Skip to main content
. 2017 Oct 4;14(10):e1002398. doi: 10.1371/journal.pmed.1002398

Table 15. Treatment effects by subgroups according to the time from the start of treatment until birth.

Outcome Time (hours) Included trials MgSO4 Control RR LCL UCL p
Death or CP 0 to <4 9,10,11,12 141/808 (17.5%) 157/822 (19.1%) 0.91 0.74 1.13 0.11
4 to <8 9,10,12 35/195 (17.9%) 27/191 (14.1%) 1.24 0.77 1.99
8 to <12 9,10,12 15/95 (15.8%) 29/84 (34.5%) 0.49 0.27 0.87
12 to <24 9,10,12 49/193 (25.4%) 45/182 (24.7%) 1.00 0.70 1.42
24 to <36 9,12 25/70 (35.7%) 15/57 (26.3%) 1.34 0.78 2.29
36+ 9,11,12 117/446 (26.2%) 116/394 (29.4%) 0.92 0.73 1.14
Death (at any time) 0 to <4 9,10,11,12 101/808 (12.5%) 104/822 (12.7%) 1.01 0.77 1.32 0.23
4 to <8 9,10,12 26/195 (13.3%) 20/191 (10.5%) 1.30 0.73 2.31
8 to <12 9,10,12 15/95 (15.8%) 26/84 (31.0%) 0.55 0.31 1.00
12 to <24 9,11,12 43/193 (22.3%) 41/182 (22.5%) 0.96 0.66 1.40
24 to <36 9,12 24/70 (34.3%) 14/57 (24.6%) 1.38 0.79 2.41
36+ 9,11,12 112/446 (25.1%) 110/394 (27.9%) 0.93 0.74 1.17
Cerebral palsy 0 to <4 9,10,11 40/558 (7.2%) 53/551 (9.6%) 0.75 0.50 1.12 0.77
4 to <12 9,10,12 9/186 (4.8%) 10/162 (6.2%) 0.71 0.30 1.68
12+ 9,10,11,12 13/320 (4.1%) 11/282 (3.9%) 1.13 0.47 2.68
Death or major neurosensory disability 0 to <4 9,10,11,12 164/808 (20.3%) 173/822 (21.0%) 1.00 0.82 1.21 0.24
4 to <8 9,10,12 44/195 (22.6%) 30/191 (15.7%) 1.32 0.86 2.03
8 to <12 9,10,12 20/95 (21.1%) 30/84 (35.7%) 0.64 0.39 1.05
12 to <24 9,11,12 52/193 (26.9%) 48/182 (26.4%) 0.99 0.70 1.39
24 to <36 9,12 29/70 (41.4%) 18/57 (31.6%) 1.27 0.79 2.03
36+ 9,11,12 126/446 (28.3%) 125/394 (31.7%) 0.94 0.76 1.16

Abbreviations: CP, cerebral palsy; LCL, lower confidence limit; RR, relative risk; UCL, upper confidence limit. The trials included are as follows: 9 = Australasian Collaborative Trial of Magnesium Sulphate (ACTOMgSO4), 10 = PREterm brain protection by MAGnesium sulphate (PREMAG), 11 = Magnesium and Neurologic Endpoints Trial (MAGNET), and 12 = MAGnesium sulphate for Prevention of Eclampsia (MAGPIE).

p-values for heterogeneity (The heterogeneity p-value for 1-stage analyses is from Wald chi-square tests for the interaction between treatment and subgroup in a generalising estimating equation [GEE] model).