Interpretation (in light of other evidence)
Sub-Saharan Africa is undergoing a rapid demographic and nutritional transition associated with rapid urbanisation. With an emerging epidemic of type 2 diabetes across the continent, hyperglycaemia in pregnancy is likely to pose a major health challenge in this region in the future. Screening and management of HIP are resource intensive. To give some context, the per capita health expenditure in Uganda is $37.6 per year, compared to $3,958 per year in the UK (2016 figures).17 Although there are no cost-effectiveness studies for GDM screening from Africa, a study from India estimated a cost of $1626 per life year gained.18 Costs of management, often requiring insulin therapy and ambulatory capillary glucose monitoring, are absorbed by women and are currently prohibitively expensive to most.5 In order to prioritise resources efficiently, it is therefore imperative to generate local evidence to ensure only women most at risk of clinically relevant adverse pregnancy outcomes are identified and treated.
In our study, the milder form of hyperglycaemia, GDM, was not associated with birthweight >90th centile. Whilst having larger babies is not a disorder in itself, it may lead to obstructed labour and operative delivery, and is therefore commonly used as a marker of perinatal adversity. This may have particularly grave consequences in the low-income setting where antenatal and perinatal care is suboptimal.19,20 Additionally, women in sub-Saharan Africa are already more prone to obstructed labour due to cephalopelvic disproportion.21–24 While the HAPO study described a clear association between glycaemia and LGA, our study did no show this relationship. The reasons for this discrepancy are unclear, but they may relate to sample size. Nonetheless, if there is association within our population, this is unlikely to have significant clinical relevance.25
However, in accord with HAPO (which revealed that BMI was a stronger predictor than glycaemia26), our data show that mid-gestational BMI was a strong predictor of LGA. In the HAPO study, 78% of LGA babies were born to mothers with normal glucose tolerance,27 whilst in our study nearly 90% of LGA infants were from normoglycaemic mothers. These data underscore the contribution of other (non-glucose) physiological determinants of fetal growth. While there are few studies that have explored associations between HIP and birthweight in Africa,11 those that do exist also point towards BMI as a more reliable predictor of LGA than glycaemia.9,28,29
GDM was also not associated with an increased risk of perinatal mortality. However, we observed few perinatal deaths, and the study may not have been sufficiently powered for this outcome. The background rate of perinatal mortality in Uganda 41 per 1,000 births (eight times greater than the HAPO study13,30), but we observed a lower than expected figure of 28 per 1,000 births – perhaps because our study was biased towards the urban population.
Mean arterial pressure (MAP) >90 mmHg (equivalent to blood pressure of >130/85 mmHg) was the strongest predictor of perinatal mortality. High mid-trimester MAP may be indicative of pre-existing hypertension, or evolving gestational hypertension or preterm pre-eclampsia. The strong association between such hypertensive disorders and perinatal mortality in this setting is well recognised.31 High blood pressure is linked to other determinants such as maternal age, BMI and hyperglycaemia. Indeed GDM is associated with a higher risk of hypertensive disorders, and treating GDM might have benefit of reducing the incidence.6However, in our study, the significant contribution of MAP to perinatal mortality was independent of GDM diagnosis, maternal age, and mid-trimester BMI. This supports identification of high blood pressure as a key strategy to reduce stillbirth and neonatal death.
Of our secondary outcomes, there was a significant association between GDM, glucose concentrations and Caesarean delivery. We did not have the depth of data to conclude whether clinicians were more likely to intervene with Caesarean delivery merely from the diagnosis of GDM,27,32 or due to true clinical indication from obstructed labour or neonatal issues. Three quarters of Caesarean deliveries were coded as ‘emergency’ rather than ‘elective’ in our study, which may suggest suboptimal delivery planning. This warrants further study as any unindicated Caesarean section in the low-resource setting carries undue risk, particularly when performed as emergencies.19 There was a tendency for a positive association between fasting glucose and c-peptide concentration in umbilical cord blood. This is in accord with previous data, including from the HAPO study, that have shown that maternal fasting glucose is a strong predictor of umbilical cord c-peptide concentration and clinical neonatal hypoglycaemia.3 The difference in reported neonatal hypoglycaemia between the GDM and normoglycaemic groups in our study is limited by the predilection to test babies born to mothers with a label of GDM. The rarity of the outcome reflects the challenges faced in this setting to detect neonatal hypoglycaemia which offers a potential argument in favour of testing for and managing GDM if the association with high cord c-peptide is true. However, no reduction in neonatal hypoglycaemia was detected in either of the large trials assessing treatment of GDM, even in the context of management strategies that would nonetheless be unfeasible in our setting.6,7
Conclusion (to include practical and research recommendations)
Our data from Uganda indicate that GDM, even when not optimally managed, is not associated with significant short term adverse outcomes, notably excessive birthweight or perinatal death. Instead, simple screening and interventions to manage obesity and hypertension may be more important in reducing adverse pregnancy outcomes. However, more studies will also be required to examine the effect of GDM beyond immediate peripartum complications, as HIP has been implicated in fetal programming – with potential to increase lifetime cardiovascular risk for both mother and baby.33,34