Table 3. Association of socio-demographic characteristics with HIV incidence during pregnancy.

Variable	Model 1: Penalized maximum likelihood logistic regression a (n = 1595 c)	Model 2: Poisson regression b (n = 1595 c)
	HIV incidence during pregnancy
	OR (95% CI), P-value	IRR (95% CI), P-value
Age d	0.98 (0.89–1.07), 0.645	0.97 (0.89–1.07), 0.578
Marital status d
Married or cohabiting (reference)	-	-
None cohabiting couple	6.22 (1.12–34.50), 0.037	8.78 (1.13–68.33), 0.038
Single, widowed or divorced	4.43 (0.59–33.10), 0.147	5.64 (0.55–58.38), 0.147
Education d
Primary or less (reference)	-	-
Secondary or more	0.27 (0.07–1.10), 0.067	0.23 (0.05–1.07), 0.061
Wealth score d
Low SES 0–3 (reference)	-	-
High SES 4–9	0.45 (0.15–1.31), 0.142	0.43 (0.14–1.31), 0.139
Partner HIV status d
Known negative (reference)	-	-
Known positive, known and not specified or unknown	1.90 (0.67–5.40), 0.231	1.82 (0.63–5.29), 0.270
Wald Chi-squared, P-value	14.86, 0.0214	-
LR Chi-squared, P-value	-	20.79, 0.0020
Deviance goodness of fit, P-value	-	119.20, 1.0000 f
Pearson goodness of fit, P-value	-	1661.09, 0.0987 f

^a Penalized maximum likelihood logistic regression is intended for rare events.

^b Poisson regression is intended for comparing rates of rare events.

^c Observations with missing values were excluded for all categorical or binary variables in the two regression models.

^d The two models are showing matching results.

^f P-value is showing the model to be a good fit.

OR, Odds ratio; CI, Confidence interval; IRR, Incidence rate ratio.