Data were available for 715/1555 (46.0%) UPBEAT participants who attended the 6-month postpartum follow-up visit. Of the 840 non-participants, 1 participant was excluded after randomisation, 19 pregnancies were affected by a major health problem, miscarriage or sudden infant death syndrome (SIDS); 100 declined participation; 701 either did not respond or were unable to participate as they had either, returned to full-time employment, living overseas or the child was not the primary responsibility of the woman. Five women had no infant feeding data recorded at the 6-month follow-up. Of those attending the 6-month postpartum visit with their child 354, 225 and 136 women were classified as BMI obesity classes I, II and III respectively at baseline (15+0 to 18+6 weeks gestation) (for study flow diagram see Fig. 1).

Summary diagram of recruitment process by maternal BMI at trial entry.

Compared to those women who did not return for the 6-month visit, the women who attended the 6-month follow-up were more likely to be of White ethnicity and less likely to be of Black or Asian ethnicity. Women were more likely to be nulliparous at study entry, less likely to be a smoker, more likely to have had GDM in pregnancy and more likely to have had an in-labour caesarean or an operative vaginal birth (Table 1).

There was no significant difference in breastfeeding intention, initiation or exclusive or partial breastfeeding at 6 months by UPBEAT group allocation (Supplementary Table 1). Therefore, data for intervention and control arms of the trial were combined and the participants treated as a cohort.

Table 2 summarises breastfeeding intention and practices by maternal BMI class at trial entry for those who completed the 6-month follow up visit. Overall, 76.3% of women stated antenatally that they intended to breastfeed, and 82.7% put their baby to the breast on at least one occasion. For those women who initiated breastfeeding, the percentage of any breastfeeding at 6-months postpartum decreased in BMI class III compared to classes I and II (class 1, 51.7%, II, 48.2% and III 30.7%, p < 0.05). The percentage of those using formula milk at the 6-month follow-up increased with BMI class, 75.0%, 80.0% and 84.0%, respectively.

Figure 2 shows a Kaplan–Meier survival curve for breastfeeding duration according to BMI category. Interval regression analysis showed duration of exclusive breastfeeding in women with class I obesity was 90.4 days compared to 74.6 days and 73.7 days in those with class II and class III obesity, respectively; mean differences were: class II −15.8 (95% confidence interval (CI) −28.5 to −3.1), p < 0.01) and class III −16.7 (95% CI −32.0 to −1.35), p < 0.05), compared to class I (Table 2). More women with obesity class I compared with class II and III were more likely offering breastfeeding in combination with solids and other fluids (e.g., water) at 6-months postpartum (24.8, 21.6 and 14.8%) (Table 2). 92% of the cohort had introduced solids by the 6-month follow-up.

Kaplan–Meier survival curve for duration of exclusive breastfeeding from birth until cessation in women in each obesity class.

Breastfeeding behaviours demonstrated associations with educational attainment; a higher education attainment was associated with a likelihood of partial or exclusive breastfeed at 6-months postpartum. Maternal age ≥30 years and cohabitation were associated with a higher percentage of breastfeeding initiation and duration (Supplementary Table 2). Whereas being of White ethnicity and smoking were associated with lower rates of breastfeeding initiation and duration.

Table 3 presents the birth outcomes stratified by BMI class. There was no statistical difference in the relationship between initiation of breastfeeding and maternal obesity category (Table 3). There were no differences between mode of birth or infant birthweight ≥4 kg when stratified by maternal BMI. Diagnosis of GDM was significantly more common in BMI classes II and III, compared to class I (Table 3).

Supplementary Table 3 summarises the data stratified by any vs. no breastfeeding, at 6-months postpartum. Women who were offering any breast milk at 6 months (n = 283) weighed 1.12 kg less than their pre-pregnancy weight. Whereas women who were not breastfeeding at 6 months demonstrated weight retention of 0.70 kg (mean difference of 1.81 kg (95% CI 0.75, 2.88) p < 0.01). Changes from baseline to 6 months in neck, mid-arm, wrist and hip circumferences were also significantly lower in women who were fully or partially breastfeeding at 6 months compared to those women who were not breastfeeding. There were no statistically significant associations between maternal skinfold thicknesses and breastfeeding behaviours (Supplementary Table 3).

Subgroup analysis between women of Black and White ethnicity showed that changes in weight retention and circumferences associated with breastfeeding apparent in the White women were not evident in women of Black African or Black Afro-Caribbean ethnicity (Supplementary Table 4). Black women who were breastfeeding at the 6-month follow-up had an average postpartum weight retention of 3.35 kg (95% CI 1.39, 5.30) compared to White women. Similar observations were apparent in mid-upper arm [0.93 (0.36, 1.51)] and hip circumferences [3.59 (2.00, 5.19)]. However, women of Black ethnicity were more likely to initiate breastfeeding (95% vs. 78%) and be breastfeeding at the 6-month follow-up (62% vs. 32%) compared to women of White ethnicity (Supplementary Table 2).

There were no significant differences in the metabolome at 6-months postpartum between women randomised to the control and intervention arms following adjustment for age, ethnicity and parity, and the data were therefore treated as a cohort. Exclusive or partial breastfeeding vs. no breastfeeding at 6-months postpartum was associated with marked changes in the NMR metabolome (Figs. 3 and 4). Breastfeeding was associated with a reduction in some metabolites and an increase in others. A marked reduction in total triglycerides was observed (Fig. 4b), reflecting a reduction within multiple subclasses of VLDL, LDL and HDL lipoprotein particles, and within the IDL lipoprotein subclass (Fig. 3b). VLDL particle 13-1 was smaller (Fig. 4b), and there were lower total lipids in VLDL (Very large, large, medium, small and very small) subclasses which was attributable to lower VLDL triglycerides, cholesterol, and phospholipids (Fig. 3a, b). HDL particle size was higher and there were higher total lipids in HDL (very large and large) subclasses (Fig. 4b) which was attributable to greater total cholesterol and phospholipid content (Fig. 3a). Apolipoprotein A-1 concentration was higher and Apolipoprotein B and the ApoA/ApoB ratio were lower (Fig. 4b). When expressed as proportions of total fatty acids, polyunsaturated fatty acids were increased (linoleic, omega-6 and PUFA), there was a decrease in mono-unsaturated fatty acids (Fig. 4a). In addition, glycoprotein acetyls, an inflammatory marker, was reduced, whereas acetate and the amino acids alanine and glycine were increased (Fig. 4a) in the breastfeeding group compared to those who did not breastfeed.

The right-hand side of the x-axis represents positive associations with breastfeeding (fully or mixed) at 6 months, compared to non-breastfeeding women, and negative associations to the left-hand side.

The right-hand side of the x-axis represents positive associations with breastfeeding (fully or mixed) at 6 months, compared to non-breastfeeding women, and negative associations to the left-hand side.

When considered according to ethnic group, a similar metabolic profile to the whole group was observed in the White women who were breastfeeding at 6-months postpartum (Supplementary Figs. 1 and 2). However, although numbers were smaller there were a number of notable differences seen in the metabolome of Black breastfeeding mothers compared to White mothers; IDL and LDL particle size, concentration and content, were little impacted by breastfeeding in White women (apart from triglyceride content), there was a tendency for these to be lower in breastfeeding women of Black ethnicity, likely as a result of lower phospholipid and total cholesterol content in these particles (Supplementary Fig. 1a, b). Reductions in total fatty acids including polyunsaturated, monounsaturated and saturated fatty acids appeared more marked in women of Black ethnicity who breastfed (Supplementary Fig. 2a). Supplementary Figs. 3 and 4 demonstrate that the metabolic profile comparing breastfeeding and non-breastfeeding women in the control arm only (n = 253), was similar to that in the entire cohort (n = 485).

Of the 124 women who did not initiate breastfeeding, 64% reported “never planned to”, 13% gave reasons including “difficulty in establishing lactation”, “feeding issues with a previous child” and “inverted nipples”. A further 5% reported being “advised not to breastfeed”. Of those who initiated breastfeeding but stopped before their child was 6-months old, the most common reason was “perceived inadequate milk supply” (23%); others were “discomfort” (7%) and “convenience” (5%). Thirty nine percent of women reported “other” reasons for cessation including: infant tongue tie, difficulty in ‘latching on’, needing to return to work and partner/family members wanting to feed the baby. Of the 290 women who maintained some breastfeeding at 6-months, the most cited reasons for continuing were that breast milk is the “best nutrition for baby” (74%), “convenience” (55%), “enjoyment” (47%), “cheaper” (37%) and “maternal weight loss” (22%).