. 2020 Aug 5;30(8):345–353. doi: 10.2188/jea.JE20180238

Table 2. Equations for the smoothed mean and standard deviation of weight gain for gestational age in each BMI category.

BMI category		Regression equation
Underweight	Mean	1.136975+0.0060688^GAs1+0.512251^GAs2-1.145976^GAs3+0.7481886^GAs4-0.0717789^GAs5+0.1023281^GAs6-0.7902897 ^*GAs7
	SD	Sqrt(0.2213189+0.000171^(GA)^2+2^(0.0054634)^*GA+0.0121334)

Normal weight	Mean	1.141124+0.0094918^GAs1+1.071354^GAs2-1.690335^GAs3+0.7131089^GAs4-0.0621935^GAs5+0.1044355^GAs6-0.6980647^*GAs7
	SD	Sqrt(0.2214575+0.0001768^(GA)^2+2^(-0.0054049)^*GA+0.0136551)

Overweight	Mean	1.302762-0.0003153^GAs1+3.656517^GAs2-4.300733^GAs3+0.6899292^GAs4+0.0400375^GAs5-0.0641639 ^GAs6-0.1666943 ^*GAs7
	SD	Sqrt(0.2519704+0.0002095^(GA)^2+2^(-0.0060106)^*GA+0.0164559)

Obese	Mean	1.504155-0.0194209^GAs1+1.798848^GAs2-2.404819^GAs3+0.6415784^GAs4-0.0174574^GAs5+0.1425709^GAs6-0.5427454^*GAs7
	SD	Sqrt(0.2431786+0.0002477^(GA)^2+2^(-0.0061593)^*GA+0.0228958)

BMI, body mass index; GA, gestational age; SD, standard deviation; Sqrt, square root.

^*Equations are used to estimate weight gain on the log scale (natural log).

^**GAs1, GAs2, and so on are derived dependent variables by applying restricted cubic spline to obtain more fitted smooth curves. The number of derived dependent variables (eg, GAs1, GAs2) depends on the optimal number of knots.