI)	‐0.92 [‐1.13, ‐0.72]
10.3 no explicit statement on iron supplementation or no iron given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.4 explicitely stated NO IRON	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 iron given differently in both study arms	1	313	Mean Difference (IV, Fixed, 95% CI)	‐1.6 [‐2.30, ‐0.90]
11 Number of RBC units transfused ‐ allocation concealment	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
11.1 adequate	15	2243	Mean Difference (IV, Fixed, 95% CI)	‐0.93 [‐1.19, ‐0.67]
11.2 unclear	10	2472	Mean Difference (IV, Fixed, 95% CI)	‐1.04 [‐1.35, ‐0.74]
12 Number of RBC units transfused ‐ masking	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
12.1 double‐blind	12	3382	Mean Difference (IV, Fixed, 95% CI)	‐0.91 [‐1.14, ‐0.69]
12.2 unblinded	13	1333	Mean Difference (IV, Fixed, 95% CI)	‐1.16 [‐1.55, ‐0.77]
13 Number of RBC units transfused ‐ intention‐to‐treat	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
13.1 ITT or less than 10% of participants per study arm excluded	24	4583	Mean Difference (IV, Fixed, 95% CI)	‐0.99 [‐1.19, ‐0.79]
13.2 more than 10% of participants per study arm excluded	1	132	Mean Difference (IV, Fixed, 95% CI)	‐0.69 [‐1.66, 0.28]
14 Number of RBC units transfused ‐ publication	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
14.1 full text publication	4	1200	Mean Difference (IV, Fixed, 95% CI)	‐0.77 [‐1.06, ‐0.48]
14.2 abstract publication	1	1269	Mean Difference (IV, Fixed, 95% CI)	‐1.60 [‐2.18, ‐1.02]
14.3 unpublished data	20	2246	Mean Difference (IV, Fixed, 95% CI)	‐1.03 [‐1.33, ‐0.73]
14.4 FDA presented data	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15 Number of RBC units transfused ‐ first 4 weeks are...	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
15.1 included in the analysis	18	2658	Mean Difference (IV, Fixed, 95% CI)	‐0.99 [‐1.23, ‐0.74]
15.2 excluded from the analysis	3	623	Mean Difference (IV, Fixed, 95% CI)	‐0.53 [‐0.94, ‐0.12]
15.3 unclear	4	1434	Mean Difference (IV, Fixed, 95% CI)	‐1.72 [‐2.27, ‐1.18]
16 Number of RBC units transfused ‐ experimental arms merged	19	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
17 Number of RBC units transfused ‐ age differentiated sensitivity analysis	25	4715	Mean Difference (IV, Fixed, 95% CI)	‐0.98 [‐1.17, ‐0.78]
17.1 only children <18 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.2 adults ≥18 years	21	3093	Mean Difference (IV, Fixed, 95% CI)	‐0.81 [‐1.04, ‐0.57]
17.3 >68% non‐elderly 18‐65 years	4	1622	Mean Difference (IV, Fixed, 95% CI)	‐1.39 [‐1.75, ‐1.02]
17.4 only non‐elderly adults	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.5 >68% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.6 only elderly > 65 J	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]

4.1. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 1 Number of RBC units transfused ‐ overall.

4.2. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 2 Number of RBC units transfused ‐ baseline Hb.

4.3. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 3 Number of RBC units transfused ‐ age differentiated.

4.4. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 4 Number of RBC units transfused ‐ age.

4.5. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 5 Number of RBC units transfused ‐ different malignancies.

4.6. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 6 Number of RBC units transfused ‐ different therapies.

4.7. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 7 Number of RBC units transfused ‐ different therapies differentiated.

4.8. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 8 Number of RBC units transfused ‐ epoetin versus darbepoetin.

4.9. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 9 Number of RBC units transfused ‐ duration of ESA medication.

4.10. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 10 Number of RBC units transfused ‐ iron supplementation.

4.11. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 11 Number of RBC units transfused ‐ allocation concealment.

4.12. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 12 Number of RBC units transfused ‐ masking.

4.13. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 13 Number of RBC units transfused ‐ intention‐to‐treat.

4.14. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 14 Number of RBC units transfused ‐ publication.

4.15. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 15 Number of RBC units transfused ‐ first 4 weeks are....

4.16. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 16 Number of RBC units transfused ‐ experimental arms merged.

4.17. Analysis.

Comparison 4 Number of red blood cell units transfused per patient, Outcome 17 Number of RBC units transfused ‐ age differentiated sensitivity analysis.

Comparison 5. Overall survival.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Overall survival ‐ overall	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
2 Overall survival updated review (adjusted results)	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
3 Overall survival ‐ baseline Hb	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
3.1 Hb < 10 g/dL	29	6144	Peto Odds Ratio (95% CI)	1.06 [0.96, 1.17]
3.2 Hb 10 to 12 g/dL	31	6418	Peto Odds Ratio (95% CI)	1.01 [0.93, 1.10]
3.3 Hb > 12 g/dL	15	5725	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
3.4 unclear	5	716	Peto Odds Ratio (95% CI)	0.83 [0.67, 1.03]
4 Overall survival ‐ different malignancies	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
4.1 solid tumours	50	11704	Peto Odds Ratio (95% CI)	1.06 [1.00, 1.13]
4.2 haematological malignancies	11	2901	Peto Odds Ratio (95% CI)	1.07 [0.90, 1.26]
4.3 MDS	1	66	Peto Odds Ratio (95% CI)	4.52 [0.38, 53.37]
4.4 mixed	18	4332	Peto Odds Ratio (95% CI)	1.02 [0.91, 1.15]
4.5 not reported	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
5 Overall survival ‐ age	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
5.1 children	1	222	Peto Odds Ratio (95% CI)	0.98 [0.14, 7.03]
5.2 adults	79	18781	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
6 Overall survival ‐ age differentiated	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
6.1 only children < 18	1	222	Peto Odds Ratio (95% CI)	0.98 [0.14, 7.03]
6.2 adults >= 18	62	13975	Peto Odds Ratio (95% CI)	1.07 [1.01, 1.13]
6.3 > 68% non elderly 18‐65	11	1611	Peto Odds Ratio (95% CI)	0.86 [0.67, 1.10]
6.4 only non‐elderly adults	4	2916	Peto Odds Ratio (95% CI)	1.02 [0.82, 1.26]
6.5 > 68% elderly > 65	2	279	Peto Odds Ratio (95% CI)	0.77 [0.51, 1.17]
6.6 only elderly > 65	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
7 Overall survival ‐ different therapies	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
7.1 chemotherapy	55	13800	Peto Odds Ratio (95% CI)	1.04 [0.98, 1.11]
7.2 radiotherapy/radiochemotherapy	14	2939	Peto Odds Ratio (95% CI)	1.03 [0.92, 1.15]
7.3 no therapy	8	1942	Peto Odds Ratio (95% CI)	1.23 [1.04, 1.45]
7.4 unclear/other	3	322	Peto Odds Ratio (95% CI)	0.79 [0.49, 1.27]
8 Overall survival ‐ different therapies differentiated	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
8.1 chemotherapy, > 70% with platinum	19	3622	Peto Odds Ratio (95% CI)	1.01 [0.91, 1.12]
8.2 chemotherapy, < 70% with platinum	7	1812	Peto Odds Ratio (95% CI)	1.14 [0.95, 1.36]
8.3 chemotherapy without platinum (all patients)	19	6706	Peto Odds Ratio (95% CI)	1.08 [0.98, 1.19]
8.4 chemotherapy, platinum and non platinum containing, no numbers given	5	910	Peto Odds Ratio (95% CI)	0.94 [0.65, 1.37]
8.5 chemotherapy no details given	5	750	Peto Odds Ratio (95% CI)	0.82 [0.61, 1.10]
8.6 radiochemotherapy	8	1193	Peto Odds Ratio (95% CI)	0.92 [0.78, 1.09]
8.7 radiotherapy	6	1746	Peto Odds Ratio (95% CI)	1.12 [0.97, 1.30]
8.8 no therapy	8	1942	Peto Odds Ratio (95% CI)	1.23 [1.04, 1.45]
8.9 unclear/other	3	322	Peto Odds Ratio (95% CI)	0.79 [0.49, 1.27]
9 Overall survival ‐ epoetin vs darbepoetin	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
9.1 Epoetin	67	14047	Peto Odds Ratio (95% CI)	1.04 [0.98, 1.10]
9.2 Darbepoetin	13	4956	Peto Odds Ratio (95% CI)	1.09 [0.99, 1.20]
10 Overall survival ‐ duration of ESA medication	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
10.1 6 to 9 weeks	13	2244	Peto Odds Ratio (95% CI)	1.10 [0.95, 1.27]
10.2 12 to 16 weeks	46	9385	Peto Odds Ratio (95% CI)	1.05 [0.98, 1.13]
10.3 more than 17 weeks	18	6963	Peto Odds Ratio (95% CI)	1.04 [0.95, 1.14]
10.4 not reported	3	411	Peto Odds Ratio (95% CI)	1.10 [0.58, 2.07]
11 Overall survival ‐ iron supplementation	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
11.1 fixed iron supplementation	10	2348	Peto Odds Ratio (95% CI)	1.15 [0.99, 1.33]
11.2 iron supplementation as necessary	56	13888	Peto Odds Ratio (95% CI)	1.07 [1.01, 1.14]
11.3 iron handled differently in the study arms	9	2050	Peto Odds Ratio (95% CI)	0.96 [0.84, 1.11]
11.4 no explicit statement on iron supplementation or no iron given	5	717	Peto Odds Ratio (95% CI)	0.56 [0.38, 0.83]
11.5 explicitly stated NO IRON	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
12 Overall survival ‐ publication	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
12.1 full text publication	20	4531	Peto Odds Ratio (95% CI)	1.07 [0.92, 1.24]
12.2 abstract publication	3	1020	Peto Odds Ratio (95% CI)	1.09 [0.87, 1.37]
12.3 unpublished data	5	318	Peto Odds Ratio (95% CI)	0.73 [0.17, 3.05]
12.4 Data presented at FDA hearing	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
12.5 Data taken from IPD review (Bohlius 2009)	50	13018	Peto Odds Ratio (95% CI)	1.06 [1.00, 1.12]
12.6 clinical trial result information	1	60	Peto Odds Ratio (95% CI)	7.39 [0.15, 372.38]
12.7 other	1	56	Peto Odds Ratio (95% CI)	0.58 [0.33, 1.03]
13 Overall survival ‐ time‐to‐event or binary mortality data	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
13.1 binary mortality data at end of study	19	1917	Peto Odds Ratio (95% CI)	1.02 [0.66, 1.60]
13.2 Data from IPD review	50	13018	Peto Odds Ratio (95% CI)	1.06 [1.00, 1.12]
13.3 Cox regression analysis, Hazard ratio, log‐rank test, p‐value	7	3141	Peto Odds Ratio (95% CI)	1.07 [0.92, 1.25]
13.4 Survival curve and p‐value	4	927	Peto Odds Ratio (95% CI)	0.99 [0.78, 1.24]
14 Overall survival ‐ allocation concealment	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
14.1 adequate	47	12424	Peto Odds Ratio (95% CI)	1.08 [1.02, 1.14]
14.2 unclear	33	6579	Peto Odds Ratio (95% CI)	0.97 [0.87, 1.09]
15 Overall survival ‐ masking	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
15.1 double‐blind	42	10525	Peto Odds Ratio (95% CI)	1.06 [0.99, 1.14]
15.2 unblinded	38	8478	Peto Odds Ratio (95% CI)	1.04 [0.96, 1.12]
16 Overall survival ‐ intention‐to‐treat	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
16.1 ITT or less than 10% of participants per study arm excluded	75	18052	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
16.2 more than 10% of participants per study arm excluded	2	385	Peto Odds Ratio (95% CI)	1.25 [0.92, 1.71]
16.3 Unclear	3	566	Peto Odds Ratio (95% CI)	0.49 [0.27, 0.89]
17 Overall survival ‐ follow up	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
17.1 follow up longer than on‐study mortality	44	13224	Peto Odds Ratio (95% CI)	1.05 [0.99, 1.10]
17.2 Short term follow up	36	5779	Peto Odds Ratio (95% CI)	1.14 [0.95, 1.36]
18 Overall survival ‐ follow up and design	80	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
18.1 Long term follow up and designed for long term follow up	25	9704	Peto Odds Ratio (95% CI)	1.06 [0.99, 1.12]
18.2 Long term follow up but not designed for long term follow up	11	2197	Peto Odds Ratio (95% CI)	1.02 [0.89, 1.18]
18.3 Long term follow up and design unclear	7	1142	Peto Odds Ratio (95% CI)	0.96 [0.79, 1.16]
18.4 Short term follow up but designed for long term follow up	3	1250	Peto Odds Ratio (95% CI)	1.32 [1.05, 1.66]
18.5 Short term follow up and not designed for long term follow up	34	4710	Peto Odds Ratio (95% CI)	0.97 [0.76, 1.23]
18.6 short term follow up and design unclear	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
19 Overall survival‐ experimental arms merged	78	19003	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
20 Overall survival‐ experimental arms merged sens pos	76	17551	Peto Odds Ratio (95% CI)	1.04 [0.98, 1.10]
21 Overall survival‐ experimental arms merged sens neg	76	18018	Peto Odds Ratio (95% CI)	1.08 [1.02, 1.14]
22 Overall survival ‐ sensitivity analysis baseline Hb	75	18287	Peto Odds Ratio (95% CI)	1.07 [1.01, 1.13]
22.1 Hb < 10 g/dL	29	6144	Peto Odds Ratio (95% CI)	1.06 [0.96, 1.17]
22.2 Hb 10 to 12 g/dL	31	6418	Peto Odds Ratio (95% CI)	1.01 [0.93, 1.10]
22.3 Hb > 12 g/dL	15	5725	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
23 Overall survival ‐ sensitivity analysis iron supplementation	75	18286	Peto Odds Ratio (95% CI)	1.07 [1.01, 1.12]
23.1 fixed iron supplementation	10	2348	Peto Odds Ratio (95% CI)	1.15 [0.99, 1.33]
23.2 iron supplementation as necessary	56	13888	Peto Odds Ratio (95% CI)	1.07 [1.01, 1.14]
23.3 iron handled differently in the study arms	9	2050	Peto Odds Ratio (95% CI)	0.96 [0.84, 1.11]
23.4 explicitly stated NO IRON	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
24 Overall survival ‐ sensitivity analysis intention‐to‐treat	77	18437	Peto Odds Ratio (95% CI)	1.06 [1.01, 1.12]
24.1 ITT or less than 10% of participants per study arm excluded	75	18052	Peto Odds Ratio (95% CI)	1.05 [1.00, 1.11]
24.2 more than 10% of participants per study arm excluded	2	385	Peto Odds Ratio (95% CI)	1.25 [0.92, 1.71]

5.1. Analysis.

Comparison 5 Overall survival, Outcome 1 Overall survival ‐ overall.

5.2. Analysis.

Comparison 5 Overall survival, Outcome 2 Overall survival updated review (adjusted results).

5.3. Analysis.

Comparison 5 Overall survival, Outcome 3 Overall survival ‐ baseline Hb.

5.4. Analysis.

Comparison 5 Overall survival, Outcome 4 Overall survival ‐ different malignancies.

5.5. Analysis.

Comparison 5 Overall survival, Outcome 5 Overall survival ‐ age.

5.6. Analysis.

Comparison 5 Overall survival, Outcome 6 Overall survival ‐ age differentiated.

5.7. Analysis.

Comparison 5 Overall survival, Outcome 7 Overall survival ‐ different therapies.

5.8. Analysis.

Comparison 5 Overall survival, Outcome 8 Overall survival ‐ different therapies differentiated.

5.9. Analysis.

Comparison 5 Overall survival, Outcome 9 Overall survival ‐ epoetin vs darbepoetin.

5.10. Analysis.

Comparison 5 Overall survival, Outcome 10 Overall survival ‐ duration of ESA medication.

5.11. Analysis.

Comparison 5 Overall survival, Outcome 11 Overall survival ‐ iron supplementation.

5.12. Analysis.

Comparison 5 Overall survival, Outcome 12 Overall survival ‐ publication.

5.13. Analysis.

Comparison 5 Overall survival, Outcome 13 Overall survival ‐ time‐to‐event or binary mortality data.

5.14. Analysis.

Comparison 5 Overall survival, Outcome 14 Overall survival ‐ allocation concealment.

5.15. Analysis.

Comparison 5 Overall survival, Outcome 15 Overall survival ‐ masking.

5.16. Analysis.

Comparison 5 Overall survival, Outcome 16 Overall survival ‐ intention‐to‐treat.

5.17. Analysis.

Comparison 5 Overall survival, Outcome 17 Overall survival ‐ follow up.

5.18. Analysis.

Comparison 5 Overall survival, Outcome 18 Overall survival ‐ follow up and design.

5.19. Analysis.

Comparison 5 Overall survival, Outcome 19 Overall survival‐ experimental arms merged.

5.20. Analysis.

Comparison 5 Overall survival, Outcome 20 Overall survival‐ experimental arms merged sens pos.

5.21. Analysis.

Comparison 5 Overall survival, Outcome 21 Overall survival‐ experimental arms merged sens neg.

5.22. Analysis.

Comparison 5 Overall survival, Outcome 22 Overall survival ‐ sensitivity analysis baseline Hb.

5.23. Analysis.

Comparison 5 Overall survival, Outcome 23 Overall survival ‐ sensitivity analysis iron supplementation.

5.24. Analysis.

Comparison 5 Overall survival, Outcome 24 Overall survival ‐ sensitivity analysis intention‐to‐treat.

Comparison 6. On‐study mortality.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 On‐study mortality ‐ overall	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
2 On‐study mortality ‐ baseline Hb	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
2.1 Hb < 10 g/dL	28	5759	Peto Odds Ratio (95% CI)	1.12 [0.96, 1.32]
2.2 Hb 10 to 12 g/dL	26	5537	Peto Odds Ratio (95% CI)	1.09 [0.91, 1.29]
2.3 Hb > 12 g/dL	13	3923	Peto Odds Ratio (95% CI)	1.37 [1.12, 1.68]
2.4 unclear	5	716	Peto Odds Ratio (95% CI)	1.20 [0.75, 1.93]
3 On‐study mortality ‐ different malignancies	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
3.1 solid tumours	44	10056	Peto Odds Ratio (95% CI)	1.21 [1.06, 1.37]
3.2 haematological malignancies	10	1598	Peto Odds Ratio (95% CI)	1.13 [0.80, 1.59]
3.3 MDS	1	66	Peto Odds Ratio (95% CI)	4.52 [0.38, 53.37]
3.4 mixed	17	4215	Peto Odds Ratio (95% CI)	1.10 [0.92, 1.31]
3.5 not reported	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
4 On‐study mortality ‐ age	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
4.1 children	1	222	Peto Odds Ratio (95% CI)	0.98 [0.14, 7.03]
4.2 adults	71	15713	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
5 On‐study mortality ‐ age differentiated	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
5.1 children	1	222	Peto Odds Ratio (95% CI)	0.98 [0.14, 7.03]
5.2 adults >= 18	59	13007	Peto Odds Ratio (95% CI)	1.18 [1.07, 1.31]
5.3 > 68% non elderly, 18‐65	10	1334	Peto Odds Ratio (95% CI)	0.76 [0.42, 1.35]
5.4 only non elderly adults	2	1372	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
5.5 > 68% elderly > 65 years	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
5.6 only elderly	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
6 On‐study mortality ‐ different therapies	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
6.1 chemotherapy	52	12058	Peto Odds Ratio (95% CI)	1.10 [0.98, 1.24]
6.2 radiotherapy/radiochemotherapy	10	1669	Peto Odds Ratio (95% CI)	1.48 [0.96, 2.27]
6.3 no therapy	8	1942	Peto Odds Ratio (95% CI)	1.34 [1.07, 1.66]
6.4 unclear/other	2	266	Peto Odds Ratio (95% CI)	1.48 [0.65, 3.37]
7 On‐study mortality ‐ different therapies differentiated	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
7.1 chemotherapy, > 70% with platinum	19	3622	Peto Odds Ratio (95% CI)	1.04 [0.85, 1.27]
7.2 chemotherapy, <70% with platinum	6	1475	Peto Odds Ratio (95% CI)	1.06 [0.76, 1.49]
7.3 chemotherapy without platinum, all patients	18	5418	Peto Odds Ratio (95% CI)	1.24 [1.04, 1.47]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	4	793	Peto Odds Ratio (95% CI)	0.74 [0.40, 1.38]
7.5 chemotherapy, no details reported	5	750	Peto Odds Ratio (95% CI)	0.82 [0.50, 1.34]
7.6 radiochemotherapy	6	822	Peto Odds Ratio (95% CI)	1.46 [0.85, 2.51]
7.7 radiotherapy	4	847	Peto Odds Ratio (95% CI)	1.51 [0.75, 3.06]
7.8 no therapy	8	1942	Peto Odds Ratio (95% CI)	1.34 [1.07, 1.66]
7.9 unclear/other	2	266	Peto Odds Ratio (95% CI)	1.48 [0.65, 3.37]
8 On‐study mortality ‐ duration of ESA medication	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
8.1 6 to 9 weeks	9	1113	Peto Odds Ratio (95% CI)	0.95 [0.57, 1.59]
8.2 12 to 16 weeks	44	8992	Peto Odds Ratio (95% CI)	1.16 [1.02, 1.33]
8.3 more than 17 weeks	16	5419	Peto Odds Ratio (95% CI)	1.22 [1.04, 1.42]
8.4 not reported	3	411	Peto Odds Ratio (95% CI)	0.82 [0.38, 1.78]
9 On‐study mortality ‐ epoetin vs darbepoetin	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
9.1 Epoetin	60	11478	Peto Odds Ratio (95% CI)	1.16 [1.03, 1.30]
9.2 Darbepoetin	12	4457	Peto Odds Ratio (95% CI)	1.20 [1.00, 1.44]
10 On‐study mortality ‐ iron supplementation	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
10.1 fixed iron supplementation	8	1911	Peto Odds Ratio (95% CI)	1.20 [0.80, 1.81]
10.2 iron supplementation as necessary	53	11954	Peto Odds Ratio (95% CI)	1.17 [1.05, 1.30]
10.3 no explicit statement on iron supplementation or no iron given	3	276	Peto Odds Ratio (95% CI)	2.16 [0.72, 6.46]
10.4 explicitly stated no iron	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
10.5 iron handled differently in the study arms	8	1794	Peto Odds Ratio (95% CI)	1.09 [0.74, 1.60]
11 On‐study mortality ‐ publication	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
11.1 full text publication	14	1689	Peto Odds Ratio (95% CI)	0.97 [0.60, 1.57]
11.2 abstract publication	2	181	Peto Odds Ratio (95% CI)	7.64 [1.29, 45.03]
11.3 unpublished data	5	318	Peto Odds Ratio (95% CI)	1.65 [0.20, 13.32]
11.4 Data presented at FDA hearing	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
11.5 IPD Bohlius 2009	51	13747	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
12 On‐study mortality ‐ time‐to‐event or binary mortality data	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
12.1 binary mortality data at end of study	21	2188	Peto Odds Ratio (95% CI)	1.14 [0.73, 1.79]
12.2 Results from IPD review	51	13747	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
12.3 Cox regression analysis, Hazard ratio, log‐rank test, p‐value	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
12.4 Survival curve and p‐value	0	0	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
13 On‐study mortality ‐ allocation concealment	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
13.1 adequate	42	11144	Peto Odds Ratio (95% CI)	1.20 [1.08, 1.34]
13.2 unclear	30	4791	Peto Odds Ratio (95% CI)	1.03 [0.82, 1.30]
14 On‐study mortality ‐ masking	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
14.1 double‐blind	39	9049	Peto Odds Ratio (95% CI)	1.19 [1.06, 1.34]
14.2 unblinded	33	6886	Peto Odds Ratio (95% CI)	1.12 [0.94, 1.34]
15 On‐study mortality ‐ intention‐to‐treat	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
15.1 ITT or less than 10% of participants per study arm excluded	69	15706	Peto Odds Ratio (95% CI)	1.16 [1.05, 1.28]
15.2 more than 10% of participants per study arm excluded	1	48	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
15.3 Unclear	2	181	Peto Odds Ratio (95% CI)	7.64 [1.29, 45.03]
16 On‐study mortality ‐ sensitivity analysis ‐ follow‐up	80	19018	Peto Odds Ratio (95% CI)	1.16 [1.05, 1.27]
16.1 Short term follow subgroup	72	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
16.2 Short term from Kaplan Meier curve	7	2966	Peto Odds Ratio (95% CI)	1.06 [0.60, 1.86]
16.3 Long term follow up	1	117	Peto Odds Ratio (95% CI)	0.94 [0.57, 1.54]
17 On‐study mortality ‐ sensitivity analysis experimental arms merged	70	15935	Peto Odds Ratio (95% CI)	1.17 [1.06, 1.29]
18 On‐study mortality ‐ sensitivity analysis intention‐to‐treat	70	15754	Peto Odds Ratio (95% CI)	1.16 [1.05, 1.28]
18.1 ITT or less than 10% of participants per study arm excluded	69	15706	Peto Odds Ratio (95% CI)	1.16 [1.05, 1.28]
18.2 more than 10% of participants per study arm excluded	1	48	Peto Odds Ratio (95% CI)	0.0 [0.0, 0.0]
19 On‐study mortality ‐ sensitivity analysis excluding Leyland and Smith	70	14007	Peto Odds Ratio (95% CI)	1.09 [0.97, 1.23]

6.1. Analysis.

Comparison 6 On‐study mortality, Outcome 1 On‐study mortality ‐ overall.

6.2. Analysis.

Comparison 6 On‐study mortality, Outcome 2 On‐study mortality ‐ baseline Hb.

6.3. Analysis.

Comparison 6 On‐study mortality, Outcome 3 On‐study mortality ‐ different malignancies.

6.4. Analysis.

Comparison 6 On‐study mortality, Outcome 4 On‐study mortality ‐ age.

6.5. Analysis.

Comparison 6 On‐study mortality, Outcome 5 On‐study mortality ‐ age differentiated.

6.6. Analysis.

Comparison 6 On‐study mortality, Outcome 6 On‐study mortality ‐ different therapies.

6.7. Analysis.

Comparison 6 On‐study mortality, Outcome 7 On‐study mortality ‐ different therapies differentiated.

6.8. Analysis.

Comparison 6 On‐study mortality, Outcome 8 On‐study mortality ‐ duration of ESA medication.

6.9. Analysis.

Comparison 6 On‐study mortality, Outcome 9 On‐study mortality ‐ epoetin vs darbepoetin.

6.10. Analysis.

Comparison 6 On‐study mortality, Outcome 10 On‐study mortality ‐ iron supplementation.

6.11. Analysis.

Comparison 6 On‐study mortality, Outcome 11 On‐study mortality ‐ publication.

6.12. Analysis.

Comparison 6 On‐study mortality, Outcome 12 On‐study mortality ‐ time‐to‐event or binary mortality data.

6.13. Analysis.

Comparison 6 On‐study mortality, Outcome 13 On‐study mortality ‐ allocation concealment.

6.14. Analysis.

Comparison 6 On‐study mortality, Outcome 14 On‐study mortality ‐ masking.

6.15. Analysis.

Comparison 6 On‐study mortality, Outcome 15 On‐study mortality ‐ intention‐to‐treat.

6.16. Analysis.

Comparison 6 On‐study mortality, Outcome 16 On‐study mortality ‐ sensitivity analysis ‐ follow‐up.

6.17. Analysis.

Comparison 6 On‐study mortality, Outcome 17 On‐study mortality ‐ sensitivity analysis experimental arms merged.

6.18. Analysis.

Comparison 6 On‐study mortality, Outcome 18 On‐study mortality ‐ sensitivity analysis intention‐to‐treat.

6.19. Analysis.

Comparison 6 On‐study mortality, Outcome 19 On‐study mortality ‐ sensitivity analysis excluding Leyland and Smith.

Comparison 7. Complete tumour response.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Complete tumour response	19	5012	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.98, 1.06]
2 Tumour‐response specific quality criteria	19	5012	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.98, 1.06]
2.1 high quality	5	2476	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.95, 1.02]
2.2 low quality	14	2536	Risk Ratio (M‐H, Fixed, 95% CI)	1.15 [0.98, 1.36]
3 Complete tumour response ‐ experimental study arms merged	15	5012	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.98, 1.06]

7.1. Analysis.

Comparison 7 Complete tumour response, Outcome 1 Complete tumour response.

7.2. Analysis.

Comparison 7 Complete tumour response, Outcome 2 Tumour‐response specific quality criteria.

7.3. Analysis.

Comparison 7 Complete tumour response, Outcome 3 Complete tumour response ‐ experimental study arms merged.

Comparison 8. Change in FACT‐Fatigue 13.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in FACT‐Fatigue (13 items) ‐ overall	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
2 Change in FACT‐Fatigue sensitivity analysis‐ Tsuboi	18	4967	Mean Difference (IV, Fixed, 95% CI)	2.10 [1.46, 2.75]
3 Change in FACT‐F 13 ‐ baseline Hb	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
3.1 Hb <= 10 g/dL	8	2484	Mean Difference (IV, Fixed, 95% CI)	1.66 [0.76, 2.55]
3.2 Hb 10 to 12 g/dL	9	2181	Mean Difference (IV, Fixed, 95% CI)	2.87 [1.89, 3.85]
3.3 Hb > 12 g/dL	1	300	Mean Difference (IV, Fixed, 95% CI)	0.0 [‐2.63, 2.63]
3.4 Hb category unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Change in FACT‐F 13 ‐ different malignancies	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
4.1 solid tumours	9	2459	Mean Difference (IV, Fixed, 95% CI)	2.29 [1.33, 3.25]
4.2 haematological malignancies	2	566	Mean Difference (IV, Fixed, 95% CI)	1.99 [0.28, 3.69]
4.3 mixed	7	1940	Mean Difference (IV, Fixed, 95% CI)	1.87 [0.87, 2.87]
4.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Change in FACT‐F 13 ‐ age	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
5.1 only children < 18 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 adults >= 18 years	17	4627	Mean Difference (IV, Fixed, 95% CI)	1.82 [1.16, 2.49]
5.3 >70% non elderly 18‐65 years	1	338	Mean Difference (IV, Fixed, 95% CI)	5.1 [2.79, 7.41]
5.4 only non‐elderly adults	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.5 >70% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.6 only elderly > 65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Change in FACT‐F 13 ‐ different therapies	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
6.1 chemotherapy	14	3515	Mean Difference (IV, Fixed, 95% CI)	2.79 [2.03, 3.55]
6.2 radiotherapy	1	300	Mean Difference (IV, Fixed, 95% CI)	0.0 [‐2.63, 2.63]
6.3 no therapy	3	1150	Mean Difference (IV, Fixed, 95% CI)	0.42 [‐0.91, 1.76]
6.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Change in FACT‐F 13 ‐ different therapies differentiated	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.07 [1.43, 2.72]
7.1 chemotherapy, >70% with platinum	4	1069	Mean Difference (IV, Fixed, 95% CI)	1.85 [0.41, 3.30]
7.2 chemotherapy, <70% platinum containing	2	411	Mean Difference (IV, Fixed, 95% CI)	2.53 [‐0.01, 5.07]
7.3 chemotherapy without platinum (all patients)	6	1468	Mean Difference (IV, Fixed, 95% CI)	3.22 [2.12, 4.32]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	3	567	Mean Difference (IV, Fixed, 95% CI)	3.27 [1.30, 5.23]
7.5 radiochemotherapy NEW	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.6 radiotherapy	1	300	Mean Difference (IV, Fixed, 95% CI)	0.0 [‐2.63, 2.63]
7.7 no therapy	3	1150	Mean Difference (IV, Fixed, 95% CI)	0.42 [‐0.91, 1.76]
7.8 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.9 chemotherapy no details given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Change in FACT‐F 13 ‐ epoetin versus darbepoetin	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
8.1 Epoetin	11	2475	Mean Difference (IV, Fixed, 95% CI)	3.25 [2.33, 4.16]
8.2 Darbepoetin	7	2490	Mean Difference (IV, Fixed, 95% CI)	0.94 [0.04, 1.84]
9 Change in FACT‐F 13 ‐ duration of ESA medication	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
9.1 6 to 9 weeks	1	115	Mean Difference (IV, Fixed, 95% CI)	3.10 [‐0.27, 6.47]
9.2 12 to 16 weeks	15	4076	Mean Difference (IV, Fixed, 95% CI)	1.99 [1.30, 2.69]
9.3 more than 17 weeks	2	774	Mean Difference (IV, Fixed, 95% CI)	2.35 [0.46, 4.25]
9.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10 Change in FACT‐F 13 ‐ iron supplementation	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
10.1 fixed iron supplementation	4	863	Mean Difference (IV, Fixed, 95% CI)	1.64 [‐0.01, 3.29]
10.2 iron supplementation as necessary	13	3881	Mean Difference (IV, Fixed, 95% CI)	2.03 [1.32, 2.74]
10.3 no explicit statement on iron supplementation or no iron given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.4 explicitly stated NO IRON	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.6 iron given differently in both study arms new	1	221	Mean Difference (IV, Fixed, 95% CI)	5.15 [1.70, 8.60]
11 Change in FACT‐F 13 ‐ allocation concealment	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
11.1 adequate	16	4493	Mean Difference (IV, Fixed, 95% CI)	2.20 [1.51, 2.89]
11.2 unclear	2	472	Mean Difference (IV, Fixed, 95% CI)	1.20 [‐0.61, 3.02]
11.3 unclear wether adequate or unclear :)	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12 Change in FACT‐F 13 ‐ masking	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
12.1 double‐blind	11	3384	Mean Difference (IV, Fixed, 95% CI)	1.33 [0.56, 2.10]
12.2 unblinded	7	1581	Mean Difference (IV, Fixed, 95% CI)	3.76 [2.60, 4.92]
13 Change in FACT‐F 13 ‐ intention‐to treat	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
13.1 ITT or less than 10% of participants per study arm excluded	4	864	Mean Difference (IV, Fixed, 95% CI)	3.30 [1.89, 4.70]
13.2 more than 10% of participants per study arm excluded	12	3689	Mean Difference (IV, Fixed, 95% CI)	1.80 [1.04, 2.56]
13.3 unclear	2	412	Mean Difference (IV, Fixed, 95% CI)	1.34 [‐0.95, 3.63]
14 Change in FACT‐F 13 ‐ publication	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
14.1 full text publication	16	4412	Mean Difference (IV, Fixed, 95% CI)	2.17 [1.49, 2.85]
14.2 abstract publication	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.3 unpublished data	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.4 FDA hearing	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.5 clinical trial result information	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.6 other source	2	553	Mean Difference (IV, Fixed, 95% CI)	1.32 [‐0.63, 3.26]
15 Change in FACT‐F 13 ‐ type of data	18	4965	Mean Difference (IV, Fixed, 95% CI)	2.08 [1.43, 2.72]
15.1 Not imputed data	7	2298	Mean Difference (IV, Fixed, 95% CI)	1.88 [0.96, 2.80]
15.2 Imputed data	11	2667	Mean Difference (IV, Fixed, 95% CI)	2.26 [1.36, 3.15]

8.1. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 1 Change in FACT‐Fatigue (13 items) ‐ overall.

8.2. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 2 Change in FACT‐Fatigue sensitivity analysis‐ Tsuboi.

8.3. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 3 Change in FACT‐F 13 ‐ baseline Hb.

8.4. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 4 Change in FACT‐F 13 ‐ different malignancies.

8.5. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 5 Change in FACT‐F 13 ‐ age.

8.6. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 6 Change in FACT‐F 13 ‐ different therapies.

8.7. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 7 Change in FACT‐F 13 ‐ different therapies differentiated.

8.8. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 8 Change in FACT‐F 13 ‐ epoetin versus darbepoetin.

8.9. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 9 Change in FACT‐F 13 ‐ duration of ESA medication.

8.10. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 10 Change in FACT‐F 13 ‐ iron supplementation.

8.11. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 11 Change in FACT‐F 13 ‐ allocation concealment.

8.12. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 12 Change in FACT‐F 13 ‐ masking.

8.13. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 13 Change in FACT‐F 13 ‐ intention‐to treat.

8.14. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 14 Change in FACT‐F 13 ‐ publication.

8.15. Analysis.

Comparison 8 Change in FACT‐Fatigue 13, Outcome 15 Change in FACT‐F 13 ‐ type of data.

Comparison 9. Change in FACT‐An 20.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in FACT‐An (20 items) ‐ overall	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
2 Change in FACT‐An 20 ‐ baseline Hb	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
2.1 Hb <= 10 g/dL	1	290	Mean Difference (IV, Fixed, 95% CI)	6.6 [3.92, 9.28]
2.2 Hb 10 to 12 g/dL	4	713	Mean Difference (IV, Fixed, 95% CI)	5.82 [3.71, 7.93]
2.3 Hb > 12 g/dL	1	82	Mean Difference (IV, Fixed, 95% CI)	6.4 [0.83, 11.97]
2.4 Hb category unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Change in FACT‐An 20 ‐ different malignancies	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
3.1 solid tumours	5	795	Mean Difference (IV, Fixed, 95% CI)	5.90 [3.92, 7.87]
3.2 haematological malignancies	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.3 mixed	1	290	Mean Difference (IV, Fixed, 95% CI)	6.6 [3.92, 9.28]
3.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Change in FACT‐An 20 ‐ age	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
4.1 only children < 18 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.2 adults >= 18 years	5	747	Mean Difference (IV, Fixed, 95% CI)	5.93 [4.00, 7.86]
4.3 >70% non elderly 18‐65 years	1	338	Mean Difference (IV, Fixed, 95% CI)	6.59 [3.79, 9.39]
4.4 only non‐elderly adults	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.5 >70% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.6 only elderly > 65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Change in FACT‐An 20 ‐ different therapies	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
5.1 chemotherapy	5	1051	Mean Difference (IV, Fixed, 95% CI)	6.21 [4.60, 7.82]
5.2 radiotherapy	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.3 no therapy	1	34	Mean Difference (IV, Fixed, 95% CI)	3.9 [‐5.75, 13.55]
5.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Change in FACT‐An 20 ‐ different therapies differentiated	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
6.1 chemotherapy, >70% with platinum	2	256	Mean Difference (IV, Fixed, 95% CI)	5.08 [1.12, 9.04]
6.2 chemotherapy, <70% platinum containing	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.3 chemotherapy without platinum (all patients)	4	795	Mean Difference (IV, Fixed, 95% CI)	6.43 [4.66, 8.19]
6.4 chemotherapy, platinum and non‐platinum containing, no numbers given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.5 radiochemotherapy NEW	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.6 radiotherapy	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.7 no therapy	1	34	Mean Difference (IV, Fixed, 95% CI)	3.9 [‐5.75, 13.55]
6.8 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.9 chemotherapy no details given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Change in FACT‐An 20 ‐ epoetin versus darbepoetin	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
7.1 Epoetin	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
7.2 Darbepoetin	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Change in FACT‐An 20 ‐ duration of ESA medication	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
8.1 6 to 9 weeks	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8.2 12 to 16 weeks	5	795	Mean Difference (IV, Fixed, 95% CI)	5.90 [3.92, 7.87]
8.3 more than 17 weeks	1	290	Mean Difference (IV, Fixed, 95% CI)	6.6 [3.92, 9.28]
8.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9 Change in FACT‐An 20 ‐ iron supplementation	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
9.1 fixed iron supplementation	1	126	Mean Difference (IV, Fixed, 95% CI)	3.53 [‐1.90, 8.96]
9.2 iron supplementation as necessary	4	744	Mean Difference (IV, Fixed, 95% CI)	6.48 [4.68, 8.28]
9.3 no explicit statement on iron supplementation or no iron given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.4 explicitly stated NO IRON	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.5 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.6 iron given differently in both study arms new	1	215	Mean Difference (IV, Fixed, 95% CI)	5.84 [1.47, 10.21]
10 Change in FACT‐An 20 ‐ allocation concealment	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
10.1 adequate	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
10.2 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.3 unclear wether adequate or unclear :)	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11 Change in FACT‐An 20 ‐ masking	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
11.1 double‐blind	3	406	Mean Difference (IV, Fixed, 95% CI)	6.41 [4.06, 8.75]
11.2 unblinded	3	679	Mean Difference (IV, Fixed, 95% CI)	5.92 [3.76, 8.08]
12 Change in FACT‐An 20 ‐ intention‐to treat	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
12.1 ITT or less than 10% of participants per study arm excluded	1	338	Mean Difference (IV, Fixed, 95% CI)	6.59 [3.79, 9.39]
12.2 more than 10% of participants per study arm excluded	5	747	Mean Difference (IV, Fixed, 95% CI)	5.93 [4.00, 7.86]
12.3 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13 Change in FACT‐An 20 ‐ publication	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
13.1 full text publication	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
13.2 abstract publication	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13.3 unpublished data	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13.4 FDA hearing	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13.5 clinical trial result information	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14 Change in FACT‐An 20 ‐ data type	6	1085	Mean Difference (IV, Fixed, 95% CI)	6.14 [4.55, 7.73]
14.1 Not imputed data	3	635	Mean Difference (IV, Fixed, 95% CI)	6.38 [4.20, 8.55]
14.2 Imputed data	3	450	Mean Difference (IV, Fixed, 95% CI)	5.88 [3.54, 8.21]

9.1. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 1 Change in FACT‐An (20 items) ‐ overall.

9.2. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 2 Change in FACT‐An 20 ‐ baseline Hb.

9.3. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 3 Change in FACT‐An 20 ‐ different malignancies.

9.4. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 4 Change in FACT‐An 20 ‐ age.

9.5. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 5 Change in FACT‐An 20 ‐ different therapies.

9.6. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 6 Change in FACT‐An 20 ‐ different therapies differentiated.

9.7. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 7 Change in FACT‐An 20 ‐ epoetin versus darbepoetin.

9.8. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 8 Change in FACT‐An 20 ‐ duration of ESA medication.

9.9. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 9 Change in FACT‐An 20 ‐ iron supplementation.

9.10. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 10 Change in FACT‐An 20 ‐ allocation concealment.

9.11. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 11 Change in FACT‐An 20 ‐ masking.

9.12. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 12 Change in FACT‐An 20 ‐ intention‐to treat.

9.13. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 13 Change in FACT‐An 20 ‐ publication.

9.14. Analysis.

Comparison 9 Change in FACT‐An 20, Outcome 14 Change in FACT‐An 20 ‐ data type.

Comparison 10. Change in FACT‐An Total 47.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in FACT‐An Total (47 items) ‐ overall	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
2 Change in FACT‐An Total 47‐ sensitivity analysis	9	1715	Mean Difference (IV, Fixed, 95% CI)	3.46 [0.96, 5.96]
3 Change in FACT‐An Total 47 ‐ baseline Hb	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
3.1 Hb <= 10 g/dL	5	978	Mean Difference (IV, Fixed, 95% CI)	2.57 [‐0.63, 5.78]
3.2 Hb 10 to 12 g/dL	4	537	Mean Difference (IV, Fixed, 95% CI)	16.22 [12.14, 20.30]
3.3 Hb > 12 g/dL	1	300	Mean Difference (IV, Fixed, 95% CI)	1.90 [‐4.19, 7.99]
3.4 Hb category unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Change in FACT‐An Total 47 ‐ different malignancies	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
4.1 solid tumours	8	1437	Mean Difference (IV, Fixed, 95% CI)	7.21 [4.58, 9.84]
4.2 haematological malignancies	1	206	Mean Difference (IV, Fixed, 95% CI)	6.10 [‐1.67, 13.87]
4.3 mixed	1	172	Mean Difference (IV, Fixed, 95% CI)	5.7 [‐0.83, 12.23]
4.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Change in FACT‐An Total 47 ‐ age	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
5.1 only children < 18 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 adults >= 18 years	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
5.3 >70% non elderly 18‐65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.4 only non‐elderly adults	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.5 >70% elderly >65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.6 only elderly > 65 years	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Change in FACT‐An Total 47 ‐ different therapies	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
6.1 chemotherapy	8	1415	Mean Difference (IV, Fixed, 95% CI)	3.77 [1.03, 6.51]
6.2 radiotherapy	1	300	Mean Difference (IV, Fixed, 95% CI)	1.90 [‐4.19, 7.99]
6.3 no therapy	1	100	Mean Difference (IV, Fixed, 95% CI)	29.90 [23.46, 36.34]
6.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Change in FACT‐An Total 47 ‐ different therapies differentiated	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
7.1 chemotherapy, >70% with platinum	3	352	Mean Difference (IV, Fixed, 95% CI)	6.91 [1.12, 12.70]
7.2 chemotherapy, <70% platinum containing	3	600	Mean Difference (IV, Fixed, 95% CI)	0.28 [‐3.90, 4.45]
7.3 chemotherapy without platinum (all patients)	3	463	Mean Difference (IV, Fixed, 95% CI)	6.10 [1.44, 10.76]
7.4 chemotherapy, platinum and non‐platinum containing, no numbers given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.5 radiochemotherapy NEW	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.6 radiotherapy	1	300	Mean Difference (IV, Fixed, 95% CI)	1.90 [‐4.19, 7.99]
7.7 no therapy	1	100	Mean Difference (IV, Fixed, 95% CI)	29.90 [23.46, 36.34]
7.8 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.9 chemotherapy no details given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Change in FACT‐An Total 47 ‐ epoetin versus darbepoetin	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
8.1 Epoetin	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
8.2 Darbepoetin	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9 Change in FACT‐An Total 47 ‐ duration of ESA medication	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
9.1 6 to 9 weeks	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.2 12 to 16 weeks	8	1625	Mean Difference (IV, Fixed, 95% CI)	3.39 [0.83, 5.96]
9.3 more than 17 weeks	2	190	Mean Difference (IV, Fixed, 95% CI)	23.64 [18.05, 29.22]
9.4 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10 Change in FACT‐An Total 47 ‐ iron supplementation	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
10.1 fixed iron supplementation	7	1217	Mean Difference (IV, Fixed, 95% CI)	7.14 [4.31, 9.96]
10.2 iron supplementation as necessary	2	378	Mean Difference (IV, Fixed, 95% CI)	5.87 [0.86, 10.87]
10.3 no explicit statement on iron supplementation or no iron given	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.4 explicitly stated NO IRON	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.5 unclear	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.6 iron given differently in both study arms new	1	220	Mean Difference (IV, Fixed, 95% CI)	7.67 [0.40, 14.94]
11 Change in FACT‐An Total 47 ‐ allocation concealment	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
11.1 adequate	4	840	Mean Difference (IV, Fixed, 95% CI)	4.69 [0.92, 8.45]
11.2 unclear	6	975	Mean Difference (IV, Fixed, 95% CI)	8.30 [5.33, 11.26]
12 Change in FACT‐An Total 47 ‐ masking	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
12.1 double‐blind	6	1078	Mean Difference (IV, Fixed, 95% CI)	8.00 [5.13, 10.87]
12.2 unblinded	4	737	Mean Difference (IV, Fixed, 95% CI)	4.83 [0.84, 8.82]
13 Change in FACT‐An Total 47 ‐ intention‐to treat	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
13.1 ITT or less than 10% of participants per study arm excluded	4	585	Mean Difference (IV, Fixed, 95% CI)	10.85 [7.29, 14.41]
13.2 more than 10% of participants per study arm excluded	4	840	Mean Difference (IV, Fixed, 95% CI)	4.69 [0.92, 8.45]
13.3 unclear	2	390	Mean Difference (IV, Fixed, 95% CI)	2.54 [‐2.81, 7.90]
14 Change in FACT‐An Total 47 ‐ publication	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
14.1 full text publication	9	1725	Mean Difference (IV, Fixed, 95% CI)	7.02 [4.63, 9.40]
14.2 abstract publication	1	90	Mean Difference (IV, Fixed, 95% CI)	4.71 [‐6.49, 15.91]
14.3 unpublished data	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.4 FDA hearing	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.5 clinical trial result information	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15 Change in Fact‐An Total 47 ‐ data type	10	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]
15.1 Not imputed data	7	1101	Mean Difference (IV, Fixed, 95% CI)	9.31 [6.45, 12.17]
15.2 Imputed data	3	714	Mean Difference (IV, Fixed, 95% CI)	2.19 [‐1.83, 6.20]
16 FACT‐An Total 47 ‐ merged experimental study arms	9	1815	Mean Difference (IV, Fixed, 95% CI)	6.92 [4.59, 9.25]

10.1. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 1 Change in FACT‐An Total (47 items) ‐ overall.

10.2. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 2 Change in FACT‐An Total 47‐ sensitivity analysis.

10.3. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 3 Change in FACT‐An Total 47 ‐ baseline Hb.

10.4. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 4 Change in FACT‐An Total 47 ‐ different malignancies.

10.5. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 5 Change in FACT‐An Total 47 ‐ age.

10.6. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 6 Change in FACT‐An Total 47 ‐ different therapies.

10.7. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 7 Change in FACT‐An Total 47 ‐ different therapies differentiated.

10.8. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 8 Change in FACT‐An Total 47 ‐ epoetin versus darbepoetin.

10.9. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 9 Change in FACT‐An Total 47 ‐ duration of ESA medication.

10.10. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 10 Change in FACT‐An Total 47 ‐ iron supplementation.

10.11. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 11 Change in FACT‐An Total 47 ‐ allocation concealment.

10.12. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 12 Change in FACT‐An Total 47 ‐ masking.

10.13. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 13 Change in FACT‐An Total 47 ‐ intention‐to treat.

10.14. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 14 Change in FACT‐An Total 47 ‐ publication.

10.15. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 15 Change in Fact‐An Total 47 ‐ data type.

10.16. Analysis.

Comparison 10 Change in FACT‐An Total 47, Outcome 16 FACT‐An Total 47 ‐ merged experimental study arms.

Comparison 11. Thrombotic events.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Thrombotic events ‐ overall	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
2 Thrombotic events ‐ baseline Hb	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
2.1 Hb < 10 g/dL	19	4231	Risk Ratio (M‐H, Fixed, 95% CI)	1.41 [1.06, 1.88]
2.2 Hb 10 to 12 g/dL	26	5491	Risk Ratio (M‐H, Fixed, 95% CI)	1.64 [1.33, 2.03]
2.3 Hb > 12 g/dL	13	5348	Risk Ratio (M‐H, Fixed, 95% CI)	1.44 [1.15, 1.80]
2.4 Hb unclear	2	428	Risk Ratio (M‐H, Fixed, 95% CI)	1.64 [1.02, 2.65]
3 Thrombotic events ‐ different malignancies	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
3.1 solid tumours	36	9121	Risk Ratio (M‐H, Fixed, 95% CI)	1.67 [1.41, 1.96]
3.2 haematological malignancies	8	2531	Risk Ratio (M‐H, Fixed, 95% CI)	1.33 [0.93, 1.90]
3.3 mixed	14	3693	Risk Ratio (M‐H, Fixed, 95% CI)	1.29 [0.98, 1.69]
3.4 MDS	2	153	Risk Ratio (M‐H, Fixed, 95% CI)	2.08 [0.23, 18.84]
3.5 not reported	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Thrombotic events ‐ age	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
4.1 children	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	2.95 [0.61, 14.28]
4.2 adults	59	15276	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.33, 1.73]
5 Thrombotic events ‐ age differentiated	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
5.1 children	1	222	Risk Ratio (M‐H, Fixed, 95% CI)	2.95 [0.61, 14.28]
5.2 adults >= 18	46	11310	Risk Ratio (M‐H, Fixed, 95% CI)	1.60 [1.38, 1.86]
5.3 > 68% non elderly, 18‐65	7	1028	Risk Ratio (M‐H, Fixed, 95% CI)	1.71 [1.02, 2.86]
5.4 only non elderly	5	2882	Risk Ratio (M‐H, Fixed, 95% CI)	1.17 [0.86, 1.60]
5.5 > 68% elderly > 65 years	1	56	Risk Ratio (M‐H, Fixed, 95% CI)	0.23 [0.01, 4.58]
5.6 only elderly	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Thrombotic events ‐ different therapies	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
6.1 chemotherapy	37	10844	Risk Ratio (M‐H, Fixed, 95% CI)	1.48 [1.27, 1.73]
6.2 radiotherapy/radiochemotherapy	11	2384	Risk Ratio (M‐H, Fixed, 95% CI)	2.02 [1.44, 2.83]
6.3 no therapy	8	1921	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [0.88, 1.78]
6.4 unclear/other	4	349	Risk Ratio (M‐H, Fixed, 95% CI)	2.09 [0.74, 5.93]
7 Thrombotic events ‐ different therapies differentiated	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
7.1 chemotherapy, > 70% with platinum	16	3136	Risk Ratio (M‐H, Fixed, 95% CI)	1.46 [1.13, 1.89]
7.2 chemotherapy, < 70% with platinum	3	919	Risk Ratio (M‐H, Fixed, 95% CI)	1.63 [0.92, 2.88]
7.3 chemotherapy, without platinum, all patients	14	6085	Risk Ratio (M‐H, Fixed, 95% CI)	1.42 [1.14, 1.76]
7.4 chemotherapy, platinum and non platinum containing, no numbers given	2	265	Risk Ratio (M‐H, Fixed, 95% CI)	4.56 [0.79, 26.20]
7.5 chemotherapy no details given	2	439	Risk Ratio (M‐H, Fixed, 95% CI)	1.79 [0.67, 4.76]
7.6 radiochemotherapy	7	1097	Risk Ratio (M‐H, Fixed, 95% CI)	2.01 [1.37, 2.96]
7.7 radiotherapy	4	1287	Risk Ratio (M‐H, Fixed, 95% CI)	2.03 [1.02, 4.07]
7.8 no therapy	8	1921	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [0.88, 1.78]
7.9 unclear/other	4	349	Risk Ratio (M‐H, Fixed, 95% CI)	2.09 [0.74, 5.93]
8 Thrombotic events ‐ epoetin versus darbepoetin	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
8.1 Epoetin	50	11055	Risk Ratio (M‐H, Fixed, 95% CI)	1.57 [1.33, 1.85]
8.2 Darbepoetin	10	4443	Risk Ratio (M‐H, Fixed, 95% CI)	1.44 [1.16, 1.79]
9 Thrombotic events ‐ duration of ESA treatment	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
9.1 6 to 9 weeks	10	1719	Risk Ratio (M‐H, Fixed, 95% CI)	2.06 [1.17, 3.64]
9.2 12 to 16 weeks	30	7223	Risk Ratio (M‐H, Fixed, 95% CI)	1.46 [1.20, 1.77]
9.3 more than 17 weeks	19	6312	Risk Ratio (M‐H, Fixed, 95% CI)	1.53 [1.27, 1.84]
9.4 not reported	1	244	Risk Ratio (M‐H, Fixed, 95% CI)	1.46 [0.15, 13.85]
10 Thrombotic events ‐ iron supplementation	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
10.1 fixed iron supplementation	4	1445	Risk Ratio (M‐H, Fixed, 95% CI)	1.53 [0.92, 2.54]
10.2 iron supplementation as necessary	47	12073	Risk Ratio (M‐H, Fixed, 95% CI)	1.51 [1.31, 1.74]
10.3 no explicit statement on iron supplementation or no iron given	4	293	Risk Ratio (M‐H, Fixed, 95% CI)	1.26 [0.39, 4.06]
10.4 iron handled differently in the study arms	5	1687	Risk Ratio (M‐H, Fixed, 95% CI)	1.69 [1.12, 2.54]
10.5 explicitly stated NO IRON	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
11 Thrombotic events ‐ concealment of allocation	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
11.1 adequate	38	10494	Risk Ratio (M‐H, Fixed, 95% CI)	1.64 [1.40, 1.92]
11.2 unclear	22	5004	Risk Ratio (M‐H, Fixed, 95% CI)	1.31 [1.04, 1.65]
12 Thrombotic events ‐ masking	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
12.1 double‐blind	32	9209	Risk Ratio (M‐H, Fixed, 95% CI)	1.40 [1.19, 1.64]
12.2 unblinded	28	6289	Risk Ratio (M‐H, Fixed, 95% CI)	1.78 [1.43, 2.23]
13 Thrombotic events ‐ intention‐to‐treat	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
13.1 ITT or less than 10% of participants per study arm excluded	55	13182	Risk Ratio (M‐H, Fixed, 95% CI)	1.50 [1.30, 1.72]
13.2 more than 10% of participants per study arm excluded	3	1589	Risk Ratio (M‐H, Fixed, 95% CI)	1.59 [0.95, 2.66]
13.3 unclear	2	727	Risk Ratio (M‐H, Fixed, 95% CI)	1.82 [1.15, 2.89]
14 Thrombotic events ‐ publication	60	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.34, 1.74]
14.1 full text publication	35	8388	Risk Ratio (M‐H, Fixed, 95% CI)	1.57 [1.32, 1.87]
14.2 abstract publication	2	1343	Risk Ratio (M‐H, Fixed, 95% CI)	1.04 [0.70, 1.55]
14.3 unpublished data	1	66	Risk Ratio (M‐H, Fixed, 95% CI)	1.43 [0.06, 33.82]
14.4 data presented at ODAC hearing	21	5645	Risk Ratio (M‐H, Fixed, 95% CI)	1.66 [1.33, 2.08]
14.5 other	1	56	Risk Ratio (M‐H, Fixed, 95% CI)	0.23 [0.01, 4.58]
15 Thrombotic events ‐ experimental arms merged	57	15498	Risk Ratio (M‐H, Fixed, 95% CI)	1.53 [1.34, 1.74]

11.1. Analysis.

Comparison 11 Thrombotic events, Outcome 1 Thrombotic events ‐ overall.

11.2. Analysis.

Comparison 11 Thrombotic events, Outcome 2 Thrombotic events ‐ baseline Hb.

11.3. Analysis.

Comparison 11 Thrombotic events, Outcome 3 Thrombotic events ‐ different malignancies.

11.4. Analysis.

Comparison 11 Thrombotic events, Outcome 4 Thrombotic events ‐ age.

11.5. Analysis.

Comparison 11 Thrombotic events, Outcome 5 Thrombotic events ‐ age differentiated.

11.6. Analysis.

Comparison 11 Thrombotic events, Outcome 6 Thrombotic events ‐ different therapies.

11.7. Analysis.

Comparison 11 Thrombotic events, Outcome 7 Thrombotic events ‐ different therapies differentiated.

11.8. Analysis.

Comparison 11 Thrombotic events, Outcome 8 Thrombotic events ‐ epoetin versus darbepoetin.

11.9. Analysis.

Comparison 11 Thrombotic events, Outcome 9 Thrombotic events ‐ duration of ESA treatment.

11.10. Analysis.

Comparison 11 Thrombotic events, Outcome 10 Thrombotic events ‐ iron supplementation.

11.11. Analysis.

Comparison 11 Thrombotic events, Outcome 11 Thrombotic events ‐ concealment of allocation.

11.12. Analysis.

Comparison 11 Thrombotic events, Outcome 12 Thrombotic events ‐ masking.

11.13. Analysis.

Comparison 11 Thrombotic events, Outcome 13 Thrombotic events ‐ intention‐to‐treat.

11.14. Analysis.

Comparison 11 Thrombotic events, Outcome 14 Thrombotic events ‐ publication.

11.15. Analysis.

Comparison 11 Thrombotic events, Outcome 15 Thrombotic events ‐ experimental arms merged.

Comparison 12. Hypertension.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Hypertension ‐ overall	37	7228	Risk Ratio (M‐H, Fixed, 95% CI)	1.30 [1.08, 1.56]
2 Hypertension ‐ merged experimental study arms	31	7228	Risk Ratio (M‐H, Fixed, 95% CI)	1.31 [1.09, 1.58]
3 Hypertension ‐ sensitivity analysis Dammacco	37	7228	Risk Ratio (M‐H, Fixed, 95% CI)	1.29 [1.10, 1.52]
4 Hypertension ‐ sensitivity analysis random effects	37	7228	Risk Ratio (M‐H, Random, 95% CI)	1.12 [0.94, 1.33]
5 Hypertension ‐ sensitivity analysis without Rose	36	7007	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.18, 1.97]

12.1. Analysis.

Comparison 12 Hypertension, Outcome 1 Hypertension ‐ overall.

12.2. Analysis.

Comparison 12 Hypertension, Outcome 2 Hypertension ‐ merged experimental study arms.

12.3. Analysis.

Comparison 12 Hypertension, Outcome 3 Hypertension ‐ sensitivity analysis Dammacco.

12.4. Analysis.

Comparison 12 Hypertension, Outcome 4 Hypertension ‐ sensitivity analysis random effects.

12.5. Analysis.

Comparison 12 Hypertension, Outcome 5 Hypertension ‐ sensitivity analysis without Rose.

Comparison 13. Thrombocytopenia or haemorrhage.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Thrombocytopenia ‐ overall	24	4507	Risk Ratio (M‐H, Fixed, 95% CI)	1.21 [1.04, 1.42]
2 Thrombocytopenia ‐ merged experimental arms	21	4507	Risk Ratio (M‐H, Fixed, 95% CI)	1.21 [1.04, 1.41]
3 Thrombocytopenia ‐ sensitivity analysis random effects	24	4507	Risk Ratio (M‐H, Random, 95% CI)	1.18 [1.02, 1.36]

13.1. Analysis.

Comparison 13 Thrombocytopenia or haemorrhage, Outcome 1 Thrombocytopenia ‐ overall.

13.2. Analysis.

Comparison 13 Thrombocytopenia or haemorrhage, Outcome 2 Thrombocytopenia ‐ merged experimental arms.

13.3. Analysis.

Comparison 13 Thrombocytopenia or haemorrhage, Outcome 3 Thrombocytopenia ‐ sensitivity analysis random effects.

Comparison 14. Rash.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Rash ‐ overall	18	2485	Risk Ratio (M‐H, Fixed, 95% CI)	1.49 [0.99, 2.24]
2 Rash ‐ merged experimental arms	16	2485	Risk Ratio (M‐H, Fixed, 95% CI)	1.50 [1.00, 2.27]

14.1. Analysis.

Comparison 14 Rash, Outcome 1 Rash ‐ overall.

14.2. Analysis.

Comparison 14 Rash, Outcome 2 Rash ‐ merged experimental arms.

Comparison 15. Seizure.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Seizure ‐ overall	8	2890	Risk Ratio (M‐H, Fixed, 95% CI)	0.77 [0.42, 1.41]

15.1. Analysis.

Comparison 15 Seizure, Outcome 1 Seizure ‐ overall.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Aapro 2008.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 463 randomised: ESA = 231; control = 232 disease: breast cancer (M1) treatment: chemotherapy mean/median baseline Hb: 11.4 g/dL
Interventions	drug: Epoetin beta dose: 30000 IU sc weekly Hb‐target: 13‐15 g/dL planned ESA duration: 24 weeks
Outcomes	primary: overall survival secondary: progression free survival, tumour response rate, QoL
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 97413)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Aapro 2008 IPD.

Methods	see Aapro 2008
Participants
Interventions
Outcomes
Notes

Abels 1993.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 124 randomised: ESA = 65; control = 59 disease: hematological malignancies, genitourinary, gastrointestinal, and other cancer; except primary myeloid malignancy or acute leukaemia (category: mixed) treatment: none mean/median baseline Hb: 9.3 g/dL;
Interventions	drug: Epoetin alpha dose: 100 IU/kg tiw s.c. Hb‐target: not reported duration: 8 weeks
Outcomes	primary: transfusion, Hct secondary: QoL, safety
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 98906)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ each patient was assigned a random identification number and was assigned to a treatment group by a computerised randomisation schedule

Abels 1993 IPD.

Methods	see Abels 1993
Participants
Interventions
Outcomes
Notes

Abels 1993 J&J 2004.

Methods	see Abels 1993
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Antonadou 2001.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 385, randomised: ESA = 190, control = 195 disease: pelvic malignancies treatment: radiotherapy baseline Hb: 9.95 g/dL , ESA 9.8, control 10.1, , categorized as < 10g/dL
Interventions	drug: Epoetin dose: 10.000 U 5x/week s.c. Hb‐target: >= 13 g/dL duration: 5‐6 weeks
Outcomes	4 years disease free survival, safety, Hb, tumour control, overall survival
Notes	abstract, poster, study number = 10176
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	NR
Allocation concealment (selection bias)	Unclear risk	NR

Aravantinos 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 47 randomised: ESA = 24; control = 23 disease: ovarian, lung, stomach, other cancer (categorised as solid) treatment: platinum‐based chemotherapy Hb baseline: 9.6 g/dL
Interventions	drug: Epoetin alpha dose: 150 IU/kg tiw sc Hb target: 14 g/dL duration: NR, approx. >9‐12 weeks, categorized: 12‐16 weeks
Outcomes	primary: transfusion requirements (secondary: ?) Hct, Hb, RBC number
Notes	full text publication, study number = 11595
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Unclear risk	Unclear

Bamias 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 144 randomised: ESA = 72; control = 72 disease: ovarian, NSCLC, SCLC, other cancer (categorized: solid) treatment: platinum‐based chemotherapy Hb baseline: 11.5 g/dL
Interventions	drug: Epoetin alpha dose: 10,000 IU tiw sc Hb target: 13 ‐ 15 g/dL duration: 21 to 24 weeks (duration of chemotherapy)
Outcomes	primary: transfusions secondary: Hb < 10 g/dL, predictors of response, optional: QoL
Notes	full text publication, study number = 16091
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Unclear risk	Unclear

Blohmer 2011.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 257, randomised: ESA = 128, control = 129 disease: cervical cancer treatment: platinum‐containing chemotherapy in all patients and radiotherapy (categorized as radiochemotherapy) baseline Hb: 11.9 g/dL , ESA 12.0 g/dL, control 11.8 g/dL, categorised as 10‐12 g/dL
Interventions	drug: Epoetin alfa dose: 10'000 IU sc. TIW Hb‐target: >14 g/dL duration: >20 weeks
Outcomes	primary: relapse free survival
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 16218
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer random‐numbers generator
Allocation concealment (selection bias)	Low risk	Yes ‐ central registration of the patients for treatment allocation

Boogaerts 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 262 randomised: ESA = 133; control = 129 disease: multiple myeloma, Non‐Hodgkin lymphoma, chronic lymphocytic leukaemia, Hodgkin disease, ovarian, bone, gastrointestinal, respiratory, other cancer treatment: chemotherapy baseline Hb: 9.0 g/dL
Interventions	drug: Epoetin beta dose: 150 IU/kg sc TIW Hb‐target: 12‐14 g/dL planned ESA duration: 12 weeks
Outcomes	primary: QoL secondary: hematologic response, hematopoietic response, Hb change, transfusions, PS, Hct
Notes	full text publication of the study previously published as abstract Coiffier 2001, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 36158)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Boogaerts 2003 IPD.

Methods	see Boogaerts 2003
Participants
Interventions
Outcomes
Notes

Boogaerts Coiffier 2001.

Methods	see Boogaerts 2003
Participants
Interventions
Outcomes
Notes	Abstract

Carabantes 1999.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 35, randomised: ESA = 20, control = 15 disease: SCLC and ovarian carcinoma treatment: platinum‐containing chemotherapy baseline Hb: (at randomisation): 10.5 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: NR control: no treatment duration: 18‐24 weeks
Outcomes	haematologic response, transfusion requirements, QoL
Notes	abstract publication, study number = 17026
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Cascinu 1994.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 100, randomised: ESA = 50; control = 50 disease: various solid tumours treatment: concomitant platinum‐based chemotherapy; some patients received G‐CSF (n = 27) mean/median baseline Hb: 8.7 g/dL
Interventions	drug: Epoetin alpha dose: 100 U/kg 3x/week s.c. Hb target: 10 ‐ 12 g/dL duration: 9 weeks
Outcomes	haematologic response, change in Hb values, transfusion requirement, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 19548
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer random‐number generator
Allocation concealment (selection bias)	Unclear risk	unclear ‐ sealed envelopes

Case 1993.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 157, randomised: ESA = 81; control = 76 disease: non myeloid hematological malignancies, breast, lung, gynaecological, gastrointestinal, other cancer treatment: non‐cisplatin chemotherapy mean/median baseline Hematocrit: 28.9%
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: Hct 38%‐40% duration: 12 weeks
Outcomes	haematologic response, change in Hct, transfusion requirement, QoL, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 34917)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ description is unclear

Case 1993 IPD.

Methods	see Case 1993
Participants
Interventions
Outcomes
Notes	study number = 34917

Case 1993 J&J 2004.

Methods	see Case 1993
Participants
Interventions
Outcomes
Notes

Cazzola 1995.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 146, randomised: control = 29 (IPD: control:30, EPO: 116), evaluated EPO:114, control: 29 ESAa = 31; ESAb = 29; ESAc = 31; ESAd = 26; ESAtotal = 117 disease: multiple myeloma, Non‐Hodgkin´s Lymphoma treatment: chemotherapy, assumed without platinum because of hematological disease mean/median baseline Hb: 9.4 g/dL
Interventions	drug: Epoetin beta dosages: a: 1000 IU sc 7x/week; b: 2000 IU sc 7x/week; c: 5000 IU sc 7x/ week; d: 10000 IU sc 7x/week Hb‐target: 11‐13 g/dL (MM), 11‐15 g/dL (NHL) a: 1000 IU sc 7x/week, b: 2000 IU sc 7x/week; c: 5000 IU sc 7x/ week; d: 10000 IU sc 7x/week duration: 8 weeks
Outcomes	primary: haematologic response secondary: Hb, Hct, transfusions, reticulocytes, iron, ferritin, safety
Notes	full text publication, additional unpublished data obtained for first Cochrane Review and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 37653)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation list
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Cazzola 1995 IPD.

Methods	seeCazzola 1995
Participants
Interventions
Outcomes
Notes

Cazzola 1995a.

Methods	see Cazzola 1995
Participants
Interventions
Outcomes
Notes

Cazzola 1995b.

Methods	see Cazzola 1995
Participants
Interventions
Outcomes
Notes

Cazzola 1995c.

Methods	see Cazzola 1995
Participants
Interventions
Outcomes
Notes

Cazzola 1995d.

Methods	see Cazzola 1995
Participants
Interventions
Outcomes
Notes

Chang 2005.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 354, randomised: ESA = 176; control = 178 disease: breast cancer, stage I‐IV treatment: chemotherapy baseline Hb: 11.3 g/dL
Interventions	drug: Epoetin alpha dose: 40,000 IU qw sc Hb target: 14 g/dL duration: 16 weeks, max 28 weeks
Outcomes	primary: QoL secondary: maintain Hb above 12 g/dL, tumour response, overall survival
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 99137)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Chang 2005 IPD.

Methods	see Chang 2005
Participants
Interventions
Outcomes
Notes	study number = 99137

Charu 2007.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 287, randomised: ESA = 228, control = 59 disease: lymphoma, breast, lung, gastrointestinal, genitourinary, gynaecologic, other cancer treatment: none baseline Hb: 10.2 g/dL
Interventions	drug: Darbepoetin alpha dose: 3.0 µg/kg sc Q2W Hb‐target: 13‐14 g/dL (women), 13‐15 g/dL (men) duration: 12 weeks
Outcomes	primary: hospitalisation days secondary: costs, QoL, transfusion, Hb, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 53081)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Charu 2007 Amgen 2007.

Methods	see Charu 2007
Participants
Interventions
Outcomes
Notes

Charu 2007 IPD.

Methods	see Charu 2007
Participants
Interventions
Outcomes
Notes

Christodoulou 2009.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 399, randomised: NR, evaluated: ESA 167, control = 170 disease: solid tumours treatment: chemotherapy, platinum and non‐platinum containing baseline Hb: 10.2 g/dL
Interventions	drug: Epoetin alfa dose: 10'000 IU TIW Hb‐target: 12 ‐ 14 g/dL duration: minimum anticipated duration 12 weeks. categorized 12‐16 weeks
Outcomes	primary: QoL secondary: transfusions, anaemia
Notes	full text publication, abstract in 2003 (Janinis), study number = 22108
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ block randomisation
Allocation concealment (selection bias)	Low risk	yes ‐ centrally randomised

Dammacco 2001.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 145, randomised: ESA = 69; control = 76 disease: multiple myeloma treatment: chemotherapy mean/median baseline Hb: 9.5 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: 14 g/dL duration: 12 weeks
Outcomes	primary: transfusion secondary: haematologic response, Hb, Hct, reticulocytes, serum erythropoietin levels, QoL, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 11220)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation schedule prepared by RWJPRI
Allocation concealment (selection bias)	Unclear risk	unclear ‐ two randomisation lists (pat. prev. transfused or not), when patient enters the study the next number was to be assigned

Dammacco 2001 IPD.

Methods	see Dammacco 2001
Participants
Interventions
Outcomes
Notes

Dammacco 2001 J&J 2004.

Methods	see Dammacco 2001
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Debus 2006 IPD.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 385, randomised: ESA = 195, control = 190 disease: NSCLC (stage III, primarily inoperable) treatment: radiochemotherapy baseline Hb: not reported, unclear
Interventions	drug: Epoetin alpha dose: 40'000 IU sc weekly Hb‐target: 12‐14 g/dL, in 11/2003 reduced to 12‐13 g/dL duration: assumed to be 12‐16 weeks
Outcomes	primary: 2‐year‐survival rate secondary: tumour response, QoL, tolerance to epoetin alpha, Hb change, transfusion, safety
Notes	only unpublished data available, were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 83322)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation code provided by OrthoBiotech
Allocation concealment (selection bias)	Unclear risk	unclear ‐ assigned envelopes, sequentially numbered, but it is unclear whether they were sealed and opaque

Debus 2006 J&J 2004.

Methods	see Debus 2006 IPD
Participants
Interventions
Outcomes
Notes

Debus 2006 J&J 2007.

Methods	see Debus 2006 IPD
Participants
Interventions
Outcomes
Notes

Del Mastro 1997.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 62, randomised: ESA = 31, control = 31 disease: breast cancer treatment: non‐platinum based chemotherapy and G‐CSF 5µg/kg d4‐d11 s.c for all patients; radiotherapy and Tamoxifen fore the majority mean/median baseline Hb: 13.1 g/dL
Interventions	drug: Epoetin (?) dose: 150U/kg 3x/week s.c. Hb target: 13‐15 g/dL duration: 14 weeks
Outcomes	change in Hb values, transfusion requirement, QoL, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 24367
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer random‐number generator
Allocation concealment (selection bias)	Low risk	Yes ‐ central allocation

Dunphy 1999.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 30, randomised: ESA = 15, control = 15 disease: head and neck cancer, NSCLC treatment: platinum‐based chemotherapy mean/median baseline Hb: 14.1 g/dL
Interventions	drug: Epoetin (?) dose: 150U/kg 3x/week s.c. Hb target: 16‐18 g/dL duration: 6 weeks
Outcomes	change in Hb values, transfusion requirement
Notes	full text publication, study number = 25455
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Unclear risk	Unclear

Engert 2009.

Methods	See Engert 2010
Participants
Interventions
Outcomes
Notes

Engert 2010.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 1,379, randomised ESA: 685, placebo: 694 disease: advanced stage Hodgkin Lymphoma treatment: chemotherapy without platinum baseline Hb: 12.5 g/dL
Interventions	drug: Epoetin alpha dose: 40,000 IU /week Hb target: 12‐13 g/dL duration: > 20 weeks
Outcomes	primary: anaemia‐related fatigue secondary: other QoL, number of transfusions needed, Hb during and after treatment, safety, freedom from treatment failure, OS
Notes	full‐text publication, additional unpublished data, study number = 27258
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear, not reported
Allocation concealment (selection bias)	Unclear risk	unclear, not reported

EPO‐GER‐20 IPD.

Methods	randomised controlled trial, not placebo‐controlled
Participants	n = 93, randomised: ESA = 45, control = 48 disease: SCLC (extensive stage) treatment: chemotherapy baseline Hb: not reported, unclear
Interventions	drug: Epoetin alpha dose: 10000 IU sc TIW Hb‐target: 12‐14 g/dL duration: during chemotherapy
Outcomes	primary: rate of patients with anaemia secondary: QoL, tolerability of ESA, transfusion, effectiveness of chemotherapy
Notes	only IPD data, study number = 31678
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ patients were assigned with a randomisation code provided by Janssen‐Cilag
Allocation concealment (selection bias)	Unclear risk	unclear ‐ assigned envelopes, sequentially numbered, but it is unclear whether they were sealed and opaque

EPO‐INT‐1 IPD.

Methods	see EPO‐INT‐1 J&J 2004
Participants
Interventions
Outcomes
Notes

EPO‐INT‐1 J&J 2004.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 246, randomised: ESA = 165, control = 81 disease: ovarian cancer (stage I‐IV) treatment: chemotherapy baseline Hb: not reported, eligibility criterion Hb < 11 g/dL or Hb drop 1.5 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: Epoetin alpha dose: a: 150, b: 300 IU/kg tiw sc Hb‐target: 14 g/dL duration: 1 month
Outcomes	primary: transfusion secondary: Hb change, Hct, QoL, survival
Notes	data presented by J&J at FDA/ODAC hearing in may 2004, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 53915)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

EPO‐INT‐3 IPD.

Methods	see EPO‐INT‐3 J&J 2004
Participants
Interventions
Outcomes
Notes

EPO‐INT‐3 J&J 2004.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 201, randomised: ESA = 136, control = 65 disease: breast, NHL, MM, ovarian, SCLC, other cancer treatment: chemotherapy, < 70% platinum containing baseline Hb: not reported, eligibility criterion Hb < 12 g/dL or Hb drop 1.5 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: Epoetin alpha dose: 150‐300 IU/kg tiw sc Hb‐target: 14 g/dL for women and 16 g/dL for men duration: 12 weeks
Outcomes	primary: transfusions secondary: mortality, disease progression, tumour response, adverse events, Hb, QoL
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 36274), clinicaltrials.gov
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ according to randomisation schedule prepared by RWJPRI
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Fujisaka 2011.

Methods	randomised controlled trial, placebo‐controlled
Participants	randomised N = 186, evaluated N = 181, ESA = 89, control = 92 disease: lung cancer, gynaecological cancer treatment: platinum‐based chemotherapy baseline Hb: 9.4 g/dL duration: 12 weeks
Interventions	drug: epoetin beta dose: 36,000 IU/week target Hb: 12.0g/dL duration: 12 weeks
Outcomes	primary: proportion of patients receiving RBCTs and/or Hb<8.0 g/dL secondary: need for transfusions, changes in Hb, QoL
Notes	full‐text publication, study id: 15478
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	not reported
Allocation concealment (selection bias)	Low risk	central randomisation system

Gebbia 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 45, randomised ESA = 22, control = 23 disease: squamous cell carcinoma of the head an neck treatment: platinum‐containing chemotherapy baseline Hb: 12.1 g/dL
Interventions	drug: rhEpo dose: 10'000IU tiw Hb‐target: 12‐14 g/dL duration: NR
Outcomes	primary: NR secondary: transfusion, QoL, clinical outcome
Notes	full text publication, study number = 29327
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Gordon 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 220, randomised: ESA = 164, control = 56 disease: non‐myeloid hematological malignancies, breast, gastrointestinal, genitourinary, lung, gynaecological, other cancer (stage I‐IV) therapy: none baseline Hb: 10.2 g/dL
Interventions	drug: Darbepoetin alpha dose: 6.75 µg/kg sc Q4W Hb‐target: 12‐13 g/dL duration: 16 weeks
Outcomes	primary: Hb response secondary: transfusion, Hb change, QoL, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 65772)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation list will be centrally generated by Amgen
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Gordon 2008 IPD.

Methods	see Gordon 2008
Participants
Interventions
Outcomes
Notes

Goss 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 104, randomised: ESA = 52, control = 52 disease:SCLC (limited disease) treatment: radiochemotherapy baseline Hb: 13.5 g/dL
Interventions	drug: Epoetin alpha dose: 40000 IU sc weekly Hb‐target: 14‐16 g/dL, in 10/2002 reduced to 12‐14 g/dL duration: during chemotherapy and radiotherapy
Outcomes	disease progression free survival, tumour response, overall survival, local disease progression, Hb, transfusion, QoL
Notes	abstract publication, additional unpublished data obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 55703)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Goss 2005 FDA 2004.

Methods	see Goss 2005
Participants
Interventions
Outcomes
Notes

Goss 2005 IPD.

Methods	see Goss 2005
Participants
Interventions
Outcomes
Notes

Goss 2005 J&J 2004.

Methods	see Goss 2005
Participants
Interventions
Outcomes
Notes

Grote 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 224, randomised: ESA = 109, control = 115 disease: SCLC (limited and extensive disease) treatment: chemotherapy baseline Hb: 12.9 g/dL
Interventions	drug: Epoetin alpha dose: 150 IU/kg sc TIW Hb‐target: 14‐16 g/dL duration: NR, assumed to be 12 weeks (drug given during 3 x 3 weeks chemo plus 3 weeks)
Outcomes	primary: assess possible stimulatory effects of ESA on solid tumour growth, tumour response secondary: overall survival, Hb, transfusion, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 73807)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ description is unclear

Grote 2005 IPD.

Methods	see Grote 2005
Participants
Interventions
Outcomes
Notes

Grote 2005 J&J 2004.

Methods	see Grote 2005
Participants
Interventions
Outcomes
Notes

Gupta 2009.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 120, randomised: ESA = 60, control = 60 disease: cervical cancer treatment: platinum‐containing in all patients plus radiotherapy baseline Hb: 10.6 g/dL
Interventions	drug: Epoetin beta dose: 30'000 IU TIW Hb‐target: unclear duration: unclear
Outcomes	primary: Hb, energy level, QoL secondary: response rate, survival, toxicities, adverse events
Notes	full text publication, study number = 30057
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Hedenus 2002.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 66, randomised: ESA = 55, control = 11 disease: lymphoma: HD, NHL, MM treatment: NR, assumed to be chemotherapy without platinum baseline Hb: 9.6 g/dL
Interventions	drug: Darbepoetin alpha dosages: a: 1.0 µg/kg qw sc; b: 2.25 µg/kg qw sc; c:Darbepoetin 4.5 µg/kg qw sc Hb target: 13‐14 g/dL for women and 13‐15 g/dL for men duration: 12 weeks
Outcomes	primary: dose response relationship of darbepoetin in haemoglobin or haematopoietic response and overall survival secondary: transfusion
Notes	full text publication, study number = 32213
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Low risk	yes ‐ central computerized system

Hedenus 2002a.

Methods	see Hedenus 2002
Participants
Interventions	Darbepoetin 1.0 µg/kg qw sc
Outcomes
Notes

Hedenus 2002b.

Methods	see Hedenus 2002
Participants
Interventions	Darbepoetin 2.25 µg/kg qw sc
Outcomes
Notes

Hedenus 2002c.

Methods	see Hedenus 2002
Participants
Interventions	Darbepoetin 4.5 µg/kg qw sc
Outcomes
Notes

Hedenus 2003.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 349, randomised: ESA = 176, control = 173 disease: lymphoma: Hodgkin disease, NHL, MM, CLL, Waldenstrom´s disease treatment: NR, assumed to be chemotherapy without platinum Hb baseline: 9.5 g/dL
Interventions	drug: Darbepoetin alpha dose: 2.25 mg/kg qw sc Hb target: 13‐14 g/dL (women), 13‐15 g/dL (men) duration: 12 weeks
Outcomes	primary: Hb response secondary: transfusion, Hb change, QoL, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 63455)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ based on a schedule specified by Amgen before the start of the study
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Hedenus 2003 FDA 2007.

Methods	see Hedenus 2003
Participants
Interventions
Outcomes
Notes

Hedenus 2003 IPD.

Methods	see Hedenus 2003
Participants
Interventions
Outcomes
Notes	63455

Henke 1999.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 50, randomised: control = 11; ESAa = 19; ESAb = 14; ESAc = 6; ESAtotal = 39 disease: various solid tumours treatment: radiotherapy mean/median baseline Hb: 11.5 g/dL
Interventions	drug: Epoetin alpha or beta dose: ESAa: 150U/kg 3x/week i.v., ESAb: 300U/kg 3x/week i.v., ESAc: 150U/kg 3x/week s.c.; Hb target: 14 ‐ 16 g/dL (men) or 13 ‐ 15 g/dL (women) duration: 8 weeks
Outcomes	haematologic response, change in Hb values
Notes	full text publication, study number = 39895
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Unclear risk	Unclear

Henke 1999a.

Methods	see Henke 1999
Participants	n = 19
Interventions	Epoetin alpha or beta a: 150U/kg 3x/week i.v.
Outcomes
Notes

Henke 1999b.

Methods	see Henke 1999
Participants	n = 14
Interventions	Epoetin alpha or beta b: 300U/kg 3x/week i.v.;
Outcomes
Notes

Henke 1999c.

Methods	see Henke 1999
Participants	n= 6
Interventions	Epoetin alpha or beta c: 150U/kg 3x/week s.c.
Outcomes
Notes

Henke 2003.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 351, randomised: ESA = 180, control = 171 disease: advanced (stage III , IV) head and neck cancer treatment: radiotherapy baseline Hb: 11.8 g/dL
Interventions	drug: Epoetin beta dose: 300 IU/kg tiw sc Hb‐target: 12‐14 g/dL (women), 13‐15 g/dL (men) duration: 7‐9 weeks
Outcomes	primary: efficacy of radiotherapy, measured as local progression free survival secondary: survival, progression free survival, Hb, safety, tolerability
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 58106)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ each centre had numbered packages per stratum, once randomised the lowest number had to be assigned. There was a randomisation list only the statistics centre had access to. In addition, there were sealed envelopes for emergencies.
Allocation concealment (selection bias)	Low risk	yes ‐ coded drug packs of identical appearance

Henke 2003 IPD.

Methods	see Henke 2003
Participants
Interventions
Outcomes
Notes

Henke 2003 Roche 2004.

Methods	See Henke 2003
Participants
Interventions
Outcomes
Notes	Data presented by Roche at FDA/ODAC hearing in May 2004

Henry 1995.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 132, randomised: ESA = 67, control = 65 disease: any type of cancer except primary myeloid malignancy or acute leukaemia treatment: platinum‐containing chemotherapy baseline Hb: 9.5 g/dL
Interventions	drug: Epoetin alpha dose: 150 IU/kg sc TIW Hb‐target: Hct 38%‐40% duration: 12 weeks
Outcomes	primary: Hct, transfusion, haematologic response secondary: correction of anaemia, response, QoL, safety
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 70332)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ medication boxes were used, but without identical appearance

Henry 1995 IPD.

Methods	see Henry 1995
Participants
Interventions
Outcomes
Notes	study number = 70332

Henry 1995 J&J 2004.

Methods	see Henry 1995
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Hernandez 2009.

Methods	randomised controlled trial, placebo‐controlled
Participants	n = 391, randomised: ESA = 196, control = 195 disease: non‐myeloid haematological malignancies, breast, lung, gastrointestinal, genitourinary, gynaecological, other cancer (stage I‐IV) treatment: chemotherapy, 36% receiving platinum baseline Hb:10.1 g/dL
Interventions	drug: Darbepoetin alpha dose: 300 µg sc Q3W Hb‐target: 12‐13 g/dL duration: 15 weeks
Outcomes	primary: transfusion secondary: Hb target achieved, number of transfusions, safety, QoL
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 37476, Taylor 2005)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Hernandez 2009 IPD.

Methods	see Hernandez 2009 was Taylor 2005 in IPD meta‐analysis by Bohlius et al 2009
Participants
Interventions
Outcomes
Notes	study number = 37476

Hoskin 2009.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 300, randomised: ESA = 151, control = 149 disease: head and neck cancer (stage I‐IV) treatment: radiotherapy, no chemotherapy baseline Hb: 13.6 g/dL
Interventions	drug: Epoetin alpha dose: if Hb < 12.5 10000 IU sc TIW; if Hb > 12.5 4000 IU sc TIW Hb‐target: 14.5 to 15 g/dL duration: 12 weeks
Outcomes	primary: local disease free survival secondary: overall survival, QoL, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 81645)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ central randomisation schedule stratified by the study site was generated by the sponsor
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

Hoskin 2009 GBR‐7 FDA 04.

Methods	see Hoskin 2009
Participants
Interventions
Outcomes
Notes	unpublished study, data obtained from FDA report for the FDA/ODAC in hearing May 2004

Hoskin 2009 IPD.

Methods	see Hoskin 2009
Participants
Interventions
Outcomes
Notes	study number = 81645

Hoskin 2009 J&J 2004.

Methods	see Hoskin 2009
Participants
Interventions
Outcomes
Notes

Huddart 2002.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 95, randomised: ESA = 48, control = 47 disease: lung, gynaecological, genitourinary, other cancer treatment: platinum‐containing chemotherapy baseline Hb: not reported, eligibility criterion Hb < 10.5 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: Epoetin alpha dose: 10,000 IU tiw Hb‐target: 12‐14 g/dL duration: max 28 weeks
Outcomes	Hb response, reticulocyte, survival, QoL, safety
Notes	abstract, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 88443)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

Huddart 2002 IPD.

Methods	see Huddart 2002
Participants
Interventions
Outcomes
Notes	study number = 88443

Iconomou 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 122, randomised: ESA = 57, control = 55 disease: lung, breast, colorectal, ovarian, unknown primary, kidney, stomach, other cancer treatment: chemotherapy, platinum & non platinum baseline Hb: 10.1 g/dL
Interventions	drug and dose: NR, assumed Epoetin alpha 10,000 IU tiw sc Hb target: NR duration: 12 weeks
Outcomes	primary: QoL secondary: Hb, transfusions
Notes	full text publication, study number = 40799
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Low risk	Randomisation was performed by a telephone call to the registry of the department of medicine

Italian 1998.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 87, randomised: ESA = 44, control = 43 disease: Myelodysplastic Syndromes treatment: none mean/median baseline Hb: 8.2 g/dL;
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: not reported duration: 8 weeks, thereafter Epo for all the patients
Outcomes	haematologic response, change in haemoglobin values, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 46703
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer random number generator
Allocation concealment (selection bias)	Low risk	Yes ‐ central registration by telephone before treatment assignment

Katakami 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 207, ESA = 103, control = 104 disease: lung and gynaecologic cancer treatment: platinum‐containing chemotherapy baseline Hb: not reported, eligibility criterion Hb <= 11 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: darbepoetin alfa dose: 2.25 ug/kg once per week sc Hb target: 13.0 g/dL (amended to 12.0 g/dL) duration: 12 weeks
Outcomes	proportion of patients reaching transfusion trigger or receiving RBCT, haematologic endpoints, adverse events, survival
Notes	abstract publication (study id: 13567) additional study reporting exactly the same: Katsumata 2009
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear, not reported
Allocation concealment (selection bias)	Unclear risk	unclear, not reported

Kotasek 2002.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 161, randomised: ESA = 129, control = 32 disease: lung, breast, gastrointestinal, genitourinary, gynaecological, other cancer (stage I‐IV) treatment: chemotherapy baseline Hb: not reported, eligibility criterion Hb <= 11 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: Darbepoetin alpha dose: a: 9 µg/kg sc Q4W, b: 12 µg/kg sc Q4W, c: 15 µg/kg sc Q4W, d: 18 µg/kg sc Q4W Hb‐target: 13‐14 g/dL (women), 13‐15 g/dL (men) duration: 12 weeks
Outcomes	primary: safety secondary: determine effective dose, effect of ESA, QoL feasibility
Notes	additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 26117)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Kotasek 2002 IPD.

Methods	see Kotasek 2002
Participants
Interventions
Outcomes
Notes

Kotasek 2003.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 259, randomised: ESA = 208, control = 51 disease: breast, gynaecological, gastrointestinal, lung, genitourinary, other cancer treatment: chemotherapy, not reported whether with or without platinum, interpreted as some patients receiving platinum baseline Hb: 9.9 g/dL
Interventions	drug = Darbepoetin alpha dose = a: 4.5 µg/kg sc Q3W, b: 6.75 µg/kg sc Q3W, c: 9 µg/kg sc Q3W, d: 12 µg/kg sc Q3W, e: 13.5 µg/kg sc Q3W, f: 15 µg/kg sc Q3W Hb‐target = 13‐14 g/dL (women), 13‐15 g/dL (men) duration = 12 weeks
Outcomes	primary: safety secondary: determine effective dose, effect of ESA, QoL feasibility
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 35466)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Kotasek 2003 IPD.

Methods	see Kotasek 2003
Participants
Interventions
Outcomes
Notes	study number = 35466

Kotasek 2003a.

Methods	see Kotasek 2003
Participants	N = 32
Interventions	Darbepoetin 4.5 µg/kg Q3W sc
Outcomes
Notes

Kotasek 2003b.

Methods	see Kotasek 2003
Participants	N = 17
Interventions	Darbepoetin 6.75 µg/kg Q3W sc
Outcomes
Notes

Kotasek 2003c.

Methods	see Kotasek 2003
Participants	N = 46
Interventions	Darbepoetin 9 µg/kg Q3W sc
Outcomes
Notes

Kotasek 2003d.

Methods	see Kotasek 2003
Participants	N = 28
Interventions	Darbepoetin 12 µg/kg Q3W sc
Outcomes
Notes

Kotasek 2003e.

Methods	see Kotasek 2003
Participants	N = 35
Interventions	Darbepeotin 13.5 µg/kg Q3W sc
Outcomes
Notes

Kotasek 2003f.

Methods	see Kotasek 2003
Participants	N = 40
Interventions	Darbepoetin 15 µg/kg Q3W sc
Outcomes
Notes

Krzakowski 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 313, randomised: ESA a = 104, ESA b = 105, control = 104 disease: lung cancer, gastrointestinal tumour, breast cancer, genitourinary, haematological and other cancer treatment: platinum and non‐platinum containing chemotherapy baseline Hb: 9.4 g/dL
Interventions	drug: Epoetin delta dose: a: 150 IU/kg tiw, b: 300 IU/kg tiw Hb‐target: 12 ‐ 14 g/dL duration: 12 weeks
Outcomes	primary: Hb, RBC, transfusions secondary: haematocrit, FACT‐An, subgroup analysis for type of cancer/ chemotherapy
Notes	full text publication, study number = 49839
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Krzakowski 2008a.

Methods	see Krzakowski 2008
Participants	ESA a = 104
Interventions	dose: a: 150 IU/kg tiw
Outcomes
Notes

Krzakowski 2008b.

Methods	see Krzakowski 2008
Participants	ESA b = 105
Interventions	dose: b: 300 IU/kg tiw
Outcomes
Notes

Kunikane 2001.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 72, evaluated: 53. randomisation is only given for those: Epoetin a = 16, Epoetin b = 18, control = 19 disease: NSCLC treatment: platinum based chemotherapy baseline Hb: 12.3 g/dL
Interventions	drug: Epoetin beta dose: 1: 100U/kg 3x/week s.c.; 2: 200U/kg 3x/week s.c. Hb‐target: 14 g/dL in women, 16 g/dL in men duration: 6 weeks
Outcomes	change in Hb values, transfusion requirement, adverse events
Notes	full text publication, additional unpublished data obtained for first Cochrane Review, study number = 51164
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Low risk	yes ‐ central registration before treatment assignment

Kunikane 2001a.

Methods	see Kunikane 2001
Participants	Epoetin 1: n = 16
Interventions	Epoetin beta 1: 100U/kg 3x/week s.c.
Outcomes
Notes

Kunikane 2001b.

Methods	see Kunikane 2001
Participants
Interventions	Epoetin beta 2: 200U/kg 3x/week s.c.
Outcomes
Notes

Kurz 1997.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 35, randomised: ESA = 23, control = 13 disease: gynaecologic tumours treatment: platinum based chemotherapy baseline Hb: 9.9 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb‐target: no upper target reported duration: 12 weeks
Outcomes	haematologic response, change in Haemoglobin values, transfusion requirement, Quality of Life, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 54819
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ random permuted blocks and a corresponding randomisation list at the randomisation office
Allocation concealment (selection bias)	Low risk	Yes ‐ registration of the patients with an institution seperate from the centre where patients were recruited before treatment assignment

Leyland‐Jones 05 J&J 04.

Methods	see Leyland‐Jones 2005
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Leyland‐Jones 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 939, randomised: ESA = 469, control = 470 disease: metastatic breast cancer (stage IV, M1) treatment: chemotherapy baseline Hb: 12.5 g/dL
Interventions	drug: Epoetin alpha dose: 40,000 IU qw sc Hb‐target = 12‐14 g/dL duration: 52 weeks
Outcomes	primary: overall survival secondary: Hb, transfusion, tumour control, QoL, time to progression
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 17100)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Leyland‐Jones 2005 IPD.

Methods	see Leyland‐Jones 2005
Participants
Interventions
Outcomes
Notes	study number = 17100

Littlewood 2001.

Methods	randomised controlled trial, placebo‐controlled
Participants	N=375, randomised: ESA = 251, control = 124 disease: NHL, MM, HD, CLL, gastrointestinal, other cancer treatment: chemotherapy without platinum baseline Hb: 9.8 g/dL
Interventions	drug: Epoetin alpha dose: 150 IU/kg sc TIW Hb‐target: 12‐15 g/dL duration: 28 weeks
Outcomes	primary: transfusion secondary: haematologic response, Hb, Hct, reticulocytes, predictors for response, QoL, adverse events, after protocol amendment also survival
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 17123)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated randomisation schedule prepared by RWJPRI
Allocation concealment (selection bias)	Low risk	yes ‐ coded drug packs of identical appearance

Littlewood 2001 IPD.

Methods	see Littlewood 2001
Participants
Interventions
Outcomes
Notes	study number = 17123

Littlewood 2001 J&J 2004.

Methods	SeeLittlewood 2001
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Machtay 2007.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 148, randomised: ESA = 77, control = 71 disease: head and neck cancer (stage I‐IV) treatment: radiotherapy, advanced stages received in addition platinum based chemotherapy baseline Hb: 12.1 g/dL
Interventions	drug: Epoetin alpha dose: 40'000 IU sc weekly Hb‐target: 12.5‐14 g/dL (women), 13.5‐16 g/dL (men) duration: 8‐10 weeks
Outcomes	primary: local regional control tumour response secondary: overall survival, patterns of failure, local‐regional progression‐free survival, Hb, toxicity, QoL
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 87660), old publication was Machtay 2004
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Machtay 2007 IPD.

Methods	see Machtay 2007
Participants
Interventions
Outcomes
Notes	study number = 87660

Milroy 2003 IPD.

Methods	see Milroy 2011
Participants
Interventions
Outcomes
Notes

Milroy 2011.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 424, randomised: ESA = 214, control = 210 disease: NSCLC (stage IIIb or IV, advanced) treatment: platinum‐based chemotherapy baseline Hb: 12.7 g/dL
Interventions	drug: Epoetin alpha dose: if body weight > 45 kg 10000 IU sc TIW, if body weight < 45 kg 5000 IU sc TIW Hb‐target: 12.5‐14 g/dL (women), 13.5‐15 g/dL (men) duration = during chemotherapy
Outcomes	primary: QoL secondary: Hb, tumour response, survival, transfusion
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 67954)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

ML17616 2006.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 60 disease: MM, NHL and Cll treatment: chemotherapy, assumed to be without platinum because of hematological disease baseline Hb: not reported, eligibility criterion Hb <10 g/dL, categorized as Hb baseline < 10 g/dL
Interventions	drug: epoetin beta dose: 150 IU/kg TIW Hb‐target: NR duration: 12 weeks
Outcomes	primary: hematologic response secondary: Hb, transfusions, safety
Notes	clinical trial result information, study number = 99765
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

ML17620 2006.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 121 disease: solid tumours treatment: platinum‐based chemotherapy baseline Hb: not reported, unclear
Interventions	drug: epoetin beta dose: 150 IU/kg TIW Hb‐target: NR duration: 12 weeks
Outcomes	primary: hematologic response secondary: Hb, transfusions, Hct, safety, blood pressure
Notes	clinical trial result information, study number = 10373
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Moebus 2007.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 643, randomised: ESA = 324, control = 319 disease: breast cancer (high risk, stage II/IIIA; M0) treatment: chemotherapy without platinum baseline Hb: 12.6 g/dL
Interventions	drug: Epoetin alpha dose: 150 IU/kg sc TIW Hb‐target: 13‐14 g/dL duration: 18 weeks
Outcomes	primary: transfusion, Hb secondary: recurrence free survival, overall survival, relapse, QoL
Notes	abstract publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 22515)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer random‐number generator
Allocation concealment (selection bias)	Low risk	yes ‐ central registration of patients before treatment allocation

Moebus 2007 IPD.

Methods	see Moebus 2007
Participants
Interventions
Outcomes
Notes

Moebus 2007 J&J 2007.

Methods	see Moebus 2007
Participants
Interventions
Outcomes
Notes

Mystakidou 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 100, randomised: ESA = 50, control = 50 disease: pancreatic, genital, colon and lung cancer treatment: none baseline Hb: 10.01 g/dL; for ESA group 9.87g/dL for control group 10.15, categorized as 10‐12 g/dL
Interventions	drug: Epoetin alfa dose: 40'000 weekly Hb‐target: 15 g/dL duration: 24 weeks
Outcomes	primary: Hb secondary: QoL, safety
Notes	full text publication, was excluded in the IPD Review, study number = 61315
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

O'Shaughnessy 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 100, randomised: ESA = 51, control = 49 disease: breast cancer, stages I‐IIIB treatment: chemotherapy baseline Hb: 12.9 g/dL
Interventions	drug: Epoetin alpha dose: 40,000 IU qw sc Hb‐target: 13‐15 g/dL duration: 12 weeks
Outcomes	primary: cognitive function, fatigue secondary: QoL
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 40730)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer generated
Allocation concealment (selection bias)	Low risk	yes ‐ coded drug packs of identical appearance

O'Shaugnessy 2005 IPD.

Methods	see O'Shaughnessy 2005
Participants
Interventions
Outcomes
Notes	study number = 40730

OBE/EPO‐INT‐03 IPD.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 72, randomised: ESA = 35, control = 37 disease: MM treatment: chemotherapy, assumed to be without platinum because of hematological disease baseline Hb: not reported, unclear
Interventions	drug: Epoetin alpha dose: 40'000 IU sc weekly Hb‐target: 12‐13 g/dL duration: during chemotherapy
Outcomes	primary: Hb change secondary: QoL, Hb response, transfusion, safety
Notes	unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 92503)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Oberhoff 1998.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 218, randomised: ESA = 114, control = 104 disease: ovarian, breast, lung, genitourinary, gastrointestinal, other cancer treatment: platinum containing chemotherapy baseline Hb: ESA arm 9.6 g/dL, control 10.3 g/dL, categorised as < 10 g/dL
Interventions	drug: Epoetin beta dose: 5000U daily s.c. Hb‐target: 14 g/dL duration: 12 weeks
Outcomes	primary: transfusion secondary: haematologic response, Hb response, safety
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 45434)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Oberhoff 1998 IPD.

Methods	see Oberhoff 1998
Participants
Interventions
Outcomes
Notes	study number = 45434

Osterborg 1996.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 144, randomised: ESA 1 = 47, ESA 2 = 48, control = 49 disease: MM, NHL, chronic lymphocytic lymphoma treatment: chemotherapy, non‐platinum containing baseline Hb: 8.8 g/dL
Interventions	drug: Epoetin beta dose: a: 10000 IU sc 7x/week, b: 2.000U daily s.c.; increased to 5.000U and 10.000U daily if no response Hb‐target: 12‐13 g/dL (women), 13‐14 g/dL (men) duration: 24 weeks
Outcomes	primary: transfusion secondary: safety, Hb, haematologic response
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 43680)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Osterborg 1996 IPD.

Methods	see Osterborg 1996
Participants
Interventions
Outcomes
Notes	study number = 43680

Osterborg 1996a.

Methods	see Osterborg 1996
Participants
Interventions	dose a: 10'000 daily sc
Outcomes
Notes

Osterborg 1996b.

Methods	see Osterborg 1996
Participants
Interventions	dose b: 2.000U daily s.c.; increased to 5.000U and 10.000U daily if no response
Outcomes
Notes

Osterborg 2002.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 349, randomised: ESA = 173, control = 176 disease: MM, NHL, CLL; treatment: chemotherapy, assumed without platinum because of hematological disease baseline Hb: 9.3 10g/dL
Interventions	drug: Epoetin beta dose: 150U/kg 3x/week s.c. Hb‐target = 13‐14 g/dL duration: 16 weeks
Outcomes	primary: transfusion free survival secondary: haematologic response, Hb change, time to response, number of blood transfusions, QoL, safety
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 77914)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation program
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Osterborg 2002 IPD.

Methods	see Osterborg 2002
Participants
Interventions
Outcomes
Notes	study number = 77914

Overgaard 2009.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 515, randomised: ESA = 255, control = 260 disease: head and neck cancer treatment: radiotherapy baseline Hb: approximately 13 g/dL
Interventions	drug: darbepoetin dose: 150 mg sc weekly Hb target: > 15.5 g/dL duration: 8 to 10 weeks
Outcomes	OS, DS, tumour control, adverse events
Notes	abstract publication, study number = 62913
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	not reported
Allocation concealment (selection bias)	Low risk	central allocation method

P‐174 IPD.

Methods	see P‐174 J&J 2004
Participants
Interventions
Outcomes
Notes

P‐174 J&J 2004.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 45, randomised: ESA = 33, control = 12 disease: CLL (any stage) treatment: NR 'other' baseline Hb category: not reported, unclear
Interventions	drug: Epoetin alpha dose: 150 IU/kg tiw sc Hct‐target: 38% to 40% duration 12 weeks
Outcomes	primary: Hct secondary: Hb, transfusion, QoL, safety
Notes	data presented by J&J at FDA/ODAC hearing in May 2004, additional unpublished data were obtained for an individual patient data meta‐analysis (Bohlius et al, 2009 study number = 60584)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

Pirker 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 600, randomised: ESA = 299, control = 301 disease: SCLC (untreated, extensive stage) treatment: platinum‐containing chemotherapy baseline Hb: 11.9 g/dL, ESA arm 12.03 g/dL, control 11.86 g/dL, categorised as 10‐12 g/dL
Interventions	drug: Darbepoetin alpha dose: 300 µg sc weekly for weeks 1‐4 then 300 µg Q3W starting week 5 onwards Hb‐target: 13‐14 g/dL duration: 19 weeks
Outcomes	primary: Hb change, survival secondary: QoL, progression‐free‐survival, tumour response, time to progression, transfusion
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 89335)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Pirker 2008 IPD.

Methods	see Pirker 2008
Participants
Interventions
Outcomes
Notes

Pronzato 2010.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 223, randomised ESA=110, control =113 disease: breast cancer (stage I‐IV) treatment: chemotherapy baseline Hb: 10.7 g/dL
Interventions	drug: epoetin alpha dose: if body weight >45kg 10,000 IU sc TIW, if body weight <45kg 5,000 IU sc TIW Hb target: 12‐14 g/dL duration: categorized: >20 weeks
Outcomes	Primary: QoL (anaemia) Secondary: hematological response, other QoL, tumour response, OS, number of patients transfuse
Notes	Full‐text publication, unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 22233)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Pronzato 2010 IPD.

Methods	see Pronzato 2010
Participants
Interventions
Outcomes
Notes

Quirt 1996.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 56, randomised: ESA = 28, control = 28 disease: lung, gynaecological, hematological malignancies, other cancer treatment: chemotherapy baseline Hb: 10.8 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c Hb‐target: 12.5‐14 g/dL duration: 16 weeks
Outcomes	primary: transfusion, Hb change secondary: QoL, costs from societal perspective, tumour response
Notes	abstract publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 80214)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

Quirt 1996 IPD.

Methods	see Quirt 1996
Participants
Interventions
Outcomes
Notes	study number = 80214

Ray‐Coquard 2009.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 218, randomised: ESA = 110, control = 108 disease: breast, sarcoma, lung, ovarian, other solid cancer and hematological malignancies treatment: chemotherapy (IPD) full text: NR baseline Hb: 10.0 g/dL, categorised as 10‐12 g/dL
Interventions	drug: Epoetin alpha dose: if body weight < 45 kg 10000 IU sc 2x/week, if body weight 45 kg to < 89 kg 10000 IU sc TIW, if body weight > 89 kg 10000 IU sc 4x/week Hb‐target: 12‐14 g/dL planned ESA duration: 12 weeks
Outcomes	primary: transfusion dependent anaemia secondary: QoL, Hb response predictors, Hb, toxicity, survival, costs
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 37491)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes
Allocation concealment (selection bias)	Low risk	yes

Ray‐Coquard 2009 IPD.

Methods	see Ray‐Coquard 2009
Participants
Interventions
Outcomes
Notes	abstract publication, IPD data

Razzouk 2004.

Methods	see Razzouk 2006
Participants
Interventions
Outcomes
Notes

Razzouk 2006.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 224, randomised ESA: 112, Control: 112 disease: solid tumours, HD, NHL, ALL treatment: chemotherapy baseline Hb: 9.7 g/dL
Interventions	drug: epoetin alpha dose: 600 IU/kg iv weekly Hb target: 13‐15 g/dL (age >12 years), 13‐14 g/dL (age<12 years) duration: 16 weeks
Outcomes	primary: QoL secondary: Hb, transfusion
Notes	Full‐text publication, additional unpublished data were obtained for an Individual Patient Data meta‐analysis study (Bohlius et al 2009). Study number: 80515
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes‐ computer generated
Allocation concealment (selection bias)	Low risk	yes‐ central randomisation and coded drug packages of identical appearance

Rose 1994.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 221, ESA = 142, control = 79 disease: CLL (stage III, IV) treatment: chemo‐ and radiotherapy, without platinum baseline Hb: 9.2 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: Hct 38%‐40% duration: 12 weeks
Outcomes	primary: Hct, haematologic response secondary: transfusion, safety, QoL
Notes	Abstract publication, additional unpublished data were obtained for the this Cochrane review and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 98358)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Unclear risk	unclear ‐ no description

Rose 1994 IPD.

Methods	see Rose 1994
Participants
Interventions
Outcomes
Notes	study number = 98358

Rose 1994 J&J 2004.

Methods	see Rose 1994
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Rosen 2003.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 90, randomised: ESA = 47, control = 43 disease: head and neck cancer treatment: radiation therapy and non platinum‐containing chemotherapy baseline Hb: 12.2 g/dL at baseline
Interventions	drug: Epo dose: 40'000 IU per week Hb‐target: NR duration: 14 weeks
Outcomes	primary: response rate, toxicity, disease free and overall survival secondary: response to Epo treatment
Notes	full text publication, was excluded in the IPD review: n = 90, study number = 72003
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Rosenzweig 2004.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 27, randomised: ESA = 14, control = 13 disease: metastatic breast cancer treatment: less than 50% of participants received chemotherapy, some received hormones, categorised as other baseline Hb: not reported, eligibility criterion Hb < 12 g/dL, categorised as Hb 10‐12 g/dL
Interventions	drug: Epoetin alpha dose: 40,000 IU qw sc Hb target: NR duration: 12 weeks
Outcomes	primary: fatigue, QoL
Notes	full text publication, study number = 76065
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Low risk	Yes ‐ using sequential, opaque, sealed envelopes with the order unknown to the investigators

Savonije 2005.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 315, randomised: ESA = 211, control = 104 disease: NSCLC, gastrointestinal, gynaecological, colorectal, SCLC, other cancer treatment: platinum based chemotherapy baseline Hb: 10.7 g/dL
Interventions	drug: Epoetin alpha dose: 10'000 IU sc TIW sc Hb‐target: 13‐14 g/dL duration: 14 weeks
Outcomes	primary: transfusion secondary: Hb, tumour response, QoL, survival
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 70724)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ randomisation centre generates a list of subject numbers and randomly allocate numbers to the two treatment groups using a block size of six
Allocation concealment (selection bias)	Low risk	Yes ‐ central randomisation

Savonije 2005 IPD.

Methods	see Savonije 2005
Participants
Interventions
Outcomes
Notes	study number = 70724

Silvestris 1995.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 54, randomised: ESA = 30, control = 24 disease: MM treatment: chemotherapy without platinum baseline Hb: <=8 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. Hb target: no target defined duration: 24 weeks
Outcomes	haematologic response, adverse events
Notes	full text publication, study number = 76441
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Smith 2003.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 86, randomised: ESA = 64, control = 22 disease: genitourinary, breast, gastrointestinal, lymphoma: myeloma, CLL, NHL treatment: none baseline Hb: 9.995; <10 g/dL for two groups and 10‐12 g/dL for the other two, categorized as <10 g/dL
Interventions	drug: Darbepoetin alpha dose: see below Hb target: 13‐14 g/L (women), 13‐15 (men) g/dL duration: 12 weeks
Outcomes	primary: haematopoietic response secondary: time to response, Hb response, Hb change, transfusions, serum darbepoetin conc. in a subset of patients
Notes	full text publication, study number = 76561
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Smith 2003 Amgen 2007.

Methods	see Smith 2003
Participants
Interventions
Outcomes
Notes

Smith 2003a.

Methods	See Smith 2003
Participants
Interventions	dose a: 6.75 µg/kg Q3W sc
Outcomes
Notes

Smith 2003b.

Methods	See Smith 2003
Participants
Interventions	dose b: 6.75 µg/kg Q4W sc
Outcomes
Notes

Smith 2003c.

Methods	See Smith 2003
Participants
Interventions	dose c: 10 µg/kg Q4W sc
Outcomes
Notes

Smith 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 989, randomised: ESA = 517, control = 472 disease: lung, hematological malignancies, breast, gastrointestinal, genitourinary, other cancer (stage III‐IV) treatment: none baseline Hb: 9.5 g/dL
Interventions	drug: Darbepoetin alpha dose: 6.75 µg/kg sc Q4W Hb‐target: 12‐13 g/dL duration: 16 weeks
Outcomes	primary: transfusion secondary: Hb, QoL, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 81215)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Low risk	Yes, adequate

Smith 2008 IPD.

Methods	see Smith 2008
Participants
Interventions
Outcomes
Notes

Strauss 2008.

Methods	randomised controlled trial, not placebo‐controlled
Participants	n = 74, randomised: ESA = 34, control = 40 disease: cervical cancer (stage IIB‐IVA) treatment: radio‐ and platinum‐containing chemotherapy baseline Hb: 11.5 g/dL
Interventions	drug: Epoetin beta dose: 150 IU/kg sc TIW Hb‐target: 14‐15 g/dL duration: 12 weeks
Outcomes	primary: tumour control failures secondary: progression‐free survival, overall response rate, relapses/metastases, overall survival, Hb change, QoL, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 70404)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ patient randomisation number were generated without reporting the method
Allocation concealment (selection bias)	Unclear risk	unclear ‐ patient randomisation numbers were to be allocated sequentially in the order in which the patients are enrolled

Strauss 2008 IPD.

Methods	see Strauss 2008
Participants
Interventions
Outcomes
Notes

Suzuki 2008.

Methods	randomised controlled trial, placebo‐controlled
Participants	N=123: ESA: 81, control=42 disease: lymphoma, breast cancer treatment: chemotherapy baseline Hb: not reported, eligibility criterion Hb <= 11 g/dL, after amendment <= 10 g/dL, categorized as Hb 10‐12 g/dL
Interventions	drug: darbepoetin alpha dose: a) 4.5 ug/kg sc Q3W or b) 6.75 ug/kg sc Q3W Hb target: <=13.0 g/dL, amended to <=12.0 g/dL duration: 12 weeks
Outcomes	Hb response, safety, QoL, survival
Notes	abstract publication, study id: 14688
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear, not reported
Allocation concealment (selection bias)	Unclear risk	unclear, not reported

Suzuki 2008a.

Methods	see Suzuki 2008
Participants	darbepoetin: 40, placebo: 42
Interventions	darbepoetin 4.5 ug/kg sc Q3W vs placebo
Outcomes
Notes

Suzuki 2008b.

Methods	see Suzuki 2008
Participants	darbepoetin: 41, placebo: 42
Interventions	darbepoetin 6.75 ug/kg sc Q3W vs placebo
Outcomes
Notes

Sweeney 1998.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 48, randomised: ESA = 24, control = 24 disease: breast, lung, prostate and cervix cancer treatment: chemotherapy for 5 patients, radiotherapy for probably all of the patients baseline Hb: ESA arm 12.07, control: 10.72 g/dL, categorized as 10‐12 g/dL
Interventions	drug: Epoetin alfa dose: 200 IU/kg/d Hb target: 14 g/dL for women and 15 g/dL for men duration: 7 weeks
Outcomes	Hb, total white blood cell count and platelets, QoL
Notes	full text publication, excluded for IPD‐Review, study number = 77932
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	computer‐generated random numbers in blocks of 10
Allocation concealment (selection bias)	Unclear risk	not reported, unclear

Ten Bokkel 1998.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 122, randomised: ESA = 88, control = 34 disease: ovarian carcinoma (stage II‐IV) treatment: platinum‐based chemotherapy baseline Hb: 11.6 g/dL
Interventions	drug: Epoetin beta dose: a: 150 IU/kg sc TIW, b: 300 IU/kg sc TIW Hb‐target: 14‐15 g/dL duration = during chemotherapy, 24 weeks
Outcomes	primary: transfusion secondary: Hb, reticulocytes, Hct, safety, tumour response, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 47852)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Ten Bokkel 1998a.

Methods	see Ten Bokkel 1998
Participants	Epoetin a: N = 45
Interventions	dose a: 150U/kg 3x/week s.c.
Outcomes
Notes

Ten Bokkel 1998b.

Methods	see Ten Bokkel 1998
Participants	Epoetin 2: N = 42
Interventions	dose b: 300U/kg 3x/week s.c.
Outcomes
Notes

Ten Bokkel1998 IPD.

Methods	see Ten Bokkel 1998
Participants
Interventions
Outcomes
Notes	study number = 47852

Thatcher 1999.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 130, randomised: ESA = 86, control = 44 disease: SCLC treatment: platinum based chemotherapy baseline Hb: 13.4 g/dL
Interventions	drug: Epoetin alpha dose: ESA a: 150 IU/kg sc TIW; ESAb: 300 IU/kg sc TIW Hb‐target: 13‐15 g/dL duration: 26 weeks
Outcomes	change in Hb values, transfusion requirement, QoL, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001) and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 65529)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ upon study entry each patient was assigned a sequential identification number which had been randomly assigned to chemotherapy with or without ESA, blocks of 6, each investigator had to treat at least 6 patients, but preferably 12 patients.
Allocation concealment (selection bias)	Unclear risk	unclear ‐ see randomisation

Thatcher 1999 IPD.

Methods	see Thatcher 1999
Participants
Interventions
Outcomes
Notes	study number = 65529

Thatcher 1999a.

Methods	See Thatcher 1999
Participants	Epoetin a: n = 42
Interventions	Epoetin alpha a: 150U/kg 3x/week s.c.
Outcomes
Notes

Thatcher 1999b.

Methods	See Thatcher 1999
Participants	Epoetin b: N = 44
Interventions	Epoetin alpha b: 300U/kg 3x/week s.c.
Outcomes
Notes

Thomas 2002.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 130, randomised: ESA = 65, control = 65 disease: breast, gastrointestinal, gynaecological, other cancer treatment: chemotherapy baseline Hb: 10.6 g/dL
Interventions	drug: Epoetin alpha dose: if body weight > 45 kg 10000 IU sc TIW, if body weight < 45 kg 5000 IU sc TIW Hb‐target: 12‐14 g/dL duration: 12 weeks
Outcomes	Hb, QoL, transfusions
Notes	abstract publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 84090)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Thomas 2002 IPD.

Methods	see Thomas 2002
Participants
Interventions
Outcomes
Notes	study number = 84090

Thomas 2008.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 114 (from IPD), full‐text: accrued: 114, 5 found subsequently not eligible randomised: ESA = 57, control = 52; planned were 460, vs IPD, vs 2006 disease: cervical cancer (stage IIB ‐ IV A, M0) treatment: platinum‐based chemotherapy plus radiotherapy baseline Hb: 10.7 g/dL
Interventions	drug: Epoetin alpha dose: 40000 IU sc weekly Hb‐target: 13‐14 g/dL duration: 8 weeks max, categorised as 6‐9 weeks
Outcomes	primary: progression‐free survival secondary: OS, local control, distant recurrences, thromboembolic events
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 21481)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description "eligible patients were randomised..."
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Thomas 2008 IPD.

Methods	see Thomas 2008
Participants
Interventions
Outcomes
Notes

Thomas 2008 J&J 2004.

Methods	see Thomas 2008
Participants
Interventions
Outcomes
Notes

Thompson 2000.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 66, randomised: ESA = 45, control = 21 disease: Myelodysplastic Syndromes treatment: no chemotherapy, GM‐CSF 0.3‐5.0 mg/kg daily in both groups mean/median baseline Hb: 9 g/dL
Interventions	drug: Epoetin alpha dose: 150U/kg 3x/week s.c. + Hb target: 12 ‐ 13 g/dL duration: 12 weeks
Outcomes	haematologic response, change in Hb values, transfusion requirement, adverse events
Notes	full text publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 82687
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer random‐number generator
Allocation concealment (selection bias)	Low risk	Yes ‐ adequate

Throuvalas 2000.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 55, randomised: ESA = 28, control = 27 disease: cervix and bladder carcinoma treatment: platinum‐based radiochemotherapy baseline Hb: 11.3 g/dL
Interventions	drug: Epoetin (?) dose: 10,000U 5x/week s.c. Hb target: NR duration: 6 weeks
Outcomes	change in Hb values, transfusion requirement, tumour response
Notes	abstract publication, additional unpublished data were obtained for the first Cochrane review (1985‐2001), study number = 83700
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ random‐number generator
Allocation concealment (selection bias)	Low risk	yes ‐ central allocation

Tjulandin 2010.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 223, randomised: Epo theta = 76, Epo beta = 73, control = 74 disease: ovarian cancer, gastric cancer, breast cancer, lung cancer. other solid cancers treatment: platinum‐based chemotherapy baseline Hb: 9.5 g/dL
Interventions	drug a): Epoetin theta, dose: 20'000 IU weekly drug b): Epoetin beta, dose: 150 IU/kg sc TIW Hb‐target: 13 g/dL duration: 12 weeks
Outcomes	primary: haematological response secondary: partial Hb response, RBCTs, number of bloods units transfused, safety, QoL
Notes	full‐text publication, study number = 19632
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	computer‐generated allocation schedule
Allocation concealment (selection bias)	Unclear risk	not reported

Tjulandin 2010a.

Methods	see Tjulandin 2010
Participants
Interventions	Epoetin theta, dose: 20'000 IU weekly
Outcomes
Notes

Tjulandin 2010b.

Methods	see Tjulandin 2010
Participants
Interventions	Epoetin beta, dose: 150 IU/kg sc TIW
Outcomes
Notes

Tjulandin 2011.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 186, randomised: ESA = 95, control = 91 disease: hematological, breast and gastric cancer treatment: chemotherapy without platinum baseline Hb: 9.2 g/dL
Interventions	drug: Epoetin theta dose: 20'000 IU weekly Hb‐target: 13 g/dL duration: 12 weeks
Outcomes	primary: haematological response secondary: partial Hb response, RBCTs, number of bloods units transfused, safety, QoL
Notes	full‐text publication, study number = 18036
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	computer‐generated
Allocation concealment (selection bias)	Unclear risk	not reported

Tsuboi 2009.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 122, randomised: ESA = 63, control = 59 disease: lung cancer, malignant lymphoma (HL and NHL) treatment: chemotherapy, both platinum and non‐platinum containing, no numbers given. baseline Hb: 10.2 g/dL
Interventions	drug: Epoetin beta dose: 36'000 IU sc weekly Hb target: >= 14 g/dL duration: 8 weeks
Outcomes	primary: Hb change secondary: hematological response, transfusions, Hb, QoL, (survival, cave: retrospective)
Notes	full text publication, abstract Watanabe 2006 was excluded for the IPD‐Review, study number = 92759
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation system

Untch 2008 IPD.

Methods	see Untch 2011_1
Participants
Interventions
Outcomes
Notes

Untch 2011_1.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 733, randomised: ESA = 356, control = 377 disease: breast cancer (M0) treatment: non platinum‐containing chemotherapy baseline Hb: 13.6 g/dL
Interventions	drug: Darbepoetin alpha dose: 4.5 µg/kg sc Q2W Hb‐target: 12.5‐13 g/dL duration: during chemotherapy, approximately > 20 weeks
Outcomes	primary: relapse free survival time, OS secondary: tumour control, safety and tolerability, transfusion, Hb level, QoL
Notes	two full text publications, in addition unpublished data were obtained for the individual patient data meta‐analysis study (Bohlius et al 2009, study number = 66960)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ no description
Allocation concealment (selection bias)	Unclear risk	unclear ‐ description is unclear

Untch 2011_2.

Methods	see Untch 2011_1
Participants
Interventions
Outcomes
Notes

Vansteenkiste 02 FDA 2004.

Methods	see Vansteenkiste 2002
Participants
Interventions
Outcomes
Notes

Vansteenkiste 2002.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 320, randomised: ESA = 159, control = 161 disease: SCLC (limited and extensive), and NSCLC (stage I‐IV) treatment: platinum based chemotherapy baseline Hb: 10.1 g/dL
Interventions	drug: Darbepoetin alpha dose: 2.25 mg/kg sc weekly Hb‐target: 13‐14 g/dL (women), 13‐15 g/dL (men) duration: 12 weeks
Outcomes	primary: transfusion secondary: Hb response, Hb, transfusion timing and quantity, QoL
Notes	full text publication, additional unpublished data were obtained for and an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 49684)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ based on a schedule specified by Amgen before the start of the study
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Vansteenkiste 2002 IPD.

Methods	see Vansteenkiste 2002
Participants
Interventions
Outcomes
Notes	study number = 49684

Welch 1995.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 30, randomised: ESA = 15, control = 15 disease: ovarian carcinoma treatment: platinum‐containing chemotherapy mean/median baseline Hb: 12.9 g/dL
Interventions	drug: Epoetin alpha dose: 300U/kg 3x/week s.c. Hb ‐ target: 12‐15 g/dL duration: 24 weeks
Outcomes	change in Hb values, transfusion requirement, adverse events
Notes	full text publication, study number = 97952
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear
Allocation concealment (selection bias)	Unclear risk	unclear

Wilkinson 2006.

Methods	randomised controlled trial, not placebo‐controlled
Participants	N = 182, randomised: ESA = 121; control = 61 disease: ovarian cancer (stage I‐IV) treatment: chemotherapy baseline Hb: 10.7 g/dL
Interventions	drug: Epoetin alpha dose: if body weight > 45 kg 10000 IU sc TIW, if < 45 kg 5000 IU sc TIW Hb‐target: 12‐14 g/dL duration: max. 28 weeks
Outcomes	primary: Hb response secondary: QoL, transfusion, tumour response
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 75688)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	unclear ‐ a prospective randomisation procedure will be employed
Allocation concealment (selection bias)	Unclear risk	unclear ‐ assigned envelopes, but it is unclear whether they were opaque and sequentially numbered

Wilkinson 2006 IPD.

Methods	seeWilkinson 2006
Participants
Interventions
Outcomes
Notes

Winquist 2009.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 56, randomised ESA: 26, control = 30 disease: prostate cancer treatment: unclear Baseline Hb: 10.4 g/dL
Interventions	drug: epoetin alpha dose: 40,000 IU sc 3 times /week Hb target: 14.0 g/dL duration: 16 weeks
Outcomes	primary: QoL secondary: Hb level, RBCTs, adverse events, survival
Notes	letter publication, study number 13321
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	not reported, unclear
Allocation concealment (selection bias)	Low risk	central telephone

Witzig 2005.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 344, randomised: ESA = 174, control = 170 disease: lung, breast, other cancer (active incurable advanced stage) treatment: chemotherapy, platinum & non platinum Hb category: 9.5 g/dL
Interventions	drug: Epoetin alpha dose: 40000 IU sc weekly Hb‐target: 13‐15 g/dL planned ESA duration: 16 weeks
Outcomes	primary: transfusions secondary: Hb change, Hb over time, predictors for response, incidence of nephrotoxicity, OS, tumour response, QoL
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 36512)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation and coded drug packs of identical appearance

Witzig 2005 IPD.

Methods	see Witzig 2005
Participants
Interventions
Outcomes
Notes

Witzig 2005 J&J 2004.

Methods	see Witzig 2005
Participants
Interventions
Outcomes
Notes	Data presented by J&J at FDA/ODAC hearing in May 2004

Wright 2007.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 70, randomised: ESA = 33, control = 37 disease: NSCLC (advanced stage IIIA, B and IV, recurrent disease) treatment: no anticancer therapy baseline Hb: 10.3 g/dL
Interventions	drug: Epoetin alpha dose: 40'000 IU sc weekly Hb‐target: 12‐14 g/dL duration = 12 weeks
Outcomes	primary: QoL secondary: Hb, Hct, transfusion, safety
Notes	full text publication, additional unpublished data were obtained for an individual patient data meta‐analysis study (Bohlius et al 2009, study number = 53572)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	yes ‐ computer‐generated
Allocation concealment (selection bias)	Low risk	yes ‐ central randomisation

Wright 2007 IPD.

Methods	see Wright 2007
Participants
Interventions
Outcomes
Notes

Wurnig 1996.

Methods	randomised controlled trial, placebo‐controlled
Participants	N = 30, randomised: ESA = 16, control = 14 disease: Ewing's or osteosarcoma treatment: chemotherapy, some platinum‐based mean/median baseline Hb: 10.8 g/dL
Interventions	drug: Epoetin beta dose: 600U/kg 2x/week i.v. Hb‐target: 11‐13.5 g/dL duration: 20 weeks
Outcomes	change in Hb values, transfusion requirement, adverse events
Notes	full text publication, no IPD data, study number = 97958
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Yes ‐ computer‐generated randomisation code
Allocation concealment (selection bias)	Unclear risk	Unclear

CLL: Chronic lymphatic leukaemia 
 D: day 
 ESA: erythropoiesis stimulating agent 
 FDA: Food and Drug Administration, USA 
 G‐CSF: Granulocyte colony‐stimulating factor 
 Hb: Haemoglobin 
 Hct: haematocrit 
 HD: Hodgkin's Disease 
 IPD: individual patient data 
 i.v.: intravenous 
 J&J: Johnson & Johnson Pharmaceuticals Ltd. 
 MM: multiple myeloma 
 NHL: non‐Hodgkin's lymphoma 
 NSCLC: non‐small cell lung cancer 
 NR: not reported 
 ODAC: Oncology Drug Advisory Committee 
 OS: overall survival 
 QoL: Quality of Life 
 qw: once per week 
 Q3W: once every three weeks 
 Q4W: once every four weeks 
 RBC: red blood cell 
 s.c.: subcutaneous 
 SCLC: small cell lung cancer 
 tiw: three times per week 
 TR: Tumour response 
 wk: week 
 wks: weeks

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aagaard 2010	nrct
Aapro 2009	nrct
Abdelrazik 2007	ineligible patient characteristics: ALL
Abraham 2011	nrct
Adamson 2009	no RCT
Alexopoulos 2004	randomised comparison of epoetin alfa 10,000 IU tiw versus darbepoetin alfa 150 mcg qw, n = 50
Anonymous 2007	no RCT
Anthony 2011	all arms ESAs
Arcasoy 2010	nrct
Arslan 2004	randomised comparison of different usage strategies
Auerbach 2004	randomised comparison of different iron applications
Australian 2010	nrct
Aziz 2001	treatment allocation not concealed IPD: too small for inclusion
Barosi 1998	nrct
Barosi 2011	nrct
Beggs 2003	randomised comparison of 40,000 IU epoetin alfa versus placebo, study too small for inclusion: n = 21
Bell 2008	no RCT
Bessho 1997	ineligible patient characteristics: only patients with aplastic anaemia included, N = 131
Bindi 2004	unclear whether this a randomised controlled trial, authors have been unsuccesfully contacted IPD: too small for inclusion
Blayney 2003	study stated to be randomised, experimental arms received darbepoetin alfa, treatment in the control group not documented, authors were contacted for clarification without success, n = 1173
Boccia 2007	no RCT
Borg 2008	no RCT
Borget 2008	no RCT
Bowen 2004	G‐CSF only in one arm
Brower 2008	nrct
Buchler 2011	nrct
Buyukpamukcu 2002	follow‐up study to Varan 1999, does not satisfy allocation concealment requirement
Cabanillas 2012	acute leukaemia
Candelaria 2005	no ESA was given
Canon 2006	compared different ESA dosages
Canon 2011	retrospective analysis of dose‐fidning study
Caravita 2009	nrct
Casadevall 2004	randomised controlled study in patients with MDS, comparing erythropoietin PLUS G‐CSF versus supportive care without erythropoietin and without G‐CSF
Cazzola 2003	randomised comparison of once weekly 30,000 IU epoetin beta versus three times weekly 10,000 IU epoetin beta
Chan 1995	very small trial, only 10 evaluable patients per study arm
Cheng 2009	yixuesheng capsule with ESA vs ESA alone
Christodoulakis 2005	ESAs were given in context with surgery
Coiffier 2006	nrct
Coleman 2009	not randomised for epo
Crawford 1997	randomised‐controlled trial, placebo controlled, double‐blind, included 25 patients with SCLC, as no sufficient data were available the study was excluded
Crawford 2003	randomised comparison of epootin alfa 40,000 IU per week versus control; patients in the control group received epoetin alfa when the Hb level decreased below 10g/dL, n = 216
Crawford 2007	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Cremieux 2003	nrct
Dahl 2008	nrct
Daneryd 1998	epoetin only given to the anaemic patients in the treatment arm
Dannemann 2004	compared different ESA dosages
Demetri 1998	community based, non randomised study including 2,370 patients with nonmyeloid malignancies
Dicato 2011	nrct
Dronca 2008	nrct
Dusenbery 1994	mixed concurrent and historical controls
Elsaid 2001	n too small
Fagnoni 2005	retrospective study design
Ferrero 2009	no RCT
Franchi 2008	no RCT, dose comparing
Freeman 2006	too small for inclusion, n = 14
Fujisaka 2004	single arm study, no RCT
Gabrilove 2001	community based, nonrandomised study including 3,012 participants with nonmyeloid malignancies
Garton 1995	very small trial, only 10 evaluable patrticipants per study arm
Gascon 2008	nrct
Gebbia 1992	n = 19, too small for inclusion
Glaser 1999	no RCT, 37 participants with advanced oral squamous carcinoma receiving neoadjuvant radiochemotherapy
Glaser 2001	no RCT, 191 patients with advanced oral squamous carcinoma receiving neoadjuvant radiochemotherapy
Glaspy 1997	community study
Glaspy 2001	dose finding study, not randomised
Glaspy 2002	darbepoetin versus erythropoietin, related to Glaspy 2002b
Glaspy 2002b	darbepoetin versus erythropoietin, related to Glaspy 2002
Glaspy 2003	darbepoetin versus erythropoietin
Glaspy 2005	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Glaspy 2006	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Glaspy 2011	nrct
Glimelius 1998	randomised study with two different EPO groups: 2.000 U sc three times per week versus 10.000 U sc tree times per week
Glossmann 2003	randomised controlled comparison in patietns with relapsed lymphoma undergoing stem‐cell supported high‐dose therapy with or without erythropoietin
Goldsmith 2011	nrct
Granetto 2003	randomised comparison of fixed versus weight‐based dosing of epoetin alfa
Gregory 2005	nrct
Grigorescu 2006	probably not randomised
Hadland 2009	nrct
Han 2008	randomised trial of amifostine vs epo
Harousseau 2005	nrct
Hellström‐Lindberg 1998	G‐CSF supplementation only in the control arm
Hellstörm‐Lindberg 2010	nrct
Henry 2004	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Henry 2006	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Henry 2007	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Henze 2002	surgery
Hesketh 2004	comparison of different darbepo dosages
Hirsh 2007	CERA
Huggett 2011	not relevant (listed as "other reasons" in prisma")
Hyer 2011	nrct
Itzykson 2009	case report, no RCT
Jacubowski 2003	comparisons of epo versus darbepo, ongoing
Jadersten 2008	nrct
Jitnuyanont 2001	randomised comparison of erythropoietin versus control in 24 anaemic cancer patients, study was split into three arms with 10 and less patients per study arm
Johansson 2001	randomised comparison of epeotin beta 1,000 IU versus 5,000 IU three times per week
Jones 2011	nrct
Juan 2006	no RCT
Justice 2005	compared different ESA products (sc vs iv)
Kara 2008	nrct
Katodritou 2008	nrct
Katodritou 2009	nrct
Kettelhack 1998	ESAs were given in context with surgery
Kim 2010	no cancer
Kosmadakis 2003	ESAs were given in context with surgery
Kotasek 2004	compared different ESA dosages
Kotasek 2007	compared different ESA dosages
Larsson 2008	n = 18, too small for inclusion
Lastiri 2002	nrct
Latagliata 2008	nrct
Lavey 1993	nrct
Lavey 2004	nrct
Leitgeb 1994	nrct
Leon 1998	historical control group
Lichtin 2008	nrct
Loibl 2006	all patients received ESAs
Ludwig 1995	nrct
Malik 1998	nrct, N = 23
Mangiameli 2002	randomised controlled study with 5 patients per study arm
Mantovani 2000	nrct
Marinaccio 2003	randomised comparison of epoetin alfa versus control for patients with ovarian cancer undergoing surgery and chemotherapy, epoetin was administered BEFORE surgery, n = 22
Markman 1993	comparison of two non‐randomised trials
Mel 2008	nrct
Merchionne 2009	nrct
Merlano 2001	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
MF4266	ineligible patient characteristics AML
Miller 2004	unclear wether this is a randomised trial
Morishima 2006	compared different ESA dosages
Morrow 2007	nrct
Muravyov 2009	single arm study without control group
Muravyov 2010	nrct
Mustacchi 2006	nrct
Nagel 2011	too many patients in experimental arm did not receive ESAs
NCT00364845 2011	no cancer
Nonoguchi 2009	nrct
Oberhoff 2005	nrct
Olsson 2002	randomised comparison of epeotin beta 1,000 IU versus 5,000 IU three times per week
Opie 2011	nrct
Ots 2008	epoetin vs darbepoetin
Pat 2009	nrct
Pierelli 1999	randomised‐controlled trial, unblinded, not placebo controlled, 50 participants with ovarian carcinoma were included, as no sufficient data were available, the study was excluded
Poirier 2010	nrct
Policarpo 2007	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Popescu 2010	nrct
Porter 1996	too small for inclusion
Puglisi 2009	subgroup analysis
Rades 2009	nrct
Rath 2010	nrct
Rau 1998	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Rearden 2004	randomised comparison of darbepoetin early and late initiating of treatment
Reed 2005	compared different ESA products (epoetin versus darbepoetin)
Ribatti 2009	nrct
Richardson 2011	nrct
Rodgers 2008	nrct
Sakai 2004	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Samper 2002	randomised comparison of erythropoetin alfa 10,000 IU three times per week versus 30,000 IU once per week
Schwartzberg 2004	darbepo versus epo
Schwartzberg 2005	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Schwartzberg 2007	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Scott 2002	randomised comparison of erythropoietin versus placebo in head and neck cancer patients undergoing surgery
Senecal 2005	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Shasha 2003	community study
Shi 2007	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Shord 2009	nrct
Spaeth 2010	nrct
Spicka 2004	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Steensma 2006	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Steensma 2011	all patients EPO
Stein 1991	very small study, fewer than 10 participants per treatment arm IPD: ineligible patient characteristics (e.g. with MDS or SAA)
Steinmetz 2011	nrct
Stokoe 2009	nrct
Stone 2008	nrct
Straus 2002	upfront epo compared to epo if Hb decreased below 9 g/dL
Straus 2006	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Stull 2010	nrct
Suzuki 2008ex	single arm study
Tsukuda 1998	very small study, fewer than 10 participants in the study arms
Tzekova 2009	nrct
Uhl 2007	nrct
Vadhan‐Raj 2003	nrct
Vadhan‐Raj 2004	surgery
Varan 1999	treatment allocation not concealed IPD: too small for inclusion
Vekeman 2009	nrct
Velilla Millan 2003	randomised comparison of epeotin 10,000 IU tiw versus 40,000 IU qw
Vorvaud 2007	nrct
Wagner 2004	no usable data for any outcome
Walsh 2010	nrct
Waltzman 2004	comment to Glaspy 2003 study IPD: Darbepoetin compared to erythropoietin
Waltzman 2005	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Wauters 2006	nrct
WHO 2011	nrct
Yang 2008	nrct
Yilmaz 2004	application of different erythropoietin alfa dosages: 150 IU/kg tiw versus 250 IU/kg tiw in children with cancer
Yurut‐Caloglu 2008	nrct
Zagari 2003	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Zaragoza 2004	n = 17, too small for inclusion
Zhang 2003	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised
Zhou 2006	ESAs were given in context with surgery, stem cell transplantation, compared different ESA dosages or ESA products (epoetin versus darbepoetin), or trials were not randomised

ALL: acute lymphatic leukaemia 
 AML: acute myeloid leukaemia 
 CERA: continuous erythropoietin receptor activator 
 ESAs: erythropoiesis stimulating agents 
 G‐CSF: granulocyte colony‐stimulating factor 
 Hb: haemoglobin 
 IPD: individual patient data 
 iv: intravenous 
 MDS: myelodysplastic syndrome 
 n: number 
 nrct: not a randomised controlled trial 
 RCT: randomised controlled trial 
 SAA: serum amyloid A 
 SCLC: small cell lung cancer 
 sc: subcutaneous 
 vs: versus

Characteristics of ongoing studies [ordered by study ID]

Blackstock.

Trial name or title	1CDR0000069148CCCWFU‐62299; NCI‐P01‐0200; CCCWFU‐BG01‐193
Methods
Participants	Solid ‐ NSCLC Chem + Rad
Interventions	Epo, unsure ‐ Epo dose unknown
Outcomes	Hb levels, disease progression, tumour response rate, overall survival, QoL, number of RBCT
Starting date	January 2002
Contact information	AR Blackstock
Notes

Boehrer 2010.

Trial name or title	GFMAzaEpo‐2008‐1 trial, NCT01015352
Methods	Randomised phase‐II trial
Participants	Lower Risk Myelodysplastic Syndrome (MDS)
Interventions	AZA vs AZA+Epoetin beta
Outcomes	Major erythroid responses (HI‐E major) after 6 courses, according to IWG 2000 criteria. Secondary endpoints included overall IWG 2000 HI‐E, including major and minor, after 4 and 6 courses, response duration, IPSS progression, survival and toxicity.
Starting date	NR
Contact information
Notes	http://abstracts.hematologylibrary.org/cgi/content/abstract/ashmtg Unclear whether it should be included or not.

Broadley.

Trial name or title	Double‐blind randomised placebo controlled trial of the effect of epoetin alfa on symptomatic anaemia and fatigue in cancer patients receiving ongoing care without planned chemotherapy
Methods
Participants	Disease: metastatic breast and prostate cancer; Treatment: none
Interventions	Epoetin alfa vs placebo
Outcomes	Change in HB values; QoL.
Starting date	October 1998
Contact information	Dr K. Broadley Palliative Medicine The Royal Marsden NHS Trust Fulham Road Chelsea London SW3 6JJ UK
Notes

Chapman 2004.

Trial name or title	NRR 2004 Issue 2 Study ID numbers: N0123138194, REC 01/05/53.C
Methods
Participants	Haem ‐ MM, Chemotherapy
Interventions	Epo alfa vs standard, epo dose unknown
Outcomes	Hb Respnse, QoL (FACT‐An)
Starting date	Jan 2002, Finish date: June 05
Contact information	Chapman C
Notes

Delarue 2009.

Trial name or title	LNH03‐6B GELA Study
Methods	Multicentric prospective randomised phase III study
Participants	Elderly patients with DLBCL treated by immunochemotherapy
Interventions	Darbepoetin vs usual treatment
Outcomes	The efficacy of DA in association with chemotherapy (R‐CHOP) as measured by the EFS at 2 years, events being defined as death from any cause, relapse for complete responders and unconfirmed complete responders, progression during or after treatment and changes of therapy during allocated treatment. Secondary objectives were OS, PFS, DFS, response rate and analysis of toxicity.
Starting date	NR
Contact information	NR
Notes	Interim analysis

EPO CAN 303.

Trial name or title	EPO‐CAN‐303;NCT00083434R
Methods
Participants	Not reported, no concomitant anticancer therapy
Interventions	Epo alfa?
Outcomes	NR
Starting date	NR
Contact information	NR
Notes

Famoyin 2004.

Trial name or title	A randomised phase II study of thalidomide with or without erythropoietin (EPO) in metastatic renal cell carcinoma (RCC)
Methods	Randomised phase II study
Participants	Metastatic renal cell carcinoma
Interventions	Thalidomide with or without EPO
Outcomes	Hb, QoL
Starting date
Contact information	C. Famoyin, C. Byrnes, S. Roberts, J. Gollob, M. Atkins, J. Mier, Y.‐J. Ko, S. Gautam and D. McDemott Beth Israel Deaconess Medical Center, Boston, MA
Notes	Journal of Clinical Oncology, 2004 ASCO Annual Meeting Proceedings (Post‐Meeting Edition). Vol 22, No 14S (July 15 Supplement), 2004: 4747

Fisch.

Trial name or title	CDR0000069409; MDA‐DM‐02331; MDA‐DM‐0038; NCI‐P02‐0225; NCI00052221MJ
Methods
Participants	Solid tumours, no concomitant anti‐ malignancy treatment
Interventions	Epo alfa vs Placebo
Outcomes	Hb response, QoL, Fatigue
Starting date	Feb 03
Contact information	Fisch S
Notes

Gallagher.

Trial name or title	The role of epoetin alpha in anaemia and fatigue in cancer patients
Methods
Participants	Disease: ovarian and cervical cancer Treatment: concomitant cisplatin chemotherapy
Interventions	Epoetin alpha vs no treatment
Outcomes	Changes in HB values; QoL
Starting date	September 1998
Contact information	Dr C Gallagher Medical Oncology Department St Bartholomew´s Hospital West Smithfield London EC1A 7BE UK
Notes

Gamucci 1993.

Trial name or title	Erythropoietin for the prevention of anaemia in neoplastic patients treated with cisplatin
Methods	Randomised controlled study
Participants	Patients with advanced tumours
Interventions	Epoetin alpha versus control
Outcomes	Hb, serum epoetin levels
Starting date	NR
Contact information	T. Gamucci, Department of Medical Oncology, 1, Regina Elena Institute for Cancer Research, Rome, Italy,
Notes	Gamucci T, Thorel MF, Frasca AM, Giannarell D, Calabresi F. Erythropoietin for the prevention of anaemia in neoplastic patients treated with cisplatin. Eur J Cancer. 1993;29A Suppl 2:S13‐4. PMID: 8398359

Gascon 2010.

Trial name or title	Amgen 20070782; ClinicalTrials.gov NCT00858364
Methods	Randomized, placebo‐controlled trial
Participants	Non‐small cell lung cancer (NSCLC) patients (pts) with anaemia concomitant with chemotherapy (ACC)
Interventions	Darbepoetin vs placebo
Outcomes	Noninferiority of ESA to placebo for overall survival (primary endpoint) and progression‐free survival (modified RECIST per investigator; secondary endpoint) will be examined when 2700 deaths occur. Other safety endpoints include tumour response and thromboembolic events. Transfusion rates are a key efficacy endpoint; Hb changes will also be reported.
Starting date	NR
Contact information	NR
Notes	No results yet; to follow up

Ghavamzadeh 2010.

Trial name or title	NR
Methods	Randomized, double blind, placebo‐controlled phase III study
Participants	Anaemic patients with lung or gynaecologic cancer receiving platinum‐containing chemotherapy
Interventions	Darbepoetin alpha
Outcomes	NR
Starting date	NR
Contact information	NR
Notes	Not retrievable from librarians. unclear whether it would be included or not

H Thomas 1997.

Trial name or title	Open label comparative evaluation of the effect of epoetin on quality of life and burden of illness in anaemic patients receiving platinum‐containing chemotherapy
Methods
Participants	Disease: cancer; Treatment: concomitant platinum‐based chemotherapy
Interventions
Outcomes	Changes in Hb values; QoL
Starting date	July 1997
Contact information	Dr Hilary Thomas Department of Clinical Oncology Imperial College School of Medicine Hammersmith Hospital Du Cane Road London W12 0HS UK
Notes

Howell.

Trial name or title	A double blind, randomised, placebo‐controlled study to evaluate the impact of maintaining haemoglobin using epoetin alpha in stage IV breast cancer subjects receiving chemotherapy
Methods
Participants	Disease: breast cancer; Treatment: concomitant chemotherapy
Interventions	Epoetin alfa vs placebo
Outcomes	Change in Hb values; Transfusion requirement; AE; OS; QoL.
Starting date	August 2000
Contact information	Prof A. Howell Christie Hospital NHS Trust Wilmslow Road Withington Manchester M20 4BX UK
Notes

Koelbl.

Trial name or title	CDR 0000257189;AGOSG‐OVAR‐MO16375‐MARCH; EI20217; ROCHE‐MO16375; ROCH‐RO2053859
Methods
Participants	Solid cancers, Cervix ca, undergoing chemo ‐ plat + Rad
Interventions	Epo beta vs Standard care , epo dose unknown
Outcomes	Hb, AE, TR, Hb response, QoL
Starting date
Contact information	H Koelbl
Notes

Lambin 2006.

Trial name or title	Lambin, was CDR0000068669 and EORTC 22996‐24002
Methods	A phase III double‐blind, randomised, placebo‐controlled study of Erythropoietin when used as an adjuvant to radiation therapy in patients with head and neck squamous cell carcinoma
Participants	Head and neck squamous cell carcinoma; Radiation
Interventions	Epo once a week +RT +/‐chemotherapy versus placebo +RT +/‐chemotherapy
Outcomes	Loco‐regional control, overall survival, Hb levels during radiotherapy, adverse effects
Starting date	February 1999 designed for 762 patients
Contact information	Dr. P‐. Lambin, EORTC, Maastricht/Heerlen, The Netherlands Dr V. G. Budach Arbeitsgemeinschaft Radiologische Onkologie Dr J. Bernier EORTC Head and Neck Cancer Group Dr J.‐H. Bourhis Groupe d'Oncologie et Radiothérapie Tête et Cou Dr J. Denham Trans‐Tasman Radiation Oncological Group Incorporated
Notes

Liang 2009.

Trial name or title	The clinical study on recombinant human erythropoietin for chemotherapy ‐ related anaemia
Methods	Randomised controlled trial, not placebo‐controlled
Participants	N = 62 Disease: NR Treatment: chemotherapy Baseline Hb: NR
Interventions	Drug: Epo Dose: 12'000 IU tiw Hb‐target: NR Duration: 6 weeks
Outcomes	NR
Starting date
Contact information
Notes	Liang J, Qing B, Shen L, Cheng H. The clinical study on recombinant human erythropoietin for chemotherapy ‐ related anaemia. Clinical Research 2009;29(1):58‐60; full text in Chinese, English abstract, study number = 54993

Nitz 2008.

Trial name or title	Adjuvant chemotherapy with or without darbepoetin in node‐positive breast cancer: a safety analysis from the phase III ARA plus trial
Methods	Randomised controlled trial
Participants	Breast cancer patients
Interventions	Darbepoetin versus control
Outcomes
Starting date
Contact information
Notes	Nitz U, Oberhoff C, Reimer T, Schumacher C, Hackmann J, Warm M, Uleer C, Runde V, Gluz O, Zuna I West German Study Group, Moenchengladbach, Germany; Marienhospital, Essen, Germany; Klinikum Suedstadt, Rostock, Germany; St. Elisabeth KH, Koeln, Germany; Uni Koeln, Koeln, Germany; HZM Pharmaservice, Wiesbaden, Germany; Praxis Gyn. Onko., Hildesheim, Germany; Wilhelm‐Anton‐H., Goch, Germany. Adjuvant chemotherapy with or without darbepoetin in node‐positive breast cancer: a safety analysis from the phase III ARA plus trial. San Antonio Breast Cancer Symposium 2008:4100.

O´Brien.

Trial name or title	Open‐label randomised group‐comparative evaluation of the effect of epoetin on anaemia and fatigue in lung cancer patients receiving palliative platinum containing chemotherapy
Methods
Participants	Disease: lung cancer Treatment: concomitant platinum containing chemotherapy
Interventions	Epoetin vs standardised intervention
Outcomes	Change in HB values; QoL; Response and tolerance to chemotherapy.
Starting date	August 1998
Contact information	Dr Mary O´Brien Medicine Section The Royal Marsden NHS Trust Downs Road Sutton Surrey SM2 5PT UK
Notes

O´Connell.

Trial name or title	Phase III randomised study of epoetin alfa in anaemic patients with advanced cancer undergoing chemotherapy
Methods
Participants	Disease: advanced cancer; Treatment: concomitant chemotherapy
Interventions	Epoetin alfa vs placebo
Outcomes	Change in Hb values; Transfusion requirement; QoL.
Starting date	December 1998
Contact information	Dr M. J. O´Connell North Central Cancer Treatment Group USA
Notes

Park 1996.

Trial name or title	Clinical evaluation of recombinant human erythropoietin (Eprex) in anaemic cancer patients receiving chemotherapy
Methods	Randomised controlled trial
Participants	Anaemic cancer patients (lung cancer, stomach cancer, colon cancer)
Interventions	Epeotin alpha
Outcomes	Haematocrit, RBC transfusion
Starting date	NR
Contact information	NR
Notes	Park HS, Hong DS, Lee SJ, Chung TJ, Choi YM. Clinical evaluation of recombinant human erythropoietin(Eprex) in anaemic cancer patients receiving chemotherapy. Ann Oncol 1996;7(Suppl 5).

Parliament.

Trial name or title	Radiation therapy with or without Epoetin alfa in anaemic patients with head and neck cancer
Methods
Participants	Disease: need and neck cancer; Treatment: concomitant radiotherapy
Interventions	Epoetin: Control: no treatment
Outcomes	Changes in Hb values; AEs; QoL; local tumour control rate; OS.
Starting date
Contact information	Cross Canada Institute Edmonton, Alberta T6G 1Z2 Canada
Notes

R Thomas 2002.

Trial name or title	EPREX trial: Open‐label comparative‐group evaluation of the effect of epoetin alfa on quality of life and burden of illness in anaemic cancer patients receiving platinum containing chemotherapy
Methods
Participants	Disease: various malignancies; Treatment: concomitant platinum‐based chemotherapy
Interventions	Epoetin alfa vs no treatment
Outcomes	QoL; Patient burden; Transfusion requirements; Change in Hb values.
Starting date	June 1997
Contact information	Dr R. Thomas Primrose Oncology Unit Bedford South Wing Hospital Kempston Road Bedford MK42 9DJ UK
Notes

Recasens 2003.

Trial name or title	Randomised comparison of epoetin alfa versus control in patients with multiple myeloma, n=91, only interim analysis available so far, costs reported
Methods	Randomised controlled trial
Participants	Patients with multiple myeloma
Interventions	Epoetin alpha versus control
Outcomes	Costs
Starting date
Contact information	pgiraldo@salud.aragon.es
Notes	Recasens V, Rubio‐Martinez A, Gomez‐Barrera M, Rubio‐Felix D, Giralt M, Giraldo P. A pharmacoeconomical analysis comparing Epoetin Alpha vs transfusion in patients with anaemia associated to multiple myeloma. Blood. 2003; Vol. 102, issue 11.

Rexer 2006.

Trial name or title	Prospektiv offene, randomisierte Phase III Studie zur Evaluation von Darbepoetin Alfa (Aranesp) als Supportivtherapie bei Paienten mit "good/intermediate prognosis" Keimzelltumoren: PEB versus PEB+ Darbepoetin alfa (Aranesp"), Leitung: Prof. Bokemeyer, University Hospital Eppendorf, Hamburg, Germany
Methods	Randomised controlled trial
Participants	Testicular cancer patients
Interventions	Darbepoetin versus control
Outcomes	Hb, RBC transfusion, QoL, disease progressions, OS.
Starting date	NR
Contact information	Ina Böhlke, i.boehlke@uke.uni‐hamburg.de
Notes	Rexer H [Darbepoetin alfa (Aranesp) as supportive therapy in patients with germ cell tumours]. Urologe A. 2006 Aug;45(8):1017‐8. MeckEvidence, Geschäftsstelle der AUO, Seestr. 11, 17252, Schwarz. AUO@MeckEvidence.de

Rudd.

Trial name or title	Evaluation of epoetin in lung cancer pts. receiving chemotherapy
Methods
Participants	Disease: lung cancer; Treatment: concomitant platinum‐based chemotherapy
Interventions	Epoetin alpha vs no treatment
Outcomes
Starting date	November 1998
Contact information	Dr R. M. Rudd Medical Oncology Department St Bartholomew´s Hospital West Smithfield London EC1A 7BE UK
Notes

Stewart.

Trial name or title	Open randomised comparative group evaluation of the effect of epoetin alfa on local disease free survival and quality of life in head and neck cancer patients receiving radical radiotherapy
Methods
Participants	Disease: head and neck cancer; Treatment: concomitant radiotherapy
Interventions
Outcomes	Local tumour control; Disease‐free survival; QoL; OS.
Starting date	August 1999
Contact information	Dr J. S. Stewart Department of Radiotherapy Charing Cross Hospital Fulham Palace Road London W6 8RF UK
Notes

UKCCCR GN308.

Trial name or title	A double‐blind, placebo controlled study to assess the effects of early intervention and/or treatment with Epoetin alfa on anaemia in cancer patients receiving non platinum containing chemotherapy
Methods
Participants	Disease: cancer Treatment: concomitant chemotherapy
Interventions	Epoetin vs placebo
Outcomes	Transfusion requirement; QoL.
Starting date
Contact information	UKCCCR Register Co‐ordinator MRC Clinical Trials Unit 222 Euston Road London NW1 2DA
Notes

Yousseff 2011.

Trial name or title	The Effectiveness of a Fixed Low Dose of Erythropoietin (EPO) in Anemic Solid Tumor Patients Receiving Concomitant Chemotherapy: A Prospective, Randomized, Controlled Study
Methods	Randomised controlled trial
Participants	Anemic solid tumour patients
Interventions	Low dose Epoetin versus control
Outcomes
Starting date
Contact information
Notes	Youssef Lama A, Hussien Dilman Haj, Sulaiman Siham. The Effectiveness of a Fixed Low Dose of Erythropoietin (EPO) in Anemic Solid Tumor Patients Receiving Concomitant Chemotherapy: A Prospective, Randomized, Controlled Study. ASH Annual Meeting Abstracts 2011;118(21):2092

Ziras 2001.

Trial name or title	Soluble transferrin receptor (sTFR) as a predictor of response to prophylactic epoetin alfa (EPO) treatment in non‐anaemic cancer patients (pts) under chemotherapy (CT). Preliminary results.
Methods	Randomised controlled trial
Participants	Non‐anaemic cancer patients
Interventions	Epoetin alpha
Outcomes	Hb
Starting date	NR
Contact information	Agii Anargiri Cancer Hospital Athens, Metaxa Cancer Hosptial, Piraeus, Greece
Notes	Proc Am Soc Clin Oncol 20:2001 (abstr 2987)

AE: adverse event 
 AZA: Azacitidine 
 DLBCL: diffuse large B‐cell lymphoma. 
 EPO: Erythropoietin 
 ESA: erythropoiesis stimulating agent 
 G‐CSF: granulocyte colony‐stimulating factors 
 Hb: haemoglobin 
 IPSS: International Prognostic Scoring System 
 OS: overall survival 
 PFS: progression‐free survival 
 Pts.: patients 
 QoL: quality of life 
 RBCS: Red blood cells 
 s.c.: subcutaneous 
 vs.: versus 
 NR: not reported 
 RBCT: red blood cell transfusions 
 RT: radiotherapy 
 TR: tumour response

Differences between protocol and review

In previous versions of this review, we required dosages of at least 300 U/kg body weight per week (epoetin‐a and beta) given for at least four weeks. For the current update this criterion was removed and we included studies or study arms with low dosages as well. In previous versions of this review, iron supplementation had to be identical in the study arms of a given trial. For the current update, we changed this criterion and included trials using iron supplementation in the experimental but not in the control arm as well.

Contributions of authors

Thomy Tonia: Quality of Life analysis, searching for trials, eligibility and quality assessment, fact checking, data extraction and analysis, drafting of QoL and discussion, revision of review

Annette Mettler: Searching for trials, eligibility and quality assessment, data extraction and analysis

Nadège Robert: Searching for trials, eligibility and quality assessment, data extraction and analysis

Guido Schwarzer: Statistical and methodological advice, data analysis, content input

Olaf Weingart: eligibility and quality assessment, fact checking, creating summary of findings table

Jerome Seidenfeld: Clinical and scientific advice, content input, revision of draft review

Chris Hyde: Protocol development, searching for trials, eligibility and quality assessment, data extraction and analysis

Andreas Engert: Clinical and scientific advice, content input

Julia Bohlius: Protocol development, searching for trials, eligibility and quality assessment, data extraction and analysis, drafting and revision of review

Sources of support

Internal sources

• Department of Internal Medicine I, University of Cologne, Germany.

• Cochrane Haematological Malignancies Group (CHMG), Germany.

• Institute of Social and Preventive Medicine, University of Bern, Switzerland.

External sources

• Department of Health, UK.

Declarations of interest

Thomy Tonia, Annette Mettler, Nadège Robert, Olaf Weingart, Guido Schwarzer, Jerome Seidenfeld, Chris Hyde, Andreas Engert, Julia Bohlius: none known.

New search for studies and content updated (conclusions changed)