Clinical Effectiveness of Continuous Glucose Monitoring in Pregnancies Affected by Type 1 Diabetes

Abstract

Background:

Continuous glucose monitoring (CGM) improves neonatal outcomes in type 1 diabetes pregnancies; however, its effectiveness has not been assessed in a real-world setting in the United States.

Objective:

The Triple C Study aimed to examine the clinical effectiveness, assessed through maternal glucose control and gestational health outcomes, of CGM use compared with self-monitoring of blood glucose (SMBG) in pregnancies associated with type 1 diabetes in a real-world setting.

Research Design and Methods:

We retrospectively identified 160 type 1 diabetes pregnancies at the Barbara Davis Center for Diabetes managed with CGM therapy (n = 109) or SMBG (n = 51) over a 6.5-year period (2014–2020). Obstetric care was provided at multiple practices. CGM use was defined as ≥60% wear in the second and third trimesters of pregnancy. Data were obtained from the electronic medical record system, hospital records, and vital statistics departments (Colorado and Wyoming). We used Student's t-test for continuous variables and chi-square test for categorical variables to compare outcomes between groups.

Results:

The CGM group had more participants meeting trimester-specific hemoglobin A1C (HbA1c) goals throughout pregnancy and postpartum (P < 0.01 in each time period). The CGM group had fewer participants never meeting HbA1c goals in any trimester than the SMBG group (P < 0.001). There were no significant differences in neonatal outcomes between groups, other than for macrosomia (12.8% CGM vs. 29.4% SMBG, P = 0.01). Infants of CGM users required a neonatal intensive care unit admission less often (52.9% CGM vs. 68.3% SMBG, P = 0.0989).

Conclusions:

CGM use was associated with improved maternal glucose levels in a diverse real-world cohort.

Introduction

Pregnancies associated with hyperglycemia have a significantly increased risk of adverse maternal and fetal health outcomes. Pregnancies associated with type 1 diabetes, in particular, are associated with high rates of adverse outcomes and with more glycemic variability than in pregnancies associated with type 2 diabetes or gestational diabetes.^1

–8 Hyperglycemia as measured by glycated hemoglobin A1C (HbA1c) levels has been historically associated with many adverse outcomes of pregnancy, including congenital malformations, fetal loss, macrosomia, and neonatal and infant death.⁹ Although HbA1c is clearly a critical overall measure of glucose control, emerging evidence indicates that it may not be the only glucose metric of importance.¹⁰

Moreover, HbA1c has limitations. One retrospective study found that HbA1c levels of ≥8.5% during early pregnancy did not predict most adverse outcomes such as preeclampsia (PE), small-for-gestational age (SGA) infants, or preterm delivery, possibly since HbA1c levels may only reflect an average of high and low glucose levels rather than persistent euglycemia.¹¹ In addition, while the level of glycemia is clearly a factor that is associated with gestational health outcomes, it is not the only one. The CARNATION study showed the benefit of comprehensive management for type 1 diabetes pregnancies such as preconception care, limiting maternal weight gain, and strict blood pressure control.¹²

Continuous glucose monitoring (CGM) is one mode of therapy that has been shown to improve gestational health outcomes with glucose metrics other than HbA1c levels. The CONCEPTT Study was a multicenter, multinational, randomized controlled trial (RCT), in which pregnant women with type 1 diabetes were randomized to CGM or self-monitoring of blood glucose (SMBG).¹³ CONCEPTT found that the use of CGM therapy was associated with significant reductions in adverse neonatal outcomes, including large-for-gestational age (LGA) infants and severe neonatal hypoglycemia, despite the fact that the HbA1c levels between groups was similarly low and changed to a similar degree (0.19% HbA1c difference between groups in change from baseline to 34 weeks favoring CGM, P = 0.0207).

There was a significantly higher percentage of maternal glucose time spent in the pregnancy-specific time-in-range (TIR; 63–140 mg/dL) and a significantly lower percentage of time in the pregnancy-specific hyperglycemic range (>140 mg/dL) among CGM users, which accounted for these improved neonatal outcomes.¹⁴ The data from this RCT were critical to implementing guidance on CGM use in pregnancies associated with diabetes. However, the CONCEPTT Study cohort was relatively homogeneous, and so, it was unclear if results were generalizable. In addition, although maternal glucose levels were improved through CGM metrics with CONCEPTT, the rates of adverse maternal health outcomes were similar between groups.¹³ When an RCT is unlikely to be repeated, it can be helpful to see if results borne out in an RCT remain consistent in real-world studies.

There is a lack of data surrounding real-world differences in HbA1c levels between CGM and SMBG therapies in pregnancy, CGM's ability to affect maternal health outcomes, and whether CGM use in more heterogeneous populations would result in similar effectiveness compared with that seen in an RCT.

The findings from the CONCEPTT Study may not be generalizable as most participants self-identified as European origin, college-educated, and nonsmokers.¹³ Many other available observational studies and RCTs looking at the effects of CGM use on pregnancies associated with type 1 diabetes are similarly less generalizable as many of them examine a specific population in a certain geographic area or were unable to recruit a diverse pool of participants.^15

–19

GlucoMOMS, another RCT investigating CGM use in pregnancies associated with diabetes, had a study cohort consisting of 95% white Europeans, while an RCT in Malaysia recruited 50 women, most of whom were of Malay ethnicity.^17,18 Furthermore, it is unclear if CGM therapy can reduce adverse maternal outcomes in addition to neonatal outcomes. Thus, we first set out to examine the clinical effectiveness, assessed through maternal glucose control and maternal and neonatal health outcomes, of CGM use compared with SMBG use in pregnancies associated with type 1 diabetes in a real-world setting with a racially, ethnically, and socioeconomically diverse group of participants in the Clinical Effectiveness and Cost-Effectiveness of CGM Use in Pregnancy (Triple C) Study.

Research Design and Methods

Study design and stratification

We conducted a retrospective chart review of the electronic medical record (EMR) system at the Barbara Davis Center for Diabetes (BDC) for women attending the Pregnancy and Women's Health Clinic for pregnancy care between January 1, 2014, and August 31, 2020. The purpose of this study was twofold: (1) To evaluate the clinical effectiveness of CGM use compared with SMBG use in pregnancies associated with T1D in a real-world study, which we herein present, and (2) to evaluate the cost-benefit of CGM use compared with SMBG use in pregnancies associated with T1D managed in the United States, which we plan to present separately. The first pregnancy visit had to be between January 1, 2014, and December 31, 2019, and the delivery date could be no later than August 31, 2020. EMR search criteria were used to identify pregnancies.

Patients who were younger than 18 years on December 31, 2019, and older than 55 years on January 1, 2014, were excluded. Additional exclusion criteria included diabetes other than type 1 diabetes (type 2 diabetes, gestational diabetes mellitus, MODY, etc.), use of basal insulin alone, use of bolus insulin alone if on multiple daily injection therapy, BDC care for preconception counseling rather than for pregnancy, failure to attend the BDC Pregnancy and Women's Health clinic for pregnancy care at least once each trimester, first trimester fetal loss during the study inclusion time period, <3 prenatal visits at the BDC during the pregnancy, and a multifetal gestational pregnancy (two or more fetuses).

The study inclusion criterion of a clinic visit at our center at least once per trimester was required so that we could examine HbA1c levels in each trimester in each group and as an indicator that our clinic protocol and best practices for pregnancy care would be implemented for all participants in both groups. Only one pregnancy per woman was allowed for inclusion, and thus, women with multiple pregnancies within the study time frame would have the first eligible pregnancy included in the final cohort, whenever possible. The final cohort of women with deliveries who met the inclusion/exclusion criteria was stratified to either CGM therapy or SMBG based on available CGM data or lack thereof. This study was approved by the Colorado Multiple Institutional Review Board (COMIRB).

BDC pregnancy and women's health clinic protocol for T1D pregnancy care

Details of the clinic care protocol for management of T1D pregnancies were provided in the supplementary data of a previous publication.²⁰ Very briefly, pregnant individuals had clinic or telehealth visits every 1–4 weeks, most commonly every 4 weeks. Point-of-care HbA1c measurements were obtained at each in-person clinic visit. Interim glucose management, including reviews of glucose meter, CGM, and insulin pump data (if applicable), was recommended every week in between clinic/telehealth visits. Thus, insulin dosing adjustments were advised every week of pregnancy, although not all individuals provided glucose and pump data to clinicians with this frequency. A registered dietitian provided nutrition counseling at least once each trimester. After national diabetes guidelines began recommending low-dose aspirin administration for PE prevention,²¹ the BDC started advising pregnant individuals with T1D to take aspirin 81 mg orally daily (the dose recommended in the guidelines) starting at 12 weeks of gestation.

Data collection and inclusion/exclusion criteria

EMR search criteria to identify cases included the following: (1) obstetric history in which at least one episode of pregnancy (start date, estimated delivery date, end date, or delivery date) fell within the specified time range of January 1, 2014, and December 31, 2019, for a first pregnancy visit with delivery no later than August 31, 2020; (2) ICD10 codes O24.011, O24.012, O24.013, O24.019, O24.311, O24.313, O24.919, O24.93, V22.0, or V23.89 within the problem list, all of which are related to type 1 diabetes pregnancies (first, second, third trimester, or unspecified); and (3) the keywords “pregnant,” “pregnancy,” “gestation,” or “gestational age” in clinic notes of visits within the specified time range. Inclusion/exclusion criteria were then applied to identify pregnancies eligible for inclusion in the study. Manual chart review was performed to verify the accuracy of the identified cases.

Baseline characteristics data were gathered through the EMR system at the BDC for all women with at least one pregnancy visit in the study time period, which includes individuals meeting the inclusion criteria with a verified type 1 diabetes pregnancy but met other exclusion criteria. HbA1c measurements were obtained through the EMR system at the BDC. These included measurements through point-of-care performed on the same machine at the BDC throughout the study time period and maintained as per the manufacturer's instructions, venous HbA1c measurements in our local laboratory, and venous HbA1c measurements obtained at outside commercial laboratories. As per the clinical care protocol, point-of-care measurements were obtained at each clinic visit (monthly for most patients in pregnancy) and venous measurements were obtained as per the provider's discretion but especially for telehealth visits.

The clinic care protocol also advises a postpartum clinic visit at 4–6 weeks postpartum, in which an HbA1C measurement is obtained, although some patients may have come outside of this window. As obstetric care was provided at multiple practices in three different states, estimated due dates to calculate gestational ages were gathered through the EMR system at the BDC, maternal pregnancy health outcomes data were gathered through the EMR system at the BDC (self-report), by manual review of labor and delivery (L&D) admission records for mothers and infants at the University of Colorado and other hospital systems for whom patients authorized data sharing, the vital statistics department of Colorado for pregnancies delivered in the state of Colorado, and the vital statistics department of Wyoming for pregnancies delivered in the state of Wyoming.

Vital statistics data were not available for women who delivered in Nebraska (n = 2). Fetal/neonatal birth and death records were collected from infant hospital admission records at the time of delivery (birth records only), the Colorado Department of Public Health and Environment Center for Health and Environmental Data and the Colorado Vital Records Department for children in the state of Colorado, and Wyoming Department of Health Vital Statistics Services for children in the state of Wyoming. When records could not otherwise be obtained through the above-stated means, women still attending the BDC clinic were invited to sign informed consent in person or electronically to obtain maternal L&D admission records and infant hospital records from the L&D admission.

Manual review of records was used to obtain and/or to verify specific outcome measures such as gestational age at first pregnancy visit, gestational age at delivery, birth weight, delivery type, pregnancy complications, labor complications, neonatal complications, and neonatal intensive care unit (NICU) admissions and length of stay. Outcomes were also collected from self-reported data at the immediate postpartum clinic visit through the EMR. Maternal HbA1c levels, retinopathy examination results, urine protein/creatinine results, self-reported occurrences of severe hypoglycemia during gestation, and self-reported occurrences of diabetic ketoacidosis during gestation were also collected from the BDC EMR system.

Initial CGM use was captured from the BDC EMR system and then verified by examination of CGM data uploaded to CGM software systems at the BDC during gestation. Raw CGM data were obtained from LibreView^® software for Abbott CGM systems, Dexcom Clarity^® software for Dexcom CGM systems, and CareLink^® Clinical software for Medtronic CGM systems since the most used CGM systems during the study period were Flash Libre, Dexcom G4 or G5 Platinum^® CGM systems, Dexcom G6^® CGM system, Medtronic Enlite™ Sensor, and Medtronic Guardian^® 3 Sensor. CGM use was defined as ≥60% wear in the second and third trimesters of pregnancy through raw CGM data or clinic uploaded reports of CGM wear.

During the time period of study, it was part of the clinical pregnancy care protocol of the BDC Pregnancy and Women's Health Clinic to advise all pregnant individuals to continue to perform SMBG testing, even while using a CGM device, as per guidelines from the American Diabetes Association.²² For CGM devices requiring blood glucose measurements for calibrations, all clinic patients were advised to perform SMBG and calibrations as per the instructions of the CGM manufacturers.

Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at the University of Colorado Denver. REDCap is a secure, web-based application designed to support data capture for research studies.

Statistical analysis

LGA was defined as greater than the 90th percentile infant birth weight. Appropriate-for-gestational age (AGA) was defined as between the 10th and 90th percentile infant birth weight. SGA was defined as below the 10th percentile infant birth weight. Birth weight centile, LGA, AGA, and SGA were calculated and individually adjusted for maternal characteristics (race/ethnicity, height, weight, parity) and infant characteristics (gestation-adjusted birth weight and sex) using a customized growth calculator from the Gestation Related Optimal Weight software from the Gestation Network. Macrosomia was defined as infant birth weight >4 kg (8 pounds 13 ounces) and >4.5 kg (9 pounds 15 ounces). PE was defined as a PE diagnosis from L&D admission records or a self-reported diagnosis in the immediate postpartum clinic visit when not otherwise available.

HbA1c pregnancy goals were defined as below 6.5% (48 mmol/mol) in trimester 1 (T1) and below 6.0% (42 mmol/mol) in trimesters 2 (T2) and 3 (T3) as per the guidelines by the American Diabetes Association.²² T1 was defined as the time period from the last menstrual period until 13 weeks 6 days of gestation, T2 as 14 weeks 0 days until 27 weeks 6 days of gestation, and T3 as 28 weeks 0 days of gestation until delivery. The arithmetic mean for all HbA1cs measured during pregnancy was calculated and reported as “average HbA1c,” and the arithmetic mean for all HbA1cs measured during each trimester was calculated and reported per trimester.

In the CGM group, pregnancy-specific TIR was defined as glucose between 63 and 140 mg/dL. Time below range (TBR) was defined as glucose <63 mg/dL and level 2 hypoglycemia was defined as glucose <54 mg/dL. Time above range (TAR) was defined as glucose >140 mg/dL and level 2 hyperglycemia was defined as glucose >180 mg/dL. Gestational weight gain was determined from the difference between weight before pregnancy and weight at delivery as defined in the vital statistics records.

Characteristics of women included in the study were compared between the CGM and SMBG groups using Student's t-test for continuous variables and chi-square test for categorical variables. For the demographic data, ANOVA tests were performed to compare characteristics between the CGM, SMBG, and excluded pregnancies groups, and t-tests were used for overall comparison of included pregnancies (CGM and SMBG) versus excluded pregnancies for continuous variables. We compared glycemic metrics, including HbA1c average during the pregnancy and by trimester using Student's t-test, and the ability or inability to meet HbA1c goals between groups using chi-square tests. Maternal and neonatal characteristics were compared between the CGM and SMBG groups using Student's t-test for continuous variables and chi-square test for categorical variables. Linear mixed models were used to examine changes in CGM metrics across the pregnancy in the CGM group.

There was no single gestational health outcome of interest and as this was a retrospective study, the sample size was determined by eligible women in our clinic population, and thus, no sample size calculations were performed. In a post hoc power analysis, the sample size of 109 CGM and 51 SMBG users provided 80% power to detect a difference in HbA1c of 0.5% assuming a standard deviation (SD) of 1.0%, and a difference in birth weight of 335 g, assuming a SD of 700 g. These effect sizes are considered medium. A P-value <0.05 was considered statistically significant.

Data and resource availability

The data sets generated during and/or analyzed in the current study are available from the corresponding author upon reasonable request.

Results

The EMR search yielded 932 subjects identified using the search criteria (Fig. 1). In total, 669 women were excluded for not being pregnant, not having type 1 diabetes in pregnancy, being pregnant outside the specified study time frame, or exclusion ages. There were 263 women found to be pregnant and having type 1 diabetes within the study time frame, for whom we collected baseline characteristics. Of these 263 women, 103 were excluded due to failure to attend the BDC Pregnancy and Women's Health Clinic for pregnancy care at least once each trimester, first trimester loss, <3 prenatal visits at the BDC, or a multiple gestational pregnancy (two or more fetuses). There were an additional 90 pregnancies identified within the same time frame from the same women (e.g., individuals with two to three pregnancies within the study time frame), and thus, these extra pregnancies were not included in the final cohort.

FIG. 1.

Methods' flowchart. The identification of pregnancies and stratification into CGM therapy and SMBG using defined exclusion criteria. CGM, continuous glucose monitoring; SMBG, self-monitoring of blood glucose.

This yielded a final cohort of 160 pregnancies from unique individuals for whom we collected maternal and fetal health outcome data. The final cohort of included pregnant women was stratified to CGM therapy (n = 109) or SMBG (n = 51). Among those stratified to the CGM group, 78 (71.6%) were using a CGM at baseline and an additional 31 (28.4%) started therapy after the first pregnancy visit.

The baseline characteristics of included participants are shown in Table 1 and those of individuals excluded for fetal loss, too few prenatal visits, missing trimester-specific prenatal visits, and multifetal gestation are in Supplementary Table S1. CGM users were significantly older, had a significantly higher rate of commercial insurance and CGM use, and a lower HbA1c, while other baseline characteristics were similar between groups (Table 1). The types of CGM devices, insulin pumps, and modes of insulin delivery with and without automation are shown in Supplementary Tables S2 and S3. The median and average number of visits for clinical care throughout pregnancy was 8 and 9, respectively, in both the SMBG and CGM groups.

Table 1.

Baseline Characteristics

	CGM (n = 109)	SMBG ^a (n = 51)	P
Age, years, mean ± SD	29.1 ± 5.3	26.9 ± 5.7	0.02
Diabetes duration, years, mean ± SD	16.0 ± 7.8	14.4 ± 7.7	0.23
BMI, kg/m², mean ± SD	27.2 ± 4.6	25.9 ± 4.1	0.10
Race/ethnicity, n (%)			0.38
American Indian/Alaska Native	1 (0.9)	0 (0)
Asian	1 (0.9)	0 (0)
Hispanic	12 (11.0)	7 (13.7)
Non-Hispanic black	1 (0.9)	3 (5.9)
Non-Hispanic white	83 (76.2)	38 (74.5)
Other	16 (14.7)	7 (13.7)
Refused to answer	0 (0)	0 (0)
Insurance, n (%)			0.007
Commercial	85 (78.0)	28 (54.9)
Medicare or Tricare	5 (4.6)	1 (2.0)
Medicaid	15 (13.8)	19 (37.3)
Other	4 (3.7)	3 (5.9)
Gestational age at first pregnancy visit, weeks, mean ± SD	6.6 ± 1.9	7.1 ± 2.2	0.17
CGM use at baseline, n (%)	78 (71.6)	12 (23.5)	<0.001
Insulin pump use at baseline, n (%)	77 (70.6)	34 (66.7)	0.61
Started insulin pump use after first pregnancy visit, n (%)	31 (28.4)	11 (21.6)	0.35
HbA1c at first pregnancy visit, n (%)	7.0 (1.2)	7.7 (1.5)	0.002

Data missing for SMBG group for gestational age at first pregnancy visit and HbA1c at first pregnancy visit (n = 1).

BMI, body mass index; CGM, continuous glucose monitoring; HbA1c, hemoglobin A1C; SD, standard deviation; SMBG, self-monitoring of blood glucose.

CGM users had a lower HbA1c across pregnancy compared with pregnant women using SMBG (6.5% CGM vs. 7.1% SMBG, P < 0.001, Table 2) and in each trimester (6.0%–6.6% range across trimesters in CGM group vs. 6.5%–7.4% range across trimesters in SMBG group, P < 0.01 for all, Table 2), and all of these associations remained statistically significant after adjusting for baseline differences in HbA1c except for the HbA1c difference between groups in the third trimester (P = 0.07 for third trimester but <0.05 for all others, data not shown). CGM users were more likely to meet HbA1c goals in all trimesters (P < 0.01 in each trimester, Table 2). More individuals in the SMBG group never met any HbA1c goals in any trimester than in the CGM group (70.6% SMBG vs. 33.9% CGM, P < 0.001, Table 2).

Table 2.

Glycemic Control

	CGM (n = 109) ^a	SMBG (n = 51) ^b	P
Average HbA1c across pregnancy (%)	6.5 ± 0.8	7.1 ± 1.0	<0.001
HbA1c in T1 (%)	6.6 ± 0.9	7.4 ± 1.2	<0.001
HbA1c in T2 (%)	6.0 ± 0.7	6.5 ± 0.8	<0.001
HbA1c in T3 (%)	6.2 ± 0.7	6.7 ± 0.9	0.001
HbA1c postpartum (%)	7.0 ± 1.2	7.8 ± 1.7	0.003
Met HbA1c goal T1 (≤6.5%)^c	53 (48.6)	11 (22.0)	<0.001^d
Met HbA1c goal T2 (≤6%)^c	61 (56.0)	10 (20.0)
Met HbA1c goal T3 (≤6%)^c	49 (46.7)	9 (18.4)
Never met HbA1c goal in any trimester	37 (33.9)	36 (70.6)
CGM metrics in T1 (%)
Average sensor readings, n	8898
Mean sensor glucose, mg/dL	135.6 ± 2.6
TBR <54 mg/dL	2.6 ± 0.2
TBR <63 mg/dL	6.4 ± 0.4	N/A	N/A
TIR 63–140 mg/dL	55.8 ± 1.7
TAR >140 mg/dL	38.9 ± 1.8
TAR >180 mg/dL	20.6 ± 1.5
CGM metrics in T2 (%)
Average sensor readings, n	15,150
Mean sensor glucose, mg/dL	134.8 ± 2.6
TBR <54 mg/dL	2.4 ± 0.2
TBR <63 mg/dL	5.6 ± 0.4^e	N/A	N/A
TIR 63–140 mg/dL	56.5 ± 1.7
TAR >140 mg/dL	39.0 ± 1.8
TAR >180 mg/dL	19.1 ± 1.5
CGM metrics in T3 (%)
Average sensor readings, n	10,354
Mean sensor glucose, mg/dL	133.1 ± 2.6
TBR <54 mg/dL	1.7 ± 0.2^f,g
TBR <63 mg/dL	4.3 ± 0.4^f,g	N/A	N/A
TIR 63–140 mg/dL	59.1 ± 1.7^e,g
TAR >140 mg/dL	37.8 ± 1.8
TAR >180 mg/dL	16.8 ± 1.5^f,g

Data are presented as mean ± SD, least square mean ± standard error (CGM data), or n (%).

Data missing for CGM group for HbA1c in T3 (n = 4) and postpartum (n = 1). Data missing for met HbA1c goal T3 (n = 4). Raw data missing for CGM metrics in T1 (n = 17), T2 (n = 13), and T3 (n = 15).

Data missing for SMBG group for HbA1c in T1 (n = 1), T2 (n = 1), T3 (n = 2), and postpartum (n = 4). Data missing for met HbA1c goal for T1 (n = 1), T2 (n = 1), and T3 (n = 2).

P-values for met HbA1c goal at T1, T2, and T3 are P = 0.0011, P < 0.0001, and P = 0.005, respectively.

P-value comparing those who never met HbA1c goals in any trimester to those meeting HbA1c goals in trimesters 1–3.

P < 0.05 compared with T1.

P < 0.001 compared with T1.

P < 0.05 compared with T2.

T1, first trimester; T2, second trimester; T3, third trimester; TAR, time above range; TBR, time below range; TIR, time in range.

Among CGM users, the least square mean ± standard error (SE) for TIR (63–140 mg/dL) changed from 55.8% ± 1.7% in the first trimester to 59.1% ± 1.7% (P < 0.001) in the third trimester with a least square mean ± SE reduction in the TBR <63 mg/dL from 6.4% ± 0.4% to 4.3% ± 0.4% (P < 0.001), respectively, from the first to third trimester (Table 2).

CGM users had a higher rate of PE (32.1% CGM vs. 17.7% SMBG, P = 0.0499). CGM users had less macrosomia by both definitions (12.8% CGM vs. 29.4% SMBG, P = 0.01 for birth weight >4000 g; 2.8% CGM vs. 15.7% SMBG, P = 0.004 for birth weight >4500 g) and trends for infants with lower mean birth weights (3382 g CGM vs. 3598 g SMBG, P = 0.07) and lower rates of LGA infants (46.8% CGM vs. 60.8% SMBG, P = 0.0978, Table 3). Fewer infants of CGM users required an NICU admission of 24 h or longer (52.9% CGM vs. 68.3% SMBG, P = 0.0989, Table 3). Other maternal and neonatal health outcomes were similar between groups.

Table 3.

Maternal and Neonatal Outcomes

Outcomes	CGM (n = 109) ^a	SMBG (n = 51) ^b	P
Maternal outcomes
Gestational weight gain, kg, mean ± SD	15.6 ± 7.4	15.0 ± 6.0	0.64
Gestational hypertension, n (%)	24 (22.0)	7 (13.7)	0.21
Preeclampsia, n (%)	35 (32.1)	9 (17.7)	0.0499
Cesarean section, n (%)	75 (68.8)	33 (64.7)	0.61
Duration of hospital stay, days, mean ± SD	4.9 ± 2.1	5.5 ± 4.8	0.45
Neonatal outcomes
Female sex, n (%)	48 (44.0)	21 (41.2)	0.73
Gestational age at delivery, weeks, mean ± SD	36.8 ± 2.1	36.9 ± 1.4	0.89
Preterm delivery, n (%)	35 (32.1)	19 (37.3)	0.52
Birth weight, g, mean ± SD	3382 ± 665	3598 ± 760	0.07
Birth weight centile (%)	74.2 ± 29.4	80.6 ± 28.3	0.19
Macrosomia >4000 g, n (%)	14 (12.8)	15 (29.4)	0.01
Macrosomia >4500 g, n (%)	3 (2.8)	8 (15.7)	0.004
Gestational size for age^c			0.14
Small-for-gestational age	4 (3.7)	3 (5.9)
Appropriate-for-gestational age	54 (49.5)	17 (33.3)
Large-for-gestational age	51 (46.8)	31 (60.8)
Duration of hospital stay, days, mean ± SD	9.2 ± 14.0	7.5 ± 8.2	0.46
NICU admission, n (%)	45 (52.9)	28 (68.3)	0.10
Duration of NICU stay, days, median [IQR]^d	5.5 [3.3–14.0]	5.0 [3.0–15.0]	0.64

Data missing for CGM group for gestational weight gain (n = 6), preeclampsia (n = 1), maternal duration of hospital stay (n = 27), infant duration of hospital stay (n = 94), percentage of NICU admission (n = 24), and median duration of NICU. Data representative of n = 40 for the median duration of NICU stay.

Data missing for SMBG group for gestational weight gain (n = 3), maternal duration of hospital stay (n = 12), infant duration of hospital stay (n = 44), and percentage of NICU admission (n = 10). Data representative of 25 for the median duration of NICU stay.

P-values for SGA and LGA are P = 0.68 and P = 0.0978, respectively.

Data representative of 40 in the CGM group and 25 in the SMBG group in the infant duration of NICU stay.

IQR, interquartile range; LGA, large-for-gestational age; NICU, neonatal intensive care unit; SGA, small-for-gestational age.

Discussion

In the Triple C Study, we found that CGM users had significantly lower HbA1c levels averaged across pregnancy (P < 0.001) and in each individual trimester of pregnancy compared with pregnant individuals using SMBG (P < 0.01 for all trimesters). Moreover, we found larger differences in HbA1c levels between the SMBG group and the CGM group than was reported in the largest RCT in a similar population (CONCEPTT Study). The mean HbA1c levels between groups differed by 0.8% in the first trimester (6.6% CGM vs. 7.4% SMBG, P < 0.001), 0.5% in the second trimester (6.0% CGM vs. 6.5% SMBG, P < 0.001), and 0.5% in the third trimester (6.2% CGM vs. 6.7% SMBG, P = 0.001) in the Triple C Study.

However, the CONCEPTT Study showed a 0.17% difference between groups at 24 weeks (6.23% CGM vs. 6.40% SMBG) and a 0.18% difference between groups at 34 weeks (6.35% CGM vs. 6.53% SMBG).¹³ Lemaitre et al. performed a retrospective, observational study of type 1 diabetes pregnancies managed with intermittently scanned CGM versus SMBG in a single French center.²³ Similar to CONCEPTT, they did not find large differences in HbA1c levels between groups in the second or third trimesters (0.1% in the second trimester, and 0% and 0.1% differences at two time points in the third trimester).²³ Sibiak et al. conducted a retrospective study with T1D pregnancies in Poland matched by baseline HbA1c and White's diabetes classification and then stratified by patient decision to use a CGM or SMBG (n = 42 per group).²⁴ They found similar HbA1c levels between groups in the matched cohort (differences of 0.07% in second trimester and 0.06% in third trimester), but a slightly larger and significant difference in HbA1c in the third trimester in an unmatched cohort (5.74% CGM [n = 75] vs. 5.96% SMBG [n = 59], P = 0.04).²⁴ It is possible that our larger HbA1c differences are a reflection of a real-world study, rather than an RCT, and/or a relatively more racially/ethnically diverse population. Of note, the rates of insulin pump use in our cohort were higher than those reported in some other pregnancy technology studies^13,15,23 and some participants were using insulin pumps with partial or full automation (e.g., basal suspend-on-low, basal predictive low glucose suspend, automated insulin delivery with hybrid closed-loop therapies), and it is unclear how these factors may have influenced glycemic levels at baseline or over time.

The use of CGM was associated with a significantly higher rate of individuals meeting trimester specific HbA1c goals throughout pregnancy in the Triple C Study (P < 0.001 for all trimesters). About half of the individuals using a CGM achieved the HbA1c goal in each trimester of pregnancy compared with about 20% of individuals using SMBG. These significant differences in HbA1c levels likely had an impact on the lower rate of LGA infants in the CGM users (46.8% CGM vs. 60.8% SMBG, P = 0.0978) in our real-world study. The prevalence of LGA infants born to individuals using a CGM in the Triple C Study (46.8%, n = 51) was similar to other large studies of type 1 diabetes pregnancies, including an RCT (53% CONCEPTT, n = 100)¹³ and a Swedish retrospective observational study with data from two clinical centers (53%, n = 186).¹⁵

On average, CGM users did not reach the pregnancy specific CGM targets of <4% TBR, >70% TIR, and <25% TAR (Table 2).²⁵ This is similar to findings from other studies. In CONCEPTT, the TIR was 68% among CGM users and 61% in the SMBG group at 34 weeks.¹³ They also found a higher TBR than is recommended for both groups at 34 weeks (6.7% CGM and 7.0% SMBG). A Swedish observational study of T1D pregnancies found TBR, TIR, and TAR in the first trimester to be 7.0%, 50.0%, and 43.0%, respectively (n = 155); with improvements in glucose levels over gestation such that in the third trimester these values were 6.5%, 59.8%, and 33.7%, respectively.¹⁵

While CGM users overall had lower HbA1c levels, it was unexpected that CGM users also had higher rates of PE compared with SMBG users. The increased risk of PE in T1D is well known, with rates up to four times higher in women with diabetes than those in the general population.²⁶ Poor placentation in the pathophysiology of PE is well documented, and there is increased evidence that chronic hyperglycemia may cause increased oxygen radical production, interfering with placental development and prolonged depression of endothelial function.²⁶ However, there are a multitude of risk factors that have also been found to contribute to PE, including genetic risk factors, triglyceride levels, cholesterol/HDL ApoE concentrations, chronic hypertension, history of acute kidney injury, family history of myocardial infarction, previous PE, diabetic nephropathy before pregnancy, and excessive gestational weight gain.^27,28

In total, evidence shows that the pathogenesis of PE is multifactorial, with other key players besides glycemic control being relevant, which may explain the higher PE rates we see in our Triple C cohort. Importantly, the American Diabetes Association Standards of Care first started recommending routine administration of low-dose aspirin (60–150 mg/day) to decrease the risk of PE in women with T1D in 2018,²¹ toward the end of our study time frame, and this was a recommendation that we incorporated into our routine care for T1D pregnancies at that time. Of note, even though the rate of PE was higher among CGM users, the number of NICU admissions and median duration of NICU stays were not.

Another important fetal health parameter that differed between groups in the Triple C Study was the number of infants admitted to the NICU for 24 h or more. We found that 52.9% (50/95) of infants born to CGM users were admitted to the NICU compared with 68.3% (28/44) of infants born to individuals using SMBG (P = 0.0989). CONCEPTT also showed a reduction in NICU admissions >24 h in CGM users (27% CGM vs. 43% SMBG, P = 0.02).¹³ The prevalence rates of NICU admissions in T1D pregnancies vary across studies, with some (such as CONCEPTT) reporting lower rates^13,15 and others reporting rates similar or higher^29
–31 to what we found in the Triple C Study.

Unfortunately, the differences in glycemic control, as reflected by HbA1c levels, did not translate to any significant beneficial differences in maternal health outcomes in the Triple C Study, including gestational weight gain, gestational hypertension, PE, cesarean deliveries, or maternal hospital length of stay. This striking lack of improvement in maternal gestational health outcomes of CGM users was also seen in the CONCEPTT Study¹³ and the French observational study.²³ The retrospective Polish study did find a lower rate of cesarean deliveries among the unmatched CGM users (64% CGM vs. 80% SMBG, P = 0.048), but no other differences in reported maternal health outcomes.²⁴ While there may be many reasons, it is possible that even greater differences in glucose control may be required, such as those seen with the use of hybrid closed-loop systems, to achieve improvements in maternal health.

This study has some limitations. This was a retrospective observational study, and thus, there may be biases that lead to CGM prescriptions or patient preferences for CGM use, although most baseline characteristics were similar between groups. The retrospective nature of the study also made it difficult to reliably determine which participants received preconception counseling, so we were unable to examine this as a possible contributor to differences observed between groups. As this was not an RCT, we also did not have equal numbers of CGM users compared with those using SMBG alone, making our SMBG sample size relatively small in comparison, which is likely a major contributor to the fact that any observed differences in maternal and neonatal health outcomes were not found to be significantly different between groups.

While the participants in the SMBG group did not use the CGM enough to meet inclusion in the CGM group, some did use these devices intermittently. RCTs of intermittent use of CGM did not have consistent results in terms of effects on gestational health outcomes and some showed no effects,¹⁶ and so, it is difficult to know if the partial use of CGM by some SMBG participants influenced our study findings. Also, there were multiple CGM systems used. This is not likely to affect the ability to lower HbA1c as has been shown with multiple different CGM systems, but it may have impacted the descriptive data of CGM metrics. While we obtained hospital records and vital statistics for most of the pregnancies, some maternal and fetal outcomes were only obtained through self-report, which may not be comprehensive or fully accurate.

Nevertheless, our study also had several strengths. We were able to obtain vital statistical records for all but two pregnancies. We obtained hospital L&D and infant records for most of the pregnancies. We verified type 1 diabetes pregnancy status through manual chart review for all pregnancies. We verified CGM use through chart review and raw data capture. Our cohort was relatively racially/ethnically diverse compared with other studies publishing data in type 1 diabetes pregnancies.

In conclusion, in this real-world study with a U.S. population of type 1 diabetes pregnancies, we found significantly lower HbA1c levels throughout pregnancy, in each trimester, and postpartum in individuals using CGM compared with those using SMBG. More CGM users were able to reach HbA1c goals in each trimester compared with those using SMBG alone. There were no significant improvements in maternal health outcomes in the Triple C Study. There were fewer babies born with macrosomia among CGM users. There were trends for fewer LGA infants and a lower prevalence of high-level neonatal admissions among CGM users in type 1 diabetes pregnancies, but these did not reach statistical significance in our cohort. More studies are needed to see if further improvements in glucose control, such as those that may be seen with hybrid closed-loop systems, can lead to improvements in maternal health outcomes and/or further improvements in neonatal health outcomes.

Footnotes

Acknowledgments

The authors thank the participants of this study for contributing their data. They thank Elizabeth Westfeldt and Lisa Sher for their assistance in data collection. The authors also thank Tim Vigers for his contribution to the computer programming for CGM data analysis.

Authors' Contributions

S.P. is responsible for the conception, design, conduct, and funding of the study. S.P. and J.K.S.-B. were involved in the methodology and formal analysis of the study. All authors were involved in the validation and data curation. V.G. and S.P. wrote the first draft of this article, and all authors edited, reviewed, and approved the final version of the article. S.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Author Disclosure Statement

S.P. reports research funding from Dexcom, Inc., Eli Lilly, JDRF, The Leona & Harry Helmsley Charitable Trust, NIDDK, and Sanofi US Services; research support from Diasome Pharmaceuticals, Inc., Medtronic MiniMed, Inc., and Sanofi U.S. Services; contributing writer: diaTribe; Medical Advisory Board: Medtronic MiniMed, Inc. (2020). V.G., C.J., E.M., and J.K.S.-B. have no conflicts.

Funding Information

This was an investigator-initiated study, in part, supported by Dexcom, Inc., through the Board of Regents at the University of Colorado Denver. This study was supported by NIH/NCRR Colorado CTSI grant number UL1 RR025780. Its contents are the authors' sole responsibility and do not necessarily represent Dexcom, Inc. or official NIH views. The funders had no role in data collection and analysis, decision to publish, or preparation of the abstract. Sarit Polsky is the guarantor of this work, has full access to all the data in the study, takes responsibility for the integrity of the data, and takes responsibility for the accuracy of the data analysis.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

References

Alessi

, Wiegand

, Hirakata

, et al. Temporal changes in characteristics and outcomes among pregnant women with pre-gestational diabetes. Int J Gynaecol Obstet, 2018; 143(1):59–65; doi: 10.1002/ijgo.12590

Gainor

, Fitch

, Pollard

. Maternal diabetes and perinatal outcomes in West Virginia Medicaid enrollees. W V Med J, 2006; 102(1):314–316.

Murphy

, Bell

, Cartwright

, et al. Improved pregnancy outcomes in women with type 1 and type 2 diabetes but substantial clinic-to-clinic variations: A prospective nationwide study. Diabetologia, 2017; 60(9):1668–1677; doi: 10.1007/s00125-017-4314-3

Nasri

, Houde Ng

, Westgate

, et al. Malformations among infants of mothers with insulin-dependent diabetes: Is there a recognizable pattern of abnormalities?. Birth Defects Res, 2018; 110(2):108–113; doi: 10.1002/bdr2.1155

Polsky

, Garcetti

. CGM, pregnancy, and remote monitoring. Diabetes Technol Ther, 2017; 19(Suppl 3):S49–S59; doi: 10.1089/dia.2017.0023

Starikov

, Dudley

, Reddy

. Stillbirth in the pregnancy complicated by diabetes. Curr Diab Rep, 2015; 15(3):11; doi: 10.1007/s11892-015-0580-y

Van Zyl

, Levitt

. Pregnancy outcome in patients with pregestational and gestational diabetes attending Groote Schuur Hospital, Cape Town, South Africa. S Afr Med J, 2018; 108(9):772–776; doi: 10.7196/SAMJ.2018.v108i9.12992

Weissgerber

, Mudd

. Preeclampsia and diabetes. Curr Diab Rep, 2015; 15(3):9; doi: 10.1007/s11892-015-0579-4

Buschur

, Stetson

, Barbour

. Diabetes in Pregnancy. In: Endotext. ( Feingold

, Anawalt

, Boyce

, et al. eds.) MDText.com, Inc.: South Dartmouth, MA; 2000. PMID: 25905241.

10.

Kitzmiller

, Block

, Brown

, et al. Managing preexisting diabetes for pregnancy: Summary of evidence and consensus recommendations for care. Diabetes Care, 2008; 31(5):1060–1079; doi: 10.2337/dc08-9020

11.

Starikov

, Inman

, Chien

, et al. Can hemoglobin A1c in early pregnancy predict adverse pregnancy outcomes in diabetic patients?. J Diabetes Complications, 2014; 28(2):203–207; doi: 10.1016/j.jdiacomp.2013.10.004

12.

Zheng

, Yang

, Luo

, et al. Association of implementation of a comprehensive preconception-to-pregnancy management plan with pregnancy outcomes among Chinese pregnant women with type 1 diabetes: The CARNATION study. Diabetes Care, 2021; 44(4):883–892; doi: 10.2337/dc20-2692

13.

Feig

, Donovan

, Corcoy

, et al. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): A multicentre international randomised controlled trial. Lancet, 2017; 390(10110):2347–2359; doi: 10.1016/S0140-6736(17)32400-5

14.

Garg

, Polsky

. Continuous glucose monitoring in pregnant women with type 1 diabetes. Lancet, 2017; 390(10110):2329–2331; doi: 10.1016/S0140-6736(17)32449-2

15.

Kristensen

, Ogge

, Sengpiel

, et al. Continuous glucose monitoring in pregnant women with type 1 diabetes: an observational cohort study of 186 pregnancies. Diabetologia, 2019; 62(7):1143–1153; doi: 10.1007/s00125-019-4850-0

16.

O'Malley

, Wang

, Ogyaadu

, et al. Assessing glycemic control using CGM for women with diabetes in pregnancy. Curr Diab Rep, 2021; 21(11):44; doi: 10.1007/s11892-021-01415-2

17.

Paramasivam

, Chinna

, Singh

AKK

, et al. Continuous glucose monitoring results in lower HbA(1c) in Malaysian women with insulin-treated gestational diabetes: A randomized controlled trial. Diabet Med, 2018; 35(8):1118–1129; doi: 10.1111/dme.13649

18.

Voormolen

, DeVries

, Sanson

RME

, et al. Continuous glucose monitoring during diabetic pregnancy (GlucoMOMS): A multicentre randomized controlled trial. Diabetes Obes Metab, 2018; 20(8):1894–1902; doi: 10.1111/dom.13310

19.

, Lv

, Liang

, et al. Continuous glucose monitoring effects on maternal glycemic control and pregnancy outcomes in patients with gestational diabetes mellitus: A prospective cohort study. J Clin Endocrinol Metab, 2014; 99(12):4674–4682; doi: 10.1210/jc.2013-4332

20.

Polsky

, Akturk

. Case series of a hybrid closed-loop system used in pregnancies in clinical practice. Diabetes Metab Res Rev, 2020; 36(3):e3248; doi: 10.1002/dmrr.3248

21.

American Diabetes

. 13. Management of diabetes in pregnancy: Standards of medical care in diabetes-2018. Diabetes Care, 2018; 41(Suppl 1):S137–S143; doi: 10.2337/dc18-S013

22.

ElSayed

, Aleppo

, Aroda

, et al. 15. Management of diabetes in pregnancy: Standards of care in diabetes-2023. Diabetes Care, 2023; 46(Suppl 1):S254–S266; doi: 10.2337/dc23-S015

23.

Lemaitre

, Faiz

, Baudoux

, et al. Intermittently scanned continuous glucose monitoring is associated with lower spontaneous abortion rate compared with conventional blood glucose monitoring in pregnant women with type 1 diabetes: An observational study. Diab Vasc Dis Res, 2022; 19(6):14791641221136837; doi: 10.1177/14791641221136837

24.

Sibiak

, Gutaj

, Mrzewka-Rogacz

, et al. Novel continuous glucose monitoring metrics and large-for-gestational-age risk: An exploratory retrospective cohort study in pregnancies with type 1 diabetes. Diabetes Technol Ther, 2022; 24(1):42–53; doi: 10.1089/dia.2021.0194

25.

Battelino

, Danne

, Bergenstal

, et al. Clinical targets for continuous glucose monitoring data interpretation: Recommendations from the international consensus on time in range. Diabetes Care, 2019; 42(8):1593–1603; doi: 10.2337/dci19-0028

26.

Mecacci

, Ottanelli

, Vannuccini

, et al. What is the role of glycemic control in the development of preeclampsia among women with type 1 diabetes?. Pregnancy Hypertens, 2021; 25:191–195; doi: 10.1016/j.preghy.2021.06.010

27.

Thilaganathan

, Kalafat

. Cardiovascular system in preeclampsia and beyond. Hypertension, 2019; 73(3):522–531; doi: 10.1161/hypertensionaha.118.11191

28.

Oppermann

, Alessi

, Hirakata

, et al. Preeclampsia in women with pregestational diabetes—A cohort study. Hypertens Pregnancy, 2020; 39(1):48–55; doi: 10.1080/10641955.2019.1704002

29.

Yamamoto

, Donovan

, Mohammad

, et al. Severe neonatal hypoglycaemia and intrapartum glycaemic control in pregnancies complicated by type 1, type 2 and gestational diabetes. Diabet Med, 2020; 37(1):138–146; doi: 10.1111/dme.14137

30.

Capobianco

, Gulotta

, Tupponi

, et al. Fetal growth and neonatal outcomes in pregestational diabetes mellitus in a population with a high prevalence of diabetes. J Pers Med, 2022; 12(8):1320; doi: 10.3390/jpm12081320

31.

Newman

, Egan

, Ahern

, et al. Retrospective national cohort study of pregnancy outcomes for women with type 1 and type 2 diabetes mellitus in Republic of Ireland. Diabetes Res Clin Pract, 2022; 189:109947; doi: 10.1016/j.diabres.2022.109947

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB