Results

  • Macrosomia overall: RR 0.67 (95% CI 0.48 to 0.94), moderate-certainty evidence. In gestational diabetes: RR 0.55 (0.35 to 0.86), high-certainty evidence, a 45 percent lower relative risk of macrosomia in the studied group with gestational diabetes.
  • Admission to neonatal intensive care: RR 0.74 (0.60 to 0.91), moderate-certainty evidence.
  • Neonatal hypoglycaemia in pre-existing diabetes: RR 0.71 (0.51 to 0.98), moderate-certainty evidence.
  • Time in range in pre-existing diabetes: plus 5.68 % (3.10 to 8.26), moderate-certainty evidence.

Caesarean birth, preterm birth, weight gain and pre-eclampsia did not change significantly. Large-for-gestational-age birth was less common overall (RR 0.77), with very-low-certainty evidence. For small-for-gestational-age birth the RR was 1.37 (0.96 to 1.95), not significant. The authors say a higher risk cannot be excluded, particularly in gestational diabetes.

Limits

Trial sequential analysis clearly limits the strength of these birth-weight findings. For macrosomia, large-for-gestational-age and small-for-gestational-age birth, the required amount of data has not been reached. For macrosomia only about 16 percent of the needed information is available, and about 12 percent in gestational diabetes. Further trials could still change the statistical result, including the effect in gestational diabetes that GRADE rated as high certainty.

The authors continue to recommend CGM in pre-existing diabetes, and in gestational diabetes as a provisional option with scientific follow-up. No CGM type was clearly better than the others.

What the evidence says for type 1 diabetes

For CGM the evidence is strong. In CONCEPTT, 215 pregnant women with type 1 diabetes were randomised to CGM plus capillary checks or capillary checks alone. With CGM, HbA1c fell a little more (minus 0.19 %), and time in range was 68 percent rather than 61 percent. Newborns were less often large for gestational age (OR 0.51; 0.28 to 0.90), less often admitted to intensive care for more than 24 hours (OR 0.48; 0.26 to 0.86) and less often hypoglycaemic (OR 0.45; 0.22 to 0.89). The investigators recommend offering CGM to all pregnant women with type 1 diabetes on intensive insulin therapy.

CONCEPTT is part of the meta-analysis. There, large-for-gestational-age birth in pre-existing diabetes was not significantly reduced in the pooled result (RR 0.89; 0.66 to 1.21). Type 1 and type 2 diabetes could not be analysed separately, because trials that included both groups reported the data together.

Automated insulin delivery was not part of the meta-analysis. In AiDAPT, 124 pregnant women with type 1 diabetes and an HbA1c of at least 6.5 percent used either hybrid closed loop (CamAPS FX) or usual insulin therapy, both with CGM. Time in the range 3.5 to 7.8 mmol/l was 68.2 percent versus 55.6 percent (adjusted difference 10.5 %; 7.0 to 14.0). Time above range, overnight values and HbA1c (minus 0.31 %) were also better. The trial was not designed for neonatal outcomes.

In CRISTAL the 95 participants were already well controlled (mean HbA1c 6.5 percent). With the MiniMed 780G, time in range did not rise significantly overall (66.5 versus 63.2 percent). Overnight, values were in range more often, and time below 3.5 mmol/l was shorter overall and overnight. No unexpected safety problems occurred.

How to read it

In type 1 diabetes, CGM in pregnancy is well supported by randomised data, including benefit for the newborn. Hybrid closed loop can improve maternal glucose further: clearly in AiDAPT, and in CRISTAL, where glucose was already good, mainly overnight and with less time in hypoglycaemia. The results apply to the system that was tested. A benefit of automated insulin delivery for the newborn is not yet shown.

In gestational diabetes, the reduction in macrosomia is the best-supported finding. Trial sequential analysis says the amount of data is not yet enough. Because the question of small-for-gestational-age birth is also open, monitoring fetal growth remains important.