Table 4.

Umbilical cord infection by treatment group stratified by timing of intervention and skin cleansing trial intervention

	Timing of intervention
	Early intervention (• 24 h)	Late intervention (• 24 h)
Algorithm 1: Moderate or severe redness
Chlorhexidine 4.0%	0.65 (0.56−0.78)	0.70 (0.49−1.01)
Soap/water	1.00 (0.84−1.18)	1.32 (0.93−1.87)
Algorithm 2: Moderate or severe redness, with pus, or severe redness alone
Chlorhexidine 4.0%	0.42 (0.32−0.54)	0.66 (0.40−1.09)
Soap/water	0.82 (0.64−1.06)	1.19 (0.79−1.81)
Algorithm 3: Severe redness with pus
Chlorhexidine 4.0%	0.13 (0.06−0.31)	1.06 (0.35−3.20)
Soap/water	1.00 (0.55−1.82)	1.12 (0.37−3.36)

	Skin cleansing trial intervention
	Chlorhexidine 0.25%	Placebo
Algorithm 1: Moderate or severe redness
Chlorhexidine 4.0%	0.71 (0.57−0.89)	0.64 (0.51−0.82)
Soap/water	1.00 (0.78−1.27)	1.07 (0.84−1.37)
Algorithm 2: Moderate or severe redness, with pus, or severe redness alone
Chlorhexidine 4.0%	0.49 (0.35−0.67)	0.43 (0.30−0.62)
Soap/water	0.86 (0.62−1.18)	0.90 (0.62−1.30)
Algorithm 3: Severe redness with pus
Chlorhexidine 4.0%	0.37 (0.16−0.86)	0.12 (0.03−0.58)
Soap/water	1.16 (0.56−2.38)	0.86 (0.36−2.10)

Data are IRR (95% CI) Generalised estimating equations (Poisson distribution, log link function) were used to adjust point estimates and CIs for the cluster-randomised design.