Reduced-energy diet in women with gestational diabetes: the dietary intervention in gestational diabetes DiGest randomized clinical trial

Abstract

Reduced-energy diets promote weight loss and improve long-term outcomes in type 2 diabetes but are untested in gestational diabetes. We aimed to identify if weight loss in pregnancy improves perinatal outcomes in gestational diabetes. We performed a multicentre parallel, randomized, controlled, double-blind trial of energy restriction in women with singleton pregnancies, gestational diabetes and body mass index ≥25 kg m^-2. Participants were randomized to receive a standard-energy control diet (2,000 kcal d^-1) or reduced-energy intervention diet (1,200 kcal d^-1) from enrollment (29 weeks) until delivery, provided as weekly diet boxes (40% carbohydrate, 35% fat, 25% protein). The randomization was performed in a 1:1 ratio, stratified by center and blinded to the participants and study team. Primary outcomes were maternal weight change from enrollment to 36 weeks and offspring birth weight. In total, 425 participants were randomized to the control (n = 211) or intervention (n = 214). Outcome data were available for 388 of 425 (90.1%) participants at 36 weeks and 382 of 425 (89.8%) at delivery. There was no evidence of a difference in maternal weight change to 36 weeks between groups (intervention effect -0.20 (95% confidence interval -1.01, 0.61); P > 0.1) and offspring standardized birth weight (intervention effect 0.005 (-0.19, 0.20); P > 0.1). A reduced-energy diet was safe in pregnancy. ISRCTN registration no. 65152174 .

Fig. 1. Flow chart of DiGest study participants.

Participants were randomized to a trial arm after baseline measurements were complete and details of numbers achieved with maternal and neonatal primary outcomes.

Fig. 2. Interaction of maternal characteristics on the effect of the intervention on the requirement for long-acting insulin.

Post hoc subgroup analysis to determine the interaction effect of participant characteristics on the effectiveness of the intervention for the requirement of long-acting insulin. Subgroup analyses were conducted for maternal baseline BMI, education, ethnicity, socioeconomic status, maternal age and study center. Estimated effect sizes were calculated using unadjusted logistic regression and are shown as ORs for each subgroup with 95% CIs. Interaction P values (P interactions > 0.05 for all subgroups) indicate no statistically significant interaction with any subgroup. IMD, Index of Multiple Deprivation. n represents the number of participants with that characteristic out of the total number of participants in that trial arm that required long-acting insulin.

Extended Data Fig. 1. The design of the DiGest randomized controlled trial.

Overview of the study protocol detailing study recruitment process, study timeline and the measurements taken at each study visit.

Extended Data Fig. 2. Post-hoc subgroup analyses to determine interaction effect of participant characteristics on the association between weight loss and pregnancy outcomes.

Subgroup analyses were conducted for baseline BMI, education, ethnicity, socioeconomic status, maternal age, and study centre upon (a) Associations between weight loss upon CGM TIR (3.5–6.7 mmol/L) at 36 weeks. (b) Associations between weight loss and LGA (Intergrowth) in offspring. (c) Associations between weight loss and HbA1c (mmol/mol) at 3 months postnatally. Estimated effect sizes were calculated using unadjusted logistic regression for categorical outcomes and unadjusted linear regression for continuous outcomes. Effect sizes are shown as odds ratios or mean difference for each subgroup with 95% confidence interval. There were no significant interactions with the categories of maternal BMI, education, ethnicity, deprivation, age and study centre across any of the outcomes (p-interaction >0.05).

Extended Data Fig. 3. Changes in weight status and glycaemic control in women who lost weight versus gained weight from study enrolment to delivery.

(a) Mean (SD) weight status at 29 weeks (n = 154;235), 32 weeks (n = 119;173), 36 weeks (n = 154;234) gestation, and 3 months postpartum (n = 108;142) is shown between women who lost weight versus gained weight, respectively. (b) Mean (SD) CGM average glucose (mg/dL) and (c) mean (SD) CGM Time in Range (63–120 mg/dL) at 29 weeks (n = 137;199), 32 weeks (n = 90;112), 36 weeks (n = 95;129) gestation, and 3 months postpartum (n = 85;103) is shown between women who lost weight versus gained weight, respectively. (d) Mean (SD) HbA1c (mmol/mol) at 29 weeks (n = 60;75) gestation and 3 months postpartum (n = 105;132) is shown between women who lost weight versus gained weight, respectively.