We employed data from a subset of women enrolled in a RCT named DHA to Optimize Mother and Infant Outcome (DOMInO). The DOMInO trial involved 2399 pregnant women, recruited from 2005 to 2008, at <20 week gestation with singleton pregnancies. DOMInO was a double-blind, multicentre RCT conducted in five Australian perinatal centres: two in South Australia and one each in New South Wales, Victoria and Queensland .
Study participants analysed in the present manuscript were the subset of women from the DOMInO trial who delivered at baby friendly hospitals and were certified in their medical record as competently breastfeeding their infant at discharge by their attending midwife. These subjects were followed until six months postpartum as part of the DOMInO study protocol. They were telephoned at six weeks and six months postpartum by a study nurse and interviewed. Information collected at interview included infant feeding status, in particular, whether they were fully or partially breastfeeding their infant and whether they had introduced any solids.
Study hypothesis and sample size
Our hypothesis was that lower level of maternal education would be a leading predictor of early discontinuation of breastfeeding in line with the largest previously published prospective study . We predicted that 70% of women who continued to breastfeed to 6 months would have achieved post secondary education, against a background rate of 60% in women using formula. To achieve power of 90% and alpha error of 0.05, a sample size of 496 was required in each group.
Classification of infant feeding
A research nurse interviewed each study participant at six weeks and six months postpartum. Based on information supplied by the mother at these assessments, feeding was coded as ‘breastfeeding’ if the mother reported that she continued to breastfeed the baby or breastfed in combination with infant formula at six months. If the mother reported that she exclusively infant formula fed, then feeding was coded as ‘formula’. Fully breastfeeding was defined as breastfeeding with no infant formula or solids, although administration of water and juices was allowed in the preceding 24 hours.
Baseline demographic and obstetric characteristics of the mother and assessments of social support, depressive symptomatology, smoking and alcohol use were established using validated tools that have been previously utilised in maternal and child health studies [11-19].
Pregnancy data in respect to infant gender, gestational age at birth, preterm birth, very preterm birth and birth weight were abstracted from the hospital medical record. Symptoms of postnatal depression were assessed using the Edinburgh Postnatal Depression Scale (EPDS), which was completed by mothers at six weeks and six months post-partum. A score of more than 12 was coded as a positive case for depression [17-19].
A trial coordinator monitored data collection, which was facilitated by a web-based management information system.
Discrete variables were presented as number and percentage and were analysed using Chi Square tests to generate p-values and Odd Ratio with 95% confidence intervals. Continuous variables with normal distribution were presented as mean and standard deviation and compared using T-tests.
Continuous variables with skewed distribution were presented as median and interquartile range and compared using Mann Whitney U tests.
Factors significantly associated with feeding type at a p-value <0.1 were included in a multivariate regression model. Statistical significance was accepted at the 0.05 levels. We then retrospectively applied a model of significant independent associations to determine the extent to which these variables identified women who prematurely ceased breastfeeding.
Ethics Committees conformed to NHMRC and Declaration of Helsinki 1995 criteria and prospective trial registration was undertaken. The Ethics Committees and approval numbers were: Women’s and Children’s Hospital, South Australia, Australia - 1657/12/2007; Flinders Medical Centre, South Australia, Australia - 199/045; Sunshine Hospital, Victoria, Australia - REC 1657/12/2010; Campbelltown Hospital, New South Wales, Australia - HE05/142; and Royal Brisbane and Women's Hospital, Queensland, Australia – 2007000424.
Written informed consent was obtained from each participant. Data in this manuscript were extracted from the DOMInO trial data set and were subjected to secondary analysis. The DOMINO trial was registered on the Australian Clinical Trials Research Network (ACTRN12605000569606) .