We used data collected as part of Project Viva, a prospective cohort study of the determinants of pregnancy outcomes and offspring health.43
Pregnant members of eight selected obstetric offices of Harvard Vanguard Medical Associates, a large, multispecialty urban/suburban group practice in eastern Massachusetts, were enrolled in the study at their first prenatal visit and followed through delivery and postpartum. Human subjects committees of Harvard Pilgrim Health Care, Brigham and Women's Hospital, and Beth Israel Deaconess Medical Center approved the study protocols. At the first study visit, directly after the initial clinical prenatal visit, a trained research assistant approached the potential participant, described the study, obtained informed consent, administered a brief interview, and provided a take-home self-administered questionnaire. Research assistants were not associated with any of the clinical offices or hospitals involved in the participants' care; they informed participants at the initial screening visit that participation in the project would have no effect on routine care and that this information would not be communicated to their healthcare providers. Subjects were blinded to the specific hypotheses, and there were no specific hypotheses about lactation at the time of recruitment. Breastfeeding was not specifically mentioned during recruitment, and none of the recruiters were associated with breastfeeding promotion.
Exclusion criteria were multiple gestation, inability to answer questions in English, plans to move out of the area before delivery, and gestational age >22 completed weeks at initial prenatal clinical appointment. We enrolled 2670 pregnant women (64% of those eligible) between April 22, 1999, and July 31, 2002. Of the 2670 participants, 329 subsequently became ineligible because of multiple gestation (n
19), transferring obstetric care to a nonstudy site (n
115), or because they were no longer pregnant (n
195). Of the 2341 remaining participants, 195 (8%) withdrew and 18 (<1%) were lost to follow-up, leaving 2128 who delivered a live infant. We excluded women from this analysis who did not complete all items on the anxiety questionnaire (n
187) and who did not return (n
366) or who did not complete one or more questions of the second trimester Edinburgh Postpartum Depression Scale (EPDS) (n
44). We also excluded women who did not respond to the question regarding previous history of depression (n
3), which we defined as self-reported depressive symptoms and either a (1) prepregnancy diagnosis of depression by a health professional or (2) prior use of prescription antidepressants. Finally, we excluded women whose postdelivery breastfeeding status could not be determined (n
20) or who were uncertain what they would feed their baby in the first week of life (n
72). Thus, we based these analyses on 1436 women.
Study representatives enrolled women immediately after the initial clinical prenatal appointment, usually during the first trimester of pregnancy. Study representatives then conducted a brief interview, and women were asked to complete a short questionnaire that included 7 items on pregnancy-related anxiety of the 10 previously used by Rini et al.44
The questionnaire included such items as concern about how the baby is growing and developing and concern about having a hard or difficult labor and delivery. For each item, the possible responses were very much, moderately, somewhat, and not at all.44–46
The mean gestational age at the time the woman completed the anxiety questionnaire was 10.4 weeks.
The second trimester study visit occurred in late second trimester, typically at 26–28 weeks. At that time, participants completed a questionnaire on history of depressive symptoms, previous diagnosis of depression by a medical professional, and previous use of antidepressant medication. Participants also completed the 10-item Edinburgh EPDS.47
We have described the study procedures in more detail elsewhere.48
The EPDS has been validated for use in prenatal patients and is 86% sensitive and 78% specific for diagnosis of prenatal depression.49
At the same second trimester visit, we determined each subject's intention to breastfeed with written questions asking if she planned to feed her infant breast milk only, mostly breast milk, some formula, mostly formula, some breast milk, or formula only during the first week of infant life or if she was uncertain about what she planned to feed her infant in the first week. Women who planned mostly or exclusively to bottle feed during the first week of life (n
159) were categorized as “planned to formula feed.” Women who indicated that they would only or mostly breastfeed during the first week of life were categorized as “planned to breastfeed.”
Information on initiation of breastfeeding was collected at postdelivery interviews. Participants were asked: “Have you breastfed your baby? By breastfeeding, we mean that you have put your baby to your breast whether or not your baby actually received breast milk, or that you have fed your baby your breast milk?” to report whether or not they had initiated breastfeeding. We defined failure to initiate breastfeeding as a response of No to this question.
We attempted to minimize bias due to self-report of breastfeeding behaviors in many ways. First, Project Viva research assistants who conducted the delivery interview typically had not met with participants prior to the delivery interview, which included the breastfeeding behavior questions. None of the research assistants involved in Project Viva were involved in breastfeeding promotion or education, and none were clinical care providers in any of the hospital or clinic sites associated with Project Viva. Participants were reminded during each of the in-person interviews that their answers were confidential and that their answers would not affect the healthcare provided to them. Research assistants also underwent a lengthy training process on interviewing and were observed several times prior to being allowed to interview participants independently. The script used for the delivery interview was written and piloted to be value neutral to attempt to encourage accurate self-report. Specifically, the research assistant's delivery interview dialogue reads: “This next section is about feeding your baby. I'd like to begin by stressing that there are no right or wrong answers. Please just answer as best you can.”
We estimated odds ratios (ORs) of prenatal intention to formula feed and failure to initiate breastfeeding from logistic regression models fit using SAS version 8.02 software (SAS Institute, Cary, NC). We categorized prenatal depressive symptoms into two groups based on the second trimester EPDS score. We considered women with scores of <13 to have no complaint of prenatal depressive symptoms, and we designated these women as the reference group. We classified women with scores of ≥13 on a 0–30 scale as having prenatal depressive symptoms, in keeping with studies validating the EPDS as a measure of major and minor depression in English-speaking populations.47,49,50
We categorized pregnancy anxiety into two groups based on the number of times a woman chose the response very much to any of the anxiety questions. We have previously used this categorical measure of anxiety because it does not require an assumption of equidistance between response levels of somewhat, moderately, and very much. We classified women with three or more very much responses as the high anxiety group. We categorized all other women as having low-moderate levels of pregnancy-related anxiety.
We considered the following variables as possible confounders or effect modifiers: maternal age, race/ethnicity, maternal country of origin, gestational age, mode of delivery, partner status, household income, and maternal educational attainment.