The study was conducted with a single-center, prospective, observational design.
The study was conducted at Pokhara Academy of Health Sciences, Pokhara, Nepal from May to December 2020. The hospital had 6500 annual births and is the referral hospital of the province. A nurse-midwife attends all births with on-call support from a general medical doctor, pediatrician and obstetrician when needed.
REFINE, a heart-rate-guided neonatal resuscitation protocol, was implemented by providing training in Helping Babies Breathe (HBB) supplemented by a half-day orientation on the use of NeoBeat at all births . With this curriculum, providers were taught to provide delayed cord clamping if an infant was vigorous, which was defined as maintaining the cord intact for at least 60 seconds in HBB. Within REFINE, periodic meetings were facilitated by the quality improvement team to discuss progress on implementation of heart-rate-guided neonatal resuscitation care using a Plan-Do-Study-Act (PDSA) process .
All women admitted to the labor room with a singleton pregnancy at gestational age 34 weeks or more were eligible for this study. Eligible women who agreed to participate provided written consent for enrollment. Infants born by caesarean section were not included as observation during childbirth in the operation theatre was not feasible. All newborns who were not vigorous at birth or required respiratory support (bag mask ventilation) were excluded from this study. Vigorous newborns were defined as those who cried immediately (within approximately 15 seconds) after birth and required no intervention beyond thorough drying immediately after birth.
Data collection and management
Study research nurses were present in the labor and delivery room to collect data using mobile device with a purpose-built Liveborn application to record real-time observations of newborn care in the first few minutes after birth as previously described . A research nurse attending the delivery noted time of birth, time of cord clamping, time of infant cry, and start and stop of resuscitation interventions such as stimulation, suctioning, and ventilation by annotating the events in the Liveborn application. After delivery the neonate was thoroughly dried by the health worker and a research nurse applied NeoBeat on the upper abdomen. NeoBeat incorporates two dry electrodes into a semi-circular carrier with an integrated digital display that is easily visible to the provider (Supplementary Fig. 1). Continuous HR data was recorded by was recorded by Neobeat, live streamed to the Liveborn application and synchronized with activities recorded in the application. Movement or handling of the baby could temporarily prevent detection or cause interference with HR.
Vigorous newborns were positioned on the mother’s chest after NeoBeat was placed. Drying of newborns was done on mother’s chest. Cord clamping was performed at the health worker’s discretion, although training on HBB recommended cord clamping after 1 minute and the majority of providers had received recent training. Demographics, obstetric history and obstetric complications during labour (Supplementary Table 1) and neonatal characteristics were collected from medical records . The Liveborn application automatically integrated heart rate with any intervention provided after time of birth. A single unique ID was provided for each participant to link the Liveborn app and demographic/medical history data for final data analysis.
For the data analysis, we excluded a participant’s data when there was an indication that the birth time registered in Liveborn application was not correct, i.e. if NeoBeat recording started before “baby born” was pressed in the Liveborn application. Other exclusion criteria included too little (<30 s), too late (>120 s after birth) or poor signal quality data. We categorized the participants into two different groups based on the timing of umbilical cord clamping, defined as early cord clamping (ECC) at 60 seconds or less and delayed cord clamping (DCC) at 61 seconds or more. The obstetric characteristics of ECC and DCC were analyzed including mean maternal age, parity, induction of labor, gestation, and obstetric complications during labor. The neonatal characteristics analyzed included gestational age in weeks, birth weight in grams, gender, Apgar score at 1 and 5 minutes, time of first HR in seconds and immediate drying after birth. Means were compared using independent t-test, medians by Mann-Whitney U test and proportions via Pearson’s Chi-square test.
We analyzed the continuous HR data for the ECC and DCC groups from 10 until 180 seconds after birth using centile curves at 3rd, 10th, 25th, 50th, 75th, 90th and 97th. The centile curves were smoothed using piecewise cubic spline fitting. Median HRs at 12 serial time points (10, 30, 45, 60, 75, 90, 105, 120, 135, 150, 165 and 180 s) were compared using Mann Whitney U test with post hoc Bonferroni correction of p-values to adjust for multiple repeated measures (p = 0.05/12 = p ≤ 0.004 for significance).
Proportion of bradycardic infants with 95% CI in the ECC and DCC groups was graphically presented using generalized additive models (GAM). The relative risk of ever being bradycardic (HR < 100 bpm) after cord clamping was determined by cord clamping group after birth using generalized linear model regression with binomial distribution and log link function. Gestational age in weeks, birthweight in grams, presence of bradycardia at time of cord clamp, and obstetric complications during labor were included as covariates. Results are presented as the crude relative risk (cRR) and adjusted relative risk (aRR), with 95% confidence intervals (CI). Analyses were performed using R version 44.1.2 (Vienna, Austria) and Matlab R2021a (Mathworks Inc, Natick, MA).