Abdominopelvic computed tomography during pregnancy and the risk of congenital malformations: protocol for a nationwide population-based cohort study in South Korea

Abstract

Introduction

The use of CT has increased markedly over recent decades, and this trend is also observed among pregnant women. According to current clinical guidelines, fetal radiation exposure below a threshold of 100 to 200 mGy does not appear to increase the risk of congenital malformations. Although the estimated fetal dose from a maternal abdominopelvic CT examination is generally lower than this range, evidence regarding the risks associated with such relatively low-dose diagnostic exposures during pregnancy remains scarce. This study aims to evaluate the association between abdominopelvic CT exposure during the first trimester of pregnancy and the risk of congenital malformations in infants.

Methods and analysis

We will conduct a nationwide population-based cohort study using the National Health Insurance Service (NHIS) database of South Korea. All live births from 2011 through 2023 will be identified and linked to their mothers. Exposure will be defined as maternal abdominopelvic CT during the first trimester of pregnancy. The primary outcome is major congenital malformations in infants, defined according to the standardised classification system of the European Surveillance of Congenital Anomalies. Secondary outcomes include organ-specific malformations and congenital malformations requiring neonatal intensive care unit admission or corrective surgery. A propensity score-based fine stratification weighting approach will be used to adjust for covariates as potential confounders, and relative risks will be estimated. Prespecified sensitivity analyses will be conducted to assess robustness of the findings.

Ethics and dissemination

This study has been approved by the Institutional Review Board of Seoul National University Bundang Hospital (Institutional Review Board No. X-2508-993-902). Informed consent was waived because only anonymised administrative claims data will be used. All analyses will be conducted within the secure NHIS research environment, and no individually identifiable data will be released. The study findings will be disseminated through peer-reviewed publications, scientific conference presentations and communication with regulatory authorities, clinicians and policymakers.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study is a nationwide population-based cohort including all live births in South Korea from 2011 through 2023.

• Various potential biases, including residual confounding, exposure misclassification and selection bias, will be addressed using multiple sensitivity analyses and a quantitative bias analysis.

• The use of the standardised European Surveillance of Congenital Anomalies classification enables a comprehensive and comparable assessment of major congenital malformations.

• The database does not contain clinical details such as imaging findings or actual fetal radiation dose.

Introduction

Over the past several decades, the use of CT has risen substantially,¹ largely attributable to advancements in imaging technology, broader clinical accessibility and defensive medical decision-making. A similar pattern is seen among pregnant individuals; for example, CT utilisation during pregnancy in North America increased by approximately 4.0-fold between 1996 and 2016, demonstrating a notable increase in diagnostic radiation exposure in this population.² Current clinical guidelines state that the use of diagnostic ionising radiation such as CT during pregnancy is justified only when the expected maternal benefit clearly outweighs the potential fetal risks and no suitable diagnostic alternative exists.³,¹⁰

Regarding the risk of congenital malformations, previous literature reviews have suggested that radiation exposure during pregnancy below a threshold of 100 to 200 mGy may not increase the risk, even when exposure occurs during the period of organogenesis.¹¹,¹³ However, the supporting evidence remains surprisingly limited, in part because such exposures and outcomes are inherently difficult to assess. Previous human epidemiologic studies seem outdated in the current medical context or lacked methodological rigour. First, these studies were conducted before the widespread adoption of CT and therefore focused primarily on diagnostic radiation exposures from conventional radiography. In contrast, CT has now become the major source of fetal radiation exposure, while other forms of diagnostic radiation have decreased among pregnant women in developed countries.² Second, most prior studies were case series or case–control analyses with small numbers of participants, limiting the ability to draw definitive conclusions. Third, the outcomes assessed in these studies were largely restricted to externally apparent congenital malformations, such as brain or skeletal malformations. Data on non-apparent malformations, such as cardiac septal defects, are almost entirely lacking, likely reflecting diagnostic limitations in earlier eras. As a result, whether such malformations occurred remains uncertain.

Abdominopelvic CT is of particular clinical concern because it is one of the most frequently performed CT examinations in women of childbearing age^{1 14} and delivers substantially higher fetal doses (approximately 10–50 mGy) compared with chest or brain CT (<1 mGy) or plain radiographs. Notably, the threshold dose for congenital malformations has been extrapolated primarily from experimental studies in mammals, mainly rodents, whereas corresponding human data remain scarce. A large epidemiologic study would therefore be valuable in determining whether abdominopelvic CT during pregnancy carries an actual risk of congenital malformations. This study aims to evaluate the association between abdominopelvic CT exposure during the first trimester of pregnancy and the risk of congenital malformations in infants.

Methods

Data source

This study will be conducted using the National Health Insurance Service (NHIS) database of South Korea, a comprehensive administrative claims database covering the entire population of more than 50 million individuals.¹⁵ The NHIS database contains comprehensive information on insurance eligibility, sociodemographic characteristics, healthcare utilisation, diagnoses and procedures, prescription records, health screening and mortality. Health behaviour variables such as smoking and alcohol consumption are available for individuals who participated in the national health screening programme. Child characteristics (eg, birth weight), obtained from national infant and child health screening records, have been available in the NHIS database since 2007. The medical diagnoses are recorded by physicians using the International Classification of Diseases, 10th revision (ICD-10) at the time of claims for medical procedures. Before any data collection, we registered the study design including the predefined outcomes.¹⁶ No external data linkage beyond the NHIS database is planned. We follow the relevant publication guidelines.¹⁷,¹⁹

Cohort formation and participants

The study population will include all live births in South Korea occurring between 1 January 2011 and 31 December 2023, identified from the NHIS database. Mothers and their children will be linked using the unique insurance identification number shared among family members. Gestational age at delivery is available for approximately 80% of pregnancies as a restricted access variable derived by a data manager at the NHIS based on the expected date of delivery recorded at the first prenatal visit. For pregnancies without recorded gestational age, the beginning of pregnancy will be estimated using validated algorithms: 273 days before delivery for term births and 245 days before delivery for preterm births.²⁰

We will apply the following exclusion criteria. First, we will exclude mother-child pairs with abdominopelvic CT exposure during the second or third trimester of pregnancy. Second, we will exclude mother-child pairs with genetic or teratogenic syndromes, including chromosomal abnormalities and microdeletions. Third, we will exclude mother-child pairs with exposure to known teratogenic medications such as misoprostol, warfarin, ACE inhibitors, systemic retinoids, valproic acid or carbamazepine (figure 1).^{21 22} Cohort entry will be defined as the beginning of pregnancy. Cohort exit will occur at the earliest of the following: diagnosis of any congenital malformation, childhood death or 1 year after birth.

Figure 1. Flow diagram. NHIS, National Health Insurance Service.

Exposure

Mother-child pairs will be classified as the exposed group if the mother has at least one claim record for abdominopelvic CT during the first trimester of pregnancy. Abdominopelvic CT exposure will be identified using predefined procedure codes (HA425, HA435, HA445, HA465, HA475, HA515, HA525, HA535, HA835). The unexposed group will consist of mother-child pairs who did not undergo an abdominopelvic CT examination during pregnancy.

Outcome

The primary outcome is the occurrence of a major congenital malformation in infants, defined according to the standardised classification system of the European Surveillance of Congenital Anomalies (EUROCAT).²³ EUROCAT categorises structural congenital malformations into 12 organ system groups, including malformations of the nervous system, eye, ear-face-neck, heart, respiratory system, orofacial clefts, digestive system, abdominal wall, urinary system, genital organs, limbs and other specified congenital malformations.

The occurrence of a major congenital malformation will be determined when at least one of the following validated criteria is met¹⁶: (1) two or more visits with any organ-specific malformation code recorded as a primary diagnosis within 1 year after delivery, (2) at least one visit with an organ-specific malformation code recorded as a primary diagnosis accompanied by a corresponding malformation-specific surgical or procedural code within 1 year after delivery or (3) at least one visit with an organ-specific malformation code recorded as a primary diagnosis in association with infant death within 1 year after delivery. Congenital malformations categorised as minor malformations in the EUROCAT exclusion list will be excluded from analysis.

Secondary outcomes will include organ-specific congenital malformations, such as congenital cardiac malformations, as defined by EUROCAT and prior studies. Additional secondary outcomes will include congenital malformations requiring neonatal intensive care unit admission or corrective surgery, identified using predefined procedure codes.

Covariates

For the analyses, we will adjust for the following covariates as potential confounders or their proxies: maternal age, year of delivery, child’s birth year, sex of the child, birth weight, healthcare utilisation (eg, number of outpatient visits and hospitalisations and hospital setting of delivery), obstetric conditions (eg, multiple gestation, nulliparity and preterm delivery), socioeconomic indicators (eg, insurance type and region of residence), the Obstetric Comorbidity Index²⁴ and exposure to CT examinations outside the abdominopelvic area, as well as exposure to radiography or nuclear medicine examinations.²⁵

Planned statistical analysis

We aim to evaluate the association between abdominopelvic CT exposure during the first trimester of pregnancy and the risk of congenital malformations using prespecified statistical methods. A propensity score-based fine stratification weighting approach to adjust for measured baseline confounders will be conducted. Covariate balance between exposed and unexposed groups will be assessed using the absolute standardised difference, with values less than 0.1 indicating adequate balance in characteristics between the two groups.²⁶ The primary analysis will estimate relative risk and 95% CI for major congenital malformations using generalised linear regression models, and we will apply robust variance estimators to account for within-woman correlations among women with multiple pregnancies. For secondary outcomes, including organ-specific malformations, multiplicity will be addressed using appropriate adjustment methods such as Bonferroni correction. Missing data patterns will be evaluated for all variables. Missing covariates will be handled using appropriate statistical methods. We will conduct sensitivity analyses to assess the robustness of the primary findings. First, we will restrict the cohort to singleton pregnancies to eliminate potential confounding from multiple gestations. Second, we will restrict the cohort to first pregnancies only to address within-woman correlation among those who contributed multiple pregnancies. Third, to account for potential exposure misclassification, we will redefine the exposure window to early organogenesis (the fourth to eighth weeks of gestation), a period during which the embryo is highly vulnerable to malformations.¹³ Fourth, we will repeat analyses among women who underwent two or more abdominopelvic CT examinations during the first trimester. Fifth, we will restrict the analyses to pregnancies with available gestational age information to reduce misclassification of exposure timing. Sixth, we will conduct a negative-control analysis by shifting the exposure window to the preconception period. A null association in this setting would indirectly suggest that the observed findings are unlikely to be explained by residual confounding. Seventh, we will restrict the cohort to women with available smoking and alcohol consumption data to address residual confounding from health behaviours. Finally, because our cohort is restricted to live births, we will conduct a quantitative bias analysis to evaluate the potential impact of selection bias arising from the exclusion of pregnancies ending in spontaneous abortion, stillbirth or termination. This analysis will be performed using the method proposed by Greenland and Khoury,^{27 28} based on assumptions regarding differential probabilities of inclusion conditional on exposure and outcome.

Ethics and dissemination

This study has received ethics approval from the Seoul National University Bundang Hospital Institutional Review Board (IRB No. X-2508-993-902; approval date: 18 August 2025). The requirement for informed consent has been waived because the analysis will use anonymised administrative claims data. The data are anonymised, and all analyses are conducted within a secure data analysis environment in accordance with NHIS policies. The results of this study will be disseminated through peer-reviewed publications and presentations at scientific conferences. Findings will also be shared with relevant clinical and public health stakeholders to support evidence-based decision-making regarding the use of diagnostic imaging during pregnancy. In accordance with NHIS policies, individual-level data will not be shared externally.

Patient and public involvement

Patients and the public were not directly involved in the design or conduct of this study, primarily because access to the NHIS database is restricted to authorised researchers. However, recurrent clinical concerns regarding fetal radiation exposure, raised by pregnant patients and clinicians in obstetric and radiologic practice, informed the motivation for this study. The findings are expected to support patient-centred decision-making and inform public health policy regarding the use of diagnostic imaging during pregnancy.

Discussion

In this study, we anticipate that the outcomes will provide robust evidence to clarify the association between abdominopelvic CT exposure during the first trimester of pregnancy and the risk of congenital malformations. Leveraging the NHIS database, we aim to obtain nationwide population data, given the epidemiological nature of the investigation and the rarity of both the exposure and the outcomes. The study findings may contribute evidence to clinical guidelines for the management of pregnant women who require CT examinations or who are inadvertently exposed to CT. These clinical scenarios are relatively common and are encountered by a wide range of healthcare professionals caring for pregnant women and women of childbearing age, including primary care providers, obstetricians, emergency physicians and radiologists, as well as by pregnant women exposed to diagnostic radiation and their families.

This study aims to investigate the following outcomes.

The impact of diagnostic radiation exposure on congenital malformations

According to current clinical guidelines, fetal radiation exposure below a threshold of 100 to 200 mGy does not appear to increase the risk of congenital malformations. Although the estimated fetal dose from a maternal abdominopelvic CT examination is generally lower than this range, evidence regarding the risks associated with such relatively low-dose diagnostic exposures during pregnancy remains scarce. Given the increasing clinical use of CT in pregnant women, the findings of this study will be particularly important. By leveraging the comprehensive NHIS database of a nationwide universal single-payer healthcare system, we will be able to extract relevant and accurate data for cohort formation and subsequent analyses.

Identifying which congenital malformations are susceptible to diagnostic radiation exposure

The primary outcome of this study is the occurrence of major congenital malformations, defined using the EUROCAT classification system. This system categorises congenital malformations into 12 organ system groups, allowing us to assess organ-specific malformations, such as congenital cardiac malformations. Furthermore, in cases of severe congenital malformations, we will identify malformations that required NICU admission or corrective surgery, which are clinically relevant and of particular interest.

Limitations

There are several limitations associated with the nature of the data used in this study. First, the NHIS database was originally established for administrative and healthcare purposes, rather than for research. Specific fetal radiation dose information is not available in the NHIS database, as it contains only claims data for abdominopelvic CT and does not include detailed information on scanner type, vendor, acquisition protocols, technical parameters or field of view. However, we will attempt to estimate fetal dose as accurately as possible using available information, including typical CT radiation dose levels in South Korea, gestational age and maternal height and weight. Furthermore, inclusion failures may occur due to missing mother-child linkages, particularly when a child is listed as a dependent under a different primary policyholder, typically the father. This situation could arise if the mother was employee-insured, self-employed, divorced from the child’s father or covered by a foreign health insurance system. Second, there may be a risk of misclassification of diagnoses. To minimise this risk, we will use a previously validated algorithm to identify diagnoses. Repeated relevant ICD-10 diagnostic codes in the claims data, diagnostic codes accompanied by corresponding surgical or procedural codes and diagnostic codes recorded in association with infant death will be used to improve the accuracy of outcome identification. Furthermore, minor malformations will be excluded because they are not of primary interest and are more prone to misclassification than major congenital malformations, which typically require further evaluation or treatment. Third, there may be missing variables in the NHIS data. For example, gestational age is available for approximately 80% of the included mother-child pairs. In such cases with missing values, we will use validated algorithms to estimate the beginning of pregnancy, a method that has been used in other large epidemiologic studies.²⁰ We also expect missing data for patient characteristics such as weight, height or alcohol consumption, and in these instances, we will apply appropriate imputation techniques. Fourth, unmeasured confounding remains a potential limitation.