Intrapartum prediction of birth weight with a simplified algorithmic approach derived from maternal characteristics.

Objective: To derive and validate a population-specific multivariate approach for birth weight (BW) prediction based on quantitative intrapartum assessment of maternal characteristics by means of an algorithmic method in low-risk women. Methods: The derivation part (n =200) prospectively explored 10 variables to create the best-fit algorithms (70% correct estimates within +/- 10% of actual BW) for prediction of BW at term; vertex presentation with engagement. The algorithm was then cross validated with samples of unrelated cases (n =280) to compare the accuracy with the routine abdominal palpation method. Results: The best-fit algorithms were parity-specific. The derived simplified algorithms were (1) BW (g) = 100 [(0.42 x symphysis-fundal height (SFH; cm)) + gestational age at delivery (GA; weeks) - 25] in nulliparous, and (2) BW (g) = 100 [(0.42 x SFH (cm)) + GA - 23] in multiparous. Cross validation showed an overall 69.3% accuracy within +/- 10% of actual BW, which exceeded routine abdominal palpation (60.4%) (P = 0.019). The algorithmic BW prediction was significantly more accurate than routine abdominal palpation in women with the following characteristics: BW 2500-4000 g, multiparous, pre-pregnancy weight <50 kg, current weight <60 kg, height <155 cm, body mass index (BMI) <18.5 kg/m(2), cervical dilatation 3-5 cm, station <0, intact membranes, SFH 30-39 cm, maternal abdominal circumference (mAC) <90 cm, mid-upper arm circumference (MUAC) <25 cm and female gender of the neonates (P < 0.05). Conclusion: An overall accuracy of term BW prediction by our simplified algorithms exceeded that of routine abdominal palpation.