Table 3

Comparison of statistical models

AIC95% prediction interval
BlnSFH, parity, MUAC, betel nut useGA=23.86946(lnSFH)−0.2278491(parity)−0.1294864(MUAC) +0.114807–43.739810.51310.51225.55547.61
DlnEUV, parity, MUAC, betel nut useGA=10.84176(lnEUV)−0.2485541(parity)−0.3396929(MUAC) +0.1303466(betel)−44.080140.51610.51515.51767.47
ElnSFH for three sequential visitsSee online supplementary webappendixNA* NA*NA*7.39
FLMP GAGA=0.9364955(LMP GA)+2.6780590.82160.82164.56584.65
GlnSFH, LMP, parityGA=3.854183(lnSFH) +0.8436906(LMP GA)−0.1075177–6.9804820.82520.82504.53074.56
  • *For Model E, refer to online supplementary webappendix eMethods 2 for statistical methods of longitudinal modelling. This involved derivation of the GA at the first SFH measurement using two separate regression models, thus a single R2, adjusted R2 and AIC are not reported.

  • AIC, Akaike information criterion; GA, gestational age; LMP, last menstrual period gestational age (weeks from first day of LMP); lnEUV, natural log transformation of estimated uterine volume; lnSFH, natural log transformation of symphysis fundal height; MUAC, maternal mid-upper arm circumference (cm); parity, previous number of deliveries of a live child.