Table 3.

Summary of 3 studies with continuous outcomes, ranked by effect size of β

First Author, Year [Reference]	Exposure¹		Outcome		β (95% CI)³	Social divisions			Notable Exclusions	Strengths	Limitations

	Definition	Period	Definition	Life Stage²		Racial/Ethnic Minorities	Low Income	Low Parental Education
Karp, 2012 [27]	Weight Change, 4^th quartile <2.5 kg	4 to 6 mo	Log( $\hat{BMI}$ )	Preschooler	1.23 (1.07,1.39)	95%	ND	ND		• Clinic serves a largely inner city immigrant population	• Minimal adjustment for confounders • BMI outcome not standardized to sex and age • % missing data unknown • Birth cohort and sociodemographics of target population undefined
Karaolis-Danckert, 2008 [18]	Rapid Weight Gain	Birth to 24 mo	BMI SDS trajectory	Toddler - Preschooler		ND, low	ND	45.30%	• Preterm • Post term • Multiple Births • Low birth weight	• Analytically examined interactions with education • Additional examination of effect modification and adjustment for important maternal factors • Only 28% missing data	• Binary exposure
		Mother not overweight, not smoker			0.95⁴
		Mother not overweight, smoker			1.10⁴
		Mother overweight, not smoker			1.01⁴
		Mother overweight, smoker			1.16⁴
Shankaran, 2010 [24]	Weight change per month (g/mo)	Birth to 4 BMI Middle	BMI	Middle Childhood		88%	ND	ND	Mothers that were: • Previously institutionalized • <18 years • Not fluent in English	• Additional adjustment for important maternal factors • Did not exclude preterm, isolated stronger association • Recruited from multiple US regions • Large prospective sample	• 31% missing data
		Born preterm			0.11 (0.05,0.17)
		Born at term			0.07 (0.01, 0.13)

BMI, Body Mass Index; SDS: Standard Deviation Score; mo, months; y, years; d, days; ND, not defined; Ideal adjustment included studies that adjusted for at least sex, birth size, and socioeconomic status

All studies have at least one period of early life growth defined as starting between birth and 7 days of age;

Outcome childhood stage: Toddlers, 2-3y; Preschoolers, 3-5y; middle childhood, 6-11y; young teens, 12-14y; teenagers, 15-17y; young adult, 18-35y;[49]

β corresponds to the most adjusted model presented. For those studies with multiple examined associations correspond to the greatest β

⁴

Conditional estimates calculated using adjusted coefficients in Karaolis-Danckert (2008), Table 4[