TABLE 3—

Longitudinal Association Between Exposure to Breastfeeding Television Spots and Exclusive Breastfeeding Practice at Commune Level by Intervention Area: Vietnam, 2011–2014

	A&T-NI Communes (n = 61), b (95% CI)		A&T-I Communes (n = 57), b (95% CI)
Variable	Model 1	Model 2	Model 3	Model 4
Constant	13.3 (4.1, 22.5)	20.4 (14.8, 26.0)	17.2 (–3.6, 38.0)	41.1 (33.3, 48.9)
Baseline EBF prevalence	0.349 (0.135, 0.563)	0.344 (0.132, 0.555)	0.493 (0.249, 0.737)	0.510 (0.269, 0.752)
Messages recalled, no.	7.97 (–1.49, 17.40)		16.70 (5.01, 28.50)
Message recalled, tercile
2nd		2.49 (–4.40, 9.39)		−0.71 (–11.20, 9.80)
3rd (highest)		8.65 (–0.12, 17.40)		13.50 (4.00, 23.00)

Note. A&T-I = Alive & Thrive-intensive; A&T-NI = Alive & Thrive-nonintensive; CI = confidence interval; EBF = exclusive breastfeeding. Bootstrapping (with 10 000 replications) linear regression models at the commune level regressed EBF prevalence in rounds 2–5 on exposure to the television spots, adjusted for EBF prevalence at baseline. We defined EBF as feeding breastmilk exclusively in the previous 24 h (i.e., no foods or liquids except medications such as drops or syrups).