Working with what you have: How the East Africa Preterm Birth Initiative used gestational age data from facility maternity registers

Respiratory distress syndrome prediction at birth by optical skin maturity assessment and machine learning models for limited-resource settings: a development and validation study

Background

A handheld optical device was developed to evaluate a newborn's skin maturity by assessing the photobiological properties of the tissue and processing it with other variables to predict early neonatal prognosis related to prematurity. This study assessed the device's ability to predict respiratory distress syndrome (RDS).

Methods

To assess the device's utility we enrolled newborns at childbirth in six urban perinatal centers from two multicenter single-blinded clinical trials. All newborns had inpatient follow-up until 72 h of life. We trained supervised machine learning models with data from 780 newborns in a Brazilian trial and provided external validation with data from 305 low-birth-weight newborns from another trial that assessed Brazilian and Mozambican newborns. The index test measured skin optical reflection with an optical sensor and adjusted acquired values with clinical variables such as birth weight and prenatal corticoid exposition for lung maturity, maternal diabetes, and hypertensive disturbances. The performance of the models was evaluated using intrasample k-parts cross-validation and external validation in an independent sample.

Results

Models adjusting three predictors (skin reflection, birth weight, and antenatal corticoid exposure) or five predictors had a similar performance, including or not maternal diabetes and hypertensive diseases. The best global accuracy was 89.7 (95% CI: 87.4 to 91.8, with a high sensitivity of 85.6% (80.2 to 90.0) and specificity of 91.3% (95% CI: 88.7 to 93.5). The test correctly discriminated RDS newborns in external validation, with 82.3% (95% CI: 77.5 to 86.4) accuracy. Our findings demonstrate a new way to assess a newborn's lung maturity, providing potential opportunities for earlier and more effective care.

Trial registration

RBR-3f5bm5 (online access: http://www.ensaiosclinicos.gov.br/rg/RBR-3f5bm5/), and RBR-33mjf (online access: https://ensaiosclinicos.gov.br/rg/RBR-33rnjf/).

Stillbirth rates and their determinants in a national maternity hospital in Phnom Penh, Cambodia in 2017-2020: a cross-sectional assessment with a nested case-control study

BACKGROUND

In Cambodia, stillbirths and their underlying factors have not been systematically studied. This study aimed to assess the proportion and trends in stillbirths between 2017 and 2020 in a large maternity referral hospital in the country and identify their key determinants to inform future prevention efforts.

METHODS

This was a retrospective cross-sectional analysis with a nested case-control study of women giving birth at the National Maternal and Child Health Centre (NMCHC) in Phnom Penh, 2017-2020. We calculated percentages of singleton births at ≥ 22 weeks' gestation resulting in stillbirth and annual stillbirth rates by timing: intrapartum (fresh) or antepartum (macerated). Multivariable logistic regression was used to explore factors associated with stillbirth, where cases were all women who gave birth to a singleton stillborn baby in the 4-year period. One singleton live birth immediately following each case served as an unmatched control. Multiple imputation was used to handle missing data for gestational age.

RESULTS

Between 2017 and 2020, 3.2% of singleton births ended in stillbirth (938/29,742). The stillbirth rate increased from 24.8 per 1000 births in 2017 to 38.1 per 1000 births in 2020, largely due to an increase in intrapartum stillbirth rates which rose from 18.8 to 27.4 per 1000 births in the same period. The case-control study included 938 cases (stillbirth) and 938 controls (livebirths). Factors independently associated with stillbirth were maternal age ≥ 35 years compared to < 20 years (aOR: 1.82, 95%CI: 1.39, 2.38), extreme (aOR: 3.29, 95%CI: 2.37, 4.55) or moderate (aOR: 2.45, 95%CI: 1.74, 3.46) prematurity compared with full term, and small-for-gestational age (SGA) (aOR: 2.32, 1.71, 3.14) compared to average size-for-age. Breech/transverse births had nearly four times greater odds of stillbirth (aOR: 3.84, 95%CI: 2.78, 5.29), while caesarean section reduced the odds by half compared with vaginal birth (aOR: 0.50, 95%CI: 0.39, 0.64). A history of abnormal vaginal discharge increased odds of stillbirth (aOR: 1.42, 95%CI: 1.11, 1.81) as did a history of stillbirth (aOR: 3.08, 95%CI: 1.5, 6.5).

CONCLUSIONS

Stillbirth prevention in this maternity referral hospital in Cambodia requires strengthening preterm birth detection and management of SGA, intrapartum care, monitoring women with stillbirth history, management of breech births, and further investigation of high-risk referral cases.

Adapting the preterm birth phenotyping framework to a low-resource, rural setting and applying it to births from Migori County in western Kenya

BACKGROUND

Preterm birth is the leading cause of neonatal and under-five mortality worldwide. It is a complex syndrome characterized by numerous etiologic pathways shaped by both maternal and fetal factors. To better understand preterm birth trends, the Global Alliance to Prevent Prematurity and Stillbirth published the preterm birth phenotyping framework in 2012 followed by an application of the model to a global dataset in 2015 by Barros, et al. Our objective was to adapt the preterm birth phenotyping framework to retrospective data from a low-resource, rural setting and then apply the adapted framework to a cohort of women from Migori, Kenya.

METHODS

This was a single centre, observational, retrospective chart review of eligible births from November 2015 - March 2017 at Migori County Referral Hospital. Adaptations were made to accommodate limited diagnostic capabilities and data accuracy concerns. Prevalence of the phenotyping conditions were calculated as well as odds of adverse outcomes.

RESULTS

Three hundred eighty-seven eligible births were included in our study. The largest phenotype group was none (no phenotype could be identified; 41.1%), followed by extrauterine infection (25.1%), and antepartum stillbirth (16.7%). Extrauterine infections included HIV (75.3%), urinary tract infections (24.7%), malaria (4.1%), syphilis (3.1%), and general infection (3.1%). Severe maternal condition was ranked fourth (15.6%) and included anaemia (69.5%), chronic respiratory distress (22.0%), chronic hypertension prior to pregnancy (5.1%), diabetes (3.4%), epilepsy (3.4%), and sickle cell disease (1.7%). Fetal anaemia cases were the most likely to transfer to the newborn unit (OR 5.1, 95% CI 0.8, 30.9) and fetal anomaly cases were the most likely to result in a pre-discharge mortality (OR 3.9, 95% CI 0.8, 19.2).

CONCLUSIONS

Using routine data sources allowed for a retrospective analysis of an existing dataset, requiring less time and fewer resources than a prospective study and demonstrating a feasible approach to preterm phenotyping for use in low-resource settings to inform local prevention strategies.

Gestational age assessed by optical skin reflection in low-birth-weight newborns: Applications in classification at birth

Introduction

A new medical device was previously developed to estimate gestational age (GA) at birth by processing a machine learning algorithm on the light scatter signal acquired on the newborn's skin. The study aims to validate GA calculated by the new device (test), comparing the result with the best available GA in newborns with low birth weight (LBW).

Methods

We conducted a multicenter, non-randomized, and single-blinded clinical trial in three urban referral centers for perinatal care in Brazil and Mozambique. LBW newborns with a GA over 24 weeks and weighing between 500 and 2,500 g were recruited in the first 24 h of life. All pregnancies had a GA calculated by obstetric ultrasound before 24 weeks or by reliable last menstrual period (LMP). The primary endpoint was the agreement between the GA calculated by the new device (test) and the best available clinical GA, with 95% confidence limits. In addition, we assessed the accuracy of using the test in the classification of preterm and SGA. Prematurity was childbirth before 37 gestational weeks. The growth standard curve was Intergrowth-21st, with the 10th percentile being the limit for classifying SGA.

Results

Among 305 evaluated newborns, 234 (76.7%) were premature, and 139 (45.6%) were SGA. The intraclass correlation coefficient between GA by the test and reference GA was 0.829 (95% CI: 0.785-0.863). However, the new device (test) underestimated the reference GA by an average of 2.8 days (95% limits of agreement: -40.6 to 31.2 days). Its use in classifying preterm or term newborns revealed an accuracy of 78.4% (95% CI: 73.3-81.6), with high sensitivity (96.2%; 95% CI: 92.8-98.2). The accuracy of classifying SGA newborns using GA calculated by the test was 62.3% (95% CI: 56.6-67.8).

Discussion

The new device (test) was able to assess GA at birth in LBW newborns, with a high agreement with the best available GA as a reference. The GA estimated by the device (test), when used to classify newborns on the first day of life, was useful in identifying premature infants but not when applied to identify SGA infants, considering current algohrithm. Nonetheless, the new device (test) has the potential to provide important information in places where the GA is unknown or inaccurate.

Newborn Skin Maturity Medical Device Validation for Gestational Age Prediction: Clinical Trial

Background

Early access to antenatal care and high-cost technologies for pregnancy dating challenge early neonatal risk assessment at birth in resource-constrained settings. To overcome the absence or inaccuracy of postnatal gestational age (GA), we developed a new medical device to assess GA based on the photobiological properties of newborns’ skin and predictive models.

Objective

This study aims to validate a device that uses the photobiological model of skin maturity adjusted to the clinical data to detect GA and establish its accuracy in discriminating preterm newborns.

Methods

A multicenter, single-blinded, and single-arm intention-to-diagnosis clinical trial evaluated the accuracy of a novel device for the detection of GA and preterm newborns. The first-trimester ultrasound, a second comparator ultrasound, and data regarding the last menstrual period (LMP) from antenatal reports were used as references for GA at birth. The new test for validation was performed using a portable multiband reflectance photometer device that assessed the skin maturity of newborns and used machine learning models to predict GA, adjusted for birth weight and antenatal corticosteroid therapy exposure.

Results

The study group comprised 702 pregnant women who gave birth to 781 newborns, of which 366 (46.9%) were preterm newborns. As the primary outcome, the GA as predicted by the new test was in line with the reference GA that was calculated by using the intraclass correlation coefficient (0.969, 95% CI 0.964-0.973). The paired difference between predicted and reference GAs was −1.34 days, with Bland-Altman limits of −21.2 to 18.4 days. As a secondary outcome, the new test achieved 66.6% (95% CI 62.9%-70.1%) agreement with the reference GA within an error of 1 week. This agreement was similar to that of comparator-LMP-GAs (64.1%, 95% CI 60.7%-67.5%). The discrimination between preterm and term newborns via the device had a similar area under the receiver operating characteristic curve (0.970, 95% CI 0.959-0.981) compared with that for comparator-LMP-GAs (0.957, 95% CI 0.941-0.974). In newborns with absent or unreliable LMPs (n=451), the intent-to-discriminate analysis showed correct preterm versus term classifications with the new test, which achieved an accuracy of 89.6% (95% CI 86.4%-92.2%), while the accuracy for comparator-LMP-GA was 69.6% (95% CI 65.3%-73.7%).

Conclusions

The assessment of newborn’s skin maturity (adjusted by learning models) promises accurate pregnancy dating at birth, even without the antenatal ultrasound reference. Thus, the novel device could add value to the set of clinical parameters that direct the delivery of neonatal care in birth scenarios where GA is unknown or unreliable.

International Registered Report Identifier (IRRID)

RR2-10.1136/bmjopen-2018-027442

Quality of routine health facility data used for newborn indicators in low- and middle-income countries: A systematic review

Background

High-quality data are fundamental for effective monitoring of newborn morbidity and mortality, particularly in high burden low- and middle-income countries (LMIC).

Methods

We conducted a systematic review on the quality of routine health facility data used for newborn indicators in LMIC, including measures employed. Five databases were searched from inception to February 2021 for relevant observational studies (excluding case-control studies, case series, and case reports) and baseline or control group data from interventional studies, with no language limits. An adapted version (19-point scale) of the Critical Appraisal Tool to assess the Quality of Cross-Sectional Studies (AXIS) was used to assess methodological quality, and results were synthesized using descriptive analysis.

Results

From the 19 572 records retrieved, 34 studies in 16 LMIC countries were included. Methodological quality was high (>14/19) in 32 studies and moderate (10-14/19) in two. Studies were mostly from African (n = 30, 88.2%) and South-East Asian (n = 24, 70.6%) World Health Organization (WHO) regions, with very few from Eastern Mediterranean (n = 2, 5.9%) and Western Pacific (n = 1, 2.9%) ones. We found that only data elements used to calculate neonatal indicators had been assessed, not the indicators themselves. 41 data elements were assessed, most frequently birth outcome. 20 measures of data quality were used, most along three dimensions: 1) completeness and timeliness, 2) internal consistency, and 3) external consistency. Data completeness was very heterogeneous across 26 studies, ranging from 0%-100% in routine facility registers, 0%-100% in patient case notes, and 20%-68% in aggregate reports. One study reported on the timeliness of aggregate reports. Internal consistency ranged from 0% to 96.2% in four studies. External consistency (21 studies) varied widely in measurement and findings, with specificity (6.4%-100%), sensitivity (23.6%-97.6%), and percent agreement (24.6%-99.4%) most frequently reported.

Conclusions

This systematic review highlights a gap in the published literature on the quality of routine LMIC health facility data for newborn indicators. Robust evidence is crucial in driving data quality initiatives at national and international levels. The findings of this review indicate that good quality data collection is achievable even in high-burden LMIC settings, but more efforts are needed to ensure uniformly high data quality for neonatal indicators.