The truth is, there’s good science and there’s bad science—but often bad or questionable science lands in the same credible science category simply because it is called a “study.” Are there weak pro-breastfeeding studies? Absolutely, but that weakness is usually in the methodology, such as the sample size, and typically not because it was funded by a for-profit company with a vested interest in a certain outcome. Unfortunately, most of us typically don’t have the know-how to properly unpack the research before us. What we do have are attention-grabbing headlines or the assertion that a news item about a research study was “most shared” on Facebook. In this environment, important distinctions and gaping research holes get lost in the media clamor about provocative findings. Important questions about the kind of study or how was it conducted often fall to the wayside in favor of the flashy headline—even when the research is flimsy. We may think “the latest” research means better research, but in doing so we are throwing out the bathwater of years of scientific evidence in favor of one new baby.
The “bathwater” of scientific research on breastfeeding, now decades deep, actually contains infallible proof of the disadvantages of formula feeding. The evidence in favor of breastfeeding as the optimal nutrition for infants is strong enough to convince the World Health Organization, the Centers for Disease Control and Prevention, the United Nations, UNICEF, the Academy of Pediatrics, and countless other international bodies to resolutely support it. The United Arab Emirates made it mandatory by law to breastfeed for two years, noting the health benefits and cost savings. While the science of breastfeeding has thoroughly convinced health authorities, government agencies, and policy makers, it remains widely disputed on the individual level, mostly by those left with the burden of actually doing it—women.
The public’s failure to understand that it is scientifically undisputed that breast milk is the best way to feed an infant is the result of our changing relationship to science, a general misunderstanding of what science should be, and the increasing influence of money and society on the direction of scientific research.
To better understand the body of evidence supporting breastfeeding, we go to the Big Bertha of breastfeeding studies—one of the most systematic and comprehensive reviews of breastfeeding research, conducted by the Agency for Healthcare Research and Quality (AHRQ), one of the twelve agencies within the U.S. Department of Health and Human Services. The depth of this study takes into account the full body of scientific evidence—including any studies that only proved weak associations. For their study, AHRQ convened eight investigators from the Tufts–New England Medical Center Evidence-Based Practice Center along with a fourteen-member technical expert team. The landmark four-hundred-page report screened over nine thousand abstracts and identified forty-three primary studies on infant health outcomes, forty-three primary studies on maternal health outcomes, and twenty-nine systematic reviews or meta-analyses that covered approximately four hundred individual studies. They looked at short-term infant health outcomes, such as infectious diseases (including ear infections, diarrhea, and lower respiratory tract infections), sudden infant death syndrome (SIDS) and infant mortality, and longer-term outcomes such as cognitive development, childhood cancer (including leukemia), types 1 and 2 diabetes, asthma, eczema, cardiovascular disease (including hypertension), hyperlipidemia, and obesity. For mothers, the researchers examined the findings on short-term outcomes such as postpartum depression and return to prepregnancy weight, and long-term outcomes, such as breast cancer, ovarian cancer, diabetes, and osteoporosis.
In this review, not all science was equally weighted or even deemed worthy of consideration. To make sure the results were applicable to the U.S. market and clearly understood by researchers, only studies conducted in developed countries and published in English were included. In the case of studies on infant feeding, studies must have had a comparative arm of formula feeding or different durations of breastfeeding included. Studies that examined only formula-fed infants were excluded. Then the expert panel developed an approach to grade the studies that made the first cut based on their methodological quality. Primary studies and meta-analyses were graded A for good, B for fair to moderate, and C for poor. “A” studies presented the least bias and the most valid results. “B” studies were susceptible to some bias but were still considered valid, and the “C” studies had significant biases that may have resulted in invalid results. There was also a checklist developed specifically for this review to evaluate the quality of the reviews of observational studies. The checklist included such questions as whether the study included an appropriate search strategy, justification for inclusion and exclusion criteria for studies, and a description of a well-defined population. The definition of breastfeeding (either at the breast or expressed milk fed with a bottle) varied across the studies reviewed, so the authors elected to use the definition of exclusive breastfeeding as provided by the authors of the studies reviewed and to qualify the conclusions based on these definitions.
What did this megareview of scientific research of various breastfeeding studies reveal regarding the body of research on breastfeeding and infant and maternal outcomes? In full-term infants, the exhaustive literature review found that a history of breastfeeding was highly associated with a reduction in the risk of acute otitis media (ear infections), nonspecific gastroenteritis, severe lower respiratory tract infections, atopic dermatitis (severe eczema), asthma in young children, obesity, types 1 and 2 diabetes, childhood leukemia, sudden infant death syndrome (SIDS), and necrotizing enterocolitis (NEC), a serious and often fatal intestinal disease. There was no substantive relationship between breastfeeding in full-term infants and cognitive performance. The relationship between breastfeeding and cardiovascular diseases was also unclear.
The specific associations range in significance. For example, the study found that there was a 23 percent reduction of otitis media when breastfeeding. When exclusive breastfeeding for three or six months’ duration was compared to exclusive formula feeding, there was a reduction of otitis media of 50 percent. For atopic dermatitis, there was a reduced risk of 42 percent in children with a family history of atopy eczema and exclusively breastfed for at least three months compared with those who were breastfed for less than three months.
For gastrointestinal infections, one case-control study analyzed and rated as good found that infants who were breastfeeding had a 64 percent reduction in the risk of nonspecific gastroenteritis compared with nonbreastfeeding infants. Furthermore, an updated meta-analysis by the authors found the risk of asthma is reduced by 27 percent in infants breastfeeding for at least three months compared with nonbreastfed infants. If there was a family history of asthma, the risk was reduced by 40 percent. The authors also conducted their own meta-analysis on seven case-control studies and found that a history of breastfeeding reduced the risk of SIDS by 36 percent. In three meta-analyses that were rated as good and moderate regarding methodological quality, breastfeeding was associated with a reduced risk of obesity in adolescence and adult life when compared with those not breastfed. The reduction ranged from 7 percent to 24 percent. For the preterm infant, the authors performed a meta-analysis of four randomized controlled trials of necrotizing enterocolitis (NEC) and found an absolute risk reduction of 5 percent. The authors point out that this small difference is significant given the high fatality rate of NEC, one of the leading causes of infant death.
The infant health outcomes where there was little or no evidence of positive outcomes related to breastfeeding included full-term and preterm cognitive development, risk of cardiovascular diseases, and types 1 and 2 diabetes.
For maternal outcomes, a history of lactation was associated with a reduced risk of breast and ovarian cancer and of type 2 diabetes. Early cessation of breastfeeding or not breastfeeding was associated with an increased risk of maternal postpartum depression. There was no relationship shown between a history of lactation and the risk of osteoporosis. The effect of breastfeeding in mot
hers on return-to-prepregnancy weight was negligible, and the effect of breastfeeding on postpartum weight loss was unclear.
The health implications of these findings are hard to ignore. For example, acute otitis media (AOM) is a very common childhood infection that begins with an upper respiratory tract infection and spreads to the inner ear. Meanwhile, the prevalence of a first attack of AOM in children under one year of age is estimated to be 44 percent, according to the report. Given that 80 percent of children have had an episode of a serious ear infection by the age of three, therefore, a 23 to 50 percent reduction in this infection, depending on duration of breastfeeding, would affect a large number of children. Among mothers, the report found the reduction in breast cancer risk was 4.3 percent for each year of breastfeeding in one meta-analysis and 28 percent for twelve or more months of breastfeeding in the other. With one in eight U.S. women, about 12 percent, estimated to develop invasive breast cancer over the course of her lifetime, the protective benefits of breastfeeding are critical.
So if a high percentage of the actual science favoring breastfeeding is so clear on eight health indicators for infants and four indicators for mothers, why does the science of breastfeeding continue to appear confusing?
Part of the problem is how the research is being conducted. The common protocol for breastfeeding research is literally stacked against breastfeeding. In public health research, it is the biological norm, not the cultural norm, that is used as the baseline control group to compare against the “treatment.” The health comparison should note whether the deviation from the baseline is helpful or harmful. Nonsmokers are compared to smokers, showing that smokers are more ill. This happens everywhere, except in breastfeeding research. Instead of using breastfeeding as the baseline, researchers have fallen into the trap of using formula feeding as the baseline and breastfeeding as the treatment. In a proper scientific landscape, it would be incumbent on infant formula manufacturers to prove that there are no or minimal risks when formula feeding is used compared with the biological baseline of breastfeeding. Instead, breast milk is studied in order to prove it to be better than formula. This backwards science creates backwards language. Because breastfeeding is the biological norm, breastfed babies should not be noted as “healthier”; artificially fed babies should be described as ill, and that they are more often and more seriously ill.
Another way the science industry has undermined breastfeeding research is with a historic overreliance on the randomized controlled trial (RCT). The RCT is widely considered science’s gold standard for study design because little is left to chance. In an RCT, subjects are assigned to treatment groups at random. One group is treated and the outcome is compared to an identical group (the control group) that did not receive treatment. Of course, this is impossible to achieve in breastfeeding. You can’t randomize who is breastfeeding—you will know you are doing it—and it would be unethical to stop a woman from breastfeeding for the sake of research. But Dr. Michael Kramer, a highly respected researcher at McGill University in Montreal, came as close as possible to achieving an RCT in breastfeeding when he released a series of groundbreaking breastfeeding studies in 2001. Kramer found a unique way to randomize his study; instead of telling each woman how to feed her child, Kramer studied women who began breastfeeding at birth on their own. Half of the mothers, however, were encouraged to breastfeed exclusively by a health professional and received ongoing supportive interventions modeled on the World Health Organization’s Baby-Friendly Hospital Initiative. The other half did not receive the supportive intervention. He followed 17,046 healthy mothers and infants born in Belarus throughout their childhood, and the resulting Promotion of Breastfeeding Intervention Trial, or PROBIT, is the largest cluster-randomized controlled trial ever conducted in the area of human lactation during infancy. The study randomized thirty-one maternity hospitals and their affiliated health centers (where children are followed for their routine health care) to either receive the experimental intervention of breastfeeding promotion or the control arm of continuing the common breastfeeding practices, which does not include, for example, postnatal home visits from health professionals. It also produced two randomized cohorts with a substantially different exposure to breastfeeding. The women who received more information and support breastfed their children for much longer, allowing researchers to study the impact of extended exposure to breast milk. Kramer’s most notable finding was that breastfed children scored 7.5 points higher on tests measuring verbal intelligence and 2.9 points higher on tests measuring nonverbal intelligence at the age of six and a half. The PROBIT study has, over the years, produced other findings about breastfeeding—some showing strong positive associations, such as a reduced risk of gastrointestinal infections and atopic eczema in the first year of life, some showing weak to minimal associations with conditions such as obesity.
Kramer’s work was significant and remains highly referenced. The lack of an RCT for breastfeeding given the difficulty of creating one has often been used to discredit the volumes of research that do exist. But RCTs cannot be the only gold standard. RCTs are not perfect, and many things simply can’t be studied using this approach, mostly because of ethical issues. And while no one suggests abandoning the RCT, more and more scientists are supplementing RCTs with other forms of evidence.
These other study designs have a widely accepted hierarchy of credibility that can help us determine the good science from the not so good. But first, a few questions that are critical in recognizing good science: What was the sample size? Bigger is always better. And, what was the methodology—how was the study designed and how was it executed? In other words, was it an RCT or a cohort study? What about the funding source? Look beyond the prestigious sounding “institute” to see who is actually funding the research. Science funded by formula companies, baby food manufacturers, and those who have commercial interests in selling other processed infant foods deserve our immediate scrutiny. I also take interest in how the study is reported by the researchers themselves and by the general public. In other words, be leery of inflammatory headlines and sensational claims.
At the top of the science heap are systematic reviews, which are considered the most reliable evidence, followed by evidence from randomized controlled clinical trials, cohort studies, and then case-control studies. A meta-analysis, or review study, is a common statistical technique for combining the findings of other independent studies over a period of time. These are considered most trustworthy when rigor has been applied to assessing the methodological quality of the design and execution of each study included in the review. Most research falls into two categories: observational studies or experimental studies. In experimental studies such as an RCT or a controlled clinical trial (CCT), researchers introduce an intervention and study the effects. Experimental studies are usually randomized. In observational studies, which include cohort studies and case-control studies, a large group of people are followed over months, years, or decades in an effort to identify associations between behaviors and certain health markers or a specific treatment. These studies are usually more “retrospective”—using data based on events that have already happened. Observational studies have become a staple of many research journals, but they are often hotly debated when it comes to breastfeeding because it is argued that some benefits may not be attributed to breastfeeding alone. For example, if a cohort study found that women who breastfed lost postpregnancy weight faster than mothers who did not, it could be questioned whether the effect was solely due to breastfeeding since women who breastfeed may be more likely to eat healthier while they do so.
Another problem with analyzing breastfeeding data is the definition of “breastfeeding” itself. Does it mean a child is fed exclusively at the breast or can it include bottle-fed breast milk? And what frequency of breastfeeding is required to be included in the cohort? Breast milk is a delicate, living fluid, and breastfeeding benefits increase with duration. So a child who was breastfed for one month may not hav
e the same health benefits as a child breastfed for twelve months—but most studies never make the distinction. In 2012 the American Academy of Pediatrics acknowledged that among several ongoing methodological issues in breastfeeding research is “the absence of distinguishing between ‘any’ and ‘exclusive’ breastfeeding.” In evaluating outcomes, it is important for breastfeeding studies to account for the frequency of breastfeeding and for whether the baby was fed at the breast or by bottle. Yet these distinctions are lacking in many of the high-profile studies that streak across social media and that women use as decision-making tools.
Dr. Miriam Labbok, former director of the Carolina Global Breastfeeding Institute at the University of North Carolina at Chapel Hill said in an interview that she’d been encouraging a standard definition of breastfeeding for research protocols since 1988—to no avail. Labbok, an epidemiologist, said the CDC and the WHO include babies fed both expressed milk and at the breast in their breastfeeding definitions, adding that policy makers at both institutions come from a perspective of nutrition, where the focus is only on receiving the food. Researchers from maternal and child health departments, who are more likely to have a deeper understanding of the nuances of lactation science, would be more concerned with method of delivery, Labbok noted. But the distinction is critical. Studies show that babies only fed by bottle eat more and thus grow differently. A 2010 analysis shows that babies fed breast milk by bottle in early infancy were more likely to empty their bottles later in infancy than their breastfeeding counterparts, which could impact weight gain. Infants fed both at the breast and with bottles of expressed milk gained weight at a similar rate to those only breastfed, but infants gained more weight per month when fed only by bottle. There are a couple explanations why: babies can get milk out of bottles quicker than from breasts, and breastfed babies eat from their feeding cues and then stop when they feel full, but caregivers tend to see an empty bottle as the sign of a completed meal.
The Big Letdown Page 8