SEARCH

SEARCH BY CITATION

Keywords:

  • Challenge test;
  • gestational diabetes mellitus;
  • screening;
  • systematic review

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Please cite this paper as: van Leeuwen M, Louwerse M, Opmeer B, Limpens J, Serlie M, Reitsma J, Mol B. Glucose challenge test for detecting gestational diabetes mellitus: a systematic review. BJOG 2012;119:393–401.

Background  The best strategy to identify women with gestational diabetes mellitus (GDM) is unclear.

Objectives  To perform a systematic review to calculate summary estimates of the sensitivity and specificity of the 50-g glucose challenge test for GDM.

Search strategy  Systematic search of MEDLINE, EMBASE and Web of Science.

Selection criteria  Articles that compared the 50-g glucose challenge test with the oral glucose tolerance test (OGTT, with a 75- or 100-g reference standard) before 32 weeks of gestation.

Data collection and analysis  Summary estimates of sensitivity and specificity, with 95% confidence intervals and summary receiver operating characteristic curves, were calculated using bivariate random-effects models. Two reviewers independently selected articles that compared the 50 g glucose challenge test to the oral glucose tolerance test (OGTT, 75 or 100 gram, reference standard) before 32 weeks of gestation.

Main results  Twenty-six studies were included (13 564 women). Studies that included women with risk factors showed a pooled sensitivity of the 50-g glucose challenge test of 0.74 (95% CI 0.62–0.87), a pooled specificity of 0.77 (95% CI 0.66–0.89) (threshold value of 7.8 mmol/l), a derived positive likelihood ratio (LR) of 3.2 (95% CI 2.0–5.2) and a negative LR of 0.34 (95% CI 0.22–0.53). In studies with consecutive recruitment, the pooled sensitivity was 0.74 (95% CI 0.62–0.87) for a specificity of 0.85 (95% CI 0.80–0.91), with a derived positive LR of 4.9 (95% CI 3.5–7.0) and negative LR of 0.31 (95% CI 0.20–0.47). Increasing the threshold for disease (OGTT result) increased the sensitivity of the challenge test, and decreased the specificity.

Author’s conclusions  The 50-g glucose challenge test is acceptable to screen for GDM, but cannot replace the OGTT. Further possibilities of combining the 50-g glucose challenge test with other screening strategies should be explored.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Gestational diabetes mellitus (GDM) is a common metabolic complication of pregnancy that affects between 2% and 9% of all pregnant women in Western countries.1–3

Hyperglycaemia in pregnancy is associated with a number of adverse perinatal outcomes, such as neonatal clinical hypoglycaemia, macrosomia, resulting an increased risk of shoulder dystocia, and the need for neonatal intensive care.4 Maternal complications associated with hyperglycaemia in pregnancy include an increased risk of caesarean delivery and pre-eclampsia.4 Furthermore, women with GDM have up to 60% risk of developing type-2 diabetes mellitus (T2D) within 5–15 years of delivery,5 and it has been suggested that children prenatally exposed to a diabetic milieu have an increased risk for the development of T2D later in life.6,7

Until recently there was a lack of evidence to demonstrate the beneficial effects of treatment for GDM. However, recently it has been demonstrated in a number of high-quality studies that treatment of GDM with diet or insulin reduces the risk of a number of important complications associated with GDM, thus improving both perinatal as well as maternal outcome.3,8,9 Crowther et al.3 showed that treatment of GDM with diet or insulin significantly reduced the risk of serious perinatal complications from 4% to 1%. Landon et al.9 showed that there were fewer cases of shoulder dystocia and fewer caesarean deliveries with the treatment of mild GDM. Identifying women with GDM in order to provide treatment has therefore become of eminent importance, but is difficult as clinical signs and symptoms are often absent.

Because of the lack of clinical signs and symptoms of GDM, screening tests are essential to identify women with GDM. One of the tests that is used in the diagnostic pathway is the 50-g glucose challenge test.10 The 50-g glucose challenge test is a glucose loading test. Women ingest a drink containing 50 g of glucose. After 1 hour the venous glucose level is measured. A 75- or 100-g diagnostic oral glucose tolerance test (OGTT) is performed when the blood glucose value is elevated after a 50-g glucose challenge test (the threshold value is often set as 7.2 or 7.8 mmol/l). The OGTT is a glucose loading test in which women ingest a drink containing 75 or 100 g of glucose. The test is performed after overnight fasting. Venous glucose levels are measured both before and at 1 and 2 hours after the ingestion of a glucose load. GDM is diagnosed if blood glucose values after an OGTT are elevated (multiple criteria; Table S1). A number of studies have evaluated the accuracy of the 50-g glucose challenge test as a screening test for GDM, reporting diverse results.

Although currently the 50-g glucose challenge test is not recommended in the majority of national guidelines, it could be a useful test in the diagnostic work-up of GDM. The aim of this study was to systematically review and meta-analyse the accuracy of the 50-g glucose challenge test for the detection of glucose intolerance in pregnancy, in order to evaluate its applicability in the diagnostic work-up of GDM. We evaluated the applicability of the 50-g glucose challenge as a first-step screening test for GDM, and as a replacement of the current diagnostic test (OGTT).

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Search strategy

A medical librarian (J.L.) undertook a systematic search of the electronic databases MEDLINE (1950–October 2010) and EMBASE (1980–October 2010) to identify studies reporting on the 50-g glucose challenge test in pregnant women. The search strategy consisted of free-text words and subject headings (MeSH, SH) related to the target disease (GDM), population (pregnant women) and screening test (the 50-g glucose challenge test). No methodological filter or other restrictions were applied, as this can lead to the omission of relevant papers.11 We systematically inspected reference lists, conducted a ‘cited reference search’ in Web of Science, applied ‘related articles/find similar feature’ in PubMed and EMBASE, and contacted authors of primary studies for further published trials. We imported all references into reference manager databases (Thomson ISI ResearchSoft, Carlsbad, CA, USA). Duplicate studies were excluded.

Study selection

Two reviewers independently screened the titles and abstracts of all of the studies retrieved (M.v.L. and M.D.L.). Based on the full-text manuscripts, we selected studies according to predefined inclusion criteria. Studies were included when they compared the 50-g glucose challenge test (index test) with either the 75 or the 100 g OGTT (reference standard) in pregnant women before 32 weeks of gestation, at any level of risk for GDM, and reported sufficient data to reproduce a 2 × 2 table from the two tests. Studies that did not report enough data for a 2 × 2 table, but for which data could possibly be obtained from the authors, were also evaluated. Studies in which the OGTT was only performed in screen-positive women were excluded. We also excluded diagnostic case–control studies in which women with GDM were compared against women without GDM, as we expected over-optimistic estimates of test accuracy in these studies.12,13 Final inclusion/exclusion decisions were made by comparison of the results of both reviewers. Disagreement was resolved by consulting a third independent reviewer (B.W.M.).

Data extraction

We extracted data on study characteristics, study quality and 2 × 2 tables of test accuracy. We used a pre-designed piloted data extraction form. If there were data missing on test accuracy or on other relevant characteristics, we contacted the corresponding author. Disagreement on data was resolved by discussion and consensus. If no consensus was reached, a third reviewer (B.W.M.) was consulted.

Quality assessment

The methodological quality of selected papers was evaluated using QUADAS, a tool for the quality assessment of studies of diagnostic accuracy.14 Included studies were evaluated on 15 items concerning patient selection, verification, description of the tests and description of the study population.15

Diagnosis GDM

The reference standard for diagnosis of GDM was the OGTT. In current clinical practice, the 75-g OGTT as well as the 100-g OGTT are used to diagnose GDM. We therefore included studies that used either the 75-g OGTT or the 100-g OGTT as reference tests.

In the past, results of the OGTT were classified as normoglycaemic, impaired glucose tolerance (IGT, intermediate category) or as GDM. Currently, the category intermediate IGT is not used. To facilitate comparison between studies and to enable meta-analysis, we considered women with IGT in older studies as either normoglycaemic or as having GDM, according to the criteria currently used.

Data synthesis and bivariate regression model

We extracted 2 × 2 tables cross-classifying the results of the 50-g glucose challenge test with the results of the OGTT. We plotted their results in receiver operating characteristic plots, and created forest plots to visualise data and to explore heterogeneity.

We used a bivariate regression model to calculate summary estimates of sensitivity and specificity, and their 95% confidence intervals, and to construct summary receiver operating characteristic (sROC) curves.16 Likelihood ratios (LRs) were derived from estimates of sensitivity and specificity. In the bivariate regression approach pairs of sensitivity and specificity are jointly analysed within a single model using a random effects approach to account for variation beyond chance. In addition to chance variation and differences in thresholds, the heterogeneity in results between studies can result from bias arising from flawed design or a variation in accuracy between different clinical subgroups. To explore these other sources of heterogeneity, the bivariate regression approach can be extended with covariates to examine whether they have an effect on sensitivity, specificity or both. We examined the following covariates for their effect on test accuracy: reference test (75- or 100-g OGTT), risk level of women in the study (consecutive inclusion versus inclusion of women with risk factors). As multiple criteria for an abnormal OGTT exist for the 75-g OGTT as well as for the 100-g OGTT, we categorised the threshold values of the OGTT to define GDM as being high or low (Table S1). This classification was also added as a covariate to the bivariate model.

We calculated summary estimates of accuracy measures using studies that reported on a threshold of 7.8 mmol/l. In order to evaluate accuracy measures over the whole range of possible thresholds, we estimated accuracy as a function of the 50-g glucose challenge test threshold values by including this value as a continuous covariate in the bivariate model. In order to avoid results being biased towards studies reporting on many different thresholds, we estimated the model in 250 stratified bootstrap samples, in which only one threshold value from each study was randomly selected. The final model was based on the average over all estimates from 250 bootstrap samples. The model parameters were used to produce sROC curves, where the increase in sensitivity and decrease in specificity reflect the shift in threshold value of the 50-g glucose challenge test in the model. Separate ROC curves reflect each type of study (studies with consecutive recruitment of patients versus studies including high-risk women, and low versus high OGTT threshold values).

Statistical analyses were performed using spss 16.0 (SPSS, Chicago, IL, USA) and sas 9.1.3 (SAS Institute Inc. Cary, NC, 2000–2004). Forest plots were made with review manager 5.0 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2008).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Figure 1 summarises the process of literature identification and study selection. Our search resulted in 745 hits. We included 26 studies, comprising 13 564 women, of whom 1027 (7.5%) had GDM.17–42

image

Figure 1.  Literature identification and study selection.

Download figure to PowerPoint

Study characteristics

Table 1 summarises the characteristics of the studies included. Twenty-four studies (92.3%) were cohort studies, one was a randomised controlled trial and one was a cross-sectional study. All studies reported prospective recruitment. Sample sizes ranged from 42 women to 3836 women (median 378 women), with the incidence of GDM varying from 3% to 33% (median 8%).

Table 1.   Key characteristics of the studies included
First author, yearCountryDesignInclusionExclusionGestational age (weeks)Sample size (n)GCT verified (%)GDM n (%)OGTT 75/100 gThreshold OGTT high/lowSensitivity 7.8 mmol/lSpecificity 7.8 mmol/lCapillary, venous plasma or blood
  1. GCT, 50-g glucose challenge test.

  2. *Testing in the second trimester (women diagnosed in the first trimester were excluded).

  3. **For a threshold of the 50-g GCT of 7.2 mmol/l. Accuracy measures were not reported for a threshold of 7.8 mmol/l.

  4. ***For a threshold of the 50-g GCT of 7.0 mmol/l. Accuracy measures were not reported for a threshold of 7.8 mmol/l.

Ayach, 200617BrazilCohortConsecutivePreterm birth, fetal death24–2834110013 (4)100Low0.770.88Plasma
Bonomo, 199818ItalyCohortConsecutiveNone24–2870410050 (7)100High0.910.73Venous plasma
Caliskan, 200419TurkeyCohortConsecutiveNone24–284229914 (3)100High0.930.76Venous
Cetin, 199720TurkeyCohortConsecutiveMedication, preterm birth, preeclampsia24–2827410017 (6)100High0.650.88Plasma
Cocilovo, 199421ItalyCohortConsecutiveNone24–302491009 (4)100High1.00.76Capillary whole blood
Espinosa, 199922MexicoCohortConsecutiveNone24–3044510043 (10)100High0.880.85Venous
Hidar, 200123FranceCohortConsecutivePreterm birth, PPROM24–289510013 (14)75Low0.690.87Plasma
Jirapinyo, 199324ThailandCohortRisk factorsNone8–38 65% 24–3039610042 (11)100High0.860.65Plasma
Keshavarz, 200625IranCohortConsecutiveNone24–284129026 (6)100Low0.880.88Unknown
Lamar, 199926USARCTConsecutiveNone24–281361005 (4)100High0.80.82Venous
Maegawa, 200327JapanCohortNot reportedNone13–26735*1008 (1*)75High0.790.85Unknown
Mathai, 199428IndiaCohortConsecutiveDelivery elsewhere26–3023210011 (5)100Low0.360.8Plasma
Perea, 200229SpainCohortConsecutiveNone24–2813810013 (9)100High1.000.77Unknown
Perea, 200229SpainCohortConsecutiveNone24–2864210053 (8)100High0.980.74Unknown
Perucchini, 199930SwitzerlandCohortConsecutiveNone24–285209353 (10)100Low0.580.91Venous plasma
Puavilai, 199331ThailandCohortRisk factorsNone24–2811510016 (14)100High0.190.95**Unknown
Ramirez, 200332MexicoCohortConsecutiveNone24–2833410024 (7)100Low0.880.64Venous
Rey, 200433CanadaCohortConsecutiveNone24–281887821 (11)75High0.790.97Unknown
Schwartz, 199434USACohortConsecutiveNone7–40, Mean 27.613210025 (19)100High0.920.52Venous plasma
Sermer, 199435CanadaCohortConsecutiveNone26–273836100265 (7)100High0.770.82Plasma
Siribaddana, 199836Sri LankaCohortConsecutiveNone24–2872110040 (6)75Low0.630.84Plasma
Tam, 200037ChinaCohortNot reportedNone24–2889395122 (13)75Low0.730.68***Unknown
Thitadilok, 199538ThailandCohortRisk factorsNone24–2830410023 (8)100High0.910.77Plasma
Uncu, 199539TurkeyCohortConsecutiveNone24–284210014 (33)100High0.790.54Plasma
van Leeuwen, 200740the NetherlandsCohortConsecutiveNone24–2812819847 (4)75Low0.700.89Venous plasma
Weerakiet, 200641ThailandCross sectionalRisk factorsChronic disease (treatment)21–2835910060 (17)100Low0.900.61Plasma

Patient recruitment was reported as being consecutive in 21 studies (81%), whereas in four studies (15%) patients were screened based on the presence of risk factors. One study did not report inclusion criteria. Rates of GDM varied from 3% to 33% in studies with consecutive recruitment, and from 11% to 17% in studies with inclusion based on the presence of risk factors.

Index test

All but two studies reported accuracy measures with the threshold of the 50-g glucose challenge test set at 7.8 mmol/l. Nineteen studies reported accuracy measures for multiple thresholds of the 50-g glucose challenge test. In the majority of the studies the 50-g glucose challenge test was performed between 24 and 28 weeks of gestation.

Reference test

In six studies the 75-g OGTT was used as a reference standard, whereas 20 studies used the 100-g OGTT as a reference standard. Ten studies were categorised as having a low threshold value of the OGTT and 16 studies were categorised as having a high threshold value of the OGTT to define GDM (Tables 1 and S1). In the majority of the studies the OGTT was performed between 24 and 28 weeks of gestation.

Quality assessment

Figure S1 summarises the results of the methodological quality assessment. Inclusion criteria were reported in 25 studies (96.1%). Verification of the 50-g glucose challenge test results was 100% in 20 studies (76.9%) and ≥90% in five studies (19%). Details on the administration of the index and the reference test were reported in 73.1% and 76.9% of the studies.

Data analysis

We could construct 125 2 × 2 tables. The sensitivity of the index test reported by the 26 individual studies ranged from 0% to 100%. The specificity ranged from 2% to 100%. The combined values of the sensitivity and specificity calculated from the 2 × 2 tables are plotted in Figure 2. The wide range of sensitivity and specificity was mainly a result of variation in threshold values of the index test used to define test positivity (range 4.0–16.0 mmol/l). Figure S2 shows sensitivity and specificity of the individual included studies for three commonly used thresholds. Studies reporting several thresholds appear multiple times in this chart.

image

Figure 2.  ROC plot of sensitivity and specificity for all studies included, irrespective of recruitment (consecutive or inclusion of women with risk factors), reference test (75- or 100-g OGTT), threshold value of the index test (high or low) and threshold value of the 50-g glucose challenge test. The width of the blocks is proportional to the inverse standard error of specificity. The height of the blocks is proportional to the inverse standard error of sensitivity.

Download figure to PowerPoint

First we estimated pooled sensitivity and specificity with a bivariate model in which we used data from the studies that reported accuracy measures of the index test for the threshold of 7.8 mmol/l. We evaluated the effect of covariates on sensitivity and specificity. The recruitment of patients was associated with specificity but not with sensitivity of the index test. The specificity of the index test was lower in studies that included women with risk factors compared with studies with consecutive recruitment (P = 0.02). There was no association between the type of reference test (75- or 100-g OGTT) and sensitivity or specificity. A higher threshold of the reference test for the diagnosis of GDM increased the sensitivity and decreased the specificity of the index test compared with a low threshold (P = 0.01 and 0.01, respectively).

For studies that included women with risk factors the pooled sensitivity was 0.76 (95% CI 0.67–0.84), with a pooled specificity of 0.76 (95% CI 0.60–0.87), and with a derived positive LR of 3.2 (95% CI 1.8–5.7) and negative LR of 0.32 (95% CI 0.21–0.47). For studies that included consecutively recruited women the pooled sensitivity was 0.76 (95% CI 0.60–0.87), with a pooled specificity of 0.85 (95% CI 0.81–0.88), and with a derived LR of 5.1 (95% CI 3.7–6.0) and negative LR of 0.28 (95% CI 0.16–0.51). With increasing the threshold of OGTT to diagnose GDM, the pooled sensitivity increased to 0.89 (95% CI 0.72–0.97), with a decreasing specificity of 0.77 (95% CI 0.63–0.86), for studies that recruited all pregnant women, and a specificity of 0.65 (95% CI 0.38–0.85) for studies that recruited women based on the presence of risk factors, with a derived positive LR of 3.9 (95% CI 2.3–6.6) and 2.5 (95% CI 1.2–5.6), and a derived negative LR of 0.14 (95% CI 0.05–0.43) and 0.17 (95% CI 0.05–0.55), respectively.

Next, we estimated accuracy measures for all reported threshold values by including the threshold value of the 50-g glucose challenge test as a covariate in the bivariate model. The effects of the covariates (patient recruitment, reference test, threshold value of the reference test) were the same as described above. The effects of threshold value of the 50-g glucose challenge test were statistically significant for sensitivity (P < 0.0001) as well as for specificity (P < 0.0001).

The pooled sensitivity and specificity for three threshold values are presented in Table 2. For an index test threshold of 7.8 mmol/l, the pooled sensitivity in studies that recruited only women with risk factors was 0.74 (95% CI 0.62–0.87), with a pooled specificity of 0.77 (95% CI 0.66–0.89), and with a derived positive LR of 3.2 (95% CI 2.05.2) and negative LR 0.34 (95% CI 0.22–0.53). The pooled sensitivity of studies including all pregnant women was 0.74 (95% CI 0.62–0.87), with a pooled specificity of 0.85 (95% CI 0.80–0.91), and with a derived positive LR of 4.9 (95% CI 3.5–7.0) and negative LR of 0.31 (95% CI 0.20–0.47). Increasing the threshold of the reference test for the diagnosis of GDM increased sensitivity to 0.83 (95% CI 0.75–0.91), and decreased specificity to 0.81 (95% CI 0.75–0.87) for studies that recruited all pregnant women (derived positive LR 4.4, 95% CI 3.2–6.0; negative LR 0.21, 95% CI 0.14–0.32), and 0.72 (95% CI 0.60–0.84) for studies that recruited women based on the presence of risk factors (derived positive LR 3.0, 95% CI 2.0–4.5; negative LR 0.24, 95% CI 0.15–0.37). Summary ROC curves that reflect the results for all possible thresholds of the 50-g glucose challenge test are displayed in Figure 3.

Table 2.   Accuracy measures for three thresholds of the 50-g glucose challenge test, estimated with a bivariate regression model
 Recruitment of all womenRecruitment of women with risk factors
Sensitivity (95% CI)Specificity (95% CI)Sensitivity (95% CI)Specificity (95% CI)
  1. The threshold of the 50-g glucose challenge test was added as a covariate. Recruitment of women (recruitment of all women versus recruitment of women with risk factors for GDM) was added a covariate to the model. Threshold of the OGTT was also added as a covariate to the model.

OGTT with a low threshold for disease
7.5 mmol/l0.78 (0.67–0.89)0.81 (0.74–0.87)0.78 (0.67–0.89)0.72 (0.58–0.85)
7.8 mmol/l0.74 (0.62–0.87)0.85 (0.80–0.91)0.74 (0.62–0.87)0.77 (0.66–0.89)
8.0 mmol/l0.72 (0.59–0.85)0.88 (0.83–0.93)0.72 (0.59–0.85)0.81 (0.71–0.91)
OGTT with a high threshold for disease
7.5 mmol/l0.85 (0.78–0.93)0.76 (0.69–0.83)0.85 (0.78–0.93)0.65 (0.51–0.79)
7.8 mmol/l0.83 (0.75–0.91)0.81 (0.75–0.87)0.83 (0.75–0.91)0.72 (0.60–0.84)
8.0 mmol/l0.81 (0.72–0.90)0.84 (0.79–0.89)0.81 (0.72–0.90)0.76 (0.65–0.86)
image

Figure 3.  sROC plots of sensitivity and specificity for various subgroups (depending on recruitment and threshold of the reference test), based on the estimates of the bivariate model.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

We systematically reviewed the literature on the accuracy of the 50-g glucose challenge test for the diagnosis of GDM. We found that the pooled estimate of sensitivity for a threshold value of 7.8 mmol/l ranged between 0.74 (95% CI 0.62–0.87) and 0.83 (95% CI 0.75–0.91), depending on the threshold for diagnosis of GDM, with a specificity ranging between 0.72 (95% CI 0.60–0.84) and 0.85 (95% CI 0.80–0.91).

Depending on the application of the test (screening or alternative diagnostic) and the consequences of false-positive and false-negative test results, certain combinations of accuracy values are preferred. These values depend on whether it is more harmful to classify women as false-positive or false-negative, taking all possible consequences of such results into account. In the case of GDM, regarding the nature and consequences of the disease, one should aim for an adequate detection rate, albeit not at the cost of an unacceptable false-positive rate. If the 50-g glucose challenge test is used as a screening test, a higher sensitivity rate than 74% would probably be warranted to accept a false-positive rate of 83%. Moreover, if one considers using the 50-g glucose challenge test as a diagnostic test for GDM, higher detection rates are required. As the prevalence of GDM in the general population is relatively low, a clinically useful test would thus have to have a high positive LR (>10) and a low negative LR (<0.10). Regarding the derived positive and negative LRs in the present study, the accuracy of the 50-g glucose test for gestational diabetes mellitus is modest. For example, if the incidence of GDM is 3%, and the positive LR of the 50-g glucose challenge test is 4.5, the post-test probability for an abnormal result on the 50-g glucose challenge test would be 12%, which is still low. If the 50-g glucose challenge test is combined with other screening methods, such as the presence of risk factors for GDM (e.g. GDM in a previous pregnancy, obesity), LRs of the risk factors are multiplied with the LR of the 50-g glucose challenge test, and thus the post-test probability might be improved.

Our findings imply that if the 50-g glucose challenge test is performed in a cohort of 1000 unselected women, with an assumed prevalence of 4%, and with a threshold of 7.8 mmol/l, 30 women would have a true positive test result and 144 would have a false-positive test result (positive predictive value of 17%). Ten women would have a false-negative test result and 816 women would have a true positive test result (negative predictive value of 99%).

In many potentially relevant studies dealing with the 50-g glucose challenge test in pregnant women, the OGTT was only performed if the 50-g glucose challenge test was considered to be abnormal. This design characteristic, known as partial verification, is encountered in many studies on diagnostic accuracy: to minimise the burden of possibly redundant additional testing in women with a negative screening test result, only abnormal screening test results are verified by the reference test. To avoid this partial verification bias, only studies that performed both a 50-g glucose challenge test and an OGTT independent of the result of the gram glucose challenge test in more than 75% of the women were included in this systematic review.

A limitation of the present study is that details on study and sample characteristics were not reported equally well in the individual studies. Because of this incomplete reporting, we were not able to evaluate every quality item in every study. Inclusion of methodologically poor studies may have affected our estimates of diagnostic accuracy.42 With the bivariate model that we used to calculate summary estimates of sensitivity and specificity, the potential influence of clinical and study characteristics (covariates) on the mean sensitivity and specificity can be evaluated. Because of the limited details that were reported, clinical variables of interest (for example age, BMI and time between the last meal and the index test), the number of covariates included in the bivariate model was limited. To evaluate the true effect of clinical variables and different threshold values, individual patient data meta-analysis is needed.

Another limitation of our study was the lack of a uniform reference test for GDM. The glucose load of the OGTT is either 75 or 100 g. Both tests are used in clinical practice. For the 75-g as well as for the 100-g OGTT, various criteria for an abnormal test result exist. Direct comparison between all studies included in our systematic review was therefore complicated. With the bivariate model we were able to account for the variation in summary estimates caused by the difference in cut-off values used for the index test. We accounted for variation in summary estimates caused by the various criteria to define an abnormal OGTT as well, by categorising the threshold for an abnormal OGTT as being high or low. Adding this variable to the bivariate model increased the fit of the model. We do not know, however, to what extent the arbitrary categorisation of the OGTT thresholds is justified. During pregnancy placental hormones cause maternal insulin sensitivity to decrease, and as a consequence postprandial glucose levels increase. Combs et al.43 showed that rising postprandial glucose values were associated with fetal macrosomia, a common feature in pregnancies complicated by GDM. A glucose loading test like the 50-g glucose challenge test in theory seems an adequate method to mimic postprandial glucose levels, and therefore to measure the degree of glucose (in)tolerance in pregnancy.

A health technology report concerning various screening strategies for GDM stated that the cost-effectiveness of a of number of studies find that screening with the 50-g glucose challenge test, and then testing screen-positives with the OGTT, was less costly than going straight to universal OGTT. However, a high-quality cost-effectiveness analysis developed by the UK’s National Institute for Health and Clinical Excellence (NICE) guideline development group found that two screening strategies dominated: selection by American Diabetes Association (ADA) criteria, followed by the 75-g OGTT; and selection by high-risk ethnicity, followed by the 75-g OGTT.

In view of these findings and as an extension to the results of the cost-effectiveness analysis of the NICE guideline development group, it would be interesting to consider the cost effectiveness of a strategy that consists of selection based on various risk factors, followed by screening with a 50-g glucose challenge test, followed by an OGTT in the case of an abnormal test result of the 50-g glucose challenge test, and to compare this in a randomised controlled trial with other screening strategies.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

As GDM is often asymptomatic, screening is necessary to identify women with GDM. High sensitivity is often warranted in screening tests, as a false-negative test result (in which disease remains undiscovered) is considered to be more harmful than a false-positive test result (in which a reference test is unnecessarily performed). Although higher detection rates would be preferable, the detection rate of the 50-g glucose challenge test of 74% might be acceptable if used as a screening test for a condition such as GDM. On the other hand one could consider a one-step method, using the OGTT for screening for example in a selected population (risk factors). This could be a lesser burden for women and more cost-effective than a two-step method in which a glucose loading test might be performed twice. To use the 50-g glucose challenge test as a definite diagnostic test for GDM (replacement of the OGTT), higher accuracy measures are warranted.

Contribution to authorship

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

MvL, BWM and BO conceived the idea for the study. MvL, BO and JL prepared the protocol, supported by MS and supervised by BWM and JR JL developed and performed the literature search. ML and MvL collected data. BO, JR and MvL performed the statistical analyses. MvL and ML drafted the article. All authors reviewed and edited the article.

Funding

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

This research was supported by grant 917.46.346 in the VIDI-program of ZonMW, the Hague, the Netherlands. The funding sources had no involvement in the design, analysis or reporting of this study.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Preliminary results of this study were presented as a poster presentation at the International Workshop Conference on Gestational diabetes diagnosis & classification, Pasadena, USA, 2008, and at the Annual Meeting of the Society for Gynaecologic Investigation, Glasgow, UK, 2009.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information
  • 1
    Jensen DM, Damm P, Sørensen B, Mølsted-Pedersen L, Westergaard JG, Korsholm L, et al. Proposed diagnostic thresholds for gestational diabetes mellitus according to a 75-g oral glucose tolerance test. Maternal and perinatal outcomes in 3260 Danish women. Diabet Med 2003;20:517.
  • 2
    Esakoff TF, Cheng YW, Caughey AB. Screening for gestational diabetes: different cut-offs for different ethnicities? Am J Obstet Gynecol 2005;193:10404.
  • 3
    Crowther CA, Hiller JE, Moss JR, McPhee AJ, Jeffries WS, Robinson JS; Australian Carbohydrate Intolerance Study in Pregnant Women (ACHOIS) Trial Group. Effect of treatment of gestational diabetes mellitus on pregnancy outcomes. N Engl J Med 2005; 352:247786.
  • 4
    Metzger BE, Lowe LP, Dyer AR, Trimble ER, Chaovarindr U, Coustan DR, et al.; HAPO Study Cooperative Research Group. Hyperglycaemia and adverse pregnancy outcomes. N Engl J Med 2008; 358:19912002.
  • 5
    Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: a systematic review. Diabetes Care 2002;25:18628.
  • 6
    Plagemann AJ. A matter of insulin: developmental programming of body weight regulation. Matern Fetal Neonatal Med 2008;21:1438.
  • 7
    Tam WH, Ma RC, Yang X, Li AM, Ko GT, Kong AP, et al. Glucose intolerance and cardiometabolic risk in adolescents exposed to maternal gestational diabetes: a 15-year follow-up study. Diabetes Care 2010;33:13824.
  • 8
    Langer O, Yogev Y, Most O, Xenakis EM. Gestational diabetes: the consequences of not treating. Am J Obstet Gynecol 2005;192:98997.
  • 9
    Landon MB, Spong CY, Thom E, Carpenter MW, Ramin SM, Casey B, et al. Eunice Kennedy Shriver National Institute of child health and human development maternal-fetal medicine units network. A multicenter, randomized trial of treatment for mild gestational diabetes. N Engl J Med 2009;361:133948.
  • 10
    Gabbe SG, Gregory RP, Power ML, Williams SB, Schulkin J. Management of diabetes mellitus by obstetrician-gynecologists. Obstet Gynecol 2004;103:122934.
  • 11
    Leeflang MM, Scholten RJ, Rutjes AW, Reitsma JB, Bossuyt PM. Use of methodological search filters to identify diagnostic accuracy studies can lead to the omission of relevant studies. J Clin Epidemiol 2006;59:23440.
  • 12
    Lijmer JG, Mol BW, Heisterkamp S, Bonsel GJ, Prins MH, van der Meulen JH, et al. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA 1999;282:10616.
  • 13
    Whiting P, Rutjes AW, Reitsma JB, Glas AS, Bossuyt PM, Kleijnen J. Sources of variation and bias in studies of diagnostic accuracy: a systematic review. Ann Intern Med 2004;140:189202.
  • 14
    Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25.
  • 15
    Whiting P, Harbord R, Kleijnen J. No role for quality scores in systematic reviews of diagnostic accuracy studies. BMC Med Res Methodol 2005;5:19.
  • 16
    Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005;10:98290.
  • 17
    Ayach W, Costa RA, Calderon Ide M, Rudge MV. Comparison between 100-g glucose tolerance test and two other screening tests for gestational diabetes: combined fasting glucose with risk factors and 50-g glucose tolerance test. Sao Paulo Med J 2006;124:49.
  • 18
    Bonomo M, Gandini ML, Mastropasqua A, Begher C, Valentini U, Faden D, et al. Which cutoff level should be used in screening for glucose intolerance in pregnancy? Definition of Screening Methods for Gestational Diabetes Study Group of the Lombardy Section of the Italian Society of Diabetology Am J Obstet Gynecol 1998;179:17985.
  • 19
    Caliskan E, Kayikcioglu F, Oztürk N, Koc S, Haberal A. A population-based risk factor scoring will decrease unnecessary testing for the diagnosis of gestational diabetes mellitus. Acta Obstet Gynecol Scand 2004;83:52430.
  • 20
    Cetin M, Cetin A. Time-dependent gestational diabetes screening values. Int J Gynaecol Obstet 1997;56:25761.
  • 21
    Cocilovo G, Leonardi MC, Vesce F, Mollica G. Cut-off values of laboratory measurements of capillary blood glucose in a glucose challenge test. Evaluation of 250 cases. J Obstet Gynaecol 1994;14:768.
  • 22
    Espinosa de los Monteros A, Parra A, Cariño N, Ramirez A. The reproducibility of the 50-g, 1-hour glucose screen for diabetes in pregnancy. Obstet Gynecol 1993;82:5158.
  • 23
    Hidar S, Chaïeb A, Baccouche S, Laradi S, Fkih M, Milled A, et al. [Post-prandial plasma glucose test as screening tool for gestational diabetes: a prospective randomized trial]. J Gynecol Obstet Biol Reprod (Paris) 2001;30:3447.
  • 24
    Jirapinyo M, Puavilai G, Chanprasertyotin S, Tangtrakul S. Predictive value of 1 hour 50 g oral glucose load screening test for gestational diabetes mellitus compared to 3 hour oral glucose tolerance test in high risk pregnant women. Asia Oceania J Obstet Gynaecol 1993;19:712.
  • 25
    Keshavarz M, Cheung NW, Babaee GR. The value of the screening test for gestational diabetes in pregnant Iranian women. Diabetes Res Clin Pract 2006;73:989.
  • 26
    Lamar ME, Kuehl TJ, Cooney AT, Gayle LJ, Holleman S, Allen SR. Jelly beans as an alternative to a fifty-gram glucose beverage for gestational diabetes screening. Am J Obstet Gynecol 1999;181:11547.
  • 27
    Maegawa Y, Sugiyama T, Kusaka H, Mitao M, Toyoda N. Screening tests for gestational diabetes in Japan in the 1st and 2nd trimester of pregnancy. Diabetes Res Clin Pract 2003;62:4753.
  • 28
    Mathai M, Thomas TJ, Kuruvila S, Jairaj P. Random plasma glucose and the glucose challenge test in pregnancy. Natl Med J India 1994;7:1602.
  • 29
    Perea-Carrasco R, Pérez-Coronel R, Albusac-Aguilar R, Lombardo-Grifol M, Bassas-Baena de León E, Romero-Diaz C. A simple index for detection of gestational diabetes mellitus. J R Soc Med. 2002; 95:4359.
  • 30
    Perucchini D, Fischer U, Spinas GA, Huch R, Huch A, Lehmann R. Using fasting plasma glucose concentrations to screen for gestational diabetes mellitus: prospective population based study. BMJ 1999;319:8125.
  • 31
    Puavilai G, Chanprasertyotin S, Jirapinyo M. An evaluation of glycosylated hemoglobin measurement by a colorimetric method as a screening test for gestational diabetes mellitus. J Med Assoc Thai 1993;76:54953.
  • 32
    Ramírez-Torres MA, Rodríguez-Pezino J, Zambrana-Castañeda M, Lira-Plascencia J, Parra A. Gestational diabetes mellitus and glucose intolerance among Mexican pregnant adolescents. J Pediatr Endocrinol Metab 2003;16:4015.
  • 33
    Rey E, Hudon L, Michon N, Boucher P, Ethier J, Saint-Louis P. Fasting plasma glucose versus glucose challenge test: screening for gestational diabetes and cost effectiveness. Clin Biochem 2004;37:7804.
  • 34
    Schwartz JG, Phillips WT, Blumhardt MR, Langer O. Use of a more physiologic oral glucose solution during screening for gestational diabetes mellitus. Am J Obstet Gynecol 1994;171:68591.
  • 35
    Sermer M, Naylor CD, Gare DJ, Kenshole AB, Ritchie JW, Farine D, et al. Impact of time since last meal on the gestational glucose challenge test. The Toronto Tri-Hospital Gestational Diabetes Project. Am J Obstet Gynecol 1994;171:60716.
  • 36
    Siribaddana SH, Deshabandu R, Rajapakse D, Silva K, Fernando DJ. The prevalence of gestational diabetes in a Sri Lankan antenatal clinic. Ceylon Med J 1998;43:8891.
  • 37
    Tam WH, Rogers MS, Yip SK, Lau TK, Leung TY. Which screening test is the best for gestational impaired glucose tolerance and gestational diabetes mellitus? Diabetes Care 2000;23:1432.
  • 38
    Thitadilok W, Techatraisak K. The screening for gestational diabetes employing 50 g glucose test. J Med Assoc Thai 1995;78:52631.
  • 39
    Uncu G, Ozan H, Cengiz C. The comparison of 50 grams glucose challenge test, HbA1c and fructosamine levels in diagnosis of gestational diabetes mellitus. Clin Exp Obstet Gynecol 1995;22:2304.
  • 40
    van Leeuwen M, Zweers EJ, Opmeer BC, van Ballegooie E, ter Brugge HG, de Valk HW, et al. Comparison of accuracy measures of two screening tests for gestational diabetes mellitus. Diabetes Care 2007;30:277984.
  • 41
    Weerakiet S, Lertnarkorn K, Panburana P, Pitakitronakorn S, Vesathada K, Wansumrith S. Can adiponectin predict gestational diabetes? Gynecol Endocrinol 2006;22:3628.
  • 42
    Rutjes AW, Reitsma JB, Di Nisio M, Smidt N, van Rijn JC, Bossuyt PM. Evidence of bias and variation in diagnostic accuracy studies. CMAJ 2006;174:46976.
  • 43
    Combs CA, Gunderson E, Kitzmiller JL, Gavin LA, Main EK. Relationship of fetal macrosomia to maternal postprandial glucose control during pregnancy. Diabetes Care 1992;15:12517.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Disclosure of interests
  9. Contribution to authorship
  10. Details of ethics approval
  11. Funding
  12. Acknowledgements
  13. References
  14. Supporting Information

Figure S1. Methodological quality of the studies included in the systematic review of the 50-g glucose challenge test as a screening test for GDM.

Figure S2. Forest plots for all studies, shown for three thresholds that are frequently applied in current clinical practice (7.5, 7.8 and 8.0 mmol/l).

Table S1. Criteria for high or low threshold of the OGTT.

FilenameFormatSizeDescription
BJO_3254_sm_FigS1.pdf109KSupporting info item
BJO_3254_sm_FigS2.pdf431KSupporting info item
BJO_3254_sm_TableS1.pdf8KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.