Postnatal care by provider type and neonatal death in sub-Saharan Africa: a multilevel analysis

BMC Public Health

Table 3 Parameter estimates (odds ratio) for multilevel weighted logistic regression models of death on days 2 to 7 and death on days 2 to 28 by PNC type, maternal health factors and socio-demographic controls (n = 31,799) ^a

	Deaths in the first week						Deaths in the first month
Explanatory variables	OR	(95% CI)		OR	(95% CI)		OR	(95% CI)		OR	(95% CI)
Fixed effects
Post natal care
Visit on first day
None	1.00						1.00
Unskilled	0.87	(0.55, 1.38)					0.68	(0.48, 0.97)	*
Skilled	0.77	(0.43, 1.38)					0.79	(0.55, 1.15)
Visit within first week
None				1.00						1.00
Unskilled				0.34	(0.23, 0.52)	***				0.34	(0.30, 0.38)	***
Skilled				0.40	(0.18, 0.88)	*				0.51	(0.35, 0.75)	***
Ever breastfed
No	1.00			1.00			1.00			1.00
Yes	0.02	(0.01, 0.04)	***	0.02	(0.01, 0.04)	***	0.03	(0.02, 0.04)	***	0.03	(0.02, 0.04)	***
Delivery characteristics
Unskilled at home	1.00			1.00			1.00			1.00
Unskilled in a facility	0.21	(0.10, 0.41)	***	0.28	(0.12, 0.66)	**	0.68	(0.36, 1.29)		0.85	(0.48, 1.50)
Skilled at home	0.37	(0.09, 1.53)		0.39	(0.10, 1.54)		0.69	(0.35, 1.36)		0.72	(0.39, 1.35)
Skilled in a facility	1.20	(0.68, 2.13)		1.39	(0.76, 2.54)		0.95	(0.67, 1.35)		1.05	(0.72, 1.52)
Antenatal care
Tetanus toxoid before/during pregnancy
No	1.00			1.00			1.00			1.00
Yes	0.66	(0.45, 0.95)	*	0.71	(0.48, 1.04)		0.63	(0.46, 0.88)	**	0.67	(0.49, 0.93)	*
Antenatal visits
< 4	1.00			1.00			1.00			1.00
≥ 4	0.58	(0.43, 0.78)	***	0.59	(0.44, 0.79)	***	0.58	(0.45, 0.74)	***	0.58	(0.45, 0.76)	***
Random effects
Country-level variance (SE)^b	0.80	(0.67, 0.95)	*	0.74	(0.63, 0.88)	***	0.74	(0.61, 0.89)	**	0.70	(0.59, 0.82)	***
Log-likelihood		-1306.96			-1293.25			-1985.77			-1969.57
AIC		2631.91			2604.50			3991.53			3959.14
Log-likelihood ratio test (Chi-square)^c		498.20	***		490.10	***		580.31	***		569.25	***

Parameters for predictors (fixed effects) are reported as odds ratio; for random effects, the parameter is the variance.
*p < 0.05; **p < 0.01, ***p < 0.001.
^aFull model including all demographic control variables: age, parity, education, residence (urban/rural), and wealth quintile.
^bSignificance of random effects evaluated by comparing model with a similar one in which random effects have been constrained to be zero.
^cCompared to null model with no covariate.

ISSN: 1471-2458