Design
The study had a prospective and longitudinal design with a consecutive data collection.
Participants and data collection
Participants were recruited between September 2012 and July 2015 from a neonatal intensive care unit or maternity unit at a Swedish university hospital, after being identified through the logbook in the hospital delivery ward. Inclusion criteria were mothers with a singleton infant born between 34 + 0 and 36 + 6 weeks of gestation. Excluded participants and dropout rates of the eligible mother-infant pairs have been described elsewhere (11, 12). The mothers were contacted by letter at term age (baseline; n = 148) and three months of corrected age (follow-up) and asked to complete paper-and-pencil questionnaires. For the present study, only the 110 mothers answering the infant state-regulation index at three months follow-up were eligible for participation. Moreover, we excluded five mothers who only fed their infant breast milk using a baby bottle, thus including a total of 105 breastfeeding mothers in the study.
Study variables
At baseline (term age), the mothers completed the BSES-SF instrument, as well as a study-specific questionnaire containing questions about clinical and demographic characteristics, breastfeeding and formula use (Table 1). The BSES-SF instrument comprises 14 questions about how confident the mother is with breastfeeding her new baby, with answers given on a 5-point scale, ranging from “not at all confident” (1 point) to “very confident” (5 points). The total score on the instrument ranges from 14 to 70 points, with higher scores indicating better self-efficacy. The study-specific questions at baseline covered the mother’s age (years), if she was married/cohabiting (yes/no), had a college or university education (yes/no), was employed before giving birth (yes/no), used tobacco (yes/no), the number of children she had, if she was primiparas (yes/no), and if she perceived that she had support for breastfeeding from anyone in her vicinity (yes/no), in which case she was given the opportunity to specify who provided the support. Additionally, questions about the infant concerned if it was a vaginal birth (yes/no), the age of the infant (weeks), his or her gestational age (GA) at birth (weeks) and birth weight (grams), and if he or she was given breast milk only by breastfeeding (yes/no).
At the three months corrected age follow-up, mothers were asked to complete the Infant state-regulation index, comprising eight questions concerning the mothers’ perceptions of their infants’ ability to state regulate: these questions related to whether the infant had difficulties with colic, persistent crying, comforting, falling asleep, sleep problems, breastfeeding, eating or poor weight gain. Answers were given on a four-point scale, ranging from “no problems” (0 points) to “very severe problems” (3 points), which were then summed to produce a total score, ranging from 0 to 24 points, with high scores indicating more problems with state-regulation.
Statistical analyses
Categorical data are presented as frequencies and percentages, n (%), while ordinal and continuous data are given as means with accompanying standard deviations (SDs). The associations between clinical, demographic, breastfeeding and formula use variables measured at baseline (term age) and the Infant state-regulation index measured at three months of corrected age follow-up were examined using adjusted and unadjusted linear regression analyses, with the results presented as slope coefficient β with accompanying 95% confidence intervals (CIs). Two different adjusted regression models are reported: a basic model including all variables having a P-value < 0.20 in the unadjusted regression analyses, and a trimmed model constructed by excluding variables with P-values ≥ 0.20 one-by-one from the basic model, starting with the variable with the highest P-value, and re-estimating the model until only variables with P-values < 0.20 remained in the model. All statistical analyses were performed using R 4.1.0 (R Foundation for Statistical Computing, Vienna, Austria), with two-sided P-values < 0.05 considered statistically significant.