Volume 185, Number 6 A m J Obstet Gynecol

427 USE OF POSTPARTUM FASTING GLUCOSE AND LOW-RISK ANTENATAL CRITERIA T O REDUCE ORAL GLUCOSE TOLERANCE TESTING GESTATIONAL DIABETES UTE, M SCHAEFER-GRAF 1 , ANNY XIANG 2, RUTH, K PETERS ~, THOMAS, A BUCHANAN 4, SIRI, L KJOSS; 1Charitr , Obstetrics, Berlin, Berlin; 2University of Sou thern California, Preventive Medicine, Los Angeles, CA; 3University Sou thern California, Preventive Medicine, Los Angeles, CA; 4University of Southern California, Medicine, Los Angeles, CA; -~University of Southern California, Obstetrics, Los Angeles, CA

OBJECTIVE: To identify threshold levels for postpar tum fasting glucose, alone and in combinat ion with known antepar tmn predictors for diabetes to reasonably exclude early diabetes without performing glucose tolerance tests (oGTF) in women with gestational diabetes (GDM).

STUDY DESIGN: 8tepwise logistic regression was per formed to confirm independent an tepar tum predictors of diabetes in 2439 women with o G T r 1-4 months postpartum. After r andom division into t raining and validation co- horts, receiver operator curve (ROC) analysis was used (training) to examine postpartum-FPG alone and in combinat ion with confirmed predictor. The best thresholds were tested on the validation cohor t for NPV.

RESULTS: The independent au tepar tum predictors for diabetes (n = 283) were: highest fasting plasma glucose (FPG), oGTT-sum ( fas t ing+lH+2H glucose), gestational age (GA) at diagnosis, and glucose challenge test (GCT). The pos tpar tum-FPG <85 m g / d l , provided the best low-risk threshold (sensitivity 98.1%, negative predictive value (NPV) 98.7%. If women met one of four low-risk an tepar tum criteria, i.e. highest FPG <105mg/dl , oGTf-sum <500 mg/d l , GCT< 165 mg/d l , or GA diagnosis<27 weeks, a h igher postpartum- FPG threshold, <100 mg /d l , could be used while mainta ining a high NPV (98.9%). If any 2 low-risk criteria were met, a yet h igher threshold, <110 m g / d l could be used, improving NPV (>99.5%), When these thresholds were tested in the validation cohort , for postpartmn-FPG thresholds alone, with one and two low-risk criteria, the respective NPVs were 99.1%, 99.4%, and 99.7% and testing could be omitted in 17.7%, 61.9% and 55.5% of the cohort .

CONCLUSION: Postpartum diabetes testing using varying FPG thresholds in women with low-risk an t epa r tum criteria, could reduce oGTTs while maintaining high NPVs.

SMFM Abstracts S197

429 MACROSOMIA CONSIDERING MATERNAL ANTHROPOMETRY BEST PREDICTS NEONATAL MORBIDITY IN PREGNANCIES WITH DIABETES MELLITUS UTE, M SCHAEFER-GRAF l, GESINE AFIF 2, RENATE BERG- MANN l, SIRI, L KJOS 3, JOACHIM, W DUDENHAUSEN l, KLAUS VETTER2; tCharit~, Obstetrics, Berlin; 2Hospital Neukoelln, Obstetrics, Berlin; 3Univer- sity of Southern California, Obstetrics, Los Angeles, CA

OBJECTIVE: Macrnsomia in diabetic pregnancies is associated with an increased risk of neonatal morbidity. O u r study investigated whether defini- t ions for macrosomia using neona ta l length or mate rna l a n t h r o p o m e t r y improved predict ion of neonatal morbidity compared to a definition based on birth weight (BW) alone.

STUDY DESIGN: The macrosomia status of 613 term born singletons of women with diabetes (94.8% GDM by O'Sullivan) was retrospectively defined according to the 90th percentile for gestational age of 1.) BW, 2.) the ratio of BW/ leng th (BW/L), 3.) the BMI, a n d 4.) BW adjusted for materna l prepregnancy weight and height (BW/MW). Stepwise regression analysis was used to de te rmine which definit ion of macrosomia best predicts neonata l morbidity, i.e. neonata l hypoglycemia (_< 30mg/d l ) , in t ravenous glucose therapy (iv glucose) to treat hypoglycemia, transfer to neonatal intensive care unit (NICU) and C-section.

RESULTS: The maerosomia rate was 20.5% by BW, 27% by BW/L, 32.3% by BMI and 18.1% by BW/MW. Twenty-four % of the infants def ined as macrosomic by BW were redefined as normosomic by BW/MW. Hypoglycemia occurred in 10.6% of the newborns, iv glucose in 1.5%, transfer to NICU in 8.2% and C-section was per formed in 16.6%. BW/MW was selected by stepwise regression as the best predic tor for each neonata l morbidity. The negative predictive value of BW/MW for hypoglycemia was 91%, for iv glucose 99.2%, for transfer to NICU 94.2%, and for C-section 86.6%.

CONCLUSION: Growth curves including baby length did not improve identification of newborns at risk compared to BW alone. However a growth curve which considers materna l a n t h r o p 0 m e t r y to define macrosomia ameliorated the predict ion of neonatal morbidity.

428 PREDICTORS FOR NEONATAL HYPOGLYCEMIA IN LARGE-FOR- GESTATIONAL-AGE NEWBORNS UTE, M SCHAEFER-GRAF l, ANNETTE WILKE2, SIRI, L KJOS 3, CHRISTOPH BfiHRER 4, JOACHIM, W DUDEN- HAUSEN 1, KLAUS VETTER2; 1Charite, Obstetrics, Berlin; 2Hospital Neukoelln, Obstetrics, Berlin; 3University of Southern California, Obstetrics, Los A~lgeles, CA; 4Charite, Neonatology, Berlin

OBJECTIVE: Testing for neonata l hypoglycemia is r e c o m m e n d e d for large-for-gestational-age newborns (LGA). We investigated the rate and tinting of hypoglycemia in LGA newborns and possible materna l or neonata l predictors.

STUDY DESIGN: In a retrospective study of 1094 LGA newborns ~ 34 weeks of gestation univariate and stepwise regression was used to determine predictors for neonatal hypoglycemia (capillary glucose < 30mg/dl ) within the first 24 hours (H) of life. Glucose testing was per formed at 1 a n d / o r 2H of life with subsequent measurements based on clinical practice. Maternal parameters tested were historical variables, current OGTT values (n - 306), C- section and gestational age (GA) at delivery. Neonatal parameters included were Apgar /5 rain, cord pH; ratio of birth weight / length and neonatal BMI > 90th and LGA > 95th percentile.

RESULTS: Hypoglycemia occurred in 16.3% (178). There was a significant stepwise decrease of the rate of hypoglycemia (first occm'rence) from 15.9% at 1H to 2.4% at 2-6H and 0,7% at 7-24H (P< .001 and P = .04). By univariate and stepwise regression, the only independent predictor for hypoglycemia was GA at delivery (P= .002) with ROC analysis showing a decreased risk for > 39 compared to < 39 weeks (14.3% vs 20.6%, P < .01). In the subgroup of women with antenatal OGTT performed, the 1H value was the only independent predictor (P < .002). ROC analysis revealed a stepwise increased risk of hypoglycemia with increasing 1H levels: < 120, 120-199 and > 200 m g / d l with rates of hypoglycemia of 0%, 13.8% and 36.7% respectively (P= ,01 and P= .02).

CONCLUSION: Routine testing of LGA newborns appears to be indicated with a 16% prevalence of neonatal hypoglycemia. The rate of hypoglycemia d r o p p e d to <1% after 6H. In women not tested for GDM, there were no clinically useful predictors for neonatal hypoglycemia. In wmnen undergo ing OGTT, the 1H value seems to be useful to identify newborns at risk.

430 PROSPECTIVE COHORT STUDY TO ESTABLISH PREDICTORS OF GLY- BURIDE SUCCESS IN GESTATIONAL DIABETES MELLITUS RAMEN CHMAIT 1, THERESA DINISE 2, SEAN DANESHMAND 1 , MATTHEW KIM 1, THOMAS MOORE1; 1University of California, San Diego, Reproduct ive Medicine, San Diego, CA; eUniversity of California, San Diego, School of Medicine, San Diego, CA

OBJECTIVE: To establish the rate of glyburide failure (GF) in achieving glycemic cont ro l in gestational diabetics (GDM), and to develop cri ter ia predict ing glyburide success (GS).

STUDY DESIGN: A protocol was established to assign glyburide to all patients between 13 and 38 weeks with Class A2 GDM. Demographic data and four-times daily glucose values were recorded. GF was defined when maximmn dose glyburide (20rag/d) did no t maintain fasting blood sugars < 90 m g / d l and one hou r postprandials < 130 mg/d l . Chi square and Student 's t test were used to assess significance, and a receiver operat ing curve was constructed to identify predictors of GS.

RESULTS: All 42 patients offered the protocol were enrolled. Aside f rom lower maternal age in GF versus GS (26.1 vs. 31.9 years, P = .017), there were no differences in demographic data, glucose tolerance test results, birthweight and delivery mode. The overall rate of GF was 16.7% (7 of 42 patients). The average gestational age of glyburide initiation was 26.6 weeks in GF versus 31.9 weeks in GS (P = .014). Mean pre-treatment fasting blood sugars (FBS) were higher in GF (118.9 mg/d l ) than in GS (96.9 mg/d l , P= .002). GS was 44% for those with pre-treatment mean FBS > 110 mg/d l , and 95% if< 110 m g / d l ( P : .005). GS was 54% if inadequate glycemic control occurred on dietary therapy before 30 week.s, and 95% if greater than 30 weeks (P= .006).

CONCLUSION: Al though oral hypoglycemic therapy was universally accepted by our GDM patients, one-sixth of them failed to achieve glycemic control on maximuna dose glyhuride, requir ing the addition of insulin. GDM patients who fail dietary therapy after 30 weeks gestation and whose mean FBS is less than 110 m g / d l have a high likelihood of achieving glycemie control on glyburide.