The study did not attempt to establish a causal effect; it is impossible to know from these studies whether nursing causes a reduction in diabetes (or that infant formula increases diabetes rates) or whether it's just an observed fact that diabetes is more common among formula-fed babies. It may be, for example, that nursing mothers tend to prepare more healthful meals for their growing children – and that healthy meals is the reason that diabetes is less frequent among those who breastfed.
As with most research on breastfeeding, there are no controlled studies to test the hypothesis that nursing is better than formula. A controlled study would take two groups of women who are all similar in demographics (mother's BMI, baby's birth weight, socioeconomic status, history of diabetes, etc) and randomly assign one group to nurse, and the other to use formula. For obvious reasons, we can't carry out a study this way-- instead we have to just compare the results of breastfed babies with those who didn't nurse. And that means that a whole host of other factors can get in the way. The differences in population that might affect the results are called confounding factors.
And here's where the research becomes hard to pin down. There were three studies considered by HHS which controlled for the most obvious confounding factors. They included mother's age, baby's sex, mother's BMI before and during pregnancy, baby's birth weight, socioeconomic status, prenatal diabetes, maternal diabetes in pregnancy, hospital duration in infancy, and maternal diet and smoking. The research did find a negative association between breastfeeding and diabetes – in other words, they found that breastfed babies were less likely to develop type 2 diabetes as adolescents or young adults. The results of the study were similar both before and after adjusting for confounders. In other words, when they did not adjust for all these confounders, the benefit was approximately the same.
Why is this important? The reason is that there were four other studies included in the HHS analysis that had not adjusted for these confounders. Normally, one would then discount the studies because they results may have been very different if they had adjusted properly. But the fact that the first three studies had similar results before and after adjusting suggest that this might also be the case with these four. While we should take these additional four studies with a grain of salt, they also suggest that breastfeeding is correlated with a lower risk of type 2 diabetes. The seven studies together showed a 39% decrease in diabetes among those who nursed. And, it seems, the benefit comes even if you only nurse a little – the comparison was made between babies who “ever” nursed to those who “never” nursed.
Overall, the research is indicative but not conclusive. Nursing does correlate with reduced rates of diabetes in those studies considered by HHS, but does nursing cause a reduction in diabetes? Does formula
lead to an increased rate of diabetes? The studies suggest this might be the case, but there's a caveat.
As with many of these studies, there are many possible confounding factors that were not taken into account. They include a person’s weight before developing diabetes, a person’s activity level, and various other indicators including cholesterol, blood pressure and glucose tolerance. However, if weight itself (or any other risk factors) is impacted by nursing – i.e. weight is in the causal pathway – then it should not be considered a confounding factor! This makes the research even harder to evaluate. Overall, HHS felt that the most influential factors not in the causal pathway for nursing were adjusted for, and the results are that nursing is associated with reduced diabetes risk.
That said, we can’t ignore the elephant in the room. A major confounder that no one considered is the kind of formula or other food introduced to these babies, as well as in their diets as children. This has changed and keeps changing as time goes on.
Formula-fed babies 50 (or even 10) years ago are not formula-fed babies today, yet six of the seven studies were on adults, most of whom were drinking formula before 1980 when Infant Formula Act of 1980 was passed to give FDA authority to control the contents and quality control in manufacturing of formula. In 2002, DHA (docosahexaenoic acid) and ARA (arachidonic acid) were introduced into some formula; these are both fatty acids that have been found in breast milk -- and babies drinking that kind of formula were not included in any of these studies.
Another important confounder is the issue of food introduction. Is it possible that some of the effect we observe in type 2 diabetes comes from parents who mix in regular milk or some cereal into the formula? While today that may be widely believed to be a no-no, we really don't know what people were doing a generation ago.
There are other confounders that couldn't be adjusted for as well, such as recall bias (when those who have diabetes are more likely to remember being formula-fed) or publication bias (when "no association" isn't worthy of publication). These possible biases were explicitly mentioned in the HHS study.
Does that mean we don't know? No, the science is telling us that diabetes is more common among formula-fed babies. We cannot be sure of the cause for this, but it is reasonable to consider this risk factor a part of a decision on whether to nurse or not. However, nothing is sure-fire. There may be some other factor behind the association between diabetes and infant formula, and the association may not even exist with today’s formula.
When it comes to type 1 diabetes, the results are far less clear. According to the HHS report, there are some studies showing that nursing for more than three months has a positive effect, and others that suggest this is not the case. There are some studies suggesting that food before three months increases the rate of type 1 diabetes – and still others that show it decreases the rate!
Overall, all the studies included in the HHS overview were published over ten years ago (bringing up the specter of changing ingredients of formula, and changing social norms for the introduction of other foods in babies' and kids’ diets). Most of the studies did not appropriately control for possible confounders. In addition, there was strong evidence of recall bias (i.e. people with type 1 diabetes more likely to report a shorter duration of breastfeeding than control subjects).
How big is the type 2 diabetes risk?
Type 2 Diabetes affects approximately one in 17 adults in the United States. It is unknown what percentage of those people were formula-fed and what percentage were nursed. However, if this were the risk for formula-fed people, and if nursing were causally related to reducing diabetes rates, you could lower your child’s risk from one in 17 to about one in 28 by nursing him.
According to the Centers for Disease Control, in 2004 3.1 percent of deaths were due to diabetes. Given that everyone dies eventually, not developing diabetes would extend life for about 1.2 percent of the population. Another way of seeing that is that nursing has a 1.2 percent chance of saving your child's life due to diabetes. Contrast that to death by heart disease, which claims the lives of over 27 percent of those who die. Making decisions that affect your offspring's likelihood to die by heart disease will have a much greater effect than nursing will.
What about those ads?
One advertisement developed by the Ad Council to promote nursing had syringes next to bottles, to suggest that formula increased the risk for diabetes. Insulin injections are often used to treat type 1 diabetes, but type 2 is typically treated with drug therapy.
But Type 1 diabetes is barely affected by nursing - the ads would have been misleading had they been aired. Should we blame the fact that they were pulled on pressure from industry interests? Perhaps the formula industry had the resources to look at the science and complain about a misleading public health message. But that doesn't make the government corrupt for pulling the ad - it misrepresented the science, and would have prompted a pointless guilt trip among an untold numbers of women who can't or don't want to nurse.
And as for the ads with a pregnant woman riding a bull, perhaps have a chuckle and take it with a grain of salt. Physical trauma rarely causes miscarriage (and formula rarely causes diabetes).
View the Technorati Link Cosmos for this entry
