Study design
This was a longitudinal study conducted in Athens-Clarke County and surrounding counties of the State of Georgia between November 2005 and December 2006. Participants recruited from the above mentioned areas reported to the Maternal and Child Nutrition Research Lab (MCNRL), Department of Foods and Nutrition, at The University of Georgia (UGA), USA, for interviews and body composition measurements. Participants were interviewed using a structured questionnaire containing both closed- and open-ended questions, which was followed by body composition measurements.
Immediately after the interview, the participant changed into a swimsuit for the body composition measurement. Height and weight were measured before entering the BOD POD for the body composition measurement. The study involved five testing days over a period of four months (36 weeks pregnancy, 2, 4, 8, and 12 weeks postpartum). All participants gave written consent before being recruited into the study after being thoroughly informed of the purpose, requirements and procedures of the study. They were also given the opportunity to read the informed consent form. The study protocol and consent form were approved by the Human Subjects Institutional Review Boards (IRBs) of the University of Georgia (UGA) and the Athens Regional Medical Center (ARMC).
Research participants were recruited via newspaper advertisements, flyers posted throughout the UGA campus, mailings to specific listservs, study brochures at doctors' offices and by word of mouth. Participants were screened for the study inclusion and exclusion criteria.
The primary outcomes were postpartum maternal weight and percent body fat changes, but we also assessed dietary intake and physical activity levels. Sample size was based on the detection of differences in postpartum maternal weight and percent body fat changes among two groups (exclusive breastfeeding and formula feeding) equivalent to 1% body fat. With a type 1 error of 0.05 and a 0.8 probability of detecting a true difference (1 – β), the required sample size per group was 8 [18]. However, because many participating mothers who were formula feeding also gave some breast milk, we ended up with exclusively breastfeeding and mixed feeding groups instead of a formula feeding group.
Eligibility and exclusion criteria
To be eligible for participation, a participant had to be at least 18 years of age, in their third trimester and not less than 36 weeks pregnant. They had to be non-smokers and not on medication which could affect their body composition such as steroids. Exclusion criteria included complications during pregnancy and delivery, a diagnosis of hypertension, pre-eclampsia or gestational diabetes, as well as pre-term delivery (≤ 37 weeks).
Study implementation and questionnaires
We collected baseline data at the time of recruitment and follow-up data at 2, 4, 8, and 12 weeks postpartum. All interviews and measurements took place in the MCNRL. During each evaluation, interviews were conducted using both closed and open – ended questionnaires to collect information on health history, infant feeding practice, and physical activity – the type of exercise, its intensity and frequency. Physical activity level was assessed based on the recommendations of the 2005 Dietary Guidelines for Americans [19]. A score for physical activity was derived by multiplying the number of days one exercised within a week by the duration of the activity [20]. For example, a participant who exercised 30 minutes a day, 3 times a week would have an activity score of 1.5 (3 × 0.5). Sedentary or inactive was defined as having an activity score of 0 – 1.4, moderately active was 1.5 – 2.9 and very active was > 3.0. Information was also gathered on dietary habits. On each visit we performed a 24 hour (24-h) dietary recall and calculated the energy and nutrient intake from foods consumed. We also used a food frequency questionnaire (FFQ) alongside the 24-h dietary recall to assess general food intake over a month period. With the 24-h dietary recall, participants were asked to recall every food they ate from morning until evening of the day before evaluation. The recall also included information on portion size as well as brand name and preparation method for each food. Overall energy intake, major macro- and micro-nutrient intakes were assessed using the Nutrition Data System for Research software. Intake of alcohol and dietary fiber were also assessed. The FFQ asked questions about specific foods and how often they were consumed (daily, weekly or monthly), and this information was used as a validation of the information collected by the 24-h dietary recall.
Body composition measurement
Body composition measurements were conducted using the BOD POD body composition system version 2.30 (Life Measurement Inc, Concord, CA). Participant height was measured to the nearest 1.0 mm using Seca 214 portable stadiometer (Itin Scale Inc, Brooklyn NY) and done in duplicates. Participant weight was measured wearing minimal clothing (bathing suit) and without shoes or jewellery on the body. Weight was measured to the nearest 0.01 kg using a calibrated electronic scale connected to the BOD POD system. Weight measurement was done during the 2-point calibration procedure for the BOD POD, described elsewhere [21].
Once these measurements were done, the computer calculated the corrected body volume (Vbcorr) which was used in calculating body density (Db) [22]. The BOD POD software then calculated the percent body fat using the Siri equation [23] for Caucasians and the general population or the Ortiz equation [24] for African Americans.
Outcome variables
The main outcome variables were changes in maternal total body weight and percent body fat. These were assessed at each evaluation time point. Changes in weight and percent body fat loss were calculated by subtracting weight or percent body fat at 2 weeks postpartum from that recorded at each evaluation time point.
Independent variables
Different one-time covariates were assessed at 36 weeks pregnancy and others at 2 weeks postpartum. These included educational background, delivery type, parity, marital status, age, pregnancy weight gain, and ethnicity. A number of time varying covariates were also assessed at each evaluation time point. These included dietary/energy intake, physical activity and infant feeding practice. The categories of infant feeding investigated in this study were exclusively breastfeeding and mixed feeding. Exclusive breastfeeding was defined as the "infant receiving only breast milk; no other liquid or solid is fed." Mixed feeding was defined as the "infant receiving some breast milk and formula as well as some solids whatever the case may be".
Statistical analysis
We performed all data entry and analyses using SPSS for Windows version 15.0 (SPSS Inc, Chicago, IL). Dietary information was analyzed using the Nutrition Data System for Research software (Nutrition Coordinating Center, University of Minnesota School of Public Health). Differences in variables by group were evaluated using t tests and analysis of variance (ANOVA) for continuous variables, and chi-square tests for categorical variables. Chi-square analyses were used to examine bivariate associations between the outcome variable and independent variables. Since the study has a longitudinal design with repeated measures over time, we conducted repeated measures analysis to assess trends in body composition changes. Multiple linear regression analysis was conducted to identify the relationship between maternal characteristics and the changes in maternal percent body fat over time. Data are reported as means ± SD. P value of ≤ 0.05 was used as a criterion for reporting statistical significance.