When a Premature Baby Decides It's Time to Eat

A look at new research following 115 premature babies to age two, comparing feeding on a schedule with feeding on the baby's own cues

Why This Question Matters

Premature babies in intensive care have traditionally been fed set amounts of milk at set times. A hospital in Israel switched to letting babies lead — feeding when they showed they were hungry and stopping when they were full — and then followed 115 children to age two. The babies who led went home almost twelve days earlier, grew normally, and their parents were less likely to be forcing food into them as toddlers.

That is the short version of a study published in Frontiers in Pediatrics in 2026 [1]. The longer version is worth telling, because it touches on something almost every NICU family eventually confronts: the moment when the tube comes out and your baby has to learn to eat.

The Problem That Made This Research Necessary

For a baby born too soon, eating is not a reflex that simply switches on. Sucking, swallowing, and breathing have to happen in a coordinated rhythm, and most premature babies cannot manage all three together until somewhere around 32 to 34 weeks of development — which, for a baby born at 29 weeks, means weeks of milk delivered through a soft tube passed through the nose into the stomach.

Getting off that tube is one of the last hurdles before going home, and for decades hospitals have approached it as a numbers problem. Babies who are growing well need a certain amount of milk and a certain number of calories each day to keep their rapidly developing brains and bodies supplied [2]. The simplest way to guarantee those numbers is to write them into an order: this much milk, this often, whether or not the baby seems interested. Nursing charts were built to record millilitres taken rather than what the baby was doing while taking them [3].

Parents who lived through this era of care will recognise the feeling. The bottle had a line on it. Whether the feed had gone well was decided by whether the baby reached the line. A baby who turned away, or fell asleep, or grew pale and stopped breathing steadily halfway through, was often encouraged to keep going — jiggled, repositioned, the bottle tipped a little more. And the parent standing beside the cot, who could see perfectly well that their baby had had enough, learned that their own reading of their child was not the one that counted.

Starting in the early 2000s, a group of clinicians began arguing that this had the logic backwards. Catherine Shaker described feeding as a conversation rather than a task, in which the baby's signals — rooting, bringing a hand to the mouth, sucking eagerly on a pacifier, or on the other side, turning away, going limp, changing colour — are the actual instructions [4]. Others turned that idea into step-by-step guidance a busy unit could follow [5]. The approach became known as cue-based feeding, and units that tried it reported that babies reached full feeding by mouth sooner [6] and, in one careful study, that this held up when tested prospectively [7].

But a serious worry persisted, and it was never unreasonable. If you let a baby stop when the baby wants to stop, does the baby get enough? An influential review of all the available research concluded that the studies were simply too small and too flawed to say [8]. And every one of them stopped counting the day the baby went home. Nobody had checked whether children fed this way were still doing well a year or two later.

What the Researchers Did

The research team, led by Dr Iris Morag, worked at Shamir Medical Center near Tel Aviv, a large hospital where around 8,000 babies are born each year and where every baby born before 35 weeks goes to the same intensive care unit [1].

In 2020 that unit changed its entire approach to feeding, moving from the traditional scheduled method to cue-based feeding for every baby. That gave the researchers a natural comparison. They looked at 58 babies born under the old approach, between June 2018 and February 2020, and 57 babies born under the new one, between January and September 2021. All were born between 27 and 34 weeks of pregnancy. The months in between, while staff were learning the new method, were left out.

They deliberately excluded babies whose feeding might be complicated by other serious problems — genetic conditions, severe bleeding in the brain, a serious bowel illness called necrotising enterocolitis, or being unusually small at birth for their stage of pregnancy, judged against a widely used growth chart [9]. The point was to look at feeding itself, not at feeding tangled up with other illness.

Then they did the part nobody had done before. In Israel, every child attends national well-child clinics where they are routinely weighed and measured, and every baby born before 36 weeks is automatically referred for nutritional follow-up. Parents shared those records, giving the researchers weights and lengths at roughly 12, 18, and 24 months. Parents also filled in a short, well-tested questionnaire about how mealtimes were going at age two — the Montreal Children's Hospital Feeding Scale, which asks about things like whether the child refuses food, whether meals drag on, and whether the parent ends up having to force the child to eat [10].

It is important to be clear about what cue-based feeding meant here, because it was not simply leaving babies to it. Feeds were offered every two to four hours when the baby seemed ready and stable. If a baby did not take enough by mouth, the difference was topped up through the feeding tube. Babies were weighed every three to four days, blood sugar was checked the first couple of times a baby went longer than three and a half hours between feeds, and if a baby lost weight the feeds were brought closer together or the tube top-ups increased. Breastfeeding directly at the breast was encouraged from about 32 weeks, with bottles introduced a week or two later.

What They Found

Babies went home sooner. Under cue-based feeding, babies were discharged at an average of 35.5 weeks of adjusted age, compared with 36.9 weeks under the old approach. Their total hospital stay was on average 11.8 days shorter — 33 days instead of 45. For a family living out of a hospital car park, nearly two weeks is a great deal of life handed back.

They were smaller in absolute terms — but not for their age. Cue-fed babies weighed about 2,154 grams at discharge, compared with about 2,485 grams. That difference disappears once you account for the fact that they were younger. Measured against what you would expect for their stage of development, the two groups were identical. They were smaller the way a two-year-old is smaller than a three-year-old.

They left breastfeeding more. Cue-fed babies were having more direct breastfeeds per day when they went home.

Growth at two years was normal in both groups — with a caveat. Every average measurement in both groups stayed comfortably within the normal range. However, the cue-fed children sat consistently a little lower on the weight and length charts throughout the two years. Importantly, the gap did not widen over time; the two groups' growth curves ran parallel rather than pulling apart. The researchers could not tell whether this reflects something about the feeding approach, something about the two groups of families, or simply chance, and they say so plainly.

Mealtimes at two years were calmer in the cue-fed group. Overall feeding-problem scores were similar between the groups. But on two specific questions, the difference was clear: parents of children fed on the old schedule were significantly more likely to say they had to force their child to eat, and significantly more likely to say that feeding affected their relationship with their child. It is a striking result — that a few weeks of practice in a hospital, two years earlier, might still be echoing at a kitchen table.

What This Means for Families

If your baby's unit uses cue-based feeding, this research offers reassurance on the question most parents actually have: your baby is not being underfed. The tube stays in as a safety net until it is genuinely not needed, weights are still checked, and babies in this study grew normally through toddlerhood.

If your unit uses the traditional approach, this study is not evidence that your baby is being harmed. Scheduled feeding is safe, it has fed generations of premature babies, and most of those children eat perfectly well. What the research suggests is that there may be a gentler route to the same place, with fewer days in hospital.

The finding about forced feeding is the one worth carrying home. Feeding difficulties are common in children born prematurely and can persist — studies have followed such children to school age and beyond and found eating problems more frequent than in children born at term [11], sometimes alongside other developmental differences [12], and reviews confirm the pattern extends into how families eat together [13]. Nutrition experts now recommend responsive feeding — watching the child, offering food, and respecting the signal that they have finished — as the general approach for young children, not just premature ones [14]. If your child was fed on a schedule in hospital, none of that is fixed. The habit of reading your child rather than the plate can be learned at home.

One practical note: because the cue-fed children in this study sat slightly lower on the growth charts, it is worth keeping post-discharge growth appointments even when everything seems fine. In Israel this follow-up happens automatically for every premature baby. Elsewhere it may need to be asked for.

What Researchers Are Working On Next

This study cannot prove cause and effect. It compared two periods of time rather than randomly assigning babies to one approach or the other, so other things that changed between 2018 and 2021 — including the pandemic, which affected how much time parents could spend at the cot — might explain part of what was seen. The mealtime questionnaire also relied on parents remembering back, and the families in the older group were remembering from further back.

The researchers are explicit that the next step is a proper randomised trial across several hospitals, comparing the two approaches head to head with growth as a formally tested outcome. Other open questions include whether formal checklists for judging feeding readiness help — this unit deliberately did not use one, relying on parents' and nurses' judgement instead [15] — and whether adding structured nutrition follow-up after discharge closes the small growth gap while keeping the calmer mealtimes.

For now, the message from this work is a modest but genuinely hopeful one. Letting a premature baby take the lead at the bottle or breast appears to get them home sooner, does not appear to cost them their growth, and may leave the family with an easier relationship around food years later — provided someone is still keeping an eye on the scales.

References

  1. Morag I, Kogan D, Weinberg E, Darsa D, Derin O, Nissimov S, Yackobovitch-Gavan M. Cue-based feeding and growth in preterm infants: a two-year follow-up study. Frontiers in Pediatrics. 2026;14:1771033. doi:10.3389/fped.2026.1771033 ↩
  2. Dutta S, Singh B, Chessell L, Wilson J, Janes M, McDonald K, et al. Guidelines for feeding very low birthweight infants. Nutrients. 2015;7(1):423–42. doi:10.3390/nu7010423 ↩
  3. Ludwig SM, Waitzman KA. Changing feeding documentation to reflect infant-driven feeding practice. Newborn and Infant Nursing Reviews. 2007;7(3):155–60. doi:10.1053/j.nainr.2007.06.007 ↩
  4. Shaker CS. Cue-based feeding in the NICU: using the infant's communication as a guide. Neonatal Network. 2013;32(6):404–8. doi:10.1891/0730-0832.32.6.404 ↩
  5. Lubbe W. Clinicians guide for cue-based transition to oral feeding in preterm infants: an easy-to-use clinical guide. Journal of Evaluation in Clinical Practice. 2018;24(1):80–8. doi:10.1111/jep.12721 ↩
  6. Kirk AT, Alder SC, King JD. Cue-based oral feeding clinical pathway results in earlier attainment of full oral feeding in premature infants. Journal of Perinatology. 2007;27(9):572–8. doi:10.1038/sj.jp.7211791 ↩
  7. Puckett B, Grover VK, Holt T, Sankaran K. Cue-based feeding for preterm infants: a prospective trial. American Journal of Perinatology. 2008;25(10):623–8. doi:10.1055/s-0028-1090583 ↩
  8. Watson J, McGuire W. Responsive versus scheduled feeding for preterm infants. Cochrane Database of Systematic Reviews. 2016;2016(8):CD005255. doi:10.1002/14651858.CD005255.pub5 ↩
  9. Fenton TR, Kim JH. A systematic review and meta-analysis to revise the Fenton growth chart for preterm infants. BMC Pediatrics. 2013;13:59. doi:10.1186/1471-2431-13-59 ↩
  10. Ramsay M, Martel C, Porporino M, Zygmuntowicz C. The Montreal Children's Hospital Feeding Scale: a brief bilingual screening tool for identifying feeding problems. Paediatrics & Child Health. 2011;16(3):147. doi:10.1093/pch/16.3.147 ↩
  11. Samara M, Johnson S, Lamberts K, Marlow N, Wolke D. Eating problems at age 6 years in a whole population sample of extremely preterm children. Developmental Medicine & Child Neurology. 2010;52(2):e16–22. doi:10.1111/j.1469-8749.2009.03512.x ↩
  12. Crapnell TL, Woodward LJ, Rogers CE, Inder TE, Pineda RG. Neurodevelopmental profile, growth, and psychosocial environment of preterm infants with difficult feeding behavior at age 2 years. The Journal of Pediatrics. 2015;167(6):1347–53. doi:10.1016/j.jpeds.2015.09.022 ↩
  13. Walton K, Daniel AI, Mahood Q, Vaz S, Law N, Unger SL, et al. Eating behaviors, caregiver feeding interactions, and dietary patterns of children born preterm: a systematic review and meta-analysis. Advances in Nutrition. 2022;13(3):875–912. doi:10.1093/advances/nmac017 ↩
  14. Pérez-Escamilla R, Segura-Pérez S, Lott M. Feeding guidelines for infants and young toddlers: a responsive parenting approach. Nutrition Today. 2017;52:223–31. doi:10.1097/NT.0000000000000234 ↩
  15. Thoyre SM, Shaker CS, Pridham KF. The early feeding skills assessment for preterm infants. Neonatal Network. 2005;24(3):7–16. doi:10.1891/0730-0832.24.3.7 ↩