Fasting capillary glucose as a screening test for gestational diabetes mellitus

Authors


Dr H Fadl, Department of Obstetrics and Gynaecology, University Hospital Örebro, SE-701 85 Örebro, Sweden. Email helena.fadl@orebroll.se

Abstract

Objective  To evaluate fasting capillary glucose as a screening test for gestational diabetes mellitus (GDM) compared with traditional risk factors and repeated random capillary glucose measurements.

Design  Cross-sectional, population-based study.

Setting  Maternal Health Care Clinics in Örebro County, Sweden.

Population  An unselected population of women without diabetes.

Methods  Fasting capillary glucose levels were measured at gestational weeks 28–32. Random capillary glucose levels were measured four to six times during pregnancy. Traditional risk factors for GDM were registered. GDM was diagnosed using a 75-g oral glucose tolerance test.

Main outcome measures  Sensitivity, specificity, likelihood ratios.

Results  In 55 of 3616 women participating in the study, GDM was diagnosed before 34 weeks of gestation. For fasting capillary glucose cutoff values between 4.0 and 5.0 mmol/l, sensitivity was in the range between 87 and 47% and specificity between 51 and 96%. Using a combined screening model of traditional risk factors with fasting capillary glucose at various cutoff values increased the sensitivity only slightly compared with using fasting capillary glucose alone.

Conclusion  In this Swedish, unselected, low-risk population, fasting capillary glucose measurements were found to be an acceptable and useful screening test for GDM.

Introduction

Gestational diabetes mellitus (GDM) is defined as ‘carbohydrate intolerance of varying severity with onset or first recognition during pregnancy’.1 This definition includes women with both mild and severe hyperglycaemia, resulting in prevalence varying from 1.4 to 14%, depending on the diagnostic criteria used and the population described.2,3 The American Diabetes Association recommends screening for GDM for all women at increased risk with the 50-g glucose challenge test (GCT),4 and most centres in Europe use either selective screening with the GCT for risk groups identified by traditional risk factors, universal screening with repeated random blood glucose measurements or one-step screening with the diagnostic 75-g oral glucose tolerance test (OGTT).

However, screening for GDM is controversial. The US Preventive Services Task Force and the UK National Institute of Clinical Excellence Antenatal Care Guidelines have made a statement that there is insufficient evidence to recommend for or against screening for GDM.5,6

Several issues need to be considered for a successful screening programme for GDM.5 There has been conflicting evidence on the impact of treatment on maternal and neonatal health,7 but the largest randomised controlled trial of the treatment of GDM showed a reduction in serious perinatal outcomes.8 The method for establishing the diagnosis of GDM and the diagnostic criteria are debated and there is a lack of a generally accepted screening test.5,9,10 If screening for GDM is to be advocated, a reliable, easy to perform and cheap screening test with high acceptance is needed to gain high screening uptake, especially when healthcare resources are finite.

The commonly used 50-g GCT has several limitations; it is relatively expensive and considered unpleasant by many women. It is also affected by the time of last meal11 and has poor reproducibility,12 and the same criticism applies to the one-step OGTT. Fasting blood glucose measurements have been proposed as a useful screening test in selected and high-risk populations,13–15 but there is a need for additional information in different populations regarding the properties of fasting blood glucose as a screening test.16

The aim of this study was to evaluate the diagnostic properties of fasting capillary glucose as a screening test in a Swedish, unselected, low-risk population and to compare fasting capillary glucose with traditional risk factors and repeated random capillary glucose measurements as screening models for GDM.17

Methods

In this cross-sectional, population-based study, all pregnant women attending the maternal healthcare clinics in Örebro County, Sweden, from 1 July 1994 to 30 June 1996, were offered a 75-g OGTT according to World Health Organization criteria between gestational weeks 28 and 32. Details of the study design have been published earlier.17,18 For a few women, due to practical reasons, the OGTT was performed up to 32 and 34 weeks of gestation. The fasting capillary glucose concentration was measured before the 75-g glucose load and the 2-hour capillary glucose concentration measurement. The diagnostic OGTT was performed irrespective of the result of the fasting capillary glucose concentrations. Instructions to patients before OGTT included intake of carbohydrate-rich food 2–3 days before and fasting after 10 pm the day before the test. Women with pre-GDM and also two women with no fasting capillary glucose value measured and four with OGTT performed after 34 weeks of gestation were excluded. Traditional risk factors for GDM (first-degree relative, obesity [≥90 kg, pre-pregnancy weight], prior LGA, large for gestational age, infant [≥4500 g or ≥mean +2SD] or prior GDM) were recorded at the first maternal healthcare visit. Random capillary glucose was measured four to six times during pregnancy, measurements starting at the end of the first trimester, with approximately 6-week interval. Capillary glucose concentration of ≥8 mmol/l at any of the measurements was defined as a positive test in accordance with a screening programme used in large parts of Sweden, and a 75-g 2-hour OGTT was performed in accordance to WHO guidelines, within the next few days.19 If this was in early pregnancy and the OGTT was negative, the OGTT was repeated at gestational weeks 28–32 and the results were included in the study.

Calculations of the performance as screening tests were made for fasting capillary glucose, repeated random capillary glucose and models based on analysis of traditional risk factors alone or in combination with fasting capillary glucose or random capillary glucose, with the result of the OGTT as the reference standard for the diagnosis of GDM.

Diagnostic criteria

Diabetes mellitus (DM) was diagnosed if fasting capillary glucose was ≥6.7 mmol/l or 2-hour capillary glucose ≥11.1 mmol/l. Impaired glucose tolerance (IGT) was diagnosed if fasting capillary glucose was <6.7 mmol/l and 2-hour capillary glucose was 9.0–11.0 mmol/l. In this study, we present GDM as the combined group of IGT and DM.

Biochemical analysis

Both random capillary glucose and fasting capillary glucose samples were analysed by Hemocue (Hemocue AB, Ängelholm, Sweden) using 5 μl capillary whole blood.20

Statistical analysis

The statistical package SPSS for Windows version 12.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis.

Sensitivity and specificity were calculated using a 2 × 2 contingency table. A receiver operator characteristics (ROC) curve was constructed to illustrate the graphic representation of the relationship between sensitivity and specificity of fasting blood glucose over all possible diagnostic cutoff values. Likelihood ratios were calculated for each cutoff value. The 95% confidence intervals (CI) were calculated using the binominal distribution for small numbers and the normal approximation for large numbers. For likelihood ratios, the 95% CI were calculated using the formula presented by Simel et al.21

The study was approved by the ethics committee in Örebro County, Sweden.

Results

During this study period, 4918 pregnant women were eligible for the study, of whom 3616 agreed to participate. Finally, the study population consisted of 3610 pregnant women without diabetes. Basal characteristics of the participants and nonparticipants are given in Table 1. The nonparticipating group tended to have a lower proportion of primiparas women with overweight or obesity, with a previous pregnancy with GDM or heredity for DM, but being to a slightly higher extent of non-Nordic origin.

Table 1.  Basal characteristics for participants and nonparticipants
CharacteristicsParticipants (n= 3616)Nonparticipants (n= 1302)P value
  1. BMI, body mass index.

  2. Results are presented as mean ± SD or percentage. Mann–Whitney U test or chi-square test.

Age in years (SD)27.9 ± 4.828.5 ± 5.00.005
Weight (kg) (SD)65.6 ± 12.164.9 ± 10.00.60
Length (cm) (SD)166 ± 6.0166 ± 6.40.53
BMI (kg/m2) (SD)23.8 ± 4.123.5 ± 3.80.18
Non-Nordic origin (%)11.214.30.001
Heredity (%)9.46.60.002
Obesity ≥ 90 kg (%)4.52.60.003
BMI ≥ 30 kg/m2 (%)7.95.50.005
Prior infant ≥ 4500 kg (%)3.21.80.008
Prior GDM (%)1.30.50.020
Primipara (%)4630.6<0.001

Of the 55 women in the study population diagnosed with GDM (1.5%), 43 (1.2%) fulfilled the criteria for IGT and 12 (0.3%) for DM.

Occurrence, sensitivity, specificity and likelihood ratio for different levels of fasting capillary glucose as a screening test for GDM is shown in Table 2.

Table 2.  The test characteristics for different levels of FCG to predict GDM (IGT plus DM), with 95% CI in parentheses
FCG (mmol/l)Occurrence (%)Sensitivity (%)Specificity (%)LR+LR−
  1. FCG, fasting capillary glucose; LR, likelihood ratio.

≥4.04987 (76–95)51 (50–53)1.8 (1.6–2.0)0.25 (0.12–0.50)
≥4.23278 (65–88)69 (67–70)2.5 (2.2–2.9)0.32 (0.19–0.53)
≥4.41967 (53–79)82 (81–83)3.7 (3.0–4.5)0.40 (0.27–0.58)
≥4.61058 (44–71)90 (89–91)6.0 (4.7–7.7)0.46 (0.34–0.63)
≥4.8649 (35–63)94 (93–95)8.5 (6.3–11.5)0.54 (0.42–0.70)
≥5.0447 (34–61)96 (96–97)13.2 (9.5–18.4)0.55 (0.43–070)

The ROC curve for all the calculated cutoff levels of fasting capillary glucose values as a screening test for GDM is shown in Figure 1.

Figure 1.

ROC curve for different fasting capillary glucose cutoff values (mmol/l) as screening test for GDM.

Calculations for fasting capillary glucose as a screening test for DM are shown in Table 3.

Table 3.  The test characteristics for different levels of FCG to predict DM, with 95% CI in parentheses
FCG (mmol/l)Occurrence (%)Sensitivity (%)Specificity (%)LR+LR−
  1. FCG, fasting capillary glucose; LR, likelihood ratio.

≥4.049100 (73–100)51 (49–52)2.0 (1.9–2.1)0 (—)
≥4.23283 (52–98)68 (66–70)2.6 (2.0–3.4)0.24 (0.07–0.87)
≥4.41983 (52–98)81 (80–83)4.5 (3.4–5.8)0.20 (0.06–0.73)
≥4.61083 (52–98)90 (89–91)8.2 (6.3–10.8)0.19 (0.05–0.66)
≥4.8683 (52–98)94 (93–95)13.5 (10.2–17.9)0.18 (0.05–0.63)
≥5.0483 (52–98)96 (95–97)20.9 (15.5–28.2)0.17 (0.05–0.62)

Sensitivity and specificity for anamnestic risk factors, fasting capillary glucose, repeated random capillary glucose, alone or in combination to predict GDM are shown in Table 4.

Table 4.  Properties of different screening models to predict GDM (IGT plus DM), with 95% CI in parentheses
Screening modelOccurrence (%)GDM
Sensitivity (%)Specificity (%)
  1. FCG, fasting capillary glucose; RCG, random capillary glucose.

FCG ≥ 4.6 mmol/l1058 (44–71)90 (89–91)
FCG ≥ 4.8 mmol/l649 (35–63)94 (93–95)
FCG ≥ 5.0 mmol/l447 (34–61)96 (96–97)
Traditional risk factors1648 (35–60)85 (84–86)
RCG ≥ 8.0 mmol/l448 (35–60)97 (96–98)
Traditional risk factors or FCG ≥ 4.4 mmol/l3074 (63–60)71 (70–71)
Traditional risk factors or FCG ≥ 4.6 mmol/l2371 (58–82)78 (76–79)
Traditional risk factors or FCG ≥ 4.8 mmol/l2064 (52–78)80 (80–81)
Traditional risk factors or RCG ≥ 8.0 mmol/l1969 (57–80)82 (81–84)

For fasting capillary glucose cutoff values between 4.0 and 5.0 mmol/l, sensitivity was in the range between 87 and 47% and specificity between 51 and 96%.

Discussion

In this cross-sectional, low-risk-population-based study, fasting capillary glucose was found to be useful as a screening test for GDM, with test properties comparable with those of other tests. Using a combined screening model of traditional risk factors with fasting capillary glucose at various cutoff values increased the sensitivity only slightly compared with using fasting capillary glucose alone. Fasting capillary glucose gives an opportunity, where near-patient testing is carried out, to proceed to full OGTT.

The strength of this study is that it is based on a large, unselected, Scandinavian population that has an attendance to maternal healthcare clinics close to 100%. The nonparticipating group was characterised by a significantly lower rate of risk factors for GDM. This means that the rate of GDM in the study population could be high in comparison with that in the total population; but sensitivity, specificity and likelihood ratios should not be affected by a difference in prevalence.

A limitation was that fasting capillary glucose was measured only once in pregnancy and in relation to the OGTT, meaning that participants had a intake of carbohydrate-rich food prior to testing. This could have influenced the absolute level of fasting capillary glucose. A disadvantage of a single test during pregnancy as late as 28–32 weeks is that undetected pre-GDM would be missed.

Earlier studies on fasting blood glucose as screening test for GDM have been performed with different diagnostic criteria and populations with different prevalences of GDM compared with those in this study. In the study by Reichelt et al.,13 WHO diagnostic criteria was used, while Perucchini et al.15 used the 3-hour 100-g OGTT. In both the studies, plasma glucose was measured in contrast to our study. According to manufacturer and a study by Fogh-Andersen,20 plasma glucose gives about 11% higher values than capillary whole blood. Thus, a fasting blood glucose value in capillary whole blood of 4.6 mmol/l equals to 5.1 mmol/l when measured in plasma. Allowing for these differences, our results are in agreement with the results of others on fasting blood glucose as a screening test.13,15

The commonly used 50-g GCT screening test for GDM has generally been considered gold standard. In the present study, fasting capillary glucose of ≥4.2 mmol/l, comparable with 4.7 mmol/l plasma value, gives approximately the same sensitivity as reported for the GCT but lower specificity.4 A cutoff level for capillary glucose of 4.6 mmol/l, comparable with 5.1 mmol/l plasma value, gives the same or better specificity than the GCT but somewhat lower sensitivity.22 Thus, the GCT may have minor advantages in test properties. However, compared with a capillary glucose value, the GCT is more expensive and considered unpleasant by many women.

Fasting plasma glucose has been directly compared with the GCT in a population-based study by Sacks et al.;23 in their study, fasting plasma glucose was analysed from the OGTT results, as in our study. They found that fasting plasma glucose performed better than the GCT based on area under curves, with sensitivity and specificity close to the results of the present study when correcting for capillary glucose values. They concluded that fasting plasma glucose might be a more discriminating and reproducible screening test for GDM than the 50-g GCT. Conflicting evidence comes from Rey et al.,24 who came to the conclusion that GCT was better than fasting plasma glucose as a screening test in their population and was cost-effective. However, their study was performed in a selected and small population.

What cutoff value to use in a screening programme depends on clinical evaluation and clinical practice. If choosing a lower cutoff value, higher sensitivity is achieved at the price of having to perform more OGTTs. If the aim is to detect the more severe cases of GDM, the DM group, a screening programme resulting in a low rate of performed OGTT and a high sensitivity is desirable. In the present study, this was achieved by a capillary glucose of ≥5.0 mmol/l as screening test. With serious conditions, the screening programme should have a high sensitivity test. In the case of screening for GDM, it may be justifiable to advocate a test with high specificity in order to avoid exposing a large number of women to OGTT and to avoid the problem of labelling pregnant women with a diagnosis of a complicated pregnancy, at least until the benefit of treating IGT has been unequivocally established.

If the aim is to detect all GDM, including both DM and IGT, it is impossible to have a high sensitivity without getting a low specificity since IGT is close to normal glucose tolerance. This underlines the need for studies evaluating the value of identification and treatment of IGT. Also, when deciding the optimal cutoff value in the ROC curve, the nature of the condition to be identified has to be considered.

For all models of screening for GDM, it is important to recognise that since the disease is rare, estimates of sensitivity will be based on small numbers and hence the 95% CI will be wide. Fasting capillary glucose has been shown in this study to be useful, with test properties comparable with those of other screening tests. Fasting capillary glucose gives an opportunity, where near-patient testing is carried out, to proceed to full OGTT, saving time and resources; hence, fasting capillary glucose as a screening test for GDM could be particularly useful where healthcare resource is scarce.

Conclusion

Fasting capillary glucose measurements were found to be a useful screening test for GDM, with test properties comparable with those of other screening tests. In this Swedish, unselected, low-risk population, it had high acceptance and could be used in clinical practice.

Acknowledgements

The study was supported by grants from the Research Committee of Örebro County Council. We are very grateful to Anders Magnusson, Clinical Research Centre, Örebro University Hospital, for statistical advice.

Ancillary