Primary outcomes

• All‐cause neonatal mortality (during initial hospital stay and up to 28 days' postnatal age)

Secondary outcomes

• Sepsis‐related mortality (death attributed to sepsis or its complications, including disseminated intravascular coagulation, refractory shock, multiorgan dysfunction, severe pneumonia, necrotizing enterocolitis, etc.) up to 28 days' postnatal age

• Duration of hospital stay (days)

• Adverse effects of BET

  • Thrombocytopenia (defined as platelet count < 150,000/mm³) (Chavda 2012)

  • Hypocalcemia (defined as total serum calcium concentration < 7 mg/dL, or as an ionized calcium concentration < 4 mg/dL (1 mmol/L)) (Abrams 2008)

  • Necrotizing enterocolitis (NEC) (defined as an acute gastrointestinal disorder that manifests clinically with systemic signs (temperature instability, apnea, bradycardia, lethargy, hypotension, metabolic acidosis, hyponatremia, thrombocytopenia, disseminated intravascular coagulation, etc.), intestinal signs (feed intolerance, gastrointestinal bleeding, abdominal tenderness, abdominal wall cellulitis, abdominal distention, etc.), and radiological features (nonspecific intestinal dilation and ileus in stage I, pneumatosis intestinalis and air in the portal tree in stage II, or pneumoperitoneum in stage III), and pathologically with intestinal necrosis} (Bell 1978; Walsh 1986)

  • Metabolic acidosis (defined as arterial or capillary pH < 7.25 and base excess worse than ‐6 mmol/L)

  • Air embolism (diagnosed as cutaneous blanching, air in intravascular catheters, an air‐filled heart and vessels on chest roentgenogram, etc.) (Carlo 2011)

  • Cardiac arrhythmia (defined as abnormal rhythm on electrocardiogram)

  • Hypotension (defined as blood pressure < 10th percentile that is appropriate for gestation and postnatally) (Nuntnarumit 1999)

  • Desaturation episode (defined as oxygen saturation as measured by pulse oximeter < 85%)

  • Adverse effects not predefined by us but reported by study authors

• Duration of antibiotic therapy in days (counted from the day antibiotics were started for the sepsis episode for which BET was done up to the day when antibiotics were stopped because the infant was declared sepsis free or the infant died)

• Long‐term neurological morbidity (long‐term outcome will be reported for all studies that have evaluated children between 18 and 24 months' or later postnatal age. Major neurological morbidity will include cerebral palsy and developmental delay (Baley or Griffith assessment > 2 standard deviations (SDs) below the mean) or intellectual impairment (intelligence quotient (IQ) > 2 SDs below the mean), blindness (vision < 6/60 in both eyes), and sensorineural deafness (requiring amplification)

Sequence generation (selection bias)

For each study, we will assess the method of sequence generation as:

• low risk: when sequences were generated using a true randomisation process;

• high risk: when any nonrandom process was used for sequence generation; or

• unclear risk: when the method of sequence generation is unclear.

Allocation concealment (selection bias)

For each study, we will assess the method of allocation concealment as:

• low risk: when adequate methods were used for allocation concealment (eg, consecutively numbered sealed, opaque envelopes);

• high risk: when allocation concealment was inadequately done; or

• unclear risk: when the method of allocation concealment is unclear.

Blinding (performance bias and detection bias)

For each study, we will categorize the methods used to blind study participants and personnel from knowledge of which intervention a participant received. We will assess the process of blinding separately for different outcomes and will categorize methods as:

• low risk, high risk or unclear risk for participants;

• low risk, high risk or unclear risk for personnel; and

• low risk, high risk or unclear risk for outcome assessors.

Incomplete outcome data (attrition bias)

For each study, we will examine the completeness of data including attrition and exclusion from analysis. We will note whether attrition and exclusions were reported, numbers included in the analysis at each stage (compared with the total number of randomized participants), whether reasons for attrition or exclusion were reported, and whether missing data were balanced across groups or were related to outcomes. We will categorize the method as:

• low risk: < 20% missing data;

• high risk: ≥ 20% missing data; or

• unclear risk.

Selective reporting (reporting bias)

For each study, we will investigate the possibility of selective outcome reporting and will categorize the method as:

• low risk: when all prespecified study outcomes and expected outcomes of interest were reported;

• high risk: when not all prespecified study outcomes were reported or one or more primary outcomes were not prespecified; or

• unclear risk: when information provided is unclear.

Other source of bias

For each included study, we will describe our concerns about other possible sources of bias. On the basis of potential sources of bias, we will describe the included study as:

• low risk;

• high risk; or

• unclear risk.

Quality of evidence

We will use the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, as outlined in the GRADE Handbook (Schünemann 2013), to assess the quality of evidence for the following (clinically relevant) outcomes.

• All‐cause neonatal mortality (during initial hospital stay and up to 28 days' postnatal age).

• Sepsis‐related mortality (death attributed to sepsis or its complications, including disseminated intravascular coagulation, refractory shock, multiorgan dysfunction, severe pneumonia, necrotizing enterocolitis, etc.) up to 28 days' postnatal age.

• Duration of hospital stay (days).

• Long‐term neurological morbidity (long‐term outcome will be reported for all studies that have evaluated children between 18 and 24 months' or later postnatal age). Major neurological morbidity will include cerebral palsy and developmental delay (Baley or Griffith assessment > 2 SDs below the mean) or intellectual impairment (IQ > 2 SDs below the mean), blindness (vision < 6/60 in both eyes), and sensorineural deafness (requiring amplification).

Two review authors will independently assess the quality of the evidence for each of the outcomes above. We will consider evidence from randomized controlled trials as high quality but will downgrade the evidence one level for serious (or two levels for very serious) limitations on the basis of the following: design (risk of bias), consistency across studies, directness of the evidence, precision of estimates and presence of publication bias. We will use the GRADEpro Guideline Development Tool to create a "Summary of findings" table to report the quality of the evidence.

The GRADE approach results in an assessment of the quality of a body of evidence according to one of four grades.

• High: We are very confident that the true effect lies close to that of the estimate of effect.

• Moderate: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of effect but may be substantially different.

• Low: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of effect.

• Very low: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.