Study design
This is a quasi-experimental study.
Settings
This study was conducted in Sichuan Province of western China from May 2017 to December 2017. As one of the pilot hospitals where the implementation of EENC was rolled out successfully since March 2016, Sichuan Provincial Hospital for Women and Children provided qualified faculties and technical support for this study. Mianyang City and Deyang City, with the highest numbers of annual live births in Sichuan Province, were chosen as research settings and randomly assigned as the intervention group (Mianyang City) and the control group (Deyang City). The conditions of medical care services and economic development were equivalent in the two groups. One city level and three county level maternal and children’s hospitals were randomly elected in each group. EENC was introduced in the intervention hospitals while the control hospitals followed routine childbirth and newborn care procedures.
Participants
Inclusion and exclusion criteria
Neonates and their mothers were included in this study according to the following criteria: 1) pregnant women had no history of medical problems and agreed to accept the implementation of EENC with written informed consent; 2) full-term (gestational age: ≥37 weeks and <42 weeks) newborns with birthweight >2500 g and <4000 g, who were delivered vaginally. The exclusion criteria included: 1) women with abnormal pregnancy who were admitted to hospitals for any reason, such as those with obstetric emergencies and/or serious underlying diseases; 2) newborns who were admitted to the neonatal unit for any problems. Caesarean deliveries were excluded because the current clinical protocols of caesarean section in China were inconsistent with EENC recommendations and hospitals were not prepared to do so.
Eligible pregnant women were recruited in the prenatal inpatient wards and allocated a number according to order of hospitalization. Those who were included in this study were selected using random number table method. If the women and/or newborns were excluded after delivery, the recruitment and sampling procedure were repeated.
Sample size considerations
The sample size of this study was calculated using the following formula (P1: exclusive breastfeeding rates before hospital discharge in the pre-EENC phase; P2: exclusive breastfeeding rates before hospital discharge in the post-EENC phase; P:; Z1-α/2/Zβ: standard normal deviance at the significance level of α/1-β; Deff: design effect):
The initial estimate of P1 (46.6%) was applied based on previous publications [22, 23]. We hoped the EENC implementation would increase this rate by 15% and thus made an assumption of P2 (63%). A minimum sample of 255 (α=0.05; β=0.10; Deff=1.2) mother-child pairs per group was therefore calculated with allowance for 10% loss to follow up. The total number of mother-child pairs in the study was 1020 (pre and post*intervention and control=255*4).
Implementation of EENC
EENC was introduced in the intervention hospitals in July 2017 by firstly coaching health professionals that included obstetricians, obstetric nurses, midwives, pediatrician/ neonatologists, and pediatric nurses using Early Essential Newborn Care Module 2 - Coaching for the First Embrace - Facilitator's Guide [24]. The key interventions of EENC include immediate and sustained skin-to-skin (SSC) of mother and newborn for at least 90 minutes after birth, timely breastfeeding when newborns exhibit feeding cues, delayed umbilical cord clamping, immediate and thorough drying, and neonatal resuscitation for those without spontaneous breathing. Mother and newborn SSC should be the continuous contact that the naked baby is placed against mother’s breast and abdomen until cues of readiness to suck, such as rooting, drooling, tonguing, and biting hands [13, 14, 22]. By September 2017, a quality control assessment was carried out by provincial facilitators to oversee the implementation of EENC and ensure that all trained staffs grasped the skills. The control hospitals implemented routine childbirth care practices. EENC was introduced in the control hospitals after the final data collection.
Data collection
Pre- and post-intervention data were collected in both groups. The baseline data of the intervention group (n=331) and control group (n=381) were collected from May to June 2017, which was before the EENC introduction. When the intervention hospitals could implement EENC with satisfying quality, we collected post-intervention data in the two groups (n=312 and 325, respectively) from October to December 2017.
The same questionnaires were applied in the pre- and post-intervention phase. Birth records on vital signs of mothers and neonates, situation of EENC implementation, and breastfeeding practices (including the time of initiation and duration of breastfeeding), were documented by midwives after delivery. Exit interviews with postpartum mothers were conducted by obstetric nurses in maternity wards to record information about feeding patterns (exclusive breastfeeding: infants were fed with only breastfeeding without any liquids or solid food, except for medicine, minerals and vitamin; artificial feeding: infants were given formula milk only; breastfeeding: infants received any breastfeeding, combined with formula) prior to discharge. Follow-up data on feeding practices, and mother and child’s health status were collected by child health care workers via telephone interviews or home visits at 1 month, 3 months, and 6 months post birth. The flowchart for data collection is shown in Figure 1. This study only focused on the effect of EENC on breastfeeding. The association of EENC with other child health indicators will be published separately.
Statistical analyses
Data were input into Epidata 3.0 with double-entry method and SPSS statistical software (version 22.0) was used for analyses. All P-values were two-sided with the significant difference at 5% level. For continuous variables, data with normal distribution were summarized by mean± standard deviance (SD), and the statistical differences between the two groups were compared by t-tests. Quantitative data with abnormal distribution were demonstrated by median and inter-quartile range (IQR), and Wilcoxon rank sum tests were carried out to ascertain the differences. For categorical variables, data were described by frequencies and proportions, and Pearson chi-squared tests were performed to compare outcomes pre- and post-EENC implementation.
Difference in difference (DID) is a statistical model specific to the nonequivalent design. It is applied to deal with baseline differences between the intervention and control groups and estimate the net effect with control of confounding variables [25-27]. With respect to continuous variables (time of SSC initiation, duration of SSC, duration of the first breastfeeding) as outcomes in this study, general linear model (formula 1) was applied to explain the effect of EENC. Logistic regression model (formula 2) was introduced to examine the intervention effect with categorical variables (successful first-time breastfeeding after SSC, breastfeeding pattern) as outcomes. Estimates of the outcome measures were performed by odds ratio (OR) and corresponding 95% confidence intervals (CI). Effects of EENC on breastfeeding were analyzed by DID with group, time, effect (group*time) as independent variables, and time of SSC initiation, duration of SSC, duration of the first breastfeeding, successful first-time breastfeeding with provision of SSC, and breastfeeding pattern as dependent variables. The definitions of variables are shown in Table 1.