Lactation

Lactation history was assessed when women enrolled in WISH by asking women about each of their children, “How old was your child when you stopped nursing?” Women responded with the number of months that they breastfed each child. Irrespective of whether a mother chooses to breastfeed, her body initiates lactogenesis after giving birth. Because under the 1993 Family and Medical Leave Act, US working women are granted only 12 weeks’ unpaid time off to care for a newborn, many US mothers find it difficult to breastfeed for more than 3 months. Thus, although the American Academy of Pediatrics recommends that women breastfeed for the first year of their infant’s life, we considered a mother to have successfully breastfed if she breastfed for 3 or more months and to have prematurely discontinued lactation if she breastfed any of her children for less than 3 months (or never breastfed at all). Thus, we considered 3 groups of mothers: those who breastfed each of her children for 3 or more months, those who had breastfed any of their children for less than 3 months, and those who had never breastfed any of their infants. Mothers’ reasons for weaning were not assessed.

Subclinical cardiovascular disease

Average carotid artery intima-media thickness (IMT) was assessed using an Acuson Sonoline Antares high-resolution duplex scanner (Siemens, Malvern, PA). Images were digitized from the near and far walls of the distal common carotid artery (1 cm proximal to the carotid bulb), far wall of the blub, and first centimeter of the far wall of the internal carotid artery (a total of 8 locations). Intima-media thickness measures were performed by electronically tracing the lumen-intima interface and the media-adventitia interface across 1 cm segments in these locations. The mean of all average readings across the 8 locations comprise the average carotid IMT.

Common carotid artery adventitial diameters were measured in the same way using the adventitial-medial interface on the near wall and the medial-adventitial interface on the far wall of the common carotid artery. Lumen diameter was measured as the distance between the lumen-intima interfaces. Carotid-femoral pulse wave velocity (cf-PWV) measures were automatically generated using a noninvasive and automated waveform analyzer (Complior SP; Artech Medical, Pantin, France). The Complior records the carotid and femoral pulse waveform using multiarray tonometers. Following 10 minutes of rest in a supine position, sonographers palpate the right femoral artery and the right carotid artery and place the handheld tonometers over these 2 pulse areas to obtain femoral and carotid pulse waveforms simultaneously. A foot-pedal is used to start the recording.

Pulse wave velocity (PWV) was calculated by time-phase analysis using volume waveforms of respective arteries (carotid and femoral arteries). PWV was calculated as the distance between arterial sites divided by the time between the foot of the respective waveforms. The distance between the sampling sites was measured over the surface of the body with a tape measure (this measure is highly reproducible between sonographers with a correlation of 0.93.) Three distances were measured: (1) from the suprasternal notch to the sampling site on the right common carotid artery; (2) from the suprasternal notch to the inferior edge of the umbilicus; and (3) from the inferior edge of the umbilicus to the sampling site on the right common femoral artery. The final carotid to femoral distance was calculated by subtracting measurement 1 from the sum of 2 and 3.

Three runs of data are used and results were averaged to reduce measurement variability. Replicate readings were performed in the ultrasound laboratory to evaluate reproducibility of the measures. The intraclass correlations were 0.88-0.99 for IMT, 0.91-0.96 for adventitial diameter, and 0.78-0.91 for cf-PWV.

Covariates

Delivery characteristics, including gestational age and infant birthweight, were abstracted from hospital birth records. Birth outcomes were categorized as being preterm, SGA, or term non-SGA infants. Years since last live birth were calculated by subtracting the date of last live birth from the date of the WISH participant visit. At the WISH participant visit, data on race/ethnicity (non-Hispanic Black, other); marital status (married or marriage-like, unmarried); maternal education (high school or less, some college or more); insurance (private, other); annual household income (<$50,000, ≥$50,000); smoking (number of cigarettes per day); multivitamin use (yes, no); and parity (number of live births) were ascertained.

Family history of diabetes, myocardial infarction, and stroke was assessed and included disease in participants’ mother, father, brothers, or sisters. Women reported the outcomes of all pregnancies before and following the target pregnancy including gestational age and were categorized as having 1 or 2 or more preterm births. Maximum weight gain during pregnancy was assessed by asking women who reported 1 or more live births, “How much weight did you gain during this pregnancy?” The maximum value across all pregnancies was used. Optimism and anxiety were measured using validated instruments (the Life Orientation Test [LOT] and the State Anxiety Score of the Spielberger State-Trait Anxiety Inventory, respectively).

Physical activity was assessed with the Pfaffenberger Physical Activity Scale and is reported as metabolic equivalent task (MET) hours per week. Height and weight were obtained by standard methods, with body mass index (BMI) calculated as weight in kilograms divided by height in meters squared. Early adult BMI was calculated as weight in kilograms collected by self-report with the question, “Approximately how much did you weigh when you left high school?” divided by current height in meters squared.

Fasting blood samples were collected and all measurements were completed at the Nutrition Laboratory in the Department of Epidemiology at the University of Pittsburgh, which is Clinical Laboratory Improvement Amendments certified and participates in the Centers for Disease Control and Prevention–National Heart, Lung, and Blood Institute Lipid Standardization and College of American Pathologists’ Proficiency Programs. Total cholesterol, high-density lipoprotein (HDL), and triglycerides were measured using standard enzymatic procedures. Low-density lipoprotein (LDL) was estimated using the Friedewald calculation, glucose was determined by an enzymatic determination, and insulin was measured using an radioimunnoassay procedure developed by Linco Research Inc. (St. Charles, MO). C-reactive protein (CRP) was measured using a high-sensitivity turbidimetric method (reagents developed by Carolina Liquid Chemicals , Brea , CA). Blood pressure was evaluated as the mean of 3 measurements following a 10 minute rest.

Statistical analysis

We used analysis of variance and χ ² tests to examine whether sociodemographic characteristics, cardiovascular risk factors, or measures of subclinical cardiovascular disease varied by lactation history. Because of the skewing of the distribution, IMT was dichotomized at the 75th percentile or less and greater than the 75th percentile (≤0.60 mm and >0.60 mm, respectively). Multivariable logistic regression was used to estimate the associations between lactation history and IMT.

Because lumen diameter, adventitial diameter, and carotid-femoral pulse wave velocity values were normally distributed, multivariable linear regression was used to estimate associations between lactation and lumen and adventitial diameter and lactation and cf-PWV. After initial adjustment for age, parity, birth outcome, gestational age, infant birthweight, additional preterm births (yes/no), years since last birth, socioeconomic covariates (race, education, income), and lifestyle variables (smoking [cigarettes per day]), physical activity, vitamin supplementation, optimism, anxiety), early adult BMI, maximum gestational weight gain, and family history (of diabetes, myocardial infarction, or stroke), a subsequent model included current BMI and traditional cardiovascular risk factors (systolic blood pressure, total cholesterol, HDL, triglycerides, CRP, glucose, and insulin).

Covariates that were not normally distributed were natural log transformed for entry into regression models. Potential issues of collinearity were examined using variance inflation factors with 10 or greater, indicative of collinearity. Subjects with missing covariate data were dropped from analyses involving that covariate. Effect modification by birth outcome was assessed using a likelihood ratio test ( α = 0.10). Analyses were performed with SAS (version 9.2; SAS Institute, Inc, Cary, NC). All tests were 2 sided with statistical significance level at P = .05.