Design
A quasi-experimental cohort design was employed using two groups. The first group, the control (bottle) group, received all oral feedings by bottle during hospitalization as that was the standard practice in the Neonatal Intensive Care Units (NICUs) where the study was conducted. The control group was studied first to avoid the exposure of the control group to the intervention, cup feeding. The second group, the intervention (cup) group, received all oral feedings by cup during hospitalization. Infants in both groups were studied weekly for six weeks after discharge.
Sample
The convenience sample consisted of 60 late preterm infants admitted to the NICU. Thirty infants were assigned to the control group and the next 30 to the intervention group. To calculate the sample size, statistical power analysis was performed using a medium effect size and a power of 80% [23]. Breastfeeding prevalence at discharge from a previous study [14] was used to conduct the power calculation. Infants met the following inclusion criteria: (a) singleton birth, (b) 34 to 37 weeks of gestation at birth, (c) maternal intention to breastfeed, (d) no supplemental oxygen required, and (e) being fed only by intermittent gavage feeding at the time of recruitment. Infants could be in open cribs, radiant warmers, or incubators. Infants who had any condition interfering with oral feeding, including an oral congenital anomaly, intracranial hemorrhage, and/or craniofacial anomalies, were excluded. All potentially eligible infants and mothers were approached sequentially until the required sample was completed for each group, with the intervention group being recruited after completion of the control group. Total attrition for the study was 22 mothers, with one mother not returning for the fourth week visit, nine more mothers not returning at week five and an additional 12 mothers were lost at week six. At week six, 25 mothers in the control group and 13 mothers in the intervention group remained in the study (Figure 1).
Setting
The study was conducted in two transitional nurseries in neonatal intensive care units at Pediatric University Hospital, Cairo, Egypt. Mothers in the current study setting were not instructed on how to express breast milk.
Instruments
Two study instruments were used for data collection. The first was The Maternal Breastfeeding Practice Questionnaire (MBP), developed for use in this study to assess daily infant feeding practices during the past week. The MBP included demographic questions as well as questions that assessed breastfeeding frequency and the number of bottle feedings, and whether artificial formula or any other type of feedings had been given to the infant. The proportion of feedings that were breastfeeding (direct breastfeeding or any expressed breast milk) was classified according to the Labbok and Krasovec schema for the definition of breastfeeding [24]. The schema divides the act of breastfeeding into three major categories: full breastfeeding (exclusive and almost exclusive), partial breastfeeding, and token breastfeeding. Exclusive breastfeeding means that nothing other than breast milk enters the infant's mouth. Almost exclusive breastfeeding means that water, vitamins or ritualistic feedings like herbal drinks are given infrequently but not for nutritional purposes. Partial breastfeeding means supplementing the infant's feedings with other foods or liquids, and includes three levels: "High," "Medium," and "Low". Partial breastfeeding levels represent the proportion of breastfeedings per day, or the relative amount of breast milk consumed to any other feeds (> 80%, 20 – 80%, < 20%). Token breastfeeding reflects minimal and irregular breastfeeding that constitute less than 15% of the total daily feedings, and using the breast primarily for infant comfort and consoling, not for nutrition.
The Premature Infant Breastfeeding Behavior Scale (PIBBS) [25] was the second instrument. The PIBBS was used to measure the infant's breastfeeding behaviors at one to six weeks after discharge. The PIBBS consists of 11 items; six of these items measure the development of preterm infant's breastfeeding behavior, while the other five items measure factors related to the breastfeeding session, such as the infant's general behavior, presence of letdown reflex, how long the infant was held, presence of any breast problem, and influence of the environment. Consequently, only the six items used for the scoring of the infant's breastfeeding behavior were used in the current study. The items were rooting, areola grasp, longest duration of latching, amount of sucking, longest sucking burst, and swallowing [25]. Face validity of the PIBBS was determined by three experts working in a WHO project on breastfeeding and demonstrated good capacity to describe maturational steps in infant breastfeeding behavior, ranging from the most immature to full term mature behavior. Inter-rater reliability of the PIBBS showed acceptable and satisfactory agreement between two observers in terms of percent of agreement and Kappa values (0.88, 0.72) respectively [26]. For the current study, reliability of the PIBBS was measured and showed good reliability (Cronbach's alpha based on standardized items 0.88).
Procedures
The study was approved by the research committee at the School of Nursing, and the Pediatric University Hospital, Cairo University. The principal investigator (PI) reviewed the eligibility of each infant admitted to the transitional nursery during December 2003 to August 2004. If the infant met the eligibility criteria, the study was described to the mother. Mothers' verbal consent was obtained before data collection which was the standard procedure for consent in that setting. For both groups, oral feedings were started when determined by the attending physician in the NICU. Infants were fed either by bottle (control group) or by cup (intervention group) from the time oral feeding was started until discharge. Bottle feedings were given either by the assigned nursing staff or by the PI. All cup feedings were given by the PI or by one of two research assistants who are staff nurses at the NICU and who had been trained in the cup feeding technique by the PI. Lang's cup feeding technique [17] was used.
After infants were discharged from the NICU, mothers were interviewed at the first outpatient visit (one week post discharge) in a private room adjacent to the NICU to recall their breastfeeding practices during the previous week. Additionally, mothers and infants were observed by the PI during one breastfeeding session weekly for six weeks for assessment of infants' breastfeeding behaviors. The observation unit was a breastfeeding session defined as beginning when the mother initiated skin-to-skin contact with her infant and ending when skin-to-skin contact was terminated. The PI sat near the mother in a position which provided the best possible visibility of the infant's face and chin, and the infant's behavior at the breast was recorded using the PIBBS. At the end of the first breastfeeding session the PI asked the mothers the questions included in the Maternal Breastfeeding Questionnaire. It included questions about the frequency of breastfeeding during the day and the night, if any bottle feedings were given to the infant since discharge from the hospital, what was given (i.e. infant formula) and the frequency.
Statistical analysis
Data were analyzed using the Statistical Package for the Social Sciences (SPSS) version 14. Descriptive statistics (frequency, percentage, mean, standard deviation, and range) were used to describe demographic characteristics of infants and mothers. Differences between groups for demographic interval data were determined using chi-square and for continuous data using t-tests. Two-way repeated-measures ANOVA was used to examine the effect of feeding method (between-subjects effect) and time (within-subjects effect) and the interaction between feeding method and time on the preterm infant breastfeeding behaviors after discharge. A Type I error of 0.05 was used as the level of statistical significance for all tests.