Fattori che influenzano il controllo glicemico nelle gravidanze singole complicate da diabete mellito gestazionale nella Cina occidentale: uno studio retrospettivo

1. Introduction

Gestational diabetes mellitus (GDM) refers to diabetes that manifests during pregnancy in individuals with normal glucose metabolism prior to conception. [ , ] Reported rates of GDM can be as high as 25%, varying across populations and diagnostic criteria, with overall rates increasing globally. [ ] The emergence of GDM poses potential impacts on short-term and long-term outcomes for both the mother and the offspring. [ ] Pregnant individuals with GDM face an increased risk of obstetric complications, including gestational hypertension, preeclampsia, polyhydramnios, shoulder dystocia, birth canal injuries, and postpartum hemorrhage. [ ] Moreover, they are at an elevated lifetime risk of developing type 2 diabetes mellitus, metabolic syndrome, and cardiovascular diseases. [ – ] Offspring of GDM-affected individuals are also at a heightened risk for complications, such as macrosomia, fetal growth restriction, neonatal hypoglycemia, and neonatal polycythemia. [ , ] Additionally, there is an increased incidence of long-term risks during the life time of offspring, including abnormal glucose tolerance, obesity, and metabolic disorders. [ , ]

Investigating whether variations in glycemic control among those with GDM impact pregnancy outcomes and understanding how this influence unfolds are crucial. Glycemic treatment target recommendations for women with GDM vary widely internationally in the current, often relying on consensus rather than high-quality trials. [ ] In the latest update review in 2023, the effects of different intensities of glycemic control in pregnant women with GDM on maternal and infant health outcomes were assessed, including 4 randomized controlled trials involving 1731 women that took place in Canada, [ ] New Zealand, [ ] Russia, [ ] and the USA. [ ] Tighter glycemic control may lead to a potential increase in hypertensive disorders of pregnancy, but it does not seem to significantly affect cesarean section rates or induction of labor rates. [ ]

However, the applicability of current results to China remains unknown. In a more nuanced exploration, there are topics lacking thorough investigation in the current, including understanding influencing factors and potential differences of glycemic control based on different classifications, and determining which classification is more advantageous for early prediction,

This study aims to explore the factors influencing glycemic control in patients with GDM and understand the intricate relationship between glycemic control levels and their consequences, to pave the way for more informed clinical interventions and preventive measures.

2. Method

### 2.1. Ethics approval and consent to participate

The study was in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the West China Second Hospital of Sichuan University (Approval No. 2021 [181]). Due to the retrospective nature of the study, the need for informed consent was waived by the Ethics Committee of the West China Second Hospital of Sichuan University in accordance with the national legislation.

### 2.2. GDM diagnosis

In accordance with the guidelines outlined in the American Diabetes Association (ADA)’s “Management of Diabetes in Pregnancy: Standards of Medical Care in Diabetes (2021),” [ ] the standard diagnostic method entails a 75 g oral glucose tolerance test (OGTT) administered between 24 to 28 weeks of gestation. GDM diagnosis is established if the OGTT results meet or exceed any of the following criteria: 0-h PG ≥ 5.1 mmol/L, 1-h PG ≥ 10.0 mmol/L, and/or 2-h PG ≥ 8.5 mmol/L. GDM A1 is defined as glucose can be controlled through diet and exercise guidance alone, while GDM A2 is defined as requiring medication to control glucose. [ ] Since metformin pharmacotherapy was not utilized, GDM A2 in this study specifically denoted the use of insulin during pregnancy.

### 2.3. Sample size

This retrospective study aims to develop a multivariate regression model incorporating 5 to 15 features. Adhering to the Event Per Variable criterion, a sample size exceeding 150 is recommended for binary outcome categories. [ ] Considering the current situation in the West China Second Hospital of Sichuan University, the incidence of GDM A1 was approximately 9 times that of GDM A2, and individuals with satisfactory glycemic control were roughly twice as prevalent as those with unsatisfactory glycemic control. Consequently, the total number of participants included in the final analysis should surpass 1500 cases.

### 2.4. Study participants

This study enrolled pregnant women diagnosed with GDM, who underwent regular antenatal examinations, and delivered at the West China Second Hospital of Sichuan University in 2019. Exclusion criteria encompassed prepregnancy diabetes, age below 18 years old, twin or multiple pregnancies, non-Chinese nationality, non-Han ethnicity, severe dysfunction of vital organs, and hospitalization not related to childbirth, such as therapeutic induced labor. Participants with overt diabetes in pregnancy, defined as hyperglycemia first recognized during pregnancy which meet the thresholds of diabetes in nonpregnant adults, were also excluded. [ ]

On the one hand, based on the difficulty of glycemic control during pregnancy in GDM patients, participants were divided into A1 and A2 groups. [ ] On the other hand, referring to ADA’s recommendations, glucose targets for GDM patients were fasting plasma glucose (FPG) < 5.3 mmol/L and either 1-h PG < 7.8 mmol/L or 2-h PG < 6.7 mmol/L. [ ] In fact, venous blood sampling was performed at least twice during the third trimester of pregnancy to measure glucose levels, in addition to using fingertip blood sampling for glucose monitoring. Participants were categorized as part of the satisfactory group if their glucose values at 32 to 37 weeks of gestation exceeded the standards for glycemic control more than once, while those who did not were classified as part of the unsatisfactory group.

### 2.5. Observation indicators

The variables examined in this study are outlined in Table S1, Supplemental Digital Content, http://links.lww.com/MD/N655 .

### 2.6. Statistical analysis

Categorical variables were presented as frequencies (percentages [%]) and assessed through χ 2 tests. Continuous variables were expressed as mean (standard deviation) and analyzed using Student t tests. Logistic regression models were constructed based on univariate analysis, providing odds ratios (OR) and corresponding 95% confidence intervals (CI). A propensity score matching (PSM) was implemented in a 1:1 ratio, and a reevaluation of pregnancy outcomes was performed to further mitigate the impact of confounding factors. All statistical analyses were conducted using SPSS, with a two-tailed P -value < .05 considered as indicative of statistically significant.

3. Results

### 3.1. Participants selection

A total of 15,796 inpatients were enrolled from the obstetric medical unit of West China Second University Hospital between January 2019 and December 2019. Among them, 3047 patients diagnosed with GDM, excluding 237 cases with prepregnancy diabetes. Exclusions were made for 317 cases involving twin pregnancies, 10 cases involving multiple pregnancies, 1 cases of non-Chinese ethnicity, 70 cases of non-Han nationality, 92 cases experiencing severe organ dysfunction, and 131 cases not intended for delivery purpose. Ultimately, 2621 patients were included in the final analysis. The participant selection process is illustrated in Figure S1, Supplemental Digital Content, http://links.lww.com/MD/N655 .

### 3.2. Influencing factors of glycemic control

The A2 group were characterized by advanced age ( P < .001), elevated prepregnancy weight ( P < .001) and prepregnancy body mass index (BMI) ( P < .001), higher gravidity ( P = .001), higher proportions of individuals with a history of GDM ( P = .001), a family history of diabetes ( P = .004), and hypertension ( P = .027). FPG levels in early pregnancy, and 3 glucose values of OGTT during pregnancy were elevated (all P ≤ .001). Moreover, the A2 group displayed a notable increase in weight gain ( P = .009), higher pre-delivery BMI ( P = .007), elevated amniotic fluid depth ( P = .022), and amniotic fluid index ( P = .026).

The unsatisfactory group exhibited higher prepregnancy BMI ( P = .022), higher gravidity ( P = .033), a lower proportion of primiparity ( P = .020), higher white blood cell count ( P = .049), and FPG ( P < .001) in early pregnancy and during OGTT. Also, higher pre-delivery BMI ( P = .006) and lower diastolic blood pressure (DBP) ( P = .028) were displayed in the unsatisfactory group (Tables and ).

### 3.3. The impact of glycemic control on pregnancy outcomes

The A2 group demonstrated prolonged hospitalization days ( P < .001), heightened incidences of cesarean section ( P = .009), placenta previa ( P = .016), and neonatal pneumonia ( P = .012).

The unsatisfactory glycemic control had lower gestational age ( P < .001), lower incidences of cesarean section ( P = .003), placenta previa ( P < .001), higher incidence of postpartum hemorrhage ( P = .002), lower neonatal length ( P = .003) and weight ( P = .002), higher incidences of premature birth ( P < .001), neonatal jaundice ( P < .001), hypoglycemia ( P = .006), pneumonia ( P < .001), neonatal respiratory distress syndrome ( P = .001), anemia ( P = .011), hospitalization ( P < .001), and hospitalization for more than 15 days in both the pediatric unit ( P = .001) and the neonatal intensive care unit ( P = .006) (Table ).

Considering the substantial baseline differences between A1 and A2 groups, PSM was employed for further analysis. The A2 group only showed a lower incidence of ICP ( P = .003). However, owing to minimal baseline differences between the satisfactory and unsatisfactory groups, and the potential substantial reduction in sample size associated with PSM, we did not employ PSM in subsequent classifications. (Tables S2 and S3, Supplemental Digital Content, http://links.lww.com/MD/N655 ).

### 3.4. Multivariate logistic regression of glycemic control (A1 and A2)

In Model 1, adjusting for original characteristics, the results revealed that age (OR = 1.053, 95% CI: 1.019–1.087, P = .002), prepregnancy BMI (1.087, 1.042–1.133, P < .001), and a history of GDM (2.194, 1.287–3.741, P = .004) were independent factors associated with the occurrence of GDM A2.

Building upon Model 1, Model 2 incorporated adjustments for laboratory indicators in early pregnancy. The findings demonstrated that age (1.059, 1.021–1.098, P = .002), prepregnancy BMI (1.067, 1.014–1.122, P = .013), history of GDM (1.992, 1.116–3.556, P = .020), and FPG (2.031, 1.456–2.832, P < .001) in early pregnancy were identified as independent factors.

In Model 3, adjustments were made for original characteristics, FPG in early pregnancy, and glucose values during OGTT. The results indicated that prepregnancy BMI (1.070, 1.019–1.122, P = .006), history of GDM (1.888, 1.052–3.389, P = .033), FPG (1.828, 1.320–2.532, P < .001), 1-h PG (1.126, 1.009–1.256, P = .034), and 2-h PG (1.181, 1.046–1.333, P = .007) were independent factors (Table ).

### 3.5. Multivariate logistic regression of glycemic control (satisfactory and unsatisfactory)

In Model 1, with adjustments for original characteristics and laboratory indicators in early pregnancy, the results indicated that FPG (1.349, 1.107–1.644, P = .003) and white blood cell count (1.049, 1.002–1.098, P = .040) were independent factors associated with unsatisfactory glycemic control.

Expanding upon Model 1, Model 2 included adjustments for OGTT results. The findings revealed that only FPG during OGTT emerged as an independent factor (1.546, 1.235–1.935, P < .001).

In Model 3, which accounted for original characteristics, FPG in early pregnancy, and OGTT results, the analysis demonstrated that only FPG during OGTT remained an independent factor (1.590, 1.273–1.985, P < .001) (Table ).

4. Discussion

Based on difficulty of glycemic control, as determined by the utilization of insulin was used during pregnancy [ ] and adhering to the glycemic control targets outlined by ADA, [ ] this study including 2621 women performed a parallel classification analysis. Elevated prepregnancy BMI, FPG in early pregnancy, 1-h and 2-h PG during OGTT, and with a history with GDM are independent factors influencing insulin utilization, while elevated 0-h PG is an independent influencing factor of unsatisfactory glycemic control. Compared with the A1 group, the A2 group has longer hospitalization, higher rates of cesarean section, placenta previa, and neonatal pneumonia ( P < .05). Compared with the satisfactory group, the unsatisfactory group has lower gestational age, lower rates of cesarean section and placenta previa, and higher rates of postpartum hemorrhage for mothers; lower length and weight, and higher rates of premature birth, jaundice, hypoglycemia, pneumonia, respiratory distress syndrome, anemia, hospitalization, and hospitalization for more than 15 days in both pediatric unit and neonatal intensive care unit for newborns.

The selection of indicators plays a crucial role in determining the generalizability of research findings. In a study conducted by Daniel PJ et al, the relationship between cardiovascular biomarkers and glucose regulation was investigated, with assessment based on third trimester HbA1c levels. [ ] However, these indicators were not routinely examined in western China. Our study opted baseline indicators, a choice more conducive to generalization within primary health care systems with limited costs. We found that prepregnancy BMI, history of GDM, FPG in early pregnancy, and 1-h PG and 2-h PG during OGTT was closely related to insulin use during pregnancy. Notably, only FPG during OGTT emerged as an independent influencing factor contributing to failure in achieving glycemic targets.

A correlation between glycemic control and pregnancy outcomes has been demonstrated to some extent. [ , ] Early identification of unsatisfactory glycemic control allows for proactive regulation of glycemic levels, which is crucial for improving maternal and neonatal outcomes. However, large-scale studies are still warranted to comprehensively investigate this association. This study offers significant strengths, including a large sample size, dual classification, detailed outcome presentation, and adjustment for multiple confounding factors. Our findings suggested that under the classification 1, pregnancy outcomes did not exhibit significant differences, which was similar to the study of Koren et al. [ ] There were minor differences in general clinical characteristics between the 2 groups under the classification 2; therefore, PSM was not performed in this classification.

It is worth noting that diverse glycemic targets can influence study outcomes. Hofer OJ, et al proposed stringent tight glycemic targets ranging between ≤5.0 and 5.1 mmol/L for FPG and ≤6.7 and 7.4 mmol/L postprandial in the included trials. Less-tight targets for glycemic control ranged between <5.3 and 5.8 mmol/L for FPG and <7.8 and 8.0 mmol/L postprandial. [ ] The evidence suggested a possible increase in hypertensive disorders of pregnancy with tighter glycemic control. [ ] In contrast, our study specifically focused on insulin utilization and ruled that patients with GDM maintained glucose levels at FPG ≤5.3 mmol/L, and 2-h postprandial PG ≤ 6.7 mmol/L, and showed no statistical difference in preeclampsia and eclampsia between groups.

Yefet et al proposed that well-controlled glucose was correlated with a long-term reduction in maternal cardiovascular risk. [ ] In our study, although DBP revealed no significant difference in classification 2, the unsatisfactory group demonstrated lower DBP ( P = .028). The question of whether greater pulse pressure differences are associated with underlying cardiovascular disease remains a topic worthy of long-term follow-up. As for outcomes for offspring, González-Quintero et al highlighted that over one-third of infants in the poorly glycemic controlled group tested positive for the composite variable, comprising macrosomia, large for gestational age, hypoglycemia, jaundice, and stillbirth, which was similar to the results of our study. [ ]

The description of pregnancy outcomes in this study concluded with the completion of delivery, but further exploration is warranted in long-term follow-ups. This should encompass postpartum metabolic changes in mothers and the growth and development of newborns, taking into account social contributors and circadian rhythm. [ , ] Moreover, the study’s handling of outcome variables separately could benefit from considering adverse pregnancy outcomes as a composite variable, defining it as the occurrence of at least one adverse pregnancy outcome.

Acknowledgments

The authors express gratitude to all research staff involved in data collection and analysis, as well as the participants who actively participated in the study. All individuals who made significant contributions to this research are acknowledged as authors and meet the criteria for authorship.

Author contributions

Conceptualization: Jiani Zhang, Chihui Mao, Qi Cao, Xiaodong Wang.

Data curation: Jiani Zhang, Chihui Mao, Guiqiong Huang.

Formal analysis: Jiani Zhang, Chihui Mao, Qi Cao.

Funding acquisition: Xiaodong Wang.

Methodology: Jiani Zhang, Qi Cao, Guiqiong Huang.

Supervision: Xiaodong Wang.

Writing – original draft: Jiani Zhang.

Writing – review & editing: Xiaodong Wang.

Supplementary Material