3. postpartum weeks and 2) to evaluate associations ceived encouragement. This project was approved by the KPMCP
between breastfeeding discontinuation and modifi- Institutional Review Board for the Protection of Human Subjects.
able factors, including maternal physical and mental
Statistical Analysis
health status, health services, and workplace issues.
We tested the specific hypotheses that mothers The primary outcomes of interest were breastfeeding discon-
tinuation at 2 and 12 weeks postpartum as previously defined. 2
whose clinicians supported breastfeeding and moth- analyses for categorical variables and the t test for continuous
ers who had fewer depressive symptoms would be variables were used to identify predictors associated with breast-
more likely to continue breastfeeding through 12 feeding discontinuation at 2 and 12 weeks. Predictors evaluated
weeks. were broadly categorized as sociodemographic, health services,
and health status variables. Demographic data examined included
maternal age, race/ethnicity, total household income, highest
METHODS level of education attained, marital status, and parity. Social fac-
tors included returning to work or school, the perceived impor-
Overview
tance of breastfeeding, reported confidence in the ability to breast-
This prospective cohort study included low-risk mothers and feed, requiring more help with their infant or with household
newborns in the Kaiser Permanente Medical Care Program (KP- chores, and maternal rating of the support received by the infant’s
MCP), Northern California Region, a group model managed care father and by family and friends to breastfeed. Health services
organization. Data were collected via 1) a face-to-face interview included the amount and quality of the breastfeeding advice that
during the postpartum hospitalization, 2) a telephone interview at mothers received by health care providers, encouragement to
2 weeks, and 3) a telephone interview at 12 weeks. Bivariate and breastfeed by health care providers, and participation in prenatal
multivariate analyses were conducted to identify predictors of breastfeeding classes. Maternal health status variables included
breastfeeding discontinuation at 2 weeks and at 12 weeks. depressive symptoms noted in the CES-D scale, having visited an
emergency department or clinic for their own health problem, and
Study Population perceived difficulty of their labor and delivery experience.
Logistic regression was performed to assess the independent
The cohort for the current study consisted of low-risk mothers effects of these predictors on breastfeeding discontinuation. Pre-
and newborns at KPMCP medical center in Sacramento, Califor- dictor variables associated with the outcome at P .15 or less in
nia, who had enrolled in a randomized controlled trial of home bivariate analyses were eligible for entry to preliminary multivar-
visits versus clinic-based follow-up during July 1996 through Sep- iate models. Each preliminary model included all eligible socio-
tember 1997. The randomized controlled trial did not find a dif- demographic predictors and one of the other predictors of interest
ference in mode of follow-up and breastfeeding continuation in a forced-entry logistic regression. Predictors of interest that
rates. Only those mother–newborn pairs who were medically and were associated with breastfeeding discontinuation at P .05 or
socially low risk were eligible. Infants who weighed 2500 or less in preliminary modeling were then entered into a final logistic
4600 g at birth, had stayed in the intensive care nursery, or had regression model. In the final model for breastfeeding discontin-
a medical problem that necessitated follow-up by a pediatrician or uation at 12 weeks, we eliminated 2 variables—perceived diffi-
nurse practitioner were excluded. Mothers who were 14 years old culty of labor and confidence in ability to feed the infant— because
or younger, had a positive toxicology screen for drugs of abuse they were highly correlated with maternal depressive symptoms.
after admission to labor and delivery, or spoke a language other The fit of the final logistic models was assessed as adequate on the
than English or Spanish were excluded. Finally, mothers and basis of Hosmer-Lemeshow tests and c-statistics. The Hosmer-
newborns whose anticipated length of stay was 48 hours, usu- Lemeshow P values for the 2- and 12-week models (.64 and .73,
ally as a result of cesarean delivery, were excluded. We did not respectively) were substantially .05. The c-statistics for the 2- and
collect information regarding breastfeeding duration on ineligible 12-week models were 0.716 and 0.725, respectively. All data anal-
mother–newborn pairs. Additional details are provided in the yses were performed in SAS, version 8.0 (SAS Institute, Cary, NC).
report of the randomized trial.11
RESULTS
Data Collection
Study Population
Research nurses used chart review and the enrollment inter-
view on the postpartum floor to collect baseline data on clinical Among the 1163 enrolled mother–newborn pairs,
and demographic variables, as well as on maternal experiences 1007 (87%) initiated breastfeeding and were included
and perceptions about prenatal care and breastfeeding. At 2 weeks in additional analyses. A total of 872 (75%) mothers
and 12 weeks postpartum, a research interviewer contacted each were breastfeeding at the 2-week interview; of these
mother by telephone to conduct a 15-minute interview about
breastfeeding, other outcomes, and satisfaction. At both 2 weeks women, 859 (99%) completed the 12-week interview.
and 12 weeks, discontinuation of exclusive breastfeeding was The study group (Table 1) was 62% white, 12% His-
defined as either giving no breast milk or still breastfeeding but panic, 6% black, 5% Asian, and 11% multicultural
giving 12 oz of formula per day. Using previous definitions of white, with a mean age of 28 6 years. Most mothers
breastfeeding, mothers who give 12 oz of formula per day
would be providing less than half of the average infant’s caloric
were married (89%) and had at least some college
intake via breastfeeding and would be considered “low partial or education (68%). Sixty-five percent of the mothers
token” breastfeeders.12,13 In the analysis, such mothers were reported household incomes of $30 000 per year;
grouped in the “breastfeeding discontinuation” category. however, a substantial group (14%) reported house-
Race/ethnicity was self-reported with the mother asked to hold incomes of $20 000 per year. Forty-one per-
name all racial or ethnic identifications that applied; for analysis,
respondents were categorized as white (non-Hispanic), black, His- cent of mothers were nulliparous, and the majority
panic, Asian, multicultural white, or other (categorization algo- (91%) had initiated prenatal care in the first trimes-
rithm available on request). The Center for Epidemiologic Studies ter.
Depression Scale (CES-D), a widely used 20-item instrument that
has been validated in English and Spanish, was used to evaluate Rates of Breastfeeding Discontinuation
maternal depressive symptoms using quartiles during the 2-week
telephone interview. We conducted a second telephone interview Among the original 1163 mothers in the cohort,
that focused on breastfeeding at 12 weeks postpartum. All data 1007 (87%) were breastfeeding at 1 to 2 days, 872
from 2 weeks to 12 weeks were collected retrospectively during (75%) were breastfeeding at the 2-week interview,
the 12-week interview. At this time, mothers were asked whether
they received encouragement to breastfeed from a doctor, nurse,
and 646 (55%) were breastfeeding at the 12-week
or breastfeeding consultant at KPMCP. Response categories were interview. In other words, of the 1007 mothers who
yes, no, or don’t know. Mothers were not asked when they re- initiated breastfeeding, 135 (13%) discontinued by 2
ARTICLES 109
Downloaded from www.pediatrics.org by on June 2, 2009
4. TABLE 1. Demographic Characteristics of Women and Associations With Breastfeeding Discontinuation at 2 and 12 Weeks, Kaiser
Permanente
Characteristic Total N N (%) P N (%) P
(%; N 1007) Discontinued Discontinued at
at 2 Weeks 12 Weeks
Mean maternal age (y) 28 ( 6) 27 ( 6) .0001 27 ( 6) .0001
Race/ethnicity .25 .28
White 628 (62) 75 (12) 193 (35)
Hispanic 121 (12) 18 (15) 45 (44)
Black 62 (6) 11 (18) 21 (41)
Asian or Pacific Islander 48 (5) 9 (19) 20 (50)
Multicultural white 111 (11) 14 (13) 33 (83)
Other 34 (3) 8 (24) 9 (39)
Prenatal care .27 .53
First trimester 921 (91) 121 (13) 292 (38)
Beyond first trimester 80 (8) 14 (18) 27 (23)
Parity .01 .03
0 previous children 414 (41) 71 (17) 143 (43)
1 previous child 348 (35) 35 (10) 110 (35)
2 or more children 245 (24) 29 (12) 68 (32)
Education .0001 .0001
Less than high school 65 (6) 16 (25) 22 (45)
High school graduate 262 (26) 50 (19) 91 (44)
Some college 411 (41) 48 (12) 144 (40)
Completed college 170 (17) 14 (8) 48 (31)
Postgraduate 99 (10) 7 (7) 16 (17)
Income .02 .0001
$20 000 144 (14) 30 (21) 67 (58)
$20 001–$30 000 167 (17) 22 (13) 60 (43)
$30 001–$40 000 156 (15) 23 (15) 51 (40)
$40 001–$55 000 213 (21) 29 (14) 64 (35)
$55 000 290 (29) 27 (9) 69 (27)
Marital status .005 .02
Married 898 (89) 111 (12) 279 (36)
Single 109 (11) 24 (22) 42 (49)
weeks postpartum. Of the 859 mothers who com- points. No significant differences were observed in
pleted the 12-week interview, another 213 mothers discontinuation among racial/ethnic groups in biva-
had discontinued breastfeeding by 12 weeks post- riate analysis.
partum (Fig 1). The sharpest drop in breastfeeding
occurred during the first 4 postpartum weeks.
Clinician Support and Maternal Mental Health
Sociodemographic Characteristics and Breastfeeding Mothers who reported (during the 12-week inter-
Discontinuation at 2 and 12 Weeks view) having received encouragement to breastfeed
Breastfeeding discontinuation at both 2 and 12 from a doctor, nurse, or breastfeeding consultant
weeks was higher among women who were were less likely to discontinue breastfeeding at 12
younger, had lower educational attainment, had weeks (P .015; Table 2).
lower household income, or were single parents (Ta- A total of 239 mothers (24%) had depressive symp-
ble 1). First-time mothers were more likely than mul- toms as measured by a score of 16 or higher on the
tiparas to discontinue breastfeeding at both time CES-D. Mothers with higher depressive symptom
scores at 2 weeks were more likely to discontinue
breastfeeding at 12 weeks (P .01).
Psychosocial and Work/School-Related Characteristics
Maternal attitudes reported at the 1- to 2-day in-
terview, including the perceived importance of
breastfeeding and confidence in ability to breastfeed,
were associated with breastfeeding continuation at
both 2 and 12 weeks (Table 2). Lack of support from
the father for breastfeeding, as reported during the
2-week interview, was associated with breastfeeding
discontinuation at 2 weeks but not at 12 weeks.
Among the 1007 mothers who initiated breastfeed-
ing, 476 (47%) had returned to work or school by the
12-week interview. Returning to work or school was
1 of the strongest predictors of breastfeeding discon-
Fig 1. Proportion of mothers who breastfed, by week, among tinuation (P .0001) at 12 weeks postpartum. Simi-
low-risk mothers, Kaiser Permanente. larly, 236 (50%) women who returned to work or
110 CLINICIAN SUPPORT AND BREASTFEEDING
Downloaded from www.pediatrics.org by on June 2, 2009
5. TABLE 2. Psychosocial and Work/School-Related Characteristics and Associations With Breastfeeding Discontinuation at 2 and 12
Weeks, Kaiser Permanente
Timing of Characteristic Total N N (%) P N (%) P
Interview (%; N 1007) Discontinued Discontinued at
at 2 Weeks 12 Weeks
1–2 d Perceived breastfeeding importance .0001 .0001
Very important 946 (94) 110 (12) 292 (36)
Not important 53 (5) 22 (42) 26 (70)
Confidence in ability to breastfeed .04 .04
Very confident 680 (68) 81 (12) 206 (35)
Not very confident 327 (32) 54 (17) 115 (42)
Breastfeeding classes .49 .09
Yes 362 (36) 45 (12) 106 (34)
No 645 (64) 90 (14) 215 (40)
2 wk Breastfeeding support from father .002 .58
of infant
A great deal 840 (83) 96 (11) 58 (11)
A little to none at all 120 (12) 26 (22) 38 (12)
Breastfeeding problems with current .001 .03
infant at 2–3 d
None at all 791 (79) 90 (11) 244 (35)
Somewhat serious 123 (12) 26 (21) 45 (47)
Serious problems 90 (9) 19 (21) 31 (44)
Maternal depression score .39 .01
Lowest quartile 228 (23) 26 (11) 57 (29)
Second quartile 238 (24) 27 (11) 68 (33)
Third quartile 299 (30) 44 (15) 108 (43)
Highest quartile 239 (24) 37 (15) 87 (43)
12 wk Returned to work or school 476 (47) NA NA 197 (48) .0001
Problems breastfeeding at work 236 (23) NA NA 136 (58) .0001
or school
Encouraged to breastfeed by health .015
care provider
Yes 181 (18) NA NA 50 (30)
No 777 (77) 268 (40)
NA indicates not applicable.
school reported problems trying to continue breast- tant at the 1- to 2-day interview (OR: 0.24; 95% CI:
feeding, and those who reported problems were sig- 0.11– 0.53). Breastfeeding discontinuation was more
nificantly more likely to discontinue by 12 weeks frequent among mothers with more depressive
(P .0001). symptoms as measured by the CES-D scale (OR: 1.18
per quartile; 95% CI: 1.01–1.37). Mothers who re-
Multivariate Analysis turned to work or school (OR: 2.4; 95% CI: 1.75–3.3)
In the final multivariate models, mothers who ex- or experienced problems breastfeeding or pumping
perienced breastfeeding problems at 2 to 3 days at work or school (OR: 3.2; 95% CI: 1.9 –5.39) also had
(odds ratio [OR]: 1.5; 95% confidence interval [CI]: increased odds of discontinuation at 12 weeks (Table
1.1–1.97) or reported a lack of confidence in their 3).
ability to breastfeed their infant at the 1- to 2-day
interview (OR: 2.8; 95% CI: 1.02–7.6) were more
likely to have discontinued breastfeeding by 2 weeks Reported Breastfeeding Problems
postpartum. Those who rated breastfeeding as im- At the 1- to 2-day interview, approximately 21% of
portant at the 1- to 2-day interview (OR: 0.29; 95% CI: mothers reported experiencing breastfeeding prob-
0.15– 0.58) were less likely to have discontinued lems that they described as either somewhat serious
breastfeeding at 2 weeks. Other independent predic- or serious. These mothers were more likely to dis-
tors of breastfeeding discontinuation at 2 weeks in- continue breastfeeding at 2 weeks and 12 weeks (Ta-
cluded lower maternal education (OR: 1.5; 95% CI: ble 2). Reasons for breastfeeding discontinuation var-
1.2–1.9) and being of Asian race/ethnicity (OR: 2.6; ied by week (Table 4). In the first postpartum week,
95% CI: 1.1–5.7). Women who reported a lack of mothers reported breastfeeding discontinuation as a
support from the father of their infant were more result of problems with their infant sucking or latch-
likely to discontinue breastfeeding at 2 weeks, but ing on (23%) and breast pain and soreness (14%).
this did not reach statistical significance (OR: 1.7; Another frequently reported problem was the belief
95% CI: 0.99 –2.91). that the infant was still hungry or that they were not
In multivariate models to evaluate predictors of producing enough breast milk (27% at 0 –1 week,
breastfeeding at 12 weeks, mothers were much less 38% at 4 – 6 weeks).
likely to discontinue breastfeeding by 12 weeks if Returning to work or school progressively became
they reported (at 12 weeks) having received encour- the greatest reason for breastfeeding discontinuation
agement from their clinician to breastfeed (OR: 0.56; by 10 to 12 weeks. The proportion of mothers who
95% CI: 0.37– 0.84) or rated breastfeeding as impor- cited return to work as the main reason for breast-
ARTICLES 111
Downloaded from www.pediatrics.org by on June 2, 2009
6. TABLE 3. Predictors of Breastfeeding Discontinuation at 2 and 12 Weeks in Multivariate Models
Timing of Predictor Odds of Breastfeeding
Interview Discontinuation at
2 Weeks 12 Weeks
(OR [95% CI]) (OR [95% CI])
1–2 d Lower maternal education* 1.5 (1.2–1.9) 1.24 (1.04–1.48)
Decreasing maternal age† 1.1 (0.89–1.31) 1.22 (1.05–1.42)
Lower household income‡ 1.0 (0.88–1.22) 1.23 (1.08–1.39)
Asian§ 2.6 (1.1–5.7) 2.3 (1.1–4.93)
Lack of confidence in ability to breastfeed 2.8 (1.02–7.6) 1.2 (0.84–1.7)
Perceived breastfeeding importance¶ 0.29 (0.14–0.58) 0.24 (0.11–0.53)
2 wk Lack of support from father of infant# 1.7 (0.99–2.91) 1.1 (0.66–1.9)
Problems breastfeeding current infant at 2–3 d** 1.5 (1.1–1.97) 1.2 (0.94–1.6)
Maternal depressive symptoms†† 1.07 (0.88–1.3) 1.18 (1.01–1.37)
12 wk Encouraged to breastfeed by health care NA 0.56 (0.37–0.84)
provider‡‡
Returned to work/school# NA 2.4 (1.75–3.30)
Problems breastfeeding at work/school# NA 3.2 (1.90–5.39)
* Education, measured as the highest grade or year completed in school, used as interval variable with
6 intervals.
† Odds of breastfeeding discontinuation for every 5-year decrease in age.
‡ Household income, measured as category that best described their total household income in 1995,
before taxes.
§ Respondents were categorized as white (non-Hispanic), black, Hispanic, Asian, multicultural white,
or other (categorization algorithm available on request). White mothers used as reference group.
Measured as a 4-category variable, not confident at all, not very confident, somewhat confident, and
very confident, and recategorized as a binary variable, very confident and not very confident.
¶ Measured as a 4-category variable, not very important, somewhat important, very important, and
extremely important, and recategorized as a binary variable, very important and not important.
# Binary variable, yes or no.
** Binary variable measured as any problems breastfeeding their infant at 1 to 2 days.
†† Scores on CES-D, by quartiles.
‡‡ Response categories were yes, no, or don’t know. Mothers were not asked when they received
encouragement.
TABLE 4. Reasons for Breastfeeding Discontinuation Vary by Week
Main Reason for Discontinuation Week of Breastfeeding Discontinuation
0–1 2–3 4–6 7–9 10–12
(n 105) (n 74)* (n 112) (n 53) (n 19)†
Infant still hungry/not enough milk 27% 18% 38% 28% 11%
Problems sucking/latching on 23% 12% 1% 1% 5%
Breast pain/soreness 14% 14% 4% 0% 0%
Mother returned to work or school 4% 14% 29% 34% 58%
Mother sick or on medication 2% 12% 6% 11% 11%
Bottle feeding easier or more convenient 8% 9% 7% 13% 0%
Lack of energy/desire to discontinue 4% 8% 7% 2% 4%
Other‡ 18% 13% 8% 11% 11%
* Mothers were asked, during the 2- and 12-week interviews, to report when they discontinued and
the reason for breastfeeding discontinuation. Of the 135 women who discontinued breastfeeding at 2
weeks, 132 (98%) women responded to this question.
† Of the 321 additional women who discontinued breastfeeding after 2 weeks, 231 (72%) responded
to this question.
‡ Other reported reasons for discontinuation included infant not gaining weight or sick, breast milk
intolerance, and infant spitting up.
feeding discontinuation increased from 14% at 2 to 3 tional videos. Many women reported receiving both
weeks to 58% at 10 to 12 weeks postpartum. The written (64%) and verbal (63%) breastfeeding infor-
majority of problems reported among women who mation during their postpartum hospital stay. Thir-
returned to work were restricted schedules and ty-five percent of women received help from a lacta-
breaks (51%) and insufficient privacy (20%). Among tion consultant, 57% received information regarding
women who returned to school, the absence of on- breast pumps, and 65% reported receiving individu-
site child care was also reported as a barrier to alized support and breastfeeding instruction from a
breastfeeding continuation (23%). health care provider. This type of individualized in-
struction was chosen by 49% of respondents as the
Health Services Used in Support of Breastfeeding most helpful service actually received in support of
A total of 165 women (16%) participated in breast- breastfeeding and named by 63% of respondents as
feeding classes, predominantly in the prenatal pe- the most helpful service that they could have been
riod, and 230 (23%) viewed breastfeeding instruc- offered by their health care providers.
112 CLINICIAN SUPPORT AND BREASTFEEDING
Downloaded from www.pediatrics.org by on June 2, 2009
7. DISCUSSION Maternal Mental Health
Major Findings Postpartum depression is a serious, common, and
Our findings suggest that clinician support and treatable condition that often goes unrecognized.23
maternal mental health status deserve attention as Although previous studies have shown a relation-
modifiable factors in promoting breastfeeding con- ship between maternal depressive symptoms and
tinuation. Women who reported that their health breastfeeding cessation,24 very few have examined
care providers encouraged them to breastfeed were the adequacy of health care resources in support of
approximately half as likely to discontinue breast- postpartum mothers.25 Our study shows that women
feeding by 12 weeks postpartum as those who did with more maternal depressive symptoms as mea-
not. Conversely, the odds of breastfeeding discontin- sured by the CES-D scale had greater odds of breast-
uation were substantially elevated for each increase feeding discontinuation by 12 weeks. Thus, clini-
in quartile in the CES-D, a measure of depressive cians’ recognition of postpartum depressive
symptoms. The observational design of the current symptoms may contribute to breastfeeding continu-
study does not permit firm causal inferences be- ation as well as prevent other adverse effects on
tween breastfeeding duration and clinician support mothers and infants.26
or maternal depressive symptoms but does suggest
that interventions in these areas deserve additional Other Factors That Influence Breastfeeding
evaluation. Discontinuation
Two important independent predictors of early
breastfeeding discontinuation in our study— breast-
Clinician Support feeding problems at 2 to 3 days and lack of confi-
Our results are in accordance with previous stud- dence in the ability to breastfeed— have been previ-
ies that suggest that clinicians and other health care ously found to be associated with early
providers may have an influential role in breastfeed- discontinuation of breastfeeding.9 Early (2–3 days)
ing initiation and continuation. An analysis of na- breastfeeding problems reported by mothers in our
tional survey data collected from parents at up to 3 study were perceptions of insufficient milk supply,
years after childbirth found that provider support breast soreness, and problems with their infants
was associated with breastfeeding initiation.14 A pro- latching on. Current guidelines on maternal and neo-
spective, observational study suggested that health natal follow-up tend to focus on catastrophic, rare
system support of breastfeeding during the postpar- events rather than on common problems such as
tum hospitalization and early postdischarge period maternal perception of insufficient milk supply.27–29
was associated with successful breastfeeding.10 Development of evidence-based recommendations
Our study is distinct from previous research in and anticipatory guidance for common, early breast-
that we evaluated the association between clinician feeding problems may improve rates of breastfeed-
support and breastfeeding continuation at 12 weeks ing continuation.
postpartum. In addition, we were able to control for This study confirms previous findings that return-
many other prospectively collected demographic ing to work is associated with lower rates of breast-
and psychosocial factors, including maternal confi- feeding initiation and continuation.7,30 The most
dence in ability to breastfeed and belief in the impor- common workplace problems reported by women in
tance of breastfeeding. our study were restricted schedules and breaks and
Behavior-oriented interventions to support breast- insufficient privacy. Given the large number of
feeding have been found effective in several random- women who return to work or school in the postpar-
ized, controlled trials.15–17 However, these interven- tum period, workplace and school policies to en-
tions have usually been conducted by health hance scheduling flexibility and privacy for breast-
educators or lactation consultants and have almost feeding mothers should be encouraged.
always involved a dedicated amount of time (for Demographic variables such as maternal age, ed-
example, a 50- to 80-minute group counseling ses- ucation, income, and race have been studied exten-
sion) outside usual prenatal or pediatric preventive sively.8 To our knowledge, our study is one of the
visits. We are unaware of any controlled trials that first to document higher breastfeeding discontinua-
have evaluated the effectiveness of clinician support tion rates among mothers of Asian race/ethnicity.31
for breastfeeding delivered during usual preventive Additional research is warranted to determine the
visits. Several studies have found that pediatricians, reasons for breastfeeding discontinuation among
obstetricians, and family practitioners lack knowl- mothers of Asian race/ethnicity.
edge and training on breastfeeding topics.18,19 Stud-
ies of other topics, including smoking cessation,20,21 Limitations
suggest that physician-delivered counseling can be Our study should be interpreted keeping in mind
effective in promoting preventive health behaviors. several limitations. First, it focused on a medically
Our study suggests that pediatrician skills to support and socially low-risk population of mother–infant
breastfeeding deserve attention in breastfeeding pro- pairs in an integrated health maintenance organiza-
motion efforts22 and that randomized, controlled tri- tion (HMO). The mothers in the study had diverse
als of interventions to enhance clinician support of racial/ethnic backgrounds, but nearly two thirds had
breastfeeding during routine preventive visits are at least some college education and 10% had less
warranted. than a high school degree. The HMO in this study
ARTICLES 113
Downloaded from www.pediatrics.org by on June 2, 2009
8. primarily serves families with employment-based, tion. Attention should also be paid to maternal men-
comprehensive health insurance coverage; the study tal health status not only for the health benefits to
population’s median income was higher than that of mothers but also as a potentially modifiable factor in
the general population. Although our population promoting breastfeeding continuation.
was ethnically diverse, our results may not be gen-
eralizable to more socioeconomically disadvantaged
populations or to families who receive care in less ACKNOWLEDGMENTS
integrated settings. Our results also may not be gen- Results from this study were presented at the Society for Pedi-
atric Research meeting in May 2002 and have been printed in
eralizable to mothers or newborns with medical com- abstract form as part of the proceedings.
plications or mothers who delivered by cesarean sec- This work was supported by the Innovation Program of Kaiser
tion. Additional research involving such populations Permanente Medical Care Program, Northern California; grants
and settings is needed. MCJ 067951 and 6 H 16 MC 00050 from the Maternal and Child
Mothers in the study were asked to recall events Health Bureau, Health Resources and Services Administration,
Department of Health and Human Services; and the Agency for
that may have taken place 12 weeks or more before Health care Research and Quality. Dr Taveras was supported by
the interview. In particular, mothers were asked dur- grant T32 PE 10018 from the Health Resources and Services Ad-
ing the 12-week interview to recall whether they ministration, Department of Health and Human Services, to the
were encouraged to breastfeed by a health care pro- Harvard Pediatric Health Services Research Fellowship Program,
Boston.
vider. The observed association between clinician We thank the many research staff, clinicians, and mothers who
support and breastfeeding continuation could have contributed support and interview data for this study.
stemmed from recall bias in that women who were
still breastfeeding at 12 weeks may have been more
likely to remember being encouraged to do so. An- REFERENCES
other possible explanation for the observed associa- 1. American Academy of Pediatrics Work Group on Breastfeeding. Breast-
tion is selection bias: mothers who intended to feeding and the use of human milk. Pediatrics. 1997;100:1035–1039
2. US Department of Health and Human Services. Developing Objectives for
breastfeed might have chosen clinicians who were Healthy People 2010. Washington, DC: US Department of Health and
more likely to encourage it. Human Services, Office of Disease Prevention and Health Promotion;
Our analysis focused on the mother–infant pairs 1997
that deviated from American Academy of Pediatrics 3. US Department of Health and Human Services. HHS Blueprint for Action
on Breastfeeding. Washington, DC: US Department of Health and Human
recommendations for exclusive breastfeeding during
Services, Office on Women’s Health; 2000
the first 6 months of life. Our definition of breast- 4. Scott JA, Binns CW. Factors associated with the initiation and duration
feeding continuation was relatively conservative in of breastfeeding: a review of the literature. Breastfeed Rev. 1999;7:5–16
that we classified mothers who were giving more 5. Avery M, Duckett LJ, Dodgson J. Factors associated with very early
than half of an average infant’s caloric intake as weaning among primiparas intending to breastfeed. Matern Child Health
J. 1998;2:167–179
formula as having discontinued exclusive breast- 6. Deshpande AD, Gazmararian JA. Breast-feeding education and
feeding. Thus, we classified token and low-partial support: association with the decision to breast-feed. Eff Clin Pract.
breastfeeding as having discontinued exclusive 2000;3:116 –122
breastfeeding. 7. Piper S, Parks PL. Predicting the duration of lactation: evidence from a
national survey. Birth. 1996;23:7–12
Given this study’s observational design, it is im-
8. Losch M, Dungy CI, Russell D, Dusdieker LB. Impact of attitudes on
portant to acknowledge the potential role of unmea- maternal decisions regarding infant feeding. J Pediatr. 1995;126:507–514
sured characteristics that might explain the observed 9. Ertem IO, Votto N, Leventhal JM. The timing and predictors of the early
associations among clinician support, maternal de- termination of breastfeeding. Pediatrics. 2001;107:543–548
pressive symptoms, and breastfeeding duration. 10. Kuan LW, Britto M, Decolongon J, Schoettker PJ, Atherton HD, Kotagal
UR. Health system factors contributing to breastfeeding success. Pedi-
Given the range of maternal demographic and psy- atrics. 1999;104:1–7
chosocial characteristics that we were able to con- 11. Lieu TA, Braveman PA, Escobar GJ, Fischer AF, Jensvold NG, Capra
sider, however, the overall pattern of findings indi- AM. A randomized comparison of home and clinic follow-up visits
cate that these 2 factors are strongly associated with after early postpartum hospital discharge. Pediatrics. 2000;105:
1058 –1065
breastfeeding duration. A recent cluster-randomized
12. Labbok M, Krasovec K. Toward consistency in breastfeeding defini-
trial in Belarus found that an intervention emphasiz- tions. Stud Fam Plann. 1990;21:226 –230
ing healthcare working assistance with breastfeeding 13. Labbok MH, Belsey M, Coffin CJ. A call for consistency in defining
increased duration of breastfeeding at 3 and 6 breast-feeding. Am J Public Health. 1997;87:1060 –1061
months.32 Controlled trials of counseling interven- 14. Lu MC, Lange L, Slusser W. Provider encouragement of breastfeeding:
evidence from a national survey. Obstet Gynecol. 2001;97:290 –295
tions by primary care providers delivered during 15. Brent NB, Redd B, Dworetz A. Breast-feeding in a low-income popula-
usual preventive visits should be performed in the tion. Program to increase incidence and duration. Arch Pediatr Adolesc
United States. Furthermore, for evaluating the suc- Med. 1995;149:798 – 803
cess of future interventions to promote breastfeed- 16. Sciacca JP, Dube DA, Phipps BL, Ratliff MI. A breastfeeding education
and promotion program: effects on knowledge, attitudes, and support
ing, more sensitive, clinically detailed measures of for breastfeeding. J Community Health. 1995;20:473– 490
breastfeeding problems are needed. 17. Kistin N, Benton D, Rao S, Sullivan M. Breastfeeding rates among black
urban low-income women: effect of prenatal education. Pediatrics. 1990;
86:741–746
CONCLUSIONS 18. Freed GL, Clark SJ, Sorenson J. National assessment of physicians’
Our study adds to the growing body of literature breast-feeding knowledge, attitudes, training, and experience. JAMA.
1995;273:472– 476
suggesting that support and encouragement to 19. Schanler RJ, O’Connor KG, Lawrence RA. Pediatricians’ practices and
breastfeed from health care providers is associated attitudes regarding breastfeeding promotion. Pediatrics. 1999;103(3).
with a higher likelihood of breastfeeding continua- Available at: http://www.pediatrics.org/cgi/content/full/103/3/e35
114 CLINICIAN SUPPORT AND BREASTFEEDING
Downloaded from www.pediatrics.org by on June 2, 2009
9. 20. Kottke TE, Battista RN, DeFriese GH. Attributes of successful smoking 26. Miller LJ. Postpartum depression. JAMA. 2002;287:762–765
cessation intervention in medical practice: a meta-analysis of 39 con- 27. American Academy of Pediatrics. Hospital stay for healthy term new-
trolled trials. JAMA. 1988;259:2883–2889 borns. Pediatrics. 1995;96:788 –790
21. Fiore MC, Bailey WC, Cohen SJ. Smoking Cessation. Clinical Practice 28. Escobar G, Braveman P, Ackerson L, et al. A randomized comparison of
Guideline No. 18. Rockville, MD: US Department of Health and Human home visits and hospital-based group follow-up visits after early post-
Services, Public Health Service, Agency for Health Care Policy and partum discharge. Pediatrics. 2001;108:719 –727
Research; 1996 29. Escobar GJ, Gonzales VM, Armstrong MA, Folck BF, Xiong B, Newman
22. Philipp BL, Merewood A, O’Brien S. Physicians and breastfeeding TB. Rehospitalization for neonatal dehydration: a nested case-control
promotion in the United States: a call for action. Pediatrics. 2001;107: study. Arch Pediatr Adolesc Med. 2002;156:155–161
584 –587 30. Meek JI. Breastfeeding in the workplace. Pediatr Clin North Am. 2001;
23. Georgiopoulos AM, Bryan TL, Wollan P. Routine screening for post- 48:461– 474
partum depression. J Fam Pract. 2001;50:117–122 31. Serdula MK, Cairns KA, Williamson DF, Fuller M, Brown JE. Correlates
24. Misri S, Sinclair DA, Kuan AJ. Breast-feeding and postpartum of breast-feeding in a low-income population of whites, blacks, and
depression: is there a relationship? Can J Psychiatry. 1997;42:1061–1065 southeast Asians. J Am Diet Assoc. 1991;91:41– 45
25. Gunn J, Lumley J, Chondros P, Young D. Does an early postnatal 32. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding
check-up improve maternal health: results from a randomised trial in Intervention Trial (PROBIT): a randomized trial in the Republic of
Australian general practice. Br J Obstet Gynaecol. 1998;105:991–997 Belarus. JAMA. 2001;285:413– 420
CHANCE TO CHOICE
“Yet another reason why the search for ethics has emerged now is that the new
science has moved us from chance to choice in many matters—for instance, repro-
duction.”
Somerville M. The Ethical Canary. Viking; 2000
ARTICLES 115
Downloaded from www.pediatrics.org by on June 2, 2009
10. Clinician Support and Psychosocial Risk Factors Associated With Breastfeeding
Discontinuation
Elsie M. Taveras, Angela M. Capra, Paula A. Braveman, Nancy G. Jensvold, Gabriel J.
Escobar and Tracy A. Lieu
Pediatrics 2003;112;108-115
DOI: 10.1542/peds.112.1.108
Updated Information including high-resolution figures, can be found at:
& Services http://www.pediatrics.org/cgi/content/full/112/1/108
References This article cites 28 articles, 13 of which you can access for free at:
http://www.pediatrics.org/cgi/content/full/112/1/108#BIBL
Citations This article has been cited by 30 HighWire-hosted articles:
http://www.pediatrics.org/cgi/content/full/112/1/108#otherarticles
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
Nutrition & Metabolism
http://www.pediatrics.org/cgi/collection/nutrition_and_metabolism
Permissions & Licensing Information about reproducing this article in parts (figures, tables) or
in its entirety can be found online at:
http://www.pediatrics.org/misc/Permissions.shtml
Reprints Information about ordering reprints can be found online:
http://www.pediatrics.org/misc/reprints.shtml
Downloaded from www.pediatrics.org by on June 2, 2009
![Clinician Support and Psychosocial Risk Factors Associated With
Breastfeeding Discontinuation
Elsie M. Taveras, MD, MPH* ; Angela M. Capra, MA‡; Paula A. Braveman, MD, MPH§;
Nancy G. Jensvold, MPH‡; Gabriel J. Escobar, MD‡; and Tracy A. Lieu, MD, MPH
ABSTRACT. Objective. Breastfeeding rates fall short sues may help to promote breastfeeding continuation
of goals set in Healthy People 2010 and other national among mothers who initiate. Policies to enhance sched-
recommendations. The current, national breastfeeding uling flexibility and privacy for breastfeeding mothers at
continuation rate of 29% at 6 months lags behind the work or school may also be important, given the elevated
Healthy People 2010 goal of 50%. The objective of this risk of discontinuation associated with return to work or
study was to evaluate associations between breastfeed- school. Pediatrics 2003;112:108 –115; breastfeeding, neona-
ing discontinuation at 2 and 12 weeks postpartum and tal, health services, clinician support, maternal depres-
clinician support, maternal physical and mental health sion.
status, workplace issues, and other factors amenable to
intervention.
Methods. A prospective cohort study was conducted ABBREVIATIONS. KPMCP, Kaiser Permanente Medical Care
Program; CES-D, Center for Epidemiologic Studies Depression
of low-risk mothers and infants who were in a health
Scale; OR, odds ratio; CI, confidence interval; HMO, health main-
maintenance organization and enrolled in a randomized, tenance organization.
controlled trial of home visits. Mothers were interviewed
in person at 1 to 2 days postpartum and by telephone at
T
2 and 12 weeks. Logistic regression modeling was per- he promotion and support of breastfeeding has
formed to assess the independent effects of the predic- emerged as a national public health priority in
tors of interest, adjusting for sociodemographic and other recent years. The American Academy of Pedi-
confounding variables. atrics has recommended exclusive breastfeeding for
Results. Of the 1163 mother–newborn pairs in the
cohort, 1007 (87%) initiated breastfeeding, 872 (75%) were
6 months, continuing to 1 year or beyond.1 In its
breastfeeding at the 2-week interview, and 646 (55%) Healthy People 2010 recommendations, the US De-
were breastfeeding at the 12-week interview. In the final partment of Health and Human Services set goals of
multivariate models, breastfeeding discontinuation at 2 75% of mothers to breastfeed exclusively in the early
weeks was associated with lack of confidence in ability postpartum period, 50% to continue to 5 to 6 months,
to breastfeed at the 1- to 2-day interview (odds ratio [OR]: and 25% to continue to 1 year.2
2.8; 95% confidence interval [CI]: 1.02–7.6), early breast- National statistics show large gaps between the
feeding problems (OR: 1.5; 95% CI: 1.1–1.97), Asian race/ reality and the goals, particularly for breastfeeding
ethnicity (OR: 2.6; 95% CI: 1.1–5.7), and lower maternal continuation. In 1998, 64% of mothers initiated
education (OR: 1.5; 95% CI: 1.2–1.9). Mothers were much
breastfeeding in-hospital, but rates of any breastfeed-
less likely to discontinue breastfeeding at 12 weeks post-
partum if they reported (during the 12-week interview) ing were only 29% by 6 months and 16% by 1 year.3
having received encouragement from their clinician to Studies have found elevated risk of breastfeeding
breastfeed (OR: 0.6; 95% CI: 0.4 – 0.8). Breastfeeding dis- discontinuation among mothers who are older, have
continuation at 12 weeks was also associated with demo- lower educational attainment, and return to work
graphic factors and maternal depressive symptoms (OR: full-time.4 –7 Lack of support from family and friends,
1.18; 95% CI: 1.01–1.37) and returning to work or school insufficient prenatal breastfeeding education, and
by 12 weeks postpartum (OR: 2.4; 95% CI: 1.8 –3.3). hospital discharge packs that contain infant formula
Conclusions. Our results indicate that support from have also been associated with breastfeeding discon-
clinicians and maternal depressive symptoms are associ- tinuation.8,9
ated with breastfeeding duration. Attention to these is-
Among the many studies of breastfeeding contin-
uation, few have evaluated factors amenable to in-
From the *Harvard Pediatric Health Services Research Fellowship Program,
tervention by clinicians in office settings.10 During
Boston, Massachusetts; ‡Division of Research, Kaiser Permanente, Oakland, the first 12 weeks postpartum, mothers must adjust
California; §Department of Family and Community Medicine, University of to many new physical, psychological, and social de-
California, San Francisco, San Francisco, California; and Center for Child mands. They and their newborns also regularly see
Health Care Studies, Department of Ambulatory Care and Prevention,
health care providers for routine preventive visits,
Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massa-
chusetts. which are an opportunity for support of breastfeed-
Received for publication Apr 24, 2002; accepted Oct 23, 2002. ing.
Address correspondence to Elsie M. Taveras, MD, MPH, Harvard Pediatric This study was designed to address gaps in exist-
Health Services Research Fellowship Program, Children’s Hospital of Bos- ing knowledge about clinician support and maternal
ton, 333 Longwood Ave, LO-240, Boston, MA 02115. E-mail: elsie.taveras@
tch.harvard.edu
mental health status as influences on breastfeeding
PEDIATRICS (ISSN 0031 4005). Copyright © 2003 by the American Acad- continuation. Our aims were 1) to describe reasons
emy of Pediatrics. for breastfeeding discontinuation during the first 12
108 PEDIATRICS Vol. 112 No. 1 July 2003
Downloaded from www.pediatrics.org by on June 2, 2009](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)