Risk of bias for analysis 1.1 Risk of necrotising enterocolitis stage ≥ 2.
| Study | Bias | |||||||||||
| Randomisation process | Deviations from intended interventions | Missing outcome data | Measurement of the outcome | Selection of the reported results | Overall | |||||||
| Authors' judgement | Support for judgement | Authors' judgement | Support for judgement | Authors' judgement | Support for judgement | Authors' judgement | Support for judgement | Authors' judgement | Support for judgement | Authors' judgement | Support for judgement | |
| Kaur 2015 | Low risk of bias | A computer‐generated block randomisation sequence with block size of 4 was prepared by a person not involved in clinical care, measurement of outcomes, or analysis of data. This randomisation sequence was kept in sequentially numbered sealed opaque envelopes. However, a fixed block size of 4 gives the chance to guess the allocation of every fourth infant in an unmasked study. | Some concerns | No data to assess if deviations arouse because of trial context | Low risk of bias | All 80 randomised infants were included in the analysis | Low risk of bias | Though masking was not done, NEC is an objective outcome and is less prone to detection bias. | Low risk of bias | All proposed outcomes were reported (personal communication) | Some concerns | Some concerns in one domain, low risk in other domains |
| Parker 2019 | Low risk of bias | Infants were randomized to 1 of 2 groups by a computer generated sequence with random‐length permuted blocks of sizes (4, 6, or 8). Randomization was concealed until the intervention was assigned. | Low risk of bias | Part 1: Eighteen (26%) infants in the "no gastric residual" group had one or more gastric residuals evaluated. However, no infant had gastric residuals evaluated for more than one day and hence this deviation was considered unlikely to have affected the outcome. Part 2: Modified intention to treat (mITT) was used (Low risk) |
Low risk of bias | All randomised neonates were accounted for. | Low risk of bias | Though masking was not done, NEC is an objective outcome and is less prone to detection bias. | Low risk of bias | All proposed outcomes were reported | Low risk of bias | Low risk in all domains |
| Thomas 2018 | Low risk of bias | Randomization was completed using a computer‐generated random number table in unequal block sizes ranging from 4 to 12. Allocation concealment was done using sequentially numbered opaque sealed envelopes | Low risk of bias | Part 1: There were no deviations from intended intervention (Low risk) Part 2: modified intention to treat analysis was used (Low risk) |
Low risk of bias | All randomised neonates were accounted for. | Low risk of bias | Though masking was not done, NEC is an objective outcome and is less prone to detection bias. | Some concerns | Study protocol had not been published | Some concerns | Some concerns in one domain and low risk in other domains |
| Torrazza 2015 | Low risk of bias | A computer‐generated block randomisation sequence with variable block sizes was used. The randomisation sequence was kept in sequentially numbered sealed opaque envelopes (personal communication) | Some concerns | No data to assess if deviations arouse because of trial context. | Low risk of bias | All 61 randomised infants were included in the analysis. | Low risk of bias | Though masking was not done, NEC is an objective outcome and is less prone to detection bias. | Low risk of bias | Study protocol had been published. All proposed outcomes were reported. | Some concerns | Some concerns in one domain, low risk in other domains |