Table 1.

Estimated relative risk ratios (RRRs) and 95% confidence intervals (CI) of belonging to a given body mass index (BMI) or fussy eating (FE) class (relative to the reference class) per 1 SD increase in birth weight, estimated using either a 1-step or 3-step approach. The classes were identified using the best fitting growth mixture model (GMM) and best fitting latent class growth analysis (LCGA) model, for log(BMI) and FE; Avon Longitudinal Study of parents and children, N = 4227 for the BMI classes and N = 5437 for the FE classes

Variable	Model	Classa	1-step				3-step
			Class %	RRR	95% CI		Class %	RRR	95% CI
Log (BMI)	GMM	1 (ref)	74.8	1			74.7	1
		2	12.4	1.17	0.96	1.42	12.6	1.06	0.86	1.31
		3	5.7	0.81	0.64	1.04	6.0	0.92	0.68	1.25
		4	7.2	1.45	1.21	1.76	6.7	1.32	1.05	1.68
	LCGA	1	18.2	0.74	0.67	0.81	17.9	0.77	0.70	0.84
		2 (ref)	33.0	1			33.3	1
		3	27.1	1.10	1.00	1.21	27.3	1.13	1.02	1.24
		4	15.7	1.03	0.92	1.17	15.8	1.04	0.93	1.16
		5	5.9	1.25	1.05	1.48	5.8	1.30	1.09	1.55
FE	GMMb	1 (ref)	65.3	1			75.1	1
		2	27.1	0.91	0.82	1.00	16.5	1.16	1.01	1.34
		3	7.6	1.12	0.86	1.46	8.4	0.96	0.83	1.10
	LCGA	1	20.7	0.99	0.89	1.10	20.9	1.04	0.93	1.16
		2	7.0	0.87	0.77	0.99	6.9	0.91	0.80	1.04
		3 (ref)	37.4	1			37.5	1
		4	5.9	0.88	0.75	1.03	5.9	0.90	0.77	1.06
		5	19.9	0.88	0.80	0.97	19.8	0.90	0.81	1.00
		6	9.0	0.97	0.84	1.13	9.1	1.00	0.85	1.17

BMI body mass index, FE fussy eating; ref: reference

^aAs in Figs. 5, 7 and 8

^bResults obtained after constraining the variance of the quadratic slope to be zero