Objective. Breastfeeding rates fall short of goals set in Healthy People 2010 and other national recommendations. The current, national breastfeeding continuation rate of 29% at 6 months lags behind the Healthy People 2010 goal of 50%. The objective of this study was to evaluate associations between breastfeeding discontinuation at 2 and 12 weeks postpartum and clinician support, maternal physical and mental health status, workplace issues, and other factors amenable to intervention.
Methods. A prospective cohort study was conducted of low-risk mothers and infants who were in a health maintenance organization and enrolled in a randomized, controlled trial of home visits. Mothers were interviewed in person at 1 to 2 days postpartum and by telephone at 2 and 12 weeks. Logistic regression modeling was performed to assess the independent effects of the predictors of interest, adjusting for sociodemographic and other confounding variables.
Results. Of the 1163 mother-newborn pairs in the cohort, 1007 (87%) initiated breastfeeding, 872 (75%) were breastfeeding at the 2-week interview, and 646 (55%) were breastfeeding at the 12-week interview. In the final multivariate models, breastfeeding discontinuation at 2 weeks was associated with lack of confidence in ability to breastfeed at the 1- to 2-day interview (odds ratio [OR]: 2.8; 95% confidence interval [CI]: 1.02–7.6), early breastfeeding problems (OR: 1.5; 95% CI: 1.1–1.97), Asian race/ethnicity (OR: 2.6; 95% CI: 1.1–5.7), and lower maternal education (OR: 1.5; 95% CI: 1.2–1.9). Mothers were much less likely to discontinue breastfeeding at 12 weeks postpartum if they reported (during the 12-week interview) having received encouragement from their clinician to breastfeed (OR: 0.6; 95% CI: 0.4–0.8). Breastfeeding discontinuation at 12 weeks was also associated with demographic factors and maternal depressive symptoms (OR: 1.18; 95% CI: 1.01–1.37) and returning to work or school by 12 weeks postpartum (OR: 2.4; 95% CI: 1.8–3.3).
Conclusions. Our results indicate that support from clinicians and maternal depressive symptoms are associated with breastfeeding duration. Attention to these issues may help to promote breastfeeding continuation among mothers who initiate. Policies to enhance scheduling flexibility and privacy for breastfeeding mothers at work or school may also be important, given the elevated risk of discontinuation associated with return to work or school.
The promotion and support of breastfeeding has emerged as a national public health priority in recent years. The American Academy of Pediatrics has recommended exclusive breastfeeding for 6 months, continuing to 1 year or beyond.1 In its Healthy People 2010 recommendations, the US Department of Health and Human Services set goals of 75% of mothers to breastfeed exclusively in the early postpartum period, 50% to continue to 5 to 6 months, and 25% to continue to 1 year.2
National statistics show large gaps between the reality and the goals, particularly for breastfeeding continuation. In 1998, 64% of mothers initiated breastfeeding in-hospital, but rates of any breastfeeding were only 29% by 6 months and 16% by 1 year.3 Studies have found elevated risk of breastfeeding discontinuation among mothers who are older, have lower educational attainment, and return to work full-time.4–7 Lack of support from family and friends, insufficient prenatal breastfeeding education, and hospital discharge packs that contain infant formula have also been associated with breastfeeding discontinuation.8,9
Among the many studies of breastfeeding continuation, few have evaluated factors amenable to intervention by clinicians in office settings.10 During the first 12 weeks postpartum, mothers must adjust to many new physical, psychological, and social demands. They and their newborns also regularly see health care providers for routine preventive visits, which are an opportunity for support of breastfeeding.
This study was designed to address gaps in existing knowledge about clinician support and maternal mental health status as influences on breastfeeding continuation. Our aims were 1) to describe reasons for breastfeeding discontinuation during the first 12 postpartum weeks and 2) to evaluate associations between breastfeeding discontinuation and modifiable factors, including maternal physical and mental health status, health services, and workplace issues. We tested the specific hypotheses that mothers whose clinicians supported breastfeeding and mothers who had fewer depressive symptoms would be more likely to continue breastfeeding through 12 weeks.
This prospective cohort study included low-risk mothers and newborns in the Kaiser Permanente Medical Care Program (KPMCP), Northern California Region, a group model managed care organization. Data were collected via 1) a face-to-face interview during the postpartum hospitalization, 2) a telephone interview at 2 weeks, and 3) a telephone interview at 12 weeks. Bivariate and multivariate analyses were conducted to identify predictors of breastfeeding discontinuation at 2 weeks and at 12 weeks.
The cohort for the current study consisted of low-risk mothers and newborns at KPMCP medical center in Sacramento, California, who had enrolled in a randomized controlled trial of home visits versus clinic-based follow-up during July 1996 through September 1997. The randomized controlled trial did not find a difference in mode of follow-up and breastfeeding continuation rates. Only those mother-newborn pairs who were medically and socially low risk were eligible. Infants who weighed <2500 or >4600 g at birth, had stayed in the intensive care nursery, or had a medical problem that necessitated follow-up by a pediatrician or nurse practitioner were excluded. Mothers who were 14 years old or younger, had a positive toxicology screen for drugs of abuse after admission to labor and delivery, or spoke a language other than English or Spanish were excluded. Finally, mothers and newborns whose anticipated length of stay was >48 hours, usually as a result of cesarean delivery, were excluded. We did not collect information regarding breastfeeding duration on ineligible mother-newborn pairs. Additional details are provided in the report of the randomized trial.11
Research nurses used chart review and the enrollment interview on the postpartum floor to collect baseline data on clinical and demographic variables, as well as on maternal experiences and perceptions about prenatal care and breastfeeding. At 2 weeks and 12 weeks postpartum, a research interviewer contacted each mother by telephone to conduct a 15-minute interview about breastfeeding, other outcomes, and satisfaction. At both 2 weeks and 12 weeks, discontinuation of exclusive breastfeeding was defined as either giving no breast milk or still breastfeeding but giving >12 oz of formula per day. Using previous definitions of breastfeeding, mothers who give >12 oz of formula per day would be providing less than half of the average infant’s caloric intake via breastfeeding and would be considered “low partial or token” breastfeeders.12,13 In the analysis, such mothers were grouped in the “breastfeeding discontinuation” category.
Race/ethnicity was self-reported with the mother asked to name all racial or ethnic identifications that applied; for analysis, respondents were categorized as white (non-Hispanic), black, Hispanic, Asian, multicultural white, or other (categorization algorithm available on request). The Center for Epidemiologic Studies Depression Scale (CES-D), a widely used 20-item instrument that has been validated in English and Spanish, was used to evaluate maternal depressive symptoms using quartiles during the 2-week telephone interview. We conducted a second telephone interview that focused on breastfeeding at 12 weeks postpartum. All data from 2 weeks to 12 weeks were collected retrospectively during the 12-week interview. At this time, mothers were asked whether they received encouragement to breastfeed from a doctor, nurse, or breastfeeding consultant at KPMCP. Response categories were yes, no, or don’t know. Mothers were not asked when they received encouragement. This project was approved by the KPMCP Institutional Review Board for the Protection of Human Subjects.
The primary outcomes of interest were breastfeeding discontinuation at 2 and 12 weeks postpartum as previously defined. χ2 analyses for categorical variables and the t test for continuous variables were used to identify predictors associated with breastfeeding discontinuation at 2 and 12 weeks. Predictors evaluated were broadly categorized as sociodemographic, health services, and health status variables. Demographic data examined included maternal age, race/ethnicity, total household income, highest level of education attained, marital status, and parity. Social factors included returning to work or school, the perceived importance of breastfeeding, reported confidence in the ability to breastfeed, requiring more help with their infant or with household chores, and maternal rating of the support received by the infant’s father and by family and friends to breastfeed. Health services included the amount and quality of the breastfeeding advice that mothers received by health care providers, encouragement to breastfeed by health care providers, and participation in prenatal breastfeeding classes. Maternal health status variables included depressive symptoms noted in the CES-D scale, having visited an emergency department or clinic for their own health problem, and perceived difficulty of their labor and delivery experience.
Logistic regression was performed to assess the independent effects of these predictors on breastfeeding discontinuation. Predictor variables associated with the outcome at P = .15 or less in bivariate analyses were eligible for entry to preliminary multivariate models. Each preliminary model included all eligible sociodemographic predictors and one of the other predictors of interest in a forced-entry logistic regression. Predictors of interest that were associated with breastfeeding discontinuation at P = .05 or less in preliminary modeling were then entered into a final logistic regression model. In the final model for breastfeeding discontinuation at 12 weeks, we eliminated 2 variables—perceived difficulty of labor and confidence in ability to feed the infant—because they were highly correlated with maternal depressive symptoms. The fit of the final logistic models was assessed as adequate on the basis of Hosmer-Lemeshow tests and c-statistics. The Hosmer-Lemeshow P values for the 2- and 12-week models (.64 and .73, respectively) were substantially >.05. The c-statistics for the 2- and 12-week models were 0.716 and 0.725, respectively. All data analyses were performed in SAS, version 8.0 (SAS Institute, Cary, NC).
Among the 1163 enrolled mother-newborn pairs, 1007 (87%) initiated breastfeeding and were included in additional analyses. A total of 872 (75%) mothers were breastfeeding at the 2-week interview; of these women, 859 (99%) completed the 12-week interview. The study group (Table 1) was 62% white, 12% Hispanic, 6% black, 5% Asian, and 11% multicultural white, with a mean age of 28 ± 6 years. Most mothers were married (89%) and had at least some college education (68%). Sixty-five percent of the mothers reported household incomes of >$30 000 per year; however, a substantial group (14%) reported household incomes of <$20 000 per year. Forty-one percent of mothers were nulliparous, and the majority (91%) had initiated prenatal care in the first trimester.
Rates of Breastfeeding Discontinuation
Among the original 1163 mothers in the cohort, 1007 (87%) were breastfeeding at 1 to 2 days, 872 (75%) were breastfeeding at the 2-week interview, and 646 (55%) were breastfeeding at the 12-week interview. In other words, of the 1007 mothers who initiated breastfeeding, 135 (13%) discontinued by 2 weeks postpartum. Of the 859 mothers who completed the 12-week interview, another 213 mothers had discontinued breastfeeding by 12 weeks postpartum (Fig 1). The sharpest drop in breastfeeding occurred during the first 4 postpartum weeks.
Sociodemographic Characteristics and Breastfeeding Discontinuation at 2 and 12 Weeks
Breastfeeding discontinuation at both 2 and 12 weeks was higher among women who were younger, had lower educational attainment, had lower household income, or were single parents (Table 1). First-time mothers were more likely than multiparas to discontinue breastfeeding at both time points. No significant differences were observed in discontinuation among racial/ethnic groups in bivariate analysis.
Clinician Support and Maternal Mental Health
Mothers who reported (during the 12-week interview) having received encouragement to breastfeed from a doctor, nurse, or breastfeeding consultant were less likely to discontinue breastfeeding at 12 weeks (P = .015; Table 2).
A total of 239 mothers (24%) had depressive symptoms as measured by a score of 16 or higher on the 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 interview, 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 breastfeeding, 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 discontinuation (P < .0001) at 12 weeks postpartum. Similarly, 236 (50%) women who returned to work or school reported problems trying to continue breastfeeding, and those who reported problems were significantly more likely to discontinue by 12 weeks (P < .0001).
In the final multivariate models, mothers who experienced breastfeeding problems at 2 to 3 days (odds ratio [OR]: 1.5; 95% confidence interval [CI]: 1.1–1.97) or reported a lack of confidence in their 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 postpartum. Those who rated breastfeeding as important at the 1- to 2-day interview (OR: 0.29; 95% CI: 0.15–0.58) were less likely to have discontinued breastfeeding at 2 weeks. Other independent predictors of breastfeeding discontinuation at 2 weeks included lower maternal education (OR: 1.5; 95% CI: 1.2–1.9) and being of Asian race/ethnicity (OR: 2.6; 95% CI: 1.1–5.7). Women who reported a lack of support from the father of their infant were more likely to discontinue breastfeeding at 2 weeks, but this did not reach statistical significance (OR: 1.7; 95% CI: 0.99–2.91).
In multivariate models to evaluate predictors of breastfeeding at 12 weeks, mothers were much less likely to discontinue breastfeeding by 12 weeks if they reported (at 12 weeks) having received encouragement from their clinician to breastfeed (OR: 0.56; 95% CI: 0.37–0.84) or rated breastfeeding as important at the 1- to 2-day interview (OR: 0.24; 95% CI: 0.11–0.53). Breastfeeding discontinuation was more frequent among mothers with more depressive symptoms as measured by the CES-D scale (OR: 1.18 per quartile; 95% CI: 1.01–1.37). Mothers who returned to work or school (OR: 2.4; 95% CI: 1.75–3.3) or experienced problems breastfeeding or pumping at work or school (OR: 3.2; 95% CI: 1.9–5.39) also had increased odds of discontinuation at 12 weeks (Table 3).
Reported Breastfeeding Problems
At the 1- to 2-day interview, approximately 21% of mothers reported experiencing breastfeeding problems that they described as either somewhat serious or serious. These mothers were more likely to discontinue breastfeeding at 2 weeks and 12 weeks (Table 2). Reasons for breastfeeding discontinuation varied by week (Table 4). In the first postpartum week, mothers reported breastfeeding discontinuation as a result of problems with their infant sucking or latching on (23%) and breast pain and soreness (14%). Another frequently reported problem was the belief that the infant was still hungry or that they were not producing enough breast milk (27% at 0–1 week, 38% at 4–6 weeks).
Returning to work or school progressively became the greatest reason for breastfeeding discontinuation by 10 to 12 weeks. The proportion of mothers who cited return to work as the main reason for breastfeeding discontinuation increased from 14% at 2 to 3 weeks to 58% at 10 to 12 weeks postpartum. The majority of problems reported among women who returned to work were restricted schedules and breaks (51%) and insufficient privacy (20%). Among women who returned to school, the absence of on-site child care was also reported as a barrier to breastfeeding continuation (23%).
Health Services Used in Support of Breastfeeding
A total of 165 women (16%) participated in breastfeeding classes, predominantly in the prenatal period, and 230 (23%) viewed breastfeeding instructional videos. Many women reported receiving both written (64%) and verbal (63%) breastfeeding information during their postpartum hospital stay. Thirty-five percent of women received help from a lactation consultant, 57% received information regarding breast pumps, and 65% reported receiving individualized support and breastfeeding instruction from a health care provider. This type of individualized instruction was chosen by 49% of respondents as the most helpful service actually received in support of breastfeeding and named by 63% of respondents as the most helpful service that they could have been offered by their health care providers.
Our findings suggest that clinician support and maternal mental health status deserve attention as modifiable factors in promoting breastfeeding continuation. Women who reported that their health care providers encouraged them to breastfeed were approximately half as likely to discontinue breastfeeding by 12 weeks postpartum as those who did not. Conversely, the odds of breastfeeding discontinuation were substantially elevated for each increase in quartile in the CES-D, a measure of depressive symptoms. The observational design of the current study does not permit firm causal inferences between breastfeeding duration and clinician support or maternal depressive symptoms but does suggest that interventions in these areas deserve additional evaluation.
Our results are in accordance with previous studies that suggest that clinicians and other health care providers may have an influential role in breastfeeding initiation and continuation. An analysis of national survey data collected from parents at up to 3 years after childbirth found that provider support was associated with breastfeeding initiation.14 A prospective, observational study suggested that health system support of breastfeeding during the postpartum hospitalization and early postdischarge period was associated with successful breastfeeding.10
Our study is distinct from previous research in that we evaluated the association between clinician support and breastfeeding continuation at 12 weeks postpartum. In addition, we were able to control for many other prospectively collected demographic and psychosocial factors, including maternal confidence in ability to breastfeed and belief in the importance of breastfeeding.
Behavior-oriented interventions to support breastfeeding have been found effective in several randomized, controlled trials.15–17 However, these interventions have usually been conducted by health educators or lactation consultants and have almost always involved a dedicated amount of time (for example, a 50- to 80-minute group counseling session) outside usual prenatal or pediatric preventive visits. We are unaware of any controlled trials that have evaluated the effectiveness of clinician support for breastfeeding delivered during usual preventive visits. Several studies have found that pediatricians, obstetricians, and family practitioners lack knowledge and training on breastfeeding topics.18,19 Studies of other topics, including smoking cessation,20,21 suggest that physician-delivered counseling can be effective in promoting preventive health behaviors. Our study suggests that pediatrician skills to support breastfeeding deserve attention in breastfeeding promotion efforts22 and that randomized, controlled trials of interventions to enhance clinician support of breastfeeding during routine preventive visits are warranted.
Maternal Mental Health
Postpartum depression is a serious, common, and treatable condition that often goes unrecognized.23 Although previous studies have shown a relationship between maternal depressive symptoms and breastfeeding cessation,24 very few have examined the adequacy of health care resources in support of postpartum mothers.25 Our study shows that women with more maternal depressive symptoms as measured by the CES-D scale had greater odds of breastfeeding discontinuation by 12 weeks. Thus, clinicians’ recognition of postpartum depressive symptoms may contribute to breastfeeding continuation as well as prevent other adverse effects on mothers and infants.26
Other Factors That Influence Breastfeeding Discontinuation
Two important independent predictors of early breastfeeding discontinuation in our study—breastfeeding problems at 2 to 3 days and lack of confidence in the ability to breastfeed—have been previously found to be associated with early discontinuation of breastfeeding.9 Early (2–3 days) breastfeeding problems reported by mothers in our study were perceptions of insufficient milk supply, breast soreness, and problems with their infants latching on. Current guidelines on maternal and neonatal follow-up tend to focus on catastrophic, rare events rather than on common problems such as maternal perception of insufficient milk supply.27–29 Development of evidence-based recommendations and anticipatory guidance for common, early breastfeeding problems may improve rates of breastfeeding continuation.
This study confirms previous findings that returning to work is associated with lower rates of breastfeeding initiation and continuation.7,30 The most common workplace problems reported by women in our study were restricted schedules and breaks and insufficient privacy. Given the large number of women who return to work or school in the postpartum period, workplace and school policies to enhance scheduling flexibility and privacy for breastfeeding mothers should be encouraged.
Demographic variables such as maternal age, education, income, and race have been studied extensively.8 To our knowledge, our study is one of the first to document higher breastfeeding discontinuation rates among mothers of Asian race/ethnicity.31 Additional research is warranted to determine the reasons for breastfeeding discontinuation among mothers of Asian race/ethnicity.
Our study should be interpreted keeping in mind several limitations. First, it focused on a medically and socially low-risk population of mother-infant pairs in an integrated health maintenance organization (HMO). The mothers in the study had diverse racial/ethnic backgrounds, but nearly two thirds had at least some college education and <10% had less than a high school degree. The HMO in this study primarily serves families with employment-based, comprehensive health insurance coverage; the study population’s median income was higher than that of the general population. Although our population was ethnically diverse, our results may not be generalizable to more socioeconomically disadvantaged populations or to families who receive care in less integrated settings. Our results also may not be generalizable to mothers or newborns with medical complications or mothers who delivered by cesarean section. Additional research involving such populations and settings is needed.
Mothers in the study were asked to recall events that may have taken place 12 weeks or more before the interview. In particular, mothers were asked during the 12-week interview to recall whether they were encouraged to breastfeed by a health care provider. The observed association between clinician support and breastfeeding continuation could have 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. Another possible explanation for the observed association is selection bias: mothers who intended to breastfeed might have chosen clinicians who were more likely to encourage it.
Our analysis focused on the mother-infant pairs that deviated from American Academy of Pediatrics recommendations for exclusive breastfeeding during the first 6 months of life. Our definition of breastfeeding continuation was relatively conservative in that we classified mothers who were giving more than half of an average infant’s caloric intake as formula as having discontinued exclusive breastfeeding. Thus, we classified token and low-partial breastfeeding as having discontinued exclusive breastfeeding.
Given this study’s observational design, it is important to acknowledge the potential role of unmeasured characteristics that might explain the observed associations among clinician support, maternal depressive symptoms, and breastfeeding duration. Given the range of maternal demographic and psychosocial characteristics that we were able to consider, however, the overall pattern of findings indicate that these 2 factors are strongly associated with breastfeeding duration. A recent cluster-randomized trial in Belarus found that an intervention emphasizing healthcare working assistance with breastfeeding increased duration of breastfeeding at 3 and 6 months.32 Controlled trials of counseling interventions by primary care providers delivered during usual preventive visits should be performed in the United States. Furthermore, for evaluating the success of future interventions to promote breastfeeding, more sensitive, clinically detailed measures of breastfeeding problems are needed.
Our study adds to the growing body of literature suggesting that support and encouragement to breastfeed from health care providers is associated with a higher likelihood of breastfeeding continuation. Attention should also be paid to maternal mental health status not only for the health benefits to mothers but also as a potentially modifiable factor in promoting breastfeeding continuation.
Results from this study were presented at the Society for Pediatric Research meeting in May 2002 and have been printed in abstract form as part of the proceedings.
This work was supported by the Innovation Program of Kaiser Permanente Medical Care Program, Northern California; grants MCJ 067951 and 6 H 16 MC 00050 from the Maternal and Child Health Bureau, Health Resources and Services Administration, Department of Health and Human Services; and the Agency for Health care Research and Quality. Dr Taveras was supported by grant T32 PE 10018 from the Health Resources and Services Administration, Department of Health and Human Services, to the Harvard Pediatric Health Services Research Fellowship Program, Boston.
We thank the many research staff, clinicians, and mothers who contributed support and interview data for this study.
- Received April 24, 2002.
- Accepted October 23, 2002.
- Address correspondence to Elsie M. Taveras, MD, MPH, Harvard Pediatric Health Services Research Fellowship Program, Children’s Hospital of Boston, 333 Longwood Ave, LO-240, Boston, MA 02115. E-mail:
- ↵American Academy of Pediatrics Work Group on Breastfeeding. Breastfeeding and the use of human milk. Pediatrics.1997;100 :1035– 1039
- ↵US Department of Health and Human Services. Developing Objectives for Healthy People 2010. Washington, DC: US Department of Health and Human Services, Office of Disease Prevention and Health Promotion; 1997
- ↵US Department of Health and Human Services. HHS Blueprint for Action on Breastfeeding. Washington, DC: US Department of Health and Human Services, Office on Women’s Health; 2000
- ↵Ertem IO, Votto N, Leventhal JM. The timing and predictors of the early termination of breastfeeding. Pediatrics.2001;107 :543– 548
- ↵Kuan LW, Britto M, Decolongon J, Schoettker PJ, Atherton HD, Kotagal UR. Health system factors contributing to breastfeeding success. Pediatrics.1999;104 :1– 7
- ↵Lieu TA, Braveman PA, Escobar GJ, Fischer AF, Jensvold NG, Capra AM. A randomized comparison of home and clinic follow-up visits after early postpartum hospital discharge. Pediatrics.2000;105 :1058– 1065
- ↵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
- ↵Schanler RJ, O’Connor KG, Lawrence RA. Pediatricians’ practices and attitudes regarding breastfeeding promotion. Pediatrics.1999;103(3) . Available at: www.pediatrics.org/cgi/content/full/103/3/e35
- ↵Fiore MC, Bailey WC, Cohen SJ. Smoking Cessation. Clinical Practice Guideline No. 18. Rockville, MD: US Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research; 1996
- ↵Philipp BL, Merewood A, O’Brien S. Physicians and breastfeeding promotion in the United States: a call for action. Pediatrics.2001;107 :584– 587
- ↵American Academy of Pediatrics. Hospital stay for healthy term newborns. Pediatrics.1995;96 :788– 790
- Escobar G, Braveman P, Ackerson L, et al. A randomized comparison of home visits and hospital-based group follow-up visits after early postpartum discharge. Pediatrics.2001;108 :719– 727
- Copyright © 2003 by the American Academy of Pediatrics