Background Large fetal size (macrosomia) is a frequent pregnancy complication in women with diabetes. High blood glucose during pregnancy is hypothesised to explain much of the association, but previous investigations have presented equivocal results. This could be due to inadequate adjustment for confounding factors or a non-uniform association between blood glucose concentration and birth weight.

This study investigated the association between blood glucose concentrations during early and late pregnancy and birth weight in women with pre-gestational diabetes, whilst accounting for a range of potentially relevant clinical and socio-demographic factors.

Methods All singleton births in women with pre-gestational diabetes delivered during 1996–2008 were identified from the Northern Diabetes in Pregnancy, a population-based survey of all pregnancies occurring in women with pre-gestational diabetes in the North of England. Cases of congenital anomaly (identified from the Northern Congenital Abnormality Survey) were excluded.

Three measures of glycated haemoglobin concentration (HbA1c) were obtained, to estimate the blood glucose concentration around conception, and during the second and third trimesters. The associations between HbA1c at these time points, a range of other clinical and socio-demographic variables, and birth weight were examined by multiple linear regression. The total and indirect associations were further examined by path-analysis. Gestational age was modelled as a three-order polynomial.

Results Increasing peri-conceptional HbA1c was associated with reduction in birth weight (adjusted regression coefficient, b=-50.4 grams per 1%, 95% CI: -71.1 to -29.6), while increasing third-trimester HbA1c was associated with increase in birth weight (b=171.9 grams per 1%, 95% CI: 132.1–211.7). There was no association between birth weight and second-trimester HbA1c.

Of the other variables in the adjusted model; male sex, increasing maternal height, increasing maternal BMI, multiparity, and later gestational age at delivery were all significantly associated with larger birth weight, while increasing maternal age, later gestational age at booking, maternal smoking, history of pre-pregnancy nephropathy or retinopathy were all significantly associated with smaller birth weight.

Maternal socio-economic status (estimated from maternal postcode at birth) was associated with a range of birth weight modifiers (maternal height, BMI, age, parity, smoking status, and periconceptional HbA1c), but as these acted in opposite directions, the overall effect on birth weight was negligible.

Conclusion Maternal blood glucose concentrations are associated with birth weight, but the association is complex, reversing as pregnancy progresses. For women with pre-gestational diabetes, maintaining good glucose control throughout pregnancy is likely to be associated with the lowest risk of pathological fetal size.