Queenan, John MD; Power, Michael L. PhD; Farrow, Victoria BS; Schulkin, Jay PhD
Breastfeeding provides a public health benefit at little or no cost. The advantages of breastfeeding over formula feeding have been demonstrated in multiple studies.1 The benefits from breast milk to the infant include immunologic, developmental, and nutritional, among others, and are so compelling that the American Academy of Pediatrics recommends exclusive breastfeeding for the first 6 months and that breastfeeding continue for at least 12 months and thereafter as long as it is mutually desired.1 The American College of Obstetricians and Gynecologists (the College) recommends that exclusive breastfeeding be continued until the infant is approximately 6 months old. A longer breastfeeding experience is beneficial.2
Recognizing the importance of breastfeeding as a public health measure, the U.S. Public Health Service set forth breastfeeding goals in Healthy People 20103; these goals have been updated in Healthy People 20204 (Table 1). This public health campaign has improved breastfeeding rates in the United States; however, the only goal that was achieved by 2010 was that 50% of mothers were breastfeeding at 6 months.5 There is a need to study why progress is so slow. To that end, we present results of a survey study regarding practice patterns and attitudes of obstetrician–gynecologists (ob-gyns) toward promoting breastfeeding in their patients. We compare the rates of breastfeeding among their patients as estimated by the physicians to the Centers for Disease Control and Prevention (CDC) state-by-state Breastfeeding Report Card5 data to assess how realistic the perceptions of ob-gyns are regarding their patients' breastfeeding behavior and to assess whether there are geographic patterns of physician attitude or practice that might affect breastfeeding rates.
MATERIALS AND METHODS
Surveys were mailed in July 2010 to 600 members of the College. The survey consisted of 58 questions covering a range of topics: physician and patient population demographics, satisfaction regarding the breastfeeding practices of their patients, opinions and knowledge of breastfeeding, opinions regarding breastfeeding education and practice responsibilities, opinions regarding discussion of breastfeeding practices, recommendations regarding the duration of breastfeeding, and ratings regarding their opportunity to provide breastfeeding education.
Participants were all members of the College's Collaborative Ambulatory Research Network, a group of college fellows who agree to participate in four to six surveys every 12 months. Collaborative Ambulatory Research Network members are a representative sample (by age, sex, and geographic location) of the College's membership, which includes more than 90% of ob-gyns in the United States. During the more than 20-year history of the Collaborative Ambulatory Research Network, comparisons of their responses on surveys with those of randomly selected fellows have rarely indicated any significant differences. Approximately half of Collaborative Ambulatory Research Network members were randomly selected for this survey. Second and third mailings were sent 3 and 7 weeks after the initial mailing to encourage nonresponders.
Responses were entered into a computer-based software package data file for analysis. The study identification number, sex, birth date, and geographic location for all physicians were entered into the database to allow comparison of basic demographic categories of responders and nonresponders. The institutional review board of Georgetown University School of Medicine approved the research.
Summary values are given as mean±standard error of the mean or ±95% confidence interval (CI) for proportion estimates. Analyses were conducted with an α set at 0.05. We tested differences in proportions between subgroups using χ2 tests and differences in means using analysis of variance F tests. Nonparametric tests were used for scale response items. Correlations were calculated using the Pearson correlation coefficient between continuous parameters and the Spearman correlation coefficient when one or more of the parameters were ordinal.
The CDC published data on the average percentage of women initiating breastfeeding by state.4 To estimate the extent to which the physician responses correspond to CDC data, each physician was assigned a value based on the state in which they practice: 1=CDC data show 75% or more of women initiating breastfeeding (high); 2=65–74% of women initiating breastfeeding (medium); and 3=64% or fewer of women initiating breastfeeding (low).
A total of 290 ob-gyns returned the survey, a response rate of 48.3%. Of the 290 who returned the questionnaire, 230 reported seeing pregnant patients. Only these responses were used in the analysis. A comparison of responders and nonresponders indicated no significant difference in age. Female ob-gyns aged 50 years or younger responded at a greater rate (51.1%) than men of the same age group (40.0%). There was no significant difference in response between male and female physicians aged 51 years or older.
Of the respondents, 55.8% were women. The average age of the sample was 50.6±0.6 years; men were significantly older than women (55.1±0.9 compared with 47.0±0.7, P<.001). Mean years in practice was 17.7±0.6 (range 2–40 years). There was a roughly even split between physicians who reported being breastfed (47.6%) as children and those who did not (43.7%). However, 87.5% of ob-gyns with children (n=216) report that they or their spouse breastfed an infant.
Physician estimates of percentage of patients on Medicaid correlated significantly to the race and ethnicity of patients with lower estimates of Medicaid eligibility associated with higher percentages of white (r=−0.31, P<.001) and Asian Pacific Islander (r=−0.23, P=.001), and higher estimates of Medicaid eligibility associated with higher percentages of African American (r=0.18, P=.01), Hispanic (r=0.30, P<.001), and Native American (r=0.20, P=.003) patients. The greatest average percentage of patients eligible for Medicaid was reported among physicians who practice in rural areas and towns of 5,000–50,000 (51.5%±3.2%) and urban inner cities (46.3%±5.5%). Physicians reported the lowest breastfeeding rates in these areas (Table 2).
The responding physicians overwhelmingly agreed with the statements that breastfeeding conveys nutritional (97.4%±2.1%) and immunologic (98.7%±1.5%) benefits to the infant. Most agreed that exclusive breastfeeding is the best option (87.4%±4.3%); however, most also agreed that breastfeeding is a personal choice (77.7%±5.4%), and 55.0%±6.4% agreed that formula is an acceptable option that will not harm the infant. Female physicians were more likely to respond that they strongly agree that exclusive breastfeeding is the best option (68.5%±6.0% compared with 44.4%±6.4%, P=.005).
The responding physicians reported that, on average, 27.4%±1.3% of their patients choose to formula-feed rather than breastfeed. Thus, on average, 72.6%±1.3% of their patients initiate breastfeeding. The estimated proportion of patients that choose to exclusively breastfeed was 50.5%±1.6%. The responding physicians estimated that 47.9%±1.5% of their patients are breastfeeding at 3 months, 26.8%±1.2% at 6 months, and only 10.6%±0.7% at 12 months. Female physicians on average estimated a higher proportion of their patients choose to initiate breastfeeding (75.1%±1.7% compared with 69.6%±2.0%, P=.038) and a higher proportion of their patients continue to breastfeed at 12 months (12.1% compared with 8.5%, P=.012).
Physicians caring for a higher proportion of African American patients report more patients who choose not to breastfeed (r=−0.261, P<.001) and fewer that exclusively breastfeed (r=−0.331, P<.001). The reverse relationship was found for Asian Americans: physicians caring for more Asian American and Pacific Islander patients report more patients choose to breastfeed (r=0.233, P=.001) and exclusively breastfeed (r=0.153, P=.023). This pattern was the same for breastfeeding at 3, 6, and 12 months with high Asian and Pacific Islander patient populations associated with high breastfeeding rates and high African American patient populations associated with low breastfeeding rates, although at 12 months, the result was only a trend for the proportion of African American patients (r=−0.129, P=.058). No significant relationships were found between choosing to breastfeed and the percentages of white, Hispanic, Native American, or multiracial patients; however, the proportion of white patients was associated with exclusive breastfeeding (r=0.262, P<.001).
Eligibility for Medicaid also was related to patient breastfeeding behavior. Physicians with a higher proportion of patients eligible for Medicaid were more likely to report a lower proportion choosing to initiate breastfeeding (r=−0.465, P<.001), exclusively breastfeed (r=−0.416, P<.001) as well as lower rates of breastfeeding at 3 (r=−0.455, P<.001), 6 (r=−0.352, P<.001), and 12 months (r=−0.295, P<.001). Because the estimated proportion of African American and Asian Pacific Islander patients correlates to the proportion of Medicaid-eligible patients (positively and negatively, respectively), we calculated the partial correlations of these ethnicity variables with initiating breastfeeding and breastfeeding at 3, 6, and 12 months controlling for the proportion of Medicaid-eligible patients and vice versa. The negative correlation between choosing to initiate breastfeeding and the proportion of African American patients remained significant (r=−0.207, P=.003). However, there were no longer any significant correlations between the proportion of African American patients and the estimated proportion of patients breastfeeding at 3, 6, or 12 months. The results for Asian Pacific Islanders remained significant in all cases. The proportion of Medicaid-eligible patients remained significantly negatively associated with choosing to breastfeed and to the proportion of patients breastfeeding at 3, 6, and 12 months after controlling for the proportion of African American and Asian Pacific Islander patients.
Physicians practicing in states with high breastfeeding initiation rates (per CDC Breastfeeding Report Card data) report on average higher proportions of their patients choose to initiate breastfeeding than do physicians who practice in medium states; physicians who practice in low states report the lowest rates of breastfeeding initiation (P<.001). These results match well with the CDC data on geographic variation in breastfeeding initiation rates (Table 3). The physicians appear to be pessimistic about their patients' breastfeeding behavior at 6 and 12 months, because they estimate lower proportions of their patients to be breastfeeding at these times than the CDC data indicate. However, the patterns were consistent with the CDC data: physicians practicing in states where CDC data shows high rates of breastfeeding at 6 and 12 months reported on average significantly higher rates (Table 3; P=.005).
Physician sex and practice location did not vary significantly with the CDC geographic breastfeeding rate. Similarly, there was no significant difference in the percentage of Medicaid-eligible patients based on the CDC geographic groupings, presumably because low income is spread throughout all geographic regions rather than clustered in specific states. However, physicians from the high breastfeeding states reported the lowest estimated proportion of African American patients (10.7%) and the highest estimated proportions of Hispanic (19.9%) and Asian American patients (5.5%). In the medium and low breastfeeding states, the values were 22.5%, 10.2%, and 2.6% for African American, Hispanic, and Asian American Pacific Islander patients, respectively (P<.001).
More physicians report being satisfied with their patients' breastfeeding behavior than unsatisfied (56.8% satisfied, 33.6% not satisfied, and 9.6% neutral). Ob-gyns who report being satisfied with the proportion of their patients who breastfeed have a significantly higher average mean percentage (62.0%±1.9%) of patients who choose to exclusively breastfeed compared with physicians who are unsatisfied (33.91%±2.3%; P<.001). Physicians who are unsatisfied with the proportion of their patients breastfeeding have a significantly lower average mean percentage (58.0%±2.2%) of patients who choose to initiate breastfeeding compared with those who are satisfied (82.7%±1.1%; P<.001). Physicians who are satisfied with their patients' breastfeeding behavior estimate a significantly higher average mean percentage of their patients to be breastfeeding at 3 months (58.4%±1.6% compared with 33.3%±2.2%; P<.001), 6 months (34.8%±1.6% compared with 16.6%±1.3%; P<.001), and 12 months (13.0%±1.1% compared with 7.9%±1.0%; P=.002).
Physicians with greater percentages of African American patients are more likely to report being unsatisfied with the proportion of their patients breastfeeding (P=.02). No differences were found with any other racial or ethnic group. Additionally, physicians who report being unsatisfied with the proportion of their patients who breastfeed estimate they have a significantly higher average mean percentage of patients who are Medicaid-eligible (51.1%±3.0% compared with 26.3%±2.2%, P<.001).
Significant differences were found in physician satisfaction levels by CDC geographic region with 72.7% of physicians practicing in high breastfeeding rate states reporting satisfaction with the proportion of their patients breastfeeding compared with 60% reporting satisfaction in medium states and only 34.3% reporting satisfaction in low states (P<.001). Even within geographic regions, physician satisfaction reflected estimated patient breastfeeding behavior. For example, physician satisfaction was associated with the estimated proportion of patients breastfeeding at 6 months in all three regions (r>0.48, P<.001 in all cases; Fig. 1.).
Physician satisfaction with breastfeeding rates was also influenced by the physicians' opinions about breastfeeding and formula. Significant differences were found in satisfaction based on the responses to the following question: “Feeding an infant formula is an acceptable option that will not negatively affect the child.” Of those who disagree that formula is an acceptable option, only 35.9% are satisfied with the proportion of their patients that breastfeed their babies compared with 59.7% for those who agree with the statement and 64.5% for those who were neutral regarding formula (P=.003).
Physician Practice and Effort
Thirteen questions on the survey were considered to reflect physicians' effort toward encouraging breastfeeding by their patients. An “effort ” score was created from these questions with a possible range of 0–14 (Table 4). Three of 10 responding physicians scored 12 points or higher. There was no association between physician age and effort score; however, women (10.2±0.2) scored higher than men (8.6±0.3, P=.001). Effort score was not associated with any patient demographic parameters. There was no difference in effort score between physicians from the different CDC geographic breastfeeding rate categories. Effort score was not associated with the estimated proportion of patients that initiate breastfeeding or who are breastfeeding at 3, 6, or 12 months. Effort score was associated with the physicians' opinions of breastfeeding and formula feeding. Both physicians that agreed that exclusive breastfeeding was the best option if possible and physicians that disagreed with the statement that formula feeding is an acceptable option had higher Effort scores (Table 5).
Breastfeeding is a valuable asset to an infant's health. Expert opinion concludes that increasing breastfeeding rates in the United States would have a significant public health benefit, especially among poor and African American populations in which rates are low. To devise evidence-based interventions to increase breastfeeding, it is important to study what factors influence the decision to choose breastfeeding or formula feeding. Because the mother's choice to breastfeed and its initiation generally occur during the period of obstetric care, ob-gyns' attitudes and performance might influence the rate of breastfeeding within their practice.
Physician reports of breastfeeding rates were consistent with the geographical data collected by the CDC. Physicians in states with higher breastfeeding rates reported higher on-average estimated breastfeeding rates and vice versa. However, the surveyed physicians appeared to be pessimistic about their patients' breastfeeding behavior, estimating significantly lower rates at 6 and 12 months than the CDC data indicate (Table 3). This may reflect their lower rate of contact with their patients postpartum, resulting in an underestimation of breastfeeding in their patients.
There were no differences in physicians' opinions regarding breastfeeding across the CDC geographic categories. However, in states with low breastfeeding rates, the physicians report greater dissatisfaction with their patients' breastfeeding behavior. This pattern held true even within CDC categories with physician satisfaction strongly associated with patients' breastfeeding behavior (Fig. 1). Physicians in the lower breastfeeding states did not appear to have lowered expectations for appropriate levels of breastfeeding.
Areas that are indicative of the physicians' approach to encouraging breastfeeding in their patients include responsibility, importance, and use of lactation consultants (Table 4). Our data indicate that physicians in low breastfeeding states provide just as much effort to educate and encourage breastfeeding as physicians in states with high breastfeeding rates. It appears that the physicians in the states with low breastfeeding rates show dedication and efforts to promote breastfeeding but have appropriate dissatisfaction over the low rates. Considering the role of these physicians, there is no apparent evidence that they contribute to their patients' low breastfeeding rates. These data support the hypothesis that sociocultural factors within patient populations are more important determinants of breastfeeding than are physician attitudes and practices.
One strength of this study is the ability to compare our survey results with established and reliable CDC breastfeeding data. Because there was agreement with many of the surveys' assessments of breastfeeding practices, it lends credibility to the softer data of trying to assess physician dedication. The availability of Collaborative Ambulatory Research Network, the College collaborative group for surveying, is a second strength. They are a valuable time-tested and validated resource. A weakness of this study is the reliance on physician self-report and the inherent subjective quality of asking physicians to rate their satisfaction with patient behavior. The survey hopefully provides accurate information, but the actual practice performances and opinions reported are not certifiable. The sample size was sufficient for statistically significant comparisons; however, the 95% CI for proportional results could be as high as 6.5%. There also is a potential for nonresponse bias.
The results of this study indicate that further research to understand why African American mothers are less likely to breastfeed is needed. Re-evaluation of the Medicaid program also is indicated to assure all practical support is offered to mothers to encourage breastfeeding. The education and promotion of breastfeeding and the use of lactation consultants by ob-gyns is to be applauded. Ob-gyns appear to be making good faith efforts to encourage breastfeeding, but their endeavors do not appear to be able to overcome other factors (eg, economic, educational, cultural, and so forth) that may be limiting the rates of breastfeeding in some populations.