Someone is finally trying.

Four years ago I was shopping for baby formula. I was horrified to discover that most formulas were mostly fructose and soybean oil. Including the fancy ‘organic’ ones. Humans digest fructose hepatically, and overconsumption of fructose has been implicated in a variety of modern health problems including diabetes and atherosclerosis. Why would we feed it to our babies as their primary sugar?

Oh right. It’s super cheap.

Soybean oil is a plant oil high in polyunsaturated fats and low in saturated fats. It and other high omega-6 oils have been implicated in numerous health problems. It is notoriously the only fat found in the tpn (intravenous) formulation given to infants, which causes liver failure and death. The FDA will not approve a tpn with the appropriate fat contents to keep infants alive, sparking outrage. Why would we feed our babies soybean oil instead of milk fat? It was pointed out to me that this was probably related to shelf-stability of the fat in question. Milk fat likely spoils faster.

Plus, soybean oil is cheap. As is fructose. And made with 'organic' vegetables…

So, I sprang for the premium formula which contained lactose instead of fructose, but still was mostly soybean oil.

Four years later, I am pleasantly surprised by the rise of much more sophisticated baby formulas. Enfamil Enspire, and it looks like some other formulas, have added a more complex mix of sugars and fats. The primary sugar is still lactose, but in addition there are galactooligosaccharides and polydextrose. These prebiotic sugars pass through the stomach undigested, but serve as a food source for bacteria in the gut and slowly release sugar as they are digested there. They are thought to lead to healthier gut bacteria and more stable blood sugar.

The primary fat is still a vegetable oil blend (still high in omega 6’s), but it now contains coconut oil which is high in saturated fat. Some milk fat has been added and supplemented with omega 3s DHA and ARA. One of the newest advancements is the addition of MFGM (milkfat globule membrane), which is a mix of gangliosides and phospholipids thought to play an important part in cognitive development. If you’ve spent any time studying or researching weird Ashkenazi recessive mutations, you’ll know that many of them are mutations in brain lipid metaboloism including: Nieman-Pick’s, Fabry’s, Tay Sach’s, Krabbe’s, and Gaucher’s diseases. It has long been hypothesized that having a single copy of these traits is related to higher IQ in the Ashkenazi population, even if having two can be fatal. Suffers of Gaucher’s disease are disproportionately represented in engineering and mathematics. In any case, you want to have good brain lipids. Let’s put them in formula.

Is the exact mix of components now used in these new formulas the best we can do with our knowledge and tech level? No. But it is a vast and obvious improvement over the previous formulas of fructose and soybean oil.

Why did it take this long, and why am I so surprised that we have it? I’m a cynic. I had assumed that formula was the way it was because of some regulations somewhere, and that companies didn’t want to deal with making changes. Children are too important to learn about, etc. Maybe this was true and demand finally got so loud that it was finally worth the extra cost and effort. When I think about the state of nutrition science twenty years ago, it’s unsurprising that formula looked the way it did. It takes time for a culture to shift as well. The people who cared most about what their babies were eating went crazy with demanding that everyone breast feed (no matter what the practical reality of your circumstance was). Breastfeeding became a moral imperative and breastmilk a sacred symbol of MotherLove, which could never be replicated in a food science lab.

What will the outcomes with the new formula be? We will probably *never* know. Children are too important to learn about. We don’t even have a good way of comparing formula to breast milk. Everything is confounded. Nothing is controlled. The best we can say right now is at least we know the outcomes on the old formula were only as bad as they are. That is to say, we are still uncertain if it was worse than breast milk, and therefore it must be at least mostly fine. In any case, I’m buying the new stuff.

New Comment
22 comments, sorted by Click to highlight new comments since:

I think it's considered low status for intelligent people to work on object level things even if remunerative. Unless it's extremely remunerative, in which case they're a genius in retrospect. But that doesn't help with the incentives at the time.

I'm assuming (but not 100% sure) this is related to the baby formula issues discussed in Inadequate Equlibria?

We don’t even have a good way of comparing formula to breast milk. Everything is confounded. Nothing is controlled.

Hmm. To make sure I understand: you could randomize babies getting formula vs breastmilk, but if you have reason to believe formula is bad for babies it's particularly unethical. And there are people who breastfeed and people who use formula that you could run a study on, but the people who self-select into formula and breastmilk have all sorts of confounders? (Makes sense, although I notice I'm not actually sure I could predict whether, say, poor mothers are more likely to breastfeed or to use formula)

It still seems like a situation where there should be enough families using each that you can attempt to control for confounders.

You can do an encouragement design similar to what was done in Belarus by randomizing some hospitals to adopt breastfeeding-friendly policies and some to not adopt them. Unfortunately, since not all parents in a breastfeeding-friendly hospital will breastfeed and not all parents in a control hospital will use formula, and since you're randomizing at the hospital level, your sample size has to be huge to detect any effect. And because many of the outcome variables you're interested in are long-term (IQ age seven, for example), you have to follow people for a long time. It's very very expensive and it takes forever.

The Belarus results are IMO the strongest results we have about the benefits of breastfeeding, and show a huge rise in IQ from three months of breastfeeding. Of course, as this post points out, the top formula brands have improved their product in the past decade, and modern babies may receive better milk than the babies of the Belarus study. ¯\_(ツ)_/¯

Ozy - sibling studies have a major problem - they don't take into account the reasons why a mother would breast-feed one child but not the other. If you ask moms about this, they always have an answer, and it is usually something like, 'Josh was very sleepy and just wouldn't suck. We had to give him a bottle to get him to eat at all.' My mother basically gives this exact story for why I was breast-fed and my brother was not.
And my brother had developmental problems and I did not. I don't think this is because he was fed formula.
Remember, weaker/sicker babies are more likely to get formula, and sicker/older/tireder/more depressed mothers are more likely to formula feed. In order to breastfeed, everything has to go right. One thing goes wrong, and it's on to formula.

I'm confused - the Belarusian study Ozy is talking about wasn't a sibling study, right?

I think the point here is that there are many circumstances that can influence the choice to use formula, and these might have a stronger effect than the choice itself.

But population-level differences in populations that were encouraged to breastfeed vs. not encouraged to breastfeed, as in the Belarusian study, should circumvent that.

Correct. I guess I'd rather have an appropriate quantification of the "encouragment" though; but I could be wrong, I will read the study design...

"Sites were randomly assigned to receive an experimental intervention (n = 16) modeled on the Baby-Friendly Hospital Initiative of the World Health Organization and United Nations Children's Fund, which emphasizes health care worker assistance with initiating and maintaining breastfeeding and lactation and postnatal breastfeeding support, or a control intervention (n = 15) of continuing usual infant feeding practices and policies."

There have been a number of randomized trials of breast feeding vs formula. I personally know a mother who was randomized c1980. She was not happy when she learned that the hypothesis being tested was that formula had a (specific) negative effect. Which it didn't. No RCT has found any effect. However, the RCT have been very wasteful. They should treat randomized children as a valuable resource, like a twin registry, to be followed for years and extensively measured, but they don't; indeed, I don't think they're allowed to contact them again.

I don't see how it's possible to really randomize this. No one's going to stick with a feeding method if they think it's best for their child to switch, just because they signed a form telling some researcher they would. Baby sleep studies have the same problem.

As far as I know, the closest we have is the Belarusian PROBIT study (as Ozy mentioned above) where it was advising that was randomized.

Do you apply this skepticism to all non-blind randomized studies? If people have an opinion on the right thing to do, they don't join the study. And studies do ask people if they followed the instructions.

I'm not talking about blinding, I'm just talking about randomizing. That's right, in areas with obvious confounders like class, baby health, and maternal stress level, and relatively small differences in outcomes between the groups anyway, I don't think correlational data is worth much.

Having parented a difficult-to-feed baby and having tried everything I could think of to get calories into her, I'm quite sure that even parents who start out willing to follow a given recommendation quickly change their mind if things don't seem to be going well. (If not, you're selecting for parents who are willing to prioritize following instructions over their baby's health, which certainly gets you a different population than is typical.)

It's a mess. In general poor people are more likely to use formula since they have to go back to work/don't have the same level of indoctrination- oops education - about the benefits of breast feeding, and breast feeding is a lot of work. Then there's the issue that sicker babies often have to be formula fed, because they have weaker sucking reflexes and/or require special high-calorie formula. Multiples are more likely to be formula fed, for obvious reasons. Babies of older mothers are more likely to be formula feed, since older moms produce less milk, etc. etc. More obsessive and more highly educated mothers are more likely to breast-feed for obvious reasons. In general, my conclusion from the (noncomprehensive) reading I've done about it indicate that breast feeding clearly reduces early respiratory and GI infections as well as reduced colic and GI distress (while breastfeeding), but has unclear impact on long term psychological, physical, and cognitive health. Overall those things look better with breast-fed babies, but attempts to control for other things often negates the effects, leading to yo-yoing articles about the supremacy of breast milk depending on the fashion of the day. However, going back to theory, it would be very strange if breast milk weren't better given human's past experience with making food-substitutes. That being said, the healthiest baby is a fed baby, and the impact of formula vs breast feeding is unlikely to outweigh many other factors in a person's life, such as milk production, needs to earn money to support the family, and mental health of the mother (depression in mothers is very highly correlated with poor long term outcomes).

Then there's the issue that sicker babies often have to be formula fed...

Why do you believe this? My son is currently in the NICU (born at 25 weeks), and they push breast-feeding extremely hard, even going as far as to make it sound like negligence if you don't attempt to breastfeed (due to higher incidences of NEC in premies, among other reasons). Babies whose mothers can't breastfeed are supplied with donor milk, not formula. When my first son was born (who wasn't a premie or otherwise sick), the nurses and doctors talked as if it were just a matter of personal preference.

My experience could be unrepresentative though, so I'd love to see some kind of justification for the above-quoted claim.

Sick babies are often too weak to suck much - and this is true even if the baby isn't sick enough to require a nicu stay. If a baby has to be in the hospital - it can be difficult logistically to breastfeed them, and of course if women aren't dedicated to it, they won't maintain milk. My son was required to stay in the nicu for 4 days (for ridiculous reasons - he was fine). I was only allowed to stay in the hospital 2 nights, and I was exhausted and needed to sleep.
I ended up allowing them to feed him formula since my milk was slow to come in - no one strongly encouraged me to stay there and breastfeed in the night. I got a 5 minute tutorial on how to use a pump, which was briefly suggested. It's great that some hospitals are encouraging breastfeeding and providing donor milk to premature babies. I don't know how universal this is. I know other women who have complained of similar problems I faced.

I was really glad you wrote this. I'm also confused about what 1-year-olds should drink:

I don't really know how to make it, and quick Googling didn't give me much, but I seem to recall hearing/reading that in China people often made and fed newborns a sort of thin rice gruel as a substitute for breastmilk when necessary. Obviously we could do better since I assume this will neglect many important micronutrients and may have a poor macronutrient profile such that the baby may need to eat much more thin rice gruel than breast milk, but it does suggest we have options that are still cheap but don't rely on such actively negative choices as in formulas of the past and (possibly) the present.

I find this topic interesting because I myself had to be formula fed due to lactose intolerance from birth; that is, unlike most people who initially produce lactase and then either mostly lose the ability (most people) or retain the ability but at a decreased level (Europeans, Tibetans, some Bantus), I simply seem to have never been able to produce lactase at all. If I had children it seems likely they would inherit this trait since it seems to be genetic. This does make me wonder too why, in the absence of lactose intolerance, we don't instead generally feed newborns who have difficulty breast feeding non-human milk (even if we first dehydrate it to make it portable the same way formula is)? It wouldn't have helped me or help my potential offspring, but it does seem strange that we should be so focused on plant-based formula given that most people have neither a medical need nor an ethical preference for it over milk.

Formula is typically based on cow's milk. Human milk has higher sugar (lactose) content than cow's milk. The nutrition for building baby cows and baby humans is different enough that infants shouldn't just be fed a balance of nutrients that works for other mammals. Some cultures use this or other mixtures like sugar water out of necessity, but it's not a good idea if you can avoid it. Around one year, once the child is eating other foods, is when they start recommending adding cow's milk.


I'm confused. Here is a quote from a UK website: "Infant formula is usually based on processed, skimmed cow’s milk. Added ingredients include vitamins, fatty acids and prebiotics (carbohydrates that can stimulate the growth of ‘good’ bacteria in the digestive system)." Are things very different in the US?

Maybe my experience of it is skewed, but there are lots of plant-based formulas sold in the US or primarily plant-based with small amounts of animal products added for nutritional reasons rather than as a base.

You can certainly buy plant-based formula, but most of the typical formulas you'll find on Amazon or at a US grocery store are based on cow's milk.