Background Women with gestational diabetes mellitus (GDM) receive enhanced antepartum care due to assumed higher risks of adverse pregnancy outcomes. Existing observational studies however report surprisingly modest associations between GDM on outcomes such as late stillbirth (fetal death ≥28 weeks’ gestation), provoking international debate about the value of proactively managing GDM. But existing studies have been performed in populations receiving enhanced care; which may be masking the true ‘untreated’ impact of the condition.

This study sought to estimate the distinct effects of raised glucose concentration and receipt of enhanced care on risk of late stillbirth in pregnant women without pre-existing (type 1 or type 2) diabetes.

Methods 291 case pregnancies ending in late stillbirth and 733 control pregnancies were recruited from 41 maternity units in England, UK during April 2014 to March 2016. 94 cases and 277 controls without pre-existing diabetes received a fasting plasma glucose (FPG) test. In England, GDM diagnosis is advised if FPG≥5.6 mmol/L, but other tests (such as 2-hour oral glucose tolerance tests) are generally preferred.

Causal mediation analysis was used to estimate the effects of raised FPG (≥5.6 mmol/L) and subsequent GDM diagnosis (as an instrument for receipt of enhanced care) on risk of late stillbirth. Odds ratios (OR) were estimated by logistic regression, conditioning on confounders identified by directed acyclic graph. The shape of association between FPG (as a continuous variable) and stillbirth was explored by locally-weighted scatterplot smoothing.

Results On average, women with raised FPG experienced twice the risk of stillbirth as women with normal FPG (OR=1.97, 95% CI=0.61–6.32) but this varied with GDM diagnosis (and hence receipt of enhanced care). Women with raised FPG not diagnosed with GDM had four-times higher risks of stillbirth than women with normal FPG (OR=4.22, 95% CI=1.04–17.02) while women with raised FPG who were diagnosed had similar risks as women with normal FPG (OR=1.10 95% CI=0.31–3.91). Stillbirth risk in women with raised FPG was thus around four-times lower for those who received a GDM diagnosis (OR=0.26, 95% CI=0.07–0.93). The risk of stillbirth increased monotonically with increasing FPG, suggesting no biological justification for one GDM diagnostic threshold over another.

Conclusion Women with raised FPG experience higher risk of late stillbirth. If diagnosed with GDM and managed accordingly, this appears to be largely mitigated. Inconsistent diagnostic practices however leave many women with borderline hyperglycaemia exposed to higher risks of stillbirth.