| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
REVIEW ARTICLE |
Department of Family Medicine (J.S.T., M.N.), Brown Medical School, Pawtucket, Rhode Island
Department of Obstetrics and Gynecology (J.E.K.), Brown Medical School, Pawtucket, Rhode Island
Department of Pediatrics, University of Rochester School of Medicine and Dentistry, Rochester, New York (R.S.L.)
Address reprint requests to: Julie Taylor, MD, MSc, Department of Family Medicine, Brown Medical School, Memorial Hospital of Rhode Island, 111 Brewster Street, Pawtucket, RI 02860. E-mail: julie_taylor{at}brown.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONBACKGROUNDDESCRIPTION OF SUBJECTREFERENCES |
|---|
Key words: breast feeding, type II diabetes mellitus, diabetes, gestational, obesity
Key teaching points:
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUNDDESCRIPTION OF SUBJECTREFERENCES |
|---|
Considerable research has investigated the link between breastfeeding and a variety of chronic diseases, including obesity and type 1 diabetes [7]. Large, well-designed studies in developed countries suggest that breastfeeding has a protective effect against obesity [8]. In addition, the association between breastfeeding and lower rates of type 1 diabetes has been established [9]. Although still controversial, there is some evidence to suggest that the biological mechanism for the relationship between type 1 diabetes and infant nutrition is immune-mediated [10–13]. Accordingly the American Academy of Pediatrics (AAP) strongly endorses breastfeeding as the primary source of nutrition for infants with a strong family history of diabetes [3].
Type 2 diabetes and gestational diabetes (GDM) are the two major types of diabetes that are distinguishable from type 1 diabetes. Some of the clinical issues pertaining to type 2 diabetes and GDM are the same as for type 1. There are also important differences. As the prevalence of diabetes increases among younger women [14–17], diabetes and maternal-child health concerns overlap more frequently [18, 19]. This review of the literature explores the relationship between breastfeeding and both type 2 diabetes and GDM in three major areas: the effect of maternal diabetes on breastfeeding rates, the impact of breastfeeding on the subsequent glucose tolerance of mothers with diabetes, and whether breastfeeding reduces the risk of type 2 diabetes in children.
|
BACKGROUND |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUNDDESCRIPTION OF SUBJECTREFERENCES |
|---|
Data Abstraction
Each of the relevant articles was reviewed by at least two of the authors. All articles were selected for inclusion in the review and critically evaluated based on 1) study design, 2) target population, 3) sample size/power, 4) specific definition of breastfeeding, 5) clear definition of diabetes, and 6) control for confounding variables. All studies involved term, singleton infants unless otherwise noted.
Definition of Breastfeeding
Many of the studies that were reviewed only reported breastfeeding versus bottle feeding, without more detailed information on exclusivity, frequency, or duration. Some studies which categorized breastfeeding as a dichotomous variable are included, although this lack of precision is noted as a limitation. Information on timing and duration of breastfeeding and the source of that information has been included where available.
Definition of Diabetes
Outcomes included impaired glucose tolerance (IGT), type 2 diabetes, NIDDM, and gestational diabetes (GDM). Table 1 summarizes the various definitions of diabetes used in individual studies of type 2 diabetes; Table 2 is a similar summary for definitions of GDM. Most, although not all, researchers used either American Diabetes Association (ADA) or World Health Organization (WHO) criteria. Oral glucose tolerance testing (OGTT) was a commonly used test, although there was considerable variation in the amount of the glucose load, the timing of the testing, and thresholds for diagnosis.
|
|
|
DESCRIPTION OF SUBJECT |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUNDDESCRIPTION OF SUBJECTREFERENCES |
|---|
Maternal Aspects
Effect of Diabetes on Pregnancy Outcomes and Breastfeeding.
A 1998 case series described the clinical outcomes of 530 infants born to 332 women with GDM and 177 women with IDDM [20]. Sixty-four percent of the infants were delivered at term, 22% between 34 and 37 weeks gestation, and 14% prior to 34 weeks. Infants of mothers with diabetes had high rates of prematurity, macrosomia, delivery by cesarean section, congenital malformations, respiratory distress, hypoglycemia, hypocalcemia, polycythemia, and hyperbilirubinemia. Many of these conditions lead to prolonged separation of mother and child immediately following birth, which can have an adverse effect on the initiation of breastfeeding. Self-reported information about infant feeding method was available for 67% of the infants; 37% of those infants for which feeding information was available were partially or exclusively breastfed. Breastfeeding was more common among women with diet-controlled GDM than in women with diabetes complicated by end-organ disease (43% versus 22%, respectively, p-value not reported). A sub-analysis of 93 breastfed infants and 90 bottle-fed infants who were assigned to routine care and enteral feeding showed that, after controlling for gestational age and class of maternal diabetes, breastfed infants had fewer episodes of hypoglycemia and less need for supplementation than bottle-fed infants. Finally, mild episodes of hypoglycemia were well controlled with either early breastfeeding or formula supplementation.
A separate study used retrospective analyses of data from a validated maternity database system to examine the relationship between maternal obesity and diabetes and adverse pregnancy outcomes in 287,213 women [21]. The authors found that increasing body mass index (BMI) correlated with incidence of GDM. This study also showed high rates of obstetric and neonatal complications associated with obesity. Obese women were significantly less likely to be breastfeeding their infants at time of hospital discharge (BMI 25–30: adjusted OR 0.86, 99% CI 0.84–0.88; BMI > 30: adjusted OR 0.58, 99% CI 0.56–0.60).
Effect of Breastfeeding on Maternal Risk of Subsequent Diabetes.
Many investigators have attempted to elucidate the metabolic effects of lactation on maternal glucose tolerance. Early studies reported a relatively high rate of abnormal glucose tolerance among postpartum women as compared to patients without a recent pregnancy [22]. However, these studies did not consider a woman’s lactation status. In addition, animal and human studies of subjects with elevated prolactin levels for reasons other than lactation demonstrated an increased rate of glucose intolerance. Because prolactin levels are elevated in breastfeeding mothers, researchers then began investigating the effect of lactation on postpartum glucose tolerance in patients with and without risk factors for diabetes.
One small cohort study assessed the influence of lactation on glucose tolerance and hormone levels in women without diabetes [23]. Of the 23 low-risk women in the study who had an OGTT eight weeks after delivery, 13 breastfed and 10 did not. The lactating women all had higher prolactin levels and significantly lower levels of estradiol (p < 0.0005) as well as lower fasting glucose and insulin levels (p < 0.05) than the non-lactating women. The authors concluded that the low levels of estradiol associated with breastfeeding may confer a protective effect with respect to glucose tolerance.
Another early cohort study evaluated 98 postpartum women with no risk factors for diabetes [22]. Fifty-two percent of the mothers were breastfeeding (exclusivity was not reported) at the time of testing. The remaining patients never initiated breastfeeding and were undergoing lactation suppression either with breast compression and diuretics (28%) or by administration of oral estrogens (20%). Each woman had a 3-hour OGTT on the fifth day following delivery. There were no differences in mean glucose levels between the breastfeeding and non-breastfeeding women (p > 0.05). However, among non-breastfeeding women, the estrogen-treated women had significantly higher rates of abnormal OGTTs (50% versus 9%, p < 0.001) and significantly higher mean glucose levels at 1 and 2 hours, but not at 3 hours. In spite of the early timing of the glucose testing and the small numbers of women studied, these results also suggest that estrogen plays an important role in insulin and glucose metabolism in the puerperium. As before, lower levels of estrogen in the postpartum period may be somewhat protective with regard to insulin resistance.
To delineate the effect of lactation on glucose metabolism, Kjos et al. studied 809 primarily Latina women with gestational diabetes using OGTTs between 4 and 12 weeks postpartum [24]. Breastfeeding was assessed as a dichotomous variable. Fifty percent of the women were breastfeeding at the time of glucose testing. Non-breastfeeding women were significantly more likely to have used insulin during pregnancy (30% versus 24%, p = 0.04), but the groups were otherwise similar. Postpartum glucose values were significantly lower in the breastfeeding group (p 
0.01). Non-lactating women developed postpartum diabetes at a 2-fold higher rate than lactating women (9.4% versus 4.2%, p = 0.01). These results persisted when controlling for BMI, age, and insulin use in pregnancy.
To evaluate the rates of and risk factors for recurrent GDM, MacNeill et al. performed a retrospective cohort study using a regional perinatal database to identify 651 women who were diagnosed with GDM and then had a subsequent pregnancy [25]. Breastfeeding was included in the analysis as a dichotomous variable. Thirty-six percent of women had diabetes in a subsequent pregnancy: 16 developed type 2 diabetes prior to the subsequent pregnancy and another 216 had recurrent GDM. The authors found no difference in the recurrence rate of GDM between women who breast or bottle-fed after the index pregnancy.
Kjos et al. performed another prospective cohort study to evaluate the association between birth control methods and type 2 diabetes among 904 Latina patients with a history of GDM [26]. Patients with postpartum diabetes were excluded using OGTTs between 4 and 16 weeks postpartum. Women were then divided into 3 groups based on type of birth control: non-hormonal contraception, progestin-only oral contraceptive pills (OCPs), or combination estrogen-progestin OCPs. All women were followed from the time of their initial postpartum visit until they developed diabetes or until closure of the study after 7.5 years. Each patient had at least one OGTT during the follow-up period. Breastfeeding was assessed at the beginning of the study and characterized as a dichotomous variable. Forty-one percent of women using non-hormonal contraception, all of the progestin-only OCP users, and none of the combination OCP users breastfed. Nineteen percent of the patients developed type 2 diabetes during follow-up, with an average incidence of 9.9% per year. With any type of OCP use, the incidence of type 2 diabetes was 11.7% compared with 8.7% in non-hormonal contraceptive users. The adjusted relative risk of developing diabetes was 2.87-times higher among progestin-only OCP users than among combination OCP users after controlling for insulin treatment in pregnancy, glucose at initial postpartum OGTT, weight change from the initial postpartum visit, completion of an additional pregnancy, and prior use of other contraceptive methods (95% CI 1.57–5.27). The authors then investigated breastfeeding as an independent risk factor among patients who chose non-hormonal forms of contraception and found that the risk of developing type 2 diabetes was not significantly different between breastfeeding women and bottle feeding women (adjusted relative risk (RR) 0.90, 95% CI 0.56–1.46). The authors concluded that progestin-only OCPs were associated with an increased risk of diabetes in breastfeeding Latinas with recent GDM.
Personal Infant Nutrition History.
A single study has looked at whether women with GDM are more likely to have been bottle-fed themselves as infants [27]. As part of a larger study on the relationship between GDM and a woman’s own birth weight, 138 women with GDM were compared with 100 women with normal glucose tolerance. Breastfeeding was self-reported in one of 5 categories (breastfed only, mainly breastfed, equal amounts of breast and bottle, mainly bottle, bottle-fed only) for the first 3 months of life. There were no significant differences between women with GDM and women with normal glucose tolerance with respect to the method by which they were fed as infants (p = 0.333). No control for confounding was reported.
Pediatric Aspects
Association between Breastfeeding and Type 2 Diabetes.
Bottle-feeding has been associated with adult obesity [28–31], and obesity is a strong risk factor for diabetes. Therefore, it makes sense that infant nutrition would affect the incidence of type 2 diabetes, although this specific association was not studied until 1997.
As part of a larger longitudinal diabetes study among the Pima Indians of Arizona, a population with a very high prevalence of type 2 diabetes, Pettitt et al. were the first to examine the association between infant feeding practices and type 2 diabetes (at that time called NIDDM) in later life [32]. Using a retrospective cohort analysis, they studied 720 people age 10–39 years old. A standard questionnaire administered to mothers by trained interviewers was used to classify infant feeding practices for the first two months of life into three groups: exclusive breastfeeding, some breastfeeding, or exclusive bottle-feeding. Each mother and child in the study had an OGTT. Results were adjusted for age, sex, birth date, parental diabetes, and birth weight. People who had been exclusively breastfed had significantly lower rates of type 2 diabetes than those who were exclusively bottle-fed in all age groups (age 10–19, 0% vs. 3.6%; age 20–29, 8.6% vs. 14.7%; age 30–39, 20.0% vs. 29.6%). Type 2 diabetes was 59% less common in exclusively breastfed people compared with those who were exclusively bottle-fed (adjusted OR 0.41, 95% CI 0.18–0.93).
The authors’ conclusion that exclusive breastfeeding for at least two months is associated with lower rates of diabetes which are clinically significant, even after controlling for other well known risk factors for diabetes, has been questioned [33]. If bottle-feeding was indeed a significant risk factor, the prevalence of type 2 diabetes should be inversely related to national breastfeeding rates in other ethnic groups, which it is not. While the authors propose a possible mechanism whereby breastfeeding leads to a caloric intake more suited to the child’s needs during a critical stage in development, critics suggest greater illness in bottle-fed infants and poor bottle preparation as alternative explanations for the difference.
A more recent article which examined prenatal and early infancy risk factors for type 2 diabetes in children also found that breastfeeding may have a protective effect [34]. This case-control study included 46 Native Canadian patients younger than age 18 with type 2 diabetes and 92 age- and sex-matched clinic controls without diabetes in the province of Manitoba. A Native nurse interviewer obtained information on the duration of any breastfeeding and the duration of exclusive breastfeeding. Potential confounders that were included in multiple logistic regression models included gestational diabetes; use of traditional diet, smoking, and alcohol during pregnancy; mother’s pre-pregnancy BMI; and birth weight. The authors found that breastfeeding for one year or longer was a significant independent predictor of future diabetes (adjusted OR 0.24, 95% CI 0.07–0.84) as was breastfeeding for 6 months or longer (adjusted OR 0.36, 95% CI 0.13–0.99).
Children of Mothers with Diabetes.
As part of the same, previously mentioned Pima Indian longitudinal diabetes study, Pettitt and colleagues also sought to evaluate the long-term effects of diabetes during pregnancy and the influence of early life events such as infant feeding on 10 to 39-year-old offspring of both mothers with diabetes and those without [35]. Of the 572 women included in this retrospective cohort analysis, 21 had GDM. These women and their children, beginning at age five, were then followed prospectively with OGTT testing biannually. Trained interviewers obtained information from mothers on infant feeding practices during the first two months of life. For these analyses, women who breastfed exclusively for at least two months were compared with women who did not breastfeed at all. Multiple logistic regression was used to control for confounding by age, sex, birth weight, birth date, and presence of diabetes in either parent. The researchers found that diabetes in the next generation was less common among breastfed children (6.9% vs. 30.1% among offspring of women without diabetes and women with diabetes, respectively) than among bottle-fed children (11.9% vs. 43.6%, respectively). Therefore, in theory, increased infant breastfeeding rates may lessen or prevent long-term adverse outcomes such as diabetes during pregnancy in the next generation. Additional studies are needed to confirm this preliminary finding.
A different prospective cohort study examined the consequences of breastfeeding during the early neonatal period on offspring of mothers with diabetes in Berlin, Germany [36]. Of the 112 mothers who participated in the study, 83 had type 1 diabetes and 29 had GDM. The children were evaluated prospectively for the impact of ingestion of either diabetic breast milk or non-diabetic banked donor breast milk during the first week of life on relative body weight and glucose tolerance at a mean age of 2 years. Infants in this study received differing amounts of milk from their biologic mothers with diabetes and banked-donor breast milk from unrelated women without diabetes. Exact mean volumes of milk ingested per day were calculated for each type of milk separately using a protocol involving pre- and post-feed weighing of the infants. Children underwent height and weight measurements and OGTTs at age two years. Body weight 
110% of age- and sex-specific standard population measures was considered overweight. After controlling for birth weight, gestational age, sex, age, type of maternal diabetes, and maternal BMI, the volume of diabetic breast milk ingested correlated positively with the risk of being overweight at age 2 (adjusted OR 2.47, 95% CI 1.25–4.87). Risk of childhood IGT decreased by increasing amounts of non-diabetic banked donor breast milk ingested neonatally (adjusted OR 0.19, 95% CI 0.05–0.70). The authors concluded that early neonatal ingestion of breast milk from mothers with diabetes as compared to breast milk from mothers without diabetes may increase the risk of becoming overweight and, consequently, developing IGT during childhood.
|
DISCUSSION |
|---|
Although no study has ever reported an increased risk of developing diabetes from breastfeeding, a single study did show that the use of progestin-only OCPs was associated with an almost 3-fold risk of developing type 2 diabetes compared with the use of combination estrogen-progestin OCPs for breastfeeding Latina women with recent GDM. In addition, the magnitude of that risk increased with the duration of uninterrupted use. As future research corroborates or refutes these findings, clinicians will need to counsel breastfeeding women at high risk for diabetes about postpartum contraception accordingly.
In spite of some criticism and alternative theory voiced with respect to the pediatric literature, children who are breastfed for at least 2 months may have lower rates of type 2 diabetes. With the increasing rates of type 2 diabetes in both pediatric and adult populations, any intervention that can potentially decrease risk should be encouraged. It is important for prenatal patients to understand all the potential benefits of breastfeeding, one of which appears to be a lower risk of type 2 diabetes in children who were breastfed.
Most of the studies reviewed have limitations that preclude generalization to all maternal-child populations. Small study populations call null conclusions into question. There were rarely references to power calculations, either before or after data were analyzed. Many of the early studies of glucose metabolism either excluded breastfeeding women or did not assess infant feeding method. Subsequently, many studies characterized breastfeeding as a dichotomous variable. Exclusivity, frequency, and duration of breastfeeding were rarely delineated. Such inadequate documentation of breastfeeding details complicates interpretation of results. If a woman who breastfed for 2 days and one who breastfed for 2 years are in the same exposure group, the true impact of breastfeeding will be diluted and study results will be biased towards the null. Last, as is shown in Tables 1 and 2, studies of diabetes utilize a variety of different standards and definitions based on ongoing research, with considerably less standardization of the definition of GDM over time as compared with type 2 diabetes. New diagnostic criteria strongly suggest that the diagnosis of type 2 diabetes be made on the basis of fasting blood glucose only [37], although the WHO does suggest continued use of the OGTT for patients with blood glucose values in the [uncertain range] [38]. Better standardization of testing and clarity of definitions will be extremely important for meaningful interpretation of future research. While pediatric studies generally had long-term follow-up, most maternal studies followed patients for less than 3 months postpartum. Such short duration of follow-up makes it difficult to assess true risk for later development of diabetes. Many studies did control for multiple confounders, although each study considered different variables.
More research on this important subject is warranted. There are many studies of patients with type 1 diabetes. As gestational and type 2 diabetes reach epidemic proportions, specific studies of these types of diabetes are needed. Although some studies of high-risk groups such as Pima Indians and Latinas were fairly large, large studies of more heterogeneous populations are needed. Better data collection on breastfeeding frequency and duration as well as formula supplementation is imperative. Because the metabolic environment surrounding glucose tolerance in pregnancy and the postpartum period is extremely complex, studies with more attention to lactation status and with longer follow-up would help clarify the hormonal interactions and subsequent risk.
Based on the literature reviewed, breastfeeding should be strongly encouraged for all women, with or without diabetes. The AAP currently recommends exclusive breastfeeding for 6 months and continued breastfeeding for at least 12 months [3]. Better prenatal education and counseling about breastfeeding should be made available to pregnant women with GDM or type 2 diabetes. In addition, hospital staff should be knowledgeable with respect to the benefits of breastfeeding and comfortable assisting higher risk women with initiation [1]. Given the higher rates of maternal and pediatric complications, early initiation and full support of breastfeeding is especially important for mothers with diabetes who wish to breastfeed. For a woman with a history of GDM, current practice guidelines recommend 6-week postpartum screening with fasting plasma glucose followed by annual screening, regardless of whether she breastfed or not. Values of 126 mg/dL or higher, confirmed by repeat testing, are diagnostic of diabetes [37].
Received September 30, 2004. Accepted June 27, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONBACKGROUNDDESCRIPTION OF SUBJECTREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
