Study design and participants
The basis for this analysis was a multi-center prospective cohort study that recruited mother-infant pairs (n=1287) from the postnatal obstetric units of four geographically distributed public hospitals in Hong Kong from 2011 to 2012. One thousand and eleven participants reported their intended duration of breastfeeding and were included in the analysis as we used this question to categorize meeting their intended duration of breastfeeding. The remaining 276 participants had missing data for the main variables and were excluded. Participants were all Cantonese-speaking Hong Kong residents who gave birth to a single full-term infant, intended to breastfeed, and had no serious medical or obstetric complications. Participants were excluded if the newborn had a birthweight of <2,500 grams, a 5-minute Apgar score of <7, and if they were admitted to the special care nursery for >48 hours after birth or were admitted to the neonatal intensive care unit.
Study participants were recruited by a trained research nurse within the first 24 hours after giving birth. In-hospital data collection consisted of a baseline self-administered demographic questionnaire, maternal and newborn data, exposure to six of the ten baby-friendly steps, and in-hospital infant feeding data. We measured the six out of ten baby-friendly steps that directly affect individual mothers including mothers initiating breastfeeding within 1 hour of birth (step 4), the hospital giving no food or drink other than breastmilk unless medically indicated (step 6), rooming in (step 7), breastfeeding on demand (step 8), giving no teats (step 9) and providing mothers with information on breastfeeding support (step 10). Maternal and newborn data, exposure to baby-friendly hospital steps, and in-hospital infant feeding data were extracted from the maternal and/or newborn health record by the research nurse and were validated with the participant. Follow-up newborn feeding data were collected over the telephone by a trained study research assistant at 1, 2, 3, 6, and 12 months post-partum or until weaned.
As part of the in-hospital demographic questionnaire, mothers were asked “How many weeks in total do you plan to breastfeed?” At the follow-up telephone interviews, participants were asked “What was the age of your baby (in weeks) when she or he was no longer receiving any breast milk?” We used these two questions to compute a proportion that reflected the achieved intended duration of breastfeeding, with a possible range from 0 to 100%. Participants were then categorized as having achieved their intended breastfeeding duration (yes/no), with those who breastfed for ≥90% of their intended duration categorized as yes and those who breastfed for <90% as no. Participants still breastfeeding at the end of the 12-month follow-up were also considered to have met their intentions, irrespective of their intended duration of breastfeeding. The main predictor variables were the six baby-friendly hospital steps, and maternal and infant data. Consistent with other studies [9-11], we measured the six out of ten baby-friendly steps that affect individual mothers including: (1) step 4: help mothers to initiate breastfeeding within the first hour after birth; (2) step 6: give newborns no food or drink other than breast milk, unless medically indicated; (3) step 7: practice of rooming-in; (4) step 8: encourage breastfeeding on demand; (5) step 9: give breastfeeding newborns no artificial teats or pacifiers; and (6) step 10: provide mothers with information about breastfeeding support on hospital discharge.
The Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster and the participating study sites ethics committees approved the study. All participants provided informed written consent.
Descriptive statistics were used to compare the sociodemographic characteristics of participants by achievement of their planned duration of breastfeeding. We used unadjusted and adjusted logistic regression models to examine the effect of baby-friendly hospital steps on participants’ meeting their planned duration of breastfeeding. Model 1 was adjusted for the six BFHI steps only and Model 2 was additionally adjusted for sociodemographic variables. We computed the proportion of participants’ who achieved their planned duration of breastfeeding by how many Baby-friendly hospital practices they experienced, and we used unadjusted and adjusted logistic regression analysis to examine the effect of the number of practices on achieving breastfeeding intentions. In all multiple logistic regression models, the Hosmer–Lemeshow test  was used to assess the goodness of fit of the models.
Data were analyzed using Stata version 14.1 statistical software (StataCorp LP, College Station, Texas, USA) . A 5% level of significance was used in all statistical tests.