Impact of earlier gestational diabetes screening for pregnant people with obesity on maternal and perinatal outcomes

Abstract

Objectives

Gestational diabetes (GDM) screening at 24–28 weeks’ gestation reduces risk of adverse maternal and perinatal outcomes. While experts recommend first-trimester screening for high-risk patients, including those with obesity, data supporting this recommendation is limited.

Methods

We implemented a systematic population intervention to encourage first-trimester GDM screening by oral glucose tolerance testing in a cohort of pregnant people with obesity in two integrated health systems from 2009 to 2013, and compared outcomes to the same population pre-intervention (2006–2009). Up to five years of postpartum glucose testing results (through 2018) were assessed among GDM cases in the post-intervention group. Primary outcomes were large-for-gestational-age birthweight (LGA); macrosomia; a perinatal composite outcome; gestational hypertension/preeclampsia; cesarean delivery; and medication treatment of GDM.

Results

A total of 40,206 patients (9,156 with obesity) were screened for GDM; 2,672 (6.6%) were diagnosed with GDM. Overall, multivariate adjusted risk for LGA and cesarean delivery were lower following the intervention (LGA: aOR 0.89 [0.82, 0.96]; cesarean delivery: 0.89 [0.85, 0.93]). This difference was more pronounced in patients diagnosed with GDM (LGA: aOR 0.52 [0.39, 0.70]; cesarean delivery 0.78 [0.65, 0.94]); insulin/oral hypoglycemic treatment rates for GDM were also higher post-intervention than pre-intervention (22 vs. 29%; p<0.0001). There were no differences for the other primary outcomes. Only 20% of patients diagnosed with GDM early in pregnancy who had postpartum testing had results in the overt diabetes range, suggesting a spectrum of diabetes detected early in pregnancy.

Conclusions

First trimester GDM screening for pregnant people with obesity may improve GDM-associated outcomes.

Introduction

Screening of pregnant patients for gestational diabetes mellitus (GDM) after 24 weeks’ gestation is recommended based on robust evidence that it improves maternal and fetal outcomes [1], [2], [3], [4]. For patients at high risk of GDM or of undiagnosed overt diabetes, including those with obesity, screening in the first trimester is recommended to advance earlier treatment [3, 4]. However, to date, there is limited evidence about how screening before 24 weeks’ gestation affects maternal and perinatal outcomes [1, 5].

While GDM is traditionally defined as any degree of glucose intolerance with onset or first recognition during pregnancy, and is traditionally diagnosed by oral glucose tolerance testing [5], some experts have suggested that newly identified diabetes in early pregnancy represents previously undiagnosed (“overt”) type 2 diabetes, and thus that patients at high risk of diabetes should be screened in the first trimester using the same tests and criteria used outside of pregnancy [3, 4, 6]. Some clinical practice guidelines identify the 2-step screening approach used to diagnose GDM in later pregnancy as valid for diabetes screening in early pregnancy, while acknowledging that the best test for screening in the first trimester is not clear [3]. Long-term data on postpartum glucose levels is needed to identify how frequently diabetes diagnosed in the first trimester continues following pregnancy, indicating overt type 2 diabetes. These data could inform screening method selection.

In this study, our primary aim was to test the hypothesis that a systematic early-GDM-screening program for pregnant patients with obesity in two integrated health systems would improve maternal and perinatal outcomes. Using 5 years of postpartum glucose screening data, we also determined how many patients diagnosed with GDM had elevated glucose postpartum, and how this varied based on timing of diagnosis.

Materials, subjects and methods

Study design

We implemented an intervention embedded in the electronic medical record (EMR) to encourage screening for GDM in the first trimester for all pregnant patients with obesity served by two integrated health systems. To assess the effects of the intervention, we compared outcomes in all pregnant patients age 18 and older who did not have pre-existing diagnosed diabetes, received GDM screening, and delivered a singleton pregnancy during the intervention period (October 2009–December 2013 at Kaiser Permanente Northwest [KPNW] and June 2010–December 2013 at Kaiser Permanente Hawaii [KPHI]) to those of pregnant patients meeting the same criteria who delivered in the three years prior to the intervention (October 2006–October 2009 in KPNW and June 2007–June 2010 at KPHI). Postpartum glucose testing (through 2018) was assessed in the intervention group.

Population and setting

Our study population consisted of all pregnant people who received care at KPNW from October 2006 to December 2013 and at KPHI from June 2007 to December 2013. Kaiser Permanente is an integrated health system that served over 500,000 members in its KPNW region (Oregon and Southwest Washington), and 200,000 members in its KPHI region during the study period, including Medicaid members. The demographic characteristics of both regions correspond closely to those of the local populations [7]. Both KPNW and KPHI maintain EMRs that track admissions, outpatient visits, laboratory tests, pharmacy dispenses, and outside claims and referrals. The research related to human use has been complied with all the relevant national regulations, institutional policies and in accordance the tenets of the Helsinki Declaration, and has been approved by the authors’ Institutional Review Boards. Specifically, Institutional Review Boards at both institutions approved the early screening intervention and evaluation as part of clinical care as the practice is clinically recommended for patients with obesity; this included approving waivers for individual consent as the intervention was minimal risk, clinically recommended, and conducted as part of clinical care.

GDM screening and treatment throughout the study period

KPNW and KPHI universally screen pregnant patients for GDM, typically at 24–28 weeks’ gestation. Before the intervention, patients who were identified as at high risk for gestational diabetes (e.g., due to prior history of GDM or macrosomia) were typically screened in the first trimester, and treated (if positive) or rescreened again at 24–28 weeks (if negative). Patients with obesity were not systematically screened in the first trimester prior to the intervention.

Throughout the study, GDM screening at KPNW and KPHI was conducted following the 2-step process described by O’Sullivan [8], irrespective of gestational age. First, patients were given a nonfasting 50 g 1-h glucose challenge test (GCT). A GCT result of >200 mg/dL was considered positive for GDM; patients with this result did not undergo further testing [9]. Patients with a GCT result ≥130 mg/dL at KPNW or ≥140 mg/dL at KPHI underwent a fasting, 100 g, 3-h oral glucose tolerance test (OGTT). GDM diagnosis was based on the Carpenter and Coustan criteria of two or more positive results of the four glucose concentrations measured with the 100 g OGTT: Fasting ≥95 mg/dL; 1 h≥180 mg/dL; 2 h≥155 mg/dL; and/or 3 h≥140 mg/dL [3, 4].

Throughout the study, regardless of timing of screening, patients diagnosed with GDM were referred to a dietician for individually-tailored lifestyle recommendations, weight management, and self-monitoring of glucose with the following glycemic targets [3, 10]: Fasting glucose <95 mg/dL; 1-h postprandial glucose <140 mg/dL; and 2-h postprandial glucose <120 mg/dL. If glycemic goals could not be achieved with lifestyle management, medication was initiated (class A2GDM), with insulin the medication of choice [10]. Recommended thresholds for initiating medication treatment were the same regardless of timing of GDM diagnosis, and remained consistent in the pre- and post-intervention periods [3, 10].

Early GDM screening intervention

Our intervention consisted of a best practice alert in the EMR recommending that obstetrics providers screen all patients with obesity for GDM in the first trimester of pregnancy. A pop-up notification recommending screening appeared within a patient’s medical record at their first prenatal visit if they had a body mass index (BMI) ≥30 kg/m². The notification included an electronic link that created a lab order for a 1-h GCT, the first step of the 2-step GDM screening protocol. Study physicians met with obstetrics providers at department-wide meetings and individual clinics in both regions to encourage intervention adherence prior to roll-out.

Perinatal outcome measures and covariates

Outcome measures were informed by prior literature [11], [12], [13] and included large-for-gestational-age birthweight (LGA; birthweight >90th percentile for sex and gestational age) [11], [12], [13], [14]; macrosomia (birthweight >4,000 g among newborns ≥37 weeks’ gestation) [12, 13]; gestational hypertension/preeclampsia (blood pressure ≥140/90 on two separate occasions after 20 weeks’ gestation among people with no prior diagnosis of essential hypertension and not taking antihypertensive medication) [3, 12, 13, 15]; cesarean delivery (ICD-9 diagnosis 669.7 or 763.4 or procedure 74.x) [11], [12], [13]; GDM requiring insulin or oral hypoglycemic treatment (class A2GDM) [3]; and a perinatal composite measure indicating at least one of the following adverse perinatal outcomes: stillbirth (ICD-9 delivery diagnosis V27.1, 634.x, 656.4x, 768.x, or 779.9, or procedure 69.02 or 69.52 occurring after 20 weeks’ gestation), neonatal death (death of newborn under 7 days old), shoulder dystocia (ICD-9 diagnosis 660.4), bone fracture (ICD-9 diagnosis 767.2, 767.3), and/or any upper-extremity nerve palsy related to a birth injury (ICD-9 diagnosis 767.6) [13]. While our study intervention was designed and conducted prior to the 2019 published core outcome set for prevention and treatment of GDM [16], our outcomes include nearly all of the currently recommended GDM core outcomes.

Covariates were assessed using clinical and administrative databases. Maternal race and ethnicity were collected at health plan enrollment. Parity and prior GDM (ICD-9 diagnosis 648.8) were based on the EMR and supplemented with data from locally validated pregnancy databases. Prior hypertension was based on EMR diagnoses (ICD-9 401, 642.0) and medication use. Pre-pregnancy BMI categories were calculated from the EMR using the most recent pre-pregnancy weight measurement up to 3 months prior to the estimated date of conception, or (if not available) the earliest weight measurement during pregnancy (up to 12 weeks’ gestation). Patients with pre-pregnancy BMI ≥30 kg/m² were classified as obese. Patients with measured BMI <30 in the 3-month pre-pregnancy interval, regardless of their BMI at the first prenatal visit, were classified as non-obese; those who had no measurement during the pre-pregnancy interval were classified as missing pre-pregnancy obesity status. Excessive gestational weight gain (GWG) was a dichotomous indicator of exceeding National Academy of Medicine (NAM) total GWG guidelines for a patient’s specific pre-pregnancy BMI category [17]. If pre-pregnancy weight or weight at delivery were missing, patients were classified as having missing GWG status.

While we initially intended to adjust for GDM in the overall population models, we discovered that HbA_1c and fasting plasma glucose (FPG) tests were ordered for some patients (n=1,488 or 3.7%) instead of the 2-step method, as these are also considered acceptable tests for early screening [3, 4]. To account for these different screening methods, we created a hierarchical glucose classification covariate based on all glucose screening and diagnostic tests conducted. Patients were classified as “GDM” if any test met 2-step OGTT GDM criteria; “FPG ≥95 mg/dL” if not classified as GDM, but FPG was measured at or above 95 mg/dL, either alone or as a component of the 3-hr OGTT; “HbA_1c ≥5.7%” if neither GDM or FPG criteria applied but HbA_1c was measured at or above 5.7%; “1+ OGTT” if prior criteria did not apply but any one component of the 3-h OGTT other than the FPG met a diagnostic threshold; and “no elevated glucose” if none of the above criteria applied.

5-Year postpartum maternal glucose testing

Because there is little data and great controversy about whether GDM diagnosed by OGTT early in pregnancy represents GDM, pre-existing type 2 diabetes, or a heterogenous mix, we evaluated glucose testing results in the 5 years postpartum among individuals diagnosed with GDM in early and late pregnancy in the post-intervention period to determine how rates of hyperglycemia differed. We categorized results of all glucose testing (2-h OGTT, FPG, or HbA_1c) occurring in the 5 years postpartum among patients diagnosed with GDM in the post-intervention group, and stratified these analyses by timing of GDM diagnosis. Tests occurring in subsequent pregnancies were excluded. Patients were classified as “positive” for diabetes if they had a one or more postpartum glucose measurements that met diabetes criteria (FPG ≥126 mg/dL, HbA_1c ≥6.5%, or 2-h post-75 g OGTT ≥200 mg/dL). They were classified as “impaired” if any test met impaired fasting or glucose tolerance criteria but not diabetes criteria (FPG 100–125 mg/dL, HbA_1c 5.7–6.4%, or 2-h post-75 g OGTT 140–199 mg/dL [4]. If all postpartum glucose testing was in the normal range, they were classified as “normal.”

Statistical methods

We compared distributional properties of demographic and clinical characteristics between the pre- and post-intervention groups using chi-square tests and t-tests, as appropriate, prior to analyses. We conducted multiple logistic regression to estimate adjusted odds ratios for each outcome between the pre- and post-intervention study periods in the full population (all pregnancies meeting inclusion criteria) and among all patients who were diagnosed with GDM. We conducted similar analyses among all patients in the population with known pre-pregnancy obesity. Models were adjusted for maternal age; race/ethnicity; nulliparity; prior hypertension; pre-pregnancy obesity (for models on the full population); site (KPNW or KPHI); and, with the exception of models evaluating insulin or oral medication use, exceeding NAM-recommended GWG guidelines. For the full population analysis, models were further adjusted for the hierarchical glucose classification described above, and all models of cesarean delivery outcomes were additionally adjusted for birthweight >4,000 g. For all analyses, missing data for pre-pregnancy obesity and excessive GWG were treated as unique categories within these variables. Sensitivity analyses, including complete case analysis and multiple imputation, were conducted to confirm that this treatment of missing data did not bias our results.

We compared postpartum glucose levels between patients with diagnosed early (<18 weeks) and later in pregnancy (≥18 weeks) using chi-square tests among GDM-diagnosed pregnancies in the post-intervention study period.

All analyses were performed using SAS Statistical Analysis System version 9.4 (SAS Institute, Cary, NC).

Results

Population characteristics and GDM screening by study period

Over the 6-year study period, 40,206 patients were screened for GDM, 9,156 of whom had pre-pregnancy obesity. In the full population, 2,672 patients (6.6%) were diagnosed with GDM. Characteristics of the pre- and post-intervention populations are shown in Table 1. Date of first glucose test shifted following the intervention to a pronounced bimodal distribution, demonstrating increased frequency of early testing in the post-intervention period (Figure 1).

Table 1:

Pregnancy characteristics by study period.

	Pre-intervention	Post-intervention	p-Value
Characteristic	n=18.404	n=21.802
Maternal age, years, mean (SD)	29.3 (5.8)	29.9 (5.7)	<0.0001
Race/ethnicity, n (%)			<0.0001
Caucasian	9,562 (52.0)	12.363 (56.7)
Native Hawaiian	2,233 (12.1)	2,277 (10.4)
Filipino	1,328 (7.2)	1,306 (6.0)
Japanese	562 (3.1)	517 (2.4)
Chinese	219 (1.2)	218 (1.0)
African American	512 (2.8)	602 (2.8)
Other Asian Pacific Islander	1,827 (9.9)	2,123 (9.7)
American Indian	189 (1.0)	204 (0.9)
Other	544 (3.0)	534 (2.4)
Unknown	1,428 (7.8)	1,658 (7.6)
Hispanica, n (%)	2,376 (16.1)	2,593 (17.0)	0.0264
Nulliparousa, n (%)	8,074 (43.9)	9,716 (45.2)	0.0075
Obesea (BMI≥30 kg/m2), n (%)
Yes	4,114 (26.0)	5,042 (25.9)	0.8570
No	11.733 (74.0)	14.443 (74.1)
Pre-pregnancy BMI groupa: n, %
Underweight: BMI<18.5 kg/m2	368 (2.3)	439 (2.3)	0.4597
Normal weight: BMI 18.5–24.9 kg/m2	7,100 (44.8)	8,617 (44.2)
Overweight: BMI 25.0–29.9 kg/m2	4,265 (26.9)	5,387 (27.7)
Obese: BMI≥30 kg/m2	4,114 (26.0)	5,042 (25.9)
Mean gestational weight gain, kg	13.4 (6.6)	13.2 (6.8)	0.0032b
Prior GDM, n (%)	244 (1.3)	317 (1.5)	0.2749
Prior hypertension, n (%)	2,160 (11.7)	2,855 (13.1)	<0.0001
Exceed NAM recommended weight gain guidelinea, n (%)
Yes	6,240 (50.6)	8,159 (50.2)	0.5092
No	6,098 (49.4)	8,100 (49.8)
Site, n (%)
Northwest	12.164 (66.1)	15.125 (69.4)	<0.0001
Hawaii	6,240 (33.9)	6,677 (30.6)
Gestational age, weeks, at 1st glucose test, mean, SD	26.4 (6.0)	23.2 (8.4)	<0.0001
GDM	1,201 (6.5)	1,471 (6.8)	0.3747

^aHispanic ethnicity unknown for n=10,207; Parity unknown for n=320; Obese and pre-pregnancy BMI group unknown for n=4,874; Exceed IOM recommended weight gain unknown for n=11.609. ^bp-Value is adjusted for gestational age. Body mass index (BMI) is measured in kilograms (kg) per meter squared (kg/m²). Gestational diabetes mellitus is abbreviated as GDM. Exceeding National Academy of Medicine (NAM) guidelines is exceeding total gestational weight gain based on a patient’s specific pre-pregnancy BMI category [17].

Figure 1:

Histogram of gestational age (in weeks) of first glucose screening test in the population before (pre) and after (post) the early gestational diabetes mellitus (GDM) screening intervention (n=40,206).

Post-intervention exhibits a shift to a bimodal distribution in the population with a markedly higher mode at 10 weeks’ gestation, representing the increased frequency of early screening in the first trimester.

Perinatal outcomes

Incidence rates of perinatal outcomes for the full population and among patients with obesity are presented in Table 2, along with incidence rates for the GDM subgroups.

Table 2:

Incidence of main outcomes before and after earlier gestational diabetes screening intervention.

	All pregnancies				Pregnancies among patients with obesity (BMI ≥ 30 kg/m2)
	Pre-intervention		Post-intervention		Pre-intervention		Post-intervention
	n=18.404		n=21.802		n=4,114		n=5,042
	Incidence	95% CIa	Incidence	95% CIa	Incidence	95% CIa	Incidence	95% CIa
Large for gestational ageb
Full population	8.0	(7.6, 8.4)	8.0	(7.6, 8.3)	12.9	(11.9, 14.0)	12.5	(11.6, 13.4)
GDM	11.9	(10.0, 13.8)	7.4	(6.0, 8.8)	16.7	(13.1, 20.3)	12.8	(9.8, 15.8)
Macrosomiac
Full population	10.7	(10.2, 11.1)	11.1	(10.6, 11.5)	16.1	(14.9, 17.3)	15.8	(14.8, 16.9)
GDM	12.9	(10.9, 14.9)	9.0	(7.4, 10.5)	17.5	(13.7, 21.3)	14.7	(11.4, 17.9)
Perinatal compositeb
Full population	2.4	(2.2, 2.7)	2.4	(2.2, 2.6)	3.0	(2.5, 3.6)	3.0	(2.5, 3.5)
GDM	2.8	(1.8, 3.8)	2.5	(1.7, 3.4)	3.0	(1.3, 4.7)	2.8	(1.2, 4.3)
Gestational hypertension/preeclampsiad
Full population	10.0	(9.6, 10.5)	10.3	(9.9, 10.7)	18.6	(17.3, 20.0)	18.0	(16.8, 19.3)
GDM	13.6	(11.4, 15.8)	11.4	(9.6, 13.3)	19.4	(15.0, 23.7)	17.5	(13.4, 21.7)
Cesarean deliveryb
Full population	26.2	(25.5, 26.8)	24.4	(23.9, 25.0)	35.6	(34.1, 37.1)	33.1	(31.7, 34.4)
GDM	35.0	(32.2, 37.7)	28.8	(26.4, 31.1)	43.3	(38.6, 48.0)	34.4	(30.3, 38.6)
Insulin or oral medication treatment among GDMe	22.5	(20.1, 24.9)	29.4	(27.1, 31.8)	35.0	(30.5, 39.5)	44.7	(40.3, 49.0)

^a95% Wald asymptotic confidence intervals are estimated using the continuity correction [31]. ^bLive births and stillbirths. ^cLive and still births of ≥37 weeks’ gestation. ^dExcludes individuals with pre-existing essential hypertension. ^eDuring the study period, 86% (604 out of 703) of GDM patients who received medications for hyperglycemia were treated with insulin. Gestational diabetes mellitus is abbreviated as GDM.

All pregnancies

After multivariate adjustment, rates of LGA and cesarean delivery in the full population were significantly lower in the post-intervention than pre-intervention period (LGA: aOR 0.89 [0.82, 0.96]; cesarean delivery: 0.89 [0.85, 0.93]; Figure 2). Risk reduction was even more pronounced post-intervention among patients with GDM, with significant differences in LGA (aOR 0.52 [0.39, 0.70]), macrosomia (aOR 0.64 [0.48, 0.85]), and cesarean delivery (aOR 0.78 [0.65, 0.94]). Rates of treatment with insulin/oral hypoglycemic medication among patients diagnosed with GDM were significantly higher post-intervention. There were no significant differences between pre- and post-intervention periods in other outcomes in the full population or among patients diagnosed with GDM.

Figure 2:

Adjusted odds ratios comparing outcomes post-intervention vs. pre-intervention for 1) all pregnancies in the full population (n=40,206) and among all patients with gestational diabetes mellitus (GDM; n=2,672) and 2) all patients with obesity (n=9,156) and patients with obesity and GDM (n=976).

Intervention was earlier GDM screening targeting patients with obesity. Odds ratios (ORs) for maternal and perinatal outcomes are adjusted for maternal age, race/ethnicity, nulliparity, prior hypertension, pre-pregnancy obesity (for models on the full population), site and, with the exception of models evaluating insulin or oral medication use, exceeding National Academy of Medicine (NAM) weight gain guidelines. For the full population groups, ORs are also adjusted for hierarchical elevated glucose, and all cesarean delivery models were also adjusted for birthweight >4,000 g. Among all patients with obesity, the interaction term of intervention period x exceeding NAM weight gain guidelines is significant for gestational hypertension/preeclampsia. Stratification by exceeding NAM weight gain guidelines yields the following OR (95% CI) for post- vs. pre- intervention: weight gain exceeds guidelines OR=0.73 (0.61, 0.88); Weight gain does not exceed guidelines OR=0.88 (0.69, 1.12); missing weight gain OR=0.65 (0.47, 0.91).

Pregnancies among patients with obesity

Among patients with obesity who were diagnosed with GDM, risk of LGA (aOR 0.62 [0.41, 0.93]) and cesarean delivery (0.59 [0.43, 0.80]) were significantly lower post-intervention than pre-intervention (Figure 2), and mean gestational age for initiating medical treatment was significantly earlier (post-intervention: 24 weeks, vs. pre-intervention: 28 weeks). Among all patients with obesity, the interaction term of intervention period x exceeding NAM weight gain guidelines was significant for gestational hypertension/preeclampsia. Stratification by exceeding NAM weight gain guidelines yielded the following ORs (95% CI) for post- vs. pre-intervention: Weight gain exceeds guidelines OR=0.73 (0.61, 0.88); Weight gain does not exceed guidelines OR=0.88 (0.69, 1.12); Missing weight gain OR=0.65 (0.47, 0.91). Thus, women with obesity who exceeded GWG guidelines (or for whom weight gain was missing) had reduced risk of hypertension/pre-eclampsia in the post-intervention period compared to pre-intervention.

Sensitivity analyses

Sensitivity analyses, including multiple imputation and complete case analyses conducted to address missing pre-pregnancy BMI and maternal weight gain data, were consistent with reported findings (data not shown).

Postpartum hyperglycemia in the 5 years following testing

Among the 1,144 patients diagnosed with GDM in the post-intervention period, 80.4% of those diagnosed early in pregnancy and 77.1% of those diagnosed later in pregnancy had at least one diabetes screening test in the 5 years following delivery. Of patients tested postpartum, significantly more who had been diagnosed early had impaired or diabetes-range postpartum glucose than those diagnosed later in pregnancy (59.4 vs. 37.6%, p<0.0001), yet, over 40% of these patients diagnosed early reverted to having normal glucose postpartum (Figure 3). Specifically, among the 234 patients diagnosed with GDM early in pregnancy who were tested postpartum, 47 (20.1%) were in the diabetes range, 92 (39.3%) had impaired glucose, and 95 (40.6%) had normal glucose postpartum. Among the 910 patients diagnosed with usual GDM who were tested postpartum, 54 (5.9%) were in the diabetes range, 288 (31.6%) had impaired glucose, and 568 (62.4%) had normal glucose postpartum.

Figure 3:

Postpartum glucose screening within 5 years of delivery among patients diagnosed with gestational diabetes mellitus (GDM) in the post-intervention period, by timing of diagnosis.

The targeted gestational age of early screening is the first trimester (<12 weeks) for high-risk patients and 24–28 weeks’ gestation for usual screening, and the 2nd diagnostic step of the oral glucose tolerance test (OGTT) is typically done about one week after initial screening. For analyses, we defined early GDM diagnosis as <18 weeks’ gestation and usual as ≥18 weeks’ gestation and the actual GDM diagnosis occurred at a median gestational 11 weeks’ gestation for early GDM and 29 weeks’ gestation for Usual GDM. Overall, 80.4% of patients diagnosed with GDM early in pregnancy (n=234) and 77.1% of those diagnosed with GDM later in pregnancy (n=910) had at least one diabetes screening test (Fasting plasma glucose (FPG), hemoglobin A1c (HbA_1c) or 75 g OGTT) in the 5 years following delivery. Results are stratified into the following three distinct categories based on postpartum testing: 1) any test meeting diabetes range criteria (positive test), 2) any test meeting impaired glucose range criteria and no test meeting diabetes range criteria, and 3) all tests were normal [4].

Discussion

An intervention to encourage first-trimester GDM screening for pregnant patients with obesity was associated with reduced rates of LGA and cesarean delivery in the full population of over 40,000 patients screened for GDM during pregnancy. As would be expected, these patterns were most pronounced among patients diagnosed with GDM, including those with obesity. Further, over 40% of patients diagnosed with GDM by early screening reverted to normoglycemia postpartum, indicating that not all cases of GDM identified early were cases of overt Type 2 diabetes. Thus, early screening of pregnant patients with obesity identified a heterogenous mix of GDM and overt diabetes cases, and led to improved perinatal outcomes in the full population, presumably through the advancement of GDM diagnosis and treatment in patients with obesity who were diagnosed early.

Standard GDM treatment included intensive weight management; initiating this gestational weight management earlier in pregnancy likely contributed to improved outcomes following the intervention. Excessive GWG has been shown to increase the risk of adverse GDM-associated maternal and perinatal outcomes, including LGA, macrosomia, and cesarean delivery in patients with and without GDM [18], [19], [20]. Weight gain in the first half of pregnancy is most related to birthweight [21], and a recent multi-racial GDM cohort study that found that GDM was associated with higher estimated fetal birthweight as early as 20 weeks’ gestation, indicating that GDM diagnosis and intervention before this point could be particularly important to improve birthweight outcomes [22]. In previously published data on this intervention, we found that patients with obesity who were diagnosed with GDM early were significantly less likely to exceed GWG guidelines than those diagnosed later [23]. Thus, diagnosing GDM early in pregnancy may have allowed for weight management during a critical intervention window for improving outcomes.

Advancing GDM diagnosis and treatment also likely led to both the earlier, and nearly 40% higher, rate of medication treatment of GDM in the post-intervention period; these differences in medication treatment in turn likely contributed to improved LGA and cesarean delivery outcomes in this period. Our finding of higher rates of medication treatment following early GDM diagnosis is consistent with prior work by our team and others suggesting that early GDM may be a more severe phenotype than GDM identified later [9, 24], [25], [26]. Nonetheless, our data suggest that early GDM treatment (dietary management, glucose monitoring, and medication treatment as needed) can successfully treat early GDM and improve several perinatal outcomes, including LGA and cesarean delivery.

The point estimates we observed for LGA, macrosomia, and cesarean delivery are also similar to findings in a recent randomized controlled trial that randomized the timing of GDM screening (14–20 or 24–28 weeks) for patients with obesity [27]; differences in a primary perinatal composite outcome were not significant in the trial.

This study addresses a previously identified need for data on the consequences of detecting and treating hyperglycemia in early pregnancy, and can inform classification and diagnosis criteria [28], [29], [30]. Results of this study show a positive population impact of early GDM screening (and thus diagnosis and treatment) of high-risk pregnant patients on maternal and perinatal outcomes. Interestingly, in a recently published pragmatic randomized clinical trial of screening after 24 weeks, we found that the 1-step GDM screening method diagnosed more than twice as many cases than the 2-step method (by diagnosing milder cases), but despite this difference in diagnosis rate, there were no significant differences between the two screening methods on the same maternal and perinatal outcomes studied here [7]. Taken together, these findings suggest that focusing care on patients at high risk for GDM early in pregnancy can have a greater impact on population health than diagnosing and treating more mild cases of GDM in later pregnancy.

Strengths of the present study include a diverse population of over 40,000 patients, including vulnerable groups, such as Medicaid members, who are often underrepresented in research. Additionally, GDM screening and treatment protocols were consistent at all gestational ages and across the full study period. One limitation of our study was that it was not appropriate to randomize pregnant patients with obesity to receive early screening or no early screening, as such screening is clinically recommended for this population [3, 4]. Thus, our observational design leaves open the possibility that temporally confounding factors could have contributed to observed differences. Additionally, some providers conducted early screening with FPG and HbA_1c in the post-intervention period, following clinical recommendations published in 2008 [4]. To account for these different screening methods, we used hierarchical glucose categories as a covariate in analyses of the full population.

In conclusion, an intervention to screen patients with obesity for GDM early in pregnancy was associated with reduced rates of LGA birthweight and cesarean delivery in the full population, with greatest effects among patients with GDM. These findings support the clinical recommendation of GDM screening in early pregnancy for patients with obesity.