A Mineral That Falls Fast in the Smallest Babies

What a Norwegian study of 120 very premature infants found about low blood phosphate in the first week, and about using a computer model to predict it

The Short Version

Very premature babies often develop a low level of phosphate — a mineral essential for growth — in their first week of life, and a Norwegian study found that one in three did. Babies who were unusually small at birth for their stage of pregnancy were affected much earlier and much more often. Researchers also tested whether a computer model could estimate phosphate from tests already being done, and found it promising but not yet ready for real care.

Why Phosphate Matters So Much, So Early

The study behind these findings comes from Oslo University Hospital in Norway, and it is worth understanding what problem it was trying to solve [1]. Phosphate is one of the minerals the body uses to build bone and to power the chemistry inside every cell. A baby normally receives most of its lifetime supply of phosphate and calcium during the last three months of pregnancy, when the placenta actively pumps these minerals across to the growing fetus. A baby born before 29 weeks of pregnancy simply never gets that delivery, which is why premature babies need unusually mineral-rich nutrition after birth [2].

The trouble begins when feeding starts. For years, neonatal units have given very premature babies protein and energy quickly after birth, because babies who do not grow well in intensive care tend to do worse later. But a randomised study found that this rapid feeding approach came with a cost: babies developed disturbances in their blood salts and minerals, and more infections [3]. What was happening is now well described. A baby who has been short of nutrition in the womb, then suddenly receives plenty, switches into rapid building mode — and phosphate is pulled out of the blood and into the cells to fuel that building. Doctors call this pattern refeeding syndrome, borrowing a term first used for starved adults; it was recognised in very small growth-restricted newborns [4] and later described in detail for newborn intensive care [5]. Researchers also showed that the amount of protein given in the first days directly influenced how phosphate and calcium behaved [6].

Not every baby is equally at risk. Babies who are small for gestational age — meaning smaller at birth than 90 out of 100 babies born at the same stage of pregnancy — were found to be especially vulnerable to early low phosphate and low potassium [7]. This makes biological sense, because being small for gestational age often reflects a placenta that has been supplying too little for weeks before birth, which is why specialists have worked to define growth restriction carefully rather than treating all small babies alike [8]. And this is not simply a number on a laboratory report. Analyses from a large nutrition trial linked early low phosphate to worse outcomes during the hospital stay [9], and a follow-up looked at how these babies were doing at two years of age [10].

Which leaves clinicians in an uncomfortable position. The only reliable way to know a baby's phosphate level is a blood test — and it is not a small one. Measuring phosphate needs a separate blood sample of about 125–250 microlitres, while the routine blood gas test that units already run several times a day needs only around 25 microlitres and already reports calcium and potassium [1]. For a baby weighing well under a kilogram — the babies in this study weighed 798 g on average at birth — repeated testing is a genuine contributor to anaemia and to needing transfusions [11]. Before this study, doctors had tried to work out which babies were most likely to have low phosphate from clinical features alone [12], but nobody had tried to have a computer estimate the actual level.

What the Researchers Did

The babies in this study came from the ImNuT trial (a study of fatty-acid supplements in very premature babies; the name is a coined short form and the researchers do not say what its individual letters stand for). It ran at a single hospital, Oslo University Hospital in Norway, recruiting between April 2018 and January 2021 [13]. All 120 babies were born before 29 weeks of pregnancy and were fed according to one standard nutrition plan for their first ten days, a plan already shown to deliver the recommended amounts of nutrients [14].

Nothing new was given to any baby, and no treatment was tested. This was a look back at information already collected. Of the 120 babies, 119 had phosphate measured during their first week, producing 607 test results with an average level of 1.81 mmol/L. Alongside these were 1,960 routine blood gas tests — between 2 and 42 per baby, with a typical baby having 16. On average the babies received 40.0 mg of phosphate per kilogram per day, ranging from 18.7 to 66.0.

The researchers then asked a computer to learn the relationship between the routine information and the phosphate result. They tried six different types of model and three ways of preparing the data, and tested each on three questions: what is the phosphate level right now, what will it be in 12 hours, and what will it be in 24 hours. Deliberately, previous phosphate results were left out, so the model had to work from ordinary data only. Finally, the models were tested on a completely separate group of 40 very premature babies who had nothing to do with the original study — a fair test of whether the models work outside the place they were built [1].

What They Found

The first finding is the most important, and it needs no computer at all. Using a cut-off of 1.4 mmol/L, 40 of the 119 babies — 33.6%, one in three — had a low phosphate level at some point during their first week. And timing differed sharply by group. Among the small-for-gestational-age babies, some fell below the threshold within the first hours after birth, and by around five days only about one in five had escaped a low reading. Among appropriately grown babies, the earliest cases appeared at around 80 hours, and the numbers rose far more slowly. This difference was very strong statistically.

One result looked backwards at first glance: babies born at 26 weeks or later seemed to have earlier and more frequent low readings than the very youngest. The explanation is that small-for-gestational-age babies happened to be concentrated in that slightly older group, and being small for gestational age mattered more than being extremely early. The difference between the age groups was in any case borderline and not clearly meaningful.

The computer models did reasonably well, and interestingly the simplest one worked best when moved to the outside group of babies. It explained roughly half of the variation in phosphate levels for the "right now" and "12 hours ahead" questions, both at home and in the outside group. The "24 hours ahead" prediction, however, fell away noticeably when tested elsewhere: in the outside group it accounted for only about a quarter of the variation (a statistic called R² fell to 0.265), so the furthest-ahead forecast is also the least dependable away from the hospital where it was built.

Where the honest limits show is in using the model as a yes-or-no test. It was very good at catching babies who did turn out to have low phosphate — around 83 to 93 out of every 100 of them. But it was poor at correctly identifying babies whose phosphate was fine, raising the alarm for many babies who were not actually affected. Most importantly, when the model said "this baby looks fine right now", it was right only about half the time in the outside group. A reassuring result from the model is therefore not a reason to skip the blood test.

What This Means for Families, and What Comes Next

If your baby was born very early, low phosphate in the first week is common, expected, and specifically looked for. It is not a complication that appears out of nowhere; it is a known consequence of a very small baby beginning to grow, and it is treated by adjusting the mineral content of the nutrition your baby receives. Units that have made this a focus have measurably reduced how often and how long it happens [15]. If your baby was small for their gestational age, the team may check phosphate earlier and more often, and this study is a good reason for that.

It is also fair to ask why so much blood is being taken. The answer, in part, is that the test that matters here needs several times more blood than the routine one — which is exactly the problem this research is trying to chip away at.

What this study does not do is change care today. It looked back at records rather than testing anything forward, it took place in one hospital using one feeding plan, and the outside group of babies was small and somewhat different. The researchers themselves say clearly that a proper forward-looking study is needed before any such model could be used in a real unit. What researchers are working on next is that prospective test, in several hospitals with different feeding practices, and the separate question of whether babies who are small at birth should simply be given more phosphate from day one [16]. For now, the practical value of this work is knowledge, not software: it tells the team which babies to watch, and how early to start watching.

References

  1. Westvik ÅS, Tomic O, Tscherning C, Moltu SJ. Predicting serum phosphate levels in very preterm infants using machine learning. Frontiers in Pediatrics. 2026;14:1881474. doi:10.3389/fped.2026.1881474 ↩
  2. Embleton ND, Moltu SJ, Lapillonne A, van den Akker CHP, Carnielli V, Fusch C, et al. Enteral nutrition in preterm infants (2022): a position paper from the ESPGHAN committee on nutrition and invited experts. J Pediatr Gastroenterol Nutr. 2023;76(2):248–68. doi:10.1097/MPG.0000000000003642 ↩
  3. Moltu SJ, Strømmen K, Blakstad EW, Almaas AN, Westerberg AC, Brække K, et al. Enhanced feeding in very-low-birth-weight infants may cause electrolyte disturbances and septicemia — a randomized, controlled trial. Clin Nutr. 2013;32(2):207–12. doi:10.1016/j.clnu.2012.09.004 ↩
  4. Ross JR, Finch C, Ebeling M, Taylor SN. Refeeding syndrome in very-low-birth-weight intrauterine growth-restricted neonates. J Perinatol. 2013;33(9):717–20. doi:10.1038/jp.2013.28 ↩
  5. Bradford CV, Cober MP, Miller JL. Refeeding syndrome in the neonatal intensive care unit. J Pediatr Pharmacol Ther. 2021;26(8):771–82. doi:10.5863/1551-6776-26.8.771 ↩
  6. Bonsante F, Iacobelli S, Latorre G, Rigo J, De Felice C, Robillard PY, et al. Initial amino acid intake influences phosphorus and calcium homeostasis in preterm infants — it is time to change the composition of the early parenteral nutrition. PLoS One. 2013;8(8):e72880. doi:10.1371/journal.pone.0072880 ↩
  7. Boubred F, Herlenius E, Bartocci M, Jonsson B, Vanpée M. Extremely preterm infants who are small for gestational age have a high risk of early hypophosphatemia and hypokalemia. Acta Paediatr. 2015;104(11):1077–83. doi:10.1111/apa.13093 ↩
  8. Beune IM, Bloomfield FH, Ganzevoort W, Embleton ND, Rozance PJ, van Wassenaer-Leemhuis AG, et al. Consensus based definition of growth restriction in the newborn. J Pediatr. 2018;196:71–6.e1. doi:10.1016/j.jpeds.2017.12.059 ↩
  9. Cormack BE, Jiang Y, Harding JE, Crowther CA, Bloomfield FH. Neonatal refeeding syndrome and clinical outcome in extremely low-birth-weight babies: secondary cohort analysis from the ProVIDe trial. J Parenter Enteral Nutr. 2021;45(1):65–78. doi:10.1002/jpen.1934 ↩
  10. Ford N, Bloomfield FH, Jiang Y, Cormack BE. Early hypophosphataemia and refeeding syndrome in extremely low birthweight babies and outcomes to 2 years of age: secondary cohort analysis from the ProVIDe trial. Arch Dis Child Fetal Neonatal Ed. 2025;110(2):157–64. doi:10.1136/archdischild-2024-327029 ↩
  11. Counsilman CE, Heeger LE, Tan R, Bekker V, Zwaginga JJ, Te Pas AB, et al. Iatrogenic blood loss in extreme preterm infants due to frequent laboratory tests and procedures. J Matern Fetal Neonatal Med. 2021;34(16):2660–5. doi:10.1080/14767058.2019.1670800 ↩
  12. Yakubovich D, Strauss T, Ohana D, Taran C, Snapiri O, Karol DL, et al. Factors associated with early phosphate levels in preterm infants. Eur J Pediatr. 2020;179(10):1529–36. doi:10.1007/s00431-020-03622-7 ↩
  13. Wendel K, Pfeiffer HCV, Fugelseth DM, Nestaas E, Domellöf M, Skålhegg BS, et al. Effects of nutrition therapy on growth, inflammation and metabolism in immature infants: a study protocol of a double-blind randomized controlled trial (ImNuT). BMC Pediatr. 2021;21(1):19. doi:10.1186/s12887-020-02425-x ↩
  14. Rossholt ME, Bratlie M, Wendel K, Aas MF, Gunnarsdottir G, Fugelseth D, et al. A standardized feeding protocol ensured recommended nutrient intakes and prevented growth faltering in preterm infants <29 weeks gestation. Clin Nutr ESPEN. 2023;53:251–9. doi:10.1016/j.clnesp.2022.12.024 ↩
  15. Lair CS, Brown LS, Edwards A, Jacob T, Brion LP, Jaleel M. Quality improvement project in a neonatal intensive care unit reduced the prevalence and duration of hypophosphatemia with significant and sustainable results. Nutr Clin Pract. 2023;38(6):1379–91. doi:10.1002/ncp.10986 ↩
  16. Wright TB, Bloomfield FH, Alexander T, Cormack BE. Association between early phosphate intake and refeeding syndrome in extremely low-birth-weight infants: a retrospective cohort study. J Parenter Enteral Nutr. 2025;49(4):460–7. doi:10.1002/jpen.2739 ↩