10.7272/Q69S1P7K
Miller, Lara
University of California, San Francisco
Migori County Referral Hospital_Gestational Age Data_Nov15-April16
Dryad
dataset
2018
Bill & Melinda Gates Foundation
https://ror.org/0456r8d26
OPP1107312
2018-03-21T04:32:12Z
2018-03-21T04:32:12Z
en
43987 bytes
1
Creative Commons Attribution 4.0 International (CC BY 4.0)
Background: Preterm birth is the leading cause of neonatal mortality
worldwide and specifically in Kenya. Preterm birth is defined by
gestational age (GA) less than 37 weeks, but GA estimates are questionable
in the absence of the gold standard of early ultrasound. In Migori County,
where the majority of women seek care at government facilities and do not
receive ultrasound dating, the accuracy of the GA estimates, and therefore
the preterm birth rates, is unknown. Methods: We conducted a retrospective
chart review of 455 preterm births from Migori County Referral Hospital, a
level-four government hospital in Western Kenya. Preterm birth was defined
in this context as all babies less than 2500g and babies greater than
2500g with a last menstrual period (LMP) calculated GA of less than 37
weeks. GA estimates from both the maternity register and the individual
inpatient chart were evaluated for data quality, agreement between
measurements, and accuracy when compared to the INTERGROWTH-21st
International Newborn Birthweight Standards as a benchmark. Results: Data
completeness ranged from 35.3% for recorded GA in the inpatient chart to
97.8% for birth weight in the maternity register. LMP and recorded GA
agreed in 16.8% of cases in the maternity register and 19.2% of cases in
the inpatient chart, while symphysis fundal height agreed in 37.1% and
50.5% of cases, respectively. Of the four GA measures evaluated, the
maternity register recorded GA was accurate in 69.4% of cases, maternity
register LMP-calculated GA in 57.9% of cases, the inpatient recorded GA in
68.8% of cases, and the inpatient LMP-calculated GA in 56.0% of cases.
Preterm birth rates calculated from the four GA measures ranged from 7.5%
to 18.8%. Conclusion: Non-ultrasound methods of estimating GA result in a
wide range of results, with up to five different estimates per woman. With
such conflicting data, clinical decision making is compromised and preterm
birth facility estimates are likely inaccurate. Widespread access to early
ultrasound, new technologies, and/or new methods of thinking about GA are
urgently needed to improve clinical care for the mother-infant dyad, and
to better understand the preterm birth burden in low-resource settings.
The dataset evaluated was compiled from two routine data sources at MCRH.
The first was the maternity register, a large book kept in the maternity
ward where providers record demographic and clinical data for every birth
at the facility. The second was the maternity unit inpatient record, the
individual patient medical chart of each woman admitted to MCRH for labor
and delivery. Both the maternity register and the inpatient record are
standardized forms from the Ministry of Health and completed by the
clinical staff. Data from both sources were entered into a Research
Electronic Data Capture (REDCap) tool, an online data collection mechanism
hosted on a secure server at the University of California, San Francisco
(UCSF) and created in advance by the investigators. During the study
period, there were 2070 total births at MCRH of which 455 were identified
as cases from the maternity register by either having a birth weight
<2500g or having a birth weight >2500g and an LMP-calculated
GA of <37 weeks. Inpatient records were found for 289 (63.5%) of
the cases with the remaining 166 missing, mislabeled, or otherwise
inaccessible. The five GA estimates analyzed were recorded GA and
calculated GA from the maternity register and recorded GA, calculated GA,
and SFH from the inpatient chart.
Anybody using this data must credit the authors and the East Africa
Preterm Birth Initiative.