Table 1b.

Duration of any breastfeeding and infections in the first 18 months of life in a sibling and full cohort analysis.

Endpoint	Model	Number of children with endpoint (%)	Odds ratio (95% CI) per month before introduction of complementary foods
Hospitalisation for any infection	Siblings matched (n=13,420)	1,107 (8.3)	0.994 (0.965–1.024)
	Full cohort (n=70,487)	5,669 (8.0)	0.985 (0.979–0.992)
Any lower respiratory tract infection	Siblings matched (n=13,043)	1,854 (14.2)	0.999 (0.973–1.026)
	Full cohort (n=68,322)	9,030 (13.2)	0.990 (0.985–0.995)
Any gastro-enteritis	Siblings matched (n=12,582)	7,542 (59.9)	0.994 (0.975–1.013)
	Full cohort (n=65,852)	39,393 (59.8)	0.990 (0.986–0.994)
Upper respiratory tract infections (≥9)	Siblings matched (n=12,857)	2,806 (21.8)	0.983 (0.960–1.007)
	Full cohort (n=67,249)	13,862 (20.6)	0.996 (0.992–1.001)
Frequent infections (≥11)	Siblings matched (n=12,857)	2,542 (19.8)	0.980 (0.955–1.005)
	Full cohort (n=67,249)	12,787 (19.0)	0.990 (0.985–0.994)

For each outcome, two models were run: a conditional logistic regression within sibling sets, and an unconditional logistic regression analysis in the full cohort with robust variance to correct for potential clustering within families. All models were adjusted for parity, gestational age (<37/≥ 37weeks), gender, caesarean section, birth weight category, day-care outside home, maternal age, maternal education and maternal smoking.

Only sibships discordant for endpoint contributed to this analysis, i.e. sibships where one child had the endpoint and at least one sibling did not.