Part Two / The First Year

Breastfeeding: A How-To Guide

Latch, pain, supply, pumping, and practical feeding support.

Chapter 05 · 13 min full read

In this note

The one thing to remember

Breastfeeding is a physical skill that two inexperienced people have to learn simultaneously, and it is genuinely hard for many women even when nothing is wrong. Most of the rules you'll be handed are weakly supported. The ones that matter: get skilled help early, don't endure persistent pain, and judge feeding by the baby's weight and hydration rather than by how full your breasts feel.

Cutting through the noise#

Two things collide. First, effort usually works — you worked hard and got into college, got the job, maybe got pregnant. Breastfeeding is one of the first problems where trying harder genuinely may not fix it, because there's another person and some biology involved. That's disorienting.

Second, the advice you get is a mix of well-supported physiology and pure folklore, delivered with identical confidence. "Nipple confusion" sounds as authoritative as "milk supply responds to demand." Only one of those survives a randomized trial.

Oster asked Twitter for breastfeeding woes and got: babies who wouldn't latch, cracked and bleeding nipples, mattresses that smelled of Parmesan, a husband sent on a thirty-minute bus ride for nettle tea, milk spurting on a bus every time any baby cried, and a woman who lost her fingerprints from handling sterilized pump parts. If it's hard, you are extremely not alone.

The 60-second recap#

  • Early skin-to-skin contact helps — one of the few things with randomized support.
  • Skilled help helps, particularly if someone watches an entire feed rather than just the first latch.
  • Early pain is common; persistent pain is not normal and usually has a fixable cause.
  • "Nipple confusion" from pacifiers or bottles is not supported by randomized evidence. This is the biggest myth in the chapter.
  • Supply responds to demand, but the feedback loop is imperfect and slow.
  • Milk takes longer than three days to come in for about a quarter of women — and a third of first-time mothers.
  • Herbal supply boosters are mostly unsupported. Alcohol makes supply slightly worse, not better.
  • Pumps are worse than babies at removing milk, and how well they work varies enormously between people for no known reason.
  • The only dietary restriction is high-mercury fish. That's it.

What to actually do#

RECOMMENDED - ModerateUse early skin-to-skin contact when medically possible
Why this call

A randomized study of 200 women in India found 72% vs. 57% still breastfeeding at six weeks. Confirmed across a review of many small studies, including after caesarean. Bonus finding: less pain during stitching

RECOMMENDED - ModerateGet skilled help with latch, positioning, transfer, and supply
Why this call

Evidence comes from many different interventions, so it's hard to say what specifically works. The general principle: someone who has seen it before, and who has slept recently, can spot things you can't

MUST - ModerateGet evaluated for pain that continues through a whole feed, lasts weeks, suddenly worsens, or comes with fever and a red swollen area
Why this call

Persistent pain usually has a cause — a yeast infection, damage, or mastitis. Mastitis needs antibiotics and is not something to push through

OPTION - ModerateUse a bottle or pacifier if it's useful
Why this call

Multiple randomized trials find no effect on breastfeeding success. Details below — this one deserves the space

TRY - LowA nipple shield as a temporary bridge if latch or pain is preventing feeding entirely; monitor weight
Why this call

Randomized evidence shows shields reduce stimulation and therefore milk transfer. But if the alternative is no feeding at all, that's the wrong comparison. Qualitative research has women crediting shields with letting them continue

OPTION - LowConsider tongue-tie release based on a functional feeding assessment, not appearance
Why this call

Four small randomized trials. Three looked at feeding: two found no difference, one found improvement. All four found reduced maternal pain, though self-reported. So: better evidence for pain than for feeding, and not a first-line move

TRY - LowExtra nursing or pumping to increase supply
Why this call

The demand-response biology is sound. There's no research on the best protocol, so any specific schedule you're given is someone's opinion

SKIP - LowFenugreek, lactation cookies, special teas, extra hydration, beer
Why this call

Fenugreek: two small randomized trials, one positive, one not. Hydration: no evidence it increases supply. Beer actively makes it slightly worse

Latch and pain#

What a good latch is: the baby needs a wide-open mouth and a whole mouthful of breast, not a delicate sip at the tip. Oster's friend Jane: "You really have to jam the kid on there." Once you've done it a while, you'll recognize it — many babies make a distinctive sigh when they get it right.

Why babies struggle: prematurity, illness, birth injury, inverted nipples, or structural issues like tongue tie or lip tie (a short frenulum limiting tongue or lip movement).

The myth to discard: you'll read that a good latch doesn't hurt, so pain means you're doing it wrong. For many women, early breastfeeding hurts whether or not the latch is correct. You cannot use pain as your latch diagnostic in the first weeks.

What's normal vs. not:

  • Pain in the first minute or two of a feed, easing after: common
  • Pain that continues through the whole feed: not normal, get it looked at
  • Cracked, sore, or bleeding nipples in the early weeks: unpleasantly common, and it does resolve
  • Still agonizing after two weeks: not typical, and not something to be told will improve if you keep trying
  • Red painful swollen breast plus high fever: mastitis. This needs medical attention, not persistence

Treatments for damaged nipples: the honest answer is that almost nothing has randomized support. Lanolin, gel pads, and the rest have no trial evidence. The one thing with any randomized support — from a single small trial — is rubbing breast milk on your nipples. There's no downside to lanolin, so use it if it feels good. Oster's friend Hilary's contribution: "MOISTURIZE THE NIPS EVERY TIME."

The genuinely reassuring finding: trials that enrolled women with severe nipple trauma — bleeding, open sores — found that for most, it resolved or became manageable within a couple of weeks regardless of what they did. If it looks grim right now, that is the most likely path.

If pain persists: many states have breastfeeding hotlines, and La Leche League can often connect you with a lactation specialist by phone if in-person feels like too much.

The nipple confusion myth, properly explained#

This one is worth understanding in detail, because it's the clearest example in the book of how a correlation becomes a rule.

Pacifiers. A trial enrolled 281 women and counseled half toward pacifier use and half against. Pacifier use differed between the groups, as intended. Breastfeeding at three months: about 80% in both groups. No effect.

Then the authors did something clever. They threw away the randomization and analyzed the same data as if it were an ordinary observational study — just comparing mothers who chose pacifiers with those who didn't. Now a difference appears: pacifier users were less likely to still be breastfeeding.

Same women. Same data. Two opposite conclusions. The difference is that in the second analysis, choosing a pacifier is a marker for something else — most likely a less intense commitment to breastfeeding. The pacifier isn't causing anything; it's revealing something.

Nearly all the literature behind the nipple confusion warning is the second kind of analysis.

Bottles. Harder, because supplementing is tangled up with already having feeding difficulty. A neat trial solved it: babies who needed supplementation were randomly assigned to get it by bottle or by cup — a cup can't cause nipple confusion. If bottles were the problem, the groups should diverge.

They didn't. Both groups breastfed for around four months, with exclusive breastfeeding for two to three weeks. The method of supplementation didn't matter.

Milk supply#

The biology: production is a feedback loop — effective milk removal signals the body to make more. This is why pumping after feeds is suggested for low supply: you're faking higher demand.

Why the loop disappoints: it's slow, it's imperfect, and it can overshoot.

Milk coming in ("lactogenesis II")#

You start with colostrum — small quantities, antibody-rich. The switch to copious milk production is expected within 72 hours, and if it takes longer you're labeled "delayed."

But look at how common "delayed" is, from a study of 2,500 women:

  • Almost a quarter of women take longer than three days
  • For first-time mothers, about 35%

A label applied to a third of first-time mothers isn't describing a disorder. It's describing a badly set expectation.

What correlates with delay (correlation, not necessarily cause): smoking during pregnancy, obesity, caesarean delivery, having had an epidural. On the other side, feeding on demand and starting within an hour of birth are associated with less delay. Some of these you can't or shouldn't change — an epidural has its own good reasons.

Why it matters: delayed onset correlates with earlier stopping. Possibly because it causes excess infant weight loss, which makes everything harder. Possibly because if you were ambivalent to begin with, this setback settles it.

Too little#

Standard first move is to use the demand loop — pump after some or all feeds. Biologically sound; no research tells you the optimal schedule.

What doesn't work: hydration (good for you, doesn't raise supply), beer and alcohol generally (slightly lowers it), most herbs. Fenugreek, shatavari, and malunggay all have mixed results across small trials. None are harmful at normal doses, so trying them costs little — just don't expect much.

What does work: domperidone has randomized support. It is not available in the US, which is unhelpful if you're American; readers in the UK and Canada can get it.

Sometimes there's genuinely little or no milk. Insufficient glandular tissue (IGT) means not enough milk-producing tissue, sometimes congenital. Breast reduction surgery can also limit supply depending on technique. Neither is common, but both are real and both come as a shock partly because nobody talks about them. If this is you, supplementation isn't a failure of effort.

Too much#

Often self-inflicted by over-enthusiastic early pumping. The problems: engorgement is genuinely painful — hard, hot, aching breasts — and it raises mastitis risk. It also makes feeding harder, because milk comes too fast for the baby to manage.

The trap: pumping relieves the discomfort but feeds the loop and prolongs the problem.

What has some evidence: cold and hot packs, and cold or room-temperature cabbage leaves kept in the fridge and wrapped around the breast. Yes, really. There's also a trial supporting gua sha, a scraping therapy that produces light bruising; Oster notes Gwyneth Paltrow endorses it and invites you to weigh that as you see fit.

For the firehose problem: pump or hand-express for a couple of minutes right before feeding. It improves as the baby grows.

Diet, alcohol, and medication#

Food. The only medically advised restriction is high-mercury fish — swordfish, king mackerel, tuna. That's the whole list. Unpasteurized cheese, sushi, rare steak, deli meat: all fine. After nine months of restrictions, this is good news that nobody tells you.

The cauliflower question. Oster got a reader email asking whether the baby's grandparents were right that cauliflower and tap water cause crying. There's exactly one paper on gassy foods and colic, based on a mail survey with a poor response rate, oversampling anxious respondents, and weak statistics. It found a hint of an effect and can be safely ignored. Eat what you want. Drink the tap water.

Alcohol. The concentration in milk is the same as in your blood — and the baby drinks the milk, not the alcohol, so exposure is very low. One paper calculated that even after four drinks in quick succession, feeding at peak blood alcohol, the baby's exposure is extremely low. (The same paper notes, reasonably, that four fast drinks impairs your ability to parent, which is the actual problem.)

So: you do not need to pump and dump. Alcohol isn't stored in milk. As your blood level falls, so does the milk level. Pumping doesn't speed that up.

If you want to be maximally cautious, wait about two hours per drink before feeding.

The studies are appropriately cautious about frequent heavy or binge drinking, where there's little data and other confounding factors.

Caffeine is fine for most, with no evidence of risk to the baby — but some babies are sensitive and get genuinely fussy. If yours is, you'll know.

Medication. Most are fine. LactMed is a searchable database for essentially any drug. Two groups worth knowing:

  • Painkillers. Tylenol and ibuprofen are well tolerated and often used in high doses. Codeine and other opioids are now generally advised against — newer data showed significant nervous-system effects in babies, mainly extreme sleepiness, with a few severe cases. After a caesarean, ibuprofen may not be enough, so this is a genuine tension to work through with your doctor rather than something with a clean answer.
  • Antidepressants. All pass into milk; there's little evidence of harm to the baby. The guidance is to use whatever works for you — if you've responded to a particular drug before, use that one. If starting fresh, sertraline and paroxetine transfer at the lowest levels. Postpartum depression is serious and treating it matters more.

Pumping#

MIT ran a hackathon to design a better breast pump. Nothing marketable has emerged. Pumps remain painful, loud, heavy, fiddly, constantly in need of cleaning, and often ineffective — before you get to the logistics of pumping in an airport bathroom while the TSA wands each bottle.

Three reasons to pump:

  1. To increase supply early on. Consider renting a hospital-grade pump — better quality, and you're not going anywhere anyway.
  2. To build a stash / enable bottles. The practical advice, in the absence of any research: pick one feeding, usually in the morning when supply is highest, and pump after it consistently. Small amounts accumulate over a week or two. Then pump during the feed the baby takes by bottle. The alternative — trying to time pumping between unpredictable feeds — is one of the more stressful things in early parenthood, and it's stressful for no good reason.
  3. To replace feeds after returning to work.

Practical notes:

  • Check flange fit and pump condition before concluding you have a supply problem
  • Get a hands-free pumping bra. At minimum you want to be able to hold your phone
  • Pump output does not measure what a baby can remove. Some women breastfeed successfully and get almost nothing from a pump. This varies wildly between people and nobody knows why
  • Advice to relax and look at photos of your baby has only distant supporting evidence — one NICU study found proximity to the baby increased production
  • Exclusive pumping (EP) is a legitimate feeding method for people who can't get latching to work. There's little research to guide it, but there are large online communities that will

Workplace reality: companies over a certain size are required to provide lactation space, but compliance varies and there's no requirement that the room be pleasant. One doctor described pumping in a coed locker room behind a towel. Thirty minutes three times a day is ninety minutes, and it's ninety minutes you're not doing something else.

Numbers in context#

NumberWhat it means
Milk in by 3 days for ~75% of womenWhich means a quarter take longer — and 35% of first-time mothers do. Being "delayed" is close to normal
Skin-to-skin: 72% vs. 57% breastfeeding at 6 weeksOne of the few interventions here with randomized support
Pacifier trial: ~80% vs. ~80% breastfeeding at 3 monthsThe clearest myth-buster in the chapter
Severe nipple trauma resolves in ~2 weeks for mostFrom trials of women with bleeding and open sores. Take the reassurance
Wait ~2 hours per drink if being cautiousAnd no need to pump and dump at all

Where the evidence runs out#

  • The best treatment for most kinds of nipple pain.
  • Which tongue ties are functionally significant versus incidentally present.
  • The most effective pumping schedule, or any non-drug way to reliably raise supply.
  • Why some women respond well to pumps and others get almost nothing.

Our family's notes#

  • Lactation support contacts (and a backup):
  • Signs we'll use to judge that feeding is working:
  • Our supplementation plan, decided in advance:
  • Pumping and return-to-work plan:
  • The pain or supply threshold at which we call for help:

Find a note

Guidance & evidence

The notes separate an action's strength from confidence in its evidence. These are the study guide's labels, not an independent clinical assessment.

MUST
Safety-critical or medically urgent.
Evidence leans clearly one way.
OPTION
Your circumstances and preferences decide.
TRY
Low risk; worth a limited experiment.
SKIP
Little value for the effort.
AVOID
Evidence of harm or a risk not worth taking.

High confidence: strong and consistent. Moderate: credible, with real limitations. Low: small, mixed, or indirect evidence.