Data source, inclusion criteria and study setting
BIRDY is a multi-center cohort study launched to address the lack of epidemiological data concerning drug-resistant neonatal and infantile bacterial infections in three low-income countries (Cambodia, Madagascar and Senegal) (16). Nested within the Birdy study, the present study consecutively recruited pregnant women in Guédiawaye (suburban neighborhood in Dakar) and Sokone (rural area near the Gambian border) primary health centers during their third trimester of pregnancy, from October 2013 to September 2018. In the BIRDY study, healthcare workers interviewed women using a standardized questionnaire and collected from them a vaginal swab for streptococcus B (GBS). In addition, in our study, we also screened women for hepatitis B, Toxoplasma gondii and Rubella on blood sample and for vaginal candidiasis and BV on vaginal swab. All samples were transported in coolers to Pasteur Institute of Dakar laboratory within 24 hours of collection, then stored in refrigerators until processed. Healthcare workers (nurses and midwives) were trained for this study.
Data collection
Collected variables were as follows: (i) sociodemographic factors: age, marital status, education status (“formal education” was defined as women with at least primary education), sanitation type (indoor latrines and latrines with flushing water were considered to be improved sanitation facilities, whereas outdoor latrines without flushing water were designated unimproved sanitation facilities (17)); (ii) active smoking; (iii) nutritional status (estimated by mid-upper arm circumference (18)); (iv) obstetric history: gravidity, history of stillbirth; (v) pregnancy follow-up: number of prenatal visits (adequate follow-up defined as at least three prenatal visits at recruitment, according to WHO recommendations (19)), and intermittent preventive treatment of malaria in pregnancy with sulfadoxine-pyrimethamine (SP-IPTp) (defined as an intake of at least one dose during pregnancy).
Samples analyses
After microscopic examination and Gram’s coloration, isolates were plated onto selective growth medium, Granada Medium (Becton Dickinson) for group B Streptococcus, and CHROMagarTM Candida for yeast isolation) and incubated for 24–48 hours at 37 °C in 5% CO2. Vaginal flora were classified into four categories, based on microscopic examination and Gram’s coloration of the vaginal smear : “Type I” Döderlein flora, “Type II” majority of Lactobacillus spp. associated with few bacteria, “Type III” minority of Lactobacillus spp., “Type IV” no Lactobacillus spp. With knowledge of the microscopic examination and culture results, the bacteriologist could diagnose BV. In our study we defined BV as a “Type IV” vaginal flora.
Blood samples were screened for HBs antigen, Toxoplasma gondii and Rubella antibody immunoglobulin (Ig) G by enzyme immuno assay techniques (chimiluminescence Abbott Architect). All analyses were performed in the Biomedical Laboratory of the Pasteur institute in Dakar.
Statistical methods
Continuous variables were expressed as median with interquartile range (IQR); discrete variables were expressed as percentage with 95% confidence interval (CI). BV-positive and BV-negative groups were compared using χ2 test or Fisher’s exact test for dichotomous variables and Student’s t-test or Wilcoxon rank-sum test for continuous variables.
All variables associated with BV in univariate analysis (p < 0.25) were then included in a backward stepwise logistic regression. Because rural/suburban setting was not associated with BV in univariate analysis (p > 0.25), it was not included in the final model. Interactions were assessed between age and gravidity. A p-value ≤ 0.05 was considered statistically significant. For univariate and multivariate analyses, continuous variables were expressed as dichotomous variables using either the median (median age was 28), or a clinically relevant threshold (in the literature, primigravidity and multigravidity are usually used as the threshold for gravidity (20, 21)). Data were analyzed using STATA Software Version 15.1 (Stata Corporation, College Station, Texas, USA).
Ethics, data protection and confidentiality
The BIRDY protocol was approved by the relevant national ethics committees for health research of Senegal and France. Women were included after receiving information about the project, agreeing to providing biological samples, and signing an informed consent form. The BIRDY data collection has been declared to the Commission Nationale de l’Informatique et des Libertés (CNIL – French national data protection authority), in accordance with French law.