**Background:** Antenatal and childhood growth are important indicators of fetal and child health and are associated with adult health outcomes. Most studies have examined antenatal or postnatal growth separately, but modelling growth as a continuum from gestation into childhood is important for understanding its aetiology and for examining associations with early-life exposures. This study aimed to demonstrate the application of multilevel linear spline models for modelling antenatal and postnatal growth trajectories using four measures of anthropometry from 20 weeks’ gestation to age 5 years.
**Methods:** Data came from the prospective follow-up of the ROLO study, a randomised controlled trial of a low glycaemic index diet in pregnancy that recruited 800 secundigravid women who had previously given birth to a baby weighing over 4 kg between 2007 and 2011 at the National Maternity Hospital, Dublin, Ireland. Women were recruited at first antenatal consultation; exclusions included underlying medical disorders, history of gestational diabetes, age <18 years, gestation >18 weeks, and multiple pregnancies. Fetal measurements (abdominal circumference [AC], head circumference [HC]) were obtained from ultrasound scans at medians of 20+6 and 34+1 weeks’ gestation; estimated fetal weight was calculated using the Hadlock 4-parameter formula. Postnatal measures were recorded at delivery, 6 months, 2 years, and 5 years. Weight was measured using a calibrated SECA 813 scale to the nearest 0.1 kg, standing height using a SECA 217 stadiometer to the nearest 0.1 cm, and HC and AC using a SECA ergonomic circumference measuring tape to the nearest 0.1 cm. All measurements were recorded three times and averaged. Multilevel models (2 levels: measurement occasion and individual) were used to examine trajectories from 20 weeks gestation to age 5 years. Linear spline models were fitted with knot points at different ages; the best-fitting model was selected using Akaike’s information criterion and comparison of observed versus predicted values. For AC, HC, and weight, the best-fitting model included 5 linear spline periods: 20–34 weeks gestation, 34 weeks–birth, birth–6 months, 6 months–2 years, and 2–5 years. For length/height, the best-fitting model included 3 spline periods: birth–6 months, 6 months–2 years, and 2–5 years. Models included random effects for intercept and slopes and allowed occasion-level measurement error to vary with age.
**Key Results:** A total of 754, 756, and 759 offspring were included in analyses of AC, HC, and weight, respectively; 720 offspring were included in length/height analyses. Over 50% of women had third-level education and >90% were of white ethnicity. Mean maternal age was 32.3 years (SD 4.2) for mothers of male babies and 32.6 years (SD 4.2) for mothers of female babies. Model fit was good, with mean differences between observed and predicted values close to zero across all periods (e.g., for AC: −0.08 cm from 20–34 weeks, 0.08 cm from 34 weeks–birth, −0.05 cm from birth–6 months; for weight: −0.07 kg from 20–34 weeks, 0.04 kg from 34 weeks–birth, −0.01 kg from birth–6 months). Growth rates were fastest in pregnancy or immediately postpartum. For AC, the fastest rate was 1.20 cm/week (95% CI 1.18 to 1.22) from 20–34 weeks; for HC, 1.01 cm/week (95% CI 1.00 to 1.02) from 20–34 weeks; for weight, 0.24 kg/week (95% CI 0.24 to 0.25) from 34 weeks–birth; for length/height, 0.66 cm/week (95% CI 0.64 to 0.68) from birth–6 months. No strong evidence of differences in AC, weight, or length/height trajectories was found between the intervention and control groups. There was some evidence of slightly greater HC among the control group at 5 years (difference 0.27 cm, 95% CI 0.03 to 0.51). Sex differences were observed: females had lower HC at 20 weeks (difference −0.27 cm, 95% CI −0.38 to −0.16), with the difference widening at birth (−0.75 cm, 95% CI −0.97 to −0.53) and persisting at 5 years (−0.91 cm, 95% CI −1.14 to −0.68). Females had lower birth weight (−0.15 kg, 95% CI −0.21 to −0.08) and lower weight at 5 years (−0.50 kg, 95% CI −0.96 to −0.05). Females were shorter at birth (−0.83 cm, 95% CI −1.17 to −0.48) and at 5 years (−1.22 cm, 95% CI −2.01 to −0.43).
**Clinical Implications:** Multilevel linear spline models provide a practical and interpretable method for modelling growth trajectories spanning the antenatal and postnatal periods, even with relatively sparse repeat measures (as few as 4 assessments). This approach maximises sample sizes by including participants with at least one growth measure under a missing-at-random assumption, reduces selection bias, and provides more precise standard errors by accounting for the non-independence of repeated measures. Joint modelling of antenatal and postnatal growth enables examination of the timing of exposure effects (e.g., whether associations have intrauterine mechanisms, postnatal mechanisms, or both) that would be missed when analysing antenatal and postnatal trajectories separately. The growth rates reported here are specific to a high-birth-weight cohort (macrosomic infants) and are slower postnatally than general population cohorts, consistent with expected ‘catch-down’ growth. The approach is broadly applicable to cohort studies and randomised controlled trials with repeat prospective assessments of growth.