Yet, no studies to date have evaluated the patterns of OCS usage during pregnancy including the indications for which they are prescribed. Thus, we aimed to describe the OCS utilization patterns in pregnant women using a large nationwide database in South Korea.

We conducted a retrospective observational study using the Health Insurance Review and Assessment (HIRA) database of South Korea from 2009 to 2022. The HIRA database, which covers almost the entire South Korean population, contains data on patient’s demographics and healthcare claims record (e.g., diagnoses, prescription records, and medical procedures) from both in- and outpatient settings. Using this nationally representative database, we identified all pregnancies with delivery records defined based on HIRA procedure codes (Supplementary Table 1) between January 1, 2010, and December 31, 2021. Pregnancies resulting in non-live births were not included in the analysis. The last menstrual period (LMP) was estimated based on the delivery date and diagnoses of preterm birth, using a validated algorithm in administrative healthcare database.4 Specifically, LMP was estimated by subtracting 245 days from the delivery date if the mother had a diagnostic code for preterm delivery (O42, O60.1, O60.3), and by subtracting 273 days for otherwise full-term delivery.

We first described the baseline characteristics of the pregnancies according to OCS exposure during pregnancy. OCS exposed pregnancies were defined as pregnancies with ≥ 1 OCS prescription from LMP to the day before the delivery date and unexposed pregnancies were defined as those without OCS prescription from LMP to the day before the delivery date. In each pregnancy, we assessed maternal age, insurance type, and obstetric conditions (e.g., nulliparity, multifetal pregnancies) at delivery. We also evaluated the indication-related conditions (e.g., autoimmune inflammatory disease, skin disease, respiratory system disease), comorbidities, and comedications from 180 days before LMP to the day before the date of delivery and healthcare utilization within 180 days prior to LMP. Baseline characteristics are presented as frequencies and percentages for categorical variables and as means and standard deviations for continuous variables. Absolute standardized differences (aSD) were calculated to quantify the differences between the OCS-exposed and unexposed pregnancies.

In addition, we described the annual prevalence of OCS exposure during pregnancy from 2010 to 2021. We further analyzed the prevalence by individual OCS. As prednisolone and methylprednisolone accounted for majority of the OCS usage (> 90%), we grouped the OCS as: 1) prednisolone, 2) methylprednisolone, and 3) others (dexamethasone, betamethasone, deflazacort, hydrocortisone and triamcinolone). We also calculated the number of OCS prescriptions during five different assessment windows: 1) pre-pregnancy period (one year before the LMP), 2) first trimester (LMP to day 90 of gestation), 3) second trimester (day 90 to day 180), 4) third trimester (day 180 to delivery), and 5) postpartum period (1 year after delivery).

Lastly, we evaluated the annual proportion of individual indication among OCS exposed pregnancies. Indications were identified based on the primary International Classification of Diseases, Tenth Revision (ICD-10) diagnosis of OCS prescription and were categorized into six groups: autoimmune inflammatory disease, asthma, diseases of the skin, diseases of the respiratory system, pregnancy-associated condition, and others. Disease of the respiratory system included all subcodes of ICD-10 Chapter J, while pregnancy-associated conditions included in vitro fertilization, female infertility, threatened abortion, and subcodes of ICD-10 code “Z3,” which are health service encounters related to reproduction. A detailed list of indications and the corresponding diagnostic codes for each indication is provided in Supplementary Table 2. We also assessed the proportion of each OCS indication among OCS exposed pregnancies (or women) during pre-pregnancy, first trimester, second trimester, third trimester and postpartum period. All analyses were performed using SAS Enterprise Guide 7.1 software (SAS Institute, Cary, NC, USA).

We identified a total of 4,574,294 pregnancies, of which 283,001 (6.2%) have been exposed to OCS during pregnancy. The median (interquartile range) duration of OCS prescription during pregnancy was 4.0 (3.0–6.0) days. Compared with unexposed pregnancies, OCS exposed pregnancies were older, more likely to have comorbidities (e.g., gastrointestinal diseases and hypertension), and had a higher number of diagnoses (distinct 3-digit ICD-10 codes) recorded (6.79 vs. 4.84, aSD 0.46) (Table 1). As expected, the prevalence of indication-related conditions was significantly higher in the OCS-exposed group.

During the study period, the prevalence of OCS exposure notably increased from 4.98% in 2010 to 6.65% in 2021. Among individuals younger than 35 years, the prevalence rose from 4.89% in 2010 to 5.77% in 2021. In those aged 35 years and older, the prevalence increased from 5.48% in 2010 to 8.34% in 2021 (Supplementary Table 3). Methylprednisolone (45.2%) and prednisolone (45.0%) accounted for the majority of OCS usage (Fig. 1A). Across the pregnancy period, OCS prescriptions sharply declined during the first trimester and gradually increased during the third trimester, specifically near the delivery date (Fig. 1B). Among all pregnancies, 230,274 (5.03%) were exposed during the first trimester, followed by 37,861 (0.83%) in the second trimester and 35,130 (0.77%) in the third trimester.

Among OCS exposed pregnancies, respiratory (40.48%) and skin (22.17%) diseases were the most prevalent indications (Fig. 2A). The proportion of pregnancy-associated conditions showed a notable increase from 9.31% in 2017 to 34.87% in 2021. The proportion of OCS usage for diseases of respiratory system increased from 12.80% in 2010 to 21.88% in 2019, but subsequently declined, with 14.26% in 2021 (Fig. 2A). Rheumatoid arthritis (0.72%) and systemic lupus erythematosus (0.60%) were the most prevalent indication among autoimmune inflammatory diseases, and dermatitis (15.57%) and urticaria (5.70%) accounted the high proportions among diseases of the skin. Acute upper respiratory tract infections (15.74%) predominated among diseases of the respiratory system, and within the pregnancy associated conditions, treatment of infertility (5.94%) had high proportions. Majority of the OCS prescriptions were used for acute conditions such as skin and respiratory diseases both during pregnancy and pre-pregnancy (Fig. 2B). The proportions of autoimmune diseases and asthma were relatively low but increased after gestation (Fig. 2B).

In this nationwide population-based study that comprises all pregnancies with live births from 2010 to 2021, the prevalence of OCS exposure during pregnancy was 6.2%. This proportion was higher than that reported in previous studies conducted in the United States and Australia.56 Throughout the study period, methylprednisolone and prednisolone were the most frequently used OCS during pregnancy, possibly due to their limited placenta transfer.7 We also observed a large decrease in OCS prescriptions after the time of gestation. This finding would indicate that many women tend to discontinue OCS after recognizing that they are pregnant, which would also be attributed to concerns on the safety of prenatal OCS use. Of note, most of the OCS prescriptions were used for acute conditions during pregnancy and the median duration of prescription was relatively short (4 days). In South Korea, OCS prescriptions for infertility treatment have been covered by health insurance from 2017, which coincides with the increase in OCS prescriptions for pregnancy-associated conditions after 2017.8 The prevalence of OCS prescriptions during pregnancy generally increased over the study period, with a more pronounced rise in older women. However, there was a slight decrease in 2021, likely driven by fewer prescriptions for respiratory system diseases. This coincides with the implementation of social distancing measures during the coronavirus disease 2019 pandemic, which resulted in reduced respiratory infection rates.910

This study has several limitations. We defined the OCS usage based on prescriptions, which may not accurately reflect actual exposure. In addition, although we identified the indications via the primary diagnoses of OCS prescriptions, it may not align with the actual indications. Lastly, because the HIRA database does not contain information on the start of pregnancy or LMP, we estimated these based on delivery dates and preterm birth diagnostic codes. While the algorithm we used has demonstrated a high positive predictive value for determining gestational age in administrative databases, there remains some potential for misclassification, and further validation of this algorithm is needed within the HIRA database.

Overall, our findings highlight the common usage of OCS during pregnancy for various indications. Further studies should be conducted to confirm the safety of OCS use in pregnant populations.