Maternal and fetal outcomes in a retrospective cohort of obstetrical patients that used an insulin pump paired with a continuous glucose monitor

Abstract Objective To assess maternal and neonatal outcomes in pregnant pregestational diabetic patients using a continuous subcutaneous insulin infusion (CSII) pump paired with a continuous glucose monitor (CGM). Methods This retrospective cohort study included 55 patients who delivered within one healthcare system from October 2019 to October 2022 with pregestational diabetes managed using CSII pumps paired with CGM. Maternal blood glucose (BG) data were analyzed for the two-week period preceding delivery. The percentage of time spent at a BG level of less than 140 mg/dL was recorded and compared between patients with and without obstetric and neonatal morbidities. Results Patients who delivered with severe preeclampsia (S. PreE) had a significantly lower mean percentage of time spent at BG < 140 mg/dL than those who did not (S. PreE 15/55, 63.1% ± 19.0 vs. 40/55, 73.6% ± 13.8; p = 0.03). Mothers who had a preterm birth (PTB) had a significantly lower mean percentage of time spent at BG < 140 mg/dL than those who delivered at term (PTB 35/55, 66.4% ± 16.4 vs 20/55, 78.3% ± 11.9; p = 0.006). The mean percentage of time spent at a BG < 140 mg/dL among mothers of neonates with respiratory distress syndrome (RDS) was significantly lower than those without RDS (RDS present 13/55, 59.7% ± 20.4 vs 42/55, 74.1% ± 12.7; p = 0.003). There was a significant correlation between a greater neonatal birth weight percentile and worse time spent at BG < 140 mg/dL (r = − 0.31; p = 0.02). No other significant differences were observed between the groups. Conclusion Improved blood glucose levels in pregestational diabetic patients using a CSII pump and CGM is associated with reduced maternal and neonatal morbidity as well as lower birth weight percentile neonates. Future studies are needed to clarify how much time each day a patient needs to spend below a given blood sugar, how long this blood glucose should be maintained, and what specific blood glucose target should be selected.


Introduction
Diabetes Mellitus is a global health problem [1] and common pregnancy complication [2].Pregnancy affected by diabetes is associated with a significant risk of maternal and fetal morbidity, as well as fetal mortality [2][3][4][5][6].Such risks may be minimized when blood glucose is maintained at or near physiologically normal values [7,8].Progressive insulin resistance is common during pregnancy [2,9] which makes accomplishing euglycemia a challenge.The traditional approach for managing diabetes in pregnancy, where insulin is required, is to use multiple daily injections (MDI) of subcutaneous insulin [2]; however, patients often find 4-6 injections per day to be a barrier to compliance with care [10].Moreover, increasing doses of subcutaneous insulin result in variable absorption, which impedes healthcare teams attempting to achieve euglycemia [11].
Non-pregnant adult insulin-dependent diabetes care has shifted from subcutaneous shots to continuous subcutaneous insulin infusion (CSII) delivery systems conveniently named insulin pumps [12].Our practice has been using CSII pumps (Minimed 670 G and 770 G produced by Medtronic, T:SLIM produced by Tandem Diabetes Care, and Omipod produced by Insulet) paired with a continuous glucose monitor (CGM, Guardian Connect, and Guardian Sensor 3 produced by Medtronic, Dexcom G6 produced by Dexcom) off-label to manage insulin-requiring diabetes in pregnancy.Our practice manages patients to maximize the amount of time they spend at a blood glucose better than 140 mg/dL.This value, although an arbitrary choice, is in line with the historic obstetrical practice of targeting a 1 h postmeal blood glucose level of 140 [2].The purpose of this study was to report the results from our cohort of patients managed using CSII pumps paired with a CGM.We hypothesized that patients with better blood glucose levels will have fewer Maternal and Fetal morbidities.

Methodology
Institutional IRB approval was granted to conduct this retrospective chart review (Nebraska Methodist Hospital IRB FWA 00003377 study #1486).Pregnant patients who received care at the Nebraska Methodist Health System Women's Hospital Perinatal Center over a 3-year period between October 2019 and October 2022, required insulin to manage pregestational diabetes, and delivered within the health care system were included.Electronic medical records were searched using hospital billing codes and diagnostic codes to identify the cohort patients.Patients were excluded if diabetes management did not include a CSII pump paired with a CGM, or if blood glucose data were not available for review.The choice of starting a CSII pump or CGM and the selection of which brand used were at the direction of each patient's Maternal Fetal Medicine provider and were not standardized.Some patients entered pregnancy using a CSII pump and CGM whereas other patients had technology initiated during pregnancy.A minimum of 2 weeks of technology use were required for a patient to be categorized as using a CSII pump and CGM.All women received routine obstetrical care at the direction of their managing Obstetrician combined with consultation with Maternal Fetal Medicine providers.
Electronic records were reviewed to obtain the following demographic information: maternal age, race, insurer, parity, body mass index (BMI), Hemoglobin A1c (HbA1c) at the onset of pregnancy, insulin regimen, diabetes type, and infant sex.The following outcome variables were selected: mode of delivery, gestational age at delivery, hypertensive disorders, severe preeclampsia, delivery due to poor control, birthweight, admission to the neonatal intensive care unit (NICU), large for gestational age, neonatal respiratory distress syndrome (RDS), neonatal hyperbilirubinemia, and neonatal hypoglycemia.Delivery due to poor control was categorized as present when delivery was recommended based solely on suboptimal blood glucose levels.Two definitions of neonatal hypoglycemia were considered: 1) neonatal blood glucose under 30 mg/dL and 2) neonatal blood glucose under 45 mg/dL; otherwise, standard neonatal and obstetrical definitions were used.
Maternal blood glucose information was obtained by reviewing the cloud-based patient data for each CGM brand.Two weeks of data were arbitrarily selected for convenience as each CGM brand's online information display provides data in 2-week blocks of time by default.In each CGM report, the time spent at blood glucose better than 140 mg/dL is reported and expressed as a percentage.If the patient had a different blood glucose target selected the online settings were reset to 140 mg/dL prior to generating a patient's report.
The Kolmogorov-Smirnov test was used to test for normal distribution.Since the data were normally distributed, the t-test and Pearson's correlation coefficient were used to compare interval variables.SPSS version 29.0 was used for data analysis.Statistical significance was set at a p-value <0.05.

Results
A total of 110 patients were identified.Fifty-five patients were excluded, leaving 55 patients in the study cohort (Figure 1).Demographic, Obstetrical, and Neonatal outcome information is presented in Table 1.The mean percentage of time spent at blood glucose (BG) less than 140 mg/dL was not significantly different between obstetrical patients with type 1 diabetes (DM) and those with type 2. (type 1 DM 26/55, 67.1% ± 12.4 vs type 2 DM 29/55, 74.0% ± 18.1; p ¼ NS) For all subsequent comparisons, patients with type 1 diabetes and type 2 diabetes were combined into a single group.

Women that delivered preterm
The mean percentage of time spent at a BG less than 140 mg/dL was significantly lower in patients delivered prior to 37 weeks (PTB) than in those delivered in the term period (PTB present 35/55, 66.4% ± 16.4 vs 20/55, 78.3% ± 11.9; p ¼ 0.006; Table 2, Figure 2).

Neonatal birth weight
There was a significant correlation between greater neonatal birth weight percentile and lower mean percentage of time spent at a BG less than 140 mg/dL (r¼ À0.31; p ¼ 0.02; Figure 3).No significant differences in the mean percentage of time spent at a BG less than 140 mg/dL were noted between patents with and without hypertensive disorder of pregnancy, delivery due to poor control, cesarean delivery, large for gestational age, admission to the NICU, neonatal hyperbilirubinemia, neonatal BG < 45 mg/dL, and neonatal BG < 30 mg/dL (Table 2).

Discussion
Principal findings of this study 1.Patients delivered with severe preeclampsia had blood glucose levels that were significantly worse than patients who did not have severe preeclampsia  2. Patients that delivered preterm had blood glucose levels that were significantly worse than patients delivered at term. 3. RDS occurred in newborns of patients that had blood glucose levels which were significantly worse than patients whose newborns did not have RDS.4.There is a significant correlation between birth weight percentile and time spent at a BG better than 140 mg/dL 5. Time spent at a BG better than 140 mg/dL did not differ in the 2 weeks before delivery between patients delivered due to poor control compared to patients who were not delivered due to poor glycemic control The principal findings of this study agree with previous work demonstrating that pregnant patients with diabetes have a higher risk of preeclampsia and preterm birth [13,14]; furthermore, neonates of diabetic  patients experience respiratory distress syndrome more often [15].This work is novel because it demonstrates that, in patients using a CSII pump paired with a CGM, blood sugar levels are significantly better in those patients who did not experience preeclampsia, preterm birth, and RDS than in those who did experience these outcomes.Additionally, these results are consistent with the finding of a significant correlation between blood glucose levels and birth weight; specifically, better blood sugar levels are correlated with a reduced neonatal birth weight percentile.
Pregestational diabetes is also a risk factor for large-for-gestational-age neonates, neonatal hyperbilirubinemia, neonatal hypoglycemia, and admission to the NICU [2].For neonatal hypoglycemia specifically, two BG thresholds, < 30 mg/dL and < 45 mg/dL, were selected, as there is ongoing controversy about how best to define neonatal hypoglycemia [16].We hypothesized that improved blood glucose levels preceding delivery are associated with fewer neonatal morbidities.Interestingly, there was no significant difference in the percentage of time spent at a BG better than 140 mg/dL for the above neonatal morbidities.Encouragingly, the correlation between lower birth weight percentile and better blood sugar levels suggests that lower blood glucose targets maintained for longer periods may be needed before improvement in these parameters is noted; however, the specifics of how low a blood glucose target needs to be and for how long remains unclear.
This study also found that there was no significant difference in the time spent at a BG better than 140 mg/dL in the 2 weeks preceding delivery between patients who were delivered for poor control and those who were not.In the current clinical setting, multiple providers see and treat pregnant patients with diabetes.The general practice pattern is to maximize the time patients spend at a blood glucose level better than 140 mg/dL.Although an arbitrary choice, this cutoff is consistent with other reports in the literature [17][18][19][20][21][22][23][24] and is in line with the historical practice of targeting a 1 h post-meal blood glucose level of < 140 mg/dL in an obstetrical patient [2].There is currently no clinical agreement on how physicians interpret diabetic blood glucose levels using CGM data in pregnancy.Consequently, the decision to recommend delivery of a patient with blood glucose levels that are not at goal varies between providers and may explain the nonsignificant difference noted in blood glucose levels in this group of diabetic patients.
The goal of this work was to focus on the blood glucose achieved during the last 2 weeks of pregnancy rather than the type of CSII pump or CGM used.This approach differs from previous publications that focused on comparing different types of diabetes treatments including: 1) MDI vs CSII [25], 2) type of CGM -real-time CGM vs intermittently viewed CGM [23], or 3) type of insulin pump -closed loop pump vs sensor augmented pump [18].
In the current clinical setting intermittently viewed CGM data were not archived for review, and these patients were excluded from participation in the study because their data were not available.This resulted in a study population that only used real-time CGM devices.Additionally, insulin pump modes of operation were not distinguished in this work.In the current clinical setting, the pump mode that produced the best blood glucose level was selected.These choices are at the discretion of the managing physician and were not standardized.
This study excluded pregnant diabetic patients treated with MDI insulin and limited comparisons to pregnant patients using CSII pumps paired with a CGM because these two groups are dissimilar in several ways.First, the type of data is different.CGM users receive a glucose reading every 5 min, which is uploaded to the cloud daily or upon patient demand, thereby creating an abundance of data.Patients using MDI insulin received instructions to follow blood glucose with point-of-care finger sticks four times a day, which resulted in a paucity of data in the best-case scenario.Second, MDI patients with blood glucose levels at goal are unlikely to be switched to a CSII pump; however, a diabetic with blood glucose levels not at goal on MDI insulin may be placed on a CSII pump.The result is that blood glucose levels are different between these groups, and it follows that MDI patient groups could have better obstetrical outcomes than CSII pump patient groups.Indeed, higher rates of large for gestational-age infants have been noted in CSII pump patients [25].Third, patients using MDI insulin receive instructions to follow blood glucose levels with point-of-care finger sticks four times a day or more, which creates significant patient compliance issues.Patients commonly struggle to complete multiple finger sticks daily; additionally, the burden is placed on the patient to bring their diabetes data to clinical visits.Both these issues lead to missing data.Furthermore, healthcare providers faced the dilemma that patient data may not be reliable.Most, if not all, diabetes providers have encountered a list of blood sugars written in the same blue ink, all odd numbers, that are universally normal and discordant from the point-of-care office reading that is abnormally high.
For these reasons, this study focused on patient groups that were as similar as possible; namely, pregnant patients who required insulin prior to pregnancy and used a CSII pump paired with a CGM.
Additional strengths of this work include the sample size reported, which is the largest US cohort to date of pregnant diabetic patients using a CSII pump paired with a CGM.Furthermore, individual outcomes rather than composite outcomes are reported herein, potentially allowing for more direct clinical application.The limitations of this study include its non-standardized clinical care, a retrospective design that prohibits statements of causality, and sample size which limit the use of multivariate tools to assess for confounding.

Conclusion and future direction
When antenatal and neonatal complications were absent, blood glucose levels were better in this retrospective cohort of pregnant patients with diabetes treated with a CSII pump paired with CGM.Several questions remain to be answered regarding blood glucose targets as they pertain to improvement in pregnancy outcomes: 1) how much time each day does a patient need to spend below a given blood sugar, 2) how long should this blood sugar target be maintained, and 3) what specific blood glucose target should be selected?Future planned work will aim to address these questions.

Figure 2 .
Figure 2. Boxplot of percent of time spent at a BG < 140 mg/dL in those with and without severe preeclampsia, preterm birth, and neonatal RDS.

Table 1 .
Participant demographics and clinical outcomes for the cohort.

Table 2 .
CGM Data for the 2 weeks preceding delivery.
Maternal and Fetal outcome (number of patients with outcome/total cohort).Data are expressed as mean percent ± 1 SD.p < 0.05 considered significant.CGM: continuous glucose monitoring system; BG: blood glucose; NS: not significant; NICU: neonatal intensive care unit; RDS: respiratory distress syndrome.