Impact of an Intensive Perinatal Handwashing Promotion Intervention on Maternal Handwashing Behavior in the Neonatal Period: Findings from a Randomized Controlled Trial in Rural Bangladesh

Table 2

Handwashing behavior as measured by structured observations, by treatment arm, Matlab, Bangladesh, 2010-11.

	Control group	Intervention group	Risk difference	Adjusted risk difference^1	Risk ratio (95% CI)	Adjusted risk ratio^1 (95% CI)
# observations completed	106	112			n/a
Mothers
Mean number of handwashing events (with or without soap)	1.7 (SD 1.6)	2.5 (SD 1.9)	0.80 (0.32, 1.27)	0.82 (0.35, 1.29)	1.48 (1.22–1.79)	1.49 (1.24, 1.81)
Mean number of events of handwashing with soap	0.20 (SD 0.52)	0.81 (SD 1.2)	0.61 (0.37, 0.86)	0.62 (0.37, 0.87)	4.10 (2.55–6.59)	4.06 (2.53, 6.54)
Handwashing with soap at recommended times^2	2.3% (18/776)	9.1% (78/854)	0.07 (0.04, 0.10)	No convergence	3.94 (2.09, 7.44)	3.86 (2.05, 7.27)
Other household members^3
# observations completed	105	111			n/a
Mean number of events of handwashing with soap	0.06 (SD 0.23)	0.32 (SD 0.93)	0.26 (0.07, 0.44)	0.26 (0.08, 0.45)	5.52 (2.32, 13.12)	5.46 (2.30, 13.00)
Handwashing with soap at recommended times	1.2% (4/343)	6.1% (23/379)	0.05 (0.02, 0.08)	No convergence	5.20 (1.80, 15.09)	5.58 (1.94, 16.11)^4

for mother’s education and water source; referent = control.
²Recommended times for handwashing: after respiratory secretion contact, before umbilical cord care, before breastfeeding, after fecal contact, and before food preparation.
³In one household in each arm, only the mother was observed during the SO.
⁴Log Poisson model used in place of log binomial model due to lack of convergence of model. Log Poisson given consistent but less efficient estimates of RR compared to log binomial models.