Data for this analysis come from the National Perinatal Surveys conducted in 2010 and 2016. These cross-sectional surveys collected data from all births at a gestational age of at least 22 weeks or with a birth weight of at least 500 grams during a single week, in all maternity units in metropolitan France and the overseas territories (9).
The data came both from the medical records, mainly on delivery, and the child's condition at birth, and a postpartum face-to-face questionnaires with parturients before their discharge. This interview focused on their prenatal care and their socioeconomic characteristics.
Women who did not speak French, were younger than age 18 and those who experienced stillbirth or termination of pregnancy were excluded from the study.
In 2010, 13 894 women with 14 142 babies were included, and in 2016, the sample comprised 15 187 women with 15 418 babies. The two survey samples were first compared to check the absence of major heterogeneity and to analyze the evolution in Down syndrome screening uptake. The two samples were then merged and analyzed together to improve the statistical power of the analysis.
We selected women who had lived in metropolitan France for at least one year when they gave birth and whose place of birth was known: 13 507 women in 2010 and 11 137 women in 2016.
Maternal characteristics studied
The mother's place of birth was classified in five categories: France, other European countries, North Africa, Sub-Saharan Africa, and the rest of the world.
In both surveys, the questionnaire used as the basis for the face-to-face interview included the following questions about Down syndrome screening: "Did you have fetal nuchal translucency measured at your first-trimester ultrasound to learn your risk of Down syndrome?" and "Did you have a blood test to learn your risk for Down syndrome (serum markers)?" The possible responses were "Yes," "No," or "I don't know". If the answer to the latter question was negative, the reasons could be: refusal; amniocentesis or chorionic villus sampling from the start; tests were not offered; late initiation of prenatal care or no prenatal care or prenatal care abroad; another reason, or "don't know".
To analyze access to Down syndrome blood screening test, we used a composite indicator built from these answers defined by Grupposo et al. (8). Accordingly, we considered that women had an opportunity to make an informed choice about Down syndrome screening if they had undergone or refused a serum marker assay or had had an initial amniocentesis or chorionic villus sampling. We considered that women had not had this opportunity when they did not know if they had had the assay or if it had not been performed for one of the following reasons: not offered, because of late antenatal care, for another reason, or for an unknown reason (Figure 1).
Adequate prenatal care was defined in accordance with French clinical practice guidelines (10) by at least 8 antenatal visits and at least 3 prenatal ultrasounds for term births, and according to gestational age at birth for preterm births (9).
The women’s characteristics were previously described in the National Perinatal Survey report (11). The other covariables considered were age, parity, and educational level.
Statistical analyses
Pearson's Chi-square tests were used to compare rates of Down syndrome screening and the opportunity to make an informed choice between 2010 and 2016, according to the mother's country of birth. Differences were considered significant when p was less than 0.05.
To study the association between the mothers' country of birth and rates of the different prenatal screening tests for Down syndrome, as well as the opportunity to make an informed choice about it, we used multivariate multinomial logistic regression models adjusted for the survey year, the individual factors (maternal age, parity, educational level, and prenatal care), and for the maternity ward characteristics (level of perinatal care: I, IIA, IIB, and III). The analysis was performed after fusion of the databases for 2010 and 2016.
For the analyzed sample, the rate of missing data per variable ranged only from 0.3% to 0.8% and thus did not justify an analysis with multiple imputations. The results presented are therefore those for the complete cases only.
The statistical analyses were performed with Stata 13 software.