Screening for Gestational Diabetes: Past, Present & Future PDF

Screening for gestational diabetes; past, present Gestational Review Diabetic DME Blackwell 0742-3071 Oxford, 19 2002 000 Article article UK diabetesLtd, Medicine Science screening Ltd 2002 F. W. F. Hanna & J. R. Peters and future F. W...

Screening for gestational diabetes; past, present Gestational Review Diabetic DME Blackwell 0742-3071 Oxford, 19 2002 000 Article article UK diabetesLtd, Medicine Science screening Ltd 2002 F. W. F. Hanna & J. R. Peters and future F. W. F. Hanna and J. R. Peters* Abstract Prince Charles Hospital, Merthyr Tydfil and Gestational diabetes is carbohydrate intolerance, with onset or first recognition *University Hospital of Wales, Cardiff, UK of hyperglycaemia during pregnancy. Several studies have suggested that gesta- Accepted 19 October 2001 tional hyperglycaemia is associated with adverse maternal and fetal outcomes, promoting the case for screening. Conversely, others argue that screening for gesta- tional diabetes may colour the clinical judgement, influencing further management, e.g. more ‘unjustified’ caesarean sections. Additionally, the lack of definitive data either on a clear-cut glycaemic threshold for the development of adverse outcomes or on the impact of intervention is emphasized by opponents of screening. This review attempts to evaluate the available data on screening for gestational diabetes. Oral glucose tolerance test is promoted on the basis that the diabetogenic stress of pregnancy is encountered during late gestation and is best recognized in the fed state. There are different tests, including the 1 h / 50-g, 2 h / 75-g and 3 h / 100-g tests, with practical limitations, including the time and cost involved and the unpleasant supra-physiological glucose load that is unrelated to body weight, and issues of reproducibility and sensitivity/specificity profiles. Despite its con- venience, the poor sensitivity of random glucose has precluded its routine use for screening. Fasting glucose appears to be promising but further testing is required to ensure satisfactory sensitivity/specificity in different populations. Despite its limitations, the oral glucose tolerance test has become established as the ‘most acceptable’ diagnostic test for gestational diabetes. More convenient methods, e.g. fasting or random or post-load glucose, have to be validated therefore against the oral glucose tolerance test to gain acceptance for routine screening. Diabet. Med 19, 351– 358 (2002) Keywords gestational diabetes, screening, glucose tolerance test insulinaemia. This results in fetal macrosomia (with its attendant Gestational diabetes mellitus obstetric and perinatal complications) together with hypo- Gestational diabetes is defined as carbohydrate intolerance, glycaemia and an increased likelihood of hypocalcaemia, hypo- resulting in hyperglycaemia of variable severity, with onset or magnesaemia, polycythamia and hyperbilirubinaemia. Long-term first recognition during pregnancy, whether or not insulin is maternal complications include postpartum Type 2 diabetes, used and regardless of whether diabetes persists after pregnancy. reported in up to 50% after 10 years. The risk is higher if insulin Unlike established diabetes with onset before pregnancy, hyper- was used and with a macrosomic fetus. Gestational diabetes has glycaemia in gestational diabetes is not established until the been reported to recur in subsequent pregnancies in 20 – 50%. late second trimester, well after organogenesis. Therefore, most Recently, endothelial dysfunction has been reported in obese studies agree that gestational diabetes is not associated with (body mass index (BMI) ≥ 27) and non-obese females with past increased rate of malformations. However, as in pregestational history of gestational diabetes even with normal postpartum diabetes, increased maternal glucose and amino acid levels can glucose tolerance test. This would argue for an underlying result in fetal pancreatic hyperplasia with consequent hyper- subclinical abnormality in insulin sensitivity, decompensating during pregnancy, to result in gestational diabetes, then evolving gradually to frank diabetes in a substantial proportion of females. Correspondence to: Fahmy W. F. Hanna, Consultant Physician, Prince Charles Whether intervention influences the fetal and/or maternal Hospital, Merthyr Tydfil CF47 9DT, UK. outcome remains to be proven in large studies of high index © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358 351  352 Gestational diabetes screening F. W. F. Hanna & J. R. Peters pregnancies. Outcome studies are therefore dependent on Table 1 Risk factors for gestational diabetes first, accepting the need for screening, and second, on evidence Obesity (> 120% ideal body weight) supporting reliable, yet practical, screening protocols for gesta- First-degree relative with diabetes tional diabetes. High-risk ethnic groups (Hispanic, Black, Native American, South-east Asian, Pacific Islander or indigenous Australian) Elevated fasting or random plasma glucose (> 7 mmol/l) Screening for gestational diabetes Glycosuria on two or more occasions Polycystic ovarian syndrome The past Previous impaired glucose tolerance or gestational diabetes, large-for-gestational age baby, or unexplained stillbirth Convenient screening tests, including glycosuria, HbA1c and random plasma glucose (to be discussed later), were not ade- quately sensitive, leading to wider acceptance of the glucose The present tolerance test. The rationale for using a glucose challenge is based on the physiological changes in normal pregnancy. The Currently, the practice and methods for screening for gesta- gestation-induced reduction in insulin sensitivity results in tional diabetes vary widely, not only across the Atlantic, but augmentation of post-prandial glucose levels, with placental also within Europe and even within the UK. In the USA, the glucose transfer to the fetus to aid growth ‘facilitated anabol- American Diabetic Association (ADA) recommends screening ism’. This is accompanied by a 10% reduction in maternal with a 50-g oral glucose load, with measurement of venous fasting glucose levels by the end of the first trimester. The plasma glucose 1 h later as the initial step. If positive, this is diabetogenic stress of pregnancy is encountered during late followed by a 3-h, 100-g glucose tolerance test. Until recently, gestation and is best recognized in the fed state, as most of the NDDG cut-off points were used but those were superseded those with gestational diabetes will demonstrate a normal fast- by the more stringent criteria of Carpenter and Coustan. Most ing glucose value. recently, a 2-h 75-g test has also been included in the ADA Initially, O’Sullivan and Mahan proposed a statistical approach recommendations, with the same fasting, 1-h and 2-h cut points to the diagnosis, designating as abnormal any two of four used in the 3-h test. blood glucose values exceeding 2 SDs above the mean of 100-g The WHO supports the use of a 75-g, 2-h glucose tolerance tolerance tests performed in 752 women during the second test. For those at risk of gestational diabetes (Table 1), it is to and third trimesters. Those criteria examined the likelihood be arranged in the first trimester, otherwise it is recommended of subsequent development of overt diabetes following the between 24 and 28 weeks of gestation. The Diabetic Preg- index pregnancy, but with no link to pregnancy outcomes. nancy Study Group of the European Association for the Study The O’Sullivan criteria were adopted by the National Diabetes of Diabetes recommended reducing the fasting venous plasma Data Group (NDDG). Those applied a factor of 1.14 to con- glucose to 6 mmol /l and increasing the 2-h value to 9 mmol /l, to vert glucose concentrations from the more dilute whole blood be more in line with the physiological changes in glucose toler- haemolysates to plasma (NDDG criteria). ance during pregnancy (Table 2 summarizes the cut-off values The lower cut-off values, proposed by Carpenter and Coustan for various glucose tolerance tests). In the UK there is general lack , were favoured over the NDDG criteria because infants of consensus; most centres employ the 75-g glucose tolerance test, of women who met the lower criteria for gestational diabetes others adopt a two-stage protocol, starting with 50-g screen, to were at similar risk of perinatal morbidity as those identified be followed, if abnormal, with 75-g glucose tolerance test. by the higher and less inclusive NDDG criteria. The subsequent accumulation of follow-up data on women with abnormal glucose tolerance helped to establish the Should we screen for gestational diabetes? glucose tolerance tests as the most ‘acceptable’ way to diag- Screening for gestational diabetes is widely debated as to whether nose gestational diabetes. it is warranted at all or whether universal or selective screening Table 2 Summary of the cut-off values for Glucose load 0h 1h 2h 3h glucose tolerance tests 100 g; NDDG (5) ≥ 5.8 ≥ 10.6 ≥ 9.2 ≥ 8.1 ADA 50 g ≥ 7.8 100 g Carpenter and Coustan ≥ 5.3 ≥ 10 ≥ 8.6 ≥ 7.8 75 g ≥ 5.3 ≥ 10 ≥ 8.6 WHO 75 g ≥7 ≥ 7.8 EASD 75 gram ≥6 ≥9 All values are based on venous plasma glucose values, converted to SI units (mmol/l) where necessary. NDDG, National Diabetes Data Group; ADA, American Diabetes Association; EASD, European Association for the Study of Diabetes. © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  Review article 353 should be adopted, with outcome data being central to the ADA criteria (7.2% vs. 2.4%). Of women positive by the WHO debate. Outcome studies have mainly focused on glucose toler- criteria, 73% were negative by ADA criteria. Conversely, 18% ance tests, being the most acceptable methods for gestational positive by ADA were negative by WHO. After adjustment diabetes. The screening debate, with supporting outcome data, for the effects of age, obesity, ethnicity (as well as for parity is summarized in that section. and educational level in pre-eclampsia and perinatal death), In a prospective analytic cohort study, 3637 pregnancies were gestational diabetes by ADA criteria predicted an increased screened with 50-g followed by 100-g test to assess the maternal risk of macrosomia (RR 1.29), pre-eclampsia (RR 2.28), and and fetal outcome of increased carbohydrate intolerance, not perinatal death (RR 3.1). Increased risk was also identified meeting the criteria of gestational diabetes (by NDDG criteria). employing the WHO criteria with relative risks of 1.45, 1.94 Increasing carbohydrate intolerance, in women without gesta- and 1.59, respectively. Therefore, it was concluded that gesta- tional diabetes, was associated with a graded rise in adverse tional diabetes, based on the 2-h 75-g test and defined by either maternal and fetal outcomes, including caesarean section, pre- WHO or ADA criteria, was associated with adverse pregnancy eclampsia, macrosomia, need for phototherapy and increased outcomes. Not only are they more inclusive, but the WHO cri- maternal and neonatal hospital stay. Multivariate analysis teria also have the added advantage of simplicity, applying the confirmed that increased carbohydrate intolerance is an inde- same criteria of diagnosing diabetes outside of pregnancy. pendent predictor of various unfavourable outcomes. The heterogeneity of study designs and populations com- Employing the NDDG criteria, Weeks and colleagues bined with variations in the obstetric care limit firm conclu- attempted to determine whether gestational diabetes with risk sions about those at highest risk of complications to be drawn. factors (for gestational diabetes) have worse glucose tolerance However, they conclude that gestational hyperglycaemia is and poorer birth outcomes than those without risk factors. They associated with adverse outcomes, supporting the screening found that patients with gestational diabetes are at increased concept. risk for adverse birth outcomes, compared with controls, includ- Paradoxically, other studies raised concern about the so- ing macrosomia (26% vs. 11%, P < 0.01), caesarean section called ‘self-fulfilling prophecy’. Knowledge of the presence (37% vs. 15%, P < 0.01) and shoulder dystochia (9% vs. 2%, of gestational diabetes may colour the clinical judgement, P < 0.05). Adverse outcomes were equally prevalent among influencing further management, e.g. higher rate of caesarean patients with and without risk factors for gestational dia- sections, even in appropriate-for-age infants. Screening betes, arguing against the notion that the high risk factors opponents also emphasize financial implications incurred, alone, rather than gestational diabetes per se, could explain the together with the stressful burden on the health care providers adverse outcomes. Based on these findings, universal, rather as well as pregnant women without clear evidence of sub- than selective screening, was promoted to avoid missing up to sequent benefit [17,18]. The debate will only be resolved by 40% of cases. more research in this area, addressing the unresolved issues A prospective, observational outcome cohort study com- (see below: Screening for gestational diabetes, the future). In pared the NDDG with the ‘modified’ Carpenter and Coustan addition, on validating the more convenient methods, e.g. criteria. The latter identified 50% more cases of gestational fasting or random glucose, against the glucose tolerance test as diabetes that had as much excess in maternal diabetes risk as the most acceptable screening method the latter’s limitations well as adverse fetal outcomes as subjects diagnosed by the have to be acknowledged. NDDG criteria. The more inclusive modified criteria were therefore more helpful in detecting those at risk of adverse outcome. Limitations of the glucose tolerance test In a prospective multicentre trial, mild gestational hyper- There are limitations in all tests of oral glucose tolerance glycaemia (only one abnormal value using the Carpenter and (Table 3). Apart from the time and cost involved, the test Coustan criteria) was more frequently associated with adverse may be unpleasant, especially if used in the first trimester. All maternal and fetal outcome, compared with controls (53.4% the tests constitute a supra-physiological glucose load that is vs. 28.7%; P < 0.01). The rate of large-for-gestational age was unrelated to body weight. In a survey of many residency pro- higher with mild gestational hyperglycaemia (22.1% vs. 11.4%; grammes in the USA, it was found that only 40% and 77% of P < 0.05), remaining significant (P < 0.05) after adjustment for the participating clinicians followed the well-established cut- confounding factors of macrosomia (prepregnancy BMI > 27, off values for the 50 and 100-g tests, respectively. Further maternal age > 35, multiparity and educational level). limitations of oral glucose tolerance testing will be discussed in Most recently, the Brazilian Gestational Diabetes Study has more detail. been reported. This cohort study evaluated the pregnancy out- comes against the diagnostic criteria of both the ADA and the Age limit for screening WHO in almost 5000 women undergoing a 75-g test, at 24 – 28 weeks gestation and followed subsequently. Individuals The American College of Obstetricians and Gynaecologists recom- with hyperglycaemia indicative of diabetes outside pregnancy mends universal screening between 24 and 28 weeks gestation. were excluded. WHO criteria identified more patients than However, the ADA recommends restricting the screening only © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  354 Gestational diabetes screening F. W. F. Hanna & J. R. Peters Table 3 Limitations of the oral glucose Expensive/time-consuming tolerance test Unpleasant (especially with pregnancy!) Not physiological Unrelated to body weight Some clinicians ignore the established cut-off values (up to 60%, see text) Age limit; testing usually limited to > 25 years old, however, recent data argue for high teenage risk (see text) Predictive value of the test varies with ethnic origin (see text) Lack of reproducibility (in up to 24%) Lowering cut-off values to enhance sensitivity invariably jeopardizes specificity (see text) Table 4 Impact of ethnicity on the results of glucose tolerance tests the sensitivity of the 50-g screen (the Staub–Traugott effect). Even the 100-g test was not reproducible in 24%, when Positive 50-g screen, % Positive 100-g test, % repeated 1 week later. Whites 27 17 Blacks 18 43 Sensitivity vs. specificity Asians 41 12 Filipinos 31 12 Lowering the cut-off value of the 50-g screen from 7.8 to 7.2 mmol /l could improve the sensitivity up to 100% but lower The results are expressed as the percentage of each ethnic group yielding positive results to the 50-g screen and the 100-g test. the specificity by 25%. Lowering the 3-h value of the 100-g test from 7.8 to 7.2 mmol/l yielded similar results. Paradoxically, the effect of relaxing the 2-h value of the to those > 25 years old. This practice was found to cut costs 75-g test (from 7.8 to 9 mmol /l) on specificity was evaluated by (by almost 50%) as well as time, with only 5% reduction in comparing the outcome of those with values of 8– 9 mmol/l sensitivity (from 79% to 74%). However, a survey of teenage with those with < 8 mmol /l. Mothers with values between Asian pregnancies in Hong Kong revealed a prevalence of gesta- 8 and 9 mmol /l were older, with higher BMI and had signific- tional diabetes of 5.4%, with a higher incidence of postpartum antly more large-for-date infants. In spite of this, there was haemorrhage (P = 0.01), blood loss at delivery (P = 0.016), no difference in neonatal morbidity, mortality or birth trauma large-for-date infants (P = 0.02) and lower first minute Apgar between the two groups. This suggests that relaxing the cut-off score (P = 0.012). Although these data would argue for value, with its beneficial cost implications, might not jeopardize universal, rather than age-limited, screening, especially in high- the fetal outcome, at least in centres offering equivalent obstetric risk groups, they carry financial and practical limitations. care to that centre in Sweden. Ethnicity and glucose tolerance Discrepancy between various glucose loads Nahum and Huffaker compared the relative prevalence of The predictive values of the WHO (75-g) and NDDG (100-g) gestational diabetes, employing the standard criteria for 50-g tests were compared in a population of 700 pregnant females screen and 100-g test, in a multiethnic group. They demon- in Bangkok. The WHO criteria detected 15.7% as having gesta- strated that the predictive value of glucose challenge varies tional diabetes, with only 1.4% detected by the NDDG (all of significantly with ethnicity (Table 4). The identification whom detected by the WHO criteria). There were 15 macro- and then implementation of different cut-off values in various somic infants, six of whom were identified as belonging to ethnic groups might be pertinent but impractical. gestational diabetic pregnancies, of which only three were identified by the NDDG criteria. One conclusion is that the WHO 2-h 75-g test has a greater predictive value than the more Reproducibility cumbersome NDDG 3-h 100-g test. This study also revealed Ninety percent of normal responses to the 50-g test but only that, even within the same population, different glucose loads 83% of abnormal results were reproducible the next day. vary widely in their predictive value. In a recent study, the sensitivity of the 50-g screen was surpris- ingly low at 59%. This was attributed to the fact that the sensitivity and reproducibility of the test would rely on the timing Other screening methods since last meal. Seventy-three percent of the population had Glycosuria their test < 2 h post-prandial. Successive exposure to carbo- hydrates, i.e. the meal then the glucose load of the test, results Detecting glycosuria is a cheap, convenient and standard nurs- in enhanced insulin sensitivity and glucose disposal, reducing ing procedure. At least it would prevent missing patients with © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  Review article 355 uncontrolled diabetes. However, it has low specificity in view tion as needing to have further glucose tolerance testing. This of the lower renal threshold with gestation. Sutherland and col- approach resulted in marginal improvement of sensitivity only leagues found that 11% of an unselected obstetric population to 29%, with the random glucose screening test identifying of 1418 women had random glycosuria. However, only 1% of only 14 out of 49 women with diabetes. They concluded that those with glycosuria had abnormal glucose tolerance test. random plasma glucose is not an efficient screening method for gestational diabetes and the predictive value would not be improved by altering the cut-off points. That was attributed Glycosylated haemoglobin (HbA1c) to poor correlation between random plasma glucose taken Although convenient and cost-effective, the HbA 1c is not used > 120 min after a meal and the 2-h value of the glucose toler- because of its poor sensitivity , given the length of time ance test (r = 0.22). required to induce changes in the glycosylation of haemoglobin. In addition, Jowett and colleagues investigated the poten- This is particularly relevant to gestational diabetes, where the tial role of random plasma glucose in screening for gestational initial abnormality is only in the post-prandial phase and hyper- diabetes. They calculated the sensitivity and specificity of random glycaemia might not be sustained enough for a change in HbA 1c. plasma glucose by correlating five timed glucose values over a 24-h period (8.00, 12.00, 15.00, 17.00, and 22.00) with 75-g test. Again, their data confirmed that random plasma glucose Random plasma glucose is not sufficiently sensitive for screening for gestational dia- This convenient approach gained popularity in the 1980s. Lind betes. Interestingly, fasting plasma glucose, at a cut-off value of and Anderson used two cut-off values for whole venous 4.1 mmol / l, appeared more encouraging (92% sensitivity and blood (6.4 and 5.8 mmol / l if evaluated within or more than 49% specificity). 2 h post-prandial, respectively). Out of 2403 pregnant women between 28 and 32 weeks gestation, 38 had elevated levels Fasting plasma glucose and proceeded to 75-g glucose tolerance test, of which 13 had values consistent with diabetes according to the WHO criteria This has gained popularity after the greater emphasis on fasting (unlike those mentioned in Table 2, earlier guidelines were plasma glucose in the latest ADA and WHO guidelines for de- implemented, with diabetes mellitus (DM) identified with 2-h tecting DM. Two recent studies (Reichelt et al. 1998 and values ≥ 11.1 mmol / l). Unfortunately, no data were available Peruchini et al. 1999 ) have evaluated the sensitivity and spe- on sensitivity or specificity. cificity of fasting plasma glucose against more standard methods Subsequently, Nasrat and colleagues correlated random plasma for screening (see Table 5 for details). Fasting plasma glucose of glucose with the 75-g test in 250 pregnant women. They 4.9 mmol / l in the first and 4.8 mmol / l in the second study yielded used venous plasma glucose cut-off values of 7.0 and 6.4 mmol / l comparable sensitivity and specificity values despite differences if evaluated within or more than 2 h post-prandial, respectively, in the ethnic origin of their respective populations. by correcting for the whole blood cut-off values designed by More recently, fasting plasma glucose, used in a ‘rule in–rule Lind and Anderson. This revealed a sensitivity of 16%, specifi- out’ algorithm, was investigated in a multiethnic population city of 96%. Alternative cut-off points (90th centile of the study (25.5% Indian subcontinent and 66.8% Arabs). A cut-off population; 6.94 and 5.8 mmol / l, respectively) were subsequently value of < 4.4 mmol / l could rule out gestational diabetes with evaluated, which would have identified 15% of the popula- > 90% sensitivity and > 5.3 mmol / l could rule it in with > 90% Table 5 Summary of two studies evaluating fasting plasma glucose in screening for gestational diabetes Reichelt et al. 1998 Peruchini et al. 1999 Population 5010 females, age > 20 years, 24–28 weeks gestation 520 females, median age 28.4 years, 24–28 weeks gestation Ethnic origin: Ethnic origin: 45% white 63% white 14% African 19% Asian 41% mixed 6% African 12% others Study design Evaluating the sensitivity and specificity of fasting Evaluating the sensitivity and specificity of fasting plasma glucose plasma glucose compared with 75-g test and 50-g screen compared with 100-g test (Carpenter and Coustan) Results Fasting plasma glucose 4.9 mmol/l: Fasting plasma glucose 4.8 mmol/l: Sensitivity 88% Sensitivity 81% Specificity 78% Specificity 76% 22% of the population had values > 4.9 mmol/l, 30% of the population had values > 4.8 mmol/l, needing to proceed to 75-g test needing to proceed to 100-g test Fasting plasma glucose 4.4 mmol/l: Sensitivity 100% Specificity 39% 65% of the population had values > 4.4 mmol/l, needing to proceed to 100-g test © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  356 Gestational diabetes screening F. W. F. Hanna & J. R. Peters Table 6 Sensitivity and specificity of various Sensitivity (%) Specificity (%) screening methods for gestational diabetes High risk features 50 66 Random plasma glucose 40 90 HbA1c 40 90 Fasting plasma glucose (4.8 mmol/l) 81 76 Fasting plasma glucose (4.9 mmol/l) 88 78 Fasting plasma glucose (4.1 mmol/l) 92 44 50-g screen (7.8 mmol/l) 59 91 50-g screen (7.5 mmol/l) 61 88 50-g screen (7.0 mmol/l) 68 82 Glucose tolerance test (NDDG) 79 83 Of the various screening methods, only the fasting plasma glucose appears to be adequately sensitive and specific to closely match that of the glucose tolerance test. specificity. Using those fasting cut-off values had more posit- gestational diabetes, further research to address these unresolved ive predictive values than the 50-g glucose challenge test and issues needs to be undertaken. Whether intervention influences offered a potential to avoid glucose tolerance testing by more the fetal and/or maternal outcome remains to be proven in than 50%. large studies of high index pregnancies. Those outcome studies The concept of fasting plasma glucose as a screening method are dependent on first, accepting the need for screening, and for gestational diabetes is not without shortcomings. Gesta- second on evidence supporting reliable, yet practical, screening tional diabetes can be considered as Type 2 DM but evolving in protocols for gestational diabetes. months, rather than years. The worsening insulin resistance with pregnancy is initially compensated for by augmentation of The Hyperglycaemia and Adverse Pregnancy Outcome insulin production, resulting in normal fasting state. Therefore, (HAPO) Study is an international epidemiological study, involv- as in Type 2 diabetes, the initial abnormality is only post- ing 16 centres in 10 countries, evaluating the effects of raised prandial. Consequently, fasting plasma glucose will miss maternal plasma glucose, less marked than frank diabetes, on some of those in the early ‘post-prandial’ phase. Perucchini pregnancy outcome. HAPO involves measurement of plasma and colleagues could achieve 100% sensitivity by dropping glucose from a 75-g test for patient selection, a random maternal the fasting plasma glucose cut-off to 4.4 mmol / l. However, sample at 34 – 37 weeks gestation, cord blood and neonatal this dropped the specificity by 27%, exposing an extra 35% blood 2 h after birth on all 25 000 participants. The ana- of their population to a glucose tolerance test. This not only lysed pregnancy outcomes will include birth weight, adiposity has unacceptable resource implications, but also demonstrates by skin fold measurement, cord blood C-peptide and early that 100% sensitivity, or nearer, could be calculated only in neonatal hypoglycaemia. The caregiver will be blinded to the retrospect. Even in the non-pregnant state, reports of the limitations result of the glucose tolerance test unless that exceeds current of fasting plasma glucose are accumulating. A recent report criteria for gestational diabetes. The study is in progress and from the DECODE study (Diabetes Epidemiology: Collaborative will report in 2004, hopefully answering the question as to Analysis of Diagnostic Criteria in Europe) has revealed that one whether screening is worthwhile. It should also identify a clear- third of the older diabetic subjects who were undiagnosed at cut threshold for the complications associated with gestational baseline by fasting plasma glucose had an isolated post-challenge diabetes (Professor D. R. Hadden, personal communication). hyperglycaemia. It has been recommended that glucose tolerance To determine their validity, screening tests should be ade- testing for those with impaired fasting glucose (6.1– 6.9 mmol / l) quately sensitive and specific, yet easy and cost-effective. Despite would detect half of those cases. its limitations, the OGTT has been established as the most Table 6 compares the sensitivity and specificity of various acceptable diagnostic test for gestational diabetes. Newer screening methods. In spite of the limitations of the oral glucose screening methods therefore have to be validated against the tolerance test (OGTT), its sensitivity/specificity profile remains much better than the more convenient clinical or baseline bio- OGTT to gain acceptance, and hence, popularity. chemical parameters (except for the recently promoted fasting In spite of the convenience of random plasma glucose, its plasma glucose). poor sensitivity has precluded routine use as a screening method for gestational diabetes. This is based, however, on only one report with the majority of patients having the samples taken Screening for gestational diabetes, > 2 h after their last meal. The lack of correlation with 2-h the future value of OGTT was thought to be the explanation of the poor Unlike Type 2 DM, where all the Wilson’s parameters for sensitivity. Random samples evaluated within 2 h of food screening are fulfilled, gestational diabetes has unresolved issues. appeared to have a better correlation with the 2-h value of the Those are discussed in Table 7, contrasting Type 2 with gesta- glucose tolerance test (r = 0.56). This needs to be evalu- tional diabetes. Rather than dismissing screening for ated prospectively. © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  Review article 357 Table 7 Unlike Type 2 diabetes mellitus, where all the Wilson’s parameters for screening are fulfilled, gestational diabetes has unresolved issues Type 2 diabetes Gestational diabetes Is there evidence suggesting Major cause of blindness, end-stage Up to 50% risk of future diabetes in the mother. that disease detection is important renal disease, lower extremity High risk of perinatal morbidity; including to public health? amputation and is a significant large-for-gestational age cause of morbidity and mortality Is there evidence of increased benefit or Glycaemic control can reduce Some evidence for improved perinatal morbidity, improved prognosis with early detection? diabetes-related complications but probable increased caesarean section rate. No evidence for influencing maternal onset of Type 2 diabetes Is there a safe, efficacious and Screening tests are available for No universally agreed guidelines acceptable method available for both detecting asymptomatic detecting asymptomatic disease? Type 2 diabetes and impaired glucose tolerance which, it has been suggested, is a precursor to Type 2 diabetes Is there evidence that screening and Evidence for Type 2 diabetes from the UKPDS Needs to be investigated related diagnostic and intervention services Evidence for intensive management in will have a positive impact on the allocation Type 1 diabetes will result in an average of of health resources? 5 years increase in life expectancy, 8 years of preserved vision, 6 years free of kidney disease and 6 years free from amputations and nerve damage Is there evidence supporting the cost effectiveness of screening when weighed against alternative strategies? Fasting glucose appears to be promising, but further testing 4 O’Sullivan JB, Mahan CM. Criteria for the oral glucose tolerance test is required to ensure satisfactory sensitivity/specificity in various in pregnancy. Diabetes 1964; 13: 278–285. 5 National Diabetes Data Group. Classification and diagnosis of dia- populations. If adopted, antenatal clinics should ensure that betes mellitus and other categories of glucose intolerance. Diabetes the fasting glucose samples are taken promptly upon arrival 1979; 18: 1039–1057. in the clinic. Patient education will enhance co-operation and 6 Carpenter MW, Coustan DR. Criteria for screening tests for gesta- compliance. tional diabetes. Am J Obstet Gynecol 1982; 144: 768–773. It would be relatively convenient to correlate either fasting 7 Magee MS, Walden CE, Benedetti TJ, Knopp RH. Influence of diag- nostic criteria on the incidence of gestational diabetes and perinatal or random plasma glucose with the glucose tolerance test. morbidity. JAMA 1993; 269: 609–615. However, to include both in a study, comparing their respect- 8 O’Sullivan JB, Mahan CM, Charles D, Dandrow RV. Screening cri- ive sensitivity/specificity profile with glucose tolerance test, would teria for high risk gestational diabetic patients. Am J Obstet Gynecol yield more information. We are currently in the process of co- 1973; 116: 895–900. ordinating a multicentre trial to evaluate those screening options. 9 American Diabetes Association. Gestational diabetes mellitus. Diabetes Care 2000; 23: S77–S79. 10 WHO Consultation. Definition, Diagnosis and Classification of Acknowledgement Diabetes Mellitus and its Complications: Report of a WHO Con- sultation. Part 1: Diagnosis and Classification of Diabetes Mellitus. The authors are grateful for the expert review and input of WHO/NCD/NCS/99.2, Geneva: World Health Organization, 1999. Professor David R. Hadden, Royal Victoria Hospital, Belfast. 11 Report of the Pregnancy and Neonatal Care Group of the European Association for the Study of Diabetes. Diabet Med 1996; 13: S43– S53. 12 Sermer M, Naylor CD, Gare DJ, Kenshole AB, Ritchie JWK, Farine D References et al. Impact of increasing carbohydrate intolerance on maternal-fetal 1 Anastasiou E, Lekakis JP, Alevizaki M, Papamichael CM, Megas J, outcomes in 3637 women without gestational diabetes. The Toronto Souvvatzoglou A et al. Impaired endothelium-dependent vasodilata- Tri-Hospital Gestational Diabetes Project. Am J Obstet Gynecol 1995; tion in women with previous gestational diabetes. Diabetes Care 173: 146–156. 1998; 21: 2111–2115. 13 Weeks JW, Major CA, de Veciana M, Morgan MA. Gestational dia- 2 Freinkel N. Banting Lecture 1980: of pregnancy and progeny. Diabetes betes: does the presence of risk factors influence perinatal outcome? 1980; 29: 1023–1035. Am J Obstet Gynecol 1994; 171: 1003–1007. 3 Kalkhoff RK, Kissebah AH, Kim HJ. Carbohydrate and lipid meta- 14 Vambergue A, Nuttens MC, Verier-Mine O, Dognin C, Cappoen JP, bolism during normal pregnancy: relationship to gestational hormone Fontaine P. Is mild gestational hyperglycaemia associated with action. In Merkatz IR, Adam PAJ eds. The Diabetic Pregnancy: a maternal and neonatal complications? The Diagest Study. Diabet Perinatal Perspective. New York: Grune & Stratton, 1979. Med 2000; 17: 203–208. © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358  358 Gestational diabetes screening F. W. F. Hanna & J. R. Peters 15 Schmidt MI, Duncan BB, Reichelt AJ, Branchtein L, Matos MC, for Gestational Diabetes Study Group of the Lombardy section of Costa e Forti A et al. for the Brazilian Gestational Diabetes Study the Italian Society of Diabetology. Am J Obstet Gynaecol 1998; 179: Group. Gestational diabetes mellitus diagnosed with a 2-h 75-g oral 179–185. glucose tolerance test and adverse pregnancy outcomes. Diabetes 28 Nord E, Hanson U, Persson B. Blood glucose limits in the diagnosis of Care 2001; 24: 1151–1155. impaired glucose tolerance during pregnancy: relation to morbidity. 16 Naylor CD, Sermer M, Chen E, Sykora K for the Toronto Tri- Acta Obstet Gynaecol Scand 1995; 74: 589–593. Hospital Gestational Diabetes Investigators. Cesarean delivery in rela- 29 Deerochanawong C, Putiyanun C, Wongsuryrat M, Serirat S, Jinayon P. tion to birth weight and gestational glucose tolerance: pathophysiology Comparison of National Diabetes Data Group and World Health or practice style? JAMA 1996; 275: 1165–1170. Organisation criteria for detecting gestational diabetes mellitus. 17 Walkinshaw SA. Diet and insulin vs diet-alone for gestational dia- Diabetologia 1996; 39: 1070–1073. betes. In Enkin MW, Keirse MJNC, Renfre MJ, Neilson JP eds. 30 Sutherland HW, Stowers JM, McKenzie C. Simplifying the clinical Pregnancy and Childbirth Module of the Cochrane Database of problem of glycosuria in pregnancy. Lancet 1970; i: 1069–1071. Systematic Reviews. The Cochrane collaboration, Issue 2. Oxford, 31 Cousins L, Dattel BJ, Holligsworth DR, Zettner A. Glycosylated UK: 1995. haemoglobin as a screening test for carbohydrate intolerance in preg- 18 Walkinshaw SA. Dietary regulation for gestational diabetes. In nancy. Am J Obstet Gynecol 1984; 150: 455–460. Neilson JP, Crowther CA, Hodnett ED, Hofmeyr GJ, Keirse MJNC 32 Lind T, Anderson J. Does random blood glucose sampling outdate eds. Pregnancy and Childbirth Module of the Cochrane Database of testing for glycosuria in the detection of diabetes during pregnancy? Systematic Reviews. The Cochrane collaboration, Issue 3. Oxford, Br Med J 1984; 289: 1569–1571. UK: 1997. 33 Nasrat AA, Johnnstone FD, Hasan SAM. Is random plasma glucose 19 Owen J, Phelan ST, Landon MB, Gabbe SG. Gestational diabetes an efficient screening test for abnormal glucose tolerance in pregnancy? survey. Am J Obstet Gynaecol 1995; 172: 615–620. Br J Obstet Gynaecol 1988; 95: 855–860. 20 Lao TT, Ho LF, Liu KL. Gestational diabetes mellitus in teenage 34 Jowett NI, Samanta AK, Burden AC. Screening for diabetes in preg- pregnancy: a case-control study. Diabet Med 1998; 15: 1036–1038. nancy: is random blood glucose enough? Diabet Med 1987; 4: 160– 21 Nahum GG, Huffaker BJ. Racial differences in oral glucose screening 163. test results: establishing race-specific criteria for abnormality in preg- 35 Reichelt AJ, Spichler ER, Branchtein L, Nucci LB, Franco LJ, nancy. Obstet Gynaecol 1993; 81: 517–522. Schmidt MI. Fasting plasma glucose is a useful test for detection of 22 Espinosa de los Monteros A, Parra A, Carino N, Ramirez A. The gestational diabetes. Diabetes Care 1998; 21: 1246–1249. reproducibility of the 50-g 1-hour glucose screen for diabetes in preg- 36 Agarwal MM, Hughest PF, Punnose J, Ezimokhai M. Fasting plasma nancy. Obstet Gynaecol 1993; 82: 515–518. glucose as a screening test for gestational diabetes in a multi-ethnic, 23 Perucchini D, Fischer U, Spinas GA, Huch R, Huch A, Lehmann R. high-risk population. Diabet Med 2000; 17: 720–726. Using fasting plasma glucose concentrations to screen for gestational 37 Pendergrass M, Fazoni E, DeFronzo R. Non-insulin-dependent dia- diabetes mellitus: prospective population based study. Br Med J betes mellitus and gestational diabetes mellitus: same disease another 1999; 319: 812–815. name? Diabetes Rev 1995; 3: 566. 24 Lewis GF, McNally C, Blackman JD, Polonsky KS, Barron WM. 38 The DECODE Study Group, on behalf of the European Diabetes Prior feeding alters the response to the 50-g glucose challenge test in Epidemiology Group. Consequences of the new diagnostic criteria pregnancy. The Staub-Traugott effect revisited. Diabetes Care 1993; for diabetes in older men and women. Diabetes Care 1999; 22: 1667– 16: 1551–1556. 1671. 25 Catalano PM, Avallone DA, Drago NM, Amini SB. Reproducibility 39 Criteria for determining the appropriateness of implementing health of the oral glucose tolerance test in pregnant women. Am J Obstet screening. http://www3.pitt.edu~dlnst5/screening.html Gynaecol 1993; 169: 874–881. 40 Elisabeth R, Trimble E for the HAPO Cooperative Research Group. 26 American College of Obstetricians and Gynaecologists. Diabetes and Hyperglycaemia and Adverse Pregnancy Outcome (HAPO) Study: Pregnancy. Technical Bulletin no. 200, December, 1994. Achieving uniformity for plasma glucose measurements. Diabetes 27 Bonomo M, Gandini ML, Mastropasqua A, Begher C, Valentini U, 2001; 50 (Suppl. 2): A539. Faden D et al. Which cut-off level should be used in screening for 41 Kjos SL, Buchanan TA. Gestational diabetes mellitus. N Eng J Med glucose intolerance in pregnancy? Definition of Screening Methods 1999; 341: 1749–1756. © 2002 Diabetes UK. Diabetic Medicine, 19, 351–358