ACOG Practice Bulletin: Clinical Management Guidelines for Obstetricians-Gynecologists PDF

interim update ACOG P RACTICE BULLET IN Clinical Management Guidelines for Obstetrician–Gynecologists Number 190, February 2018 (Replaces Practice Bulletin Number 180, July 2017) Committee on Practice Bulletins—Obstetrics. This P...

interim update ACOG P RACTICE BULLET IN Clinical Management Guidelines for Obstetrician–Gynecologists Number 190, February 2018 (Replaces Practice Bulletin Number 180, July 2017) Committee on Practice Bulletins—Obstetrics. This Practice Bulletin was developed by the American College of Obstetricians and Gynecologists Committee on Practice Bulletins—Obstetrics with the assistance of Aaron B. Caughey, MD, PhD, and Mark Turrentine, MD. INTERIM UPDATE: This Practice Bulletin is updated as highlighted to reflect a limited, focused change to clarify and pro- vide additional information on the pharmacologic treatment of gestational diabetes mellitus. Gestational Diabetes Mellitus Gestational diabetes mellitus (GDM) is one of the most common medical complications of pregnancy. However, debate continues to surround the diagnosis and treatment of GDM despite several recent large-scale studies address- ing these issues. The purposes of this document are the following: 1) provide a brief overview of the understanding of GDM, 2) review management guidelines that have been validated by appropriately conducted clinical research, and 3) identify gaps in current knowledge toward which future research can be directed. Background sity and sedentary lifestyles, the prevalence of GDM among reproductive-aged women is increasing globally. Definition and Prevalence Maternal and Fetal Complications Gestational diabetes mellitus is a condition in which carbohydrate intolerance develops during pregnancy. Women with GDM have a higher risk of developing Gestational diabetes that is adequately controlled with- preeclampsia (9.8% in those with a fasting glucose out medication is often termed diet-controlled GDM or less than 115 mg/dL and 18% in those with a fast- class A1GDM. Gestational diabetes mellitus that requires ing glucose greater than or equal to 115 mg/dL) and medication to achieve euglycemia is often termed class undergoing a cesarean delivery (25% of women with A2GDM. Because many women do not receive screening GDM who require medication and 17% of women with for diabetes mellitus before pregnancy, it can be chal- diet-controlled GDM underwent cesarean delivery ver- lenging to distinguish GDM from preexisting diabetes. sus 9.5% of controls) (4, 5). Furthermore, women with However, it has been estimated that in 2009, 7% of preg- GDM have an increased risk of developing diabetes nancies were complicated by any type of diabetes and that (predominantly type 2 diabetes) later in life. It is esti- approximately 86% of these cases represented women mated that up to 70% of women with GDM will develop with GDM (1). Additionally, the prevalence of GDM var- diabetes within 22–28 years after pregnancy (6–8). The ies in direct proportion to the prevalence of type 2 diabetes progression to diabetes also is influenced by race, ethnic- in a given population or racial or ethnic group. Caucasian ity, and obesity. For example, 60% of Latin American women generally have the lowest rates of GDM. There is women with GDM may develop type 2 diabetes within an increased prevalence of GDM among Hispanic, African 5 years of their index pregnancy (9). American, Native American, and Asian or Pacific Islander The offspring of women with GDM are at increased women (2). Gestational diabetes also increases with the risk of macrosomia, neonatal hypoglycemia, hyperbiliru- same risk factors seen for type 2 diabetes such as obesity binemia, shoulder dystocia, and birth trauma. There also and increased age (3). With a greater prevalence of obe- is an increased risk of stillbirth, although how much this VOL. 131, NO. 2, FEBRUARY 2018 OBSTETRICS & GYNECOLOGY e49 is related to glycemic control is debated (10). The results factors, including those with a prior history of GDM of the Hyperglycemia and Adverse Pregnancy Outcome (see Box 1) (16, 20). However, the best test for early study (HAPO), an international, multicenter study, dem- GDM or type 2 diabetes screening is not clear. The test- onstrated a continuous relationship between maternal ing used to diagnose type 2 diabetes in nonpregnant indi- glucose levels on each of the three values of the 75-g, viduals (ie, a fasting blood glucose followed by a 75-g 2-hour oral glucose tolerance test (OGTT) and cesarean glucose load and a 2-hour plasma glucose measurement) delivery, birth weight greater than the 90th percentile, could be used for early pregnancy screening (21). Many clinical neonatal hypoglycemia, and fetal hyperinsu- obstetricians or obstetric care providers use the two- linemia (11). Other studies have demonstrated that fetal step screening process that is used for GDM and start exposure to maternal diabetes contributes to childhood with a 50-g OGTT. The American Diabetes Association and adult-onset obesity and diabetes in offspring, which (ADA) has noted that measurement of hemoglobin A1C is independent of risks associated with obesity and also can be used, but it may not be suitable for use alone genetic predisposition (12, 13). because of decreased sensitivity compared with OGTT approaches (20). Even if the results of early testing are Screening Practices, Diagnostic negative, GDM screening still is recommended at 24–28 Thresholds, and Treatment Benefits weeks of gestation because of the large proportion of Historically, screening for GDM consisted of obtaining women who had negative early pregnancy screening but the patient’s medical history and focused primarily on who will go on to develop GDM (22). In women who past obstetric outcomes and a family medical history of have positive 50-g screening test results, but negative type 2 diabetes. A 1973 study proposed the use of the follow-up test results early in pregnancy, it is common 50-g, 1-hour OGTT as a screening tool for GDM (14). to use the follow-up test at 24–28 weeks of gestation This test has since become widely accepted, and 95% of without repeating the 50-g screening test. obstetricians in the United States use it as the tool for The two-step approach to testing for GDM that is universal screening of pregnant women (15, 16). commonly used in the United States is based on first The use of historic factors (family or personal his- screening with the administration of a 50-g oral glucose tory of diabetes, previous adverse pregnancy outcome, solution followed by a 1-hour venous glucose determi- glycosuria, and obesity) to identify GDM will fail to nation. Women whose glucose levels meet or exceed an identify approximately one half of women with GDM institution’s screening threshold then undergo a 100-g, (17). Although certain factors place women at low risk 3-hour diagnostic OGTT. Gestational diabetes mellitus of GDM, it may not be cost effective to screen that group is most often diagnosed in women who have two or more of women with glucose tolerance testing. However, such abnormal values on the 3-hour OGTT. low-risk women represent only 10% of pregnant women Institutional screening thresholds for the 1-hour and identifying those who should not be screened may glucose challenge vary from 130 mg/dL to 140 mg/dL, add unnecessary complexity to the screening process with a range of sensitivities and specificities reported. (18). Therefore, in 2014, the U.S. Preventive Services However, there are no randomized trials that have Task Force made a recommendation to screen all preg- examined whether one cutoff is more effective than oth- nant women for GDM at or beyond 24 weeks of gesta- ers. Data regarding the ideal threshold value to screen tion (16). for gestational diabetes in order to improve pregnancy outcomes also are insufficient, although standardization of a screening threshold has been recommended (23). Clinical Considerations and For example, one cohort study showed that a value of 140 mg/dL had lower false-positive rates and improved Recommendations positive predictive values across various racial and How is gestational diabetes mellitus ethnic groups. This analysis also showed that sensitivi- ties were only marginally improved when using lower diagnosed? thresholds (ie, 130 mg/dL and 135 mg/dL) (24). Using All pregnant women should be screened for GDM with a higher standardized threshold of 140 mg/dL may a laboratory-based screening test(s) using blood glucose lower the rate of false-positive screening test results and levels. Screening for GDM generally is performed at unnecessary administration of the 3-hour OGTTs, which 24–28 weeks of gestation (19). Early pregnancy screen- has been shown to be associated with increased mater- ing for undiagnosed type 2 diabetes, preferably at the nal stress and dissatisfaction regarding the process of initiation of prenatal care, is suggested in over- screening for and diagnosing GDM (25–27). However, weight and obese women with additional diabetic risk in the absence of clear evidence that supports one e50 Practice Bulletin Gestational Diabetes Mellitus OBSTETRICS & GYNECOLOGY cutoff value over another (ie, 130 mg/dL, 135 mg/dL, or 140 mg/dL) for the 1-hour glucose screening test, Box 1. Screening Strategy for obstetricians and obstetric care providers may select one Detecting Pregestational Diabetes or of these as a single consistent cutoff for their practice, Early Gestational Diabetes Mellitus ^ using factors such as community prevalence rates of Consider testing in all women who are overweight or GDM when making their decision. obese (ie, have a body mass index greater than 25 or Different cutoffs for the 3-hour OGTT also have greater than 23 in Asian Americans) and have one been proposed. Table 1 (19, 20) lists the diagnostic or more of the following additional risk factors: thresholds established for the 3-hour OGTT by the Physical inactivity National Diabetes Data Group and by Carpenter and First-degree relative with diabetes Coustan, with the latter using lower thresholds that High-risk race or ethnicity (eg, African American, subsequently result in higher rates of GDM diagnosis Latino, Native American, Asian American, Pacific (20). In the absence of clear comparative trials, one set Islander) of diagnostic criteria for the 3-hour OGTT cannot be Have previously given birth to an infant weighing clearly recommended over the other. For example, in a 4,000g (approximately 9 lb) or more cross-sectional study that compared the two sets of crite- Previous gestational diabetes mellitus ria in more than 26,000 women found that the diagnosis Hypertension (140/90 mm Hg or on therapy for of GDM increased on average by 50% with the use of hypertension) the Carpenter and Coustan thresholds (28). However, a High-density lipoprotein cholesterol level less than study that examined the clinical outcomes showed that 35 mg/dL (0.90 mmol/L), a triglyceride level greater the women in whom GDM would have been incremen- than 250 mg/dL (2.82 mmol/L) tally diagnosed by the Carpenter and Coustan criteria Women with polycystic ovarian syndrome alone had higher rates of perinatal complications than the A1C greater than or equal to 5.7%, impaired glucose women with values below these diagnostic thresholds tolerance, or impaired fasting glucose on previous (29). Women who have even one abnormal value on the testing 100-g, 3-hour OGTT have a significantly increased risk Other clinical conditions associated with insulin of adverse perinatal outcomes compared with women resistance (eg, prepregnancy body mass index greater without GDM (29). Although a higher level of scrutiny than 40 kg/m2, acanthosis nigricans) may be focused on this subset of women, further research History of cardiovascular disease is needed to clarify the risk of adverse outcomes in If pregestational or gestational diabetes mellitus is not patients with one abnormal value on the 100-g, 3-hour diagnosed, blood glucose testing should be repeated at OGTT and whether they would benefit from treatment. 24–28 weeks of gestation. Given the benefits of standardization, practitioners Adapted with permission from the American Diabetes Association. and institutions should select a single set of diagnostic Classification and Diagnosis of Diabetes. Diabetes Care 2017;40 criteria, either plasma or serum glucose levels desig- (Suppl. 1):S11–S24. Copyright 2017 American Diabetes Association. nated by the Carpenter and Coustan criteria or the Table 1. Proposed Diagnostic Criteria for Gestational Diabetes Mellitus* ^ Plasma or Serum Glucose Plasma Level Level Carpenter and National Diabetes Coustan Conversion Data Group Conversion Status mg/dL mmol/L mg/dL mmol/L Fasting 95 5.3 105 5.8 1 hour 180 10.0 190 10.6 2 hours 155 8.6 165 9.2 3 hours 140 7.8 145 8.0 *A diagnosis generally requires that two or more thresholds be met or exceeded, although some clinicians choose to use just one elevated value. Adapted with permission from the American Diabetes Association. Classification and Diagnosis of Diabetes. Diabetes Care 2017;40 (Suppl. 1):S11–S24. Copyright 2017 American Diabetes Association. VOL. 131, NO. 2, FEBRUARY 2018 Practice Bulletin Gestational Diabetes Mellitus e51 plasma levels established by the National Diabetes Data be studied before they are proposed at a national level. Group, for consistent use within their patient popula- However, individual practices and institutions may tions. Considerations for selection of one set of diag- choose to use the IADPSG’s recommendation, if appro- nostic criteria over the other could include, but are not priate, for the population they serve. limited to, the baseline prevalence of diabetes in their specific communities and the availability of resources What are the benefits of treating gestational to appropriately manage women in whom GDM will diabetes mellitus? be diagnosed by any given protocol. This approach, The 2005 Australian Carbohydrate Intolerance Study although imperfect, avoids establishment of a single set in Pregnant Women trial, the first large-scale (1,000 of diagnostic criteria across all populations based on women), randomized treatment trial for GDM (34) expert opinion alone. found the treatment was associated with a significant A one-step approach to establishing the diagnosis reduction in the rate of the primary outcome, a compos- of GDM using a 75-g, 2-hour OGTT has been used ite of serious newborn complications (perinatal death, and promoted by other organizations. For example, in shoulder dystocia, and birth trauma, including fracture 2010, the International Association of Diabetes and or nerve palsy). Treatment also reduced preeclampsia Pregnancy Study Group (IADPSG) recommended that (from 18% to 12%) as well as reduced the frequency of a universal 75-g, 2-hour OGTT be performed during infants who were large for gestational age (LGA) (from pregnancy and that the diagnosis of GDM be established 22% to 13%) and who had a birth weight greater than when any single threshold value was met or exceeded 4,000 g (from 21% to 10%). A subsequent random- (fasting value, 92 mg/dL; 1-hour value, 180 mg/dL; ized, multicenter treatment trial of 958 women with or 2-hour value, 153 mg/dL) (30). Overall, using the mild GDM conducted in the United States found that, proposed IADPSG criteria would identify approxi- although there were no differences in the frequency of mately 18% of pregnant women in the United States the primary composite outcome (perinatal death, neo- as having GDM; in some subpopulations, the propor- natal hypoglycemia, elevated umbilical cord C-peptide tion of women in whom GDM is diagnosed would be level, and birth trauma), several significant differences even higher. In 2011, the ADA endorsed these criteria in secondary outcomes were observed with treatment, while acknowledging that adopting these cutoffs would including a lower frequency of LGA infants, lower fre- significantly increase the prevalence of GDM (31). quency of birth weight exceeding 4,000 g, and reduced The additional women in whom GDM would be diag- neonatal fat mass (35). Moreover, the rates of cesarean nosed may be at a lower risk of adverse outcomes delivery, shoulder dystocia, and hypertensive disorders than and may not derive similar benefits from diag- were significantly reduced in women who were treated nosis and treatment as women in whom GDM was for GDM. A U.S. Preventive Services Task Force sys- diagnosed by traditional criteria (32). As of 2017, the tematic review underscored the demonstrated benefits ADA continues to recognize that there is an absence of of treating GDM and highlighted the reduced risks of clear evidence that supports the IADPSG-recommended preeclampsia, shoulder dystocia, and macrosomia (36). approach versus the more traditional two-step screening The treatment in such studies has consisted of dietary approach (20). counseling with specific nutritional approaches (37–39) In 2013, a Eunice Kennedy Shriver National Institute and exercise (40, 41). Based on this evidence, women in of Child Health and Human Development Consensus whom GDM is diagnosed should receive nutrition and Development Conference on Diagnosing Gestational exercise counseling, and when this fails to adequately Diabetes recommended that obstetricians and obstetric control glucose levels, medication should be used for care providers continue to use a two-step approach to maternal and fetal benefit. It is important to note that screen for and diagnose GDM. The report underscored in both trials described above, women with elevated the lack of evidence that the use of the one-step 75-g, glucose values were treated with insulin, not oral agents, 2-hour OGTT to diagnose GDM leads to clinically when medical nutrition treatment did not control glucose significant improvements in maternal or newborn out- values. comes and highlighted the significant increase in health care costs that would result (23). Additionally, a 2015 How should blood glucose be monitored in a Cochrane review supported that no specific screening woman with gestational diabetes mellitus? strategy has been shown to be optimal (33). In light of this, the American College of Obstetricians and Once a woman with GDM begins nutrition therapy Gynecologists (ACOG) supports the two-step process (dietary counseling), surveillance of blood glucose lev- and recommends that implications of suggested changes els is required to confirm that glycemic control has been e52 Practice Bulletin Gestational Diabetes Mellitus OBSTETRICS & GYNECOLOGY established. However, there is insufficient evidence to demonstrated a reduction in large-for-gestational-age define the optimal frequency of blood glucose testing neonates, macrosomia (defined as 4,000 g or more), and in women with GDM. Based on the data available, the neonatal fat mass in neonates born to women random- general recommendation is for daily glucose monitoring ized to lifestyle interventions (48). Additionally, women four times a day, once after fasting and again after each randomized to the lifestyle interventions were more meal. likely to meet postpartum weight goals 1 year after preg- Mean fasting glucose values may be useful for nancy. Despite these promising findings, the specific managing diabetes in pregnant women because they are dietary and exercise approaches are less well studied. predictive of increased neonatal fat mass in the women’s The goal of medical nutrition therapy in women offspring. Neonatal fat mass has been shown to be with GDM is to achieve normal blood glucose levels, associated with the development of childhood obe- prevent ketosis, provide adequate weight gain, and sity and diabetes (42). Another study, a randomized contribute to appropriate fetal growth and develop- controlled trial that compared the value of prepran- ment. The ADA recommends nutritional counseling by dial versus postprandial measurements for blood glu- a registered dietitian and development of a personal- cose monitoring of women with GDM, showed that use ized nutrition plan based on the individual’s body mass of the 1-hour postprandial measurement was associated index for all patients with GDM (19). In some clinical with better glycemic control, a lower incidence of LGA settings in which a dietitian is not readily available, a infants, and lower rates of cesarean delivery for cepha- clinician should be able to provide recommendations to lopelvic disproportion (43). Given this evidence, fasting the patient based on three major nutritional components: and postprandial values should be used for monitor- 1) caloric allotment, 2) carbohydrate intake, and 3) ing blood glucose in women with GDM. Assessment of caloric distribution. postprandial blood glucose can be undertaken at either A diet composed of 50–60% carbohydrates often 1 hour or 2 hours after meals. No study to date has will result in excessive weight gain and postpran- demonstrated the superiority of either approach (44–46), dial hyperglycemia. Therefore, it has been suggested and this may be because postprandial glucose peaks that carbohydrate intake be limited to 33–40% of at approximately 90 minutes, between the two time calories, with the remaining calories divided between points (47). protein (20%) and fat (40%) (49); however, the actu- Once the patient’s glucose levels are well controlled al dietary composition that optimizes perinatal out- by diet, the frequency of glucose monitoring may be comes is unknown. For example, a randomized trial of modified depending on gestational age, overall concerns 99 women with GDM that compared a low-glycemic for adherence, and likely need for future adjustments to index nutrition plan with a conventional high-fiber diet care. It is unusual to recommend obtaining fewer than found that both produced similar pregnancy outcomes two measurements per day. (39). A small, recent randomized trial demonstrated In addition, no controlled trials have been per- that women with GDM randomized to a complex car- formed to identify optimal glycemic targets. The ADA bohydrate diet had lower fasting glucose values as and ACOG recommend that fasting or preprandial compared with those on a conventional diet (50). Given blood glucose values be below 95 mg/dL and post- these findings and the results of other treatment trials, prandial blood glucose values be below 140 mg/dL at complex carbohydrates are recommended over simple 1 hour or 120 mg/dL at 2 hours to reduce the risk of carbohydrates because they are digested more slowly, macrosomia (19). Generally, these values are reviewed are less likely to produce significant postprandial hyper- weekly; however, when there are many abnormal val- glycemia, and potentially reduce insulin resistance (38). ues, more frequent review is common. Alternatively, There is little evidence evaluating or supporting differ- with stable, normal values, less frequent review is ent dietary approaches to the treatment of GDM (37). acceptable. In practice, three meals and two to three snacks are recommended to distribute carbohydrate intake and to What nonpharmacologic treatments are reduce postprandial glucose fluctuations. effective in managing gestational diabetes Although there are multiple randomized trials that have examined exercise and lifestyle interventions in mellitus? adults with diabetes who are not pregnant, there are Most commonly, GDM management begins with the few published exercise trials in women with GDM. nonpharmacologic approaches of dietary modifications, Even though most of these published trials have small exercise, and glucose monitoring. A recent meta-analysis sample sizes, they do appear to show improvement of lifestyle modification trials in women with GDM in glucose levels (40, 51–54). In adults with diabetes VOL. 131, NO. 2, FEBRUARY 2018 Practice Bulletin Gestational Diabetes Mellitus e53 who are not pregnant, exercise—particularly weight intermediate-acting insulin, NPH insulin has been the training—increases lean muscle mass and improves tis- mainstay, but more recently insulin glargine and insulin sue sensitivity to insulin. In overweight or obese women detemir have been described for long-acting use (57– with GDM, exercise also may be able to improve glyce- 59). For short-acting insulin, insulin analogues—includ- mic control. Therefore, a moderate exercise program is ing insulin lispro and insulin aspart—have been used recommended as part of the treatment plan for women in pregnancy, and these insulin analogues do not cross with GDM (19). Such a plan should mirror diabetes the placenta. Insulin lispro and insulin aspart should be care in general, and women with GDM should aim for used preferentially over regular insulin because both 30 minutes of moderate-intensity aerobic exercise at have a more rapid onset of action, enabling the patient least 5 days a week or a minimum of 150 minutes to administer her insulin right at the time of a meal per week (31). Simple exercise such as walking for rather than 10–15 minutes before an anticipated meal. 10–15 minutes after each meal can lead to improved This provides better glycemic control and helps avoid glycemic control and is commonly recommended (55). hypoglycemic episodes from errors in timing (60, 61) (Table 2). What pharmacologic treatments are effective in managing gestational diabetes mellitus? Oral Antidiabetic Medications Oral antidiabetic medications (eg, metformin and gly- Pharmacologic treatment is recommended when target buride) increasingly are being used among women with glucose levels cannot be consistently achieved through GDM, despite the fact that they have not been approved nutrition therapy and exercise. However, a system- by the U.S. Food and Drug Administration for this indi- atic review found no conclusive evidence for a spe- cation (62) and even though insulin continues to be the cific threshold value at which medical therapy should ADA-recommended first-line therapy (19). be started (56). Insulin historically has been considered Metformin is a biguanide that inhibits hepatic the standard therapy for GDM management in cases gluconeogenesis and glucose absorption and stimu- refractory to nutrition therapy and exercise and this has lates glucose uptake in peripheral tissues. Historically, continued to be reinforced by the ADA (19). metformin primarily has been used in women with pre- Insulin, which does not cross the placenta, can gestational diabetes or those with polycystic ovary syn- achieve tight metabolic control and traditionally has drome and infertility. In women with polycystic ovary been added to nutrition therapy if fasting blood glucose syndrome, metformin is often continued until the end levels consistently are greater than or equal to 95 mg/dL, of the first trimester, despite only limited evidence to if 1-hour levels consistently are greater than or equal to suggest that such use decreases the risks of adverse 140 mg/dL, or if 2-hour levels consistently are greater pregnancy outcomes, including first-trimester loss (63). than or equal to 120 mg/dL. These thresholds largely have Metformin crosses the placenta with levels that can been extrapolated from recommendations for managing be as high as maternal concentrations. The long-term pregnancy in women with preexisting diabetes. If insulin metabolic influence on the offspring is unknown (64); is used throughout the day in women in whom fasting however, one recent study found similar developmental and postprandial hyperglycemia are present after most outcomes by 2 years of age (65). This concern about the meals, a typical starting total dosage is 0.7–1.0 units/kg daily. This dosage should be divided with a regimen of Table 2. Action Profile of Commonly Used Insulin Agents ^ multiple injections using long-acting or intermediate- acting insulin in combination with short-acting insulin. Onset of Peak of Duration of Type Action Action (h) Action (h) However, if there are only isolated abnormal values at a specific time of day, focusing the insulin regimen Insulin lispro 1–15 min 1–2 4–5 to correct the specific hyperglycemia is preferred. For Insulin aspart 1–15 min 1–2 4–5 example, in women with only elevated fasting values, Regular insulin 30–60 min 2–4 6–8 nighttime administration of intermediate-acting insulin, Isophane insulin 1–3 h 5–7 13–18 such as NPH insulin, may be adequate. Similarly, in suspension women with elevated values only for breakfast postpran- (NPH insulin) dial, short-acting insulin before breakfast may be the Insulin glargine 1–2 h No peak 24 only insulin needed. Regardless of the starting dosage, Insulin detemir 1–3 h Minimal peak 18–26 subsequent dosage adjustments should be individual- at 8–10 ized according to the woman’s monitored blood glucose Modified from Gabbe SG, Graves CR. Management of diabetes mellitus compli- levels at particular times of the day. For long-acting and cating pregnancy. Obstet Gynecol 2003;102:857–68. e54 Practice Bulletin Gestational Diabetes Mellitus OBSTETRICS & GYNECOLOGY fetal exposure to metformin and the absence of long- twice daily. Because metformin generally is not used in term neonatal follow-up after in-utero metformin expo- patients with chronic renal disease, creatinine often is sure is one reason the ADA continues to recommend that checked at baseline to ensure adequate renal function. when pharmacologic treatment of GDM is indicated, The most common adverse effects of metformin are insulin is considered the preferred treatment for diabetes abdominal pain and diarrhea, which are minimized by in pregnancy (19). slowly increasing the dosage. Such adverse effects were In one large trial, 751 women with GDM were ran- reported in 2.5–45.7% of patients enrolled in studies domly assigned to receive insulin therapy or metformin of metformin in pregnancy (68), and it is common to (plus insulin if needed). Both groups experienced similar recommend taking the medication with meals to reduce rates of a composite outcome of perinatal morbidity, symptoms. If higher doses are needed, the maximum consisting of neonatal hypoglycemia, respiratory dis- dose is usually 2,500–3,000 mg per day in two to three tress, need for phototherapy, birth trauma, prematurity, divided doses. In women who decline insulin therapy or and low Apgar scores (66). In another prospective trial, who the obstetricians or obstetric care providers believe women randomized to metformin had lower mean glu- will be unable to safely administer insulin, or for women cose levels, less gestational weight gain, and neonates who cannot afford insulin, metformin is a reasonable with lower rates of hypoglycemia than those randomized alternative choice. to insulin (67). Glyburide is a sulfonylurea that binds to pancre- Meta-analyses comparing metformin to insulin have atic beta-cell adenosine triphosphate potassium chan- been conflicting dependent upon whether unpublished nel receptors to increase insulin secretion and insulin studies or women with type II diabetes mellitus are sensitivity of peripheral tissues. It should not be used included. In an initial meta-analysis that included only in patients who report a sulfa allergy. Previous meta- published data, the differences between neonates deliv- analyses have noted increased risks of macrosomia and ered to women randomized to metformin versus insulin hypoglycemia with glyburide compared with insulin in were minimal (68, 69). Yet, women randomized to met- the treatment of GDM (66, 67); whereas a more recent formin experienced a higher rate of preterm birth (risk meta-analysis only demonstrated higher rates of neonatal ratio [RR], 1.5), but a lower rate of gestational hyperten- hypoglycemia (72). These worse outcomes are despite sion (RR, 0.53) (68). the fact that individual trials comparing glyburide In a recent meta-analysis that included unpublished with insulin failed to show any significant difference trials, a network meta-analysis was performed (70). This in degree of glycemic control (73–75). Observational method combines information across multiple treatments studies have reported higher rates of preeclampsia, simultaneously with the analysis of direct evidence hyperbilirubinemia, and stillbirth with use of glyburide (which comes from studies directly randomizing treat- as compared with insulin, but many other outcomes have ments of interest) and indirect evidence (which comes not been statistically significantly different (62, 76–81). from studies comparing treatments of interest with a The common dosage of glyburide is 2.5–20 mg daily in common comparator). The effect size did not demon- divided doses, although pharmacokinetic studies during strate superiority when metformin was compared with pregnancy indicate daily doses up to 30 mg may be nec- insulin on the outcomes of large for gestational age, essary to achieve adequate control (82). Additionally, macrosomia, neonatal hypoglycemia, or cesarean deliv- 4–16% (or more) women required the addition of insulin ery. Interestingly, in the dichotomous meta-analysis to maintain good glycemic control when glyburide was performed, no difference in preterm delivery was dem- used as initial treatment (73, 77, 83, 84). Despite the onstrated (RR 1.37, 95%; CI 0.62–3.01). A subsequent increased use of glyburide over the past decade (62), the meta-analysis with trials that included women with type evidence indicates that glyburide treatment should not be II diabetes and GDM also noted no increase in preterm recommended as a first-choice pharmacologic treatment delivery (71). Thus, although metformin may be a rea- because, in most studies, it does not yield equivalent sonable alternative approach to treat gestational diabetes, outcomes to insulin or metformin. it is important to counsel women about the lack of supe- Concerns also have been raised about the safety of riority when compared with insulin, the placental transfer oral antidiabetic agents during pregnancy. For exam- of the drug, and the absence of long-term data in exposed ple, although an initial study that analyzed umbilical offspring. Additionally, in the aforementioned prospec- cord blood revealed no detectable glyburide in exposed tive trials, between 26% and 46% of women who took pregnancies (73), a subsequent study demonstrated that metformin alone eventually required insulin (66, 67). glyburide does cross the placenta (82). As mentioned The dosage for metformin usually starts at 500 mg previously, metformin also has been found to freely cross nightly for 1 week at initiation, then increases to 500 mg the placenta, and the fetus is exposed to concentrations VOL. 131, NO. 2, FEBRUARY 2018 Practice Bulletin Gestational Diabetes Mellitus e55 similar to maternal levels (85). Theoretic concerns include GDM with poor glycemic control. Additionally, because the potential effects of in utero metformin exposure on those women who are treated medically with insulin or long-term glucose homeostasis of developing offspring. oral agents had suboptimal glycemic control at some It also is not yet known whether oral antidiabetic medi- time, fetal surveillance usually is recommended for cations affect the progression to type 2 diabetes later these patients as well (87). Antenatal fetal testing in in life in women who were treated during pregnancy. women with poorly controlled or medication-requiring A recent Cochrane meta-analysis, reporting data on GDM without other morbidities usually is initiated at 7,381 women, that compared insulin versus any type of 32 weeks of gestation. If other factors associated with oral antidiabetic pharmacological therapies noted similar increased risk of adverse pregnancy outcome are pres- effects on health outcomes. This investigation combined ent, it may be reasonable to start surveillance earlier in women taking either metformin, glyburide, or both, and pregnancy. acarbose (86). Individually these oral antidiabetic medica- Studies have not specifically demonstrated an tions have been noted to have different clinical efficacies increase in stillbirth with well-controlled A1GDM before on maternal and neonatal outcomes with diverse safety 40 weeks of gestation. Thus, antepartum fetal testing may profiles, hence pooling these trials may have a confound- not be necessary in these women. There is no consensus ing effect limiting the conclusions drawn from this meta- regarding antepartum fetal testing among women with analysis. Although current data demonstrate no adverse well-controlled GDM who are not medically treated short-term effects on maternal or neonatal health from (A1GDM). If antepartum testing is to be used in such oral antidiabetic therapy during pregnancy, long-term patients, it is generally started later than in women with outcomes are not yet available. Thus, health care provid- A2GDM. The specific antepartum test and frequency ers should counsel women of the limitations in safety data of testing may be chosen according to local practice; when prescribing oral agents to women with GDM. however, because polyhydramnios can result from fetal Taking into account that oral antidiabetic medi- hyperglycemia, it is common for clinicians to use testing cations are not approved by the U.S. Food and Drug that incorporates serial measures of amniotic fluid. Administration for the treatment of GDM, cross the placenta, and lack long-term neonatal safety data; and What are delivery considerations in considering that summaries of the current medical lit- pregnancies complicated by gestational erature note poor trial quality while not being designed diabetes mellitus? to assess equivalence or noninferiority when comparing oral agents to insulin; insulin is considered the preferred Women with GDM with good glycemic control and no treatment when pharmacologic treatment of GDM is other complications are commonly managed expectantly indicated. Although this recommendation aligns with until term (88, 89). In most cases, women with good the ADA recommendation, ACOG recognizes that clini- glycemic control who are receiving medical therapy cal situations may occur that necessitate the use of oral do not require delivery before 39 weeks of gestation. agents. As aforementioned, in women who decline insu- The recent GINEXMAL trial of GDM-only patients lin or who the obstetricians or obstetric care providers randomized women to induction of labor at 38 weeks believe will be unable to safely administer insulin, or of gestation versus expectant management up to for women who cannot afford insulin, metformin (and 41 weeks of gestation (90). Although the study did not rarely glyburide) is a reasonable alternative choice in the achieve its intended sample size, there was no difference context of discussing with the patient the limitations of in cesarean delivery rates (12.6% versus 11.8%, P=.81) the safety data and a high rate of treatment failure that or many other outcomes. There was, however, a higher requires insulin supplementation. rate of hyperbilirubinemia in the induced group (10.0% versus 4.1%, P=.03). In a randomized trial in which Is fetal assessment indicated in pregnancies women with insulin-treated GDM and fetuses believed complicated by gestational diabetes mellitus? to be of appropriate weight for gestational age were ran- domized at 38 weeks of gestation to induction of labor Antepartum fetal testing is recommended for patients within 1 week or expectant management, there was no with pregestational diabetes. Because the increased risk difference in cesarean delivery rates (91). However, of fetal demise in patients with pregestational diabetes there was a smaller proportion of LGA infants in the is related to suboptimal glycemic control, it would be induction group. Furthermore, a multiple time series expected that women with GDM who have poor glyce- cohort study showed that there were no significant mic control also would be at increased risk. Therefore, differences in either macrosomia or cesarean delivery fetal surveillance may be beneficial for women with rates among women with insulin-treated GDM who e56 Practice Bulletin Gestational Diabetes Mellitus OBSTETRICS & GYNECOLOGY underwent induction of labor at 38–39 weeks of gestation LGA at birth (100). Additionally, in women whose when compared with expectantly managed historic con- fetuses received an LGA diagnosis, the risk of cesar- trols (92). Shoulder dystocia was experienced by 10% ean delivery was increased independent of actual birth of the expectant management group after more than weight. It has been estimated that up to 588 cesarean 40 weeks of gestation versus 1.4% in the group with deliveries would be needed to prevent a single case of labor induction at 38–39 weeks of gestation. A systemat- permanent brachial plexus palsy for an estimated fetal ic review later confirmed these findings (93). However, weight of 4,500 g, and up to 962 cesarean deliveries a recent study that compared induction of labor before would be needed for an estimated fetal weight of 4,000 g 40 weeks of gestation with expectant management dem- (101, 102). Based on the available data, it is not possible onstrated a reduction in cesarean delivery among women to determine whether the potential benefits of planned with GDM who were induced (94). A decision analysis cesarean delivery at a given estimated fetal weight are demonstrated that delivery of women with GDM at similar for women with GDM and women with pre- 38 weeks or 39 weeks of gestation would reduce overall existing diabetes. Therefore, it appears reasonable to perinatal mortality without increasing cesarean delivery recommend that women with GDM should be counseled rates (95). Although persuasive, these data have not regarding the risks and benefits of a scheduled cesarean been confirmed by large randomized trials. Therefore, delivery when the estimated fetal weight is 4,500 g or the timing of delivery in women with GDM that is more (103). controlled with only diet and exercise (A1GDM) should not be before 39 weeks of gestation, unless otherwise How should women with a history of gesta- indicated. In such women, expectant management up to tional diabetes mellitus be screened and 40 6/7 weeks of gestation in the setting of indicated counseled postpartum? antepartum testing is generally appropriate. For women with GDM that is well controlled by medications Although the carbohydrate intolerance of GDM fre- (A2GDM), delivery is recommended from 39 0/7 weeks quently resolves after delivery, up to one third of to 39 6/7 weeks of gestation. affected women will have diabetes or impaired glucose In contrast, expert opinion has supported earlier metabolism at postpartum screening. It has been esti- delivery for women with poorly controlled GDM (88, mated that between 15% and 70% will develop diabe- 89). But clear guidance about the degree of glycemic tes (predominantly type 2) later in life (8, 104–107). control that necessitates earlier delivery is lacking, and Another study showed that women with a history of the recommendations about timing of delivery lack spe- GDM have a sevenfold increased risk of developing cific guidance as well (96). In light of this, consideration type 2 diabetes compared with women without a history of timing should incorporate tradeoffs between the risks of GDM (108). Therefore, screening at 4–12 weeks post- of prematurity and the ongoing risks of stillbirth. In such partum is recommended for all women who had GDM to a setting, delivery between 37 0/7 weeks and 38 6/7 identify women with diabetes, impaired fasting glucose weeks of gestation may be justified, but delivery in the levels, or impaired glucose tolerance (IGT) (Fig. 1) late preterm period from 34 0/7 weeks to 36 6/7 weeks of (19). A fasting plasma glucose test and the 75-g, 2-hour gestation should be reserved for those women who fail OGTT have been used for diagnosing overt diabetes in-hospital attempts to improve glycemic control or who in the postpartum period. Although the fasting plasma have abnormal antepartum fetal testing. glucose test is easier to perform, it lacks sensitivity for Because macrosomia is distinctly more common detecting other forms of abnormal glucose metabolism. in women with GDM and because shoulder dystocia Results of the OGTT can confirm an impaired fasting is more likely at any given fetal weight in pregnancies glucose level and impaired glucose tolerance. Therefore, complicated by diabetes than in pregnancies not compli- the Fifth International Workshop on Gestational Diabetes cated by diabetes (97–99), it is reasonable for clinicians Mellitus recommends that women with GDM undergo a to assess fetal growth by ultrasonography or by clinical 75-g, 2-hour OGTT in the postpartum period (109). This examination late in the third trimester to attempt to iden- usually should include a fasting plasma glucose as well. tify macrosomia among women with GDM. However, All women who had GDM should follow up with data are insufficient to determine whether cesarean a primary care physician. Additionally, women with delivery should be performed to reduce the risk of birth impaired fasting glucose, IGT, or diabetes should be trauma in cases of suspected macrosomia. Although referred for preventive or medical therapy. Women with the use of ultrasonography to estimate fetal weight is impaired fasting glucose or IGT may respond to life- common, one recent study found that among cases of style modification and pharmacologic interventions to ultrasonography-diagnosed LGA infants, only 22% were decrease incident diabetes. Women with frank diabetes VOL. 131, NO. 2, FEBRUARY 2018 Practice Bulletin Gestational Diabetes Mellitus e57  Gestational diabetes FPG or 75-g, 2-hr OGTT at 4–12 weeks postpartum FPG >125 mg/dL or FPG 100–125 mg/dL or FPG 199 mg/dL 2-hr glucose 140–199 mg/dL 2-hr glucose