For infant feeding, breastfeeding is the physiologic norm. Infants who are not breastfed are more likely to develop otitis media and nonspecific gastroenteritis. Among mothers, not breastfeeding is associated with an increased risk of type 2 diabetes mellitus and breast and ovarian cancer.1 Based on these data, all major medical organizations recommend 6 months of exclusive breastfeeding followed by continued breastfeeding for at least 1 year.2–4 However, although 75% of women in the United States now initiate breastfeeding, continuation rates at 6 and 12 months drop dramatically (43% and 23%, respectively). These rates fall far short of the Healthy People 2020 goals of 61% at 6 months and 34% at 1 year.5,6
A history of short breastfeeding or not breastfeeding is associated with postpartum depression.1 This condition affects approximately 7–15% of women in the first 3 months after birth and may result in maternal anxiety, depressed mood, poor concentration, and hyperawareness of pain. These psychologic impairments may create difficulties with mother-child bonding,7,8 and they may also affect breastfeeding.9–14 The underlying neuroendocrine mechanism linking breastfeeding difficulties with maternal mood has not been studied. However, neurotransmitters in the brain are thought to contribute to both pain and depression. Thus, stress or pain, which may be associated with breastfeeding, may result in a decrease in serotonin levels with resultant anxiety and depression.15 Early identification of at-risk women is important to both decrease the negative sequelae of postpartum depression and, potentially, to increase breastfeeding success.
No studies have evaluated the role of specific neonatal breastfeeding experiences and postpartum depression.9 We therefore estimated the association between early breastfeeding experiences and postpartum depression in a longitudinal sample of women who initiated breastfeeding. We hypothesized that negative early breastfeeding experiences are associated with postpartum depression.
MATERIALS AND METHODS
This study used data from the Infant Feeding and Practices Study II, a publically available longitudinal data set sponsored by the Centers for Disease Control and Prevention, designed to understand infant feeding patterns, infant health, mother's health, and mother's diet. Approximately 4,902 women in the Infant Feeding and Practices Study II were recruited between May 2005 and June 2007 and followed from the seventh month of pregnancy through the child's first year of life. Participants were identified from a nationally representative consumer panel of more than 500,000 households in the United States.16 Compared with a nationally representative sample, participants in the Infant Feeding and Practices Study II were more likely to be white, middle class, and employed women. Details of the study design have been published elsewhere.16
From participating mothers in the Infant Feeding and Practices Study II cohort, we identified women who initiated breastfeeding after delivery. To participate in the Infant Feeding and Practices Study II, women had to be healthy and had given birth to either a full-term or nearly full-term healthy neonate weighing at least 5 pounds at birth. The parent study collected information on infant feeding and health, breastfeeding cessation, infant formula, sleeping arrangements, child care, employment, and health over the child's first year of life. With the exception of a telephone interview that was conducted at the time of the neonate's birth, all data were collected by mailed self-report questionnaires. Questionnaires were sent to participating mothers approximately 3 weeks postpartum, monthly between 2 months and 7 months postpartum, and then every 7 weeks between 7 and 12 months postpartum.
The original cohort included 4,902 healthy pregnant women who agreed to participate in the study. Three thousand four hundred fifty-two women delivered a singleton full term or nearly full-term live birth of at least 5 pounds and met inclusion criteria for the study. Among these women, 87.6% (n=3,033) completed the 2-month postnatal questionnaire. Maternal mood was assessed on the 2-month questionnaire. Of the 3,033 women who completed the neonatal questionnaire, 2,552 qualified and responded. Women in this sample had a higher level of education, were older, were more likely to be employed, were more likely to be white, and were less likely to smoke compared with a nationally representative sample from the National Survey of Family Growth (1998–2000). Our cohort included 2,586 women who initiated breastfeeding and completed the neonatal questionnaire.
Early breastfeeding experiences were obtained from a neonatal questionnaire completed by the mother approximately 3 weeks after birth. We assessed several types of early breastfeeding experiences among women who initiated breastfeeding after delivery, including the level of pain involved with breastfeeding, the mother's overall feelings about breastfeeding, the amount of breastfeeding support received, and the amount of time it took for a mother's breast milk to come in after delivery. Women were asked how they felt about breastfeeding during the first week after birth on a 5-point Likert scale ranging from “disliked very much” to “liked very much.” This variable was coded as a dichotomous variable with a Likert score of 1–3 indicating “disliked breastfeeding” and a Likert score of 4–5 indicating “liked breastfeeding.” Women also reported their level of pain with breastfeeding the first day, the first week, and the second week after delivery on a 10-point Likert scale. Pain with breastfeeding was modeled as a four-level categorical variable: no pain, mild pain (Likert level 1–2), moderate pain (Likert level 3–4), and severe pain (Likert level 5–10). This variable was not coded using quartiles because the distribution of the variable was strongly right-skewed. We also assessed whether women received help or advice with breastfeeding while in the hospital or information about breastfeeding support groups and services before a mother was discharged. Information on time until a mother's milk came in was also ascertained from the neonatal questionnaire.
The 2-month Infant Feeding and Practices Study II questionnaire included the Edinburgh Postnatal Depression Scale. This screening tool is a 10-item instrument than can be administered in 5 minutes and is widely used in the mental health profession. The instrument was validated by Cox and colleagues in 1987.17 A cut point of 9–10 was recommended as a threshold for mild depression, and a score of 12–13 was recommended as a cut point for major depression.18 The sensitivity and specificity of the instrument as a screening tool for major depression has been reported as 75% and 84%, respectively, for a cutoff score of 13.18 For this analysis, we categorized women with an Edinburgh Postnatal Depression Scale score of 13 or greater as having major depression. Women with a score below 13 were considered not to be depressed.
Baseline covariate information was collected from the demographic questionnaire that was administered to the national opinion panel from which the sample was drawn. If the mother was not the panelist in the national opinion panel survey, a separate questionnaire was sent to the mother to collect the demographic information. The following covariates were considered in this analysis: household size modeled as a categorical variable (categories), marital status (categories), race and ethnicity (categories), maternal age modeled as a categorical variable (categories), parity (categories), education (categories), employment status (categories), maternal occupation (categories), and postnatal WIC participation. All sociodemographic covariates were ascertained from the national opinion panel data with the exception of educational level and marital status, which were ascertained from the prenatal questionnaire.
We described the frequency distribution of all covariates by depression status. We modeled the association between early breastfeeding experiences and postpartum depression as a complete case analysis using logistic regression in SAS 9.1. Separate logistic models were generated for each type of breastfeeding experience including pain with breastfeeding, time until a woman's milk came in, breastfeeding support, and general feelings with breastfeeding. We generated odds ratios (ORs) and 95% confidence intervals (CIs) for both crude and multivariable adjusted models. We evaluated the assumption of linearity imposed by the logit model by plotting the log odds of postpartum depression by levels of our ordinal variables. All ordinal variables displayed a linear relationship. The following covariates were considered to affect both a mother's early breastfeeding experiences and the risk of developing postpartum depression: maternal age, parity, education, ethnicity, and postnatal WIC participation. These covariates were included as confounders in the multivariable logistic regression models.
We also used multiple imputation to evaluate the sensitivity of the estimates obtained from the complete case analysis. Data were imputed in SAS 9.1 using proc IVeware across five imputations. Effect estimates across the five imputations were averaged and standard errors were calculated accounting for both within-imputation and between-imputation variability.19,20
To determine whether parity, feelings about breastfeeding, or breast pain modified the effect of breastfeeding support on postpartum depression, we tested for interactions using a cross-product term.
Our cohort included 2,586 women who completed the 2-month Edinburgh Postnatal Depression Scale and reported ever trying to breastfeed their infant in the hospital or birth center or after the mother and child returned home. The mean age of women was 28.8 years. Women in the cohort were predominantly white (83.5%) and married (80.2%) (Table 1). Two months after delivery, 74.3% of women with available breastfeeding data (n=2,185) continued to breastfeed.
Based on a positive Edinburgh Postnatal Depression Scale screen at 2 months, two hundred twenty-three (8.62%) mothers who initiated breastfeeding met criteria for postpartum depression. Women with postpartum depression were more likely to participate in postnatal WIC (postpartum depression: 44.4% compared with no postpartum depression: 35.3%) and were twice as likely to have less than a high school education (postpartum depression: 4.5% compared with no postpartum depression: 2.0%) (Table 1). The mean age of the mother when her child was born was similar between depressed and nondepressed women (postpartum depression: 28.0 years compared with no postpartum depression: 29.1 years).
At 2 months postpartum, mothers with postpartum depression were less likely to still be breastfeeding (68.6%) compared with mothers without (74.9%) depressive symptoms (P=.04).
In multivariable logistic regression analyses, we found that negative early breastfeeding experiences were associated with depressive symptoms at 2 months postpartum. Severe breastfeeding pain in the neonatal period was associated with a twofold increase in odds of postpartum depression. Specifically the odds of postpartum depression at 2 months among women who reported severe pain with breastfeeding on day 1 was 1.96 (95% CI 1.17–3.29) times that of women who reported no pain with breastfeeding on day 1. We similarly found an increased odds (OR 2.24, 95% CI 1.18–4.26) of postpartum depression among women with severe breast pain 2 weeks after delivery (Table 2). Thus, on average, 13.87 and 14.57 additional postpartum women would have to experience severe breastfeeding pain on day 1 and week 1, respectively, for one additional mother to experience postpartum depression at two months.21 Women who experienced moderate pain with breastfeeding were 22–85% more likely to experience postpartum depression 2 months after delivery; however, these effect estimates were not statistically significant at conventional levels (P=.05). Effect estimates approximating the crude association between breastfeeding pain and the odds of postpartum depression 2 weeks after delivery were similar to the adjusted estimates.
A mother's feelings about breastfeeding in the first week after birth were also associated with increased odds of postpartum depression at 2 months. Women who disliked breastfeeding were 1.42 (95% CI 1.04–1.93) times as likely to have postpartum depression 2 months after delivery compared with women who reported “liking breastfeeding.” On average, 32.04 mothers would have to dislike breastfeeding for one additional postpartum woman to experience postpartum depression. There was no association between timing of when a mother's milk came in and odds of postpartum depression. Receiving information regarding breastfeeding support groups at hospital discharge showed a small (OR 0.87, 95% CI 0.63–1.20), although a nonsignificant, protective effect (Table 3). Overall, help with breastfeeding in the hospital was not associated with depressive symptomatology.
The strength of the association between early breastfeeding experience and postpartum depression was similar in magnitude when comparing the complete case with the multiple imputation analyses. Both severe pain with breastfeeding and “disliking breastfeeding” continued to be statistically significant in the imputed analysis.
Breastfeeding support, specifically help with breastfeeding in the hospital, did not appear to have a beneficial effect among mothers regardless of parity. Moreover, the associations among feelings about breastfeeding, receipt of support group information, and postpartum depression at 2 months did not differ significantly by parity (all cross-product P>.15). However, there was a small but statistically significant protective effect with receiving help with breastfeeding in the hospital among women with moderate to severe breast pain during nursing on day 1 (moderate: OR 0.22, 95% CI 0.05–0.94, P=.04; severe: OR 0.17, 95% CI 0.04–0.75, P=.02). However, the 95% CIs were quite wide. Within the subset of women who disliked breastfeeding in the neonatal period (n=1,264), breastfeeding help in the hospital was not significantly associated with postpartum depression at 2 months (OR 0.83, 95% CI 0.49–1.40; P=.49).
In a large sample of U.S. women, we found that negative early breastfeeding experiences were associated with increased odds of depressive symptoms at 2 months postpartum. These results suggest that early breastfeeding difficulties may indicate an increased risk for postpartum depression. Targeting these women for early screening of postpartum depression may identify women at risk and allow clinicians to reduce the morbidity associated with both postpartum depression and curtailed breastfeeding.
Our study confirms and extends earlier work evaluating the association between curtailed breastfeeding and postpartum depression. Specifically, limited breastfeeding duration, low breastfeeding self-efficacy, and concerns over breastfeeding were associated with depressive symptomatology.9–12,14 In our large sample, women who were depressed were also less likely to continue breastfeeding at 2 months compared with those women without depressive symptoms. Moreover, our results support previous work that concerns about breastfeeding and poor self-efficacy were associated with depressive symptomatology. More specifically, compared with women with no early neonatal signs of breastfeeding difficulty, we found that women who had negative feelings about breastfeeding and reported severe pain while nursing soon after birth were more likely to experience postpartum depression at 2 months. The receipt of breastfeeding help and information about support groups did not, in general, offer a protective effect. However, we found that help with breastfeeding in the hospital did offer a small beneficial effect for those mothers with moderate to severe pain with nursing.
The mechanisms underlying our results remain to be elucidated. However, previous work shows that feelings of pain and depression share the same descending pathways in the central nervous system.15 Neurochemical imbalances in key neurotransmitters such as serotonin may contribute to feelings of pain as well as depression. Specific to pregnancy, these findings are consistent with earlier work linking childbirth-associated pain and catastrophizing with subsequent depressive symptoms.22–24 The association between pain and depression is complex and may be mediated, in part, by differences in central nociception pathways. Such differences have been identified in functional magnetic resonance imaging studies of patients with fibromyalgia.25 If severe early breastfeeding-associated pain reflects pain catastrophizing, then aberrations in central nociception pathways may lead to both curtailed breastfeeding and perinatal depression.
Our study has several strengths. This is the first study to measure the association between a woman's early breastfeeding experiences and prevalence of postpartum depression 2 months after delivery. We used data from a prospective cohort in which information on infant feeding patterns was collected at frequent intervals throughout the child's first year of life. This most likely limited poor recall of feeding practices that may occur if women are asked detailed information about infant feeding later in her child's life. The Infant Feeding and Practices Study II collected extremely detailed information on breastfeeding from the prenatal period through 12 months postpartum, which is unique to this cohort. In the analysis, we considered multiple covariates that may confound the association between early breastfeeding experience and postpartum depression based on the published literature and included these covariates in our multivariable logistic regression models.
However, these data were also subject to several limitations. Although the data are from a longitudinal cohort, researchers did not obtain the mother's baseline depression status. Thus, we were not able to assess the temporality of the relationship. For example, if a mother was depressed at the time of birth, this may have led to breastfeeding difficulties. Regardless of the temporality of the events, women who exhibit signs of breastfeeding difficulties should be screened and offered breastfeeding support. According to the published literature, social and peer support may have not only positive effects on breastfeeding outcomes, but also postpartum depression26,27
The potential for residual confounding is a second limitation of these data. We included relevant covariates in our regression models that may confound the relationship between breastfeeding experiences and postpartum depression. However, there may be residual confounding of the effect estimates as a result of unmeasured confounding.
Finally, to estimate the association between a woman's early breastfeeding experiences and odds of postpartum depression, we considered only those observations with complete data. Women who completed the Edinburgh Postnatal Depression Scale were less likely to participate in postnatal WIC, to have a larger family size, to have less than a high school education, to be single, and to be African American. However, we examined the sensitivity of our findings to missing data by using multiple imputation. Effect estimates were of similar magnitude, and both breast pain and feelings about breastfeeding remained statistically significant predictors of postpartum depression at 2 months.
In a large sample of U.S. women, we found that negative early breastfeeding experiences were associated with depressed mood at 2 months postpartum. Our study supports the findings of Dennis and McQueen in 2007 who found that women with postpartum depression were more likely to report problems with breastfeeding and to discontinue breastfeeding 4–8 weeks postpartum.28 Extreme breast pain during breastfeeding as well as a general dislike of breastfeeding may identify women who are more likely to experience postpartum depression. Our results suggest that women with breastfeeding difficulties should be screened for postpartum depression, and women with depressive symptoms should be offered breastfeeding support. Screening and treatment of women with early breastfeeding difficulties may reduce the severity of postpartum depression and enable women to meet their breastfeeding goals, thereby improving health outcomes across two generations.
1. Ip S, Chung M, Raman G, Trikalinos TA, Lau J. A summary of the Agency for Healthcare Research and Quality's evidence report on breastfeeding in developed countries. Breastfeed Med 2009;4(suppl 1):S17–30.
2. American Academy of Pediatrics. Breastfeeding and the use of human milk. Pediatrics 2005;115:496–506.
3. American College of Obstetricians and Gynecologists (ACOG). Breastfeeding: Maternal and Infant Aspects. Special report from ACOG. ACOG Clin Rev 2007;12:1–16S.
4. American Academy of Family Physicians. Breastfeeding (position paper); 2001. Available at: http://www.aafp.org/online/en/home/policy/policies/b/breastfeedingpositionpaper.html
. Retrieved May 10, 2011.
5. Breastfeeding among US children born 1999–2007. CDC national immunization survey. Atlanta (GA): Centers for Disease Control and Prevention. Available at: http://www.cdc.gov/breastfeeding/data/NIS_data/index.htm
. Retrieved May 10, 2011.
6. Healthy People 2020. Washington, DC: U.S. Department of Health and Human Services. Available at: http://www.healthypeople.gov/2020/default.aspx
. Retrieved May 10, 2011.
7. Gaynes BN, Gavin N, Meltzer-Brody S, Lohr KN, Swinson T, Gartlehner G, et al. Perinatal depression: prevalence, screening accuracy, and screening outcomes. Evid Rep Technol Assess (Summ) 2005:1–8.
8. Meltzer-Brody S, Leserman J. Psychiatric comorbidity in women with chronic pelvic pain. CNS Spectr 2011;Feb 1:pii: Meltzer-Brody [Epub ahead of print].
9. Dennis CL, McQueen K. The relationship between infant-feeding outcomes and postpartum depression: a qualitative systematic review. Pediatrics 2009;123:e736–51.
10. Taveras EM, Capra AM, Braveman PA, Jensvold NG, Escobar GJ, Lieu TA. Clinician support and psychosocial risk factors associated with breastfeeding discontinuation. Pediatrics 2003;112:108–15.
11. Papinczak TA, Turner CT. An analysis of personal and social factors influencing initiation and duration of breastfeeding in a large Queensland maternity hospital. Breastfeed Rev 2000;8:25–33.
12. Chaudron LH, Klein MH, Remington P, Palta M, Allen C, Essex MJ. Predictors, prodromes and incidence of postpartum depression. J Psychosom Obstet Gynaecol 2001;22:103–12.
13. Dennis CL. The breastfeeding self-efficacy scale: psychometric assessment of the short form. J Obstet Gynecol Neonatal Nurs 2003;32:734–44.
14. Dai X, Dennis CL. Translation and validation of the Breastfeeding Self-Efficacy Scale into Chinese. J Midwifery Womens Health 2003;48:350–6.
15. Bair MJ, Robinson RL, Katon W, Kroenke K. Depression and pain comorbidity: a literature review. Arch Intern Med 2003;163:2433–45.
16. Fein SB, Labiner-Wolfe J, Shealy KR, Li R, Chen J, Grummer-Strawn LM. Infant Feeding Practices Study II: study methods. Pediatrics 2008;122:S28–35.
17. Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression. Development of the 10-item Edinburgh Postnatal Depression Scale. Br J Psychiatry 1987;150:782–6.
18. Cox JL, Chapman G, Murray D, Jones P. Validation of the Edinburgh Postnatal Depression Scale (EPDS) in non-postnatal women. J Affect Disord 1996;39:185–9.
19. Klebanoff MA, Cole SR. Use of multiple imputation in the epidemiologic literature. Am J Epidemiol 2008;168:355–7.
20. He Y. Missing data analysis using multiple imputation: getting to the heart of the matter. Circ Cardiovasc Qual Outcomes 2010;3:98–105.
21. Bender R, Blettner M. Calculating the ‘number needed to be exposed’ with adjustment for confounding variables in epidemiological studies. J Clin Epidemiol 2002;55:525–30.
22. Ferber SG, Granot M, Zimmer EZ. Catastrophizing labor pain compromises later maternity adjustments. Am J Obstet Gynecol 2005;192:826–31.
23. Boudou M, Teissedre F, Walburg V, Chabrol H. Association between the intensity of childbirth pain and the intensity of postpartum blues [in French]. Encephale 2007;33:805–10.
24. Flink IK, Mroczek MZ, Sullivan MJ, Linton SJ. Pain in childbirth and postpartum recovery: the role of catastrophizing. Eur J Pain 2009;13:312–6.
25. Gracely RH, Geisser ME, Giesecke T, Grant MA, Petzke F, Williams DA, et al. Pain catastrophizing and neural responses to pain among persons with fibromyalgia. Brain 2004;127:835–43.
26. Cohen S, Gottlieb B, Underwood L. Social relationships and health. In: Cohen S, Underwood L, Gottlieb B, editors. Social Support Measurement and Intervention: A Guide for Health and Social Scientists. Toronto (Ontario): Oxford University Press; 2000.
27. Dennis CL, Hodnett E, Kenton L, Weston J, Zupancic J, Stewart DE, et al. Effect of peer support on prevention of postnatal depression among high risk women: multisite randomised controlled trial. BMJ 2009;338:a3064.
28. Dennis CL, McQueen K. Does maternal postpartum depressive symptomatology influence infant feeding outcomes? Acta Paediatr 2007;96:590–4.