Professional Documents
Culture Documents
ABSTRACT
Purpose: To identify risk factors for lack of breastfeeding initiation and duration of 1 month
among a racially diverse cohort of women. In particular, our interest was to examine depres-
sive symptoms during pregnancy as a potential risk factor for not initiating or continuing
breastfeeding.
Methods: Survey and medical record data from a cohort of pregnant women from the San
Francisco Bay area who delivered a singleton infant (n 1448) were analyzed to examine lack
of breastfeeding initiation and duration of 1 month.
Results: In this study, 5.6% of women did not initiate breastfeeding, and 11.1% of women
who initiated breastfeeding had a duration of breastfeeding of 1 month. There were no racial
or ethnic differences in initiation of breastfeeding after adjusting for demographic and clinical
characteristics. At 1 month postpartum, African American women were more likely than white
women to have a duration of breastfeeding lasting 1 month. Depressive symptoms during or
prior to pregnancy had no effect on initiation of breastfeeding even when symptoms were per-
sistent. Women with persistent depressive symptoms (symptoms at two time points, including
one prior to delivery) were more likely to have breastfeeding duration of 1 month (odds ra-
tio [OR] 1.77, 95% confidence interval [95% CI] 1.10-2.86), whereas depressive symptoms at a
single time point were not associated with breastfeeding duration of 1 month.
Conclusions: Addressing depressive symptoms experienced by women both during and af-
ter pregnancy may improve the duration of breastfeeding.
B REASTFEEDING IS AN IMPORTANT,
yet unrealized
public health goal in the United States. In
goals of 75% during the early postpartum period,
50% at 6 months, and 25% at 1 year.1 In 2001, the
1Department of Medicine, Brigham and Womens Hospital, Department of Pediatrics, Massachusetts General Hos-
pital, Department of Medicine, Childrens Hospital Boston, and Harvard Medical School, Boston, Massachusetts.
2Division of General Medicine and Primary Care, Department of Medicine, Brigham and Womens Hospital, Boston,
Massachusetts.
3Departments of Obstetrics, Gynecology and Reproductive Sciences, and Epidemiology and Biostatistics, Univer-
cisco, California.
This work was supported by the National Institute of Child Health and Human Development (R01 HD37389).
754
6303_09_p754-762 8/3/06 1:35 PM Page 755
symptoms were assessed by the short-form Cen- (subjects were asked if they ever lacked money
ter for Epidemiologic Studies-Depression Scale for food). Additional measures included parity
(CES-D), a screening tool consisting of 10 ques- (nulliparous, parous), method of delivery (ce-
tions.36 The long form CES-D is considered to be sarean section vs. other), breastfeeding instruc-
both reliable and valid, and the short form CES- tion or preparation (yes/ no), smoking during
D has been shown to generate similar results to pregnancy or postpartum (yes/ no), presence of
those obtained through use of the long form.36 a chronic disease during pregnancy (including
The short form has also demonstrated favorable hypertension, asthma, thyroid disease, lung dis-
test characteristics, with an overall test-retest ease, anemia, arthritis, epilepsy, kidney disease,
score correlation of 0.71.36 Studies have generally liver disease, heart disease, cancer, and HIV), self-
supported the reliability of both the short and reported health status postpartum (excellent or
long form CES-D among various racial/ethnic good vs. fair or poor), and perceived stress (none,
groups.3740 The CES-D has been used among re- some, high), which was measured using the
productive age women both during pregnancy short-form Perceived Stress Scale.45,46
and in the postpartum period.4144 The first sur-
vey focused on symptoms experienced in the Data analysis
month prior to pregnancy, and subsequent sur-
Multivariate logistic regression models were
veys focused on symptoms experienced during
built with variables shown in the literature to
the week prior to survey administration. For the
be associated with breastfeeding or found to
initiation model, women were classified as hav-
be associated in bivariate analysis (p
ing depressive symptoms at either zero, one, or
0.05).47,9,10,13,1521 To prevent overfitting of the
two time points based on their responses to the
models resulting from the modest number of wo-
first three surveys, which were all conducted dur-
men who did not breastfeed, we limited the num-
ing pregnancy and reflect symptoms either prior
ber of variables in final models by including only
to pregnancy (survey one) or during pregnancy
those variables with a p 0.05 for both initiation
(surveys two and three). One hundred forty-nine
and duration, with the exception of three vari-
women did not complete both surveys two and
ables (age, lack of money for food, and depres-
three, which resulted in the choice of two time
sion [our variable of primary interest]). We tested
points, rather than three, as the upper limit of the
the interaction between depressive symptoms
depression scale. If survey three was available, it
and race/ethnicity in both models. Statistical
was used; if not, survey two was used. For the
analyses were performed using SAS for Win-
duration model, women were classified as hav-
dows, version 8.0 (SAS Institute, Inc., Cary, NC).
ing depressive symptoms at zero, one, or two or
more time points based on their responses to the
four surveys. The fourth survey reflects symp-
RESULTS
toms in the postpartum period. Again, if survey
three was available, it was used; if not, survey
Response and retention rates
two was used. Given the small number of women
(61) with depressive symptoms at three time Of 2854 eligible women, 1809 (63%) enrolled in
points, they were combined in the analysis with the study. Postpartum surveys were available for
the women who met criteria at two time points. 1480 of the 1648 women who subsequently de-
A woman was considered to have depressive livered at one of the participating sites (90%).
symptoms if her CES-D score was 10, the Thirty-two women were excluded from this
threshold used for this short form.36 analysis because of identification of race/ethnic-
Other covariates included sociodemographic ity as Native American (n 4), missing breast-
characteristics, such as age (categorized in 5-year feeding data (n 1), missing CES-D score during
intervals, except the oldest group included 11 pregnancy (n 21), and a short (1 month) or
years because of smaller sample size), race/eth- missing interval between delivery and the post-
nicity (white, Latina, African American, Asian), partum interview (n 6). The total sample size
level of educational attainment ( high school, was 1448. Respondents and nonrespondents were
high school graduate or equivalent, some college, of similar age. Nonrespondents were more likely
college graduate), marital status (married or liv- than respondents to be Asian American and less
ing with a partner vs. other), and economic means likely to be African American, Latina, or white.34
6303_09_p754-762 8/3/06 1:35 PM Page 757
Description of the sample D short form criteria at multiple time points. Fac-
tors that were significantly associated with a lack
Overall, 5.6% of women did not initiate breast-
of breastfeeding initiation included lower educa-
feeding, and among those who initiated breast-
tional attainment, insufficient money for food, de-
feeding, 11.1% stopped by 1 month postpartum
livery by cesarean section, smoking during or af-
(Table 1). The cohort was racially diverse, com-
ter pregnancy, and chronic disease during
prising 35.8% Latinas, 14.3% Asian American wo-
pregnancy. Participation in breastfeeding in-
men, 16.6% African American women, and 33.2%
struction had a protective effect on breastfeeding
white women. The majority of women were 2435
initiation (OR 0.40, CI 0.24-0.65).
years of age, had completed high school or some
Before accounting for confounders, African
college, were married or lived with a partner, and
American women were also significantly more
reported having enough money for food. Most of
likely to have a breastfeeding duration of 1
the women were nonsmokers. With regard to de-
month. After adjustment for age, educational at-
pressive symptoms, 25.2% of women reported a
tainment, lack of money for food, marital status,
history of depressive symptoms during preg-
the receipt of breastfeeding instruction, cesarean
nancy and 14.0% in the postpartum period.
section, smoking status, and presence of chronic
In unadjusted analyses, women were more
disease, African American race was significantly
likely not to initiate breastfeeding if they were
associated with breastfeeding duration of 1
African American, less educated, or not married.
month (OR 1.86, CI 1.04-3.33) (Table 2). Rates of
Maternal health factors associated with a lack of
breastfeeding duration at 1 month among Asian
initiation included parity, a history of chronic dis-
and Latina women did not significantly differ
ease during pregnancy, delivery by cesarean sec-
from those of white women. Having depressive
tion, poor or fair postpartum health status, de-
symptoms at a single time point was not a sig-
pressive symptoms, and smoking. Women were
nificant risk factor for having a breastfeeding du-
also more likely to not initiate breastfeeding if
ration of 1 month, but having depressive symp-
they did not receive breastfeeding instruction or
toms at two time points was significantly
if they lacked money for food. In unadjusted
associated with breastfeeding for 1 month (OR
analyses, risk factors for a breastfeeding duration
1.77, CI 1.10-2.86). Other significant risk factors
of 1 month were similar to those for lack of ini-
included having a high school or partial college
tiation except that they also included younger age
education as compared with graduating from col-
and more stress during pregnancy. In addition,
lege and smoking during or after pregnancy.
several factors associated with lack of initiation
Older age, being married or living with a part-
were not significant risk factors in unadjusted
ner, participation in breastfeeding instruction,
analyses of duration 1 month. These included
and a lack of money for food were associated with
parity, lack of money for food, and mode of de-
a longer duration of breastfeeding. In both the ini-
livery.
tiation and the duration model, the interaction be-
tween race/ethnicity and depressive symptoms
Risk factors for not breastfeeding was not significant, suggesting that the role of de-
Before adjusting for confounders, African pressive symptoms on breastfeeding decisions
American women were significantly more likely does not differ between racial/ethnic groups.
to not initiate breastfeeding compared with white
women (odds ratio [OR] 3.52, 95% confidence in-
terval [CI] 1.91-6.48). However, after adjustment DISCUSSION
for age, educational attainment, lack of money for
food, marital status, the receipt of breastfeeding In this cohort of ethnically and socioeconomi-
instruction, cesarean section, smoking status, and cally diverse women, depressive symptoms dur-
presence of chronic disease, there were no sig- ing or prior to pregnancy were not found to be
nificant differences in breastfeeding initiation associated with breastfeeding initiation, even
rates between white women and either African among women who had persistent symptoms.
American, Latina, or Asian women (Table 2). De- Having a breastfeeding duration of 1 month,
pression during or prior to pregnancy was not however, was found to be significantly associated
significantly associated with a lack of breastfeed- with persistent depressive symptoms (occurring
ing initiation, even among women who met CES- at a minimum of two time points, with at least
6303_09_p754-762 8/3/06 1:35 PM Page 758
TABLE 2. FACTORS ASSOCIATED WITH LACK OF BREASTFEEDING INITIATION AND WITH DURATION 1 MONTH
one time point occurring prior to delivery). symptoms and breastfeeding duration, even
African Americans were more likely to have a when symptoms were persistent.33
breastfeeding duration of 1 month. Our study differs from prior work because we
The body of literature addressing depression found that there was a black-white disparity in
during pregnancy and breastfeeding is very lim- breastfeeding duration but no disparity in initia-
ited. One study of low-income, primarily African tion of breastfeeding. We defined breastfeeding
American women examined the relationship be- initiation as having evereven only once
tween depressive symptoms and various infant breastfed; this lenient definition may have mini-
healthcare practices, including breastfeeding.33 mized differences in initiation between racial
This large study by Chung et al. was similar to groups. In addition, the overall rates of breast-
ours in that it examined the persistence of de- feeding initiation and duration were high among
pressive symptoms. Its focus, however, was pri- women in the WISH cohort; in such a population,
marily on postpartum depression, with only one disparities may be less likely. Of note, the high
of the three surveys occurring during pregnancy. rate of breastfeeding initiation in the WISH co-
Unlike our study, the Chung et al. study did not hort is comparable to the overall rate for the San
find a significant association between depressive Francisco County population, which was 85% in
6303_09_p754-762 8/3/06 1:35 PM Page 760
ues to increase into the new millennium. Pediatrics 20. Madden JM, Soumerai SB, Lieu TA, Mandl KD, Zhang
2002;110:1103. F, Ross-Degnan D. Effects on breastfeeding of changes
4. Forste R, Weiss J, Lippincott E. The decision to breast- in maternity length-of-stay policy in a health mainte-
feed in the United States: Does race matter? Pediatrics nance organization. Pediatrics 2003;111:519.
2001;108:291. 21. Heck KE, Schoendorf KC, Chavez GF, Braveman P.
5. Li R, Grummer-Strawn L. Racial and ethnic dispari- Does postpartum length of stay affect breastfeeding du-
ties in breastfeeding among United States infants: Na- ration? A population-based study. Birth 2003;30:153.
tional Health and Nutrition Examination Survey, 22. Celi AC, Rich-Edwards JW, Richardson MK, Klein-
19881994. Birth 2002;29:251. man KP, Gillman MW. Immigration, race/ethnicity,
6. Ertem IO, Votto N, Leventhal JM. The timing and pre- and social and economic factors as predictors of
dictors of the early termination of breastfeeding. Pe- breastfeeding initiation. Arch Pediatr Adolesc Med
diatrics 2001;107:543. 2005;159:255.
7. Saunders-Goldson S, Edwards QT. Factors associated 23. Henderson JJ, Evans SF, Straton JA, Priest SR, Haan
with breastfeeding intentions of African-American R. Impact of postnatal depression on breastfeeding
women at military health care facilities. Mil Med duration. Birth 2003;30:175.
2004;169:111. 24. Falceto OG, Giugliani ER, Fernandes CL. Influence of
8. Brownell K, Hutton L, Hartman J, Dabrow S. Barriers parental mental health on early termination of breast-
to breastfeeding among African American adolescent feeding: A control study. J Am Board Fam Pract
mothers. Clin Pediatr 2002;41:669. 2004;17:173.
9. Noble L, Hand I, Haynes D, McVeigh T, Kim M, Yoon 25. Taveras E, Capra A, Braveman P, Jensvold N, Esco-
JJ. Factors influencing initiation of breastfeeding bar G, Lieu T. Clinician support and psychosocial risk
among urban women. Am J Perinatol 2003;20:477. factors associated with breastfeeding discontinuation.
10. Park YK, Meier ER, Song WO. Characteristics of Pediatrics 2003;112:108.
teenage mothers and predictors of breastfeeding ini- 26. Harlow BL, Cohen LS, Otto MW, Spiegelman D,
tiation in the Michigan WIC Program in 1995. J Hum Cramer DW. Early life menstrual characteristics and
Lact 2003;19:50. pregnancy experiences among women with and with-
11. Kugyelka JG, Rasmussen KM, Frongillo EA. Maternal out major depression: The Harvard study of moods
obesity is negatively associated with breastfeeding and cycles. J Affect Disord 2004;79:167.
success among Hispanic but not black women. J Nutr 27. Taj R, Sikander KS. Effects of maternal depression on
2004;134:1746. breast-feeding. J Pak Med Assoc 2003;53:8.
12. Humphreys AS, Thompson NJ, Miner KR. Intention 28. Galler JR, Harrison RH, Biggs MA, Ramsey F, Forde
to breastfeed in low-income pregnant women: The V. Maternal moods predict breastfeeding in Barbados.
role of social support and previous experience. Birth J Dev Behav Pediatr 1999;20:80.
1998;25:169. 29. Cooper PJ, Murray L, Stein A. Psychosocial factors as-
13. Sharps PW, El-Mohandes AA, Nabil El-Khorazaty M, sociated with the early termination of breast-feeding.
Kiely M, Walker T. Health beliefs and parenting atti- J Psychosom Res 1993;37:171.
tudes influence breastfeeding patterns among low-in- 30. Jacobson SW, Jacobson JL, Frye KE. Incidence and
come African-American women. J Perinatol 2003; correlates of breast-feeding in socioeconomically dis-
23:414. advantaged women. Pediatrics 1991;88:728.
14. Bentley ME, Dee DL, Jensen JL. Breastfeeding among 31. Houghton MD, Graybeal TE. Breast-feeding practices
low income, African-American women: Power, be- of Native American mothers participating in WIC. J
liefs and decision-making. J Nutr 2003;133:305S. Am Diet Assoc 2001;101:245.
15. Kum-Nji P, Mangrem CL, Wells PJ, White P, Herrod 32. Grossman LK, Larsen-Alexander JB, Fitzsimmons
HG. Breastfeeding initiation: Predictors, attitudes, SM, Cordero L. Breastfeeding among low-income,
and practices among blacks and whites in rural Mis- high-risk women. Clin Pediatr (Phila) 1989;28:38.
sissippi. South Med J 1999;92:1183. 33. Chung EK, McCollum KF, Elo IT, Lee HJ, Culhane JF.
16. Kuan LW, Britto M, Decolongon J, Schoettker PJ, Maternal depressive symptoms and infant health
Atherton HD, Kotagal UR. Health system factors con- practices among low-income women. Pediatrics
tributing to breastfeeding success. Pediatrics 1999; 2004;113:523.
104:e28. 34. Haas JS, Fuentes-Afflick E, Stewart AL, et al. Pre-preg-
17. Lu MC, Lange L, Slusser W, Hamilton J, Halfon N. nancy health status and the risk of preterm delivery.
Provider encouragement of breastfeeding: Evidence Arch Pediatr Adolesc Med 2005;159:58.
from a national survey. Obstet Gynecol 2001;97:290. 35. Haas JS, Jackson RA, Fuentes-Afflick E, et al. Changes
18. Schwartz JB, Popkin BM, Tognetti J, Zohoori N. Does in the health status of women during and after preg-
WIC participation improve breastfeeding practices? nancy. J Gen Intern Med 2004;20:45.
Am J Public Health 1995;85:729. 36. Andresen EM, Malmgren JA, Carter WB, Patrick DL.
19. Beal AC, Kuhlthau K, Perrin JM. Breastfeeding advice Screening for depression in well older adults: Evalu-
given to African American and white women by ation of a short form of the CES-D (Center for Epi-
physicians and WIC counselors. Public Health Rep demiologic Studies Depression Scale). Am J Prev Med
2003;118:368. 1994;10:77.
6303_09_p754-762 8/3/06 1:35 PM Page 762
37. Long Foley K, Reed PS, Mutran EJ, DeVellis RF. Mea- Day NL. Long-term effects of chronic depressive
surement adequacy of the CES-D among a sample of symptoms among low-income childrearing mothers.
older African-Americans. Psychiatry Res 2002;109:61. Matern Child Health J 2005;2:1.
38. Cheng ST, Chan AC. The Center for Epidemiologic 44. NICHD Early Child Care Research Network.
Studies Depression Scale in older Chinese: Thresholds Chronicity of maternal depressive symptoms, mater-
for long and short forms. Int J Geriatr Psychiatry nal sensitivity, and child functioning at 36 months.
2005;20:465. Dev Psychol 1999;35:1297.
39. Callahan CM, Wolinsky FD. The effect of gender and 45. Ware JE Jr. SF-36 health survey: Manual and inter-
race on the measurement properties of the CES-D in pretation guide. Boston: The Health Institute, 1993.
older adults. Med Care 1994;32:341. 46. Cohen S, Kamarch T, Mermelstein R. A global mea-
40. Crockett LJ, Randall BA, Shen YL, Russell ST, Driscoll sure of perceived stress. J Health Soc Behav 1983;
AK. Measurement equivalence of the Center for Epi- 24:385.
demiologic Studies Depression Scale for Latino and
Anglo adolescents: A national study. J Counsult Clin
Psychol 2005;73:47. Address reprint requests to:
41. Marcus SM, Flynn HA, Blow FC, Barry KL. Depres- Jennifer S. Haas, M.D.
sive symptoms among pregnant women screened in Division of General Medicine and Primary Care
obstetrics settings. J Womens Health 2003;12:373. Brigham and Womens Hospital
42. Holzman C, Eyster J, Tiedje LB, Roman LA, Seagull 1620 Tremont Street
E, Rahbar MH. A life course perspective on depres- Boston, MA 02120-1613
sive symptoms in mid-pregnancy. Matern Child
Health J 2006;7:1.
43. Seto M, Cornelius MD, Goldschmidt L, Morimoto K, E-mail: jhaas@partners.org