Burden, risk factors and maternal and offspring outcomes of gestational diabetes mellitus (GDM) in sub-Saharan Africa (SSA): a systematic review and meta-analysis

Gestational diabetes mellitus (GDM) is defined as “any degree of glucose intolerance that sets in or is first diagnosed during pregnancy” [1]. Estimates suggest that GDM prevalence is 7.0% in North America [2], 5.4% in Europe [3] and 11.5% in Asia [4]. Differences in GDM prevalence across regions are, at least in part, due to methodological variations as there is currently little consensus on the appropriate methods to screen and diagnose GDM [1, 5, 6, 7, 8, 9, 10, 11, 12]. Two-step screening and diagnosis methods, for example, are based on measurement of glucose concentration following a 50 g glucose challenge test (GCT) and then again after a 100 g oral glucose tolerance test (OGTT) [13]. On the other hand, one-step approaches only rely on the OGTT. In 2010, the International Association of Diabetes in Pregnancy Study Groups (IADPSG) endorsed a more stringent one-step screening and diagnostic criteria using 75 g OGTT [9] and this recommendation was adopted by the WHO in 2013 [12]. Adoption of the IADPSG 2010/WHO 2013 criteria is growing, although use of the other criteria still exists in many contexts.

There are limited data on the burden of GDM in sub-Saharan Africa (SSA). In 2015, a review by Mwanri and colleagues suggested a prevalence of 14% among high risk individuals [14], but the prevalence in the general population is largely unknown. Similarly, the risk factors for GDM among Africans have not been adequately documented. Classical factors such as maternal age, overweight or obesity and family history for type 2 diabetes have been reported to be important risk factors of GDM in SSA [14], as they are in other populations [4]. It is possible that other local drivers such as malnutrition and infections may play a role, although these have not been sufficiently explored [15]. There is increasing evidence that undernutrition in early life can lead to later risk of cardio-metabolic disorders like diabetes [16]. Similarly, chronic infections (such as in HIV or TB that are highly prevalent in the region), perhaps via inflammation and immune activation, are thought to increase risk of diabetes [17].

GDM is known to adversely impact maternal and offspring outcomes [18]. Infants born to GDM women are more likely to be macrosomic i.e. birthweight ≥4.0kgs [19]. Macrosomic infants are more likely to suffer from birth-related injuries such as shoulder dystocia. They are also more likely to be admitted to the neonatal intensive care unit with metabolic complications [20]. Because of increased baby weight, women with GDM are more likely to deliver by caesarean section (CS) and to suffer from vaginal lacerations and postpartum haemorrhage. Most women with GDM revert to normal glycaemic status after giving birth, but they remain at increased risk of developing type 2 diabetes in the long term [2].

Since the review by Mwanri and colleagues was published, a number of studies have been published assessing the burden or risk factors of GDM in SSA (such as [21, 22, 23, 24]); thus, there is need for integrating these new findings into what is already known from previous efforts. Furthermore, much as some studies have examined maternal and offspring outcomes of GDM in SSA, to our knowledge, no one has comprehensively summarised this evidence. Therefore, in this paper, we will provide a current update integrating new evidence on the burden and determinants of GDM in SSA (including the extent to which each identified risk factor increase GDM risk), as well as undertake a rigorous review of the impacts of GDM on maternal and offspring outcomes.

Initially, we identified 271 papers from PubMed (Fig. 1). Additional 12 papers were identified through reverse-forward (recent) citations, checking of reference lists of relevant original papers and other reviews papers, adding up to 283 papers. After applying our inclusion and exclusion criteria, we ended up with 33 eligible articles [21, 22, 23, 24, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61] for inclusion in this systematic review. Of these, 28 papers contributed towards estimation of GDM prevalence [21, 23, 24, 33, 34, 35, 36, 37, 38, 40, 41, 42, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 57, 58, 59, 60, 61], 20 towards assessment of risk factors of GDM [21, 22, 23, 24, 39, 41, 44, 45, 47, 48, 49, 50, 52, 53, 54, 55, 57, 58, 60, 61] and 6 towards the evaluation of the impacts of GDM on maternal and offspring outcomes [39, 43, 47, 51, 57, 61].

Open image in new window

The IADPSG/WHO 2013 diagnostic criteria for GDM were the mostly used (in 13 studies); these were followed by the WHO 1985 to WHO 2006 criteria (10 studies) and then the O’Sullivan & Mahan criteria (or their adaptation by Carpenter and Coustan (CC) or the National Diabetes Data Group (NDDG)) in 5 studies. Fasting glucose (FG) concentrations alone were used to diagnose GDM in 4 studies. In one Ethiopian study [22], the screening and diagnostic criteria for GDM was not reported.

In terms of country of the study, 11 studies were from Nigeria alone [35, 39, 40, 42, 43, 45, 48, 50, 52, 54, 58], 8 studies from South Africa [24, 33, 34, 37, 41, 44, 59, 61] and 3 studies from Tanzania [21, 36, 49]. Cameroon [23, 47] and Ethiopia [22, 38] contributed two studies each. The other 7 countries (Democratic Republic of Congo [46], Djibouti [51], Ghana [53], Kenya [56], Rwanda [55], Uganda [57] and Zimbabwe [60]) contributed one study each.

Prevalence of GDM in sub-Saharan Africa

Our meta-analysis combining data from 28 studies estimates the overall prevalence of GDM in SSA to be 9% (95%CI, 7–12%) (Fig. 2). Further subgroup analyses suggest that the GDM prevalence is 3% (2–5%) in studies published between 1969 and 2009 and 13% (9–17%) for studies from 2010 to 2018 (Additional file 3). Looking at the diagnostic criteria used in included studies, studies employing the O’Sullivan and Mahan method (or its modification by Carpenter and Coustan or the National Diabetes Data Group) suggest a GDM prevalence of 4% (2–75%); those using the WHO 1985 to WHO 2006 criteria have a combined prevalence of 4% (2–6%); and, those relying on fasting blood alone suggest a prevalence of 7% (6–9%). On the other hand, studies using the IADPSG or WHO 2013 criteria have a combined GDM prevalence of 16% (11–21%) (Additional file 4).

Open image in new window

We estimate the prevalence of GDM in SSA to be 9% (95%CI: 7–12%) with risk factors that include having a family history of type 2 diabetes and previous pregnancies complicated by GDM, macrosomia, stillbirth and abortion. Factors such as being overweight or obese, or older than 25 years or hypertensive were associated with a higher risk of GDM. Lastly, GDM women have increased risk of macrosomia in comparison to those without GDM.

Our meta-analytic approaches suggested a combined GDM prevalence of 9%; however, there was a lot of heterogeneity (I² = 96.9%, Fig. 2) among included studies. Possibly and because of this variability, differences in GDM prevalence can be seen in individual studies and exist across and within countries. It is as low as 0% in a Tanzania [36] and as high as 46% in Djibouti [51]. Even in the same country, different estimates of prevalence exist: 2 to 27% in South Africa [41, 59]. We aimed to investigate potential sources of variation in studies on GDM prevalence via sub-group analyses. Based on when the study was published (a proxy of when the study took place), it appears the prevalence of GDM has increased significantly since around 2010. This increase may reflect a true increase in the burden of GDM, for example, because of increasing prevalence of risk factors such as obesity. High rates of overweight and obesity among African women have been reported in some contexts in SSA [62]. Another source of heterogeneity might relate to recent changes in how GDM is screened and diagnosed. Indeed, as demonstrated in this review and others [4], adoption of the IADPSG criteria in 2010 has greatly influenced estimates of GDM burden, significantly increasing the number and proportion of individuals diagnosed with the condition.

Subgroup analyses related to when the study was conducted or diagnostic criteria did not help to eliminate the significant heterogeneity across studies in the subgroups (I² remained greater than 40% in most sub-analyses). We performed meta-regression analyses (data not shown) to identify any further factors majorly influencing the estimate of GDM prevalence in the region. Meta-regression in this case considered both study (sample) size and study quality (STROBE score); however, neither variable was found to significantly influence the estimate of GDM prevalence. Meta-regression was considered not appropriate for analyses on the risk factors and complications associated with GDM. This is because there were very few studies included in each risk factor or complication analysis, and meta-regression requires many studies to implement [63]. Epidemiological approaches, rather than statistical methods, will be required to reduce variation across studies on GDM burden, determinants and complications in SSA. These will include, for example, more collaborative research, standardization of protocols and methodologies and studies conducted in more than one site within and across all SSA countries.

The estimated GDM prevalence in this review is lower than the 14% prevalence reported by Mwanri and colleagues for high risk women in SSA [14]; this should be expected since our analyses were not restricted to the risk profile of participants in included studies. The combined prevalence of GDM this review is also lower than that reported for Asia (11.5%) [4], but higher than that observed in European studies (5^.4%) [2, 3]. These discrepancies could be due to methodological variations, but may also reflect differences in susceptibility to GDM in different populations. For instance, it has been suggested that Asian women are more likely to develop GDM than their Caucasian or African-American counterparts [64].

Most of the identified risk factors for GDM in SSA (such as family history of type 2 diabetes, obstetric history factors, age and BMI category) are well-known determinants of GDM risk and have been studied in other contexts [4]. The direction and magnitude of effects of these factors would have been expected a priori; and, these classical factors will continue to guide risk factor based approaches to screening for GDM in SSA. However, few existing studies have examined non-classical risk factors for GDM among SSA populations. For example, there continues to be limited data exist on the impact of exposure to in-utero and early childhood undernutrition or chronic infections (such as HIV, malaria and others) and lifestyle factors (such as local patterns of smoking, alcohol and dietary intake) on GDM risk in SSA.

We found that GDM is significantly associated with increased risk of macrosomia and a non-significant increase in the risk of CS delivery. This is in accord with well-established literature [19]. However, delivery of large babies may represent a particular problem SSA contexts, where the burden of cephalopelvic disproportion is already high and access to obstetric and early neonatal care are still a major challenge [65].

This is the largest systematic review (to present) on the burden and risk factors of GDM in SSA. It is also the first to systematically summarize the risk that GDM poses on maternal and offspring outcomes. Although we only searched PubMed because it is publicly available and accessible to us, our paper includes more moderate to high quality studies than previous efforts on the topic [14]. Even then, there are still very few studies of good quality conducted in SSA (for example in comparison to studies carried out in Asia [4]) and most of the available evidence was generated from Nigeria and South Africa. Also, there were not enough SSA studies to combine and assess the impact of GDM on most neonatal morbidities including macrosomia or CS births [66] or on maternal and offspring outcomes that happen well after the neonatal period (such as risk of type 2 diabetes [2], infant adiposity [67] or breastfeeding rates [68]). As scientific awareness and attention to GDM increases in Africa, new high quality studies documenting the burden, risk factors and complications of GDM and in a breadth of African countries will emerge. This will enable future systematic reviewers to be more selective and report less variability across retrieved studies when estimating the burden, risk factors and impacts of GDM in the region.

Findings from this review suggest a GDM prevalence of 9% in SSA and that GDM is, to a large extent, driven by classical risk factors of the disease in other contexts. Although there are limited data on neonatal outcomes, macrosomia appears to be a common complication. More SSA studies are clearly required to rigorously document trends in GDM prevalence, characterize risk factors (both classical and emerging) and to better understand impacts on the mother and her offspring.