Skip to main content

Table 2 Results of extrapolating early-pregnancy weights in Study I and interpolating early-pregnancy weights in Study II

From: Methodological approaches to imputing early-pregnancy weight based on weight measures collected during pregnancy

Imputation method Mean Absolute Error (kg)
Study I weight extrapolation
N=1472
Study II weight interpolation
N=2131
Mixed-effects models All weights included (n=6272) Dropping third trimester weights (n=3375) All weights included (n=11,775) Dropping third trimester weights (n=8125)
 3 knots (quartiles) 1.99 2.01 1.69 1.64
 4 knots (quintiles) 2.08 2.05 1.66 1.59
 5 knots (sextiles) 2.18 N/Aa 1.60 1.70
 3 knots (5th, 50th, 95th) 2.00 2.00 1.67 1.63
 4 knots (5th, 35th, 65th, 95th) 2.00 1.98 1.62 1.66
 5 knots (5th, 27.5th, 50th, 72.5th, 95th) 2.25 1.98 1.60 1.81
 Linear 2.02 2.01 1.67 1.65
 Quadratic 2.02 1.95 1.66 1.62
 Cubic 2.02 6.46 1.62 N/Aa
Marginal models with GEE Mean residual Nearest weight residual Mean residual Nearest weight residual
 3 knots (quartiles) 2.94 3.94 2.03 1.98
 4 knots (quintiles) 2.94 3.94 2.00 1.97
 5 knots (sextiles) 2.94 3.94 1.96 1.96
 3 knots (5th, 50th, 95th) 2.94 3.94 2.02 1.97
 4 knots (5th, 35th, 65th, 95th) 2.94 3.93 1.97 1.96
 5 knots (5th, 27.5th, 50th, 72.5th, 95th) 2.94 3.92 1.95 2.01
 Linear 2.94 3.94 2.01 2.03
 Quadratic 2.94 3.94 2.02 1.98
 Cubic 2.93 3.92 1.97 1.96
Assigning the nearest weight measure 2.46 2.14
Arithmetic imputation using the nearest two weight measures 2.91 2.00
  1. Abbreviations: GEE generalized estimating equation
  2. aModel failed to converge