Types of studies

We will include randomised or quasi‐randomised controlled trials. We will not include cluster trials. We will not apply any language restrictions.

Types of participants

1. Preterm infants (less than 35 weeks PMA).

2. Evidence of peripartum hypoxia, with each enrolled infant satisfying at least one of the following criteria:

   1. Apgar score of 5 or less at 10 minutes;

   2. mechanical ventilation or resuscitation at 10 minutes;

   3. cord pH < 7.1, or an arterial pH < 7.1 or base deficit of 12 or more within 60 minutes of birth.

3. A physical exam with evidence of hypotonia or lethargy or seizures indicative of evolving HIE.

4. No major congenital abnormalities recognisable at birth.

Types of interventions

1. Doses of EPO:

   1. low doses of EPO (500 IU/kg/week to 1000 IU/kg/week);

   2. high doses of EPO (1001 IU/kg/week to 3000 IU/kg/week);

   3. very high doses of EPO (> 3000 IU/kg/week).

2. EPO will be administered subcutaneously (or intravenously) either daily or every other day.The initiation of EPO (using high or very high doses) is used as the treatment group, while low doses or placebo or no intervention are used as the control group.

3. Timing of starting intervention (< 6 hours, 6 to 48 hours, > 48 hours and < 72 hours).

Types of outcome measures

Selection of studies

We will use the standard review methods of the Cochrane Neonatal Review Group.

Two review authors (Zhangbin Yu, Junjie Lu) will separately assess all abstracts and published studies identified by the literature search as potentially relevant for inclusion in the review. We will resolve any disagreement by discussion.

Data extraction and management

Two review authors (Zhangbin Yu, Junjie Lu) will separately extract, assess and code all data for each study using a form designed specifically for this review. We will replace any standard error of the mean by the corresponding standard deviation. We will resolve any disagreement by discussion. For each study, final data will be entered into RevMan 5.1 (RevMan 2011) by one review author (Zhangbin Yu) and then checked by a second review author (Junjie Lu).

Assessment of risk of bias in included studies

We will use the standard criteria and methods of the Cochrane Collaboration and the Cochrane Neonatal Review Group for assessing the risk of bias of included trials (Higgins 2011). Each study reviewed in full will be assessed and data initially abstracted independently by two of the review authors (Zhangbin Yu, Junjie Lu). For each included study, inclusion criteria will be confirmed, and information will be sought regarding the study design, method of randomisation, masking of allocation, masking of intervention, completeness of follow‐up and masking of outcome assessment. Discrepancies will be resolved by discussion and consensus.This information will be included in the table 'Characteristics of Included Studies'.

In addition, we will evaluate the following information and add this to the Risk of Bias table: 

1. Random sequence generation (selection bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. There is a low risk of selection bias if a random (unpredictable) assignment sequence was used. Examples of low risk methods are: referring to a random number table, using a computer random number generator, coin tossing, shuffling cards or envelopes, throwing dice, drawing of lots, minimization (minimization may be implemented without a random element, and this is considered to be equivalent to being random).

   2. Criteria for the judgement of ‘High risk’ of bias. The investigators describe a non‐random component in the sequence generation process. Usually, the description would involve some systematic, non‐random approach, for example: Sequence generated by odd or even date of birth; date (or day) of admission; hospital or clinic record number. Other non‐random approaches happen much less frequently than the systematic approaches mentioned above and tend to be obvious.  They usually involve judgement or some method of non‐random categorization of participants, for example: Allocation by judgement of the clinician; preference of the participant; the results of a laboratory test or a series of tests; availability of the intervention.

   3. Criteria for the judgement of ‘Unclear risk’ of bias. Insufficient information about the sequence generation process to permit judgement of ‘Low risk’ or ‘High risk’.

2. Allocation concealment (selection bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. There is a low risk of selection bias if the assignment was generated by an independent person not responsible for determining the eligibility of the patients. This person had no information about the persons included in the trial, no influence on the assignment sequence or eligibility of the patient, and could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web‐based, and pharmacy‐controlled, randomisation); sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, or sealed envelopes.

   2. Criteria for the judgement of ‘High risk’ of bias. Examples of high risk methods are: 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, or other explicitly unconcealed procedure. Knowledge of the allocated interventions was adequately prevented during the study (performance bias).

   3. Criteria for the judgement of ‘Unclear risk’ of bias. Insufficient information to permit judgement of ‘Low risk’ or ‘High risk’. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement – for example if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially numbered, opaque and sealed.

3. Blinding of participants and personnel (performance bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. Any one of the following: 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.

   2. Criteria for the judgement of ‘High risk’ of bias. Any one of the following: 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.

   3. Criteria for the judgement of ‘Unclear risk’ of bias. Any one of the following: Insufficient information to permit judgement of ‘Low risk’ or ‘High risk’; the study did not address this outcome.

4. Blinding of outcomes assessment (detection bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. Any one of the following: 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.

   2. Criteria for the judgement of ‘High risk’ of bias. Any one of the following: 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.

   3. Criteria for the judgement of  ‘Unclear risk’ of bias. Any one of the following: Insufficient information to permit judgement of ‘Low risk’ or ‘High risk’; the study did not address this outcome.

5. Incomplete outcome data (attrition bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. Any one of the following: 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.

   2. Criteria for the judgement of ‘High risk’ of bias. Any one of the following: 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 standardized 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.

   3. Criteria for the judgement of ‘Unclear risk’ of bias. Any one of the following: Insufficient reporting of attrition/exclusions to permit judgement of ‘Low risk’ or ‘High risk’ (e.g. number randomised not stated, no reasons for missing data provided); the study did not address this outcome.

6. Selective reporting (reporting bias)

   1. Criteria for the judgement of ‘Low risk’ of bias. Any of the following: 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).

   2. Criteria for the judgement of ‘High risk’ of bias. Any one of the following: 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.

   3. Criteria for the judgement of ‘Unclear risk’ of bias. Insufficient information to permit judgement of ‘Low risk’ or ‘High risk’. It is likely that the majority of studies will fall into this category.

7. Other sources of bias:

   1. Criteria for the judgement of ‘Low risk’ of bias. The study appears to be free of other sources of bias;

   2. Criteria for the judgement of ‘High risk’ of bias. There is at least one important risk of bias. For example, the study: had a potential source of bias related to the specific study design used; or has been claimed to have been fraudulent; or had some other problem;

   3. Criteria for the judgement of ‘Unclear risk’ of bias. There may be a risk of bias, but there is either: insufficient information to assess whether an important risk of bias exists; or insufficient rationale or evidence that an identified problem will introduce bias.

Measures of treatment effect

We will perform statistical analyses using Review Manager software. We will express dichotomous data as risk ratios (RR) with their 95% confidence intervals (CI). We will calculate the risk difference (RD) and their 95% CI if the RD was statistically significant and we will convert the RD into the number needed to treat to benefit (NNTB) or the number needed to treat to harm (NNTH). We will analyse continuous data using mean difference (MD). We will report the 95% CI on all estimates.

Dealing with missing data

We will contact authors of trials for missing data, or if clarification is required. If contact is impossible, then we will exclude the study from synthesis. There may be other missing information such as method of imputation for intention‐to‐treat analyses which should be reported in the article, but if not, we will contact the author.

Assessment of heterogeneity

We will perform heterogeneity tests. We will report the results of the I² statistic. The degree of heterogeneity estimates will be assessed as: no heterogeneity < 25%; low heterogeneity 25% to 49%; moderate 50% to 74%; high heterogeneity 75%. If we detect statistical heterogeneity, we will explore the possible causes (for example, differences in study quality, participants, intervention regimens, or outcome assessments) using post hoc subgroup analyses.

Assessment of reporting biases

We will investigate reporting biases (such as publication bias) using funnel plots if there are ten or more studies in the meta‐analysis. We will assess funnel plot asymmetry visually, and using formal tests. For continuous outcomes we will use the test proposed by Egger 1997, and for dichotomous outcomes we will use the test proposed by Harbord 2006. If we detect asymmetry in any of these tests or by a visual assessment, assessment we will perform exploratory analyses to investigate.

Data synthesis

We will summarise data statistically if they are available, sufficiently similar and of sufficient quality. A meta‐analysis may be feasible if more than one eligible trial is identified and there is sufficient homogeneity among the studies with respect to participants and reported outcomes. We will perform statistical analysis according to the statistical guidelines referenced in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

We will use the standard methods of the Cochrane Neonatal Review Group, using a fixed‐effect model. In assessing the treatment effects for dichotomous outcomes we will use the RR and RD, with 95% CI. For outcomes measured on a continuous scale we will use the MD, with 95% CI.

Subgroup analysis and investigation of heterogeneity

We plan to conduct subgroup analyses on the basis of:

1. The dose of EPO:

   1. placebo or standard supportive treatment versus low doses of EPO (> 500 IU/kg/week, ≤ 1000 IU/kg/week);

   2. placebo or standard supportive treatment versus high (> 1000 IU/kg/week, ≤ 3000 IU/kg/week) doses of EPO;

   3. placebo or standard supportive treatment versus very high (> 3000 IU/kg/week) doses of EPO.

2. Severity of HIE (mild versus moderate versus severe) according to Sarnat staging (Sarnat 1976; Finer 1981).

3. Inclusion criteria: electrophysiological plus clinical criteria versus clinical criteria alone.

4. Postmenstrual age (PMA): PMA < 30 weeks versus PMA ≥ 30 weeks.

Sensitivity analysis

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

1. repeating the analysis excluding unpublished studies;

2. repeating the analysis taking account of risk of bias in included studies, as specified above (see Assessment of risk of bias in included studies).