Breastfeeding has many advantages to infants [1]. In 2010,

approximately 77% of US infants were breastfed at least once, a

substantial increase from 64% in 1998 [2,3]. Despite this progress,

breastfeeding continues to fall short of national goals for duration

and exclusivity set in initiatives such as Healthy People 2020 [2,4].

One possible reason for failure to consistently meet these goals is

the rise in complications women face as they enter pregnancy,

including diabetes, obesity, and hypertension. Breastfeeding

initiation rates are lower and breastfeeding duration is generally

shorter among women with these conditions [5–8]. Six percent of

births are complicated by diabetes [9], 3%–5% of pregnant

women have hypertensive disorders [10–12], and 19%–39% of are

obese when they become pregnant [13]. Clinical management of

these conditions and associated complications may necessitate

greater intrapartum or neonatal intervention, which could affect

care for the woman or infant in the immediate postpartum period,

including breastfeeding [14–19]. Breastfeeding

The decision to breastfeed is highly personal and affected by

many factors, including anticipated barriers to or support for

breastfeeding, hospital practices, medical issues occurring either

PLOS ONE | 1 August 2014 | Volume 9 | Issue 8 | e104820

before or during pregnancy, and complications during labor and

delivery [1,20–26]. One program that has been successful in

encouraging breastfeeding is the Baby-Friendly Hospital Initiative

(BFHI), a global program to encourage and recognize hospitals

that have policies to provide evidence-based care to support infant

feeding and mother-baby bonding [1,20,24,25,27]. The program,

for example, instructs mothers on breastfeeding, allows babies to

spend the first hour after birth in their mothers arms; provides

newborns no food or drink other than breast milk, unless medically

indicated; practices ‘‘rooming in’’ by allowing mothers and infants

to remain together 24 hours per day; gives no pacifiers or artificial

nipples to breastfeeding infants; and refer mothers to breastfeeding

support groups on discharge from the hospital or clinic. Greater

adoption of these practices is also a focus of Healthy People 2020

[28]. Yet despite the success of these measures, fewer than 7% of

U.S. births currently occur in facilities with an official BFHI

designation [28]. This study examines the relationship between

entering pregnancy with complicating health conditions and early

infant feeding behaviors, focusing on women’s breastfeeding

intentions and supportive hospital practices as potential mediators. Breastfeeding

Materials and Methods

Conceptual Model Figure 1 presents the conceptual model for the analysis. The

model focuses on women’s breastfeeding intentions and hospital

support practices during the intrapartum period and how these

factors and their effects may differ for women who enter

pregnancy with diabetes, hypertension or obesity.

Data Data are from the Listening to Mothers III survey, a nationally

representative sample of women who gave birth to a singleton in a

US hospital between July 1, 2011, and June 30, 2012 (N = 2400).

The survey was commissioned by Childbirth Connection and

conducted by Harris Interactive between October and December

2012. The survey documented pregnancy, labor, and birth

experiences in US hospitals, including information about breast-

feeding decisions and pre-existing medical conditions. Data from

this survey have been widely used in clinical and public health

research, including studies of breastfeeding and the role of

supportive hospital practices [26,29,30]. However, this was the

first wave of the survey to include information about medical

conditions prior to pregnancy. Detailed information about the

survey’s methodology, implementation, and questionnaires is

available at Breastfeeding

The data used in this analysis were de-identified. Therefore, the

University of Minnesota Institutional Review Board granted this

study exemption from review (Study No. 1011E92983).

Variable Measurement Pregnancy Complexity. We defined pregnancy complexity

from available survey data relating to 3 common medical risk

factors: (1) taking prescription medication for blood pressure

during the month before pregnancy, (2) having either type 1 or

type 2 diabetes before pregnancy or gestational diabetes, or (3)

having a prepregnancy body mass index higher than 30. Our main

analysis included a dichotomous measure of pregnancy complexity

for women reporting any of these 3 conditions. We also

constructed indicators for each of the conditions for separate

analysis (see following description of sensitivity analyses). Breastfeeding

Breastfeeding Intention. Women were asked at the time of

the survey to recall their intentions about infant feeding at the end

of pregnancy. We created dichotomous variables indicating (1) any

intent to breastfeed (exclusively or not) and (2) women’s intent to

breastfeed exclusively. Supportive hospital practices and infant

feeding status were assessed among women who reported any

intention to breastfeed (n = 1990), and exclusive breast milk

feeding status at 1 week postpartum was assessed among women

who intended to exclusively breastfeed (n = 1418).

Supportive Hospital Practices. Among women who in-

tended to breastfeed, we examined supportive hospital practices

consistent with BFHI standards. We measured supportive hospital

practices using an 8-point composite measure corresponding to 7

of the 10 BFHI steps. Measures for the remaining 3 steps were not

assessed in the Listening to Mothers surveys because they require

knowledge of hospital administrative policies beyond the scope of

women’s knowledge and experiences. However, data from these

Figure 1. Conceptual Model. doi:10.1371/journal.pone.0104820.g001

Medically Complex Pregnancies and Early Breastfeeding


PLOS ONE | 2 August 2014 | Volume 9 | Issue 8 | e104820

surveys have previously been used to successfully approximate Breastfeeding

BFHI hospital practices [26,30]. See Table 1 for detailed

information about the 10 BFHI steps and the 8 items assessed in

the data and used in this analysis.

To assess general concordance with supportive breastfeeding

practices in the hospital, we created a composite measure in which

higher scores indicate that the woman perceived a higher level of

breastfeeding-supportive hospital practices. Scores were not

normally distributed, so we constructed a dichotomous variable

on the basis of the top quintile of responses. Scores of 7 to 8 were

categorized as ‘‘high hospital support,’’ indicating practices

broadly consistent with BFHI standards. We also assessed the

distribution of the items in the composite measure and tested the

stability of the measure by modeling hospital support as a

continuous variable (0–8) and by using a lower threshold (i.e.,

scores of 6–8 for high levels of support from the hospital). Results

were robust to alternative specifications. Breastfeeding

Feeding Status 1 Week Postpartum. Two dichotomous

measures of infant feeding status were based on women’s responses

to questions regarding (1) whether they were feeding their

newborn any breast milk (either exclusively or in combination

with formula) 1 week postpartum, and (2) whether they were

feeding their newborn breast milk only 1 week postpartum. This

definition allows for both direct breastfeeding and feeding

expressed breast milk to infants. Breastfeeding

Control Variables. We controlled for labor and delivery

factors that may affect the initiation of breastfeeding, including

cesarean delivery, epidural use, and admission to a neonatal

intensive-care unit [31–34]. We assessed these variables from

maternal self-report. We also included several self-reported

sociodemographic and birth-related covariates, including age;

race/ethnicity (white, black, Hispanic, or other/multiple race);

education (high school or less, some college, bachelor’s degree, or

graduate education); 4-category census region (Northeast, South,

Midwest, West); nativity (foreign- or US-born); partnership status

(unmarried with no partner, unmarried with partner, or married);

parity (first-time pregnancy); pregnancy intention (unintended or

intended pregnancy); agreement with the statement ‘‘birth is a

process that should not be interfered with unless medically

necessary;’’ doula support; and primary payer for maternity care

(private, public, or out-of-pocket). Breastfeeding

Analysis We first explored associations between the predictors, outcomes,

and covariates for the overall sample using 1- and 2-way

tabulation. We used Pearson’s x2 tests to determine whether differences based on pregnancy complexity were statistically

significant. We used logistic regression to estimate the adjusted

odds of breastfeeding intention based on pregnancy complexity.

Among women intending to breastfeed, we estimated the adjusted

odds of breastfeeding status 1 week postpartum. To test for

mediation by hospital support, we added a variable indicating high

levels of support for breastfeeding at the hospital. In the final

multivariate models of breastfeeding status 1 week postpartum, we

included only covariates that were statistically significantly

associated with the outcomes. We conducted sensitivity analyses,

estimating the same regression models using indicator variables for

prepregnancy obesity, hypertension, and diabetes as the predictors

rather than the combined ‘‘complex pregnancy’’ variable; results

were substantively unchanged. All analyses used a p-value of 0.05

to determine statistical significance, were conducted using Stata

v.12, and weighted to be nationally representative Breastfeeding

Also check: Health Policy