First, it is important to emphasize that the BFHI-accredited maternity units included in our assessment are not evenly distributed across the country; in fact, the majority are concentrated in northern France. This meant that a key part of this evaluation would involve taking into account the inequalities that exist across different regions.

We used data from the 2010, 2016, and 2021 National Perinatal Surveys in France, each of which included more than 10,000 births (13,075, 10,919, and 10,209, respectively). The BFHI maternity units included were the 49 already accredited in mainland France. Those in the process of obtaining certification were considered non-accredited (20 in 2010, 40 in 2016, and 31 in 2021). The online database of the French National Institute of Statistics and Economic Studies (INSEE) was used to obtain data on the French departments and their characteristics.

We used a series of five mixed-effects multinomial regression models, with adjustment covariates added gradually: 1) maternity unit with BFHI status (yes/no) and survey year; 2) characteristics of the infant and the mother; 3) characteristics of the maternity unit; 4) characteristics of the French department and the spatial random effect; 5) interaction terms. In model 4, accounting for the spatial random effect at the departmental level allowed us to account for variations in breastfeeding rates across departments and the correlation between breastfeeding rates in neighboring departments. By further adjusting the model based on variables related to the French department and the random effect, the positive impact of the BFHI initiative on breastfeeding rates became clearer. This can be explained by the wide variations in breastfeeding rates across regions in mainland France and by the fact that maternity units with BFHI accreditation are unevenly distributed throughout the country. This meant that it was essential to adjust the spatial structure of the data to isolate the effect of the BFHI. In particular, the spatial random effect accounted for more variability in the model than the selected departmental variables (percentage of the population who are migrants, percentage with a university education).

In BFHI-accredited maternity units, the crude rates of exclusive breastfeeding were higher than in non-accredited maternity units, while the crude rates of mixed feeding were lower than in non-accredited maternity units. The same applies to rates adjusted for the variables mentioned above, in all models. However, the crude rate of breastfeeding of any kind was lower in accredited maternity units; yet, when adjusted to account for individual covariates (maternal age, educational level, country of birth, marital status, circumstances in late pregnancy, household income, parity, body mass index, mode of delivery, time between delivery and interview, birth weight, gestational age, neonatal transfer), the rate was higher. This difference indicates that there was variation between mothers giving birth in BFHI-accredited maternity units and those giving birth in non-accredited maternity units in terms of factors associated with breastfeeding.

In our study, low birth weight babies were breastfed less than average weight babies in both BFHI-accredited and non-accredited maternity units. This is consistent with previous studies conducted in France, Spain, and Brazil (1–3). We found that the increase in exclusive breastfeeding rates and the decrease in mixed feeding rates associated with BFHI accreditation had a more significant impact on low birth weight newborns, helping to narrow the gap between this vulnerable group and normal weight babies.

BFHI accreditation had a positive impact on breastfeeding rates among mothers with a medium to high level of education (two or more years of higher education). For mothers with no education (no schooling at all) or a primary education—accounting for around 2% of the study sample—the rate of exclusive breastfeeding was lower in accredited maternity units, although with a wide confidence interval. These mothers are part of a minority group with a specific profile that needs to be analyzed in greater depth.

Similarly, a cross-sectional study conducted in Belgium (4) showed that BFHI accreditation improved rates of initiation of exclusive breastfeeding mainly in subgroups of mothers who were already more likely to breastfeed (particularly those with a university education).

France has one of the lowest breastfeeding rates in Europe, with significant disparities across different regions and socioeconomic groups. Our study results show that BFHI-accredited maternity units do have a positive impact on the rate of exclusive breastfeeding. It also found a lower rate of mixed feeding compared to hospitals without accreditation. We have demonstrated for the first time that this positive impact is even greater among low birth weight infants. Our research has shown that implementing the BFHI can improve breastfeeding practices in maternity units, with a particular focus on vulnerable groups, such as mothers of low birth weight infants, who tend to have lower breastfeeding rates.