Kochan 2019.

Study characteristics
Methods	Randomised parallel clinical trial, conducted at the Children’s Hospital of The King’s Daughters in Norfolk, Virginia (USA) and including 180 extremely low birth weight infants in 2012 to 2015. Block randomised using a block randomisations table.
Participants	Extremely low birth weight infants (< 1000 g) admitted to the NICU Baseline characteristics Mean (SD) gestational age was 25.9 weeks (1.7) and 25.4 (1.6) in the midline and control group, respectively (P = 0.06) Mean (SD) birth weight was 739 g (170) and 726 (150) in the midline and control group, respectively (P = 0.58) The incidence of pre‐eclampsia was higher in the midline group (36/90, 40%) than control (22/90, 24%) (P = 0.026) The incidence of prolonged rupture of membranes was lower in the midline group (9/90, 10%) than control (21/90, 23%) (P = 0.018) Multivariate analysis of maternal pre‐eclampsia and prolonged rupture of membranes showed no significant effect on outcomes. Infants with congenital anomalies were excluded.
Interventions	Infants admitted to the NICU were placed supine in the incubator while umbilical lines were placed and radiographs obtained. Midline: supine, elevation of the head/body 30 degrees above the horizontal with the head maintained in the midline) for the first 4 days of life. Control: supine/flat with turning of the head 90 degrees right or left every 4 h) Cranial US were performed immediately after randomisations by a mean (SD) age of 2.9 h (1); there was no significant difference in the age at the time of admission between the two study groups (P = 0.35). Follow‐up ultrasounds were performed daily for the first 4 days of life and on day 7. Normal scans were repeated at 4 weeks of age, abnormal scans were repeated weekly. All cranial ultrasound studies were interpreted by paediatric radiologists all blinded to the randomisations group. Outcome data were reported according to intention‐to‐treat analysis. Protocol variations were reported in 26 infants in the midline group as follows. 19 infants had brief periods of flat positioning: 7 for reintubation 6 for repeated surfactant administration 6 for peripherally inserted central catheter placement 1 infant had a respiratory deterioration associated with a pulmonary haemorrhage 1 infant underwent abdominal drain placement 1 infant was kept in the flat position when a pulmonary haemorrhage occurred on day 3 of life 4 infants were placed flat when respiratory support had to be increased to high‐frequency ventilation (2 on day 2 of life, 2 on day 3 of life) After the fourth day, all infants were placed in the flat position.
Outcomes	Primary outcome: incidence of GM‐IVH any grade Secondary outcomes: severe IVH; survival on the first 4 days and to discharge; cystic periventricular leukomalacia on brain ultrasound in the first month of life; retinopathy of prematurity requiring treatment
Notes	The nursing staff documented protocol variations in 26 infants randomised to the ELEV group. Data from all infants were maintained in their original randomisations group
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Infants were randomized using a block randomizations table"
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not feasible for the nature of the intervention
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote. "All cranial ultrasound studies were interpreted by one of five pediatric radiologists all blinded to the randomizations group". All main outcomes likely to be assessed by physicians unaware of the intervention.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcomes were reported for all randomised infants
Selective reporting (reporting bias)	Unclear risk	The protocol for the study was not registered on an online database (information provided by study authors). We cannot judge whether there were any deviations from the protocol
Other bias	Low risk	Appears free of other bias