Participants
Twenty pregnant women and eighteen non-pregnant, non-lactating control women volunteered to participate in this observational research study. Two pregnant women were excluded from data analysis due to incomplete data or development of a gestational condition (e.g., gestational diabetes, preeclampsia) that could alter fluid balance. All pregnant women enrolled in the study by 16 weeks of gestation and had singleton pregnancies. Participants were excluded for tobacco use, participation in strenuous exercise (≥7 h per week), or presence of a health condition (e.g., diabetes, kidney dysfunction or disease, urinary tract infection, electrolyte abnormality, etc.) or prescription of a medication (e.g., diuretics) that could alter fluid balance. The university institutional review board approved this investigation, and all participants provided written informed consent. All study procedures were performed in accordance with ethical standards specified by the Declaration of Helsinki.
Study design
Data were collected from pregnant and subsequently lactating participants at five time points: (1) the end of the first (15 ± 2 weeks gestation) (2) second (26 ± 1 weeks gestation), and (3) third trimesters (37 ± 1 weeks gestation) as well as at (4) 3 ± 1 weeks and (5) 9 ± 1 weeks postpartum. Control women were matched to pregnant participants on the basis of age, height, and weight; these participants visited the laboratory at similar time intervals to the pregnant and lactating women and only during the early follicular phase of the menstrual cycle. This was done to limit effects of exogenous estrogen on fluid balance [
24] and to ensure each visit occurred at the same point in their cycle. All control women were taking a combination drug oral contraceptive.
Participants reported age, pre-pregnancy body mass, and parity. Researchers instructed participants to eat and drink according to their normal habits throughout the investigation and also instructed participants how to maintain an accurate fluid intake record. Participants recorded the fluid they consumed 1 day before each visit to allow calculation of 24-h total fluid intake (TFI; water consumed from drinking water and other beverages). Total fluid intake was chosen as the variable of interest given its practicality, ease of measurement by women in their daily lives, and that it generally accounts for 80 % of total water intake (i.e., water from both foods and fluids), which requires dietary analysis. During this time, participants also collected their urine for 24 h using the following procedure. On the morning before each laboratory visit, participants awoke, voided, and discarded this first morning void. All subsequent voids produced throughout the day and overnight were collected in a single urine collection container. On the morning of the study visit, participants awoke, voided, and included this first morning void to complete a full 24-h collection. Participants then returned the urine collection container to investigators at the beginning of each laboratory visit; no participant reported missed voids.
Each laboratory visit occurred during the morning and at a similar time across all five visits. Participants were fasted for at least 4 h before each visit but were allowed to consume water ad libitum. At the beginning of each visit, researchers recovered the 24-h urine collection, reviewed each participant’s fluid intake record, checked any entries that were unclear, and probed for possible omissions. Participants measured fluid volumes with wet measuring cups when the exact volume was not available on a container and recorded their fluid logs in real time throughout the 24 h. Body mass was recorded to the nearest 0.01 kg (Health O Meter, Model 349KLX, Alsip, IL), and height was measured on the first visit via stadiometer. Body mass index was calculated as body mass (kg) divided by height2 (m2). Venous blood was drawn into Vacutainer® (BD Biosciences, Franklin Lakes, NJ) tubes without additive (centrifuged for serum) and with K2EDTA (for whole blood). Osmolality (S
OSM) via freezing point depression and total protein (S
TP) via refractometer were measured in the serum. Hematocrit (Covidien, Microhematocrit Tube Reader, Minneapolis, MN) was measured in K2EDTA-treated whole blood.
The 24-h urine sample was assessed for volume (
U
VOL) gravimetrically (Ohaus Corporation, Ranger 3000, Parsippany, NJ), osmolality (
U
OSM) by freezing point depression (Advanced Instruments Inc., Model 3320, Norwood, MI), and specific gravity (
U
SG) by manual refractometer (Reichert Technologies, Model TS-400, Depew NY). Urine color (
U
COL) of 24-h samples was observed by a single investigator in a well-lit room by observing urine in a clear container against a white background, adjacent to a previously published 8-category urine color chart [
25]. The investigator recorded the number of the color that best matched the sample; the darker of two colors was recorded if the sample color was determined to fall between two colors on the chart for consistency of technique. At visit 4 (3 ± 1 weeks postpartum), lactating women also recorded 24-h breast milk volume (
M
VOL), measured directly at time of expression if pumped, or estimated by test-weighing the infant to the nearest 10 g (Newline Digital Weight Track, Model SHAEBSA-20, Edgewood, NY) at each feeding as previously described [
20], during the same 24 h in which they collected urine and recorded their fluid consumption. Research staff trained the mothers on how to use the infant scale and on the procedure for test-weighing infants before and after feedings.
Statistical methods
Data analyses were performed with SAS (version 9.4, SAS Institute Inc., Cary NC). Differences in anthropometric and demographic characteristics were assessed at the first visit during pregnancy (visit 1) and during lactation (visit 4) with independent samples t tests. We utilized a repeated measures analysis of covariance with three covariates (TFI, group, and time). Individual intercepts and slopes of TFI were fitted for each combination of group and time. F tests were performed to determine whether there was a significant relationship (nonzero slope) between TFI and each hydration biomarker. If a significant relationship (nonzero slope) was present, homogeneity of regression coefficients via F tests determined whether the relationships (slopes) were similar between groups and time. For group comparisons, pregnant (PREG; end of the first, second, and third trimesters) were compared to non-pregnant (NP; visits 1–3) women, and lactating (LACT; 3 ± 1 and 9 ± 1 weeks postpartum) were compared to non-lactating (NL; visits 4–5) women to test for between-group differences, within each visit. Time comparisons of slopes were performed separately for pregnancy (visits 1–3) and lactation (visits 4–5), within each group (PREG, LACT, NP, NL). If significant differences between groups or time were noted, post hoc comparisons with a Bonferroni adjustment were utilized to determine where differences in the relationship existed. We utilized a simple linear regression to determine whether a relationship between TFI and M
VOL was present in LACT at 3 ± 1 weeks postpartum.