Description of setting and sample
Data were collected from January 2008 to June 2010 in Ontario, Canada at five sites: two small community hospitals (~ 300-400 births per year each), a large community hospital (~ 2500 births per year), and a teaching hospital with two sites (~6800 births combined sites) [34
]. One hundred and sixty-four women registered for the study, and 109 families completed data collection. Reasons for loss of registered participants included: 37% who intended to continue but did not receive a baby scale; 23% who stopped because of illness (e.g., infant's prematurity, unexpected caesarean section, postpartum hemorrhage); 17% who changed their mind about the study or opted to formula feed; and for 23% the exact reason is not known. The group that completed the study was comparable to the lost participants based on age, amount of breastfeeding, maternal education level, and family income (using independent T-test and chi-squared, p < 0.05). There was a difference between the two groups based on being in a committed relationship (100% versus 88% respectively, p > 0.05). Demographic characteristics of the participants, mothers and newborns who completed the study, are in Table .
Characteristics of mothers and newborns
Weight loss and maternal fluids
Tables and present the average weight loss from birth (in grams and by percentage) and the amounts of maternal fluid intake during parturition, respectively. Bivariate analyses compared millilitres of maternal intake with grams of newborn weight lost. Oral fluids alone were not significant. Intravenous fluids (both total from admission and two-hour pre-birth amounts) and combined IV and oral fluids were statistically significant at 60 hours (see Table ).
Average weight loss from birth (N = 109)
Amounts of maternal fluids in millilitres (N = 109)
Newborn weight loss in grams correlated to maternal fluid types (N = 109)
Maternal fluids and neonatal output
Details of neonatal output are presented in Table . Discrete outputs (i.e., daily total amounts of voids and stools) on Day 1 were positively correlated to two-hour pre-birth IV amounts, but not to other fluid categories. Neither Day 2 nor Day 3 neonatal outputs were correlated to any maternal IV fluids amounts (see Table ).
Average neonatal output (N = 109) in grams
Maternal fluid amounts correlated to neonatal output (N = 109)
Neonatal output and newborn weight loss
Using bivariate analyses, newborn weight loss as a percentage for each 24 hour period was correlated to neonatal output for the same period (see Table ). For Day 1, there was a positive relationship between neonatal output and newborn weight loss (i.e., as weight of diaper increases, weight loss increases). This result indicates that as neonatal output increases, newborn weight loss also increases during the first 24 hours. On Day 2, there was no relationship between the two variables. On Day 3, there was a statistically significant negative correlation between output and weight loss. Since we are looking for weight loss, this final result is essentially a double negative indicating as weight of diaper increases, weight loss decreases (i.e., newborn gains).
Percentage of newborn weight loss correlated to neonatal output (N = 109)
Comparing based on groups
By forming two groups based on total amount of fluids received, we could compare the two groups and look for differences in amounts of newborn weight loss. We used the point of maximum weight loss (60 hours) and chose 1200 mls as the maternal fluid amount to compare. Our rationale was multifold: 1) the 25th percentile of fluids was 1200 mls; 2) this amount represented the median of oral fluids plus half of the IV amounts; and 3) most births were within 12 hours of admission - 100 mls per hour is a reasonable maintenance quantity. For the mothers who had 1200 mls or less, the average percentage of newborn weight loss at 60 hrs was 5.51% (n = 21). Whereas the group with more than 1200 mls total fluids, their babies averaged a 6.93% weight loss (n = 53). The difference of 1.42% was statistically significant (p = 0.03).
Women were asked if they noticed the day their "milk came in" (birth counted as Day 1), and 95% reported the day they noticed first day of breast fullness. There was a significant positive correlation between late onset of lactogenesis II (> Day 3, reported by 41% of sample) and percentage of newborn weight loss at 72 hours (rs(97) = 0.380, p < 0.001, two-tailed Spearman's rho). Additionally, the reported late onset of lactogenesis II was positively related to the total amounts of maternal fluids from admission to birth (rs(78) = 0.307, p = 0.006, two-tailed Spearman's rho).
In addition to the two independent variables of interest, maternal fluids and neonatal output, several variables were identified in the literature as predictors of weight loss. The variables cited in the literature were used in the conceptual model (see Figure ): parity [35
] gestational age [9
], oxytocin [induction or augmentation] use [35
], epidural use [36
], type of birth [37
], infant sex, [9
] birth weight [9
], feeding type (i.e., supplemented) [9
], timing of lactogenesis II [38
], and time with skin-to-skin [39
Bivariate analyses show a positive correlation between weight loss at 60 hours and the two key independent variables, maternal two-hour pre-birth IV fluids (rs(38) = 0.383, p = 0.018) and the first day of neonatal output (rs(95) = 0.287, p = 0.005, two-tailed Spearman's rho). When these two variables are analyzed together in a regression analyses, the fluid remains significant (p = 0.05), but output is not significant (p = 0.202). Using the variables in this model and holding neonatal output at 94.82 mls, we predict that 250 mls of maternal fluids results in 5.78% weight loss; whereas, 2500 mls of maternal fluids results in 8.03% weight loss.
We also ran a multiple linear regression analysis to determine predictive variables for percentage of weight loss at 60 hours - the nadir of loss. Gestational age, birth weight, and onset of lactogenesis II were included in the model with two-hour pre-birth IV fluids. We eliminated variables when a bivariate analysis resulted in p-values of greater than 0.20. Consequently, parity, oxytocin, infant sex, and supplemented (yes/no) were not included in the model. Epidural (p = 0.157, n = 95) and birth type (p = 0.008, n = 95) met the initial criteria, but both variables were strongly associated with the two-hour pre-birth IV fluids (r > 0.70, p < 0.001) and thus excluded. There was inadequate data to include skin-to-skin in the model.
Overall, the regression model was significant (see Table for details). The two-hour pre-birth maternal fluids remained predictive of percentage of weight lost when gestational age, birth weight, and timing of lactogenesis II were included in the model. Gestational age, birth weight, and timing of lactogenesis II were not predictive at 60 hours postpartum. In contrast, at 72 hours, the maternal fluids were not predictive (p = 0.384), but timing of lactogenesis II was predictive (p = 0.019; model not shown). When the dependent variable in the model was changed to grams of weight lost at 60 hours, both maternal two-hour pre-birth IV fluids and birth weight were predictors of weight loss (p = 0.022 and p = 0.036, respectively).
Regression analysis of predictor variables for percentage of neonatal weight loss at 60 hours postpartum (n = 37)