Limitations
There are several limitations when considering our study’s findings. The study group was not a probability sample fully representative of a defined population. It was of generally high socioeconomic status and, representative of Iowa, had little minority inclusion. Respondents were more educated than nonrespondents.39 Although response rates were generally favorable, approximately 100 to 300 did not respond at a given time point, resulting in censoring of 26% of the cases. Data on breastfeeding, sucking, childcare, and so forth were collected at 3 discrete time points and not on a more frequent, daily, weekly, or monthly basis. Although recall bias was limited by the short-term nature of recall with 6-week and 3-month intervals, it could have an effect on results. Since so few infants exclusively breastfed, any breastfeeding was the only suitable dependent variable. No maternal employment data were collected and quantification of pacifier use was not included.
Only our study and that of Howard and colleagues33 reported results from the United States. The statistical analyses by Howard and colleagues concerning pacifier use adjusted for a number of factors, including plans to return to work, family and paternal preferences for breastfeeding, and breastfeeding goal. Our study adjusted for plans to breastfeed and demographic factors while assessing the effects of pacifier use, digit sucking, and number of child care days. However, neither study specifically assessed reasons for use of the pacifier, in particular, in relation to work and child care requirements. So pacifiers could have been used to facilitate weaning, thus resulting in the association with reduced breastfeeding. Also, there may be other confounding differences between those using pacifiers and those who did not.
Although decisions by mothers to return to work, or for other reasons, have their infants attend child care were not generally associated with reductions in breastfeeding, our results suggest that child care has an important impact on determining the relationships between non-nutritive sucking behaviors and cessation of breastfeeding. It has been suggested that infants’ abilities to easily and successfully breastfeed are adversely affected by non-nutritive sucking, resulting in reductions in the frequency and consistency of the breastfeeding sessions. Our data support the concept. However, it is important to acknowledge that decisions to stop breastfeeding (often prior to return to work) may have preceded and led to the increase in non-nutritive sucking, rather than sucking leading to cessation of breastfeeding. That is, after the decision has been made to stop breastfeeding, a pacifier may be introduced to ease the transition to bottle feeding.
Additional studies involving in-depth interviews concerning initial and subsequent breastfeeding, employment, and child care plans would be warranted to address this question further. In addition, more controlled studies to determine whether there is any biological relationship between non-nutritive sucking and breastfeeding difficulties are warranted. Clearly, the social, biological, and economic factors involved in decisions to initiate and cease breastfeeding are complex and will require more study, both in the United States and throughout the world.
Acknowledgments
Our study was supported in part by National Institutes of Health grants #RO1-DE09551 and #P30-DE10126 and the University of Iowa’s Obermann Center for Advanced Study. We thank the staff of the Iowa Fluoride Study for their assistance in implementing the study, and Tina Craig for manuscript preparation.