Publications

Abstract

Introduction: Preterm babies are a vulnerable population that experience significant short and long-term morbidity. Rehospitalisations constitute an important, potentially modifiable adverse event in this population. Improving the ability of clinicians to identify those patients at the greatest risk of rehospitalisation has the potential to improve outcomes and reduce costs. Machine-learning algorithms can provide potentially advantageous methods of prediction compared to conventional approaches like logistic regression. 

Objective: To compare two machine-learning methods (least absolute shrinkage and selection operator (LASSO) and random forest) to expert-opinion driven logistic regression modelling for predicting unplanned rehospitalisation within 30 days in a large French cohort of preterm babies. 

Design, Setting and Participants: This study used data derived exclusively from the population-based prospective cohort study of French preterm babies, EPIPAGE 2. Only those babies discharged home alive and whose parents completed the 1-year survey were eligible for inclusion in our study. All predictive models used a binary outcome, denoting a baby’s status for an unplanned rehospitalisation within 30 days of discharge. Predictors included those quantifying clinical, treatment, maternal and socio-demographic factors. The predictive abilities of models constructed using LASSO and random forest algorithms were compared with a traditional logistic regression model. The logistic regression model comprised 10 predictors, selected by expert clinicians, while the LASSO and random forest included 75 predictors. Performance measures were derived using 10-fold cross-validation. Performance was quantified using area under the receiver operator characteristic curve, sensitivity, specificity, Tjur’s coefficient of determination and calibration measures. 

Results: The rate of 30-day unplanned rehospitalisation in the eligible population used to construct the models was 9.1% (95% CI 8.2-10.1) (350/3,841). The random forest model demonstrated both an improved AUROC (0.65; 95% CI 0.59-0.7; p = 0.03) and specificity vs. logistic regression (AUROC 0.57; 95% CI 0.51-0.62, p = 0.04). The LASSO performed similarly (AUROC 0.59; 95% CI 0.53-0.65; p = 0.68) to logistic regression. 

Conclusions: Compared to an expert-specified logistic regression model, random forest offered improved prediction of 30-day unplanned rehospitalisation in preterm babies. However, all models offered relatively low levels of predictive ability, regardless of modelling method.

Abstract

In high-income countries, breastfeeding has been shown to be positively associated with socioeconomic position. However, less is known about breastfeeding practices and their associated factors among extremely disadvantaged populations. We aimed to assess the associations of cultural origins and socioeconomic factors with any breastfeeding initiation and duration in homeless families. We analyzed data from 456 children aged 6 months to 5 years from the cross-sectional ENFAMS survey, conducted in 2013 among a random sample of homeless families in shelters in the Greater Paris area. Data were collected by bilingual interviewers in 17 languages. Four nested multivariable robust Poisson regression models were run in a hierarchical framework to determine the factors associated with breastfeeding initiation and with any breastfeeding for 6 months or more. Most of the children (86.0%) had previously been or were currently being breastfed at the time of the survey; 58.9% were fed with breast milk ≥6 months. A higher maternal age and African origin were positively associated with breastfeeding ≥6 months, although the relation to the region of origin was moderated by education level. Migration to escape war, unrest or other violence and the child’s birth in France were inversely associated with breastfeeding ≥6 months. Any breastfeeding by these homeless mothers seems influenced predominantly by their cultural origin and complicated by a difficult migration trajectory. The possible influence of poor material circumstances and cumulative hardship should encourage interventions targeted at homeless mothers that emphasize social/family support with a commitment to improving the family’s living conditions.

Abstract

Objective: To assess twins’ perinatal outcomes after planned vaginal delivery according to chorionicity.

Methods: JUMODA (JUmeaux MOde D’Accouchement) was a national prospective population-based cohort study of twin deliveries performed in 176 maternity units in France from 02/2014 to 03/2015. In this planned secondary analysis, we compared perinatal outcomes of twins born at or after 32 weeks of gestation after planned vaginal delivery when the first twin was in cephalic presentation according to chorionicity. To select a population with well recognized indications of planned vaginal delivery, we applied the same exclusion criteria as in the international randomized trial, the Twin Birth Study. Finally, twin-to-twin transfusion syndromes and twin anemia-polycythemia sequences defined complicated monochorionic twin pregnancies and were excluded. The primary outcome was a composite of intrapartum mortality and neonatal morbidity and mortality. We used multivariable logistic regression models to control for potential confounders. We conducted subgroup analyses according to birth order, first or second twin, and gestational age at delivery, before or after 37 weeks of gestation.

Results: Among 3873 women at or after 32 weeks with a cephalic first twin and a planned vaginal delivery according to the Twin Birth Study criteria, 729 (18.8%) were uncomplicated monochorionic and 3144 (81.2%) dichorionic twin pregnancies. Composite intrapartum mortality and neonatal morbidity and mortality did not differ between uncomplicated monochorionic (27/1458 (1.9%)) and dichorionic (107/6288 (1.7%)) twin pregnancies (aRR 1.07, 95% CI 0.66-1.75). No significant difference between groups was shown in subgroup analyses according to birth order and gestational age at delivery.

Conclusions: In case of planned vaginal delivery at or after 32 weeks of gestation when the first twin is in cephalic presentation, uncomplicated monochorionic twin pregnancies are not associated with higher composite intrapartum mortality and neonatal morbidity and mortality than dichorionic twin pregnancies. This article is protected by copyright. All rights reserved.

Abstract

This study aims to describe the incidence of acute respiratory infections (ARI) during the first year in infants born before 32 weeks’ gestation, and to analyze and study the risk factors as well as factors associated with oxygen requirement among infants with an ARI, in the palivizumab era. This study included 2571 infants from a nationwide French population-based cohort (Epipage 2). ARI at 1-year corrected age was identified by parental questionnaires. Risk and severity factors included those already known, and detailed information about neonatal morbidities. ARI occurred in 52.2% (n = 1349) of infants. Oxygen therapy was used in 33.2% (n = 391) of infants with an ARI. Risk factors for AII were male sex, bronchopulmonary dysplasia, presence of siblings at home, and childcare in the community together with incomplete treatment palivizumab. Mechanical ventilation in the neonatal period, bronchopulmonary dysplasia, and discharge between October and March were associated with more frequent oxygen requirement. No other factors describing neonatal morbidities were associated with risk of ARI or oxygen requirement.Conclusion: ARIs are still very common during the first year of life of very preterm children, and oxygen therapy is frequently needed. Educational strategies are needed in all families with a very preterm infant. What is Known: • Acute respiratory infections (ARIs) are the first cause of rehospitalizations in preterm children, with bronchopulmonary dysplasia being the main risk factor. • Palivizumab prophylaxis has proven its effect against severe RSV infections, but it is not universal. What is New: • No factor describing neonatal morbidity, except BPD, was associated with ARI occurrence or severity. • BPD and discharge during RSV season were the only factors associated with O₂ requirement during ARI.

Abstract

Introduction: Fetal bradycardia due to sentinel events such as placental abruption, cord prolapse or uterine rupture is associated with an increased risk of acidemia at birth. In the absence of a sentinel event, data regarding neonatal prognosis are scarce, and it seems plausible that the depth of bradycardia might be associated with an increased risk of acidosis at birth. The objective was to determine whether the depth of bradycardia is associated with a higher risk of umbilical artery acidemia at birth in term singleton pregnancies requiring cesarean delivery during labor.

Material and methods: A retrospective comparative study of all cesarean deliveries for bradycardia in an academic tertiary center in the 6-year period of 2013-2018, among term singleton pregnancies. Bradycardia associated with a sentinel event such as placental abruption, cord prolapse or uterine rupture, were excluded. The nadir of the bradycardia was defined as the lowest fetal heart rate baseline lasting at least 3 minutes during bradycardia. Women who delivered an infant with an umbilical pH at birth <7.00 (acidosis group) were compared with women who delivered an infant with an umbilical pH at birth ≥7.00 (non-acidosis group).

Results: Among 111 eligible cases, 32 women in the acidosis group were compared with 79 in the non-acidosis group. The median nadir of the bradycardia was lower in the acidosis than in the non-acidosis group (60 bpm, interquartile range [56-65] vs 70 [60-76], P < .01). A bradycardia nadir <60 bpm emerged as the optimal threshold for predicting acidemia and was more frequently observed in the acidosis than in the non-acidosis group (10 [31%] vs 10 [13%], P = .02). In the multivariable analysis, a nadir <60 bpm was independently associated with an umbilical artery pH <7.00 (adjusted OR 3.16, 95% CI 1.10-9.04).

Conclusions: A bradycardia nadir <60 bpm was associated with a tripled risk of umbilical artery acidemia at birth.

No abstract available

Abstract

In this study, for the first time we report on a comprehensive overview of different strategies to hyphenate droplet-based sample handling and preparation with electrophoretic separation in different formats (i.e. microchip and capillary electrophoresis). Droplet-interfaced electrophoresis is an emerging technique in which micro/nanometric droplets are used as a bridge and carrier of target analytes between sample treatment and electrokinetic separation steps, thus being expected to overcome the challenges of working dimension mismatch and low degree of module integration. This review covers all works on this topic from 2006 (the year of the first communication) up to 2020, with focus being given to three principal interfacing strategies, including droplets in immiscible phases, digital microfluidics with electrowetting-on-dielectric principle and inkjet droplet generation. Different instrumental developments for such purpose, the viewpoints on pros and cons of these designs as well as application demonstrations of droplet-interfaced electrokinetic strategies are discussed.

PLoS One. 2021 Jan 22;16(1):e0244933. doi: 10.1371/journal.pone.0244933. eCollection 2021.

Abstract

Background: This study aimed to compare incidence, management and outcomes of women transfused their blood volume or more within 24 hours during pregnancy or following childbirth.

Methods: Combined analysis of individual patient data, prospectively collected in six international population-based studies (France, United Kingdom, Italy, Australia, the Netherlands and Denmark). Massive transfusion in major obstetric haemorrhage was defined as transfusion of eight or more units of red blood cells within 24 hours in a pregnant or postpartum woman. Causes, management and outcomes of women with massive transfusion were compared across countries using descriptive statistics.

Findings: The incidence of massive transfusion was approximately 21 women per 100,000 maternities for the United Kingdom, Australia and Italy; by contrast Denmark, the Netherlands and France had incidences of 82, 66 and 69 per 100,000 maternities, respectively. There was large variation in obstetric and haematological management across countries. Fibrinogen products were used in 86% of women in Australia, while the Netherlands and Italy reported lower use at 35-37% of women. Tranexamic acid was used in 75% of women in the Netherlands, but in less than half of women in the UK, Australia and Italy. In all countries, women received large quantities of colloid/crystalloid fluids during resuscitation (>3·5 litres). There was large variation in the use of compression sutures, embolisation and hysterectomy across countries. There was no difference in maternal mortality; however, variable proportions of women had cardiac arrests, renal failure and thrombotic events from 0-16%.

Interpretation: There was considerable variation in the incidence of massive transfusion associated with major obstetric haemorrhage across six high-income countries. There were also large disparities in both transfusion and obstetric management between these countries. There is a requirement for detailed evaluation of evidence underlying current guidance. Furthermore, cross-country comparison may empower countries to reference their clinical care against that of other countries.

Abstract

Background: Adverse birth outcomes are major causes of morbidity and mortality during childhood and associate with a higher risk of noncommunicable diseases in adult life. Maternal periconception and antenatal nutrition, mostly focusing on single nutrients or foods, has been shown to influence infant birth outcomes. However, evidence on whole diet that considers complex nutrient and food interaction is rare and conflicting. We aim to elucidate the influence of whole-diet maternal dietary inflammatory potential and quality during periconceptional and antenatal periods on birth outcomes.

Methods and findings: We harmonized and pooled individual participant data (IPD) from up to 24,861 mother-child pairs in 7 European mother-offspring cohorts [cohort name, country (recruitment dates): ALSPAC, UK (1 April 1991 to 31 December 1992); EDEN, France (27 January 2003 to 6 March 2006); Generation R, the Netherlands (1 April 2002 to 31 January 2006); Lifeways, Ireland (2 October 2001 to 4 April 2003); REPRO_PL, Poland (18 September 2007 to 16 December 2011); ROLO, Ireland (1 January 2007 to 1 January 2011); SWS, United Kingdom (6 April 1998 to 17 December 2002)]. Maternal diets were assessed preconceptionally (n = 2 cohorts) and antenatally (n = 7 cohorts). Maternal dietary inflammatory potential and quality were ranked using the energy-adjusted Dietary Inflammatory Index (E-DII) and Dietary Approaches to Stop Hypertension (DASH) index, respectively. Primary outcomes were birth weight and gestational age at birth. Adverse birth outcomes, i.e., low birth weight (LBW), macrosomia, small-for-gestational-age (SGA), large-for-gestational-age (LGA), preterm and postterm births were defined according to standard clinical cutoffs. Associations of maternal E-DII and DASH scores with infant birth outcomes were assessed using cohort-specific multivariable regression analyses (adjusted for confounders including maternal education, ethnicity, prepregnancy body mass index (BMI), maternal height, parity, cigarettes smoking, and alcohol consumption), with subsequent random-effects meta-analyses. Overall, the study mothers had a mean ± SD age of 29.5 ± 4.9 y at delivery and a mean BMI of 23.3 ± 4.2 kg/m2. Higher pregnancy DASH score (higher dietary quality) was associated with higher birth weight [β(95% CI) = 18.5(5.7, 31.3) g per 1-SD higher DASH score; P value = 0.005] and head circumference [0.03(0.01, 0.06) cm; P value = 0.004], longer birth length [0.05(0.01, 0.10) cm; P value = 0.010], and lower risk of delivering LBW [odds ratio (OR) (95% CI) = 0.89(0.82, 0.95); P value = 0.001] and SGA [0.87(0.82, 0.94); P value < 0.001] infants. Higher maternal prepregnancy E-DII score (more pro-inflammatory diet) was associated with lower birth weight [β(95% CI) = -18.7(-34.8, -2.6) g per 1-SD higher E-DII score; P value = 0.023] and shorter birth length [-0.07(-0.14, -0.01) cm; P value = 0.031], whereas higher pregnancy E-DII score was associated with a shorter birth length [-0.06(-0.10, -0.01) cm; P value = 0.026] and higher risk of SGA [OR(95% CI) = 1.18(1.11, 1.26); P value < 0.001]. In male, but not female, infants higher maternal prepregnancy E-DII was associated with lower birth weight and head circumference, shorter birth length, and higher risk of SGA (P-for-sex-interaction = 0.029, 0.059, 0.104, and 0.075, respectively). No consistent associations were observed for maternal E-DII and DASH scores with gestational age, preterm and postterm birth, or macrosomia and LGA. Limitations of this study were that self-reported dietary data might have increased nondifferential measurement error and that causality cannot be claimed definitely with observational design.

Conclusions: In this cohort study, we observed that maternal diet that is of low quality and high inflammatory potential is associated with lower offspring birth size and higher risk of offspring being born SGA in this multicenter meta-analysis using harmonized IPD. Improving overall maternal dietary pattern based on predefined criteria may optimize fetal growth and avert substantial healthcare burden associated with adverse birth outcomes.

Abstract

Group B Streptococcus (GBS) is the major cause of human neonatal infections. A single clone, designated CC17-GBS, accounts for more than 80% of meningitis cases, the most severe form of the infection. However, the events allowing blood-borne GBS to penetrate the brain remain largely elusive. In this study, we identified the host transmembrane receptors α5β1 and αvβ3 integrins as the ligands of Srr2, a major CC17-GBS specific adhesin. Two motifs located in the binding region of Srr2 were responsible for the interaction between CC17-GBS and these integrins. We demonstrated, in a blood-brain barrier cellular model, that both integrins contributed to the adhesion and internalization of CC17-GBS. Strikingly, both integrins were overexpressed during the post-natal period in the brain vessels of the blood-brain and blood-cerebrospinal fluid barriers and contributed to the juvenile susceptibility to CC17-meningitis. Finally, blocking these integrins decreased CC17-GBS crossing into the juvenile mice central nervous system in an in vivo model of meningitis.Our study demonstrates that CC17-GBS exploits integrins for crossing the brain vessels leading to meningitis. Importantly, it provides host molecular insights into neonate’s susceptibility to CC17-GBS meningitis, thereby opening new perspectives for therapeutic and prevention strategies of GBS-elicited meningitis.