A retrospective analysis of the incidence and risk factors for amniotic fluid embolism in cesarean deliveries

Abstract

Background

Amniotic-fluid embolism (AFE) is a rare occurrence but life-threatening disease that can occur in process of delivery and postnatal women. The topic of cesarean section surgery and its associated risk factors has been neglected due to the infrequent occurrence of AFE. However, AFE significantly contribute to maternal morbidity and mortality.

Purposes

We aim to examine the correlation between clinical issues, prenatal complications, comorbidities, medical perioperative complications, and cesarean section surgery of AFE in the US, utilizing a Nationwide Inpatient Sample (NIS) database.

Methods

This study conducted a retrospective cohort analysis on cesarean deliveries in the United States using data from the NIS of the Healthcare Cost and Utilization Project (HCUP) to examine the incidence and risk factors of AFE during cesarean section surgery. The analysis included the period from January 1, 2010, to December 31, 2019. We employed a multivariable logistic regression to evaluate the main outcome, which encompassed the clinical, prenatal, comorbidity, and medical perioperative AFE undergoing cesarean deliveries.

Results

We identified AFE in 269 out of the 2,462,005 women whose cesarean deliveries we investigated, with an incidence rate of 0.0113%. In the AFE group, the median patient age at the cohort level was 32 years (IQR, 27–36 years). The in-hospital mortality rate for patients with AFE following cesarean delivery was significantly higher than for those without AFE (14.9% vs. 0.0%, P < 0.001). In univariable analysis, P < 0.05 served as the initial selection criterion. A multivariable analysis revealed that AFE at the time of cesarean deliveries was significantly correlated with chronic blood loss anemia, coagulopathy, congestive heart failure, other neurological disorders, fluid and electrolyte disorders, weight loss, pulmonary circulation disorders, abruptio placentae, and polyhydramnios.

Conclusion

This contemporaneous, nationwide investigation verified the incidence of cesarean deliveries by AFE and corroborated previously identified risk factors for AFE. Although the absolute risk of AFE is minimal, clinicians should be aware of the identified risk factors, such as chronic blood loss anemia, coagulopathy, and polyhydramnios, to enhance preparedness and optimize patient counseling, particularly in high-risk cases.

Introduction

In the United States, maternal mortality has more than doubled over the last three decades. According to the most recent national figures, 17.3 maternal deaths per 100,000 live births occurred in 2017 [1]. Even though amniotic fluid embolisms (AFE) only occur in about 7 to 8 births per 100,000, they are a major cause of maternal morbidity and mortality. AFE frequently manifests as abrupt cardiovascular collapse, respiratory distress, and coagulopathy [1,2,3,4,5]. The mechanisms and risk factors of AFE are both unclear.

When AFE was initially discovered about a century ago, it was believed that the cause was fetal debris obstructing the mother’s pulmonary circulation [5,6,7]. After fetal antigens enter the mother’s bloodstream, AFE appears to include an aberrant activation of humoral and immunologic pathways that causes a severe inflammatory response [8, 9]. The researchers discovered a high correlation between placental pathology (the structural, histological, and molecular changes in the placenta), specifically placental accreta spectrum (PAS), and AFE after delivery [1]. With the progression of pregnancy, a reduction in uteroplacental perfusion occurs, resulting in placental ischemic conditions. Concurrently, there is potential for pathological disruption of the foetal-maternal vascular network and the induction of a systemic anti-angiogenic state [10]. Maternal-fetal circulatory dysfunction may be aggravated by metabolic and microvascular disturbances, alongside possible sclerotic changes and structural modifications in chorionic villi, uterine vasculature, and placental tissue associated with diverse manifestations of maternal diabetes [11]. While traditionally linked to the process of childbirth, it has also been observed in cases of abortion, maternal trauma, amniocentesis, cervical lacerations, and manual extraction of the placenta [1, 4, 12, 13]. The identified risk factors for AFE is cesarean section in Werner and Stefan’s study [14]. It is currently unclear whether decreasing the frequency of cesarean section surgery would mitigate the risk of AFE.

The AFE is rare during cesarean section surgery, it remains a leading cause of maternal mortality, and there is a paucity of large-scale, population-based studies examining its incidence and risk factors in cesarean deliveries. The objective of this study was to utilize the the Nationwide Inpatient Sample (NIS) database [15], a comprehensive and nationally representative database, our study aims to provide robust epidemiological data on AFE during cesarean section surgery, identify modifiable and non-modifiable risk factors, and contribute to the development of targeted prevention and management strategies. We conducted an analysis on several interrelated factors, including patient demographics (age and race), hospital characteristics (admission type and payer, bed size, teaching status, location, and region), length of stay (LOS), total charges incurred during hospitalization, in-hospital mortality, comorbidities, and perioperative complications.

Methods

Data

This study utilized the NIS, a dataset provided by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project. The study design was retrospective and cohort-based [15]. The data is weighted using survey weights derived from a random sample of 20% of hospitalized patients annually. We have stripped all identifying information from the dataset, making it accessible to the public. As a result, the Hospital of DongGuan Tungwah Institutional Review Board determined that this study is exempt from review. This study followed the reporting criteria outlined in the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [16].

Study individuals

Women who underwent cesarean delivery between January 1, 2010, and December 31, 2019, were included in this study (Table S2). We chose the study period to align with the deployment of the International Statistical Classification of Diseases and Related Health Problems, 9th Revision (ICD-9) and 10th Revision (ICD-10) codes by the NIS program. ICD-9 is used to code information up until the end of 2015, while ICD-10 is used to code information from the end of 2015 until the end of 2016, in accordance with US coding standards. The identification of cesarean delivery cases was established by utilizing the International Statistical Classification of Diseases, 9th and 10th Revision, Clinical Modification (ICD-9-CM and ICD-10-CM) codes, the Procedure Classification System (ICD-10-PCS and ICD-10-PCS) codes, and the diagnosis-related group codes.

Evaluation criteria

The main criterion used to evaluate the results was the identification of AFE. The patients who experienced AFE were identified using the ICD-9 code 673.1 and the ICD-10 code O88.1 [1]. It is crucial to note that the code does not provide the precise diagnostic criteria for AFE.

Medical data

Covariates were chosen based on previous studies [5, 17,18,19] examining factors associated with AFE. The information collected includes the following categories: (1) patient demographics, (2) hospital characteristics, (3) prenatal complications, (4) comorbidities, (5) medical and surgical perioperative complications.

The patient’s baseline demographic data (Table S1) included age, which was categorized as follows: under 25 years old, aged 25–29, aged 30–34, and 35 years or above [1], year of cesarean delivery (2010–2019), The NIS program categorizes race and ethnicity by utilizing the following criteria: Asian, Pacific Islanders, Hispanic, Black, White, and other ethnicities, including American Indians. The program additionally determines the main anticipated payer as Medicaid, Medicare, private (including health maintenance organizations), self-pay, or other sources [1, 20, 21].

Patients who had Cesarean section (CS: ICD-9codes (6697, 66970, 66971, 74, 740, 741,742 744, 749, 7499) and ICD-10codes (10D00Z0, 10D00Z1, 10D00Z2)) (Table S2) from 2010 to 2019 (n = 2,462,005). Excluding missing data, a total of 58,464 persons were included, including 269 for age, 12 for length of stay, 38317 for total charge, 3454 for type of insurance, 7795 for bed size of hospital, 7973 for type of admission, 644 for died, and 0 for other items. Excluding 26876 of age less than 18 years (Fig. 1). We evaluated race and ethnicity because these factors are associated with pregnancy features and outcomes. Approximately 20% of the records in our analysis lacked data regarding the race and/or ethnicity of the patients. Several states exclude the race variable from their discharge records in order to protect patient anonymity [21]. In order to tackle this problem, we established a new classification called “others” to record the racial information of these individuals.

Fig. 1

Flow and inclusion/exclusion of all patients had cesarean section

The hospital data included the number of available beds categorized as small, medium, or big, the hospital’s geographical location, which could be either urban or rural, nonteaching or teaching, and the hospital’s area in the United States, which could be either the Northeast, Midwest, South, or West (Table S1) [1, 21].

The prenatal problems encompassed several maternal factors such as abruptio placentae (the placenta from the uterine wall before delivery), placenta previa, hypertension of pregnancy, pre-eclampsia or eclampsia, excessive vomiting in pregnancy, early or threatened labor, late pregnancy, multiple gestations, polyhydramnios (an excessive accumulation of amniotic fluid in the amniotic sac during pregnancy), and premature rupture of membranes (Table S1, S2).

The study identified the existence of 29 distinct comorbidities (Table S1) by categorizing them using the Agency for Healthcare Research and Quality Comorbidity Software, version 3.7. This software detects the specified comorbidities by using the ICD-9-CM and ICD-10-CM diagnosis codes.

The medical perioperative complications encompassed cardiac arrhythmia, sepsis, gastrointestinal bleeding, deep vein thrombosis, shock, cardiac arrest, pneumonia, and stroke. The surgical perioperative complications encompassed wound rupture, wound infection, genitourinary system disorders, bleeding, and blood transfusion (Table S1).

The sample size calculation

This study is a retrospective study based on public databases. We did not formally calculate the sample size because we intended to include all eligible patients from our database in order to maximize statistical power. We also took care to ensure that the multivariable model had more than 10 events per variable in order to prevent overfitting [22, 23].

Statistical analysis

IBM Inc. developed the SPSS statistical software, version 25.0, for conducting the statistical analysis. We used the Wilcoxon rank test to identify notable distinctions between the two groups based on continuous data, and the chi-square test for categorical data. We created univariate and multivariate logistic regression models to evaluate the correlation between AFE and medical and surgical perioperative problems. The initial stage of the analysis involved identifying the autonomous attributes linked to AFE during childbirth. The initial selection criteria were based on a significance level of P < 0.05 in univariable analysis. We used a binary logistic regression analysis to conduct multivariable analysis. We utilized a binary logistic regression analysis to ascertain the independent risk factors for AFE. We computed odds ratios (ORs) and 95% confidence intervals (CIs). A p-value less than 0.05 was deemed to be statistically significant. We used the statistical program R, version 3.5.3, to map the forest. Our primary analysis utilized multivariable logistic regression models that adjusted for an extensive set of pre-specified potential confounders, including demographic characteristics, hospital features, prenatal complications, and medical comorbidities. The selection of these variables was based on prior literature [24, 25] and clinical expertise.

Results

Trends of AFE

According to national estimates, the investigation included a comprehensive examination of 2,462,005 cesarean births. We recorded a total of 269 occurrences of AFE during the study period (Fig. 1, Table S8). The incidence rate of AFE was 0.0113% (95% CI: 0.0100–0.0126%), equivalent to 11.3 cases per 100,000 cesarean deliveries (Table 1). The occurrence of AFE declined from 0.015 to 0.008% during 2010 and 2011, representing a decrease from 15 to 8 cases per 100,000 cesarean deliveries (Fig. 2, Table S8). Nevertheless, the occurrence of AFE rose from 0.008 to 0.016% between 2014 and 2019, representing an increase from 8 to 16 cases per 100,000 cesarean deliveries (Fig. 2). Table 1 and Table S1, respectively, include the study covariates and codes.

Table 1 Patient characteristics and outcomes (2010–2019)

Fig. 2

Annual incidence of AFE after cesarean section

Demographics

Cesarean deliveries impacted by AFE had a higher average age of 32 years compared to 30 years (P < 0.001) (Table 1). Out of a total of 269 cases of AFE in women, 41 (15.2%) were Hispanic, 20 (7.4%) were Asian or Pacific Islander, 120 (44.6%) were white, 57 (21.2%) were black, and 28 (10.4%) were of other races. With a p-value of less than 0.03, we found this race distribution to be statistically significant. 48.7% of the individuals had private insurance, and 93.3% were located at urban teaching centers. Cesarean births affected by AFE led to hospital stays that were 2 days longer (5 days vs. 3 days, P < 0.001) (Table 1) and incurred significantly higher total costs ($76,582 vs. $20,278, P < 0.001) (Table 1). The patients diagnosed with AFE have a lower probability of being admitted for elective procedures compared to those without AFE (48.3% vs. 56.8%, P = 0.005). In addition, the occurrence of AFE after cesarean births was more likely to happen in teaching hospitals (65.4% vs. 60.2%, P = 0.078) and among individuals with three or more comorbidities (51.3% vs. 13.3%, P < 0.001) (Table 1). Patients who suffered from AFE after undergoing cesarean deliveries exhibited a markedly elevated rate of mortality during their hospital stay, in comparison to those who did not experience AFE. The death rate among patients with AFE was significantly greater than that of patients without AFE, with a rate of 14.9% compared to 0.0% (P < 0.001) (Table 1).

Variables associated with AFE following Cesarean deliveries

Concerning the criteria for AFE, the likelihood of experiencing perioperative AFE with cesarean deliveries is relatively low. This study conducted both univariate and multivariate analyses on the variables (Table S3-6), and those with p < 0.05 were selected for re-inclusion in the multivariate regression analysis. The multivariate analysis revealed that AFE during cesarean deliveries was independently associated with several factors. These factors include chronic blood loss anemia (odds ratio [OR] = 1.79; confidence interval [CI] = 1.27–2.51, p = 0.001), coagulopathy (OR = 13.396; CI = 10.18–17.63, p < 0.001), congestive heart failure (OR = 3.567; CI = 1.81–7.02, p < 0.001), other neurological disorders (OR = 3.702; CI = 2.48–5.53, p < 0.001), fluid and electrolyte disorders (OR = 15.069; CI = 10.83–20.96, p < 0.001), and pulmonary circulation disorders (OR = 4.245; CI = 1.93–9.34, p < 0.001). Additionally, abruptio placentae (OR = 1.946; CI = 1.28–2.95, p = 0.002), weight loss (OR = 2.357; CI = 1.08–5.14, p = 0.031), and polyhydramnios (OR = 2.336; CI = 1.43–3.82, p = 0.001) were also identified as significant risk factors for AFE during cesarean deliveries (Fig. 3, Table S7).

Fig. 3

Risk factors associated with AFE after cesarean section

AFE during cesarean deliveries was found to have a low independent risk in people aged 25 to 29 (OR = 0.65; CI = 0.45–0.95, P = 0.025), people with one comorbidity (OR = 0.23; CI = 0.13–0.40, P < 0.001), people with two comorbidities (OR = 0.57; CI = 0.37–0.86, P = 0.007), people with deficiency anemia (iron, vitamin B12, and folate deficiency anemia) (OR = 0.44; CI = 0.27–0.72, P < 0.001), and people with pre-eclampsia or eclampsia (OR = 0.46; CI = 0.32–0.67, P = 0.007) (Fig. 3, Table S7).

Discussion

Principal findings

The primary findings of the present investigation are as follows: A total of 2,462,005 cesarean deliveries incidence rate of AFE was 0.0113%, or 11.3 per 100,000 cesarean deliveries—is a descriptive, observational finding based solely on data from a large-scale national database (Table 1). The incidence of AFE showed a decline from 0.015% (15 cases per 100,000 cesarean deliveries) in 2010 to 0.008% (8 cases per 100,000 cesarean deliveries) in 2011. Subsequently, there was an increase from 0.008% in 2014 to 0.016% (16 cases per 100,000 cesarean deliveries) in 2019. The initial decline in AFE incidence from 2010 to 2011 may reflect improvements in obstetric care and early detection of risk factors. The subsequent rise from 2014 to 2019 could be attributed to heightened awareness and better recognition of AFE, as well as the inclusion of more severe cases that might have been previously underdiagnosed. Furthermore, the rate of death in patients who experienced AFE after cesarean births was considerably greater compared to those who did not experience AFE, with a mortality rate of 14.9% compared to 0.0%, respectively (P < 0.001) (Table 1). Next, we will provide the findings of a comprehensive study that examined the risk factors and sequelae of AFE in a large population-based cohort. Chronic blood loss anemia, coagulopathy, congestive heart failure, various neurological illnesses, fluid and electrolyte abnormalities, weight loss, pulmonary circulation disorders, abruptio placentae, and polyhydramnios were found to have a substantial association with AFE during cesarean deliveries (Fig. 3, Figure S1).

Risk factors

The correlation between AFE and cesarean deliveries, as it had not been previously documented. The data presented here support the conclusions of previous investigations that have shown a connection between placental anomalies [1, 26, 27], polyhydramnios [5], and operative deliveries [5, 26], with AFE. Nevertheless, this investigation has shown a novel connection between abruptio placentae, polyhydramnios, cesarean births, and AFE. This finding is significant and merits special consideration in the existing body of literature.

An explanation for the link between abruptio placentae and AFE could be that the disturbance of the connection between the mother and fetus could lead to greater exposure of fetal antigens and/or allow amniotic fluid to enter the mother’s bloodstream [5]. A clear link was seen between the severity of abruptio placentae and AFE, suggesting that the level of damage to the maternal-fetal interface directly affects the likelihood of AFE occurrence [1]. Nevertheless, the proposed correlation is presently a hypothesis-generating observation that requires cautious interpretation. Further research is required to see if there is a pathophysiologic mechanism that can explain this observation.

External validity was shown for the current study’s results on additional risk factors for AFE, as they agree with previous research from other researchers. Another interesting finding is the link between polyhydramnios and AFE. The main way that AFE happens is when amniotic fluid gets into the mother’s bloodstream, which then causes clinical symptoms to appear. Conventionally, it has been postulated that the occurrence of AFE necessitates a disruption in the physical barriers separating the maternal and fetal compartments [5], mainly at the level of endocervical veins [28], uterine trauma sites [29], and the placental attachment site [29]. It is hypothesized that a pressure gradient exists, which facilitates the movement of amniotic fluid from the uterus into the maternal circulation. The concept is substantiated by the evidence obtained from the research carried out by Talbert et al. [30], found that significant amounts of 125I-labeled albumin were transferred from the amniotic fluid to the mother’s bloodstream following the introduction of hypertonic saline by intraamniotic injection. This phenomenon likely accounts for the increased likelihood of AFE among women who have undergone surgery, experienced placenta previa, placental abruption, cervical laceration, uterine rupture, or polyhydramnios. Annick et al. [31]fund that losing weight during pregnancy was linked to lower perinatal risks for obese women from class III in this wealthy area, but it was not linked to the prevalence of low birth weight or small-for-gestational-age neonates. In our study, weight loss was associated with AFE. It was considered to be caused by abnormal nutrition deficiency, electrolyte imbalance and complications during pregnancy. It is important to think about stratifying the optimal gestational weight increase ranges for pregnant woman, although this remains speculative and requires further investigation.

Multivariate regression analysis showed that the protective effect of few comorbidity, deficiency anemia and pre-eclampsia/eclampsia appears counterintuitive. Upon further analysis, we believe that these factors may be associated with closer monitoring and more aggressive management during pregnancy, which could potentially mitigate the risk of AFE [32,33,34]. Generally, these diseases are treated with early intervention and termination of pregnancy in advance. However, we acknowledge that this is speculative, and further research is needed to explore this relationship.

Incidence

At the cohort level, we identified AFE in 6 out of every 100,000 deliveries, regardless of whether they were vaginal, surgical (including vacuum-assisted and forceps), or cesarean. The study found that the incidence rate of AFE was 0.0113%, which is equivalent to 11.3 cases per 100,000 cesarean deliveries (Table 1). It’s a little high compared to the previous data (0.7–8.8 per 100,000 deliveries) [1, 5, 13, 17, 18, 35,36,37,38]. This discrepancy may be attributed to several factors. First, differences in study populations could play a significant role. Second, variations in diagnostic criteria or case ascertainment methods across studies may contribute to the observed differences. Third, differences in healthcare settings, such as access to advanced obstetric care or variations in clinical practices, could also influence the reported incidence rates. These factors highlight the importance of contextualizing incidence rates within the specific characteristics of the study population and methodology.

The occurrence of AFE declined from 0.015 to 0.008% between 2010 and 2011 (Between 15 and 8 cases per 100,000 cesarean deliveries) (Fig. 2). The occurrence rate of AFE rose from 0.008 to 0.016% between 2014 and 2019 (8 to 16 cases per 100,000 cesarean deliveries) (Fig. 2). The study covariates and relevant codes are listed in Table S1. The suspected cause is thought to be linked to the diagnostic criteria applied retrospectively following AFE (AFE occurrences coinciding with cardiac arrest and coagulopathy) [1].

Clinical and research implications

Because AFE is rare, many healthcare practitioners, particularly those without specialized training, may be unfamiliar with this specific condition. The presence of overlapping clinical symptoms can exacerbate the lack of expertise, resulting in delayed identification. The present study advances our understanding of the associated factors (such as congestive heart failure, chronic blood loss anemia, fluid and electrolyte imbalances, diverse neurological disorders, coagulopathy, pulmonary circulation disorders, weight loss, polyhydramnios, and abruptio placentae) and mortality linked with AFE. The notable mortality rate underscores the critical need for improved training of obstetricians and anesthesiologists, such as through simulation exercises, to enhance a standardized approach to management [1, 39,40,41].

These findings have important implications for clinical practice, as they suggest that patients with the risk factors (such as congestive heart failure, chronic blood loss anemia, fluid and electrolyte imbalances, diverse neurological disorders, coagulopathy, pulmonary circulation disorders, weight loss, polyhydramnios, and abruptio placentae) may require differentiated surveillance and management strategies. Future research should focus on validating these findings and exploring the effectiveness of targeted interventions in reducing adverse outcomes. By integrating these insights into clinical decision-making, healthcare providers may be better equipped to identify and manage at-risk patients, ultimately improving maternal and neonatal outcomes.

Limitations

There are various constraints in this study. The diagnosis of AFE was primarily determined using the ICD-9 and ICD-10 codes from an administrative-type database, without doing a detailed examination of the medical records. We acknowledge that the use of administrative codes may lead to some misclassification, particularly given the lack of standardized diagnostic criteria for AFE. Furthermore, the absence of precise diagnostic criteria for AFE in each facility could potentially impact the relationship between exposure and result. Previous studies [18, 42] indicated that the occurrence of AFE seems to differ depending on the diagnostic criteria used. Similarly, the study did not assess the data accuracy for other variables, thus precluding the exclusion of potential misclassification in an unspecified number of patients, which might introduce an uncertain degree of bias.

This type of study is inherently prone to unmeasured bias. Specifically, the lack of available chronology between delivery factors and AFE hindered the capacity to establish causality [1]. Information regarding the neonatal details, the specific type of healthcare provider involved, and the precise cause of death were all unavailable. Other pertinent characteristics not investigated in this study include prolonged labor duration, placental implantation sites other than placenta previa, and the type of anesthesia used during delivery.

This study exclusively focused on identifying risk factors and mortality outcomes. Subsequent research [1, 18, 43] has underscored the importance of conducting effective cardiopulmonary resuscitation, managing coagulopathy with tranexamic acid administration, and employing extracorporeal membrane oxygenation in the treatment of AFE. However, the present study did not evaluate these specific methods. The presence of ascertainment bias is another restriction. Moreover, it is conceivable that the clinical diagnosis of AFE was more commonly made in the presence of pre-existing known risk factors, alongside considerations of the data capture methodology. However, the assessment of the generalizability of the data was not conducted and requires additional investigation. To address the infrequency of AFE, it would be beneficial to establish collaborative initiatives at both national and international levels in order to mitigate the occurrence of type 2 errors. It is imperative to establish and disseminate universal diagnostic criteria for AFE. Lastly, a limitation of our study is the absence of sensitivity analyses. our analysis did not control for certain potential confounders, such as prior history of AFE, other obstetric comorbidities, or variations in anesthetic utilization. These unmeasured factors could have influenced the observed outcomes and may limit the generalizability of our findings. Future studies should aim to incorporate these variables, and comprehensive sensitivity analyses to provide a more comprehensive understanding of the factors influencing AFE.

Conclusions

This cohort study demonstrates that although AFE remains a rare obstetric complication, it carries substantial maternal morbidity and mortality. We identified several independent risk factors, including cardiac, hematologic, neurological, and obstetric conditions. Management of high-risk patients requires a multidisciplinary approach with comprehensive pre-delivery planning, including detailed pre-anesthetic evaluation and surgical preparation.

Future research priorities should focus on: establishing standardized diagnostic criteria across institutions, investigating the molecular mechanisms underlying AFE development, developing predictive models for risk stratification, and evaluating the effectiveness of preventive strategies in high-risk populations through multicenter trials.