Associations between nighttime or weekend deliveries and adverse maternal, birth, and neonatal outcomes: secondary analysis of the MANGO study in western Kenya

Table 4 The association between nighttime/weekend deliveries and adverse neonatal outcome of neonatal death, MANGO study, September 2020-November 2023 (n = 25,911 neonates)

Characteristics	Adverse outcome	No adverse outcome	Unadjusted RR (95% CI)¹	p-value	Adjusted RR (95% CI)²	p-value
Nighttime/weekend delivery
Yes	100 (0.63)	15675 (99.37)	0.92 (0.67. 1.26)	0.61	0.96 (0.70, 1.30)	0.79
No	65 (0.69)	9390 (99.31)	Ref.		Ref.
Poor apgar score (< 7 at 5 min.)
Yes	123 (13.53)	786 (86.47)	92.7 (64.06, 134.15)	< 0.001	68.00 (36.77, 125.75)	< 0.001
No	35 (0.15)	23943 (99.85)	Ref.		Ref.
NICU admission
Yes	80 (4.56)	1673 (95.44)	13.39 (9.83, 18.25)	< 0.001	1.18 (0.70, 2.00)	0.52
No	77 (0.34)	22525 (99.66)	Ref.		Ref.
Parity
0	63 (0.63)	10008 (99.37)	1.24 (0.83, 1.87)	0.29	0.97 (0.64, 1.47)	0.9
1	36 (0.50)	7132 (99.50)	Ref.		Ref.
2	26 (0.60)	4275 (99.40)	1.20 (0.73, 1.99)	0.47	1.02 (0.65, 1.68)	0.95
3 or more	40 (1.09)	3635 (98.91)	2.17 (1.38, 3.39)	< 0.001	1.81 (1.17, 2.80)	0.01
Gravidity³
1	57 (0.60)	9476 (99.40)	1.14 (0.76, 1.73)	0.52
2	37 (0.52)	7038 (99.48)	Ref.
3	28 (0.65)	4311 (99.35)	1.23 (0.76, 2.01)	0.39
4 or more	43 (1.01)	4221 (98.99)	1.93 (1.24, 2.99)	0.003
Maternal HIV status
Yes	9 (0.91)	978 (99.09)	1.42 (0.73, 2.77)	0.31
No	156 (0.64)	24087 (99.36)	Ref.
Maternal transfer-in status
Yes	45 (2.70)	1623 (97.30)	5.32 (3.79, 7.47)	< 0.001	2.23 (1.58, 3.15)	< 0.001
No	119 (0.51)	23344 (99.49)	Ref.		Ref.

Abbreviations: RR = risk ratio; CI = confidence interval; NICU = neonatal intensive care unit
Bold text indicates estimates with p-values < 0.05.
¹ The unadjusted models were modeled for each variable separately for the outcome using Poisson regression with natural logarithm (log) as the link function, robust standard errors, and clustered on pregnant persons for multiple gestations.
² The adjusted model included the primary exposure (nighttime/weekend delivery) and any covariates with p-value < 0.20 in the unadjusted models, and used Poisson regression with natural log link, robust standard errors, and clustered on pregnant persons for multiple gestations.
³ Though parity and gravidity each have p-values < 0.20 in the unadjusted models, we included only parity in the adjusted model based on presumed collinearity between the two variables

ISSN: 1471-2393