Table 4. Adjusted odds ratios (95% CI) for the association between clean delivery kit use and hand washing, with maternal mortality obtained from logistic regression models with and without multiple imputation.

Clean delivery practices	Model type	Pooled data		India		Bangladesh		Nepal
		AOR (95% CI)	p a	AOR (95% CI)	p a	AOR (95% CI)	p a	AOR (95% CI)	p a
Use of clean delivery kit	Logistic regression b , e	1.26 (0.62–2.56)	0.519	0.66 (0.15–2.93)	0.587	1.61 (0.71–3.68)	0.256	f
	Multiple imputation c , e	1.18 (0.62–2.24)	0.612	0.68 (0.15–2.99)	0.605	1.45 (0.63–3.30)	0.381	f
Washing hands prior to delivery	Logistic regression b	0.51 (0.28–0.93)	0.028	0.15 (0.02–1.11)	0.063	0.57 (0.27–1.23)	0.154	0.83 (0.19–3.56)	0.800
	Multiple imputation c , d	0.48 (0.26–0.90)	0.022	0.15 (0.02–1.13)	0.066	0.58 (0.27–1.25)	0.162	0.91 (0.23–3.65)	0.898

^a Wald test.

^b Adjusted for maternal age, maternal education, parity, number of antenatal care visits, household assets, and for the pooled analysis, study site.

^c Multiple imputation models taking into account variables describe in b, as well as predictors of missingness including obstetric haemorrhage, and skilled birth attendant

^d Multiple imputation models also included clean delivery kit use as predictor of missingness.

^e It was not possible to include Nepal in the pooled analysis of kit use due to convergence issues caused by large numbers of missing/unknown data.

^f Model would not converge due large number of deliveries with missing/unknown data on kit use