Antenatal Corticosteroid and Risk of Fractures in Children: A Nationwide Register Study

Antenatal corticosteroids (ACs) improve the outcomes of preterm neonates [1]. However, in recent years, large population-based studies have also associated AC treatment with later harms. These harms include, for example, an increased risk for infectious diseases, neurosensory disorders, mental disorders, and delayed psychological development [2]. Most of these risks have not been confirmed in randomised trials, such as a recent 50-year follow-up study after exposure to AC or placebo [3]. However, an increased risk of osteoporotic fractures was seen after exposure to AC in comparison to placebo [3]. Since corticosteroids have a direct impact on bone mineralization as they impact both osteoclast and osteoblast functions, we aimed to investigate the risk of fractures close to the AC exposure in early childhood in a large population-based cohort. For this nationwide register-based cohort study, we included all neonates born alive in Finland between 2005 and 2017 from the Finnish Medical Birth Register. It includes information on all pregnancies ending in birth after gestational week 21 + 6. We excluded neonates who died within the first 28 days of life. Fractures were included from the Finnish Care Register of Healthcare until December 2017. This register includes all visits to public secondary and tertiary level care in Finland. The complete list of diagnoses included has been published in our previous report [4]. We compared the cumulative incidence of fractures between antenatally exposed and unexposed children using Kaplan–Meier analysis. The indication for AC administration in Finland has been imminent preterm birth and thus, the term born AC-exposed neonates had received AC due to the risk of preterm birth, which itself did not occur. Cox proportional hazards regression was used to calculate hazard ratios (HRs) for the first fracture event. Follow-up ended at the first occurrence of: a fracture event, the fifth birthday, death, or the study end date (December 31, 2017). Non-adjusted analyses were stratified between preterm (GW < 37 + 0) and term (GW ≥ 37 + 0). Gestational week was included as a continuous variable to adjust the Cox model. Proportional hazards assumptions were checked by visual inspection and were not violated [5]. All analyses were performed with SPSS v29.0 statistical software. A total of 704 905 children were included, of whom 17 688 (2.5%) were exposed to ACs. Of the exposed children, 10 916 (61.7%) were born preterm, and 6772 (38.3%) were born at term (Table 1). Overall, the cohort included 14 384 children (649 born preterm < 37 + 0 gestational weeks) who had at least one fracture during the study period. In preterm children, the cumulative incidence of the first fractures was nearly similar regardless of AC exposure (Figure 1A). Among children born at term, the cumulative incidence appeared to be marginally higher during the first 5 years of life in the corticosteroid-exposed group (Figure 1B); however, this difference was minimal. The crude HR for fracture was 1.02 (CI 0.86–1.21) among children born preterm, and 1.10 (CI 0.93–1.30) among children born at term, CIs including a possible increase of 30% in the fracture risk. Gestational age-adjusted HR was 1.03 (CI 0.91–1.16) for AC. Based on the findings of this large population-based study, exposure to AC was not associated with an increased incidence of bone fractures in children under 5 years and born preterm. In children born at term, the risk was marginally higher but not statistically significant, indicating that an increased risk of up to 30% for fractures cannot be excluded. The main limitation of the present study is that the use of AC was only binary information in the medical birth register. Thus, the information on the exact dates and doses was not available. This introduces a clear possibility for uncontrollable confounding to the analyses, which should be interpreted with caution. The second limitation was the missing information on fractures treated in public primary care and private clinics. However, these fractures are typically treated in specialised public care in Finland in this patient population. There was a notable difference in the size of the exposed and unexposed cohorts, which may cause bias in the estimates. The final limitation was the lack of information on potential confounders (maternal smoking and maternal socioeconomic status). The recent RCT on long-term harms of antenatal ACs indicated a long-term elevated risk of fractures after exposure [3]. It could be hypothesized that the impact of AC would not be seen as an increased fracture risk during the metabolically active bone growth period in childhood, but it may reduce the peak bone mass or lead to earlier bone reduction in bone density. Together with the observations of the present study, further studies are essential to clarify the possible long-term effects of ACs on bone health, especially in children born at term. I.K. and U.S. had the original idea. R.S. and U.S. provided the data. I.K. was in charge of the statistics, and R.S. consulted on the statistical choices and methods. I.K. wrote the original draft. R.S. and U.S. commented on and revised the manuscript critically for important content. All authors have approved the final version to be submitted. This study adheres to STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidance. Open access publishing facilitated by Ita-Suomen yliopisto, as part of the Wiley - FinELib agreement. This study was granted research permissions by the Finnish Institute of Health and Welfare to gain access both to the Medical Birth Register and the Care Register for Healthcare. Statistics Finland granted research permission to use data on mortality. Ethical committee evaluation was waived for this study by the Northern Savonia Healthcare Municipality Ethical Committee, as this study was completely based on routinely collected healthcare registers. The authors declare no conflicts of interest. Individual patient level data can't be made available due to the restrictions of data handling in Finland. Aggregated additional information is available from the corresponding author upon reasonable request.