As any new parent knows, you are supposed to feed your newborn 8-12 times per day, which is every 2-3 hours. As any newborn parent will tell you, especially one whose baby takes 30-60 minutes per feeding, such frequency is all encompassing, leaving slivers of time for adult eating and hygiene, much less anything higher in the hierarchy of needs. By the fourth day of waking our baby up to feed on this schedule, my wife and I started to wonder where this recommendation comes from. Caring for Your Baby and Young Child, the bible of the American Academy of Pediatrics, simply asserts this standard without citations to research. The vast wonderland of information that was supposed to be the internet turns out to just be the same piece of information surrounded by different advertisements. Since I do not like assertions without evidence, have the confidence to consume primary (academic) literature, and am blessed to have access to that paywalled literature thanks to my job, I decided to find out why Society tells me 8-12 feedings per day.

Before critiquing mainstream advice, one disclaimer is in order: a pediatrician friend says that the 8-12 recommendation is biologically grounded. Newborns’ livers do not function at 100% of their capacity, which is why they are jaundiced for the first few days. In addition to processing bilirubin, the liver is important for the production of glucose (sugar, our bodies’ main energy source) via a process known as gluconeogenesis. (Fun fact: newborns require more glucose per pound than adults because their heads are so large relative to the rest of their body.) Since the liver is weak, it cannot efficiently create glucose, so frequent feeding is needed to provide enough glucose precursors to prevent hypoglycemia. (Fun fact 2: almost all newborns are hypoglycemic immediately after birth because they do not have to produce glucose in the uterus.) After two hours of Google Scholar, however, I could not find an exposition of how liver development means feeding 8-12 times per day. The closest I found is the Academy of Breast Medicine’s 2014 “guidelines for blood glucose monitoring and treatment of hypoglycemia in term and late-preterm neonates.”, though they do not provide good citations and so I feel I hit a dead end.

Moreover, even if I did find a clear explanation connecting liver performance to feeding frequency, the liver operates normally within a few days of birth. The 8-12 recommendation makes no mention of this very short period. In fact, as the below review shows, there is very little subtlety to the recommendation, meaning parents are not aware that the potential biological grounding of the 8-12 recommendation is only relevant at the very beginning of a baby’s life.

Now, on to the flimsy to non-existent observational evidence for the 8-12 recommendation.

Fortunately, the website Parenting Science attempts to answer this question with citations to academic literature, so I started there. The closest that article gets is to cite de Carvalho et al. 1982 (“Milk intake and frequency of feeding in breastfed infants”) and Casiday et al. 2004 (“Do early infant feeding patterns relate to breast-feeding continuation and weight gain? Data from a longitudinal cohort study”) which find that the 8-12 frequency increase infant weight gain in subsequent weeks. Both articles are very useful for learning how frequently women breastfeed their newborns, but they should not be used to recommend that feeding occur 8-12 times per day.

Before proceeding, some notes about the following studies are in order. The studies all suffer from dropout, meaning some mothers and infants who start in the study do not finish it and therefore are not reported in the statistics. Few of the studies analyze this dropout; one concern is that the missing women may breastfeed so much that they are too exhausted to continue participating, making average breastfeed rates seem lower than they actually are. In addition, most studies do not start reporting results until babies are two weeks old, probably because that is when they are supposed to have regained their birth weight. (Where does the 2 week threshold come from?!?! I should investigate that in a later post.) The average number of breastfeeding sessions may therefore be higher in a baby’s initial days. Finally, all links in this post to academic articles are to copies I have downloaded, meaning they are not paywalled for you, dear reader. Please download them if you’re interested!

There are several problems with Carvalho et al. 1982 and the 8-12 recommendation. First, the 46 mothers in Carvalho et al. 1982 feed an average of 9.8 times per day with a standard deviation of 2.5 feeds. The 2.5 means that approximately 95% of women should feed between 4.8 and 14.8 times per day, a much wider range than 8-12. 8-12 therefore seems to be simply adding and subtracting 2.5 from 9.8, which would cover 68% of women. In other words, extrapolating from 46 women to 95% of women everywhere at all times means if one feeds “only” 5 times per day, that is fine. Second, look at the below screenshot of Figure 2, the distribution of feeding frequency at 2 and 4 weeks. The key here is to notice that both are right skewed, meaning the 9.8 average is affected by large values such as 14, 15, and 16. Notice as well that the two most common feeding frequencies are 7 and 8 times per day; they occur 9 times each for 39% of the observations. Including the 4 women who feed 6 times per day, almost half of the babies in the study feed less than the minimum value used to generate the 8-12 range.

Third, and perhaps most damning for the status quo, are Tables 1 and 2. Reproduced below, they show a positive correlation between feeding frequency and subsequent weight gain. These correlations, however, are not statistically different from zero, meaning it is just as likely that there is not a positive correlation between feeding frequency and weight gain. Another problem is that these numbers are correlations and not effect sizes: while the positive sign means we know an increase in feeding frequency is associated with an increase in weight gain, the tables do not tell us the size of this relationship. That is, it is not clear if one extra feeding per day leads to a weight gain of 50 grams, 200 grams, or 1 gram. (The prose of the article does not either.) Remember, it appears that the crux of the 8-12 recommendation is because of feeding frequency’s affect on weight gain, yet the actual evidence presented in de Carvalho et al. 1982 shows there is no relationship between feeding frequency and weight gain.

Cassiday et al. 2004 is no better. For the 172 exclusively breastfed babies, there is a mean feeding interval of 2.7 hours (2 hours and 42 minutes) with a standard deviation of 40 minutes; their subjects have a range of 5-14 feedings per day. In other words, their mean is 8.89 feeds per day, once again towards the bottom of the 8-12 range. The feeding frequencies are even lower for bottle and mixed fed babies.

As for weight gain, the relationship is as unclear as in de Carvalho et al. 1982. Table 5, reproduced below, is what the paper presents to show associations between feeding and weight gain. It shows weight gain 12 days and 6 weeks after the first week of feeding by breast only, bottle only, and mixed feeding groups. There are two problems with this analysis. First, TI is their measure, and its construction is very opaque. Nowhere in the article is this measure transformed back into a weight measure, so the reader has no idea what .021 or .169 mean. Second, the statistical test they run, ANOVA, only tells one that the outcomes compared are statistically significantly different from each other. It could be the case, for example, that bottle fed babies drive the weight gain. (While the 8-12 recommendation does not distinguish between type of feeding, Society also tells women to exclusively breastfeed, so that is what I am focusing on.) To summarize, the reader has no idea how much weight is gained and whether or not the statistically significant relationship is driven by breastfeeding.

Now convinced that there is no clear basis for how many times per day to breastfeed, I turned to finding studies that measure how often women breastfeed. (I again started with the Parenting Science article’s citations.) These studies were more informative and reinforce the fact that the 8-12 range does not reflect the real world. The following paragraphs briefly summarize the first studies I found.

Nichols et al. 1984 (“Human milk intake and growth in exclusively breast-fed infants”) analyzes intake for 45 exclusively breast fed infants from the Baylor Milk Bank Program (Texas, I think). Table 1, reproduced below, shows that infants 1 month old feed 8.3 +- 1.9 times per day, with the frequency declining until it stabilizes at 4 months.

Hornell et al. 1999 (“Breastfeeding patterns in exclusively breastfed infants: a longitudinal prospective study in Uppsala, Sweden”) follows 506 infants’ daily feeding pattern for 6 months. The relevant result from them is Figure 2, reprinted below. It shows that the most frequent feeding occurs at 2 weeks of age. 50% of the infants feed 7-9 times per day, with the frequency decreasing to 6-8 by week 14.

Manz, van’t Hof, and Haschke 1999 (“The mother-infant relationshiop: who controls breastfeeding frequency?”) reports breastfeeding averages from their Euro-Growth Study, the authors’ study that followed 1152 infants from 1991-1993 across 12 countries. This brief report provides average feeding frequency at 2 months of age. The highest average number of feedings is in Porto, Portugal (8.55, 95% are from 8.14-8.95 per day); only two other sites, Athens and Dublin, have average feedings of more than 8 per day. The lowest are in Rostock, Germany (5.49-5.91), Hadj-Bihar, Hungary (6.02-6.33), and Nancy, France (5.9-6.74). The table below shows their results.

Kent et al. 2006 (“Volume and Frequency of Breastfeedings and Fat Content of Breast Milk Throughout the Day”) studies 71 infants from western Australia. Though their main interest is not breastfeeding frequency, they find that babies consumed 7.9 +- 1.8 meals per day. (Two of the authors conducted a similar study of 8 infants in 1999 and find similar results.) The authors say “meal” because they define breastfeeding as the suckling of one breast, and a meal could entail one or two breasts. Table 1 shows the breastfeed frequency per day, which is not quite the same thing.

To summarize, the rationale for 8-12 breastfeeds per day is based on subsequent infant weight gain, yet the evidence of that gain is flimsy at best and non-existent at worse. In the lived world, women appear to breastfeed much closer to 8 times per day, and it is not uncommon for babies to feed 5 and 6 times per day.

Of course, new parents need some guidelines: tearing down an old system is no good if there is nothing to replace it. The answer is to focus on outputs – pees and poops – instead of inputs (number of breastfeeding sessions). Since urination and defecation reflect the body processing breastmilk, their frequency is informative of the amount of breastmilk consumed, which is what the 8-12 recommendation is really getting at. It follows from focusing on outputs that parents do not need to be slaves to the tyranny of the 8-12 range. Instead of interrupting a baby’s sleep if more than 3 hours have passed between feedings, for example, let the baby continue sleeping. So long as the baby is producing pipi and caca, let it sleep. Baby and parents will be happier.