Table 3. Ranking by association strength (P value) of the stage 1 tests with p < 0.05: Outcome associations with BMI allele score for original dataset.

Rank	Outcome variable (original data with variable N)	SD change of inverse normal transformed outcome for a 1 SD change of BMI allele score
		Sample size	SD change	95% CI	P value (adjusted P value)a	Permutation P valueb
1	Leptin, 9c	4,249	0.138	0.11, 0.17	<0.001	–
2	CRP, 9c	4,250	0.083	0.05, 0.11	<0.001	–
3	Age menarchec	2,946	−0.083	−0.12, −0.05	<0.001	–
4	HDL, 9	4,250	−0.067	−0.10, −0.04	<0.001 (0.002)	0.008
5	SBP, 7c	6,013	0.049	0.02, 0.07	<0.001	–
6	IL6, 9	4,240	0.053	0.02, 0.08	0.001, (0.091)	0.010
7	Enjoyment of School Score, 4	5,807	0.041	0.02, 0.07	0.002, (0.255)	0.022
8	Self Esteem: Scholastic Competence, 8	5,222	0.042	0.02, 0.07	0.002, (0.323)	0.025
9	Apolipoprotein B, 9	4,250	0.043	0.01, 0.07	0.005, (0.788)	0.028
10	Triglycerides, 9	4,250	0.042	0.01, 0.07	0.006, (0.970)	0.037
11	VLDL, 9	4,250	0.042	0.01, 0.07	0.006, (1)	0.048
12	Apolipoprotein al, 9	4,250	−0.038	−0.07, −0.01	0.012, (1)	0.046
13	Insulin, 15c	2,859	0.047	0.01, 0.08	0.012	–
14	Attention/activity symptoms score, 11	4,541	0.037	0.01, 0.07	0.013, (1)	0.055
15	SDQ emotional symptoms score, 6	5,748	−0.030	−0.06, 0.00	0.022, (1)	0.063
16	Hygiene Score, 4	6,231	0.028	0.00, 0.05	0.024, (1)	0.070
17	Self Esteem: Global Self Worth Score, 8	5,214	0.031	0.00, 0.06	0.025, (1)	0.079
18	FVC: lung function, 8c	5,276	0.030	0.00, 0.06	0.030	−
19	Burden of compulsions/obsessions score, 7	5,684	0.028	0.00, 0.05	0.031, (1)	0.080
20	Particular fears score, 7	5,734	−0.028	−0.05, 0.00	0.033, (1)	0.086
21	Glucose, 15c	2,862	0.041	0.11, 0.17	0.038	–

Abbreviations: BMI, body mass index; CI, confidence interval; SD, standard deviation; VLDL, very low density lipoprotein; IL6, interleukin 6; SBP, systolic blood pressure; HDL, high density lipoprotein; VLDL, very low density lipoprotein; SDQ, Strengths and Difficulties Questionnaires; LF, lung function; FVC, forced vital capacity. Full names of variables are given in Supplementary Table 3. All outcomes are transformed to normal distributions using a rank-based inverse normal transformation. Exposure and outcome variables are standardised. Outcome as dependent variable, BMI allele score as independent variable.

^aAdjusted P values are adjusted for the 160 tests performed using the Bonferroni correction: p_corrected = p_original·160. Adjusted P values greater than 1 are rounded to 1. Outcomes in validation set are excluded from Bonferroni correction.

^bPermutation P values are generated with permutation testing. Null hypothesis: The outcome variable at rank i in this table would be ranked lower than i if no association was found with the allelic score. We exclude outcomes in validation set such that such that, for instance i = 1 for HDL because all three outcomes ranked higher than HDL are in the validation set).

^cVariables that are in the ‘validation set’, that were chosen for inclusion using apriori knowledge of their association with BMI. Adjusted P value are not given for these are they are not part of the main outcome dataset.