 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
PEDIATRICS Vol. 104 No. 5 November 1999, pp. 1070-1076
, and
From the * Division of Neonatology, University of Utah Medical
Center, Salt Lake City, Utah; the Primary Children's Medical
Center, Salt Lake City, Utah; and the § Department of Psychology,
University of Nebraska, Lincoln, Nebraska.
 |
ABSTRACT |
|---|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
Objective. To evaluate the relationship between early perinatal hospital discharge and several parenting outcomes during infancy, including breastfeeding, mother-infant interaction, and mother-infant attachment.
Study Design. A prospective, longitudinal, nonrandomized
study of mother-infant dyads discharged 
36 hours after birth (early
discharge), compared with those discharged >36 hours after birth (late
discharge).
Methods. Demographic, perinatal, and psychosocial factors were determined from medical record review and maternal questionnaires. Questionnaires also assessed maternal perceptions of the hospital stay and breastfeeding rates. Mother-infant interaction was assessed at 3 months after birth using the NCAST Feeding Scale and at 9 months after birth using the NCAST Teaching Scale. Security of attachment was measured in the Ainsworth Strange Situation at 12 months after birth.
Results. Early and late discharge groups were similar with respect to major demographic, perinatal, and psychosocial characteristics and perceptions of the hospital stay. Even after adjusting for these factors in regression analyses, no significant association was found between early discharge and breastfeeding at 3 months, NCAST scores at 3 and 9 months, and security of attachment at 12 months.
Conclusion. Parenting outcomes, such as breastfeeding, mother-infant interaction, and attachment, are not influenced by early perinatal hospital discharge. Key words: hospital discharge, newborn, parenting, mother-infant interaction, attachment, breastfeeding.
The recent controversy in the United States over early
hospital discharge of healthy term newborns and their mothers, which culminated in the Newborns' and Mothers' Health Protection Act of
1996,1 followed a trend toward shorter perinatal hospital stays that spanned several decades. Although most recently prompted by
financial limitations imposed by third party payers, early discharge
was initiated as early as the 1950s because of bed space limitations at
busy hospitals in several Western countries.2 During the
1970s and 1980s, the short perinatal stay became more popular for
social rather than for financial reasons, in part, because of the
growing women's movement and the desire of women to regain some
control over the childbirth process.3,4 Among the
advantages of early discharge touted in the medical literature during
this period were improved parent-infant bonding,5,6 improved family integration, more opportunity for parent-infant interaction,6-10 and improved
breastfeeding.11,12 Elective early discharge programs
became more popular because many families preferred to spend more time
with the newborn at home.13
During the 1990s, however, the more universal abbreviation of the
perinatal stay led to concerns about the possible detrimental effects
of this practice on the family, especially among those at high social
risk.14,15 Conceivably, inadequate time in the hospital to
prepare for breastfeeding, parenting, and infant care could, in some
cases, predispose the infant to more overt medical problems, such as
dehydration or jaundice, morbidities that have increased with shorter
hospital stays.16,17 The most recent discharge statement
of the Committee on the Fetus and Newborn of the American Academy of
Pediatrics has emphasized the importance of evaluating for social risk
factors and parenting capabilities before discharge
planning.18 Yet recognizing the paucity of existing data
relating such factors to the length of stay, the Newborns' and
Mothers' Health Protection Act of 19961 emphasized the
need to study the "diversity of negative or positive outcomes
affecting mothers, infants, and families." Unfortunately, most
studies on early discharge have failed to control for psychosocial confounding factors and few have addressed parenting and family outcomes.
In light of these considerations, we sought to evaluate the effect of
early perinatal discharge on several parenting outcomes in a cohort of
mother-infant dyads that were characterized extensively with respect
to psychosocial factors. As outcomes, we evaluated breastfeeding,
mother-infant interaction, and attachment during infancy.
Characterization of the Study Cohort
The data presented are from a prospective, nonrandomized
longitudinal study of mothers and infants during the first postnatal year, performed between 1986 and 1988. Appropriate approvals were obtained from institutional review boards, and informed consent was
obtained in all cases. Mothers were recruited over an 18-month period
during prenatal office visits at 28 to 32 weeks' gestation from a
private obstetric clinic, a university hospital obstetric clinic, and a
community health clinic. Together, providers from these clinics
performed ~1600 deliveries annually. The clinics served a population
that was 3% black, 72% Caucasian, non-Hispanic, 21% Hispanic, 2%
Native American, and 2% other ethnicity; 14% of the mothers giving
birth in the community served were Signs advertising the study were posted in all clinics, together with
introductory letters with attached interest slips. Mothers completing
an interest slip were contacted by phone and provided more detailed
information. Only mothers expecting uncomplicated singleton deliveries
were offered the opportunity to enroll. Cases were excluded if
complications at the time of birth required cesarean section, preterm
delivery, or transfer of the infant to a newborn intensive care unit.
All deliveries occurred at either a university hospital or a nearby
affiliated large private community hospital. Policies and procedures
for newborn and maternal care and for maternal teaching were similar at
both hospitals. During the hospital stay, study mothers were allowed
unlimited access to their infants as desired. None of the mothers or
infants had medical problems that required an extended length of
hospital stay. A postnatal length of stay was considered short (early
discharge) when its duration was Information regarding demographic characteristics, socioeconomic
status, psychosocial risk factors, and maternal perceptions of the
length of the hospital stay was obtained from questionnaires administered prenatally, after birth, and at 3 months postpartum. Supplemental information, including medical data and length of stay,
was gathered by review of maternal and infant hospital records. Demographic and perinatal characteristics included infant gender, maternal age, ethnicity, educational status, insurance status, family
income, attendance at prenatal and parent effectiveness or parenting
classes, trimester of onset of prenatal care, birth setting, and infant
feeding type at discharge. The psychosocial characteristics included
parity, smoking, history of drug abuse, maternal physical or sexual
abuse as a child, history of recent domestic violence, whether the
mother was married and/or living with the father of the child, maternal
indication of serious problems with her mate, maternal indication of a
poor relationship with her mate, unplanned pregnancy, and lack of
choice of infant name before discharge.
Although all analyses compared early and late discharge groups, some
multivariate analyses also controlled for maternal perceptions of the
length or quality of the hospital stay. Mothers were asked to rate the
quality of their stay on a 10-point scale, yielding a quality of stay
score. To explore the perceived appropriateness of the length of stay,
they were asked to indicate whether the stay was too short, too long,
or just right. For analysis, this parameter was dichotomized to
indicate satisfaction with the length of stay (just right) or lack of
satisfaction (too short or too long).
Maternal questionnaires also included a number of formal instruments.
The Hollingshead Four Factor Index of Socioeconomic Status, which
combines information on gender, marital status, education, and
occupation, supplemented family income as an indicator of socioeconomic
status.20 The Family Apgar, an instrument designed to
measure the perception of family support systems,21 was
modified to assess maternal perceptions of her own parental support
when she was a child, as well as current support from the baby's
father, friends, and extended family. Social support, assessed by the
Family Apgar, is considered inadequate when the score on this
instrument is <7.21 For the subjects studied, the
internal consistency reliability of this instrument ranged between 0.87 and 0.92. Family life events were evaluated using a Schedule of Recent
Experiences score,22,23 a cumulative score which depicts
the numerical frequency of certain life events within the past 2 months. Maternal self-concept was assessed by the Tennessee
Self-Concept Scale,24 used by permission from Western
Psychological Associates, with the total positive score taken as the
best reflection of the overall level of self-esteem. For the subjects
studied, the internal consistency reliability was 0.93. All these
measures have been validated extensively and utilized during the
perinatal period.25
Measurement of Outcomes
Mothers were contacted initially by phone and then visited in
the home at 3, 9, and 12 months postpartum, at which times various outcomes were assessed.
Breastfeeding rates were assessed by medical record review and maternal
questionnaires administered at 3 months postpartum. Dichotomous
variables indicated whether the mother was breastfeeding (any
breastfeeding) and breastfeeding exclusive of supplementation (exclusive breastfeeding); failure to breastfeed at 3 months was taken
as an adverse outcome.
Mother-infant interaction was evaluated in the home at 3 months after
birth using the NCAST Feeding Scale and at 9 months after birth using
the NCAST Teaching Scale.26,27 These instruments have been
used extensively during infancy and evaluate maternal sensitivity to
cues, response to distress, social-emotional growth fostering, and
cognitive growth fostering, in addition to infant clarity of cues and
responsiveness to parent. Low NCAST scores have been associated with
adverse parenting outcomes, such as nonorganic failure to thrive and
child abuse.26,27 Sessions were videotaped and later
scored by at least 2 trained observers who were blind with respect to
perinatal information. In addition to using the total scores and the
parental and infant contributions to those scores in analyses, an
adverse outcome was defined as a total score that was less than the
published 10th percentile for the scales. For the subjects studied, the interrater reliability ranged from 0.89 to 0.92, and the internal consistency reliability ranged between 0.80 and 0.82.
Security of attachment was evaluated using the Ainsworth Strange
Situation,28 performed in the laboratory at 12 months
after birth and scored by trained observers from videotaped sessions.
The Ainsworth Strange Situation is a standard measure for assessment of
mother-infant attachment and has been validated highly at 1 year after
birth and beyond.29 Observer differences were reconciled
by discussion and consensus. In addition to standard classifications in
use at the time (secure, resistant, avoidant), dyads also were
categorized as either secure or insecure, with insecure attachment
taken as the adverse outcome.
Statistical analyses were performed using SPSS on the Power
Macintosh.30 Bivariate analyses included
Characteristics of Study Groups
Prenatally, 353 mothers expressed an interest in the study and 209 were enrolled. At the time of delivery, 146 subjects remained in the
study. With respect to length of stay, 65 dyads (44.5%) were in the
early discharge group and 81 dyads (55.5%) were in the late discharge
group at this time. There was some attrition during the study period,
such that 139 mother-infant dyads continued to participate at 3 months
(60 early, 79 late), 133 at 9 months (59 early, 74 late), and 128 at 12 months (58 early, 70 late). The postnatal discharge times (mean ± standard error) of the early and late groups were 23.9 ± 0.6 and
51.0 ± 0.7 hours, respectively.
Among study mothers, 4.6% were <18 years of age, 37.5% were between
18 and 25, 48.7% were between 26 and 35, and 9.2% were >35 years of
age. With respect to ethnicity, 72.8% were Caucasian, non-Hispanic,
23.8% Hispanic, and 3.4% other, proportions very similar to those of
the community from which the subjects were recruited. The mean annual
family income was $25 788 ± 1684, somewhat lower than the median
household income of the United States of $30 056 in
1989.32 From an educational standpoint, 17.1% of the
mothers had not completed high school, 14.5% had completed high school
only, 42.1% had attended some college, and 26.3% were college
graduates. The mean Hollingshead score was 36.33 ± 1.43. Of the
dyads, 64.3% had public or private insurance, a percentage lower than
the national value of 87.1% in 1987.33
Prenatal childbirth classes had been completed by 66.9% of the
mothers, and 27.9% had completed parent effectiveness classes. Prenatal care was initiated before 24 weeks' gestation by 93.8% of
the mothers; 24.7% were primiparous and 83.4% were breastfeeding at
the time of hospital discharge.
When evaluating their perinatal hospital stay at 3 months postnatally,
68.7% of the mothers indicated that the length was "just right",
21.8% that it was "too short", and 9.5% that it was "too
long". The mean value of the quality of stay score was 7.45 ± 0.16.
With respect to psychosocial characteristics, 17.5% of the mothers
were smokers, 5.9% gave a history of illicit drug use, and 16.4% were
abused physically or sexually as children. With respect to marital
relationships, a recent history of domestic violence was given by
11.5%, 20.3% of the mothers indicated that they had major problems
with their mate, and 13.7% indicated that their relationship with
their mate was poor. Of the mothers, 21% were not married to the
father of the infant and 13.1% were not living with the father of the
infant. The pregnancy was unplanned for 33.1% of the mothers, and
22.5% had not chosen a name for the infant at the time of discharge.
As defined by low scores on the Family Apgar, inadequate social support
was perceived by the following percentages of study mothers from the
indicated sources: parents during childhood, 48.7%; father of the
child, 23.1%; extended family, 28.2%; and friends, 26.5%. On the
other hand, only 3.6% of the mothers had all Apgar scores <7,
suggesting that most mothers had adequate support from at least one
source.
The mean number of total life events was 2.13 ± 0.23, with
1.16 ± 0.14 positive and 0.91 ± 0.16 negative events. The
mean self-concept score was 352.50 ± 3.70, slightly higher than
the published mean score for this subscale of 345.57.23
Demographic, perinatal, and psychosocial factors in early and late
discharge groups are compared in Tables 1
and 2. With respect to demographic
characteristics, the early discharge dyads were significantly more
likely to have no insurance, compared with the late discharge group.
With respect to psychosocial characteristics, mothers discharged early
were more likely than those with a late stay to perceive the support of
their parents as low during their childhood. There were no other
significant differences between the early and late discharge groups
with respect to these characteristics.
TABLE 1 TABLE 2
METHODS
Top
Abstract
Methods
Results
Discussion
References
19 years of age and 24% were
>30. Of the mothers, 7% had completed <10 years of education, and
27% were unmarried. Prenatal care was initiated before the third
trimester for 93%, and 39% were primiparous.19
36 hours; stays >36 hours were
considered routine (late discharge). However, the average postpartum
length of stay for the population from which the study subjects were
drawn was 43 hours, with 39% discharged within 36 hours. Mothers and
infants were not routinely seen before the first well baby visit at 2 weeks after birth.
2 tests and Fisher's exact test for
categorical variables and Student's t tests for continuous
variables. In multivariate analyses, logistic regression was used for
dichotomous outcomes and linear regression for continuous outcomes.
Power analyses were conducted using G-Power,31 with
calculations based on unadjusted outcomes. An 
significance level of
0.05 was chosen for all analyses.
RESULTS
Top
Abstract
Methods
Results
Discussion
References
Demographic and Perinatal Characteristics
Psychosocial Characteristics
Outcomes
At 3 months postpartum, 57.9% of the mothers were breastfeeding and 43.4% were breastfeeding exclusively without supplementation. The mean 3-month NCAST Feeding Scale score was 58.00 ± 0.64, and the mean 9-month NCAST Teaching Scale score was 54.61 ± 0.58. These were slightly lower than the published mean values of 64.29 ± 0.26 and 59.56 ± 0.22, respectively.26,27 The proportions of subject scores on both scales that fell below the published 10th percentiles for the NCAST were high,26,27 ranging from 11.9% to 25%. The proportions of dyads classified as secure, avoidant, and resistant in the Ainsworth Strange Situation were 54.8%, 25.2%, and 16.4%, respectively, similar to published values for populations from the United States.28,29
In bivariate analyses, there were no significant differences between early and late discharge groups with respect to breastfeeding, NCAST scores, or attachment categories (Table 3). In addition to comparisons with respect to actual length of stay, these outcomes were compared between mothers satisfied and unsatisfied with the length of stay. In this analysis, there were no significant differences in outcomes with respect to maternal satisfaction (data not shown).
|
In multivariate analyses, a variety of regression models were formulated for each outcome that included the length of stay variable and individual dyadic characteristics as covariates. Because maternal age, ethnicity, parity, and socioeconomic status have been implicated as potential predictors for each of the parenting outcomes studied,26,27,29,34,35 they were included in all models, with maternal education taken as a proxy for socioeconomic status. Using this approach, odds ratios were calculated for the major adverse dichotomous outcomes (not breastfeeding, NCAST Feeding and Teaching Scale scores below the 10th percentile, and insecure attachment) and are shown in Table 4. Similarly, regression coefficients were determined for the NCAST Feeding and Teaching Scale scores and are given in Table 5. None of these analyses indicated a significant effect of early discharge. Similarly, regression models in which NCAST parent and infant subscale scores were used as outcomes revealed no significant effects (data not shown). For all outcomes, models that controlled for maternal perceptions of the hospital stay (satisfaction with length or quality) also failed to show a significant effect of early discharge (Tables 4 and 5). Finally, analyses that controlled for a variety of interactions between individual dyadic characteristics and early discharge revealed no significant results (data not shown).
|
|
Because of these nonsignificant findings, a power analysis was
performed for the major outcome variables. With an of 0.05 and a
of 0.2, the sample size was adequate to detect the following percentage differences (observed-expected) in the outcomes measured: any breastfeeding, 23%; exclusive breastfeeding, 23%; NCAST Feeding Scale <10th percentile, 17%; NCAST Teaching Scale <10th percentile, 12%; and attachment not secure, 17%. The sample was also adequate to
detect a 10% change in total NCAST scores at both time periods with a
of 0.1. These findings indicate a moderate power to detect an
appreciable difference in each of the outcomes studied.
| |
DISCUSSION |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
This study of a cohort of mothers and term newborns suggests that early discharge, defined as occurring within 36 hours of delivery, is not associated with alterations in breastfeeding at 3 months, mother-infant interaction at 3 or 9 months, or attachment at 12 months after birth. A major strength of the study is the great detail with which the discharge groups were characterized with respect to demographic, perinatal, and psychosocial variables, confounding factors that were similar in both groups. No previous study of early discharge has assessed these factors in such detail. Although the results of the study were negative, sample size analysis revealed a moderate power to substantiate the lack of significant differences in the selected outcomes between early discharge and late discharge groups.
Although some studies have suggested that early discharge may be associated with decreased breastfeeding, others have found either no association or an increase in breastfeeding rates among mothers and infants discharged early.11,12,36,37 There was considerable variation in both discharge times and methodology among these studies, rendering comparison of results difficult. Conceivably, breastfeeding could be decreased among women discharged early, if there is inadequate time for lactation education during the short hospital stay. In contrast, it is possible that many mothers who elect early discharge may have either more lactation knowledge or experience, compared with those with longer stays. In the largest study to address this question, Mandl et al37 evaluated discharge times similar to ours and found no effect of discharge time on breastfeeding at 3 weeks postpartum. These findings are consistent with our failure to demonstrate an association between early discharge and breastfeeding at 3 months postpartum, and taken together the two studies suggest that the impact of hospital stay on breastfeeding may be minimal. It is likely that other factors, such as postdischarge lactation support, may be more important in influencing this outcome.
The NCAST Feeding and Teaching Scales are reliable and valid measures of mother-infant interaction during infancy.26,27 In this regard, they may be taken as indices of parenting in the context of the dyadic interaction. If early discharge promotes mother-infant interaction as proposed during the 1980s, then higher NCAST scores might have been expected. On the other hand, lower scores could have resulted if early discharge leads to interruption of in-hospital experiences that could foster improved mother-infant interaction. Clearly, no impact of early discharge on interaction as assessed by these instruments was demonstrable.
Only one previous study has evaluated the relationship between early perinatal discharge and mother-infant attachment.38 Compared with mothers discharged at 24 to 47 hours without their infants, mothers and infants discharged together at 24 to 47 hours with home follow-up visits had significantly lower mean scores on the Avant maternal attachment scale during the second postnatal week. Neither of these groups differed significantly from a conventional discharge group that left the hospital together at 48 to 72 hours after birth. These results may suggest a potential effect of early mother-infant separation, but they do not support an effect of discharge time on attachment. Although arguments similar to those given above for mother-infant interaction might be constructed for or against early discharge, our results suggest that discharge time does not influence attachment as assessed by the Ainsworth Strange Situation.
Liu et al39 found that newborns discharged <30 hours after birth were more likely to be rehospitalized during the first month after birth. Subgroups at increased risk for rehospitalization after early discharge included newborns born to primigravidas and to mothers <18 years of age. The reasons for the increased risk among these groups are unknown, but inadequate parenting may have been a potential mediator. If so, our finding that neither of these risk factors contributed significantly to the outcomes measured suggests that either: 1) the effect of parenting inadequacy (although present within the first few postnatal weeks) is not demonstrable by 3 months or later; or 2) such parenting inadequacies are not reflected by the outcomes used in our study. Also, although there were a moderate number of primiparous mothers in our study sample, there were few mothers <18 years of age.
Although this study found that early discharge did not influence the selected outcomes in the population studied, caution must be exercised in generalizing these results to other populations. The study population was primarily Caucasian, non-Hispanic, and as noted above, there were few very young mothers in the sample. Socioeconomic status, as reflected by family income and insurance status, was only slightly lower than the national average. Psychosocial risk factors were not uncommon, yet most subjects had adequate social support from at least one source. Furthermore, although confounding psychosocial and demographic variables were well characterized and similar in early and late discharge groups, the cohorts evaluated were relatively small. It is possible that study of larger samples, less homogeneous groups, or groups at greater social risk might yield different results. Moreover, selection bias was almost certainly present, and it is likely that the mothers, who enrolled electively in the study, constituted a subgroup that had greater parenting capabilities than the population from which they were taken. Because the study was not randomized, it is also possible that mothers at highest risk for poor parenting were less likely to be discharged early, a factor that might have diminished any differences between the early and late discharge groups and contributed to the negative findings of the study. The discharge time difference between early and late groups, although consistent with those debated in the recent early discharge controversy,3,414-16 was relatively small, and it is possible that groups differing more substantially in length of stay might demonstrate different outcomes. Finally, the outcome variables were studied at times distant from birth and may very likely have been influenced by unexplored intervening factors during infancy.
Despite these limitations, the data presented constitute the most comprehensive evaluation of parenting outcomes as they relate to the perinatal length of stay and suggest that the effects of length of stay may be minimal. Further studies will be needed to assess these and other parenting outcomes in more socially diverse and high risk populations.
| |
ACKNOWLEDGEMENT |
|---|
This work was supported in part by Grant MCJ-040523-03-0 from the Maternal and Child Health Bureau.
| |
FOOTNOTES |
|---|
Received for publication Aug 28, 1998; accepted Apr 6, 1999.
Address correspondence to John R. Britton, MD, PhD, 3878 E Adonis Dr, Salt Lake City, UT 84124.
| |
REFERENCES |
|---|
|
Top
Abstract
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. R. Britton, H. L. Britton, and V. Gronwaldt Breastfeeding, Sensitivity, and Attachment Pediatrics, November 1, 2006; 118(5): e1436 - e1443. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Sheehan, P. Krueger, S. Watt, W. Sword, and B. Bridle The Ontario Mother and Infant Survey: Breastfeeding Outcomes J Hum Lact, August 1, 2001; 17(3): 211 - 219. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 1999 American Academy of Pediatrics. All rights reserved. |
||
|
||
