Study design, setting and population

This case-control study will be conducted in Tu Du Hospital in Ho Chi Minh City located in the Southern region of Vietnam. Tu Du Hospital is a specialised obstetrics and gynaecological tertiary referral centre and one of the largest hospitals in Asia, with approximately 55,000 births in 2022.

Inclusion criteria—cases

Cases are defined as pregnant or recently pregnant women with sepsis recognised between the onset of labour and/or time of delivery/cessation of pregnancy for up to 42 days post-partum. A diagnosis of sepsis will require the woman to have:

• A suspected or confirmed infection being treated with intravenous antibiotics
• An Obstetrically Modified Sequential Organ Failure Assessment (SOFA) Score of ≥ 2 (Table 1)
• Requested cultures of any bodily fluid, including swab specimens

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Obstetrically modified SOFA score [15].

https://doi.org/10.1371/journal.pone.0305411.t001

Inclusion criteria—controls

Controls will include pregnant or recently pregnant women who are up to 42 days post-partum who do not have sepsis.

Controls will be matched by age (within 2-year age bands), location (urban vs rural), mode of delivery (normal vaginal delivery or caesarean section), parity (primiparous vs multiparous) and gestational age rounded to the closest week (where possible). For each case, two controls will be randomly selected to ensure the comparability of the controls to the general obstetric population.

Case and control exclusion criteria

1. Women aged < 18 years of age
2. Women who are < 28 weeks gestation
3. Women who have an uncomplicated localised or chronic infection
4. Women who have an infection with SARS-CoV-2 (COVID-19)

Screening

Women with a suspected or confirmed infection will be assessed for inclusion into the study by screening for possible markers of organ failure using the Obstetrically Modified quick Sequential Organ Failure Assessment (OMqSOFA) score [15]. The OMqSOFA scores one point for each of the following criteria (respiratory rate ≥ 25 breaths per minute; systolic blood pressure <90mmHg; any non-alert mental status). A score of 2 or more indicates a higher risk of infection and sepsis and prompts further screening.

For women who have an OMqSOFA score of ≥ 2 points, organ failure screening will be conducted using the Obstetrically Modified SOFA score (Table 1). A diagnosis of maternal sepsis will require the woman to have a suspected or confirmed infection plus an Obstetrically Modified SOFA score of ≥ 2. The Obstetrically Modified SOFA score has been used in accordance with recommendations from the Society of Obstetric Medicine in Australia and New Zealand [15]. In order to demonstrate evidence of organ dysfunction a score of ≥ 2 is required [15]. To simplify screening, we removed all individual organ failure scores of 3 and 4 from the screening sheet, as an overall score of ≥ 2 across all organ systems is required for study inclusion. Modified SOFA parameters for pregnant or recently pregnant woman include a reduction in the creatinine threshold, as serum creatinine levels are significantly reduced in pregnancy with normal ranges of 35–80 μmol/L and a simplified neurological assessment as Glasgow Coma Score is not routinely assessed on maternity wards. For this study, we also modified the respiratory SOFA score component, substituting the PaO₂/FiO₂ with the SPO₂/FiO₂ ratio [16], due to lack of availability of arterial blood gas monitoring in the hospital.

Outcomes

The primary outcome is to identify risk factors associated with the development of maternal sepsis. Secondary outcomes include the frequency of adverse outcomes due to maternal sepsis, including death and treatment with invasive organ support, bacterial etiology and their AMR profiles comparing cases and controls. Timely and appropriate management of maternal sepsis including the time to blood cultures, antibiotic administration, primary antimicrobial use, dose and duration of antimicrobial treatment, perinatal outcomes including complications in pregnancy and neonatal outcomes will also be assessed.

Clinical samples will be collected and subjected to microbiological culture in accordance with routine clinical care and procedures. Where clinically indicated for patient management vaginal swabs will be collected. Bacterial identification and AMR testing will be performed in-house using the BD Phoenix™ automated identification and susceptibility testing system. Bacterial isolates identified from sepsis cases will undergo bacterial and phenotypic confirmatory tests and molecular investigation (AMR gene typing, genome sequencing). For controls, two clinical samples in addition to standard care (vaginal and rectal swabs) will be taken. Samples from controls will undergo bacterial identification and AMR testing. In the control population we will assess colonising bacteria in vaginal and rectal swabs e.g. Group A, B, C and G Streptococcus, Escherichia coli, Enterobacter spp, enterococcus, Chlamydia trachomatis and staphylococcus aureus. These colonising bacteria will be subjected to whole-genome sequencing. To better understand what causes maternal sepsis we will assess the genetic relatedness of the same bacterial species identified from cases and controls.

Statistical analysis plan

Data from the case-control study will be cleaned and a descriptive analysis conducted. Approximately symmetric continuous variables will be reported as mean ± standard deviation. Skewed continuous variables will be reported as median and interquartile range (IQR). Categorical data will be reported as proportions (%). Outcomes will be censored at the end of the postpartum period (42 days) or at discharge, transfer to another treating facility or at time of death, whichever comes first.

Risk factors associated with sepsis will be compared between cases and controls. Risk factors will include sociodemographic, pregnancy, labour and delivery characteristics. Due to the uneven case-control distribution, and also due to the matched allocations, conditional logistic regression models will be used to compare the actions of putative risk factors between the cases and controls. We will assess individual risk factors, including but not limited to the mothers baseline sociodemographic and physiological pregnancy characteristics. Statistical adjustment will be made for potential confounders such as the presence of anaemia and other complications of pregnancy (hypertensive disorders and gestational diabetes) as well as matching variables. A two-sided p-value of <0.05 will be considered statistically significant. Analysis will be conducted using Stata software.

Bacterial isolates and antimicrobial resistance profiles identified via microbiological cultures will be reported and compared between cases and controls. Bacterial and phenotypic confirmatory tests, molecular investigation (AMR gene typing, whole-genome sequencing) of major infecting organisms will be compared to commensal organisms collected from controls.

We chose to study 100 cases and 200 controls primarily because of the anticipated availability of suitable cases and controls within the limitations of the research environment. With this sampling allocation we will have 80% power, at the 5% significance level and using a two-sided test, to detect an odds ratio of 3, or less, whenever the risk factor has a prevalence of 8%, or more.

Data management

Data management will be undertaken by the Oxford University Clinical Research Unit (OUCRU) in Ho Chi Minh City. Data will be collected on paper-based Case Report Forms at site and transcribed to the electronic study database (CliRes data management system) by the study coordinators and hosted at OUCRU. A separate master log of enrolments containing identifying data will be stored securely in a password protected excel file by the study coordinator on site. This will be referred to by personnel at the study site to address questions of data integrity or to verify data, as requested by the coordinating centre.

A copy of the Case Report Form is provided in the Supporting Information. Data collection is broadly grouped into the following categories; women’s sociodemographic characteristics, relevant medical history, labour and delivery details, including complications, clinical identification of infection/sepsis and treatment, laboratory investigations, ICU data collection for those admitted to ICU with infection/sepsis, length of stay outcomes, Edinburgh depression score at 42 days and infant outcomes.

Ethical considerations

The study is approved by Oxford Tropical Research Ethics Committee (approval number: 501–22) and the Ethics Committee of the Tu Du hospital (approval number: 2979/BVTD-HDDD). This is a low-risk research project, where the only foreseeable risk is one of discomfort. The site Principal Investigator will seek written informed consent from patients (or a delegated legal representative where appropriate) prior to enrolment into the study. All patient information sheets and consent forms will be written in Vietnamese.

Study status and timeline

A planned one year recruitment period commenced in September 2023, with termination of recruitment after 100 cases and 200 controls are enrolled.