Clinical characteristics and predictors of neonatal outcomes in chorioamnionitis at term gestation: A cohort study

To investigate the association between clinical and laboratory characteristics of chorioamnionitis in deliveries at term gestation with adverse neonatal outcomes.


| 1081
TERM DELIVERIES COMPLICATED BY CHORIOAMNIONITIS Chorioamnionitis should be suspected in the presence of maternal intrapartum fever in combination with clinical signs of infection, such as maternal leukocytosis, purulent cervical drainage, 2 uterine tenderness, maternal tachycardia, maternal malaise, elevated C-reactive protein (CRP), foul-smelling water or discharge, 1,3,5,6 and/or signs of fetal distress (i.e. abnormal fetal heart rate pattern). 7 Between 1% and 5% of term deliveries are estimated to be complicated by chorioamnionitis, which is associated with a significantly increased risk of maternal, fetal and neonatal morbidity and mortality. 2,3 Neonates exposed to chorioamnionitis are at increased risk of severe infections in the neonatal period, 3,[8][9][10] and are likewise at an increased risk of impaired short-and long-term neurological outcomes, hypothesised to be a consequence of the fetal hyperinflammatory response to the infection. 11,12 Undoubtedly, the early identification and timely treatment of chorioamnionitis in combination with close collaboration between obstetric and paediatric care is essential to improve neonatal outcomes, 13 but precise clinical predictors of neonatal complications are lacking. Most studies have focused on chorioamnionitis complicating preterm deliveries, whereas only a few studies have investigated the potential impact of clinical and laboratory characteristics of term deliveries complicated by chorioamnionitis. Risk prediction models based on gestational age, 14 duration of membrane rupture, highest maternal intrapartum temperature and timing of intrapartum antibiotic administration have been proposed, 2,3 but except for maternal intrapartum temperature, no study has investigated the association between clinical and laboratory characteristics of chorioamnionitis in term deliveries and the risk of adverse neonatal outcomes.
The aim of our study was therefore to describe the clinical and laboratory characteristics of term deliveries complicated by chorioamnionitis, and to assess the association between these characteristics and the risk of adverse neonatal outcomes.

| Design
We performed a retrospective cohort study of term deliveries complicated by chorioamnionitis. The study was approved by the Swedish Ethical Review Authority. There was no public or patient involvement in the development, progress or reporting of this study.

| Setting
Annually, approximately 115 000 women give birth in Sweden. The vast majority of all pregnant Swedish women attend the free-of-charge maternal health programme and essentially all women give birth in hospital-based delivery wards. The Stockholm region has six delivery wards with approximately 28 000 deliveries annually, accounting for approximately one-quarter of all deliveries in Sweden.

| Data sources
The Swedish Pregnancy Register (www.gravi ditet sregi stret. se) is a national quality register, established in 2013. It receives data from multiple sources. Most variables are directly transferred from the electronic medical records (EMR). The register contains detailed information of the pregnancies, entered prospectively into the EMR by midwives and physicians in a standardised manner, starting with the first antenatal visit, and thereafter at every subsequent visit, including ultrasound examinations, delivery and postnatal care visits. Additionally, the register includes diagnoses and procedure codes for both mothers and infants, coded using the International Classification of Diseases (ICD) system (with v10 in use since 1997). Information on clinical characteristics (symptoms and clinical findings such as maternal and fetal heart rate pattern) and laboratory measurements (CRP, leukocyte count and cultures) were extracted manually from medical records by the main author (PB).

| Study participants
All term deliveries in the Stockholm Region between 1 January 2014 and 31 August 2020 complicated by chorioamnionitis, defined as a registered diagnosis in the Swedish pregnancy register (ICD-10 O41.1) were identified. From these, a sample of 500 was taken, using simple random sampling, to make up the study cohort. The diagnosis of chorioamnionitis is registered based on the responsible obstetrician's assessment and management. Recently, an expert panel of maternal and neonatal experts recommended dividing chorioamnionitis into the three separate categories 'Isolated maternal fever', 'Suspected chorioamnionitis' and 'Confirmed chorioamnionitis' (Table S2 in Supplementary informaion). 2 Based on this categorisation, and for the purpose of this study, 'Isolated maternal fever' is defined as at least one registered maternal temperature between 38.0°C and 38.9°C, without any additional signs of infection. 'Suspected chorioamnionitis' is defined as: (i) one or more registered maternal temperature of ≥39.0°C; or (ii) one or more registered maternal temperature between 38.0°C and 38.9°C in combination with at least one clinical sign of infection, including fetal tachycardia of >160 beats per minutes (bpm) for 10 minutes or longer, maternal leukocytosis (with a leucocyte count of >15 000 cells/mm 3 ) or purulent discharge from the cervical os. 'Confirmed chorioamnionitis' is defined as: (i) a positive culture from the cervix or amniotic cavity, confirming the presence of bacteria; or (ii) placental histopathological changes typical for chorioamnionitis. Histopathological diagnosis of chorioamnionitis was defined as acute inflammatory lesions in the chorion, amnion or villous tree, indicating chorioamnionitis as diagnosed by the pathologist. The registered chorioamnionitis diagnosis in the Swedish Pregnancy Register has not previously been validated, and therefore a validation of chorioamnionitis diagnosis was performed (Methods S3 in Supplementary information).

| Exposure and covariates
Baseline characteristics were extracted from the Swedish Pregnancy Register. Maternal age was categorised into <25, 25-30, 31-35 and >35 years of age at delivery. Parity was categorised into nulliparous, primiparous or multiparous (≥2 para). Early pregnancy body mass index (BMI, kg/m 2 ) was categorised according to the World Health Organization (WHO) nutritional status categories: underweight (BMI < 18.5 kg/m 2 ), normal weight (BMI = 18.5-24.9 kg/m 2 ), overweight (BMI = 25.0-29.9 kg/m 2 ) and obese (BMI ≥ 30 kg/m 2 ). Information on education level, smoking status in early pregnancy and selected prepregnancy comorbidities was also obtained from the register (Table S5 in Supplementary informaion) according to the self-reported information collected and transferred by the midwife at the first antenatal visit. Education level was categorised into fewer than 9 years of schooling, at least elementary school (or equivalent) and up to high school (9-12 years of schooling), and higher level of education, i.e. university, college or equivalent (>12 years of schooling). Self-reported presence of prepregnancy diabetes mellitus, essential hypertension and recurrent urinary tract infection was reported as yes/no, as was smoking status in early pregnancy.
Delivery characteristics were also obtained from the Swedish Pregnancy Register. Gestational age at delivery was further stratified into early term (between 37 +0 and 38 +6/7 weeks of gestation), full term (between 39 +0 and 40 +6/7 weeks of gestation), late term (between 41 +0 and 41 +6/7 weeks of gestation) and post-term (≥42 +0 weeks of gestation) delivery. 15 The delivery onset was stratified into spontaneous, medically induced and caesarean delivery before onset of contractions. Mode of delivery was categorised into non-instrumental vaginal delivery, instrumental delivery (vacuum extraction) and caesarean delivery, according to registration in the Swedish Pregnancy Register.
Specific clinical and laboratory characteristics used to diagnose chorioamnionitis and CRP (which is not among the diagnostic criteria for chorioamnionitis but is used for diagnosing infectious diseases) were collected from medical charts. Information on maternal temperature during delivery was obtained and reported as: (i) no reported temperature of ≥38.0°C; (ii) at least one reported temperature of between 38.0°C and 38.9°C; (iii) a reported temperature of between 38.0°C and 38.9°C persisting for at least 30 min; and (iv) at least one reported temperature of 39.0°C or higher. First and highest leucocyte count and CRP level was extracted from medical charts and categorised into tertiles. First reported values were, in general, based on blood samples taken in conjunction with diagnosis (during labour, prior to or in conjunction with delivery), whereas the highest laboratory values were based on follow-up laboratory values, including postpartum care. Depending on the delivery time point and clinical presentation, the blood testing interval differed between study participants. Fetal tachycardia was defined as a fetal heart rate above 160 beats per minute (bpm) (and for a duration of ≥10 min) during a period of maternal fever, and was extracted from the cardiotocography (CTG) registration. Maternal tachycardia was defined as heart rate of 100 bpm or above in the presence of maternal fever, and was extracted from text, registered clinical parameters or, if captured, from CTG registration (by pulse oximetry). Clinical findings and symptoms of chorioamnionitis (purulent cervical drainage, uterine tenderness, maternal malaise (a lack of well-being and a feeling of weakness or overall discomfort, as a sign of infection) and/or foul-smelling amniotic fluid or discharge) was registered as present if specified in the medical chart text. A positive culture, cervical or blood, was defined as the presence of bacteria confirming the diagnosis microbiologically. Positive pathological-anatomic diagnosis (PAD) was defined as the presence of acute inflammatory changes in the amnion or chorion in a pathological analysis of the placenta.

| Outcomes
Adverse neonatal outcomes were stratified into neonatal infection and asphyxia-related conditions, defined by ICD-10 diagnoses recorded in the Swedish Pregnancy Register. Neonatal infection included necrotising enterocolitis, congenital pneumonia, sepsis, urinary tract infection, skin infection, umbilical infection and unspecified infections during the neonatal period. Asphyxia-related conditions included meconium aspiration, hypoxic ischaemic encephalopathy, convulsions/seizures, hypoglycaemia, jaundice and respiratory distress syndrome (RDS) (ICD-codes specified in Table S6 in Supplementary information).

| Statistical analyses
Descriptive statistics were compiled, tabulated and presented as absolute numbers, with means or medians and corresponding standard deviations or interquartile ranges, depending on the covariate distribution. To assess the association between clinical and laboratory characteristics of deliveries complicated by chorioamnionitis and adverse neonatal outcomes, logistic regression models were used to estimate odds ratios (ORs). For inflammatory laboratory measurements the lowest tertile was used as a reference. A multivariate model adjusted for maternal age at delivery, smoking status, BMI and parity was performed. Statistical significance was set at p < 0.05. All analyses were performed using SAS® 9.4 (SAS Institute, Cary, NC, USA).

| Study participants
Between 1 January 2014 and 31 August 2020, 133 382 term singleton deliveries in the Stockholm Region were registered in the Swedish Pregnancy Register, 629 (0.47%) of which had a registered chorioamnionitis diagnosis. A random sample of 500 of these made up the study cohort ( Figure S1 in Supplementary information).

| Baseline characteristics
Baseline characteristics, including the demographic, pregnancy and delivery characteristics of all study participants, are presented in Table 1. The mean maternal age at delivery was 31.5 years (SD 5.2 years) and 392 (79.2%) study participants were nulliparous. In total, 243 (48.6%) study participants had a normal BMI, and 221 (44.2%) were overweight (29.8%) or obese (14.4%). Half of the study participants (50.8%) reported an educational level of >12 years of schooling, and 15 (3%) study participants reported smoking in early pregnancy. Five (1%) study participants had diabetes mellitus prior to pregnancy, and 66 (13%) study participants reported prepregnancy recurrent urinary tract infection. In total, 28 (5.6%) study participants developed gestational diabetes and 49 (9.8%) study participants were diagnosed with pregnancy-induced hypertensive disorder.

| Clinical and laboratory characteristics of chorioamnionitis
Information on clinical and laboratory characteristics and data on neonatal and maternal outcomes of all study participants are presented in Table 2. In total, 49 (9.8%) study participants fulfilled the criteria of isolated maternal fever, 397 (79.4%) study participants fulfilled the criteria of suspected chorioamnionitis and 151 (30.2%) study participants fulfilled the criteria of confirmed chorioamnionitis. The corresponding positive predictive value of chorioamnionitis overall was excellent (0.92, 95% CI 0.89-0.94) ( Table S4 in Supplementary informaion). The vast majority (480/500, 96%) of the cohort received intravenous antibiotics. In total, 348 (70%) cervical cultures were sent for analysis, 133 (38%) of which were positive. The most frequently isolated bacteria were Group B streptococci (GBS), present in 60% of positive cervical cultures, followed by Escherichia coli (12%) (all specific bacteria detected in the positive cervical cultures are presented in Table S7 in Supplementary informaion).

| Clinical and laboratory characteristics of chorioamnionitis and risk of adverse neonatal outcomes
In total, 29 (6%) neonates had an Apgar score of <7 at 5 minutes, and 10% suffered from an infection in the neonatal period. In 58 (12%) neonates the umbilical artery blood gas indicated fetal acidaemia, and 110 (22%) neonates suffered from a complication, which could indicate asphyxia ( Table 2).
The association between each clinical or laboratory characteristic and risk of composite neonatal infection and/or asphyxia-related complications is presented in Tables 3 and  4. Maternal intrapartum temperature persisting between 38.0°C and 38.9°C, or a single maternal temperature of ≥39°C was not associated with an increased risk of neonatal infection. A first leukocyte count in the second tertile was associated with an approximately doubled risk of neonatal infection (aOR 2.14, 95% CI 1.02-4.49), whereas there was no association between a first leukocyte count in the third tertile and the risk of neonatal infection. There was no significant association between the highest maximum leukocyte count or first CRP level and risk of neonatal infection. The highest CRP in the third tertile was associated with an approximately fourfold increased risk of neonatal infection (aOR 4.01, 95% CI 1.66-9.68). There was a borderline significant association between fetal tachycardia and neonatal infection (aOR 1.99, 95% CI 0.99-4.00). A positive cervical culture was associated with a doubled risk of neonatal infection (aOR 2.22, 95% CI 1.10-4.48).
Neither maternal fever nor first or highest leukocyte count were associated with an increased risk of asphyxiarelated complications. Likewise, there was no association between the first measure of CRP and the outcome. A highest CRP in the third tertile was associated with an approximately doubled risk of composite asphyxia-related complications (aOR 1.93, 95% CI 1.09-3.41). Fetal tachycardia was associated with an increased risk of asphyxia-related complications (aOR 1.63, 95% CI 1.01-2.65).

| Main findings
In this cohort study of chorioamnionitis in term singleton deliveries, we found that the prevalence of neonatal complications was high. Moreover, we found that elevated inflammatory laboratory values and positive cervical culture were both associated with an increased risk of neonatal infection, and that elevated inflammatory laboratory values and fetal tachycardia were both associated with an increased risk of asphyxia-related complications. Our findings suggest that maternal CRP could be useful to incorporate in the management of women and neonates exposed to chorioamnionitis, and that continuous collaboration between obstetric and neonatal care should be promoted, to facilitate the early detection and prevention of neonatal complications.

| Strengths and limitations
This study has multiple strengths. First, we were able to include the majority of cases with chorioamnionitis at term delivery in the Stockholm Region, accounting for 25% of all deliveries in Sweden, during the study period. We performed random sampling, in which all individuals of the source population had the same odds of being selected, which minimised the risk of sampling bias. Second, the Swedish Pregnancy Register contains prospectively collected detailed information from the pregnancy and postpartum period, which precludes the risk of recall bias. Specific clinical information was extracted retrospectively from medical charts, but the information on neonatal outcomes was added separately after medical chart review, which is why we regard the risk of recall bias as being minor in this study. Third, our validation of registered chorioamnionitis diagnosis against medical chart data revealed excellent validity, thereby confirming the internal validity of the study. Despite these methodological advantages, our results should be interpreted in light of some potential study limitations. The validity of neonatal diagnoses in the Swedish Pregnancy Register has not been assessed. There may be cases of delayed presentation of adverse neonatal outcomes, which might not be captured by the Swedish Pregnancy Register, potentially leading to an overall underestimation of adverse neonatal outcomes. It is possible that there were patients with chorioamnionitis who were misdiagnosed (false negative) during the study period, and consequently not included in the study. Only 59 placentas were sent for histopathological analysis, the majority of which indicated chorioamnionitis. Adverse neonatal outcomes might increase the chance of requesting histopathological analysis, which could lead to histopathological chorioamnionitis being overestimated. One of the researchers extracted the data from medical charts, which could introduce bias in terms of misclassification of covariates or misinterpretation of information. To prevent such bias, a validation of the extraction file was performed prior to data collection, in which two of the authors extracted data independently of each other. The consistency of the data with this validation was excellent. Finally, our regional cohort comprising deliveries within Stockholm county, makes it difficult to fully extrapolate our results to other settings because of potential differences in demographic characteristics among pregnant women in different regions and countries and among different diagnostic traditions. We do, however, think that our main findings -an association between specific signs of chorioamnionitis and neonatal complications -may be generalisable to other populations.

| Interpretation
The chorioamnionitis prevalence of 0.5% in our study is considerably lower than previously reported rates of between 1% and 5%. However, these previously reported rates are, in general, based on older studies of placental findings in primiparous women, and are consequently not readily comparable with our results. 16 Maternal fever is the most important clinical sign of chorioamnionitis, and was present in approximately 90% of the study participants. Surprisingly, maternal fever was not noted in the remaining 10%, which may be a consequence of inadequate medical chart documentation rather than an absence of elevated maternal temperature. Maternal and fetal tachycardia was highly prevalent (65% and 67%, respectively), which is consistent with previously reported rates of between 50% and 80%. 17,18 Likewise, in line with previous reports, the presence of subjective clinical signs, such as uterine tenderness and foul-smelling amniotic fluid, were rare. 18 Uterine tenderness was only noted in 2% of the women with chorioamnionitis, but can be difficult to determine during active labour, and the presence was only recognised if explicitly stated in the medical chart text. Hence, the prevalence of these highly subjective clinical signs is presumably underestimated.
According to current guidelines, chorioamnionitis diagnosis can only be confirmed postpartum by histopathological analysis of placenta and/or positive cervical culture. 2 Therefore, we must rely on clinical signs and laboratory markers of infection when diagnosing and managing patients with suspected chorioamnionitis in clinical practice. Suspected chorioamnionitis should prompt initiating treatment with broad-spectrum intravenous antibiotics and careful monitoring, and labour progression should be ensured. 2 Although standard obstetric indications for caesarean delivery apply in deliveries complicated by chorioamnionitis, the myometrial contractility can be affected by the inflammation, leading to an increased risk of labour dystocia, caesarean delivery and postpartum haemorrhage, 19,20 which presumably explains the observed high rates of caesarean delivery and postpartum haemorrhage in our study.
Chorioamnionitis is an established risk factor for early-and late-onset neonatal sepsis, 10 and its presence motivates intensified monitoring, testing and sometimes empiric antibiotic treatment of the neonate. Multiple attempts to develop prediction models of the risk of developing neonatal sepsis have been made, not the least to reduce the usage of empiric antibiotic treatment, which has been associated with an increased risk of negative childhood outcomes, including asthma. 21 The impact of clinical and laboratory characteristics of chorioamnionitis on the risk of neonatal complications has not been addressed in detail previously. We found that a moderately elevated first leukocyte count, highest maximum CRP level and positive cervical culture were all associated with neonatal infection, and that a high maximum CRP and fetal tachycardia were both associated with asphyxia-related complications, which is a novel finding. Fetal tachycardia in the presence of maternal fever is a known predictor of adverse neonatal outcomes, and may be a sign of fetal inflammatory response, which has been proposed as a mechanism of impaired short-and long-term neurological outcomes. 22,23 A release of fetal pro-inflammatory cytokines, in response to inflammation, has been hypothesised to have a direct toxic effect on the brain, leading to adverse neurological outcomes. 12 A culture was sent in only 70% of cases and of these, only 40% were positive. During the medical chart review it was noted that a majority of patients were treated with intravenous antibiotics prior to the culture being taken, which could distort bacterial growth and lead to false-negative cultures. There is emerging evidence of chorioamnionitis being a clinical syndrome with various aetiologies, with intra-amniotic infection being one and with sterile intraamniotic inflammation being another. 24 In fact, similar to our results, previous studies report low rates of bacteria isolated from cervical or amniotic fluid cultures in patients with chorioamnionitis. 25,26

| Research and clinical implications
Our results pinpoint the difficulties in identifying clinical tools for the timely prediction of adverse neonatal outcomes associated with chorioamnionitis. We observed a strong association between the highest measured CRP and positive cervical culture with neonatal infection. Our findings reflect the dynamic nature of CRP, with a characteristic slow increase in response to inflammation. Thus, CRP might have limited usefulness in predicting neonatal outcomes prior to delivery but, regardless, should be further evaluated for incorporation in the management of chorioamnionitis. Streamlined follow-up blood testing of women with chorioamnionitis in combination with close collaboration between obstetric and neonatal care could contribute to the early identification of neonatal complications and facilitate preventive measurements.

| CONCLUSION
We found that elevated inflammatory markers were associated with both neonatal infection and asphyxia-related complications, and fetal tachycardia additionally predicted asphyxia-related complications. Our findings highlight the importance of a close collaboration between obstetric and neonatal care, extending past delivery, to detect maternal dynamic laboratory changes that could impact neonatal outcomes.

AU T HOR C ON T R I BU T ION S
All authors contributed substantially to this study.

F U N DI NG I N FOR M AT ION
ÄM was supported by Region Stockholm. The funders had no role in study design, analysis and interpretation of the data, or writing the article.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared. Completed disclosure of interests form available to view online as supporting information.

DATA AVA I L A BI L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

E T H IC S A PPROVA L
This study was approved by the Swedish Ethical Review Authority.