Prevalence and risk factors of chorioamnionitis in Dhaka, Bangladesh

Abstract

Objective:

To estimate the association between chorioamnionitis, maternal risk factors and birth outcomes.

Study Design:

A cross-sectional study of 600 pregnant women was conducted at a maternity center in Dhaka from January to October 2011. Outcomes included histologic, microbiologic and clinical chorioamnionitis. Log-binomial models assessed the association between risk factors and histologic chorioamnionitis (HC).

Results:

Of the 552 women with placental specimens, 70 (12.7%) were classified with HC: 46 (65.7%) with and 24 (34.3%) without fetal involvement. HC was associated with non-physician care (relative risk [RR] 2.04, 95% confidence interval [CI] 1.04 to 4.00), home slab or hanging latrine (RR 1.69, 95% CI 1.10 to 2.62), and lack of tetanus toxoid (RR 1.80, 95% CI 1.03 to 3.14). Women with fever (RR 2.30, 95% CI 1.18 to 4.50) or discolored amniotic fluid (RR 1.74, 95% CI 1.08 to 2.81) had a higher risk of HC. Microbiologic and clinical chorioamnionitis were unreliable HC measures.

Conclusion:

Prevalence of HC is high; many cases are not captured by clinical diagnosis or microbiologic cultures.

Introduction

Chorioamnionitis, a common complication of pregnancy, is associated with significant maternal and perinatal adverse outcomes including puerperal and postpartum infections, stillbirth, preterm birth and neonatal sepsis.^{1, 2, 3, 4, 5, 6} Chorioamnionitis refers to acute inflammation of the membranes and chorion of the placenta, typically from an ascending polybacterial infection in the setting of membrane rupture. The definition varies according to diagnostic criteria—clinical (presence of clinical characteristics), microbiologic (positive culture of microbes from amniotic fluid or chorioamnion) or histologic (microscopic evidence of chorioamnionic inflammation, with or without fetal vasculitis, within the placenta).¹ Although the three diagnostic criteria overlap, women with histologic chorioamnionitis (HC) often present asymptomatically. Women without clinical signs of intrauterine infection may still experience an underlying histologic process of inflammation secondary to an infectious process.

In high-income settings, chorioamnionitis is commonly diagnosed based on histologic findings to capture both sub-clinical (clinically unapparent) as well as clinical chorioamnionitis.^{7, 8} Microscopic examination of the placenta assesses both the origin of the inflammatory cells (maternal versus fetal) and the severity of the inflammation with different sites and severity scores.⁹ In low-, middle- income countries, the incidence of puerperal sepsis (chorioamnionitis and postpartum endometritis) during pregnancy is unclear but estimated to be high ranging from 0.1 to 10% due to limited antenatal care (ANC) coverage and discrepancies in diagnostic testing.^{10, 11} The case fatality from puerperal sepsis is as high as 30 to 50%.¹¹ In these settings, most cases of chorioamnionitis are diagnosed based on nonspecific clinical signs as trained pathologists and histopathology laboratories are often unavailable. There is a great need to explore the epidemiology, risk factors and diagnostic methods for life-threatening maternal infections. This paper describes the prevalence of chorioamnionitis using histologic, microbiologic and clinical measures; the associations between maternal risk factors and chorioamnionitis; and the associations between chorioamnionitis and birth outcomes in Dhaka, Bangladesh.

Materials and Methods

Study design

A cross-sectional study was conducted within a larger cohort study on the maternal origins of neonatal infection at a maternity center in Dhaka, Bangladesh following 600 mother–newborn pairs from 15 January 2011 to 31 October 2011. The maternity center was operated by our partner non-governmental organization, Shimantik, and averaged 150 deliveries per month. In Bangladesh, ~29% of deliveries occur in facilities.¹² We enrolled pregnant women at the maternity center during ANC visits or upon presentation in labor.³ Women who were pregnant ⩾30 gestational weeks and planned to deliver at the maternity center were included in the study. Women with fetal distress, obstructed labor, hemorrhage or severe pre-eclampsia were excluded to facilitate their need for urgent care. This study received ethical approval from the Johns Hopkins Bloomberg School of Public Health Committee on Human Research and the International Centre for Diarrheal Disease Research, Bangladesh Ethical Review Committee. All study participants provided written informed consent (IRB No. 00003084, approved November 2010).

Shimantik recruited four paramedics and five community health workers for primary data collection. Paramedics completed higher secondary school (12 years) and the national paramedic course. Community health workers completed at least secondary school (10 years) education. Two medical officers supervised field activities and monitored quality assurance. Staff received 2 weeks of intensive training by a pediatrician and local medical officer. Sessions included presentations, exercises, role-plays and hospital visits to recognize clinical signs and symptoms of infection.¹³ Written exams and standardized observations on staff skills were done to assess competency.

For the primary aim of the parent cohort study, 600 mother–newborn pairs were required to observe the estimated risk of early onset neonatal infection with 80% power while allowing for 5% type one error. Six hundred mother–newborn pairs detected the predicted prevalence of HC of 10%±3.5% precision with 80% power. Additional details about the study design are discussed elsewhere.³

Maternal risk factors

Study paramedics interviewed women to collect demographic data on maternal age, maternal education, ANC provider type and receipt of tetanus toxoid as a proxy for access to health care. Wealth quintiles were created using principal components analysis conducted with the following demographic variables: housing materials of primary residence; source of water supply; type of latrine; number of household inhabitants; number of household inhabitants under 5 years; and number of rooms where household members sleep.¹⁴

To assess maternal risk factors and clinical signs during labor, a study paramedic was present 24 h a day, 7 days a week in the labor and delivery room to measure duration of labor, rupture of membranes, number of vaginal exams performed and handwashing by health workers. Paramedics also collected data on neonatal characteristics such as sex, birth weight to the nearest 100 g and gestational age based on ultrasound report.

Chorioamnionitis

To classify HC, four sections of the placenta were collected immediately after delivery: a section from the extraplacental membranes, a cross-section of the umbilical cord and two sections of the central placental disc (one from the fetal surface and one from the maternal surface). Sections were fixed in 10% buffered formalin and then embedded in paraffin until processing. A pathologist, who was masked to patient characteristics and clinical signs, reviewed the histopathology slides. An expert placental pathologist reviewed a random sample (10%) of the slides for quality assurance. HC was defined as maternal neutrophilic inflammation in the extraplacental chorioamnionic membrane layers and/or with maternal neutrophil migration into chorionic plate. HC with fetal involvement was defined as cases with maternal neutrophilic inflammation as defined above and with concomitant migration of fetal neutrophils through the umbilical vessels and/or fetal vessels of the chorionic plate.

To classify microbiologic chorioamnionitis, surface cultures of placental membranes were taken with sterile swabs in the space between the chorionic and amniotic membranes immediately after delivery of the placenta. Swabs were transported daily from the maternity center to the microbiology laboratory at Dhaka Shishu Hospital using Amies transport medium and processed within 24 h of collection. Primary inoculation of the swabs was done on MacConkey and blood agar media with 5% sheep blood and incubated at 37 °C for 18 to 24 h. Identification of organisms was done by gram staining and standard biochemical reactions.

Clinical chorioamnionitis was diagnosed based on the presence of clinical signs during labor and delivery—maternal fever and one or more of the following symptoms: uterine fundal tenderness, maternal tachycardia (>120 bpm), fetal tachycardia (>160 bpm), purulent, foul or discolored amniotic fluid or vaginal discharge.

The primary outcomes were as follows: (1) HC and (2) HC with fetal involvement. Secondary outcomes were microbiologic chorioamnionitis and clinical chorioamnionitis.

Birth outcomes

We explored associations between chorioamnionitis and potential complications such as preterm birth, small for gestational age (SGA) and low birth weight. Preterm birth was defined as <37 weeks gestational age and very preterm as <35 weeks of gestational age. SGA was calculated using the Intergrowth-21st standards, which accounts for gestational age and sex. We defined SGA as <10th percentile and very SGA as <3rd percentile. Low birth weight was defined as <2500 g at birth and very-low-birth weight as <2000 g at birth.

Statistical analysis

We explored each of the risk factor variables, clinical signs and potential complications using t-tests, χ²- or fisher’s exact tests. We did not make any adjustments for multiple comparisons.¹⁵ Multivariable analyses using log-binomial regression models were conducted to examine the effect between maternal risk factors on HC. Potentially significant variables in the bivariate analyses (P<0.20) were entered into a backward stepwise selection procedure (with a threshold at P<0.05). The final model was adjusted for the potential confounding effects of wealth. Analyses were repeated for HC with fetal involvement, comparing women with HC with fetal involvement to women without HC with fetal involvement (women without HC and women with HC but no fetal involvement). Analyses were then repeated for secondary outcome measures—microbiologic and clinical chorioamnionitis. We assessed correlations between histologic classification, and microbiologic chorioamnionitis and clinical chorioamnionitis using kappa coefficient, which provides a measure of concordance above that expected by chance. We conducted sensitivity and specificity tests with each of the chorioamnionitis measures with HC as the gold standard. Analyses were performed using STATA v13 (StataCorp, College Station, TX, USA).

Results

We enrolled 600 mother–infant pairs. Forty-eight women were missing placental samples; of these women, 19 women (39.6%) delivered elsewhere, placentas were torn in two cases (4.2%), the paramedic was not available at three deliveries (6.3%) and 24 (50%) were missing data.

Prevalence of chorioamnionitis

Of the 552 mother–infant pairs with placental samples, 70 (12.7%) were classified as having HC. Forty-six of the 70 HC cases (65.7%) were classified as having HC with fetal involvement. Of the 533 women with placental cultures, 57 (10.7%) women had microbiologic chorioamnionitis. The most common organisms were Escherichia coli (n =19, 33.3%), Pseudomonas species (n =11, 19.3%) and Klebsiella pneumoniae (n =7, 12.3%) (Table 1). A small proportion, 13 (2.4%) of the 552 with placental samples had clinical symptoms of chorioamnionitis (fever and one or more of the following: maternal tachycardia >120 bpm, fetal tachycardia >160 bpm, purulent or foul smelling amniotic fluid or vaginal discharge, or uterine tenderness).¹⁶

Table 1 Distribution of organisms from placenta cultures

Maternal risk factors and clinical signs associated with histologic chorioamnionitis

Histologic chorioamnionitis

HC was associated with ANC by a non-physician (relative risk [RR] 2.04, 95% confidence interval [CI] 1.04 to 4.00), slab or hanging latrine at home (RR 1.69, 95% CI 1.10 to 2.62) and no receipt of tetanus toxoid during pregnancy (RR 1.80, 95% CI 1.03 to 3.14).

HC was not associated with maternal age, maternal education, maternal tobacco use, source of drinking water, household wall, floor, roof material, medications received during ANC visits or handwashing by providers (Table 2).

Table 2 Risk factors for chorioamnionitis (n =552)^a

Clinical signs such as fever ⩾99 °F (RR 2.30, 95% CI 1.18 to 4.50) and discolored amniotic fluid (RR 1.74, 95% CI 1.08 to 2.81) were associated with HC (Table 3). Clinical signs such as uterine tenderness, foul smelling vaginal discharge and purulent vaginal discharge were not associated with HC with or without fetal involvement (Table 3).

Table 3 Clinical signs and symptoms of chorioamnionitis (n =552)^a

After adjusting for wealth, ANC by a non-physician (RR 2.13, 95% CI 1.08 to 4.21), no receipt of tetanus toxoid during pregnancy (RR 1.95, 95% CI 1.09 to 3.47), discolored amniotic fluid (RR 1.88, 95% CI 1.18 to 3.00) and fever ⩾99 °F (RR 2.88, 95% CI 1.56 to 5.32) were significantly associated with a higher risk of HC (Table 4).

Table 4 Log-binomial multivariable models on the risk factors and clinical signs for histologic chorioamnionitis, histologic chorioamnionitis with fetal involvement and microbiologic chorioamnionitis adjusting for wealth

Histologic chorioamnionitis with fetal involvement

HC with fetal involvement was associated with risk factors such as maternal receipt of ANC by a non-physician health worker (RR 3.17, 95% CI 1.16 to 8.69), tube water source rather than tap water (RR 1.95, 95% CI 1.10 to 3.44) and slab or hanging latrine at home (RR 2.20, 95% CI 1.26 to 3.84). Fetal involvement was not associated with maternal age, maternal education, maternal tobacco use, household wall, floor, roof material, no receipt of tetanus toxoid during pregnancy, medications received during ANC visits or handwashing by providers (Table 2).

HC with fetal involvement was associated with one clinical sign—discolored amniotic fluid (RR 2.04, 95% CI 1.13 to 3.68) (Table 3). After adjusting for wealth, ANC by a non-physician health worker (RR 2.95, 95% CI 1.07 to 8.11) and discolored amniotic fluid (RR 2.01, 95% CI 1.11 to 3.64) was significantly associated with a higher risk of HC with fetal involvement (Table 4).

Maternal risk factors and clinical signs associated with microbiologic chorioamnionitis

Microbiologic chorioamnionitis was associated with handwashing by providers (RR 1.91, 95% CI 1.16 to 3.16) (Table 2). Uterine tenderness was associated with microbiologic chorioamnionitis (RR 0.41, 95% CI 0.19 to 0.86) (Table 3). When adjusting for wealth, microbiologic chorioamnionitis remained significantly associated with uterine tenderness (RR 0.42, 95% CI 0.20 to 0.91) (Table 4). Microbiologic data were not a reliable measure of HC (k=−0.004). Using HC as the gold standard, microbiologic chorioamnionitis had a sensitivity of 10.9% and positive predictive value of 12.3%.

Maternal risk factors and clinical signs associated with clinical chorioamnionitis

Clinical chorioamnionitis was not associated with any risk factors (Table 2). After adjusting for wealth, there was still no association with any risk factors. Clinical chorioamnionitis was not a reliable measure of HC (k=0.037). Using HC as the gold standard, clinical signs of chorioamnionitis had a sensitivity of 4.3% and positive predictive value of 23.1%. Among 65 women with HC and data on clinical signs of chorioamnionitis, 62 (95.3%) were clinically asymptomatic.

Chorioamnionitis and birth outcomes

HC with fetal involvement and clinical chorioamnionitis were both significantly associated with a higher risk of preterm birth <35 weeks (RR 5.70, 95% CI 1.09 to 29.75 and RR 10.9, 95% CI 1.49 to 79.69, respectively). There were no classifications of chorioamnionitis associated with a higher risk of SGA<10th percentile, SGA<3rd percentile, low birth weight, very-low-birth weight or preterm birth <37 weeks (Table 5).

Table 5 Chorioamnionitis and birth outcomes (n =552)

Discussion

In the sample of 552 placentas, the prevalence of HC was estimated to be 13% and HC with fetal involvement to be 8%. In prior studies, the prevalence of chorioamnionitis, based on placental pathology, varied widely from 7 to 85% depending on gestational age, race and ethnicity, and the threshold of polymorphonuclear leukocytes per high power field.¹⁷ Most studies have been done in high-income settings in relation to stillbirths and preterm births.¹⁸ In a study of 7505 placentas delivered after 20 weeks gestation, the prevalence of HC was 40% between 25 and 28 weeks gestation, 35, 10 and 4% at gestational weeks 29 to 32, 33 to 36 and 37 weeks or greater, respectively.¹⁹ In a case–case-control study in Sweden, HC was diagnosed in 82% of stillbirths and 68% of ‘healthy’ deliveries.²⁰ Results were similar in a case-control study in Mozambique; chorioamnionitis was diagnosed in 96% of stillbirths and in 67% of term deliveries.²¹ In our study in Dhaka, HC with fetal involvement was significantly associated with preterm birth <35 weeks. The large difference in the prevalence of HC can be explained by methodological differences in definitions and histologic methods, and differences in the populations and contexts where the study took place.

The etiology of HC is not clear, however it is believed to be associated with a microbial cause, based on the high correlation with preterm births. Prior studies found that HC with colonization was strongly associated with neonatal sepsis.³ Several clinical signs of chorioamnionitis were associated with HC, including amniotic fluid color and maternal temperature ⩾99 °F. Hygiene and sanitation as measured by the type of latrine at home was a risk factor associated with chorioamnionitis. Some studies have found term HC to be from a non-infectious cause.²²

Overall, clinical signs of chorioamnionitis had a low positive predictive value for HC. The poor association between clinical chorioamnionitis and HC suggests that women with chorioamnionitis may be asymptomatic or may indicate non-infectious causes of inflammation.²² Microbial chorioamnionitis diagnosed by placental cultures also had a low positive predictive value for HC. The swab sample between the chorion and amnion provides a sterile site, however it is a difficult site to collect and detect bacteria leading to an underestimation of chorioamnionitis. Misclassification of the microbial chorioamnionitis outcome may bias our results towards the null.

This is the first study in Bangladesh to estimate the prevalence of chorioamnionitis with fetal involvement as defined by histologic classifications. In addition to histologic examinations, chorioamnionitis was defined using microbiological cultures and clinical signs, which allowed us to compare the results across the different definitions of chorioamnionitis. The study had several limitations. This was a facility-based study and results may not be generalizable to all pregnancy outcomes that occur at home. As the majority of women deliver at home in Dhaka, women who delivered in the facility may differ in their socioeconomic status, education or clinical acuity. We excluded women who presented in fetal distress or obstructed labor to avoid interfering with medical care. However, as fetal distress and prolonged labor may be strongly associated with chorioamnionitis, excluding this sample may bias our results. Women who were missing placental samples may be different from women with samples. The sample size was calculated to study the risk factors of early onset neonatal infection and was not powered to detect predictors of HC. Given the small sample size, there was a limited number of parameters to test. A larger study is needed to assess the variables associated with chorioamnionitis-related adverse outcomes and to better identify women who would benefit from interventions such as intrapartum antibiotic prophylaxis.

The prevalence of HC is high in Dhaka, Bangladesh and well within the range of other reported studies. Diagnosis of histologic chorioamnionitis is only possible after examination of the placenta postpartum. Timely diagnosis of chorioamnionitis, especially during labor, would provide a window of opportunity for interventions. However, clinical signs of chorioamnionitis during labor are poor measures of chorioamnionitis, and currently no better clinical criteria are available to diagnose chorioamnionitis. Additional studies are needed to understand the causes of HC and its relationship with infections (clinical and microbiologic data). Future studies may need to include additional microbiologic and histologic data to better understand the relationship between risk factors and chorioamnionitis, and develop strategies aimed at preventing chorioamnionitis.

More than half of the early stillbirths and a smaller but significant component of the late preterm and term stillbirths globally are associated with HC.²³ However, there are few strategies that have been consistently shown to reduce chorioamnionitis and the preterm births and stillbirths associated with this condition. Reducing chorioamnionitis should be the focus of a major research effort.

As many as 42% of deaths in the first week of life may occur from infections.²⁴ The causes of neonatal infection, the risk factors associated with them and their modes of transmission remain poorly understood in developing countries, delaying the development of effective interventions to prevent and treat neonatal infections. During labor, mothers with clinical chorioamnionitis are two times more likely to have newborns with sepsis.³ Women with elevated temperatures, cloudy or discolored amniotic fluid during delivery may benefit from closer monitoring and treatment with antibiotics. Future studies include developing a prediction model to identify women who are at high risk of developing chorioamnionitis. In high-income countries, women in labor with fever are treated with antibiotics, which has reduced the incidence of Group B Streptococcus early onset neonatal sepsis.²⁵ In settings where the ability to detect and treat early neonatal sepsis is limited and the case fatality from maternal and neonatal infections is high, an implementation trial testing the effect of antibiotics during labor among women with clinical signs of sepsis is needed.