Childhood body mass index trajectories: modeling, characterizing, pairwise correlations and socio-demographic predictors of trajectory characteristics

Table 2 Mixed effect models with the best fractional polynomial function for childhood BMI trajectory, by model degree^a

Degree	No. of candidate models	Included age terms in the mixed effect model with the best fractional polynomial function								Goodness of fit (smaller is better)^b
		Age^(-2)	Age^(-1)	Age^(-0.5)	log(Age)	Age^0.5	Age	Age²	Age³	-2 Log likelihood	BIC
Boys (N = 1,680)
3 rd degree	56				×	×	×			150196	150218
4th degree	70		×		×	×	×			149688	149710
5th degree	56	×	×	×	×	×				147836	147858
6th degree	28	×	×	×	×	×	×			148889	148911
7th degree	8	×	×	×	×	×	×	×		161668	161690
8th degree	1	×	×	×	×	×	×	×	×	166173	166181
Girls (N = 1,609)
3 rd degree	56				×	×	×			141787	141809
4th degree	70		×	×	×	×				139990	140012
5th degree	56	×	×	×	×	×				138131	138153
6th degree	28	×	×	×	×	×	×			140079	140101
7th degree	8	×	×	×	×	×	×	×		152402	152424
8th degree	1	×	×	×	×	×	×	×	×	156241	156248

BIC, Bayesian information criterion
^a Data from 3,289 children with at least 18 well-child visits from age 1 week to 18 years
^b The best-fitting models are in bold font

ISSN: 1471-2288