Erythropoiesis stimulating agents for anaemia in adults with chronic kidney disease: a network meta-analysis

  • Protocol
  • Intervention

Authors

  • Suetonia C Palmer,

    Corresponding author
    1. University of Otago Christchurch, Department of Medicine, Christchurch, New Zealand
    • Suetonia C Palmer, Department of Medicine, University of Otago Christchurch, 2 Riccarton Ave, PO Box 4345, Christchurch, 8140, New Zealand. suetonia.palmer@otago.ac.nz.

    Search for more papers by this author
  • Georgia Salanti,

    1. University of Ioannina School of Medicine, Department of Hygiene and Epidemiology, Ioannina, Greece
    Search for more papers by this author
  • Jonathan C Craig,

    1. The University of Sydney, Sydney School of Public Health, Sydney, NSW, Australia
    2. The Children's Hospital at Westmead, Cochrane Renal Group, Centre for Kidney Research, Westmead, NSW, Australia
    Search for more papers by this author
  • Dimitris Mavridis,

    1. University of Ioannina, Department of Hygiene and Epidemiology, School of Medicine, Ioannina, Greece
    Search for more papers by this author
  • Giovanni FM Strippoli

    1. The University of Sydney, Sydney School of Public Health, Sydney, NSW, Australia
    2. The Children's Hospital at Westmead, Cochrane Renal Group, Centre for Kidney Research, Westmead, NSW, Australia
    3. University of Bari, Department of Emergency and Organ Transplantation, Bari, Italy
    4. Mario Negri Sud Consortium, Department of Clinical Pharmacology and Epidemiology, Santa Maria Imbaro, Italy
    5. Diaverum, Medical-Scientific Office, Lund, Sweden
    Search for more papers by this author

Abstract

This is the protocol for a review and there is no abstract. The objectives are as follows:

We aim to conduct a systematic review and meta-analysis of all RCTs that compare ESAs (epoetin alfa, epoetin beta, darbepoetin alfa, or methoxy polyethylene glycol-epoetin beta) against placebo, no treatment, or other ESA in adults with CKD. Using network meta-analysis, we aim to assess the safety and efficacy of ESAs to treat anaemia and to generate a clinically useful ranking off available ESAs according to their safety and efficacy.

Background

Description of the condition

Anaemia, literally meaning "lack of blood", is defined as "a condition in which the number of red blood cells or their oxygen-carrying capacity is insufficient to meet physiological needs" (http://www.who.int/topics/anaemia/en/). Circulating red blood cells transport oxygen to tissues bound to iron ions within the metalloprotein, haemoglobin. In anaemia, insufficient numbers of circulating red blood cells or inadequate quantities of iron or functional haemoglobin are available to transport and release oxygen to tissues, which is essential for aerobic (oxygen-dependent) metabolism. Anaemia, defined by the World Health Organization as a haemoglobin level below 130 g/L in men and below 120 g/L in women, affects approximately one-quarter of the world's population, particularly children and pregnant women (WHO 2008). Anaemia is common in the expanding global populations of chronic disease including people affected by solid malignancies (50%), blood cancers (60% to 70%) (Ludwig 2004), human immunodeficiency virus (HIV) causing acquired immunodeficiency syndrome (40%) (Shah 2007), chronic heart failure (20%) (Ezekowitz 2003), and nearly all individuals who have advanced chronic kidney disease (CKD). Symptoms caused by insufficient oxygen delivery to tissues in anaemia include weakness and fatigue, breathlessness, light-headedness, and palpitations. Observational cohort data show that anaemia in people who have chronic disease is also consistently associated with negative effects on quality of life (Lefebvre 2006), role function (Ludwig 2004; Semba 2005), and survival (Caro 2001; Groenveld 2008; Locatelli 2004; Melekhin 2012).

Description of the intervention

Recombinant erythropoietin and its synthetic derivatives (epoetin alfa, epoetin beta, darbepoetin alfa, methoxy polyethylene glycol-epoetin beta; collectively known as erythropoiesis-stimulating agents (ESAs)), are widely used to treat anaemia. Erythropoietin is a glycoprotein made by peritubular cells in the kidney (with an additional smaller contribution from liver cells (15% total)) and is released in response to low tissue oxygen levels (hypoxia) through the actions of hypoxia-inducible factor and stimulates the formation and viability of red blood cells in the bone marrow (erythropoiesis). The average red blood cell survives in the circulation for 120 days although red cell survival is reduced by chronic disease. Causes of anaemia are numerous and include: reduced production of erythropoietin in response to hypoxia (CKD; chronic inflammatory conditions); abnormal bone marrow responses to the actions of erythropoietin (chronic inflammatory conditions, bone marrow failure due to infiltration or drug-related therapy); insufficient iron stores; abnormal production or function of haemoglobin (thalassaemia or haemoglobinopathies); excessive red blood cell losses (destruction within the circulation or haemorrhage); or reduced red blood cell survival (Figure 1).

Figure 1.

Overview of anaemia in chronic disease

Before the development of recombinant human erythropoietin in the late 1980s (Eschbach 1987), blood transfusions and iron supplementation (both oral and intravenous (IV)) were the mainstay of treatment for anaemia in populations with severe CKD, in which haemoglobin levels were commonly in the range of 70 to 80 g/L. Androgen treatment for anaemia was also used in CKD but provided small and unsustained responses in haemoglobin levels and was poorly tolerated (Neff 1981). In the pre-recombinant-erythropoietin era, blood transfusions effectively increased haemoglobin levels to provide acute symptom relief but were associated with hospitalisation, iron overload, antibody formation against blood cell antigens, sensitisation to transplant antigens, and transfusion-related infections, particularly viral hepatitis. Technological advances and successful cloning of the erythropoietin gene allowed for large-scale production of human recombinant erythropoietin which effectively and rapidly increases haemoglobin levels when administered intravenously or subcutaneously. The United States Food and Drug Administration (FDA) approved human recombinant erythropoietin for the treatment of anaemia in people with CKD on dialysis in 1989 and broadened approval to use in people with CKD without dialysis, and in patients with HIV and anaemia on zidovudine (AZT) in 1990.

Clinical guidelines published soon after initial drug approval suggested that patients with CKD and haemoglobin concentrations below 80 g/L who were symptomatic should receive erythropoietin in conjunction with sufficient iron supplementation once other causes of anaemia were excluded (Macdougall 1990). However, rapid widespread uptake of ESAs occurred in numerous clinical settings and, by 2007, clinical practice guidelines recommended the use of erythropoietin to achieve target haemoglobin levels of 110 to 120 g/L in people with CKD (KDOQI 2007). Erythropoietin prescription also subsequently expanded to treat anaemia in cancer and heart failure populations, as well as for people undergoing surgery likely to require blood transfusion who could not undergo pre-operative blood collection. Presently, epoetin alfa is approved by the FDA for treatment of anaemia due to CKD, zidovudine in HIV-infected patients, effects of concomitant myelosuppressive chemotherapy and to reduce red blood cell transfusions in patients undergoing elective, noncardiac, nonvascular surgery. Darbepoetin alfa is currently approved by the FDA for the treatment of anaemia resulting from CKD or myelosuppressive chemotherapy (FDA website).

How the intervention might work

Despite an association between low haemoglobin levels and higher mortality in uncontrolled studies, prompting speculation that correcting anaemia with ESAs might lower cardiovascular events and mortality, the opposite was observed in subsequent meta-analyses of randomised controlled trials (RCTs) for ESAs (Bohlius 2009; Palmer 2010; Phrommintikul 2007; Strippoli 2006). Correction of anaemia and maintenance of haemoglobin levels to near normal levels with ESAs reduced the need for red blood cell transfusions, but increased mortality, cardiovascular events and cancer-progression, without consistently improving quality of life. The precise mechanisms for treatment-related harm are not understood, but observational studies suggest that impaired haemoglobin responses to erythropoietin treatment, together with higher erythropoietin doses are associated with increased treatment-related toxicity (Kilpatrick 2008; Szczech 2008).

Treatment guidelines for ESAs to treat anaemia have become more conservative over the last decade and FDA labelling now suggests that ESA treatment should be considered in people with CKD when the haemoglobin level is less than 100 g/L, and treatment objectives are to increase haemoglobin levels sufficient to reduce the need for red cell transfusions (FDA website). Clinical practice guidelines have also responded to increasing evidence of harm when higher haemoglobin levels are targeted by ESA treatment (Bohlius 2009; Palmer 2010; Phrommintikul 2007). Recent clinical practice guidelines for the use of ESAs to treat anaemia in CKD suggest the potential benefits of reducing blood transfusions and anaemia-related symptoms should be balanced against the risks of harm (e.g., stroke, vascular access loss and hypertension) for individual patients. Currently guidelines do not suggest specific haemoglobin targets for patients not treated with dialysis, while for dialysis patients the recommended approach is to use ESA therapy to avoid a haemoglobin level below 9.0 g/dL (KDIGO 2010).

Why it is important to do this review

Darbepoetin alfa and methoxy polyethylene glycol-epoetin beta (a continuous erythropoietin-receptor activator (CERA)) are newer synthetic forms of naturally-occurring erythropoietin that have a longer duration of action (Macdougall 2008). These agents have similar effects on haemoglobin concentrations as epoetin alfa and beta and require less frequent administration (Macdougall 2001; Levin 2007). Darbepoetin alfa treatment in people with earlier stages of CKD and diabetes mellitus has been shown to nearly halve the risk of blood transfusion but has no beneficial effects on survival and increases the risk of stroke and death related to cancer recurrence (TREAT 2009).

The apparent narrow therapeutic balance between potential treatment benefits (avoidance of red blood cell transfusions) and hazards (cardiovascular events, mortality, and cancer disease progression or recurrence) together with the availability of several agents in this drug class (epoetin alfa, epoetin beta, darbepoetin alfa and methoxy polyethylene glycol-epoetin beta) to treat anaemia builds the case for a comprehensive and systematic head-to-head comparison of the available ESAs to treat anaemia. However, large-scale trials directly comparing different ESAs have been relatively uncommon and the comparative efficacy and safety of each agent relative to each other is poorly understood. In the era of increasing caution in prescribing ESAs to increase haemoglobin levels due to the potential hazards of targeting higher haemoglobin levels, it is plausible that a patient could be randomised in a single multi-arm trial to receive any ESA to treat anaemia.

We will conduct a systematic review and meta-analysis that aims to compare multiple ESAs for the treatment of anaemia in adults with CKD and if deemed appropriate and feasible, we will undertake a network meta-analysis (NMA) to rank the efficacy and acceptability of all available treatments.

Objectives

We aim to conduct a systematic review and meta-analysis of all RCTs that compare ESAs (epoetin alfa, epoetin beta, darbepoetin alfa, or methoxy polyethylene glycol-epoetin beta) against placebo, no treatment, or other ESA in adults with CKD. Using network meta-analysis, we aim to assess the safety and efficacy of ESAs to treat anaemia and to generate a clinically useful ranking off available ESAs according to their safety and efficacy.

Methods

Criteria for considering studies for this review

Types of studies

We will include all RCTs comparing ESAs versus ESAs, placebo or no treatment to treat anaemia in people with CKD. We will not restrict inclusion based on language of publication. We will not include quasi-RCTs (trials in which treatment allocation was by date of birth, alternation, or similar predictable method). We will include trials in which allocation to treatment was not adequately concealed but will consider trial methodological quality in our analyses and discussion.

Types of participants

Inclusion criteria

Patients aged 18 years or older with anaemia or who are at risk from anaemia due to CKD will be included. Participants will have CKD, defined according to National Kidney Foundation Kidney Disease Outcomes Quality Initiative (KDOQI) criteria (NKF 2002) including recipients of a kidney transplant, people with end-stage kidney disease (ESKD) treated with dialysis (CKD stage 5D), and those with kidney disease characterised by clinically-relevant proteinuria, haematuria, and/or structural kidney disease with or without reduced glomerular filtration rate (GFR) and not treated with dialysis or transplantation (CKD stages 1 to 5).

Exclusion criteria

We will exclude trials in which follow-up for specified outcomes was fewer than three months. We will exclude paediatric trials as interventions and treatment strategies for anaemia in children are likely to be systematically different from those in adult CKD patients.

Types of interventions

We will consider trials of ESAs (epoetin (alfa or beta), darbepoetin alfa, and methoxy polyethylene glycol-epoetin beta) to treat or prevent anaemia in CKD administered via any route (IV or subcutaneous), compared with another ESA or to placebo or no treatment. We will include trials in which an ESA regimen is compared with different ESA regimen (either comparing two different ESAs or a single ESA using two different haemoglobin targets). Dose adaptation of ESAs and non-randomised iron supplementation depending on haematological response will be allowed. We will include trials in which iron was administered as a randomised intervention in all arms of the trial. We will stratify all interventions according to the ESA dosing strategy (target haemoglobin level, ceiling haemoglobin level, fixed dose, achievement of clinical outcome), to detect inequalities in dosing that could affect comparative efficacy.

We will include trials that assess one of more of the following interventions

  • epoetin alfa

  • epoetin beta

  • darbepoetin alfa

  • methoxy polyethylene glycol beta

  • placebo

  • no treatment

Figure 2 shows the overall network of eligible comparisons in the review. We will also explore the impact of low or high haemoglobin target levels in all ESA treatments (Figure 3). We will consider the interventions of 'placebo' or 'no treatment' separately from treatment guided by a 'low haemoglobin target' in this network. In addition, we will construct a network to detect differences between the four different ESA types (Figure 4).

Figure 2.

Full network to identify the ranking of treatments according to effectiveness and acceptability – to rank all doses and treatments

Figure 3.

Collapsed network 1: to identify the ranking of high versus low haemoglobin targets with placebo or no treatment according to effectiveness and acceptability – to evaluate whether the low and high doses make a difference

Figure 4.

Collapsed network 2: to identify the ranking of treatments according to effectiveness and acceptability – detect differences between the 4 different ESA types

We will code the comparisons within a study where iron is a randomised co-intervention in all trial arms as follows:

  • ESA1 plus iron (any route) versus ESA2 plus iron (any route) = ESA1 versus ESA2

  • ESA plus oral iron versus oral iron = ESA versus no treatment

  • ESA plus oral iron versus oral iron plus placebo injection = ESA versus placebo

  • ESA plus IV iron versus IV iron plus placebo injection = ESA versus placebo

  • ESA plus IV iron versus IV iron = ESA versus no treatment

We will exclude trials in which iron therapy is a randomised co-intervention combined with an ESA in a single arm of the trial (e.g., ESA plus iron versus ESA alone, ESA plus iron versus placebo).

Types of outcome measures

We will evaluate the following outcomes occurring at any time during trial follow-up.

Primary outcomes
Response to treatment

Proportion of patients who require one or more red blood cell transfusions (number of events)

Safety

All-cause mortality

Secondary outcomes
Response to treatment
  1. End of treatment haemoglobin (g/L)

  2. Fatigue (as defined by study authors)

  3. Dyspnoea (as defined by study authors)

Safety
  1. Proportion of patients with cardiovascular mortality (as adjudicated by investigators)

  2. Proportion of patients with one or more major cardiovascular events (as adjudicated by investigators)

  3. Proportion of patients with one or more myocardial infarctions (fatal or non-fatal)

  4. Proportion of patients with one or more stroke (fatal or nonfatal)

  5. Proportion of patients with dialysis vascular access thrombosis

  6. Proportion of patients who progress to ESKD (CKD stage 5)

  7. Proportion of patients with one or more cancers

  8. Proportion of patients withdrawing from treatment due to serious adverse events (as defined by study authors)

  9. Proportion of patients experiencing other adverse events (as described in individual trial reports)

Search methods for identification of studies

Electronic searches

We will search the following databases for review using search terms designed by a specialist Trials' Search Co-ordinator using search terms relevant to this review without language restriction.

  1. Cochrane Renal Group’s Specialised Register

  2. Cochrane Central Register of Controlled Trials (CENTRAL)

  3. Database of Abstracts of Reviews of Effects (DARE)

  4. Health Technology Assessment (HTA) database

  5. NHS Economic Evaluation Database

The Cochrane Renal Group's specialised register contains studies identified from:

  • Quarterly searches of the Cochrane Central Register of Controlled Trials CENTRAL

  • Weekly searches of MEDLINE OVID SP

  • Handsearching of renal-related journals and the proceedings of major renal conferences

  • Searching of the current year of EMBASE OVID SP

  • Weekly current awareness alerts for selected renal-journals

  • Searches of the International Clinical Trials Register (ICTRP) Search Portal and ClinicalTrials.gov

Studies contained in the Cochrane Renal Specialised register are identified through search strategies for CENTRAL, MEDLINE, and EMBASE based on the scope of the Cochrane Renal Group. Details of these strategies as well as a list of handsearched journals, conference proceedings and current awareness alerts are available in the specialised register section of information about the Cochrane Renal Group.

MEDLINE will be searched from 2005 onwards and EMBASE from 2010 onwards, as this review also includes populations other than CKD.

See Appendix 1 for search terms used in strategies for this review.

Searching other resources

  1. Review articles and relevant studies.

  2. Letters seeking information about unpublished or incomplete trials to investigators known to be involved in previous studies.

Data collection and analysis

We have drafted this protocol in advance of any trial selection or data analyses.

Selection of studies

The search strategy described will be used to obtain titles and abstracts of studies that may be relevant to the review. The titles and abstracts will be screened independently by two authors, who will discard studies that are not applicable; however studies and reviews that might include relevant data or information on trials will be retained initially. Two authors will independently assess retrieved abstracts and, if necessary the full text, of these studies to determine which studies satisfy the inclusion criteria. Systematic reviews will be screened to identify any studies not retrieved by the electronic database search.

Data extraction and management

Data extraction will be carried out independently by two authors using standard data extraction forms in an excel spread sheet.

The following data will be extracted.

  • Population: iron status at baseline (iron deficient versus iron replete), stage of CKD (CKD stages 1 to 3, CKD stages 4 to 5, CKD stage 5D, transplant), mean age, proportion of men, baseline haemoglobin level (< 10 g/dL, 10 to 12 g/dL, > 12 g/dL)

  • Intervention: ESA, dosing strategy (ceiling haemoglobin level, target haemoglobin range, fixed dose, dosing adjustment based on clinical outcome (e.g. avoidance of transfusion)), route (subcutaneous versus IV)

  • Comparison: ESA (different ESA or same ESA with different haemoglobin target), placebo or no treatment

  • Co-intervention: iron supplementation (fixed dose, as required, not clear)

  • Study design: inclusion criteria; primary endpoint; duration of ESA treatment (12 to 16 weeks; 16 to 24 weeks; 24 to 48 weeks; > 48 weeks); duration of follow-up (≥ 12 months versus < 12 months); number of participants; date of publication; number of centres; sequence generation; allocation concealment; blinding of outcome assessors; attrition; selective outcome reporting; source of funding; trial registration; publication (full text publication, abstract publication, unpublished data); period of collection of clinical outcomes (total duration of follow-up, specific phase(s) of follow-up)

Data will be cross checked between authors and discussed. Studies reported in non-English language journals will be translated before assessment. Where more than one publication of one study exists, reports will be grouped together and the publication with the most complete data will be used in the analyses. Where relevant outcomes are only published in earlier versions, these data will be used. Any discrepancy between published versions will be highlighted. Any disagreements in data extraction will be discussed with a third author. We will generate descriptive statistics for trial and study population characteristics across all eligible trials describing the types of comparisons and some important variables, either clinical or methodological (such as population, age, study quality, funding source).

Assessment of risk of bias in included studies

The following items will be independently assessed by two authors using the risk of bias assessment tool (Higgins 2011) (see Appendix 2).

  • Was there adequate sequence generation (selection bias)?

  • Was allocation adequately concealed (selection bias)?

  • Was knowledge of the allocated interventions adequately prevented during the study (detection bias)?

    • Participants and personnel

    • Outcome assessors

  • Were incomplete outcome data adequately addressed (attrition bias)?

  • Are reports of the study free of suggestion of selective outcome reporting (reporting bias)?

  • Was the study apparently free of other problems that could put it at a risk of bias?

Measures of treatment effect

Direct comparisons of treatment effects (ESA versus placebo/no treatment or other ESA)

First, we will conduct pair-wise meta-analyses by synthesising studies that compare the same interventions using a random-effects model (DerSimonian 1986). We will compare treatments that used the same haemoglobin target (e.g., epoetin high target versus darbepoetin high target) and different haemoglobin targets (e.g., epoetin high target versus epoetin lower target) if sufficient studies are available.

For dichotomous outcomes (one or more red blood cell transfusions; one or more hospitalisations for red blood cell transfusion; achievement of haemoglobin target; all-cause and cardiovascular mortality; major cardiovascular event; fatal or nonfatal myocardial infarction; fatal or nonfatal stroke; dialysis vascular access thrombosis; ESKD) results will be expressed as an odds ratio (OR) with 95% confidence intervals (CI). Where continuous scales of measurement are used to assess the effects of treatment (end of treatment haemoglobin level; time to red blood cell transfusion; number of transfusions/participant), the mean difference (MD) will be used, or the standardised mean difference (SMD) if different scales have been used. Analyses based on means are appropriate for data that are at least approximately normal distributed and for data from very large trials. If there is indication that continuous data are skewed (means and medians differ, or study reports 25th and 75th percentiles suggesting asymmetry), we will transform data on the log-scale and report geometric means and rations of geometric means. Log-transformed and untransformed data cannot be mixed in a meta-analysis. We will transform some the effect sizes so that everything is measured on the same scale, using the methods presented in Higgins 2008.

Adverse effects will be tabulated and assessed with descriptive techniques, as they are likely to be different for the various agents used. Where possible, the risk difference (RD) with 95% CI will be calculated for each adverse effect, either compared with no treatment or another agent.

We will evaluate the presence of heterogeneity within meta-analyses using the Cochran Q test and I² statistic (Higgins 2003). In the presence of statistical and clinical heterogeneity, and if sufficient studies are available we will consider the following potential sources of heterogeneity:

  • Mistakes and inconsistencies in data extraction and entry

  • Population: baseline stage of CKD (CKD stages 1-3, CKD stages 4-5, CKD stage 5D, transplant); baseline haemoglobin (< 10 g/dL; 10 to 12 g/dL; > 12 g/dL); mean age; gender; proportion with diabetes or cardiovascular disease

  • Intervention: dose, frequency or route; dose strategy (fixed dose, ceiling dose, target range, dosing adjustment based on clinical outcome (e.g., avoidance of blood transfusion)); iron supplementation strategy (fixed iron treatment, iron treatment as necessary, or not clear)

  • Risk of bias: allocation concealment, blinding of outcome assessment, attrition, premature termination of trial, publication (full text publication, abstract publication, unpublished data)

  • Funding source

  • Study design: duration of ESA treatment (12 to 16 weeks; 16 to 24 weeks; 24 to 48 weeks; > 48 weeks); duration of follow-up (≥ 12 months versus < 12 months); number of participants; date of publication.

Indirect and mixed comparisons of treatment effects

Network meta-analysis is a method of synthesising information from a network of trials addressing the same questions but involving different interventions. Joint analysis of data within a network framework allows novel inferences on treatment comparisons that have not been previously addressed directly in any studies, and it increases precision for comparisons with few data (Caldwell 2010; Lu 2004; Salanti 2008).

For a given comparison, say A versus B, direct evidence is provided by trials that compare these two treatments directly (epoetin alfa versus darbepoetin alfa) as in standard direct comparisons meta-analysis. In addition, indirect evidence for A versus B can be provided if studies that compare A versus C and B versus C are analysed jointly (e.g., epoetin alfa versus placebo trials and darbepoetin alfa versus placebo trials can allow indirect comparison of epoetin alfa versus darbepoetin alfa via the use of placebo). Network meta-analysis aims to combine the direct and indirect evidence into a single effect size and thus helps to increase the precision of the comparison, while randomisation is respected. The combination of direct and indirect evidence for any given treatment comparison can be extended when ranking more than three types of treatments according to their effectiveness or safety; every study contributes evidence in the network about a subset of the competing treatments.

We will perform network meta-analysis in STATA (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX) using the mvmeta command (White 2012) and self-programmed STATA routines available at http://www.mtm.uoi.gr. Results from network meta-analysis will be presented as summary effect sizes (odds ratios, mean differences, or standardised mean differences) and their 95% confidence intervals. A common heterogeneity variance will be presented. Ranking of treatments will be estimated using the surface below the ranking curve (SUCRA) (Salanti 2011).

Evaluating the assumption underlying network meta-analysis

The assumption of transitivity - that one can learn about treatment A versus treatment B via treatment C (e.g., learning about epoetin alfa versus darbepoetin alfa via placebo) - underlies network meta-analysis (Salanti 2012). Evaluation of the assumption is important and its plausibility determines the validity of the network meta-analysis results. In this context we expect that the transitivity assumption will hold assuming that:

  1. The common treatment used to compare different ESAs indirectly is similar when it appears in different trials (e.g., darbepoetin alfa is administered in a similar way in darbepoetin alfa versus epoetin alfa trials and in darbepoetin alfa versus methoxy polyethylene glycol-epoetin beta trials).

  2. All pair-wise comparisons comparisons do not differ with respect to the distribution of effect modifiers (e.g., the design and study characteristics of darbepoetin versus placebo trials are similar to epoetin alfa versus placebo trials)

  3. Participants in the network could in principle be randomised to any of the treatments being compared. For example an adult CKD patient with anaemia could be equally likely to be randomised to epoetin alfa, epoetin beta, darbepoetin alfa, methoxy-polyethylene glycol epoetin beta, placebo or no treatment.

The assumption of transitivity will be evaluated epidemiologically by comparing the clinical and methodological characteristics of sets of studies grouped by treatment comparisons.

Consistency is a prerequisite to calculating a valid mixed estimate of treatment effects (synthesising direct and indirect estimates within a single overall estimate of treatment comparisons). Lack of transitivity can manifest as disagreement between direct and indirect estimates. Joint analyses of direct and indirect evidence can be misleading if the network is inconsistent. Statistical methods can be employed to evaluate consistency in a network. We will use the following methods:

  • Loop-specific approach: A loop of evidence is formed by at least three treatment pairs which have been compared in studies forming a closed path. Indirect evidence can be contrasted to direct evidence and their difference defines their disagreement (inconsistency factor). To infer whether the inconsistency factor is incompatible with zero, we will look at the 95% confidence interval and a loop-specific z-test (often called the Bucher method)(Bucher 1997).

We will extend analysis to all closed loops assuming a loop-specific heterogeneity and examine the estimates of inconsistency together with 95% confidence intervals for each loop using a graphical representation (Salanti 2009), This approach can be easily applied and indicate loops with large inconsistency, but cannot infer consistency of the entire network or identify the particular comparison that is problematic. It should be noted that in a network of evidence there may be many loops and estimates of inconsistency factors and with multiple testing there is an increased likelihood that we might find an inconsistent loop by chance. Therefore, we will be cautious deriving conclusions from this approach.

  • Design by treatment interaction model: As we anticipate that our network meta-analysis will include some studies with more than two treatment arms, we will also evaluate inconsistency using a ‘design by treatment interaction model’ as fully explained in Higgins 2012 (pages 102 to 103). As that paper states 'The presence of multi-arm trials in an evidence network complicates the definition of loop inconsistency. Loop inconsistency cannot occur in a multi-arm trial'. While the loop inconsistency approach above evaluates whether direct and indirect evidence agree with each other, design inconsistency asks the question whether a particular choice of treatments in a study is associated with different effect sizes for a given treatment comparison. An ABC trial is a different design from an AB or BC trial; evaluation of design inconsistency asks whether the relative effectiveness of A versus B is different when estimated in AB or ABC studies. The design by treatment interaction model provides a global test for inconsistency.

If we find significant inconsistency, we will investigate the possible sources. We will investigate the distribution of prespecified clinical and methodological variables that we suspect may be potential sources of either heterogeneity or inconsistency in each comparison-specific group of trials. If sufficient studies are available we will consider the following potential sources of heterogeneity and/or network inconsistency.

  • Mistakes and inconsistencies in data extraction and entry.

  • Population: iron status at baseline (iron replete versus iron deficient); stage of CKD (CKD stages 1 to 3, CKD stage 4 to 5, CKD stage 5D, transplantation); baseline haemoglobin (< 10 g/dL, 10 to 12 g/dL, > 12 g/dL); mean age; gender; proportion with diabetes or cardiovascular disease

  • Intervention: dose, frequency or route; iron supplementation (fixed iron treatment, iron treatment as necessary, or not clear)

  • Risk of bias: allocation concealment; blinding of outcome assessment; attrition; premature termination of trial; publication (full text publication, abstract publication, unpublished data); funding source

  • Study design: duration of ESA treatment (12 to 16 weeks; 16 to 24 weeks; 24 to 48 weeks; > 48 weeks); duration of follow-up (≥ 12 months, versus < 12 months); number of participants; date of publication.

In the event of important inconsistency, we will attempt to control for potential confounders using multiple-treatments meta-regression where data are sufficient and/or relaxing the consistency assumption and extend the model as described in Lu 2006.

Dealing with missing data

Any further information required from the original authors or sponsors of trials included in the review will be requested by written correspondence (e.g. emailing and/or writing to corresponding author/s) and any relevant information obtained in this manner will be included in the review. Data requested will include numbers of events and numbers of participants at risk for important dichotomous clinical outcomes (all-cause mortality, cardiovascular mortality, fatal or nonfatal stroke, fatal or nonfatal myocardial infarction, dialysis vascular access thrombosis, ESKD). We will also request additional information on the use of iron supplementation in treatment arms where this is not clear from reading the trial report.

Evaluation of important numerical data such as number of screened patients, number of randomised patients as well as intention-to-treat (ITT), as-treated, and per-protocol (PP) populations will be carefully performed. Attrition rates (e.g., drop-outs, losses to follow-up and withdrawals) will be investigated. Issues of missing data and imputation methods (e.g., last-observation-carried-forward (LOCF)) will be critically appraised (Higgins 2011).

Sensitivity analysis

In some trials, investigators may exclude the number of participants receiving red blood cell transfusions during the early phase of ESA treatment from the calculation of all events. We will conduct sensitivity analyses excluding trials in which blood transfusions from early phases of treatment were not included in trial results.

We will also perform sensitivity analyses in order to explore the influence of the following factors on effect size.

  • Repeating the analysis taking account of risk of bias, as specified above.

  • Repeating the analysis excluding any very long or large studies to establish how much they dominate the results.

  • Repeating the analyses based on the population included in the outcome evaluation e.g., intention-to-treat, as-treated, and per protocol populations.

Acknowledgements

We wish to acknowledge the support of the editorial office at the Cochrane Renal Group. In particular we wish to thank Narelle Willis (managing editor) and Ann Jones. We are also grateful to our specialist Trials Search Coordinator, Ruth Mitchell.

We wish to thank the referees for their comments and feedback during the preparation of this protocol

Appendices

Appendix 1. Electronic search strategies

DatabaseSearch terms
CENTRAL
  1. an*emi*:ti,ab,kw

  2. erythropoie*:ti,ab,kw

  3. epoietin:ti,ab,kw

  4. darbepo*etin:ti,ab,kw

  5. ("EPO" or "rhEPO"):ti,ab,kw

  6. "CERA":ti,ab,kw

  7. iron:kw

  8. (ferric or ferrous):kw,ti,ab

  9. "Ferrosoferric Oxide":kw

  10. 0ferumoxytol:kw,ti,ab

  11. (iron and (gluconate* or fumarate* or dextran* or sucrose* or saccharate*)):ti,ab

  12. (iron near/3 (supplement* or therap* or replacement)):ti,ab

  13. (magnetite or "ferriferous oxide"):kw,ti,ab

  14. hematinic*:ti,ab,kw

  15. (#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)

  16. (#1 AND #15)

MEDLINE
  1. Anemia/

  2. exp Anemia, Hypochromic/

  3. exp Anemia, Refractory/

  4. an?emi*.tw.

  5. or/1-4

  6. exp Erythropoietin/

  7. erythropoie*.tw.

  8. epo?etin.tw.

  9. darbepoetin.tw.

  10. EPO.tw.

  11. rhEPO.tw.

  12. CERA.tw.

  13. exp Ferric Compounds/

  14. exp Ferrous Compounds/

  15. Hematinics/

  16. Iron-Dextran Complex/

  17. Iron/

  18. Iron Compounds/

  19. Ferrosoferric Oxide/

  20. (iron adj5 (gluconate$ or fumarate$ or dextran$ or sucrose$ or saccharate$)).tw.

  21. (iron adj5 (supplement$ or therap$ or replacement)).tw.

  22. ((ferric or ferrous) adj5 gluconate$).tw.

  23. (ferumoxytol or magnetite or ferriferous oxide).tw.

  24. or/6-23

  25. 5 and 24

EMBASE

1. anemia/

2. iron deficiency anemia/

3. refractory anemia/

4. refractory anemia with excess blasts/

5. an?emi*.tw.

6. or/1-5

7. erythropoietin/

8. recombinant erythropoietin/

9. novel erythropoiesis stimulating protein/

10. erythropoie*.tw.

11. epo?etin.tw.

12. darbepoetin.tw.

13. EPO.tw.

14. rhEPO.tw.

15. CERA.tw.

16. iron/

17. iron therapy/

18. iron derivative/

19. exp antianemic agent/

20. (iron adj5 (supplement$ or therap$ or replacement)).ab.

21. (iron adj5 (gluconate$ or fumarate$ or dextran$ or sucrose$ or saccharate$)).tw.

22. ((ferric or ferrous) adj5 gluconate$).tw.

23. (ferumoxytol or magnetite or ferriferous oxide).tw.

24. or/7-23

25. 6 and 24

Appendix 2. Risk of bias assessment tool

Potential source of bias Assessment criteria

Random sequence generation

Selection bias (biased allocation to interventions) due to inadequate generation of a randomised sequence

Low risk of bias: Random number table; computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimisation (minimisation may be implemented without a random element, and this is considered to be equivalent to being random)
High risk of bias: Sequence generated by odd or even date of birth; date (or day) of admission; sequence generated by hospital or clinic record number; allocation by judgement of the clinician; by preference of the participant; based on the results of a laboratory test or a series of tests; by availability of the intervention
Unclear: Insufficient information about the sequence generation process to permit judgement

Allocation concealment

Selection bias (biased allocation to interventions) due to inadequate concealment of allocations prior to assignment

Low risk of bias: Randomisation method described that would not allow investigator/participant to know or influence intervention group before eligible participant entered in the study (e.g. central allocation, including telephone, web-based, and pharmacy-controlled, randomisation; sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes)
High risk of bias: Using an open random allocation schedule (e.g. a list of random numbers); assignment envelopes were used without appropriate safeguards (e.g. if envelopes were unsealed or non-opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure
Unclear: Randomisation stated but no information on method used is available

Blinding of participants and personnel

Performance bias due to knowledge of the allocated interventions by participants and personnel during the study

Low risk of bias: No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding; blinding of participants and key study personnel ensured, and unlikely that the blinding could have been broken
High risk of bias: No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding
Unclear: Insufficient information to permit judgement

Blinding of outcome assessment

Detection bias due to knowledge of the allocated interventions by outcome assessors.

Low risk of bias: No blinding of outcome assessment, but the review authors judge that the outcome measurement is not likely to be influenced by lack of blinding; blinding of outcome assessment ensured, and unlikely that the blinding could have been broken
High risk of bias: No blinding of outcome assessment, and the outcome measurement is likely to be influenced by lack of blinding; blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement is likely to be influenced by lack of blinding
Unclear: Insufficient information to permit judgement

Incomplete outcome data

Attrition bias due to amount, nature or handling of incomplete outcome data.

Low risk of bias: No missing outcome data; reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias); missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; missing data have been imputed using appropriate methods
High risk of bias: Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; ‘as-treated’ analysis done with substantial departure of the intervention received from that assigned at randomisation; potentially inappropriate application of simple imputation
Unclear: Insufficient information to permit judgement

Selective reporting

Reporting bias due to selective outcome reporting

Low risk of bias: The study protocol is available and all of the study’s pre-specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre-specified way; the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified (convincing text of this nature may be uncommon)
High risk of bias: Not all of the study’s pre-specified primary outcomes have been reported; one or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not pre-specified; one or more reported primary outcomes were not pre-specified (unless clear justification for their reporting is provided, such as an unexpected adverse effect); one or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta-analysis; the study report fails to include results for a key outcome that would be expected to have been reported for such a study
Unclear: Insufficient information to permit judgement

Other bias

Bias due to problems not covered elsewhere in the table

Low risk of bias: The study appears to be free of other sources of bias
High risk of bias: Had a potential source of bias related to the specific study design used; stopped early due to some data-dependent process (including a formal-stopping rule); had extreme baseline imbalance; has been claimed to have been fraudulent; had some other problem
Unclear: Insufficient information to assess whether an important risk of bias exists; insufficient rationale or evidence that an identified problem will introduce bias

Contributions of authors

  1. Draft the protocol: SP, GS, JC, GFMS

  2. Study selection: SP

  3. Extract data from studies: SP

  4. Enter data into RevMan: SP

  5. Carry out the analysis: SP, GS, DM

  6. Interpret the analysis: SP, DM, GS, JC, GFMS

  7. Draft the final review: SP

  8. Disagreement resolution: GS

  9. Update the review: SP

Declarations of interest

Suetonia Palmer receives a Fellowship from the Consorzio Mario Negri Sud from an unrestricted grant from Amgen Dompe, Italy. SP is a L'Oreal UNESCO For Women in Science Fellow Australia and New Zealand for 2012-2013.

Sources of support

Internal sources

  • Cochrane Renal Group, Australia.

External sources

  • Consorzio Mario Negri Sud, Italy.

    Suetonia Palmer receives a Fellowship from the Consorzio Mario Negri Sud from an unrestricted grant from Amgen Dompe

  • L'Oreal UNESCO For Women in Science Fellowship Australia and New Zealand, Australia.

    Suetonia Palmer is a 2012 For Women in Science Fellow

  • Georgia Salanti and Dimitris Mavridis receive research funding from the European Research Council Starting Grant (Grant Nr. IMMA 260559), Not specified.

Ancillary