Styled Import Test
Uncertainty about the optimal methods for identifying women most likely to benefit from treatment of GDM has led to a range of different approaches being recommended worldwide. In addition to the question of the appropriate screening test and thresholds for diagnostic criteria, any such approaches can be offered either to higher-risk women (selective screening) or to all eligible women within a population (universal screening). Limiting diagnostic testing only to women at high risk of diabetes may be cost saving compared with universal testing, and more convenient, as completion of an OGTT requires pregnant women to fast overnight and attend clinic for at least two hours. On the other hand, testing all pregnant women may result in more women with GDM being identified and treated to reduce hyperglycaemia, in turn reducing adverse outcomes.
While large observational studies have reported a continuous association between levels of maternal hyperglycaemia and perinatal complications (see section 5.2.1) it remains unclear whether screening women without risk factors and treating milder cases of GDM also leads to improved maternal and fetal outcomes to the same extent.
Some organisations, for example NICE, recommend selective screening of women with known risk factors for hyperglycaemia during early pregnancy using OGTT and repeat testing later in pregnancy (usually at 24–28 weeks’ gestation) for those with risk factors who were not screened positive at first testing.14 Other organisations, for example the ADA58 and the Australian Diabetes in Pregnancy Society (ADIPS)59 recommend that after similar screening of women with risk factors during early pregnancy, all pregnant women should be offered an OGTT in mid-pregnancy irrespective of risk factors.
The NICE guideline includes advice that the following independent risk factors for GDM should be recognised by healthcare professionals.14
It is acknowledged that universal screening approaches with lower diagnostic thresholds will identify women with levels of hyperglycaemia that may be considered ‘milder’ than those identified with higher thresholds. A retrospective observational study compared GDM diagnoses in Switzerland after transition from a selective, two-step approach using a glucose challenge test (GCT) to a universal approach with less strict diagnostic criteria (IADPSG).60 The authors noted that including and treating more mild cases of hypoglycaemia in Switzerland with the IADPSG criteria slightly reduced GDM-related events only in women with risk factors. They speculated that the relationship between adverse perinatal outcomes, glycaemia during pregnancy and the IADPSG diagnostic thresholds might differ with the risk factors observed in the screened population.
Additional risk factors recommended for inclusion in the diagnostic pathway by various organisations are explored below.
Polycystic ovary syndrome
…
East Asian ethnicity
…
Age
…
While large observational studies have reported a continuous association between levels of maternal hyperglycaemia and perinatal complications (see section 5.2.1) it remains unclear whether screening women without risk factors and treating milder cases of GDM also leads to improved maternal and fetal outcomes to the same extent.
Some organisations, for example NICE, recommend selective screening of women with known risk factors for hyperglycaemia during early pregnancy using OGTT and repeat testing later in pregnancy (usually at 24–28 weeks’ gestation) for those with risk factors who were not screened positive at first testing.14 Other organisations, for example the ADA58 and the Australian Diabetes in Pregnancy Society (ADIPS)59 recommend that after similar screening of women with risk factors during early pregnancy, all pregnant women should be offered an OGTT in mid-pregnancy irrespective of risk factors.
The NICE guideline includes advice that the following independent risk factors for GDM should be recognised by healthcare professionals.14
It is acknowledged that universal screening approaches with lower diagnostic thresholds will identify women with levels of hyperglycaemia that may be considered ‘milder’ than those identified with higher thresholds. A retrospective observational study compared GDM diagnoses in Switzerland after transition from a selective, two-step approach using a glucose challenge test (GCT) to a universal approach with less strict diagnostic criteria (IADPSG).60 The authors noted that including and treating more mild cases of hypoglycaemia in Switzerland with the IADPSG criteria slightly reduced GDM-related events only in women with risk factors. They speculated that the relationship between adverse perinatal outcomes, glycaemia during pregnancy and the IADPSG diagnostic thresholds might differ with the risk factors observed in the screened population.
Evidence level 2+:
A systematic review and meta-analysis assessed the predictive accuracy of different combinations of risk factors to identify women at high risk of GDM. In addition to noting that risk factors for GDM differ with the diagnostic criteria used, the authors reported that no evidence was identified that screening strategies using several risk factors or risk prediction models offered significant benefit over the simpler strategy of identifying one or two risk factors. Individual patient data analyses suggest that the risk factor combination of maternal age and BMI (≥25 years and BMI ≥25 kg/m2) would identify the majority of women with GDM, but would mean inviting most women for an OGTT. Although this is as effective as more complex strategies (risk prediction models for example) it may not vary greatly from offering all women an OGTT. As sensitivity increases (and more women are identified), the number needed to receive a diagnostic test also increases. To achieve a sensitivity of over 90%, nearly all women would need to undergo an OGTT.61
Polycystic ovary syndrome
…
East Asian ethnicity
…
Age
…
Existing diagnostic and screening criteria for gestational diabetes
Optimal diagnostic criteria
Given the continuous relationship between glucose and some maternal and neonatal outcomes demonstrated by the HAPO study, it is unsurprising that the diagnostic level may be set at different levels. There is an argument that the precise level should reflect underlying risk in the population and may therefore be different in different populations.81
In Scotland, standard practice for diagnosis of GDM involves offering a 75 g OGTT at 24–28 weeks’ gestation and reviewing postload glucose levels against IADPSG thresholds in those with any of the following risk factors:
While IADPSG approaches have generally been favoured due to increased identification of women potentially at risk of GDM, the publication of a revised approach by NICE based on increased cost effectiveness challenges healthcare professionals in Scotland to compare and evaluate the strengths and weaknesses of these standards.
Several systematic reviews and RCTs were identified which offer information relating to the impact of different diagnostic criteria for GDM. The evidence base is difficult to synthesise as outcomes vary by population, by screening strategy used (risk-factor-based or universal) and application of screening in the first trimester. A number of RCTs investigate screening approaches which do not align with methods used in Scotland, for example with oral GCT prior to OGTT. Nevertheless, these trials potentially inform approaches to criteria with lower (diagnosing larger part of population as GDM) and higher glucose thresholds. In general, due to the nature of managing women differently according to the diagnostic strategy groups to which they were allocated in trials, it is difficult to maintain blinding across all participants, clinicians and researchers and the evidence is therefore susceptible to provider bias.
In addition, a large volume of observational studies conducted in a wide range of countries and settings was identified which provides information on the prevalence of GDM according to diagnostic criteria used, and some information on sensitivity and positive predictive value of different diagnostic thresholds. The applicability and quality of these studies varied.
Table 2: Selected diagnostic and screening criteria for GDM
Universal testing approaches
Organisation | Screening test and threshold | Diagnostic test and threshold |
IADPSG, WHO, ADIPS, FIGO, JDS, EBCOG, Endocrine Society, China Ministry of Health | One-step diagnostic test | 75 g two-hour OGTT Fasting glucose ≥5.1 mmol/L One-hour glucose ≥10 mmol/L Two-hour glucose ≥8.5 mmol/L One abnormal value required for diagnosis |
ADA58 | Either one-step diagnostic test,
or two-step: 50 g GCT with screen positive threshold at ≥7.2–7.8 mmol/L | 75 g two-hour OGTT,
Fasting glucose ≥5.1 mmol/L One-hour glucose ≥10 mmol/L Two-hour glucose ≥8.5 mmol/L One abnormal value required for diagnosis
or 100 g three-hour OGTT Fasting glucose: ≥5.3 mmol/L One-hour glucose: ≥10 mmol/L Two-hour glucose: ≥8.6 mmol/L Three-hour glucose: ≥7.8 mmol/L
Two abnormal values required for diagnosis |
NICE | BMI >30 kg/m2, previous macrosomia (≥4,500 g), previous GDM, family history of diabetes, and family origin with a high prevalence of diabetes (South Asian, Black Caribbean, Middle Eastern) | 75 g two-hour OGTT Fasting glucose ≥5.6 mmol/L Two-hour glucose ≥7.8 mmol/L One abnormal value needed for diagnosis |
NICE v IADPSG criteria
A number of studies compared prevalence of GDM or clinical outcomes for women when applying IADPSG or NICE diagnostic criteria for GDM. Eight studies reported that use of the NICE diagnostic criteria led to a smaller proportion of women being diagnosed with GDM based on the same glucose levels compared with the IADPSG criteria.82-89 One study showed that NICE criteria identified a larger proportion of women with GDM.90 In populations of pregnant women who underwent universal screening, IADPSG criteria resulted in a 1.07 to 2.4-fold increase in prevalence, and a 4.2-fold increase in risk factor-based screening.
A number of observational studies looked at how outcomes of women who would not have been diagnosed with either criterion differed from outcomes in women diagnosed with one criterion but not the other, or both (ie negative in both criteria ‘IADPSG- NICE-‘; compared with those diagnosed with IADPSG criteria but not NICE criteria ‘IADPSG+ NICE-‘; and those diagnosed with NICE criteria but not IADPSG criteria ‘IADPSG- NICE+’; and those diagnosed by both criteria, IADPSG+ NICE+’). Notably all of these types of analysis are often difficult to interpret as they are generally carried out in treated populations.
One study reported similar outcomes in treated women using either criterion.91 Several studies noted increased risk in women with fasting glucose levels of 5.1–5.5 mmol/L (IADPSG+ NICE-) compared with women without GDM by either criterion.84,86-89,92 By contrast there were no significant adverse maternal and perinatal outcomes observed in women diagnosed as GDM by NICE criteria but not IADPSG criteria (IADPSG- NICE+) compared with women without GDM.89
Other screening strategies
Evidence level 1++:
A high-quality RCT randomised women to assessment for possible GDM using two criteria. The lower diagnostic thresholds matched with IADPSG criteria based on a 1.75 odds ratio of the mean values for adverse perinatal outcomes in the HAPO study (see section 5.2.1), while the higher thresholds were a fasting glucose level of ≥5.5 mmol/L or a two-hour level of ≥9.0 mmol/L. GDM was diagnosed in 15.3% of women in the lower threshold group (IADPSG) and 6.1% of women in the higher threshold group.93
Large for gestational age infants were born to 178 of 2,019 women (8.8%) in the lower-glycaemic-criteria (IADPSG) group and to 181 of 2,031 women (8.9%) in the higher-glycaemic-criteria group (unadjusted RR 0.99, 95% CI 0.81 to 1.21; p=0.91). The risk of an LGA infant was similar in the adjusted analyses (aRR, 0.98, 95% CI 0.80 to 1.19; p=0.82).
In a subgroup analysis which included women in both groups whose OGTT results fell between lower and higher diagnostic thresholds it was possible to compare outcomes of those receiving treatment and those who did not. The characteristics of these women were similar. Among the women included in the subgroup analysis (those women in both groups whose OGTT results fell between the lower and higher glycaemic criteria), those in the lower-threshold group gave birth to fewer LGA infants than those in the higher-threshold group (6.2% vs 18.0%; adjusted RR, 0.33, 95% CI 0.18 to 0.62). The number of women needed to diagnose and treat GDM in order to prevent one LGA infant in this subgroup was 4 (95% CI 2 to 17). Results of a number of other outcomes favoured the lower-threshold group, including lower maternal weight gain during gestation, lower incidence of pre-eclampsia, a lower proportion of infants with macrosomia, and higher pharmacological treatment for GDM and use of health services. Neonatal hypoglycaemia was detected and treated more often in the lower-threshold group, perhaps reflecting the fact that mothers in this group were diagnosed with GDM which led to infants being screened for possible hypoglycaemia.
Interpretation of the trial results is not straightforward. The authors note that results of the subgroup analysis suggest clinically important, short-term maternal and infant health benefits for the women who received a diagnosis of a milder degree of GDM and also received treatment, compared with those who did not. However, based on results on the primary outcome they also note that “Overall, the risks of giving birth to an LGA infant and of other infant or maternal complications were not lower with the lower glycaemic criteria than with the higher glycaemic criteria”.
Large for gestational age infants were born to 178 of 2,019 women (8.8%) in the lower-glycaemic-criteria (IADPSG) group and to 181 of 2,031 women (8.9%) in the higher-glycaemic-criteria group (unadjusted RR 0.99, 95% CI 0.81 to 1.21; p=0.91). The risk of an LGA infant was similar in the adjusted analyses (aRR, 0.98, 95% CI 0.80 to 1.19; p=0.82).
In a subgroup analysis which included women in both groups whose OGTT results fell between lower and higher diagnostic thresholds it was possible to compare outcomes of those receiving treatment and those who did not. The characteristics of these women were similar. Among the women included in the subgroup analysis (those women in both groups whose OGTT results fell between the lower and higher glycaemic criteria), those in the lower-threshold group gave birth to fewer LGA infants than those in the higher-threshold group (6.2% vs 18.0%; adjusted RR, 0.33, 95% CI 0.18 to 0.62). The number of women needed to diagnose and treat GDM in order to prevent one LGA infant in this subgroup was 4 (95% CI 2 to 17). Results of a number of other outcomes favoured the lower-threshold group, including lower maternal weight gain during gestation, lower incidence of pre-eclampsia, a lower proportion of infants with macrosomia, and higher pharmacological treatment for GDM and use of health services. Neonatal hypoglycaemia was detected and treated more often in the lower-threshold group, perhaps reflecting the fact that mothers in this group were diagnosed with GDM which led to infants being screened for possible hypoglycaemia.
Interpretation of the trial results is not straightforward. The authors note that results of the subgroup analysis suggest clinically important, short-term maternal and infant health benefits for the women who received a diagnosis of a milder degree of GDM and also received treatment, compared with those who did not. However, based on results on the primary outcome they also note that “Overall, the risks of giving birth to an LGA infant and of other infant or maternal complications were not lower with the lower glycaemic criteria than with the higher glycaemic criteria”.
Evidence level 2++:
A meta-analysis of 55 observational studies evaluated the impact of several diagnostic criteria for GDM on the risk of adverse neonatal outcomes.94 Regardless of GDM diagnostic criteria used, the risk of adverse neonatal outcomes in¬cluding LGA infants, neonatal intensive care unit admission, preterm birth, neonatal hypoglycaemia, birth trauma, macrosomia, hyperbilirubinaemia and respiratory distress syndrome significantly increased in women with GDM compared with the non-GDM group. Similar results were seen across all diagnostic criteria analysed. Notably, meta-regression revealed that the magnitude of the risk of these adverse neonatal outcomes in the subgroup of women diagnosed using IADPSG criteria was not significantly different to those identified by other less strict diagnostic criteria.
Evidence level 1++:
A large cluster randomised non-inferiority trial which included 35,528 pregnant women in Iran compared outcomes in women diagnosed with GDM using a fasting glucose threshold >5.1 mmol/L (IADPSG) with less strict criteria (fasting glucose threshold >5.6 mmol/L).95 While prevalence of GDM was higher when the IADPSG criterion was used, the less strict criteria were non-inferior to IADPSG for macrosomia and Caesarean section births and not significantly different for all other maternal and neonatal outcomes analysed. The authors suggest that the consequences of diagnosing women who have FPG levels of 5.2–5.6 mmol/L may increase the prevalence of GDM without any positive effect on adverse pregnancy outcomes.
Evidence level 1++:
While GDM is traditionally assessed at 24–28 weeks’ gestation and individuals receiving a diagnosis are subsequently managed, a further large, high-quality, multinational RCT recruited a population of pregnant women before 20 weeks’ gestation with at least one risk factor for GDM.97 Based on results of an OGTT completed during this early pregnancy period, women who met IADPSG criteria for GDM were randomised to immediate treatment (intervention arm) or to control groups. A follow up OGTT was carried out in those allocated to the control group at 24 to 28 weeks’ gestation and individuals were further allocated to deferred treatment (for those whose results met the IADSPG criteria at this time point) or no treatment for those who did not meet the diagnostic criteria.
Women were stratified according to glycaemic range based on the 1.75 and 2.0 odds ratios for adverse pregnancy outcomes at 24 to 28 weeks’ gestation as identified in the HAPO study (see sections 1.2.3 and 5.2.1). Women in the higher glycaemic range had a fasting glucose level of 5.3 to 6.0 mmol/L, a one-hour glucose level of ≥10.6 mmol/L, or a two-hour glucose level of 9.0 to 11.0 mmol/L (ie HAPO 2.0). Women in the lower glycaemic range had a fasting glucose level of 5.1 to 5.2 mmol/L, a one-hour glucose level of 10.0 to 10.5 mmo/L, or a two-hour glucose level of 8.5 to 8.9 mmol/L (ie HAPO 1.75, which is equivalent to IADPSG criteria) and did not meet any criteria for the higher range.
Significantly fewer women in the early treatment group experienced an adverse neonatal outcome event (24.9%) compared with the control group (30.5%) (adjusted risk difference, −5.6%, 95% CI −10.1 to −1.2). There were also reductions in severe perineal injury among women in the early treatment group (0.8%) compared with control (3.6%) (adjusted mean difference −2.8%, 95% CI −4.1 to −1.5) and median number of bed days in the NICU or special care nursery (adjusted treatment difference −0.8 bed days, 95% CI −1.3 to −0.3). There was no significant between group differences in pregnancy-related hypertension or neonatal lean body mass.
Exploratory subgroup analyses reported a significant effect of early treatment for GDM on the primary composite outcome of adverse neonatal outcomes in the (HAPO 2.0) higher glycaemic range group (RR 0.77, 95% CI 0.67 to 0.89) but not the (HAPO 1.75/IADPSG) lower glycaemic range group (RR 0.91, 95% CI 0.60 to 1.38). The results also suggest the possibility of an increased risk of SGA infants among mothers who had OGTT results that were in the lower glycaemic range.
At 24–28 weeks’ gestation, GDM was diagnosed in 78.0% of the women in the subgroup with a higher glycaemic range and in 51.4% of those in the subgroup with a lower glycaemic range. The authors note that the results suggest the possibility that treatment may be more likely to benefit women with higher glucose levels at early screening and may be more likely to confer harm among those with lower values.
Women were stratified according to glycaemic range based on the 1.75 and 2.0 odds ratios for adverse pregnancy outcomes at 24 to 28 weeks’ gestation as identified in the HAPO study (see sections 1.2.3 and 5.2.1). Women in the higher glycaemic range had a fasting glucose level of 5.3 to 6.0 mmol/L, a one-hour glucose level of ≥10.6 mmol/L, or a two-hour glucose level of 9.0 to 11.0 mmol/L (ie HAPO 2.0). Women in the lower glycaemic range had a fasting glucose level of 5.1 to 5.2 mmol/L, a one-hour glucose level of 10.0 to 10.5 mmo/L, or a two-hour glucose level of 8.5 to 8.9 mmol/L (ie HAPO 1.75, which is equivalent to IADPSG criteria) and did not meet any criteria for the higher range.
Significantly fewer women in the early treatment group experienced an adverse neonatal outcome event (24.9%) compared with the control group (30.5%) (adjusted risk difference, −5.6%, 95% CI −10.1 to −1.2). There were also reductions in severe perineal injury among women in the early treatment group (0.8%) compared with control (3.6%) (adjusted mean difference −2.8%, 95% CI −4.1 to −1.5) and median number of bed days in the NICU or special care nursery (adjusted treatment difference −0.8 bed days, 95% CI −1.3 to −0.3). There was no significant between group differences in pregnancy-related hypertension or neonatal lean body mass.
Exploratory subgroup analyses reported a significant effect of early treatment for GDM on the primary composite outcome of adverse neonatal outcomes in the (HAPO 2.0) higher glycaemic range group (RR 0.77, 95% CI 0.67 to 0.89) but not the (HAPO 1.75/IADPSG) lower glycaemic range group (RR 0.91, 95% CI 0.60 to 1.38). The results also suggest the possibility of an increased risk of SGA infants among mothers who had OGTT results that were in the lower glycaemic range.
At 24–28 weeks’ gestation, GDM was diagnosed in 78.0% of the women in the subgroup with a higher glycaemic range and in 51.4% of those in the subgroup with a lower glycaemic range. The authors note that the results suggest the possibility that treatment may be more likely to benefit women with higher glucose levels at early screening and may be more likely to confer harm among those with lower values.
An economic analysis in the UK has reported that use of the universal IADPSG/WHO testing approach is less cost effective than NICE’s selective screening approach, although will identify more women potentially at risk of adverse perinatal outcomes.98 Despite using similar methods to those used in the economic modelling in the NICE guideline this analysis yields quite different results.
A large NHS health technology assessment included a cost utility analysis to assess the cost- effectiveness of a wide range of screening, testing and diagnostic threshold strategies for GDM.66 The analysis indicated that while generating improved health outcomes, none of the included strategies are cost effective compared with no testing or treatment, when the willingness to pay for health sat in the conventional ranges (£20,000 to £30,000 per quality adjusted life year (QALY)). This included the diagnostic strategies recommended by NICE and IADPSG. There are generalisability issues with the modelled population which simulated women in Bradford, who may differ from women in Scotland. In particular, over 50% women included in the Bradford cohort are of South Asian ethnicity.
The authors report having tested several scenarios in sensitivity analyses. One of the most significant was the inclusion of additional benefits from the early detection of T2DM in mothers. Inclusive of those benefits, intervention became cost effective when the willingness to pay was
£24,000 per QALY or greater. It was unclear which screening, testing and diagnostic thresholds strategies that applied to. Further, those results appeared to be highly linked to the underlying risk of T2DM, which may be higher in the modelled population than in Scotland due to ethnic differences. Similarly, the data used to estimate the treatment effect were from Bradford and Ireland leading to external validity problems.
These results support the view that although intervention at lower glucose thresholds does improve health outcomes, the resources required result in the displacement of greater health outcomes elsewhere in the NHS. The authors note that if clinicians use a lower diagnostic glucose threshold than that suggested by the model then the result will be a greater volume of women being treated, and hence an increase in the absolute volume of resources required and, correspondingly, an increase in the absolute amount of health displaced elsewhere in the NHS.
The evidence reviewed in the health technology assessment of identification and treatment of women for GDM is not sufficient to justify the cost of treatment at a cost-effectiveness threshold of £20,000 per QALY. However, if longer-term outcomes are included in the model (although evidence is limited) and costs of providing GDM treatment are reduced by more efficiently deploying existing resources then it may be cost effective to intervene in populations with a high prevalence of glucose intolerance.
Summary and interpretation
The guideline development group notes:
In forming a recommendation, the guideline development group considered a number of practical issues, including that:
Therefore, the guideline development group sought to set a minimal reasonable standard where evidence of benefit appears clear. In doing so they considered diagnostic levels early and later in pregnancy and whether there was sufficient evidence to recommend early testing in all women with risk factors. They also considered whether adoption of different diagnostic criteria in early and late pregnancy might potentially lead to confusion – with a preference to a single set of criteria unless strong evidence existed that two criteria were appropriate.
It was concluded that:
Recommendation
View definition
The diagnosis of GDM is made using a single-step 75 g OGTT when one or more of the following results are recorded in those with risk factors during routine testing:
· fasting plasma glucose ≥5.3 mmol/L
· (one-hour post 75 g oral glucose load ≥10.6 mmol/L, where used)
· two-hour post 75 g oral glucose load ≥9.0 mmol/L.
· fasting plasma glucose ≥5.3 mmol/L
· (one-hour post 75 g oral glucose load ≥10.6 mmol/L, where used)
· two-hour post 75 g oral glucose load ≥9.0 mmol/L.
Good practice
View definition
In light of developing evidence that earlier treatment of GDM may be beneficial, amendment of the current testing windows to the earlier points of 10–14 weeks (for women with prior GDM) and the earlier part of the current testing window (24–26 weeks rather than 24–28 weeks) is suggested.
First Trimester
An evidence review was conducted to investigate whether pregnant women with moderately raised HbA1c (but below the diagnostic threshold for diabetes) in the first trimester of pregnancy are at risk of adverse pregnancy outcomes. Three systematic reviews of observational studies99-101 and 12 cohort studies53,102-112 were identified, however most studies were designed to assess HbA1c as an indicator for the development of GDM in the third trimester rather than to predict risks of adverse pregnancy outcomes.
Evidence level 2++:
One systematic review included data from seven cohort studies and one non-systematic review on the use of HbA1c as a screening tool in the first trimester.99 There was wide variation in the populations, methods and quality of included studies with poor follow up reported. The authors note that there is no evidence to support use of HbA1c as a screening tool in early pregnancy. The validity of HbA1c as a marker for future adverse pregnancy outcomes may vary throughout pregnancy and between population subgroups. Studies have concluded that HbA1c in healthy pregnant women is generally lower than in non-pregnant women due to a combination of increased haemoglobin turnover in pregnancy and younger mean age than the general (non-pregnant) population. There are also natural variations in HbA1c between trimesters of pregnancy which make it harder to establish thresholds of a ‘normal’ range.
Evidence level 2++:
Another systematic review evaluated the overall accuracy of HbA1c in the diagnosis of GDM and included data from eight studies of 6,406 women of whom 1,044 had GDM.101 There was high heterogeneity among the studies due to variations in ethnicities, different criteria for OGTT interpretation and the individual performance of HbA1c methods. The diagnostic accuracy of HbA1c was reported at different thresholds ranging from 36 mmol/mol (5.4%) to 42 mmol/mol (6.0%), and the area under the curve (AUC) was 0.825 (95% CI 0.75 to 0.90), indicating a good level of overall accuracy. The pooled sensitivities and specificities are shown in Table 3. The authors note that HbA1c presents high specificity but low sensitivity regardless of the threshold used to diagnose GDM.
Evidence level 2+:
A further systematic review, which included 11 high-quality studies, examined the use of HbA1c in early pregnancy as a predictor of GDM.100 HbA1c values between 39 mmol/mol (5.7%) and 46 mmol/mol (6.4%) in early pregnancy consistently identified patients who went on to develop GDM.100 The evidence that particular levels are associated with adverse outcomes was less robust. Adverse pregnancy outcomes were associated with elevated HbA1c levels in four of six studies and included pre-eclampsia, induced labour, shoulder dystocia, Caesarean section birth, LGA birth weight, macrosomia, congenital anomalies, and perinatal death. Two studies found no association with adverse events.
Evidence level 2+:
In a post-hoc analysis of data from the vitamin D And Lifestyle Intervention for GDM prevention (DALI) trial, 900 women with singleton pregnancies, aged >18 years, with a BMI of ≥29 kg/m2 who were attending a participating antenatal clinic before 20 weeks of gestation participated.113 A two-hour, 75 g OGTT was carried out at baseline, at 24–28 weeks, and at 35–37 weeks’ gestation. Women fulfilling the criteria for GDM by IADPSG criteria or for overt diabetes were excluded from the DALI trial interventions and received treatment. The main outcome measure for this observational study was the development of GDM.
At a mean gestation of 15 weeks, the mean baseline HbA1c was 33 mmol/mol (5.2%) (range 23–45 mmol/mol (4.3–6.3%)), while 12.8% (N=111) had an HbA1c ≥39 mmol/mol (5.7%) and 4.3% (N=37) had an HbA1c >41 mmol/mol (5.9%).
The baseline HbA1c showed a poor AUC for identifying women with GDM. An HbA1c threshold of 39 mmol/mol (5.7%) showed low sensitivity (15.9%) but high specificity (89.4%) for GDM at any time during pregnancy. Overall, 51.4% of the women in the HbA1c ≥39 mmol/mol (5.7%) group developed GDM, and 72% of these cases were detected before 20 weeks. Women with a higher (≥39 mmol/mol (5.7%)) HbA1c in early pregnancy had a 1.7 times higher risk for GDM sometime in pregnancy compared with women with an HbA1c of <39 mmol/mol (5.7%) (adjusted odds ratio (aOR) of 1.72 (95% CI 1.02 to 2.89)).
There was no significant association between a higher HbA1c (≥39 mmol/mol (5.7%)) and the risk of adverse pregnancy outcomes.
The authors note that their results clearly show the poor sensitivity of HbA1c measured in early pregnancy for detecting GDM. While the 39 mmol/mol (5.7%) cutoff was highly specific for GDM, this threshold did not correctly identify most of the cases of GDM with a false negative rate of 81.8% before 20 weeks and 84.1% for GDM at any time. In this study, women with an HbA1c of ≥39 mmol/mol (5.7%) were not at increased risk of adverse pregnancy outcomes. Among those with negative OGTT results using IADPSG criteria, there was no relationship between higher HbA1c and adverse pregnancy outcomes.
At a mean gestation of 15 weeks, the mean baseline HbA1c was 33 mmol/mol (5.2%) (range 23–45 mmol/mol (4.3–6.3%)), while 12.8% (N=111) had an HbA1c ≥39 mmol/mol (5.7%) and 4.3% (N=37) had an HbA1c >41 mmol/mol (5.9%).
The baseline HbA1c showed a poor AUC for identifying women with GDM. An HbA1c threshold of 39 mmol/mol (5.7%) showed low sensitivity (15.9%) but high specificity (89.4%) for GDM at any time during pregnancy. Overall, 51.4% of the women in the HbA1c ≥39 mmol/mol (5.7%) group developed GDM, and 72% of these cases were detected before 20 weeks. Women with a higher (≥39 mmol/mol (5.7%)) HbA1c in early pregnancy had a 1.7 times higher risk for GDM sometime in pregnancy compared with women with an HbA1c of <39 mmol/mol (5.7%) (adjusted odds ratio (aOR) of 1.72 (95% CI 1.02 to 2.89)).
There was no significant association between a higher HbA1c (≥39 mmol/mol (5.7%)) and the risk of adverse pregnancy outcomes.
The authors note that their results clearly show the poor sensitivity of HbA1c measured in early pregnancy for detecting GDM. While the 39 mmol/mol (5.7%) cutoff was highly specific for GDM, this threshold did not correctly identify most of the cases of GDM with a false negative rate of 81.8% before 20 weeks and 84.1% for GDM at any time. In this study, women with an HbA1c of ≥39 mmol/mol (5.7%) were not at increased risk of adverse pregnancy outcomes. Among those with negative OGTT results using IADPSG criteria, there was no relationship between higher HbA1c and adverse pregnancy outcomes.
Recommendation
View definition
HbA1c in early pregnancy (first trimester) should be considered to detect overt diabetes in pregnancy and to identify a cohort at risk of GDM.
· Women with HbA1c ≥48 mmol/mol should be diagnosed with overt diabetes and managed as such.
· Women with HbA1c 42–47 mmol/mol are at high risk of GDM. Glucose monitoring and dietary management is recommended from the second trimester.
· Women with HbA1c ≥48 mmol/mol should be diagnosed with overt diabetes and managed as such.
· Women with HbA1c 42–47 mmol/mol are at high risk of GDM. Glucose monitoring and dietary management is recommended from the second trimester.
Good practice
View definition
Women with HbA1c 42–47 mmol/mol and who have glucose levels above treatment thresholds
Diet
…
Exercise
…
Myo-inositol
…
Probiotics
…
…
Metformin, insulin and glibenclamide
…
Metformin, insulin and glibenclamide
…