Development, construct validity, and reliability of the questionnaire on infant feeding : a tool for measuring contemporary infant-feeding behaviors / Elizabeth J. O'Sullivan, Kathleen M. Rasmussen

Background: The breastfeeding surveillance tool in the United States, the National Immunization Survey, considers the maternal-infant dyad to be breastfeeding for as long as the infant consumes human milk (HM). However, many infants consume at least some HM from a bottle, which can lead to health outcomes different from those for at-the-breast feeding. Objective: Our aim was to develop a construct-valid questionnaire that categorizes infants by nutrition source, that is, own mother's HM, another mother's HM, infant formula, or other and feeding mode, that is, at the breast or from a bottle, and test the reliability of this questionnaire. Design: The Questionnaire on Infant Feeding was developed through a literature review and modified based on qualitative research. Construct validity was assessed through cognitive interviews and a test-retest reliability study was conducted among mothers who completed the questionnaire twice, 1 month apart. Participants/setting: Cognitive interviews were conducted with ten mothers from upstate New York between September and December 2014. A test-retest reliability study was conducted among 44 mothers from across the United States between March and May 2015. Main outcome measures: Equivalence of questions with continuous responses about the timing of starting and stopping various behaviors and the agreement between responses to questions with categorical responses on the two questionnaires completed 1 month apart. Statistical analyses performed: Reliability was assessed using paired-equivalence tests for questions about the timing of starting and stopping behaviors and weighted Cohen's κ for questions about the frequency and intensity of behaviors. Results: Reliability of the Questionnaire on Infant Feeding was moderately high among mothers of infants aged 19 to 35 months, with most questions about the timing of starting and stopping behaviors equivalent to within 1 month. Weighted Cohen's κ for categorical questions indicated substantial agreement. Conclusions: The Questionnaire on Infant Feeding is a construct-valid tool to measure duration, intensity, and mode of infant HM consumption and duration of maternal HM production that is reliable within 19 to 35 months postpartum. Criterion-validity testing of these questions will improve the utility of the Questionnaire on Infant Feeding as a surveillance tool.

Nutrition.