Dietary Therapy

The goal of dietary therapy is to avoid single large meals and foods with a large percentage of simple carbohydrates. A total of 6 feedings per day is preferred, with 3 major meals and 3 snacks to limit the amount of energy intake presented to the bloodstream at any interval. The diet should include foods with complex carbohydrates and cellulose, such as whole grain breads and legumes.

Carbohydrates should account for no more than 50% of the diet, with protein and fats equally accounting for the remainder.[4, 5] However, moderate restriction of carbohydrates to 35–40% has been shown to decrease maternal glucose levels and improve maternal and fetal outcomes.[48]

Nutritional therapy should be supervised by a trained professional, ideally a registered dietitian, with formal dietary assessment and counseling provided at several points. For obese women (BMI >30 kg/m2), a 30–33% calorie restriction (to 25 kcal/kg actual weight per day or less) has been shown to reduce hyperglycemia and plasma triglycerides with no increase in ketonuria.

Insulin Therapy

Patients with preexisting diabetes require modification of their pharmacologic regimen to meet the changing metabolic demands of pregnancy. In gestational diabetes, early intervention with insulin or an oral agent is key to achieving a good outcome when diet therapy fails to provide adequate glycemic control. Determine the choice of insulin and regimen based on the patient's individual glucose profile.

The goal of insulin therapy during pregnancy is to achieve glucose profiles similar to those of nondiabetic pregnant women. Given that healthy pregnant women maintain their postprandial blood sugar excursions within a relatively narrow range (70-120 mg/dL), reproducing this profile requires meticulous daily attention by both the patient and clinician.

  1. lisproaspartregular, and neutral protamine hagedorn (NPH) are well-studied in pregnancy and regarded as safe and effective. Insulin glargine is less well-studied, and given its long pharmacologic effect, may exacerbate periods of maternal hypoglycemia.[55] Insulin detemir is safe and comparable to NPH insulin in pregnancy.[56]

As pregnancy progresses, the increasing fetal demand for glucose and the progressive lowering of maternal fasting and between-meal blood sugar levels increases the risk of symptomatic hypoglycemia. Upward adjustment of short-acting insulin doses to control postprandial glucose surges within the target band only exacerbates the tendency to interprandial hypoglycemia. Thus, any insulin regimen for pregnant women requires combinations and timing of insulin injections quite different from those that are effective in the nonpregnant state. Further, the regimens must be continuously modified as the pregnancy progresses from the first to the third trimester and insulin resistance rises. Strive to stay ahead of the rising need for insulin, and increase insulin dosages preemptively.

When more than 20% of postprandial blood glucose levels exceed 130 mg/dL, administer lispro insulin (4-8 U subcutaneously [SC] initially) before meals. If more than 10 U of regular insulin is needed before the noon meal, adding 8-12 U of NPH insulin before breakfast helps achieve control. When more than 10% of fasting glucose levels exceed 95 mg/dL, initiate 6-8 U NPH insulin at bedtime (hs). Titrate doses as needed according to blood glucose levels.[55]

In women with type 1 diabetes, postprandial glucose control is significantly impaired in late gestation. It is largely accounted for by slower glucose disposal.[57] Early prandial insulin should help accelerate glucose disposal and potentially improve or ameliorate postprandial hyperglycemia in late pregnancy.

Insulin pump

In a select group of patients, use of an insulin pump may improve glycemic control while enhancing patient convenience. These devices can be programmed to infuse varying basal and bolus levels of insulin, which change smoothly even while the patient sleeps or is otherwise preoccupied.

The effectiveness of continuous subcutaneous insulin infusion in pregnancy is well established.[58, 59] Hieronimus et al reported similar HbA1C levels, macrosomia rates, and cesarean rates in 33 pregnant women managed with insulin pump, compared with 23 receiving multiple insulin injections,.[60] Lapolla et al found no differences in glycemic control or perinatal outcome between 25 women treated with insulin pump in pregnancy and 68 women who received conventional insulin treatment.[61]

Oral therapy


The efficacy and safety of insulin have made it the standard for treatment of diabetes during pregnancy. Nevertheless, the oral agents glyburide and metformin are gaining popularity. Trials have shown these agents to be effective and no evidence of harm to the fetus has been found, although the potential for long-term adverse effects remains a concern.[6]

  1. Glyburide is a second-generation sulfonyurea that is minimally transported across the human placenta. This is probably largely due to the high plasma protein binding coupled with a short half-life.[62, 63] In addition, a human placenta perfusion study demonstrated active glyburide transport from the fetus to the mother.[64]

A 2000 randomized trial comparing glyburide to insulin in 404 pregnancies found no difference between the groups in mean maternal blood glucose levels, the percentage of infants who were LGA, birth weights, or neonatal complications. Only 4% of patients in the glyburide study arm required addition of insulin to achieve glucose control.[63] Since this study, several prospective and retrospective studies involving more than 775 pregnancies have concluded glyburide is as safe and effective as insulin. All studies comparing glyburide to traditional insulin have demonstrated similar levels of glycemic control. Most studies show no differences in maternal or neonatal outcomes with glyburide.[65]

Success rates for achieving glycemic control with glyburide vary from 79% to 86%. Studies evaluating predictors of failure with glyburide cite the following risk factors[66, 67, 68, 69] :

  • Advanced maternal age
  • Earlier gestational age at diagnosis
  • Higher gravidity and parity
  • Higher mean fasting glucose level

Glyburide should not be used in the first trimester, because its effects, if any, on the embryo are unknown. Research in this area, although needed, has been difficult given the known teratogenic effects of first-generation sulfonylurea, which readily crossed the placenta.

Glyburide has been shown to be safe in breastfeeding, with no transfer into human milk.


  1. Metformin is a biguanide, which functions mainly by decreasing hepatic glucose output. Metformin crosses the placenta, and umbilical cord levels have been shown to be even higher than maternal levels.[70]

An initial retrospective study comparing glyburide, metformin, and insulin in pregnancy raised concern, because of increased rates of preeclampsia and perinatal mortality when metformin was used in the third trimester.[71] It should be noted that in this study the patients on metformin had a higher body mass index and were older than the patients on glyburide or insulin. Since this initial study, however, several other prospective and retrospective studies involving over 300 pregnant patients have not confirmed the increased rates of preeclampsia or perinatal mortality.[72, 73, 74, 75, 76] These subsequent studies have demonstrated similar efficacy, safety, and maternal and fetal outcomes with metformin.

Moore et al compared the effect of metformin and glyburide in women with gestational diabetes who did not achieve glycemic control with diet. Between the 2 groups, patients who achieved glycemic control did not differ with regard to mean fasting and 2-hour postprandial blood glucose level. However, the percentage of women who did not achieve glycemic control and required insulin was 2.1 times higher with metformin (34.7%) than with glyburide (16.2%).[6]

Prenatal obstetric management

Periodic fetal biophysical testing

The goals of management of third-trimester pregnancies in women with diabetes are to prevent stillbirth and asphyxia while minimizing maternal and fetal morbidity associated with delivery. Monitoring fetal growth is essential to select the proper timing and route of delivery. This is accomplished by frequent testing for fetal well-being and serial ultrasonographic examinations to follow fetal size.

Various fetal biophysical tests are available to the clinician to ensure that the fetus is well oxygenated, including fetal heart rate testing, fetal movement assessment, ultrasonographic biophysical scoring, and fetal umbilical Doppler ultrasonographic studies (see the table below). If applied properly, most of these tests can be used with confidence to provide assurance of fetal well-being while awaiting fetal maturity.

Table 3. Biophysical Tests of Fetal Well-Being for Diabetic Pregnancy (Open Table in a new window)

Test Frequency Reassuring Result Comment
Fetal movement counting Every night from 28 weeks 10 movements in < 60 min Performed in all patients
Nonstress test (NST) Twice weekly 2 heart rate accelerations in 20 min Begin at 28-34 weeks with insulin-dependent diabetes, and begin at 36 weeks in diet-controlled GDM.
Contraction stress test Weekly No heart rate decelerations in response to 3 contractions in 10 min Same as for NST
Ultrasonographic biophysical profile Weekly Score of 8 in 30 min 3 movements = 2

1 flexion = 2

30 s breathing = 2

2 cm amniotic fluid = 2

Initiate testing early enough to avoid significant stillbirth but not so early that a high rate of false-positive test results is encountered. In patients with poor glycemic control, intrauterine growth restriction, or significant hypertension, begin formal biophysical testing as early as 28 weeks. In patients who are at lower risk, most centers begin formal fetal testing by 34 weeks. Fetal movement counting is performed in all pregnancies from 28 weeks onward.

There is no consensus regarding antenatal testing in patients with gestational diabetes that is well controlled with diet.

Assessing fetal growth

Monitoring fetal growth continues to be a challenging and imprecise process. Although currently available tools (serial plotting of fetal growth parameters based on ultrasonographic measurement) are superior to those used previously for clinical estimations, accuracy is still only within ±15%.[77]

In the obese fetus, the inaccuracies are further magnified. In 1992, Bernstein and Catalano reported that significant correlation exists between the degree of error in the ultrasonogram-based estimation of fetal weight and the percentage of body fat on the fetus.[78] Perhaps this is the reason no single formula has proven to be adequate in identifying a macrosomic fetus with certainty.

Despite problems with accuracy, ultrasonogram-based estimations of fetal size have become the standard of care. Estimate fetal size once or twice at least 3 weeks apart in order to establish a trend. Time the last examination to be at 36-37 weeks' gestation or as close to the planned delivery date as possible.

Timing and route of delivery

Select the timing of delivery to minimize morbidity for the mother and fetus. Delaying delivery to as near as possible to the expected date of confinement helps maximize cervical maturity and improves the chances of spontaneous labor and vaginal delivery. However, the risks of advancing fetal macrosomia, birth injury, and in utero demise increase as the due date approaches.

Although delivery as early as 37 weeks might reduce the risk of shoulder dystocia, it increases the likelihood of failed labor induction and poor neonatal pulmonary status. Because fetal growth from 37 weeks onward may be 100-150 g/wk, the reduction in net fetal weight and the risk of shoulder dystocia by inducing labor 2 weeks early may theoretically improve outcome.

By comparing the outcomes associated with labor induction in patients with gestational diabetes at 38 weeks versus expectant management with fetal testing, Kjos et al compared found that expectant management increased the gestational age at delivery by 1 week, but it did not significantly reduce the cesarean delivery.[79] However, the prevalence of macrosomia was significantly greater among infants in the expectantly managed group (23%) than among those in the active induction group (10%). This suggests that routine induction of women with diabetes on or before 39 weeks' gestation does not increase the risk of cesarean delivery and may reduce the risk of macrosomia.[79]

If the fetus is not macrosomic and the results of biophysical testing are reassuring, the obstetrician can await spontaneous labor. In patients with gestational diabetes mellitus and superb glycemic control, continued fetal testing and expectant management can be considered until 41 weeks' gestation. In the fetus with an abdominal circumference significantly larger than the head circumference or an estimated fetal weight above 4000 g, consider induction. After 40 or more weeks, the benefits of continued conservative management are likely to be outweighed by the danger of fetal compromise. Induction of labor before 41 weeks' gestation in pregnant women with diabetes, regardless of the readiness of the cervix, is prudent.

An optimal time for delivery of most diabetic pregnancies is typically on or after the 39th week. Deliver a patient with diabetes before 39 weeks' gestation without documented fetal lung maturity only for compelling maternal or fetal indications. For elective induction before 38.5 weeks, fetal lung maturity should be verified via amniocentesis.

Because the risk of shoulder dystocia and fetal injury in labor is increased 3-fold in diabetic pregnancy, elective cesarean section should be considered if the fetus is suspected to be significantly obese. The American College of Obstetricians and Gynecologists recommends offering cesarean delivery to diabetic patients if the fetal weight is estimated to be 4500 g or more.

Prevention of Shoulder Dystocia

Although ultrasonographic measurements of the fetus have proven to be poor predictors of the risk of shoulder dystocia, this technique continues to be the mainstay for assessing risk in pregnancy for women with diabetes. The commonly used formulas derived from a multivariate regression multiply multiple coefficients together, with the resultant product (estimated fetal weight) typically having an accuracy that is seldom less than within 15%. Fetuses predicted to weigh 4000-4500 g based on ultrasonographic findings actually weigh that much only 50% of the time.

In a study involving more than 300 fetuses who weighed more than 4000 grams at birth, ultrasonography was found to have a sensitivity of only 65% in identifying macrosomia. However, a sensitivity of approximately 80% is typically associated with a specificity of 50-60%. This means a false-positive rate of 30-50% occurs even with the more predictive formula, possibly requiring an unnecessary cesarean delivery of more than 100 fetuses in order to prevent 1 from having permanent Erb palsy.

Thus, current data do not support a policy of early induction of labor in cases of possible fetal macrosomia. If one accepts that 8-20% of infants of diabetic mothers born weighing 4500 g or more will experience shoulder dystocia, 15-30% of these will have recognizable brachial plexus injury, and 5% of these injuries will result in permanent deficit, approximately 333-1667 cesarean deliveries would have to be performed for possible macrosomia to prevent 1 case of permanent injury due to shoulder dystocia. However, if fetal weight is estimated to be 4500 g or more, the risks and benefits of cesarean delivery should be discussed with the patient.

Intrapartum glycemic management

Maintenance of intrapartum metabolic homeostasis optimizes postnatal infant transition by reducing neonatal hyperinsulinemia and subsequent hypoglycemia. The use of a combined insulin and glucose infusion during labor to maintain maternal blood sugars in a narrow range (80-110 mg/dL) is a common and clinically efficient practice. Typical infusion rates are 5% dextrose in Ringer lactate solution at 100 mL/h and regular insulin at 0.5-1.0 U/h. Capillary blood sugar levels are monitored hourly in these patients.

For patients with diet-controlled gestational diabetes mellitus or mild type 2 diabetes, avoiding dextrose in intravenous fluids normally maintains excellent blood glucose control. After 1-2 hours of monitoring, no further assessments of capillary blood sugar typically are necessary.

For patients with diet-controlled gestational diabetes, myoinositol improves insulin resistance and increased adiponectin levels.[80]

Management of the neonate

The most critical metabolic problem that affects infants of diabetic mothers is hypoglycemia. Unmonitored and uncorrected hypoglycemia can lead to neonatal seizures, brain damage, and death. The strongest predictor of neonatal hypoglycemia is maternal mean blood glucose level during labor. Infants of diabetic mothers also appear to have disorders of both catecholamine and glucagon metabolism and have a diminished capability to mount normal compensatory responses to hypoglycemia.

Thus, current recommendations specify frequent blood glucose checks and early oral feeding when possible (ideally from the breast), with infusion of intravenous glucose if oral measures prove insufficient. Most neonatologists maintain strict monitoring of the glucose levels of newborn infants of diabetic mothers for at least 4-6 hours (frequently 24 h), often necessitating admission to a newborn special care unit.

Current evidence indicates that with proper encouragement, sustained breastfeeding is possible for a significant proportion of patients with overt diabetes. In fact, evidence indicates that breast-fed infants have a much lower risk of developing diabetes than those exposed to cow's milk proteins.

Studies of breastfeeding women with diabetes indicate that lactation, even for a short duration, also has a beneficial effect on overall maternal glucose and lipid metabolism. For postpartum women who had gestational diabetes mellitus during their pregnancies, breastfeeding may offer a practical low-cost intervention that helps reduce or delay the risk of subsequent diabetes.

In a longitudinal study comparing breastfeeding habits among women with diabetes and without diabetes, Webster et al reported that diabetic women breastfed at least as commonly and for as long as women without diabetes. At discharge, 63% of diabetic mothers and 78% of mothers without diabetes were breastfeeding. At 8 weeks, the proportions of each were nearly identical (58% and 56%, respectively). At 3 months, 47% percent of mothers with diabetes and 33% mothers without diabetes continued to breastfeed.[81]

A study by Gunderson et al found that a higher intensity of lactation among exclusively or mostly breastfeeding (< 6 oz formula per 24 h) ­mothers improved insulin sensitivity and glucose metabolism. This may reduce the future diabetes risk after gestational diabetes.[82]

Prevention of gestational diabetes mellitus

Prevention of gestational diabetes is an attractive concept, but no progress has been made, despite attempts in small studies. Because body fat and diet contribute to the risk of gestational diabetes mellitus, patients who lose weight before pregnancy and follow an appropriate diet may lower their risk of gestational diabetes mellitus. However, the hormone levels in pregnancy impose such a high degree of insulin resistance that in very susceptible individuals, even marked weight loss and attention to diet are not likely to be successful.

Additionally, a large study by Stafne et al found that a 12-week standard exercise program during the second half of pregnancy had no benefit in preventing gestational diabetes in healthy women with normal BMI.[83]