Risk factors for primary cesarean delivery in women with gestational diabetes mellitus: a predictive model for clinical risk assessment

Introduction

Gestational diabetes mellitus (GDM) is a transient glucose intolerance that typically occurs in the second or third trimester of pregnancy, particularly between 24 and 28 weeks of gestation [1], 2]. The pathophysiology involves increased insulin resistance due to the placental secretion of various anti-insulin hormones – such as human placental lactogen, cortisol, and estrogen – coupled with an inadequate compensatory insulin response [3], [4], [5]. This imbalance ultimately leads to hyperglycemia and the onset of GDM. Studies have shown that women with a history of GDM face an approximately 50 % risk of developing type 2 diabetes within 10 years postpartum [6], 7]. Although GDM is temporary, its long-term health implications necessitate early clinical intervention and continuous postpartum follow-up.

Clinical and epidemiological studies have identified GDM as a significant independent risk factor for cesarean section [8]. In a retrospective cohort study, the cesarean section rate was 54.5 % in the GDM group and 41.5 % in the control group, with a statistically significant difference between the two groups (p=0.002) [9]. Similarly, American College of Obstetricians and Gynecologists (ACOG) data suggest that GDM independently predicts higher cesarean delivery rates [10]. Research published in The Lancet Diabetes & Endocrinology further demonstrated that GDM remains associated with increased cesarean rates even after adjusting for fetal weight, suggesting the involvement of additional factors such as metabolic abnormalities or dysfunctional uterine contractility [11].

A constellation of clinical factors influences the decision for cesarean delivery in GDM pregnancies. Among these, maternal body mass index (BMI) is strongly associated with increased cesarean risk, with overweight and obese women being more prone to complications such as macrosomia or cephalopelvic disproportion [12]. Other significant contributors include advanced maternal age, nulliparity, previous cesarean section, and a history of multiple miscarriages [12], 13]. Furthermore, poor glycemic control, gestational hypertension, polyhydramnios, abnormal fetal presentation, and estimated fetal macrosomia are also recognized as relevant clinical risk factors [14], 15]. Regarding laboratory findings, hematological and coagulation parameters – such as HGB, PLT, and D-dimer levels – have been explored as auxiliary predictive markers. In addition to physiological and clinical variables, psychosocial factors, including maternal preference, physician recommendation, and access to obstetric care, may also influence the mode of delivery.

Given the multifactorial nature of cesarean section decisions in GDM, constructing a comprehensive predictive model incorporating demographic, clinical, and laboratory data may aid in the early identification of high-risk patients and improve perinatal management strategies [16], 17].

This study aims to identify the determinants associated with cesarean section among women with GDM and to establish a predictive model for clinical use. By retrospectively analyzing baseline data – including maternal age, BMI, parity, abortion history, and family history of diabetes – alongside third-trimester laboratory parameters such as HbA_1c, HGB, white blood cell (WBC) count (WBC), PLT, prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen (FIB), and D-dimer, we seek to evaluate the relationship between these variables and cesarean delivery. Our findings may contribute to the early detection and individualized management of high-risk GDM pregnancies and offer valuable insights for clinical application.

Materials and methods

Results

Baseline characteristics stratified by BMI

Based on BMI classification, 154 patients (6.77 %) were underweight (BMI<18.5), with a mean age of 29.78±3.84 years; 1,249 patients (54.88 %) were in the normal weight group (BMI 18.5–24.0), mean age 30.76±3.74 years; 522 patients (22.93 %) were overweight (BMI 24.0–28.0), mean age 31.44±3.87 years; and 351 patients (15.42 %) were obese (BMI>28.0), mean age 30.62±4.40 years. The differences in age across the four groups were statistically significant (p<0.05), with details shown in Table 1 and Figure 1.

Figure 1:

Flow diagram of patient numbers by BMI category.

Analysis of BMI change (delivery BMI minus pre-pregnancy BMI) also revealed a statistically significant difference among groups (p<0.001). The proportion of cesarean section increased with rising BMI, showing a significant upward trend across the four groups (p<0.05): 3 cases (1.95 %) in the underweight group, 72 (5.76 %) in the normal group, 42 (8.69 %) in the overweight group, and 90 (25.64 %) in the obese group.

In addition, statistically significant differences were found among BMI groups in adjusted BMI values (pre-vs. post-pregnancy), number of pregnancies, number of abortions, and family history of diabetes (all p<0.05).

Regarding laboratory findings, significant differences were observed across BMI groups in HbA_1c (%), HGB (g/dL), and PLT ( × 10⁹/L) (all p<0.05). However, no significant difference was found in the WBC ( × 10⁹/L) count (p=0.664). Furthermore, no significant differences were observed among the four groups in coagulation indicators (all p>0.05), as detailed in Table 1.

Diagnostic performance of the predictive models

A predictive model (Model 1) was constructed using the six independent variables associated with cesarean section. The area under the ROC curve (AUC) of Model one was 0.792, with a sensitivity of 81.77 % and specificity of 71.00 %. However, the diagnostic performance was slightly lower than expected.

Clinical considerations were incorporated to improve the model, and a new model (Model 2) was established by including BMI category, abortions (times), and HbA_1c levels. Model two achieved an improved AUC of 0.842, with a sensitivity of 85.50 % and specificity of 76.72 %. Although some individual indicators (e.g., HGB, with a specificity of 91.91 %) showed relatively high diagnostic value, none outperformed Model two in overall predictive ability (Table 4 and Figure 2).

Table 4:

Detailed information on the ROC curve.

Parameters (systole)	AUC	SE	95 % CI	Sensitivity, %	Specificity, %	Youden	AC	Z	p-Value
BMI, kg/m2	0.689	0.022	0.669 – 0.708	43.00	88.13	0.311	>27.68	8.665	<0.001
Adjusted BMI	0.544	0.023	0.523 – 0.565	44.00	70.61	0.146	>5.87	1.876	0.061
Abortion (times)	0.579	0.020	0.558 – 0.600	63.50	56.08	0.196	–	3.901	0.001
PLT ( × 109/L)	0.617	0.022	0.595 – 0.639	37.02	80.21	0.172	>237	5.428	<0.001
HbA1c, %	0.799	0.019	0.782 – 0.816	70.50	79.21	0.497	>5.6	16.149	<0.001
HGB, g/L	0.616	0.024	0.594 – 0.638	26.52	91.91	0.184	≤111	4.895	<0.001
PT	0.646	0.026	0.622 – 0.670	47.85	78.21	0.261	≤11.1	5.722	<0.001
APTT	0.604	0.024	0.579 – 0.628	54.60	65.44	0.200	≤30.1	4.287	<0.001
Model 1	0.813	0.017	0.795 – 0.830	81.77	71.00	0.528	>0.078	18.604	<0.001
Model 2	0.842	0.014	0.826 – 0.857	85.50	76.72	0.622	>0.082	24.845	<0.001

1. AUC, area under curve; SE, standard error; AC, associated criterion; PLT, platelet. The bold “P value” indicates a statistically significant difference between the groups.

Figure 2:

Area under the curve (AUC) of the detection indicator. Model 1 [BMI + Abortion (times) + HbA_1c + Adjusted BMI + HGB + PLT + PT + d-dug/ml]; model 2 [BMI + Abortion (times) + HbA_1c]. BMI, body mass index; HbA_1c, hemoglobin A1c; PLT, platelet; PT, prothrombin time.

Discussion

This study found that as BMI increased, the proportion of GDM patients delivered by cesarean section also increased. This suggests that higher BMI may be an independent factor influencing the decision for cesarean delivery in GDM patients. This finding is consistent with the research by Ehrenberg et al., who demonstrated that both obesity and diabetes independently increase the risk of cesarean section [18]. Furthermore, rapid weight gain during pregnancy is also considered a predictor of cesarean section [19]. Therefore, in GDM patients, controlling pre-pregnancy and gestational weight gain may help reduce the incidence of cesarean sections. However, it should be noted that high BMI may indirectly increase the risk of cesarean delivery through mechanisms such as affecting fetal weight and the progress of labor, and the specific mechanisms require further research [20], [21], [22].

In the blood routine tests of this study, significant differences in HbA_1c, HGB, and PLT were observed across different BMI groups (all p<0.05), suggesting that these indicators may be involved in the metabolic and hemodynamic changes associated with high BMI. However, the direct relationship between these indicators and cesarean section remains unclear. Although some studies have shown that elevated HbA_1c and D-dimer levels are associated with fetal growth restriction (FGR) in GDM patients, their direct relationship with cesarean section still needs further exploration. A study by Ying Zhang et al. [23]explored the relationship between HbA_1c and D-dimer (D-D) levels and FGR in GDM patients. This retrospective analysis of clinical data from 164 GDM patients found that 63 had FGR and 101 had normal fetal weight. Their research found that HbA_1c and D-D levels were significantly higher in the FGR group than in the normal group, suggesting that these two indicators are closely related to the occurrence of FGR. Furthermore, HbA_1c levels above 5.80 % and D-D levels above 3.04 mg/L were considered effective thresholds for predicting FGR. This study showed that HbA_1c levels>5.6 % were a turning point in influencing the choice of cesarean section, which is consistent with the findings of the aforementioned study.

GDM patients are often accompanied by elevated levels of inflammatory factors such as CRP, IL-6, and TNF-α, indicating a systemic inflammatory state [24]. White blood cell count (WBC), an important indicator reflecting immune and inflammatory status, tends to increase physiologically during pregnancy. However, in GDM patients, elevated WBC is more likely to reflect a potential chronic low-grade inflammatory state, which is considered closely related to the development of GDM and pregnancy complications [25], [26], [27]. In recent years, researchers have begun to focus on whether WBC levels can also serve as a potential biomarker for predicting the risk of cesarean section in GDM patients. However, research on the direct relationship between WBC and cesarean section in GDM patients is inconsistent [28]. Some studies did not observe significant statistical differences between WBC and delivery method, which may be related to sample size, dynamic changes in WBC during delivery, and GDM subtypes (e.g., whether insulin therapy is involved) [29]. Additionally, the metabolic characteristics of GDM (such as hyperglycemia and hyperinsulinemia) may induce systemic inflammation, but whether WBC levels fully reflect this inflammatory state remains controversial. In this study, no significant differences in WBC count or coagulation indicators were found across BMI groups, suggesting that inflammation may not be a major factor influencing cesarean section in the study population.

Regarding coagulation, the high insulin state in GDM patients may stimulate fibrinogen (FIB) synthesis and activate platelet aggregation, leading to a hypercoagulable state. Previous studies have shown that D-dimer levels are significantly higher in GDM patients than those with normal pregnancies, indicating endothelial dysfunction and hypercoagulability [30], 31]. Elevated D-dimer levels are not only associated with thrombotic events during pregnancy but are also considered an early sign of placental dysfunction and placental abruption, which may indirectly contribute to the indication for cesarean section. In this study, although the coagulation factors PT and D-Dimer were identified as independent factors influencing cesarean section, the diagnostic performance of these two indicators was limited (both AUC<0.70).

Through regression model analysis, this study identified BMI, adjusted BMI, HGB, PLT, PT, and D-dimer as independent predictors of cesarean section in GDM patients. These factors span multiple physiological dimensions, including metabolism, circulation, and coagulation, indicating that the decision for cesarean section in GDM patients is multifactorial. Although no collinearity was found among these factors in the model (all VIF values<5), there may still be interactions between them. For example, high BMI may indirectly increase the risk of cesarean section through its effect on hematological indicators, and the specific mechanisms require further study. While the predictive model constructed in this study performed well in sensitivity and specificity, its feasibility and practicality in clinical application need to be considered. This study highlights the potential value of BMI and related hematological indicators in predicting the risk of cesarean section in GDM patients, providing data support for personalized delivery planning.

However, it should be noted that as a retrospective study, this research has certain limitations, such as selection bias in sample size and data completeness, which may lead to bias in the results and conclusions. The data were collected from a single-center source, which could introduce a certain degree of bias. Therefore, multicenter collaboration would enhance the reliability of the model. Future research should adopt a prospective design and consider more potential confounding factors to validate the findings of this study. Additionally, the effects of interventions such as weight management and blood glucose control in reducing the risk of cesarean section in GDM patients should be explored.