ORIGINAL ARTICLE

The National Institute of Child Health and Human Development Maternal-Fetal Medicine Unit Network Randomized in Progress Standard therapy versus no therapy for mild gestational diabetes

1 5 MARK B. LANDON, MD GARLAND ANDERSON, MD rate for GDM that 2 ELIZABETH THOM, PHD FOR THE MATERNAL-FETAL MEDICINE would qualify as preexisting diabetes ac- 3 CATHERINE Y. SPONG, MD UNITS NETWORK,THE NATIONAL 4 cording to World Health Organization MARSHALL CARPENTER, MD INSTITUTE OF CHILD HEALTH AND 2 criteria (6–9). Below this threshold, the LISA MELE, MS UMAN EVELOPMENT ETHESDA 1 H D ,B , extent to which untreated GDM is accom- FRANCEE JOHNSON, RN ARYLAND 4 M * panied by excess is un- JOANN TILLINGHAST, RN certain. GDM, if untreated or not recognized, may also be associated with an increased t is recognized that women with gesta- apy and insulin as required versus those risk of several morbidities such as macro- tional diabetes mellitus (GDM) who randomized to no specific treatment. somia, birth trauma, neonatal hypoglyce- I have significantly elevated fasting This study aims to clarify whether there mia, hyperbilirubinemia, hypocalcemia, blood glucose levels are at increased is utility in identifying and treating , and respiratory distress risk for fetal macrosomia and perinatal women with a normal fasting glucose syndrome (7). The association of GDM morbidity if treatment is not provided level who meet standard criteria for with accelerated fetal growth resulting in (1,2). The association of milder forms of GDM. We plan to compare perinatal large infants at risk for birth trauma and GDM with perinatal morbidity and outcomes in women who have been ran- long-term metabolic effects of obesity has mortality remains unclear, primarily domized to diet and/or insulin therapy received the most clinical attention. How- because the condition is often con- with women who have been random- ever, the of GDM for ex- founded with other risk factors such as ized to no specific treatment. A random- maternal obesity, age, and parity. ized treatment trial of mild GDM will cessive fetal growth is not entirely clear Screening for GDM is recommended for clarify whether identification and treat- (10–12). Race, parity, BMI, and maternal most pregnant women, yet it is un- ment of mild GDM reduce perinatal weight gain have all been associated with known whether there is a benefit to the morbidity. This information will assist an increased risk of macrosomia (13). identification and treatment of mild car- in determining appropriate thresholds Sacks et al. (13) reported that only a weak bohydrate intolerance during preg- for the treatment of GDM. relationship exists between maternal glu- nancy (3,4). The present report is an cose levels and birth weight centiles, after update of our previous description of a BACKGROUND AND STUDY controlling for confounding variables. current ongoing randomized treatment RATIONALE — Overall, with broader Similarly, Casey et al. (12) found that only trial for mild GDM (5). A randomized identification and aggressive treatment, 12% of the excess risk for macrosomia in clinical trial of women with mild GDM perinatal mortality rates associated with their GDM population could be attrib- (fasting glucose Ͻ95 mg/dl) is being GDM appear to be similar to the nondia- uted to maternal carbohydrate intoler- undertaken that compares perinatal betic population (1). Several analyses of ance. The relationship between other outcomes in those receiving diet ther- 20 years ago did document an increased morbidities such as neonatal hypoglyce- mia and hyperbilirubinemia with mater- ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● nal glucose levels in GDM has not been From the 1Departments of and Gynecology, Ohio State University, Columbus, Ohio; the 2George well characterized. Retrospective studies Washington Biostatistics Center, Washington, DC; the 3National Institute of Child Health and Human suggest that intensive treatment of GDM 4 5 Development, Bethesda, Maryland; Brown University, Providence, Rhode Island; and the University of may reduce these morbidities (3). Texas at Galveston, Galveston, Texas. *Other members of the National Institute of Child Health and Human Development MFMU Network are As women with GDM represent a listed in the APPENDIX. metabolically heterogeneous group, this Address correspondence and reprint requests to Mark B. Landon, MD, The Ohio State University, College likely translates into a broad range of peri- of Medicine, Means Hall, 5th Floor, 1654 Upham Dr., Columbus, OH 43210-1228. E-mail: natal risk. This contributes to the contro- [email protected]. Received for publication 28 March 2006 and accepted in revised form 5 June 2006. versy surrounding whether a treatment This article is based on a presentation at a symposium. The symposium and the publication of this article benefit exists for pregnancies complicated were made possible by an unrestricted educational grant from LifeScan, Inc., a Johnson & Johnson company. by this disorder. Despite longstanding Abbreviations: GDM, gestational diabetes mellitus; MFMU, Maternal-Fetal Medicine Unit. recognition of this problem, little A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion factors for many substances. progress has been made in developing ev- DOI: 10.2337/dc07-s215 idence-based guidelines for screening and © 2007 by the American Diabetes Association. treatment (4). Nearly 15 years ago, an in-

S194 DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 Landon and Associates

Figure 1—Reprinted with permission from Landon et al. (5). ternational consensus conference on the and the National Institute of Diabetes and treat gestational diabetes mellitus in order adverse effects of gestational diabetes co- Digestive and Kidney Disease concluded to prevent adverse perinatal effects cannot sponsored by the National Institute of that “the sensitivity, specificity, and cost- be resolved without additional carefully Child Health and Human Development effectiveness of efforts to diagnosis and designed studies” (14). Similar conclu-

DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 S195 MFMU Network randomized clinical trial sions have been reached by the Canadian failed to blind practitioners to the diagno- ordinating Center of George Washington Task Force on periodic health examina- sis, which can affect obstetric decision University. Four groups of women are be- tion and by the U.S. Preventive Services making. Thus, a trial blinding the control ing enrolled. Women with mild GDM are Task Force (15). Most women in the U.S. group and practitioners to the oral glu- centrally randomized to one of two are subjected to glucose screening during cose tolerance test results was critical to groups (denoted groups I and IIA, respec- followed by formal diagnostic our study design. In selecting criteria for tively): Group 1: Formal nutritional testing in ϳ15–20% of the population. inclusion, it was apparent that practitio- counseling and diet therapy, along with GDM is being diagnosed with increasing ners and institutional review boards insulin if required; Group IIA: No specific frequency based on the less stringent re- would not accept blinding of test results treatment. vised criteria proposed by the Fourth In- in women with fasting hyperglycemia. In addition, women with a positive ternational Workshop-Conference. Most Moreover, many women with fasting hy- 50-g glucose challenge test, but with a of these additional cases represent mild perglycemia might have underlying overt normal subsequent oral glucose tolerance GDM, which undergo a variety of obstet- diabetes with associated increased perina- test, will be enrolled centrally in an obser- rical and medical interventions through- tal risk. However, a prospective clinical vational cohort (denoted group IIB) out the remainder of their pregnancy. trial involving women with milder forms matched by race and BMI (Ͻ27 kg/m2, Such interventions for GDM are currently of hyperglycemia was acceptable to both Ն27 kg/m2) to the women randomized to without much scientific justification and, the Steering Committee of the MFMU “no treatment.” Participating centers will in fact, may present risk to the patient so Network and Advisory Board based on be notified that a patient has been en- diagnosed. Studies have demonstrated a the lack of current evidence supporting rolled in group II and will thus be un- consistently higher rate of cesarean sec- treatment of this group of women. aware of whether the individual is a mild tion for those women designated as hav- GDM subject. ing GDM (16). STUDY DESIGN — We have previ- A group of nondiabetic control sub- The Maternal-Fetal Medicine Unit ously reported the design of this clinical jects with a negative 50-g glucose toler- (MFMU) Network of the National Insti- trial (5). The study aims to address the ance test (value Ͻ120 mg/dl) will also be tute of Child Health and Human Devel- primary research question, “Do women enrolled by the clinical centers. Recruit- opment conducts multicenter clinical with a singleton pregnancy diagnosed ment will be controlled by the Biostatisti- studies that address important issues in with mild gestational diabetes between 24 cal Coordinating Center for nondiabetic maternal-fetal medicine and obstetrics. and 29 weeks of gestation receiving diet patients in each race/BMI category. The Over a decade ago, the MFMU Network modification and performing self- race/BMI categories will be determined began planning a randomized clinical monitoring of blood glucose have a re- individually by center according to the ra- treatment trial for GDM. Numerous retro- duction in the of neonatal cial distribution among the gestational di- spective studies as well as a pilot random- morbidity and mortality, as compared abetic women at that center. The ized clinical trial were carefully with standard obstetrical care?” Mild nondiabetic control group (group III) will considered in the planning of our study. GDM is defined as a 3-h oral 100-g glu- Garner et al. (17) conducted a pilot study cose tolerance test with normal fasting have demographics similar to those pa- of strict glycemic control and tertiary care level (i.e., Ͻ95 mg/dl) and two of the tients randomized to treatment (group I) versus routine obstetric care in the man- three post–glucose load determinations (Table 1). agement of women with normal fasting exceeding thresholds established by the After central randomization (by the glucose levels (diet-controlled GDM). Fourth International Workshop- Biostatistical Coordinating Center after The aim of Garner’s pilot study was to Conference on Gestational Diabetes (17). verification of eligibility), the clinical cen- prepare for a multicenter trial by assessing The study will also address secondary ter will be notified only that a patient has patient acceptance, determining realistic research questions including whether been randomized to diet therapy (group I) accrual rates, and detecting any major ad- treatment of mild GDM reduces the risk or that she is in group II (no treatment). verse outcomes in the control group who of a large-for-gestational-age infant After enrollment by the Biostatistical Co- received routine obstetric care. Among and/or macrosomia (birth weight Ͼ4,000 ordinating Center, women in group I re- 300 GDM women randomized, there g), neonatal intensive care unit admis- ceive nutritional counseling and are were no differences in mean birth weight, sion, and maternal complications includ- instructed on the technique of performing macrosomia, or birth trauma between the ing cesarean section and preeclampsia. self-monitoring of blood glucose in both study groups. The mode of delivery was The study design also permits an analysis the fasting and postprandial states. Insu- also similar between the two groups, of whether mild GDM is associated with lin therapy is reserved for individuals in while the treatment group did have lower an increase in macrosomia and/or large- which the majority of either fasting levels preprandial and postpartum glucose lev- for-gestational-age infants compared with are Ն95 mg/dl or 2-h postprandial glu- els during the third trimester. a similar normal nondiabetic population. cose levels are Ͼ120 mg/dl after 1 week of A serious shortcoming of Garner’s The study represents a combination diet therapy. Oral hypoglycemic agents study was that women in the control arm of a randomized multicenter clinical treat- are not used for treatment. were not blinded to their diagnosis and ment trial and observational Women with mild GDM not random- could have been self-treating by modify- (Fig. 1). Glucose screening is performed ized to treatment (group IIA) are not clin- ing their diet. Moreover, these women ac- between 24 weeks, 0 days, and 29 weeks, ically distinguishable from those with an tually performed self-monitoring of blood 6 days, followed by a diagnostic 3-h glu- abnormal screen and normal oral glucose glucose and also could have modified cose tolerance test. The oral glucose tol- tolerance test results (group IIB). Both of their behavior based on testing results. erance test is analyzed centrally with these groups as well as the non-GDM Nearly all previous studies of GDM have results forwarded to the Biostatistical Co- (group III) are not receiving any specific

S196 DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 Landon and Associates

Table 1—Study group allocation

Oral glucose tolerance test Group Population Glucose challenge test Fasting 1, 2, 3 h Treatment Selection I Mild GDM Ն135 mg/dl Normal Positive Yes Randomized A Mild GDM Ն135 mg/dl Normal Positive No Randomized II B Non-GDM Ն135 mg/dl Normal Normal No Matched to I and IIA Non-GDM Ͻ120 mg/dl N/A Normal No Matched to I Reprinted with permission from Landon et al. (5). dietary therapy except for general nutri- The secondary outcome measures for SUMMARY — There is apparent tional recommendations for pregnancy. this study include fetal/neonatal outcomes widespread acceptance of universal labo- Routine antepartum fetal surveillance such as macrosomia, large for gestational ratory screening for GDM in the U.S., (nonstress testing) is not undertaken for age, neonatal ponderal index, brachial despite both the 2001 ACOG Technical group I until 40 weeks’ gestation unless plexus injury, fetal distress, and respiratory Bulletin on Gestational Diabetes and the obstetric indications exist. Women in all distress syndrome. Maternal secondary out- Fourth International Workshop- study groups are instructed on daily as- comes include cesarean section or operative Conference on Gestational Diabetes fail- sessment of fetal activity. Similarly, ultra- delivery, maternal weight gain, preeclamp- ing to endorse this practice (4,18). The sonography to assess fetal growth is sia, and length of stay. clinical dilemma of GDM was summa- limited to standard obstetrical indica- rized during the Fourth International tions. As ultrasonography may be used in STATISTICAL Workshop-Conference, which concluded treated individuals to assess fetal growth CONSIDERATIONS — We con- that “although there is a general consen- characteristics, these data will be col- ducted a survey of the literature to deter- sus that of GDM is increasing lected. mine the frequency of each component of globally, there is considerable contro- Cord blood is being collected for all the composite outcome in both treated versy about the clinical importance of groups and is forwarded centrally for C- and untreated GDM. The frequency of GDM and the magnitude of its impact on peptide determination. Infant birth these morbidities varies widely, due to mother and offspring” (14,18). weight is collected after delivery along differences in treatment as well as in the Similarly, the Canadian Task Force with neonatal anthropometric assess- populations studied. The morbidities are on Periodic Health Examination stated ment, including skinfold measurements, unlikely to be independent so that the fre- that “further research is needed to estab- length, head circumference, and upper quencies cannot be summed for an overall lish the of neonatal and peri- mid-arm circumference. A sample of heel frequency. However, it is assumed that natal illness in relation to various degrees stick blood is obtained for all infants to in the untreated mild GDM population, of sub-diabetic elevations in maternal assess glucose and bilirubin levels. the rate of hypoglycemia, hyperbiliru- blood glucose levels. The quality of avail- A composite end point with clinically binemia, hyperinsulinemia (C-peptide able evidence cannot support a recom- relevant components has been chosen as Ͼ95th percentile), and birth trauma is mendation to include universal screening the primary outcome. This consists of likely to be 20–30%. for gestational diabetes” (15). Instead, this perinatal mortality and morbidities that If the composite outcome in the un- panel suggested that a decision to proceed have been demonstrated to be associated treated group is 25%, a sample size of 950 with screening for this diagnosis must be with maternal hyperglycemia, specifically (475 per group) would be sufficient to made on other nonspecific grounds. This stillbirth, neonatal mortality, hypoglyce- detect a 30% reduction in the composite vague recommendation was accompa- mia, hyperbilirubinemia, neonatal hyper- outcome. An effect size smaller than a nied by an admission that the Task Force insulinemia, and birth trauma. 30% reduction would suggest that there is recognized that a proportion of women Macrosomia was not chosen for the marginal benefit to treatment of mild with various degrees of carbohydrate in- primary outcome, primarily because it GDM (type 1 error of 5% and power of tolerance during pregnancy will have ad- represents a secondary result associated 80%). verse outcomes and might benefit from with birth trauma and therefore is not a The primary evaluation of the clinical screening. health-based outcome. Comparison of trial will be an intent-to-treat analysis In 2003, the U.S. Preventative Ser- macrosomia rates could also yield a bi- based on the total cohort of patients ran- vices Task Force acknowledged that no ased result. Because obstetricians may be domized or enrolled. Standard statistical well-conducted randomized controlled more likely to proceed with delivery ear- methods for rates and proportions will be trial existed that provided direct evidence lier in women with GDM, fetuses destined applied for dichotomous variables, and for the health benefits of screening for to reach 4,000 g or a selected weight cut- the Wilcoxon’s rank-sum test will be used GDM (2). Whereas insulin therapy may off in the treated group may never reach to compare continuous variables. Statisti- decrease the incidence of fetal macroso- this end point. Therefore, it is possible cal analysis will be adjusted for any differ- mia in those pregnancies complicated by that the treatment could have an effect on ences in pretreatment factors among significant hyperglycemia, the magnitude the rate of macrosomia as a result of phy- groups. Subgroup analyses will also be of any effect on maternal and neonatal sician practice characteristics. conducted if a treatment effect exists. health remains uncertain. Moreover, the

DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 S197 MFMU Network randomized clinical trial

Task Force report noted that insufficient terpretation of any treatment effect in the Vanderbilt University: S. Gabbe; The evidence exists to determine if a health subset of GDM with mild disease. George Washington University Biostatistics benefit accompanies treatment for the To date, few additional studies have Center: C. Cobb, L. Leuchtenburg; Na- large number of women with GDM with examined the effectiveness of intensive tional Institute of Child Health and Human milder degrees of hyperglycemia. A ran- treatment among women with mild Development: S. Pagliaro. domized controlled trial is necessary to GDM. A summary of four trials including answer this question (2). With mild GDM 612 women with mild GDM found no ϳ affecting 1–3% of pregnancies, 50– benefit in dietary treatment in preventing References 100,000 women annually are being iden- adverse health outcomes (20). The ongo- 1. Brody SC, Harris R, Lohr K: Screening for tified and treated for this diagnosis, ing population-based Hyperglycemia and gestational diabetes: a summary of the ev- making this an important clinical issue. Adverse Pregnancy Outcome (HAPO) idence for the U.S. Preventive Services Most recently, Crowther et al. (19) re- study is designed to determine levels of Task Force. Obstet Gynecol 101:380–392, ported results on a 10-year multicenter carbohydrate intolerance during preg- 2003 randomized trial designed to determine nancy associated with adverse perinatal 2. Langer O, Yogev Y, Mast O, Yenakis E: whether treatment of GDM reduces the outcomes. Specifically, at what level of Gestational diabetes: the consequence of risk of perinatal complications. The pri- maternal glycemia is there fetal and/or not treating. Am J Obstet Gynecol 192: mary outcome of this study was a com- maternal risk? A continuum of risk is an- 989–997, 2005 3. ACOG Practice Bulletin Number 30: Gesta- posite of perinatal morbidity and ticipated as a likely result in the Hyper- tional Diabetes, September 2001. Washing- mortality (stillbirth, shoulder dystocia, glycemia and Adverse Pregnancy ton, DC, American College of Obstetricians bone fractures, nerve palsy, neonatal in- Outcome study. The previously cited and Gynecologists, 2001 tensive care unit admission, and jaun- Crowther study and the ongoing MFMU 4. American Diabetes Association: Gesta- dice). The authors found the composite trial address whether treatment of GDM is tional diabetes mellitus (Position State- rate of “serious” perinatal complications effective at reducing perinatal risk. Given ment). Diabetes Care 26:S103–S105, was lower among the infants of the 490 that the MFMU Network trial considers 2003 treated women compared with the 510 only women with normal fasting glucose 5. Landon MB, Thom E, Spong CY, Gabbe SG, infants in the routine care group (1 vs. levels, it should help clarify whether a Leindecker S, Johnson F, Lain K, Mi- 4%). This study represents the first large- benefit exists to treatment of mild GDM. odovnik M, Carpenter M: A planned ran- domized clinical trial of treatment for mild scale randomized treatment trial for The trial should also complement the Hy- gestational diabetes mellitus. J Matern Fetal GDM. The sample size is remarkably sim- perglycemia and Adverse Pregnancy Out- Neonatal Med 11:226–231, 2002 ilar to the ongoing MFMU trial; however, come study data in arriving at some 6. O’Sullivan JB, Mahan CM, Charles D: the Crowther study is different with re- consensus for selecting thresholds for di- Screening for criteria for high-risk gesta- spect to design and analysis of outcome agnosis and treatment of carbohydrate in- tional diabetic patients. Am J Obstet Gy- data. Most importantly, the criteria used tolerance during pregnancy. necol 116:895–891, 1973 for the Crowther study include women 7. Abell DA, Beischer NA, Wood C: Routine with more significant hyperglycemia than APPENDIX — Other members of the testing for gestational diabetes, preg- the MFMU trial. Thus, our study should National Institute of Child Health and nancy, hypoglycemia, and fetal growth re- provide additional information regarding Human Development MFMU Network tardation and results of treatment. J Perinat Med 4:197–212, 1976 the effect of treatment of women with are as follows: Ohio State University:J. 8. Oats JN, Beischer NA: Gestational diabetes. milder GDM. Iams, C. Latimer; University of Alabama at AustNZJObstet Gynaecol 26:2–9, 1986 The perinatal outcome composite in Birmingham: D. Rouse, W. Andrews, A. 9. Hod M, Merlob P, Friedman S, Schoenfield Crowther’s study includes shoulder dys- Northen, J. Sheppard; University of Texas A, Ovadia J: Gestational diabetes mellitus, a tocia, a subjective diagnosis, which in Southwestern Medical Center: B. Casey, K. survey of perinatal complications in the turn weighed heavily on the composite Leveno, J. McCampbell, L. Moseley, D. 1980s. Diabetes 40:74–78, 1991 difference observed among study groups. Bradford; University of Utah: M. Varner, 10. Leiken EL, Jenkins JH, Graves WL: Pro- An increased rate of admission to the neo- M. Belfort, B. Oshiro, K. Anderson, J. Par- phylactic insulin in gestational diabetes. natal intensive care unit was observed in sons; University of Pittsburgh: S. Caritis, M. Obstet Gynecol 70:587–592, 1987 the treatment group; however, no differ- Cotroneo, M. Bickus; Wake Forest Univer- 11. Lucas MJ, Lowe TW, Bone L, McIntire DD: Class A1 gestational diabetes: a ence was observed in the rate of hypogly- sity: M. Harper, P. Meis. M. Swain, K. meaningful diagnosis? Obstet Gynecol 28: cemia (secondary outcome) requiring Lanier; Drexel University: A. Sciscione, M. 260, 1993 intravenous therapy. Finally, details re- DiVito, M. Talucci; Wayne State Univer- 12. Casey BM, Lucas MJ, McIntire DD, Lev- garding several antepartum sity: Y. Sorokin, S. Blackwell, G. Norman, eno KJ: Pregnancy outcomes in women suggest these may not have been related P. Lockhardt; Columbia University:R. with gestational diabetes compared with to untreated GDM such that the overall Wapner, R. Berkowitz, S. Bouseleiman, V. the general obstetric population. Obstet conclusion of a reduction in “serious” Carmona; Brown University: E. Chien, D. Gynecol 90:873, 1997 perinatal complications in treated GDM Catlow; Northwestern University: A. Peace- 13. Sacks DA, Greenspoon JS, Abu-Fadil S, must be interpreted with caution. In con- man, W. Grobman, G. Mallett. P. Simon; Menry HM, Wolde-Tsadik G, Yao JFF: trast to the Crowther study, the MFMU University of North Carolina: J. Thorp, K. Toward universal criteria for gestational diabetes: the 75-gram glucose tolerance Network trial primary outcome includes Moise, K. Dorman, S. Brody, S. Timlin; test in pregnancy. Am J Obstet Gynecol two markers of fetal response to maternal University of Texas at Houston: S. Ramin, L. 172:607–614, 1995 hyperglycemia: neonatal hypoglycemia Gilstrap, D. Cross-Soebbing, J. Martinez; 14. Blank A, Grave GD, Metzer BE: Interna- and cord C-peptide levels. These out- Case Western Reserve University:B.Mer- tional NIH Consensus Conference: Ad- comes should allow for a meaningful in- cer, P. Catalano, C. Milluzzi, C. Santori; verse Perinatal Outcome of Gestational

S198 DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 Landon and Associates

Diabetes Mellitus. Diabetes Care 18: tes Investigators. JAMA 275:1165– tional Diabetes Mellitus. Diabetes Care 21 1227–1229, 1995 1170, 1996 (Suppl. 2):B161–B167, 1998 15. Canadian Task Force on the Periodic 17. Garner P, Okun N, Keely E, Wells G, Per- 19. Crowther CA, Hiller JE, Moss JR, McPhee Health Examination: Periodic heath ex- kins S, Sylvain J, Belcher J: A randomized AJ, Jeffries WS, Robinson JS, Australian aminations update. I. Screening for gesta- controlled trial of strict glycemic control Carbohydrate Intolerance Study in Preg- tional diabetes mellitus. Can Med Assoc J and tertiary level obstetric care in the nant Women (ACHOIS) Trial Group: Ef- 147:435–443, 1992 management of gestational diabetes: a pi- fect of treatment of gestational diabetes 16. Naylor CD, Sermer M, Chen E, Sykora lot study. Am J Obstet Gynecol 177:190– mellitus on pregnancy outcomes. N Engl K: Cesarean delivery in relations to birth 195, 1997 J Med 352:2477–2486, 2005 weight and gestational glucose toler- 18. Metzger BE, Coustan DR: Summary and 20. Cochrane Database: Dietary regulation for ance: pathophysiology or practice style? recommendations of the Fourth Interna- gestational diabetes. Cochrane Database Toronto Trihospital Gestational Diabe- tional Workshop-Conference on Gesta- Syst Rev 2000; 2:CD000070

DIABETES CARE, VOLUME 30, SUPPLEMENT 2, JULY 2007 S199