Friendly Hospital initiative (BFHI)24 and can comprise professional or lay support, educational or supportive interventions, or any combination of these. Breastfeeding support is more likely to be effective if it is proactive, delivered face to face and provided on an ongoing basis.25 However, the majority of studies evaluating the effects of breastfeeding support are restricted to healthy term infants or they evaluate the impact of breastfeeding given in the neonatal unit (NNU) to preterm infants.26 In the UK, at the time of the survey, support and advice about breastfeeding is usually provided initially by midwives during pregnancy. Advice may also be offered during antenatal classes, either provided through the NHS, or by third sector organisations. After the birth of an infant, breastfeeding support would initially be provided by midwives or midwife care assistants within a hospital setting. Most hospitals offer breastfeeding support sessions including breastfeeding counsellors. At around day 10 of life, care of the mother and infant is transitioned from midwifery to the health visiting team who continue to care for the child until the age of 5 years. Health visitors can offer guidance and support with breastfeeding. In addition, breastfeeding support groups including counsellors and peer support are often provided in children’s centres or by third sector organisations the details of which would be provided to the mother by either the midwife or health visitor. Given the increased vulnerability in the late preterm population and higher likelihood of breastfeeding difficulties, it is particularly necessary to understand which breastfeeding support is the most effective for this group. The objective of this study was to investigate the association between different forms of breastfeeding support on breastfeeding rates at 10 days and 6 weeks in late preterm and term infants. METHODS Study design This was a cross-sectional study involving secondary analysis of data from the UK 2010 Infant Feeding Survey (IFS).17 The IFS is undertaken every 5 years in order to monitor breastfeeding prevalence and infant feeding practices. In 2010, it comprised a nationally representative unclustered sample of 30 760 infants with oversampling of mothers from the most deprived quintile of the Index for Multiple Deprivation (IMD). The use of weights in the survey analysis allowed for this oversampling and differential non-response according to mother’s age and IMD, and aims to makes the survey representative of the UK population. The IFS included three postal questionnaires assessing feeding status of the infant at 6–10 weeks, 4–6 months and 8–10 months. They were completed by self-report, usually by the mother of the infant. The IFS questionnaire in 2010 was based on the previous version and was piloted in all four countries of the UK prior to being rolled out. The original survey was approved by the Ethics committee, Department of Health Sciences at the University of York. This study primarily used data from the first questionnaire, which a total of 15 724 mothers responded to (51% response rate). This questionnaire contained 150 questions over 39 pages (see web appendix 1: Questionnaire) including questions on breastfeeding duration and breastfeeding support accessed. For infants who were less than 6 weeks at the first questionnaire, data were also used from the second questionnaire (3382 infants). The analysis was restricted to singleton births born at more than 34+0 weeks gestation (15 104 infants) (see web appendix 2: flow chart). Only those infants who initiated breastfeeding were included in the analysis of breastfeeding at 10 days. Only those still breastfeeding at 10 days were included in the subsequent analysis at 6 weeks. Measures Breastfeeding support variables were identified from questions in the stage 1 questionnaire27 (box 1). Specific variables were chosen to reflect different times at which support was given, from antenatal and immediately following delivery, to subsequent support given at home. To fully explore the association of breastfeeding support, these variables were also chosen to reflect the different nature of support ranging from direct support such as skin to skin contact to provision of information. Outcomes included the prevalence of breastfeeding at 10 days and 6 weeks. Infants were classified as having initiated breastfeeding if any breastmilk (direct or expressed) was received after birth, even if only once.28 Subsequently, infants were classified as breastfed if they were receiving any breastmilk as part of their nutrition, regardless of other fluids or solids. This included infants who were regarded as ‘partial breastfeeding’, but also those who were exclusively breastfed (when no other food or drink, not even water, except breastmilk is received). Outcome timepoints were chosen as the transition from midwife care to the health visiting team occurs at 10 days postpartum and it is accepted that breastfeeding can take 6 weeks to be fully established. Potential confounding factors were divided into sociodemographic characteristics and pregnancy and delivery characteristics. Sociodemographic characteristics included the infant’s ethnicity, maternal age, marital status, age at leaving full time education, socioeconomic status (as defined by maternal occupation using the NSSEC system29), and