Objective. Some mothers have difficulty initiating lactation even when highly motivated to breastfeed. The purpose of this study was to determine the incidence of and risk factors for suboptimal infant breastfeeding behavior (SIBB), delayed onset of lactation, and excess neonatal weight loss among mother-infant pairs in a population with high educational levels and motivation to breastfeed.
Methods. All mothers residing in Davis, California, who gave birth to a healthy, single, term infant at 1 of 5 area hospitals during the 10-month recruitment period in 1999 were invited to participate if they were willing to attempt to breastfeed exclusively for at least 1 month. Lactation guidance was provided and data were collected in the hospital (day 0) and on days 3, 5, 7, and 14. Infant breastfeeding behavior was evaluated by trained lactation consultants using the Infant Breastfeeding Assessment Tool. Onset of lactation was defined based on maternal report of changes in breast fullness. Infant weight loss was considered excessive if it was ≥10% of birth weight by day 3.
Results. Of the 328 eligible mothers, 280 (85%) participated in the study. The prevalence of SIBB was 49% on day 0, 22% on day 3, and 14% on day 7. SIBB was significantly associated with primiparity (days 0 and 3), cesarean section (in multiparas, day 0), flat or inverted nipples, infant status at birth (days 0 and 3), use of nonbreast milk fluids in the first 48 hours (days 3 and 7), pacifier use (day 3), stage II labor >1 hour (day 7), maternal body mass index >27 kg/m2 (day 7) and birth weight <3600 g (day 7). Delayed onset of lactation (>72 hours) occurred in 22% of women and was associated with primiparity, cesarean section, stage II labor >1 hour, maternal body mass index >27 kg/m2, flat or inverted nipples, and birth weight >3600 g (in primiparas). Excess weight loss occurred in 12% of infants and was associated with primiparity, long duration of labor, use of labor medications (in multiparas), and infant status at birth. The risk of excess infant weight loss was 7.1 times greater if the mother had delayed onset of lactation, and 2.6 times greater if the infant had SIBB on day 0.
Conclusions. Early lactation success is strongly influenced by parity, but may also be affected by potentially modifiable factors such as delivery mode, duration of labor, labor medications, use of nonbreast milk fluids and/or pacifiers, and maternal overweight. All breastfeeding mother-infant pairs should be evaluated at 72 to 96 hours’ postpartum.
The first week postpartum is a critical period for the establishment of breastfeeding. Normally, the amount of milk produced is minimal for the first 1 to 2 days’ postpartum, but increases dramatically by ∼2–3 days’ postpartum as lactogenesis II occurs in response to the drop in progesterone after delivery.1 During this time, both the mother and the infant are learning how to breastfeed. Sociocultural factors are strongly associated with the initiation of breastfeeding, but lactation problems are common even among mothers who are highly motivated to breastfeed. Problems such as delayed onset of lactation2 and suboptimal breastfeeding behavior among newborns, especially those exposed to labor medications during delivery,3 are frequently reported. If the situation is not handled appropriately, inadequate milk transfer can lead to excessive infant weight loss, dehydration, and serious medical complications, even death.4 Some reports suggest that the incidence of breastfeeding malnutrition has increased as shorter hospital stays have become more common.5 Although serious adverse outcomes are rare, lactation difficulties during the first week postpartum are associated with greater risk of early termination of breastfeeding6,7 and lower breastfeeding success with subsequent children.6
On average, breastfed newborns in industrialized countries lose 5% to 7% of birth weight in the first few days of life.8,9 Most clinicians judge weight loss to be of concern when it exceeds ∼10% of birth weight. There has been very little research on the incidence of excess weight loss or the lactation problems associated with it. In a hospital-based study in Italy, 8% of exclusively breastfed newborns lost ≥10% of birth weight during the first 3 to 5 days.9 The investigators determined that 26% of these cases were attributable to inadequate maternal milk volume and 74% were attributable to poor breastfeeding technique by either the mother or infant. Delayed onset of milk production (>72 hours’ postpartum) was reported by 31% of breastfeeding mothers in Connecticut.2 Several risk factors have been associated with delayed onset, including primiparity,2,10,11 cesarean section delivery,2,11,12 stress during labor and delivery,2,10,13 maternal diabetes,14 and high maternal body mass index (BMI).2,15 These studies have demonstrated that physiologic factors, not just behavioral factors, can strongly influence early lactation success.
To date, no study has documented all of the key breastfeeding outcomes simultaneously during the first week postpartum (infant breastfeeding behavior, onset of milk production, and infant weight loss), and none has used a community-based sample of women. Because of this lack of information, we undertook a prospective, community-based study of early lactation success in a population where motivation to breastfeed exclusively is very high. Our objectives were to determine the incidence of, and the risk factors associated with, poor lactation outcomes in the absence of sociocultural barriers to exclusive breastfeeding.
Study Design and Recruitment of Subjects
All mothers who gave birth between February and December, 1999 were recruited for the study if they met the following selection criteria: a) residence in Davis, California; b) mother willing to attempt to breastfeed exclusively for at least 1 month; c) single infant >37 weeks’ gestation, with no significant perinatal morbidity; d) English-speaking; and e) phone in the home. Subjects were recruited from 5 area hospitals that collectively deliver >95% of all births to women residing in Davis. Before and during the recruitment period, the study was publicized throughout the community via newspaper articles, contacts with health care providers, and letters sent to pregnant women. Recruitment occurred by contacting each of the 5 hospitals daily, 7 days a week, to identify potential subjects, who were then visited in the hospital by a member of the research team, usually within 24 hours after delivery (94%). Mothers who delivered at another hospital or at home were also potentially eligible for the study if we were notified of the birth within 72 hours.
On recruitment, an International Board Certified Lactation Consultant and a research assistant completed the initial interview, measured the infant’s weight, collected relevant data from the medical chart and observed the infant breastfeeding (whenever possible). The lactation consultant provided breastfeeding guidance emphasizing a) correct positioning and latching on to the breast, b) the importance of demand feeding and avoidance of supplements, and c) contact information if difficulties or questions arose. On days 3, 7, and 14 (counting 0–24 hours as day 0) subjects were visited in their homes (or in the hospital if they had not yet been discharged by day 3) to collect additional data and provide ongoing lactation guidance. In addition, mothers were contacted by phone on day 5. Guidance provided by the 7 lactation consultants was standardized through an initial training period and by weekly meetings to discuss challenging cases. The research protocol was approved by the Human Subjects Review Committee at the University of California, Davis.
At the time of the hospital visit, information was collected from the medical record concerning the duration of stage I labor (from the time regular contractions began to cervical dilation of 10 cm) and stage II labor (from full dilation to the delivery of the infant), as well as mode of delivery, labor medications, presence of meconium in amniotic fluid, infant resuscitation, gravida, parity, pregnancy weight gain, and infant birth weight. The accuracy of scales used to measure birth weight in each hospital was checked using standard weights at the beginning of the study and every 3 months thereafter. During the hospital interview, information was collected from the mother regarding: a) the interval without sleep before delivery; b) infant feeding patterns since birth (when first put to breast; infant interest in nursing; number of breastfeeds; use of supplemental fluids such as water, glucose water, or formula; and use of a pacifier); c) presence of a support person during labor and delivery; d) previous experience breastfeeding; and e) her confidence in her ability to exclusively breastfeed her infant for at least 4 weeks (on a scale from 1 = no confidence to 5 = very confident). During the breastfeeding observation, the lactation consultant recorded data on nipple type (normal versus flat or inverted), maternal and infant positioning, suckling behaviors and mother’s report of pain during the feeding. The Infant Breastfeeding Assessment Tool (IBFAT)16 was used to rate the infant’s breastfeeding behavior. The IBFAT includes ratings for arousal, rooting, time to latch and feed well (fixing), and suckling effectiveness, each worth 0 to 3 points (0 = did not exhibit target behavior; 3 = readily exhibited target behavior) for a total maximum score of 12. Matthews16 reported interrater reliability for the IBFAT to be 91% between mothers and trained observers. To enhance the accuracy of recall data collected during subsequent contacts, each mother was provided with nursing diary forms for the next 7 days and encouraged to record the timing of each feeding, the infant’s breastfeeding behavior, the time when her breasts first felt noticeably fuller, and any supplements given to the infant.
As soon as possible after completing 72 hours’ postpartum (ie, on day 3), each mother was visited to collect information on: a) the infant’s breastfeeding behavior and nursing frequency; b) her level of breast fullness (on a scale from 1 = no change, to 3 = noticeably fuller, to 5 = uncomfortably full) and the approximate time postpartum when breast fullness first reached 3 (if this had occurred); c) supplemental fluids given to the infant during each of the past 3 days; d) use of a pacifier; e) the level of support she received for breastfeeding from family and friends (1 = most not supportive; 2 = some supportive, some not; 3 = a lot of support); and f) any difficulties or discomfort she experienced with breastfeeding. The infant’s weight was measured on an electronic scale accurate to the nearest gram, which was validated daily using standard weights. The lactation consultant observed a breastfeed, recording the same type of information as described above for day 0, including the IBFAT items, and provided lactation guidance. When onset of milk production was not apparent by the day 3 visit, a follow-up visit was arranged for the next day. All cases of infant weight loss ≥10% of birth weight were brought to the attention of the infant’s pediatrician. The lactation consultant worked in conjunction with the pediatrician in providing appropriate follow-up care for these infants.
Mothers were contacted by phone on day 5 to collect recall data for days 3 and 4 regarding the same questions described above for the day 3 visit. At the day 7 visit, this information was again collected by recall for days 5 and 6. The day 7 visit also included a breastfeeding observation and completion of a brief medical history for the mother. The day 14 visit included measurement of maternal height (to the nearest mm, using a portable stadiometer) and weight (to the nearest 100 g, using an electronic scale).
Data were analyzed using the SAS System for Windows, Release 8.1 (SAS Institute Inc, Cary, NC). The main outcome variables were:
Suboptimal infant breastfeeding behavior (SIBB)—day 0: IBFAT score ≤10 at the breastfeeding observed during the first 24 hours’ postpartum. This cutoff was chosen because we were interested in identifying any behavior that was suboptimal.
SIBB—day 3: IBFAT score ≤10 at the day 3 visit.
SIBB—day 7: IBFAT score ≤10 at the day 7 visit.
Delayed onset of milk production: Breasts not noticeably fuller (3 on the scale of 1–5) by 72 hours’ postpartum. We previously demonstrated that the time postpartum when breasts were first noticeably fuller was strongly correlated with milk volume on day 5 (r = −0.66; P < .0001) and the first appearance of milk casein (r = 0.49; P = .0001), a biochemical marker of lactogenesis.10 In a separate preliminary study of several markers of the time postpartum when milk transfer to the infant first exceeded 15 g per feed, breast fullness level 3 was closest in average timing (53 hours for breast fullness; 58 hours for intake >15 g) and showed the highest correlation with this indicator (r = 0.60; n = 23, unpublished data). Based on these two studies, we concluded that the breast fullness marker was the best choice for studies in which test-weighing is not feasible. Others have since validated using maternal recall to assess onset of milk production.17,18
Excess infant weight loss—day 3: Weight loss of 10% or more since birth. Infants who consumed >2 oz of nonbreast milk (NBM) fluids in the first 72 hours were excluded (n = 27), as this could influence weight loss patterns.
Bivariate associations between the 5 outcome variables described above and the 27 independent variables shown in Table 1 were examined using χ2 tests. Multiple logistic regression analysis was used to control for associations among the independent variables. Stepwise selection was used to construct the preliminary models (P level for inclusion or removal = 0.10). As described below, the independent variables for each preliminary model were chosen based on whether: a) there was a theoretical basis for an association between the independent variable and the outcome; b) the independent variable preceded the outcome variable in occurrence; and c) there was little possibility of reverse causation. In addition to the main effects of the 27 independent variables, potential interaction effects with parity were examined if: 1) there was a theoretical basis to suspect an interaction, and 2) the bivariate analyses suggested that the association of an outcome with a given independent variable differed in primiparas versus multiparas. In preliminary analyses, potential interactions with delivery mode were also examined, but these were not included in the regression models because there was little evidence that the association of the outcomes with any of the independent variables differed in mothers with vaginal versus cesarean section delivery. Associations with site of delivery were examined, but they were not significant for any of the outcomes. The final models included main effects significant at P < .05 and interactions significant at P < .10. Estimated relative risks (RRs) and 95% confidence intervals were calculated from the adjusted odds ratios of these final models, following the procedures described by Zhang and Yu.19 Post hoc analysis of significant interaction effects was done using the contrast option in SAS Proc Logistic.
For the logistic regressions examining SIBB on day 0, the age of the infant at the time of the IBFAT assessment (in hours) was forced into the model. Of the 27 independent variables, 19 variables were judged eligible for inclusion. The variables excluded (and the reasons) were a) pregnancy weight gain and prenatal breast enlargement (no theoretical basis); b) use of NBM fluids in first 24 or 48 hours, pacifier use on days 0 to 2 and nipple type on day 7 (did not precede the outcome); and c) nursing frequency on day 0 and breastfeeding confidence (possible reverse causation). The interactions meeting the criteria for initial inclusion in the stepwise logistic regressions were parity with delivery mode and parity with nipple type.
For the logistic regressions examining SIBB on day 3, 23 independent variables were eligible for inclusion (pregnancy weight gain, prenatal breast enlargement, breastfeeding confidence, and nipple type on day 7 were excluded). The models were run both with and without SIBB on day 0, using a path analysis approach. When SIBB on day 0 was included in the model, the interaction of parity with SIBB on day 0 was also evaluated. No other interactions met the criteria for initial inclusion. For SIBB on day 7, the same variables were included except that nipple type on day 7 rather than on day 0 was used (nipple type was not assessed on day 3). The models were run both with and without breastfeeding confidence on day 0. No interactions met the criteria for inclusion.
For the logistic regressions examining delayed onset, 26 independent variables were included (nipple type on day 7 was excluded). The models were run both with and without SIBB on days 0 and 3. In addition, the interactions of parity with the following variables met the criteria for initial inclusion in the models: a) maternal BMI, b) duration of stage II labor, c) birth weight category, and d) SIBB on day 3.
For the logistic regressions examining excess infant weight loss, the same 26 variables were included. The regressions were run both with and without delayed onset and SIBB on day 0 or day 3. In addition, the interactions of parity with the following variables met the criteria for initial inclusion in the models: a) use of labor pain medications, b) total duration of labor, and c) maternal rating of exhaustion during labor.
In total, 386 mother-infant pairs were screened during the recruitment period. Of these, 58 were ineligible for the study and 48 refused to participate. The reasons for ineligibility were a) did not plan to breastfeed exclusively for at least 1 month (n = 17), b) preterm delivery (n = 13), c) did not speak English (n = 12), d) perinatal morbidity (n = 9), e) multiple birth (n = 3), f) moving within 1 month (n = 3), and g) no phone (n = 1). The reasons for refusal were a) not interested (n = 26), b) time constraints (n = 16), c) refused anthropometric measurements (n = 4), d) mother ill (n = 1), and e) religious restrictions (n = 1). The analysis is based on the 280 subjects who were enrolled in the study, of whom 271 remained in the study on day 3 postpartum. Nine subjects were lost to follow-up on day 3 because they could not be contacted (n = 3) or they dropped out because of time constraints/inconvenience (n = 3), infant illness (n = 2), or lack of interest (n = 1).
The characteristics of the 280 mother-infant pairs are shown in the frequency column of Table 1. Mean maternal age was 30.6 ± 4.7 years, and 56% of the mothers were primiparous. The ethnic/racial breakdown was 78% non-Hispanic white, 11% Hispanic, 8% Asian, and 3% black. Average educational level was high (16.6 ± 2.1 years), reflecting the fact that Davis is a university town. Attitudes toward breastfeeding were strongly positive: average values were 11.0 ± 5.4 months for intended duration of breastfeeding, 4.6 ± 0.7 (out of 5) for confidence in ability to breastfeed exclusively, and 2.9 ± 0.3 (out of 3) for support for breastfeeding from family and friends.
Labor and delivery outcomes are shown by parity in Table 2. The cesarean section rate was 14%, and 44% of mothers had an unmedicated labor and delivery. Primiparas had a longer duration of labor, were more likely to have a cesarean section and to receive regional anesthesia, narcotic analgesia and pitocin, and were less likely to breastfeed the infant within the first hour than multiparas.
Average breastfeeding frequency was 8.3 ± 4.0 feeds on day 0 (adjusted for the time of the interview) and 10.5 ± 3.2 feeds on day 2. Water or glucose water supplementation was infrequent (2% on day 0 and 3% on day 2). Supplements of infant formula were given to 9% of infants on day 0 and 10% of infants on day 2. Pacifiers were used by 6% of subjects on day 0 and 19% of subjects on day 2.
Onset of milk production, as judged by breast fullness, is shown in Fig 1. For the majority of women (59%), milk production began at 49 to 72 hours’ postpartum. Delayed onset (>72 hours) was experienced by 22% of the women (33% of primiparas vs 8% of multiparas, P = .0001).
Figure 2 shows the distribution of weight change between birth and day 3 (excluding those who consumed >2 oz of NBM fluids during the first 72 hours). On average, infants lost 5.5 ± 3.8% of birth weight. Only 5% of infants gained weight. The majority (82%) lost <10% of birth weight, but 12% lost ≥10% of birth weight.
We were able to assess infant breastfeeding behavior in 79% of infants on day 0, in 89% of infants on day 3, and in 81% of infants on day 7. Nearly half (49%) of the infants assessed had SIBB on day 0, but this percentage decreased to 22% on day 3 and 14% on day 7.
Table 3 shows the associations among the 5 dichotomous outcomes. Compared with infants with optimal breastfeeding behavior, infants with SIBB on day 0 were 1.8 times more likely to have SIBB on day 3, and 2.6 times more likely to have excess weight loss. The association of SIBB on day 0 with delayed onset of milk production was not significant, but mothers of infants with SIBB on day 3 were 2.6 times more likely to have delayed onset than mothers of infants with optimal breastfeeding behavior on day 3. SIBB on day 0 was not associated with SIBB on day 7, but infants with SIBB on day 3 were 2.9 times more likely to have SIBB on day 7. The risk of excess infant weight loss was 7.1 times greater if the mother had delayed onset of milk production. Among the 30 infants with excess weight loss, 63% of their mothers had delayed onset of milk production, compared with 13% of the 210 infants who did not have excess weight loss.
Table 1 shows the bivariate associations of each of the independent variables with the 5 dichotomous outcomes. It is important to note that there were significant associations among many of the independent variables (not just the obvious associations, such as maternal age with parity, or cesarean section with use of labor pain medications). For example, women with a BMI >27 kg/m2 (n = 76) were significantly more likely to have had a cesarean delivery (22 vs 10%; P = .007), use NBM fluids on day 0 (17 vs 7%; P = .02) and in the first 48 hours of life (29 vs 13%; P = .002), be <30 years of age (55 vs 33%; P = .001), and not be a college graduate (28 vs 14%; P = .01), as compared with women with a BMI ≤27 kg/m2 (n = 164). Thus, for each of the outcome variables the description of the bivariate results will be followed immediately by the multivariate results. Only variables meeting the criteria for inclusion in the multivariate models will be mentioned (though all independent variables are shown in Table 1).
In the bivariate analyses, SIBB on day 0 was significantly associated (P ≤ .05) with primiparity, cesarean section, use of regional anesthesia during labor, and flat or inverted nipples. In the multiple logistic regression (Table 4), the association with flat or inverted nipples remained significant (RR: 1.56) and there was a significant interaction between parity and delivery mode: those at lowest risk were multiparous mothers with vaginal deliveries (RR: 1.0), with all other subgroups (multiparous mothers with cesarean sections [RR: 2.46]; primiparous mothers with either vaginal [RR: 1.72] or cesarean section [RR: 1.68] delivery) being at significantly higher risk for SIBB. In addition, the risk of SIBB was higher among infants whose amniotic fluid at delivery was clear rather than stained with meconium (RR: 1.55).
SIBB on day 3 was significantly associated with the following variables in the bivariate analyses: use of labor pain medications (particularly intravenous [IV]/intramuscular [IM] analgesia), flat or inverted nipples, lower maternal breastfeeding confidence, use of NBM fluids in the first 48 hours and pacifier use. In the multiple logistic regression (Table 5, Model 1), the significant variables were flat or inverted nipples (RR: 3.02), NBM fluids in the first 48 hours (RR: 2.26), pacifier use (RR: 1.95) and an Apgar score >7 at 1 minute of age (RR: 2.10). When SIBB on day 0 was included (Model 2), the RRs for NBM fluids and pacifier use were not reduced, but the RR for flat or inverted nipples was somewhat lower (2.57) and the RR for Apgar score was no longer significant. In addition, there was a significant interaction between parity and SIBB on day 0: the latter was a significant predictor of SIBB on day 3 among primiparas (P = .02) but not among multiparas (P = .82).
In bivariate analyses, SIBB on day 7 was significantly associated with maternal age <30 years, maternal BMI >27 kg/m2, flat or inverted nipples on day 7, lower maternal breastfeeding confidence, and use of NBM fluids in the first 48 hours. In the multiple logistic regression excluding breastfeeding confidence (Table 6, Model 1), the associations with maternal BMI (RR: 2.58), flat or inverted nipples (RR: 6.57) and use of NBM fluids (RR: 2.55) remained significant, and in addition there were significant associations with stage II labor >1 hour (RR: 3.11) and birth weight <3600 g (RR: 2.69). Neither SIBB day 0 nor SIBB day 3 was significantly associated with SIBB day 7 in the multiple logistic regression analyses. When breastfeeding confidence was included in the model (RR: 4.10), the results for the other variables were generally similar (Model 2) except that birth weight became marginally significant and use of NBM fluids was no longer significant. The association with breastfeeding confidence (assessed on day 0) could be attributable to reverse causation, but this is less likely for SIBB on day 7 than for SIBB on days 0 and 3 because there was no significant association between SIBB on day 0 and SIBB on day 7 (Table 3).
Delayed onset of milk production was significantly associated with the following variables in the bivariate analyses: primiparity, greater pregnancy weight gain, maternal BMI >27 kg/m2, cesarean delivery (particularly an urgent cesarean section), use of labor pain medications, augmentation of labor with pitocin, longer duration of labor (both total and stage II), a longer interval without sleep before delivery, flat or inverted nipples, and use of NBM fluids in the first 48 hours. In the multiple logistic regression (Table 7, Model 1), the significant variables were long duration of stage II labor (RR: 2.26), cesarean delivery (RR: 2.01), high maternal BMI (RR: 2.46), flat or inverted nipples (RR: 2.26) and the interaction of parity with birth weight. The latter interaction is shown in Fig 3, which illustrates that a primiparous mother is at greatest risk for delayed onset if she has a large infant, whereas giving birth to a larger infant is not a significant risk factor for delayed onset among multiparas. In models using a path analysis approach, SIBB on day 0 did not enter the model, but SIBB on day 3 was a significant predictor of delayed onset in the logistic regression (RR: 2.66). With SIBB on day 3 included (Model 2), the RR for flat or inverted nipples became nonsignificant and the RR for cesarean delivery became marginally significant (the latter was because of the smaller sample size for Model 2, not to a decrease in the magnitude of the RR for cesarean section).
Excess infant weight loss was significantly associated with the following variables in bivariate analyses: maternal BMI >27 kg/m2, cesarean delivery, use of labor pain medications (particularly regional anesthesia), longer duration of labor (≥14 hours), infant not given oxygen postdelivery, and flat or inverted nipples on day 7. In the multiple logistic regression (Table 8, Model 1), the significant variables were longer duration of labor (RR: 2.41), infant not given oxygen (RR: 8.33), and the interaction of parity with use of labor pain medications. The latter interaction revealed that labor pain medications were a risk factor for excess infant weight loss among multiparas (RR: 4.06) but not primiparas; among women who did not receive labor pain medications, primiparity was a risk factor (RR: 4.51) for excess infant weight loss. In models using a path analysis approach, SIBB on day 3 did not enter the model, but both SIBB on day 0 and delayed onset were significant predictors of excess infant weight loss in logistic regressions. With delayed onset in the model (but not SIBB on day 0, see Model 2), the RR for duration of labor became nonsignificant, but those for infant oxygen and the interaction of parity with labor pain medications remained significant. With both delayed onset (RR: 6.13) and SIBB on day 0 (RR: 2.43) included (Model 3), none of the other variables was a significant predictor of excess weight loss. However, the nonsignificant P values for infant oxygen and the above interaction were primarily attributable to the smaller sample size for Model 3, as the RRs for these variables did not decrease when SIBB on day 0 was added to the model if confined to the same group of subjects (N = 191).
These results indicate that lactation difficulties during the first week postpartum are not uncommon, even among women who are highly motivated to breastfeed exclusively and receive good lactation guidance. In this sample of 280 mother-infant pairs, SIBB was observed in 49% on day 0, in 22% on day 3, and in 14% on day 7. Delayed onset of lactation occurred in 22% of women, and 12% of the infants lost ≥10% of birth weight. These outcomes were strongly linked. For example, infants of mothers with delayed onset of lactation were 7.1 times more likely to have excess weight loss, and those with SIBB on day 0 were 2.6 times more likely to have excess weight loss than infants without these risk factors. In fact, of the 26 infants with excess weight loss who had a breastfeeding assessment done on day 0, all but 2 had either SIBB or a mother with delayed onset (or both). Thus, 92% of such cases could be predicted on the basis of these 2 evaluations combined.
The incidence of delayed onset in our population was somewhat lower than observed in a sample of breastfeeding women in Connecticut (31%),2 which could be explained by the more comprehensive lactation guidance provided in our study. On the other hand, the incidence of excess infant weight loss in our study was greater than that observed in a sample of 686 exclusively breastfed infants born in a breastfeeding-friendly hospital in Italy (8%).9 In that hospital, no study infant received glucose or formula during the first 3 days of life (compared with 21% of infants in our study), and newborns were normally discharged on day 3 or 4 if delivered vaginally (compared with within the first 2 days in our study). These practices may prevent some cases of excess weight loss by promoting frequent breastfeeding and ensuring that mothers receive lactation guidance beyond the first 24 hours’ postpartum.
The multivariate analyses illustrate that parity, events during labor and delivery, and characteristics of the mother and infant have strong effects on SIBB, delayed onset of lactogenesis, and excess infant weight loss. Table 9 summarizes the risk factors identified. Primiparas were at greater risk than multiparas for all outcomes except SIBB on day 7. The greater risk of SIBB on day 3 among primiparas was evident only among those whose infants already had SIBB on day 0, whereas SIBB on day 0 was not a risk factor for SIBB on day 3 among multiparas. This implies that multiparas whose infants have SIBB on day 0 are able to improve the situation, presumably because of their past breastfeeding experience, whereas first-time mothers need special assistance to do so. The relationship between primiparity and excess infant weight loss was mediated by the greater risk of delayed onset and SIBB on day 0 among primiparas. The group at lowest risk for excess infant weight loss was multiparas with an unmedicated labor and delivery. The strong associations we observed between parity and early lactation success are consistent with other studies showing that primiparity or lack of previous breastfeeding experience are risk factors for delayed onset of lactation2,10,11 or excess infant weight loss.9
Cesarean section delivery was a risk factor for SIBB on day 0 and delayed onset of lactation. Other investigators have also reported that cesarean section is linked with delayed onset2,11,12 and excess infant weight loss,9 although not all studies have shown an association with difficulties initiating lactation.20 With adequate guidance, mothers who have a cesarean section can generally overcome early breastfeeding problems and successfully establish breastfeeding.21
A long duration of stage II labor predisposed mothers to delayed onset of lactation and infants to SIBB on day 7 (though not on day 0 or 3). The association with delayed onset is consistent with findings of previous studies.2,10 We also found a 2.4-fold greater risk of excess infant weight loss among mothers with a total duration of labor >14 hours, which was explained by the link to delayed onset. We have previously shown that a long duration of labor is stressful to both the mother and the infant, resulting in higher cortisol levels in both.10 Higher cortisol levels have been linked to delayed onset of lactation.10,22 The highest incidence of delayed onset in our study (56%) was observed among mothers who had an urgent cesarean section (as compared with 27% among those with a scheduled cesarean section). This relationship with urgent cesarean section delivery, which is undoubtedly the most stressful delivery experience, has also been reported by Chapman and Pérez-Escamilla.2
The influence of labor pain medications on breastfeeding has been controversial. In the bivariate analyses, we found that SIBB on day 0 was more common among those whose mothers received regional anesthesia during labor, SIBB on day 3 was more common among those whose mothers received IV or IM analgesia during labor, and delayed onset of lactation was more common in mothers who received both regional and IV/IM labor medications. None of these associations remained significant in the multivariate analyses when controlling for mode of delivery and other variables. However, within the subgroup of infants delivered vaginally, IV/IM analgesia was a significant predictor of SIBB on day 3 in multivariate analyses (P = .03; data not shown). Moreover, there was a significant association between labor medications and excess infant weight loss among multiparas, even after controlling for mode of delivery, duration of labor, and other potential confounders. Because of variability in the type and timing of dosage of medications used, it is difficult to determine which medications are responsible for this association, but the data suggest that regional anesthesia is more likely than IV/IM analgesia to affect infant weight loss. It is possible that administration of IV fluids during labor, which is more common in women given labor pain medications, increases the hydration status of the newborn and leads to greater weight loss subsequently. However, this would not explain why we saw the association between labor pain medications and excess weight loss only in multiparas. Previous studies have focused on the relationship between labor medications and infant suckling behavior or breastfeeding duration, not infant weight loss. Some have shown an association,3,23,24 whereas others have not.25,26 Our data suggest that the association may not be evident among primiparas (who are at higher risk for early lactation difficulties regardless of labor medications), but may be significant among multiparas. Conflicting results may also be attributable to the potential indirect effects of labor medications on breastfeeding via their influence on duration of labor and the mode of delivery, which would not be evident when controlling for the latter variables. In a recent systematic review of unintended effects of epidural medications during labor, Lieberman and O’Donoghue27 concluded that epidurals increase the duration of labor and reduce the likelihood of a spontaneous vaginal delivery, particularly in primiparas.
Maternal obesity has been linked to impaired lactogenesis in both animal and human studies,15 but the mechanisms for this relationship are unclear. We found that women with a BMI >27 kg/m2 were 2.5 times more likely to have delayed onset of lactation than women with a lower BMI, and their infants were 3 times more likely to have SIBB on day 7 (though not on day 0 or 3). Because women in our study with a higher BMI differed in many ways from those with lower BMI (eg, age, education, use of NBM fluids), we cannot completely rule out behavioral factors, but when these variables were included in the analyses the RRs associated with high BMI did not decrease. Endocrinologic aberrations linked with overweight are a possible cause of delayed onset of lactation. In rats, impaired lactogenesis in obese dams has been linked with altered regulation of glucose, the primary substrate for milk synthesis.28 Nonetheless, with lactation guidance the vast majority of overweight women are able to successfully establish exclusive breastfeeding.
Flat or inverted nipples were associated with SIBB on days 0, 3, and 7, and with delayed onset of lactation. This implies that a woman with flat or inverted nipples should receive special assistance until the infant is able to latch on effectively. The effect of early use of NBM fluids and/or pacifiers on breastfeeding success has been controversial. Because of the possibility of reverse causation (ie, infants doing poorly at the breast are more likely to be offered supplements) and potentially confounding variables (such as lower motivation to breastfeed exclusively among women who use supplements or pacifiers), it has been difficult to determine the causal pathway underlying the inverse associations found in observational studies.29 Our results showed that infants who were given NBM fluids in the first 48 hours or offered pacifiers were 2 to 3 times more likely to have SIBB on days 3 and 7, even after controlling for their breastfeeding score on day 0. Because there is no medical reason to give supplemental fluids to normal, healthy breastfed newborns during the first 48 hours, and we controlled for suckling difficulties on day 0, these findings support the hypothesis that supplemental fluids and pacifiers can interfere with the establishment of effective breastfeeding. On the other hand, we did not find a relationship between excess infant weight loss and supplemental fluids or pacifier use. Thus, if there is an effect of such practices on breastfeeding behavior, with appropriate lactation guidance it may be short-lived or too subtle to affect other outcomes, except perhaps in high-risk mother-infant pairs. Kramer et al29 concluded that pacifier use is a marker of breastfeeding difficulties, not a true cause of cessation of breastfeeding by 3 months. However, they did not specifically examine the effect of pacifier use during the first week of life, when the infant is learning how to suckle effectively.
A lower infant birth weight was associated with SIBB on day 7, but high infant birth weight (>3600 g) was associated with delayed onset of lactation among primiparas (though not among multiparas), even when controlling for duration of labor. One possible explanation for the latter finding is that delivering a large infant is more difficult (particularly for primiparas) regardless of the duration of labor, and leads to greater maternal and/or infant stress. Our findings regarding birth weight and delayed onset conflict with those of Chapman and Pérez-Escamilla,2 who found a higher risk of delayed onset in mothers of infants with birth weight <8 lb. However, they did not examine this relationship only within the primiparas.
Curiously, several indicators reflective of better infant status at birth were associated with a greater risk of poor outcomes in our study. Clear amniotic fluid (vs meconium staining) was associated with SIBB on day 0, 1-minute Apgar score >7 was associated with SIBB on day 3, and no use of oxygen (ie, resuscitation by mask or blow-by) was associated with excess infant weight loss. All 3 of these indicators were strongly interrelated, which is to be expected given that infants with low Apgar scores or evidence of stress (eg, meconium-stained amniotic fluid) were much more likely to be given oxygen. One possible explanation is that, when controlling for other risk factors linked with infant status (such as cesarean section, birth weight, and duration of labor), there is a benefit to administering oxygen that has a positive influence on recovery from the birth experience and thus breastfeeding. This is not to imply that all infants should receive oxygen at birth, but simply that breastfeeding outcomes may be enhanced by administration of oxygen to high-risk infants. With regard to meconium staining, another possible explanation is that infants who pass meconium sooner are less likely to have high bilirubin levels30 and thus are more interested in breastfeeding.31 It is also possible that these associations are simply attributable to chance.
The results of this study reinforce the need for special attention and follow-up of mother-infant pairs at risk for breastfeeding difficulties during the first week postpartum. High-risk groups include primiparas (particularly those delivering large infants), mothers who have a long labor or deliver by cesarean section (particularly an urgent cesarean section), multiparas who receive labor medications, overweight mothers, women with flat or inverted nipples, and infants with SIBB during the first 24 hours. These findings imply that modifications of the labor and delivery experience that reduce the duration of labor (both total and stage II) and increase the likelihood of a spontaneous, unmedicated vaginal delivery should improve early lactation success. The results also provide support for the recommendation to avoid supplemental fluids unless medically indicated32 and to discourage pacifier use during the first week of life. Given that early lactation difficulties were common even in women who were highly motivated to breastfeed and received lactation guidance during the hospital stay, it is essential that all breastfeeding mother-infant pairs be followed up at 72 to 96 hours’ postpartum. At the follow-up evaluation, mothers should be asked if their milk has come in, and if the answer is no, be provided with immediate assistance by a health care provider trained in lactation management. With appropriate assistance, nearly all women with delayed onset of lactation can eventually establish an adequate milk supply, but the phenomenon should be taken seriously not only because it can lead to excess infant weight loss in the short-term, but because it has been linked with shorter breastfeeding duration.33
The study was funded by National Institutes of Health grant RO1 HD35962 and the World Health Organization (Department of Nutrition for Health and Development).
We thank all of the mothers and infants who participated in the study, as well as our dedicated team of International Board Certified Lactation Consultants (Karen Farley, Ann Gorrell, Laura Ortiz, Jeanette Panchula, and Carla Turoff) and research assistants (Swati Deshpande, Cindy Duke, Kathy Harris, Beth Tohill, Nanise Tomlinson, and Tracey Wang). We also thank the staff at the participating hospitals (Sutter Davis Hospital, Woodland Memorial Hospital, Kaiser Permamente Hospitals of Sacramento, and the University of California at Davis Medical Center) for their cooperation with recruitment procedures. We thank Janet Peerson for her statistical guidance.
- Received September 9, 2002.
- Accepted May 8, 2003.
- Address correspondence to Kathryn G. Dewey, PhD, Department of Nutrition, University of California, One Shields Ave, Davis, CA 95616-8669. E-mail:
- ↵Neville MC, Morton J. Physiology and endocrine changes underlying human lactogenesis II. J Nutr.2001;131 :3005S– 3008S
- ↵Neifert MR. Prevention of breastfeeding tragedies. In: Schanler RJ, ed. The Pediatric Clinics of North America, Breastfeeding 2001, Part II: The Management of Breastfeeding, Volume 48. Philadelphia, PA: WB Saunders; 2001:273–298
- ↵Cooper WO, Atherton HD, Kahana M, Kotagal UR. Increased incidence of severe breastfeeding malnutrition and hypernatremia in a metropolitan area. Pediatrics.1995;96 :957– 960
- ↵Maisels MJ, Gifford K, Antle CE, Leib GR. Jaundice in the healthy newborn infant: a new approach to an old problem. Pediatrics.1988;81 :505– 511
- ↵Chen DC, Nommsen-Rivers L, Dewey KG, Lonnerdal B. Stress during labor and delivery and early lactation performance. Am J Clin Nutr.1998;68 :335– 344
- ↵Dewey K. Maternal and fetal stress are associated with impaired lactogenesis in humans. J Nutr.2001;131 :3012S– 3015S
- ↵Hartmann P, Cregan M. Lactogenesis and the effects of insulin-dependent diabetes mellitus and prematurity. J Nutr.2001;131 :3016S– 3020S
- ↵Rasmussen KM, Hilson JA, Kjolhede. Obesity may impair lactogenesis II. J Nutr.2001;131 :3009S– 3011S
- ↵Chapman DJ, Pérez-Escamilla R. Maternal perception of the onset of lactation is a valid, public health indicator of lactogenesis stage II. J Nutr.2000;130 :2972– 2980
- ↵Pérez-Escamilla R, Chapman DJ. Validity and public health implications of maternal perception of the onset of lactation: an international analytical overview. J Nutr.2001;131 :3021S– 3024S
- ↵Grajeda R, Pérez-Escamilla R. Stress during labor and delivery is associated with onset of lactation among urban Guatemalan women. J Nutr.2002;132 :3055– 3060
- ↵Lieberman E, O’Donoghue C. Unintended effects of epidural analgesia during labor: a systematic review. Am J Obstet Gynecol.2002;186(5 suppl Nature) :S31– S68
- ↵Shaw MA, Rasmussen KM, Myers TR. Consumption of a high-fat diet impairs reproductive performance in Sprague-Dawley rats. J Nutr.1997;127 :64– 69
- ↵Academy of Breastfeeding Medicine. Clinical protocol number 3—hospital guidelines for the use of supplementary feedings in the healthy term breastfed neonate. Academy of Breastfeeding Medicine [web site]. Available at: http://www.bfmed.org/proto3.html. Accessed May 17th, 2002
- Copyright © 2003 by the American Academy of Pediatrics