May 2010: Volume 36, Number 5
A New Definition of Gestational Diabetes
Will Revised Diagnostic Criteria Help or Hurt?
By Genna Rollins
Obstetricians and diabetologists long have recognized the need for a more consistent and evidence-based approach to diagnosing gestational diabetes, defined as glucose intolerance that first occurs in pregnancy. Nearly 50 years ago, the pioneering work of John O’Sullivan and Claire Mahan led to development of diagnostic thresholds for the disorder. However, these criteria were based on the mother’s likelihood of developing type 2 diabetes later in life, not on adverse pregnancy outcomes, and were never universally adopted. In addition, glucose assays have changed considerably from that time, going from measurements in whole blood to today’s high-throughput, serum- and plasma-based enzymatic methods. These issues have held back research and resulted in inconsistent approaches to the disorder.
“Different hospitals, different countries, all use different diagnostic criteria, which makes comparing research results impossible,” explained John Kitzmiller, MD, an obstetrician at Santa Clara Valley Medical Center in San José, Calif. “The original diagnostic criteria O’Sullivan used, which became the most popular in America—but were not used universally—were changed. The recommended changes were never proven in any kind of randomized trial as being the best, but some people use them and others don’t, so we’ve just had a mess.”
Clearing Up the Diagnostic Landscape
The muddled diagnostic picture for gestational diabetes soon may become much clearer, thanks to several ground-breaking clinical trials and careful deliberations by the International Association of Diabetes and Pregnancy Study Groups (IADPSG), an umbrella organization created to facilitate collaboration between regional and national groups with a focus on diabetes and pregnancy. In March, the IADPSG released recommendations for diagnosing gestational diabetes that the group hopes eventually will lead to one world-wide standard, as well as further research that will clarify cost-effective treatment strategies for the condition, optimal glycemic treatment targets, and appropriate post-natal follow-up of mothers and children (Diabetes Care, 2010;33:676–682).
The new recommendations call for diagnosis of hyperglycemia in pregnancy when thresholds are exceeded for at least one of three glucose measurements, including fasting plasma glucose, and 1-hour and 2-hour values on a 75-g oral glucose tolerance test. The thresholds are 92 mg/dL for fasting plasma glucose, 180 mg/dL for 1-hour plasma glucose, and 153 mg/dL for 2-hour plasma glucose. The group also proposed thresholds for diagnosis of overt diabetes in pregnancy (See Table, below).
New Diagnostic Criteria
The International Association of Diabetes in Pregnancy Study Groups recently recommended new thresholds for diagnosing gestational diabetes and overt diabetes in pregnancy.
Proposed Diagnostic Thresholds
Type of Glucose Test
Glucose Concentration Threshold
|Fasting Plasma Glucose
|1-h Plasma Glucose
|2-h Plasma Glucose
Overt Diabetes in Pregnancy
Fasting Plasma Glucose ≥7.0 mmol/L (≥126 mg/dL)
Random Plasma Glucose ≥11.1 mmol/L (≥200 mg/dL) + confirmation
|Adapted from Diabetes Care 2010;33:676–682|
“This will be considered by professional groups across the world, and each group will make its own decision about endorsing the recommendations, but if that happens we’ll finally have agreement about the diagnostic criteria for gestational diabetes,” observed Patrick Catalano, MD, chairman of obstetrics and gynecology at MetroHealth Medical Center and professor of reproductive biology at Case Western Reserve University, both in Cleveland. “From there, future research is going to look at, now that we have the data, how do we best utilize it in a clinical way that makes sense for the patient and society, and for long-term and cost issues.” Catalano was a member of the IADPSG writing committee.
A Surge in New Cases
Assuming key U.S.-based organizations like the American Diabetes Association (ADA) and American College of Obstetricians and Gynecologists (ACOG) adopt the recommendations, the number of American women diagnosed with gestational diabetes could increase substantially. About 5–8% of pregnant women currently qualify as having gestational diabetes, but >16% would be classified with the disorder under the IADPSG recommendations. Key studies that informed the group’s recommendations also revealed that another 1.7% of women had overt diabetes in pregnancy.
Given the prevalence of obesity-driven type 2 diabetes, a higher incidence of gestational diabetes comes as no surprise, according to Donald Coustan, MD, professor of obstetrics and gynecology at the Warren Alpert Medical School of Brown University in Providence, R.I. “Should we be defining a disease in which such a high proportion of the population is affected? My view is, there’s an epidemic of type 2 diabetes going around the world, so it isn’t surprising that we’d be seeing more gestational diabetes, which is a forme fruste of type 2 diabetes,” he said. “The latest data from ADA indicates that about 29 percent of the population is diabetic or pre-diabetic, so having nearly 17 percent of women with gestational diabetes isn’t such a leap.”
Building the Evidence Base
Although clinicians and researchers were well aware of problems associated with the patchwork of different diagnostic methods and criteria, it took considerable effort to mount a clinical trial of sufficient size and scope to establish the evidence base that IADPSG needed to suggest comprehensive new diagnostic criteria. That investigation came in the form of the Hyperglycemia and Adverse Pregnancy Outcomes (HAPO) study, which involved >25,000 pregnant women at 15 centers in nine countries (N Engl J Med 2008;358:1991–2002). HAPO was designed to assess the risks of adverse outcomes associated with various degrees of maternal glucose intolerance less severe than in overt, pre-existing diabetes. Participants underwent a 75-g oral glucose tolerance test at 24–32 weeks’ gestation and had a random plasma glucose measurement at 34–37 weeks’ gestation.
The HAPO researchers found significant graded relationships between increasing maternal glucose levels and the frequency of four primary and five secondary outcomes. For example, with a 1 standard deviation increase in maternal fasting, 1-hour, and 2-hour plasma glucose levels, there was a corresponding 38%, 46%, and 38% increased risk, respectively, in the primary outcome of birth weight >90th percentile, and a 5%, 18%, and 16% increased risk, respectively, of the secondary outcome of premature delivery before 37 weeks gestation. There were no obvious thresholds at which risks increased or decreased, although associations with the secondary outcomes tended to be weaker.
For many in the field, this was the Eureka moment they had been awaiting for decades. “Clearly the mother’s glucose levels measured on a 75-gram oral glucose tolerance test in the early third trimester of pregnancy were related to the baby’s size, weight, and many other factors,” said Steven G. Gabbe, MD, senior vice president for health sciences and CEO of Ohio State University Medical Center in Columbus. “It was very exciting to see this data support the relationship between the mother’s glucose and the baby’s outcome.” Gabbe chaired the HAPO data safety monitoring committee and was a senior author of the IADPSG writing group.
Linking Diagnosis and Treatment
The HAPO results alone might have been interesting but still not galvanized the field to make specific diagnostic recommendations had not the results of two other multi-center trials become available at about the same time as the HAPO findings. These trials—one American and one Australian—tested the effect of treatment on gestational diabetes (N Engl J Med 2005;352:2477–86; N Engl J Med 2009; 361:1339–48). Although the studies looked at slightly different outcomes and diagnostic thresholds, both concluded that treatment of gestational diabetes reduced risks to both mothers and babies. “So we have proof now that we didn’t have before in big, modern trials showing that yes, treatment has benefits, so yes, diagnosis has benefit. That’s the clincher,” said Kitzmiller, who also served on the IADPSG writing group.
Significantly, treatment for most women in the intervention arms of both studies consisted of dietary advice and glucose monitoring, rather than insulin therapy. This finding contradicts arguments raised by some that the proposed diagnostic criteria could lead to more harm than benefit. For instance, an editorial provocatively titled “Problem Solved or a Pandora’s Box?” that accompanied the IADPSG recommendations suggested that increased diagnosis of gestational diabetes could lead to more interventions such as Cesarean-sections and babies admitted to special care units. “Could these real hazards offset some of the potential advantages?” the author asked (Diabetes Care 2010;33:690–91).
Concerns also have been expressed about the negative psychological consequences for women who receive a diagnosis of gestational diabetes. However, the Australian study found that women in the intervention group reported at 3-months post partum lower rates of depression and higher quality of life. Other studies, including one conducted by Coustan’s lab, have had similar findings.
What’s the Right Cutoff?
In reviewing the HAPO results and other studies that reached comparable conclusions, the IADPSG working group faced some challenges in agreeing on acceptable odds ratios and diagnostic cutoffs. There was a bit of controversy because of the continuous, graded relationship between maternal glucose levels and adverse perinatal outcomes observed in HAPO and other trials. “There was a lot of discussion—healthy discussion—about what thresholds to recommend, because it’s going to be fairly arbitrary what you choose. What convinced people is that if we used an odds ratio of 1.5 we would include so many patients and when it boiled down to adverse outcomes, you'd prevent some, but it wasn't as dramatic of an advantage,” explained Coustan. “But if you used 2 rather than 1.75, the proportion of patients with adverse outcomes wasn't all that different. Yet by using 1.75, you could identify about two times as many patients with potential adverse outcomes.” Coustan was a member of the IADPSG consensus panel.
Similarly, there was discussion about rounding the glucose concentration thresholds up or down so that measurements in mg/dL would be in even numbers and easier for clinicians to remember. “The committee determined that we couldn't do that for two reasons. First, it would change the diagnostic power of the test. Since there's a continuum and linear relationship, moving the threshold up or down by even two milligrams per deciliter would make a difference,” explained Gabbe. “Second, around the world, the measurement used is millimoles per liter, so by rounding milligrams per deciliter to an even number, the millimoles per liter measurement would still be an odd number.”
The writing committee also carefully evaluated whether all three glucose tests—fasting, 1-hour and 2-hour—were needed. “We had a lot of debate about whether you need that two-hour test or not, but it adds about fourteen or fifteen percent predictive value, so in the U.S., that many women would be missed if you only used the fasting or one-hour tests,” explained Kitzmiller, who was a member of the writing group. As it turned out, 11.1% of HAPO subjects had only one elevated test result, 3.9% had two elevated results, and just 1.1% had elevated results for all three measures. “No single test picked-up everybody, so you can’t say if the fasting blood glucose is off analytically it doesn’t really matter because we’ll pick them up with the one- or two-hour tests,” observed David Sacks, MB, ChB, associate professor of pathology at Harvard Medical School. “That wasn’t the case, and each of those values provides independent prognostic information.”
The IADPSG statement also emphasized the need for early identification of women with overt diabetes unrelated to their pregnancy, due to their higher risk of diabetes-related complications. However, it stopped short of recommending a specific test, such as HbA1c, and instead proposed consensus thresholds for fasting plasma glucose, HbA1c, and random plasma glucose (See Table, p. 6). “People will have to decide locally how they want to handle that based on their own preferences, but it's important, as the paper discusses, to find these undiagnosed type 2 diabetics when they come into pregnancy, because they have a very different risk than women with gestational diabetes,” observed Kitzmiller.
Analytics to the Fore
Given the HAPO findings that slight elevations in maternal glucose levels were associated with adverse perinatal outcomes, the IADPSG recommendations also emphasize the importance of analytics and pre-analytics in the setting of gestational diabetes. “To achieve reliable diagnosis and classification…clinical laboratories must measure venous plasma or serum glucose using an enzymatic method with high accuracy and precision,” the writing group noted. According to several committee members, this was aimed at discouraging the use of point-of-care glucose testing in this population. “Glucose meters clearly are not good enough for this. They just should not even be considered,” advised Sacks (See Blood Glucose Meters, this issue).
Sacks also pointed out the rigorous spe-cimen collection and processing standards that were used in the HAPO trial (Clinical Trials 2006;3:397–407). Samples were collected in sodium fluoride-containing tubes, kept on ice until the plasma separated within 1 hour, and then frozen at -20ºC. Field center labs processed and aliquoted all samples and analyzed aliquots of fasting and 2-hour oral glucose tolerance tests and random plasma glucose samples. The central lab in Belfast, U.K. performed all biochemical analyses except the random plasma glucose sample. This lab also monitored the quality control for the glucose analysis performed by all the field labs.
Sacks is concerned that these standards will be difficult to duplicate in practice. “This is an area where clinical laboratorians need to communicate with physicians and the detailed protocols need to be developed by close collaboration between the groups. For example, many clinicians believe that once you put a sample in sodium fluoride, there's no glycolysis. That's a very widely held misconception,” said Sacks. In fact, some glycolysis occurs up to about 4 hours after collection even in the presence of sodium fluoride, he indicated.
Catalano agreed that physician-patient-lab communication and physician clinical judgment will be crucial. “The things that happen around the testing are as important from the clinical point of view as the accuracy of the results,” he said. “We always have to look at whether the data make sense. If there’s a very high fasting result and the rest of the values don’t look abnormal, you may need to talk with the patient to find out if she was really fasting, or if she got hungry and ate between the first and second-hour tests and forgot to tell anyone,” he explained. Sacks also expressed concerns about pre-analytical factors. “The glucose tolerance test hasn’t been as studied in pregnant women, but in non-pregnant women with diabetes it has very poor reproducibility because of the large number of factors that influence it, ranging from the time of day the woman takes the test and her prior activity to whether she smokes and the amount of water in which the glucose is dissolved,” he noted.
For Further Information
American Diabetes Association Position Statement: Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 2010; 33:S62–S69.
American College of Obstetricians and Gynecologists Practice Bulletin on Gestational Diabetes 30; September, 2001.
Cutchie WA, Cheung NW, Simmons D. Comparison of international and New Zealand guidelines for the care of pregnant women with diabetes. Diabet Med 2006; 23:460–68.
HAPO Study Cooperative Research Group. Hyperglycaemia and Adverse Pregnancy Outcomes (HAPO) Study: associations with maternal body mass index. BJOG 2010;117:575–84.
Scott DA, Loveman E, McIntyre L, Waugh N. Screening for gestational diabetes: a systematic review and economic evaluation. Health Technol Assess 2002;6:1–161.
Screening for Gestational Diabetes Mellitus: U.S. Preventive Services Task Force Recommendation Statement. Ann Intern Med 2008;148:759–765.
World Health Organization: Diabetes Mellitus: Report Of A WHO Study Group. Geneva, World Health Org., 1985, (Tech. Rep. Ser., no 727).
A Major Change in Practice?
In addition to analytical concerns, the recommendations also could spell major procedural changes for obstetricians and clinical labs in the U.S. According to Coustan, ACOG surveys have indicated that the vast majority of obstetrical practices currently use a 50-gram, 1-hour screening test followed by a 3-hour glucose tolerance test for those patients with glucose levels >130 mg/dL or ≥140 mg/dL on the screening test. Under the IADPSG recommendations, all pregnant women would undergo a three-step process of fasting, 1-hour and 2-hour plasma glucose tests during one appointment or lab visit. In the long run, this should simplify testing, but it will require adjustments on the part of both labs and obstetrical offices. For instance, Gabbe explained that OSU clinics currently don't offer a 75-gram test. “We'll have to identify where we can get that and have it available,” he said. In addition, the 50-gram test commonly in use today is non-fasting, so labs and physicians will need to ensure that patients understand that the 75-gram test is to be taken in a fasting state.
Kitzmiller believes the recommendations ultimately will streamline the gestational diabetes diagnostic process. “It makes it easier to organize if you have a system where every patient coming into your office is tested at the same time,” he said.
Timing of the test will be important as well, according to Gabbe. “If the mom has this test at 26 weeks gestation, we know in the baby the excessive growth occurs starting at 28 weeks. So we have to get right back to the mom and get treatment started as soon as possible,” he indicated.
Although experts consulted for this article agreed that if adopted by major professional organizations, the IADPSG recommendations will require some adjustments in the field, overall they are a positive development in the diagnosis and treatment of gestational diabetes. “This will increase the benefit to pregnant women to learn how to eat properly so they can control excess weight gain and have less complications in themselves and their babies,” said Kitzmiller. “It's wonderful that they will get that benefit, and it's my hope that every pregnant woman in the U.S. by 28 weeks will have a 75-gram, two-hour glucose tolerance test.”