When pregnancy complications threaten preterm delivery, clinicians routinely counsel expectant mothers about the use of steroids, medications that accelerate fetal lung maturation and reduce life-threatening respiratory complications in premature infants.
Yet clinicians face a conundrum: the same treatment that prevents brain hemorrhages and saves lives when babies are delivered very early may increase neurodevelopmental risks in the half of pregnancies that reach term.
New evidence from Danish registries, Scottish population studies, and American randomized trials reveals that one critical variable is not whether steroids are given, but when the baby will actually be delivered. The problem? Clinicians currently lack tools to predict which pregnancies truly need intervention.
Small risks, large uncertainties
Laugesen and colleagues tracked more than one million Danish births over 17 years, finding autism spectrum disorder diagnoses reached 6.6% among children exposed to antenatal steroids versus 4.3% in unexposed groups.[1] The increase translates to just 20 additional cases per 1,000 exposures, yet these registry associations cannot prove causation. The health of women receiving steroids often fundamentally differs from that of those who do not, and despite sophisticated statistical adjustments, residual confounding by disease severity remains possible.[1]
The gestational age paradox
The picture clarifies when researchers examine actual delivery timing rather than exposure alone. Ninan's meta-analysis of 1.25 million children revealed steroids reduce neurodevelopmental impairment by 31% when babies deliver early (before 28 weeks), precisely the population who benefits most from lung maturation.[4] Yet the same exposure is associated with increased neurocognitive concerns when delivery occurs slightly preterm or at term.
Frier's team confirmed these patterns in 285,637 Scottish children, finding protective effects at 28 to 33 weeks and potential harm at 37 to 38 weeks.[5] The research exposes a clinical dilemma: approximately half of pregnancies receiving steroids for the threat of preterm delivery ultimately reach full term, exposing substantial populations to potential neurodevelopmental risk without respiratory benefit.[4]
Randomized evidence offers partial reassurance
The ALPS trial followed children who received betamethasone at 34 to 36 weeks, finding no increase in autistic traits at age seven.[2] This carefully controlled study provides stronger causal evidence than registry associations, but its applicability is limited to late preterm administration with single-course exposure. The study did not address earlier gestational exposures, repeat courses, or different molecules, and it cannot resolve the fundamental problem that clinicians cannot predict at the time of administration whether a pregnancy will deliver days or weeks later.
Molecules matter, evidence remains incomplete
Postnatal steroid research reveals profound differences between molecules. Puia-Dumitrescu documented adverse effects on cognitive scores with dexamethasone exposure, while prednisolone and methylprednisolone showed no neurodevelopmental associations.[7] Watterberg's hydrocortisone trial found neither benefit nor harm for long-term neurodevelopment.[6] These molecular distinctions suggest that pharmacological characteristics beyond anti-inflammatory potency determine neurodevelopmental safety. To date, there have been no systematic prenatal research studies comparing glucocorticoids.
Better prediction, not more associations
The emerging evidence suggests the field needs better prognostic tools more than it needs association studies. If clinicians could accurately identify which pregnancies will deliver within seven days, steroids could be targeted to populations most likely to benefit. Currently, prediction remains imprecise, making overtreatment and unnecessary exposure an unavoidable risk.
Research gaps extend beyond prediction. Long-term follow-up rarely extends past early childhood. Systematic molecular comparisons for prenatal use are absent. Dose-response relationships, although emerging for postnatal use, lack comparable evidence for prenatal use. The absolute risks remain small, the relative risks modest, and the causal evidence incomplete. Yet the population-level exposure is substantial, and the long-term consequences of widespread use without precise targeting remain unknown.
References
- Laugesen K, Skajaa N, Petersen I, et al. Mental disorders among offspring prenatally exposed to systemic glucocorticoids. JAMA Netw Open. 2025;8(1):e2453245.
- Gyamfi-Bannerman C, Clifton RG, Tita ATN, et al. Neurodevelopmental outcomes after late preterm antenatal steroids: the ALPS follow-up study. JAMA. 2024;331(19):1629-1637.
- Brito AR, Vairo GPT, Dias APBH, Olej B, Nascimento OJM, Vasconcelos MM. Effect of prednisolone on language function in children with autistic spectrum disorder: a randomized clinical trial. J Pediatr (Rio J). 2021;97(1):22-29.
- Ninan K, Liyanage SK, Murphy KE, Asztalos EV, McDonald SD. Evaluation of long-term outcomes associated with preterm exposure to antenatal corticosteroids: a systematic review and meta-analysis. JAMA Pediatr. 2022;176(6):e220483.
- Frier EM, Guthrie CAG, Diaz-Gomez D, et al. Associations of antenatal corticosteroids with neurodevelopment in children aged 27-30 months: a population-based cohort study. BJOG. 2025;132(3):352-362.
- Watterberg KL, Walsh MC, Li L, et al. Hydrocortisone to improve survival without bronchopulmonary dysplasia. N Engl J Med. 2022;386(12):1121-1131.
- Puia-Dumitrescu M, Smith PB, Zhao J, et al. Dexamethasone, prednisolone, and methylprednisolone use and 2-year neurodevelopmental outcomes in extremely preterm infants. JAMA Netw Open. 2022;5(3):e220947.