An educational intervention to prevent overweight in pre-school years: a cluster randomised trial with a focus on disadvantaged families

Table 5 Statistical tests of change difference between control and intervention at the two-year follow-up

Intraclass correlation (ICC):	Min. n estimate		Max. n estimate
Control	0.0396		0.0443
Intervention	0.0000		0.0000
	Estimate	SE	95% CI		P value
Analysis at the centre level:
Outcome = final BMI z-score – baseline BMI z-score
Control	0.55	0.12	0.16	0.94
Intervention	−0.17	0.10	−0.49	0.16
Change difference (centres paired)	0.72	0.21	0.04	1.40	0.043
Change difference (centres unpaired)	0.72	0.16	0.33	1.10	0.004
Analysis at the child level:
Outcome = final BMI z-score – baseline BMI z-score, assuming ICC = 0
Control	0.56	0.10	0.35	0.77
Intervention	−0.18	0.15	−0.48	0.11
Change difference (centres unpaired)	0.74	0.18	0.38	1.10	< 0.001
Outcome = final BMI z-score – baseline BMI z-score, assuming ICC > 0
Centre level variance	0.00	0.00
Child level variance	0.70	0.11
Change difference (centres unpaired)	0.74	0.19	0.37	1.11	< 0.001
Outcome = final BMI z-score, baseline BMI z-score as covariate, assuming ICC > 0
Centre level variance	0.00	0.00
Child level variance	0.47	0.07
Baseline BMI z-score	0.60	0.06
Change difference (centres unpaired)	0.49	0.16	0.17	0.80	0.002

ISSN: 1471-2458