Preconception computed tomography exposure and risk of stillbirth in a nationwide population based cohort study

Abstract

Fetal radiation exposure can cause pregnancy loss, malformation, growth disturbance, and mutagenic/carcinogenic effects. Insufficient guidance on radiation protection for women of childbearing age complicates risk assessment. This population-based retrospective cohort study investigated preconception CT radiation on stillbirths using Taiwan National Health Insurance Database from 1995 to 2018. A total of 1,218,744 births (live and stillbirth) of 940,280 women were eligible, including 1,179,443 live births of 913,811 women in the control group and 13,631 stillbirths(≥ 20 weeks of gestation or weight ≥ 500 g) of 13,046 women in the study group. CT exposure within 90 days, 1 year, 3 years, and 5 years before conception occurred in 0.6%, 2.5%, 6.1%, and 9.0% of stillbirth mothers, versus 0.5%, 2.1%, 5.4%, and 8.1% of controls. In adjusted analyses, CT exposure within 1 year (aOR = 1.14; 95% CI: 1.01–1.28; p = 0.0325) and 3 years (aOR = 1.08; 95% CI: 1.00–1.17; p = 0.0482) before pregnancy was modestly associated with stillbirth, while the 90-day (aOR = 1.12; 95% CI: 0.88–1.42; p = 0.355) and 5-year (aOR = 1.06; 95% CI: 0.99–1.13; p = 0.0917) windows were not significant. These small effect sizes, together with potential residual confounding and lack of consistent dose–response, suggest that any true association between preconception CT radiation and stillbirth is weak. Further research with detailed dose assessment and indication data is warranted.

Introduction

Computed tomography (CT) scan is quick, reliable and commonly used in trauma triage and illness diagnosis. The use of CT in detecting, diagnosing, and monitoring disease has multiplied worldwide in recent years. However, the inevitable ionizing radiation(IR) used in CT scans can cause biological DNA damage, including point mutations, chromosomal translocations, and gene fusions leading to radiation-induced cancers¹. Previous studies have shown increased cancer risks in adults receiving radiation exposure from CT^2,3. The risk versus benefit assessment is significant when practicing IR imaging in sensitive patient groups, including children and pregnant women. Numerous studies have indicated that radiation from CT scans could lead to a subsequent risk of cancers in children^4–8. Some guidelines and recommendations are available when performing medical imaging on children and pregnant or lactating women^9–11. The general principle of diagnostic imaging during pregnancy is to minimize radiation exposure to “As Low As Reasonably Achievable” (ALARA). Despite concerns about fetal radiation exposure, patients and health care providers should not avoid necessary medical imaging examinations solely for this reason.

The potential risk of IR on the fetus varies depending on the pregnancy’s radiation dose and gestational age. Excess radiation exposure to the fetus can lead to pregnancy loss, congenital malformation, disturbance of development or growth, or even mutagenic and carcinogenic effects. Pregnancy loss is likely to occur with radiation exposure during an early stage of gestation (< 2 weeks). Malformations and developmental delays occur during the organogenesis period (2 weeks to 8 weeks) and depend on the radiation dose¹². However, there is limited evidence regarding the harmful risks to adolescent girls and women of childbearing age receiving ionizing radiation from CT scans.

The loss of a pregnancy is a significant adverse outcome that profoundly impacts both parents and their families. Pregnancy loss includes miscarriage (defined as a nonviable intrauterine pregnancy less than 20 weeks in gestation) and stillbirth (the loss of the fetus ≥ 20 weeks of gestation or weight ≥ 500 g in Taiwan¹³. Stillbirths are a significant public health problem in low/middle-income countries but remain common in developed countries¹⁴. Unfortunately, most of the causes of stillbirth are unknown or unexplained^15–17. The most common causes of stillbirths are maternal factors (age, obstetric history, obesity, infection, preexisting medical conditions, anemia, malnutrition and substance use), poor or restricted fetal growth, chromosomal and genetic abnormalities, placental complications, and external incidents (antepartum injuries/trauma or birth asphyxia and obstetric trauma)^18,19. Safety and protection during pregnancy are common concerns. Unfortunately, most physicians do not have sufficient awareness when referring women of reproductive age to medical imaging using IR²⁰. To our knowledge, there is no definite guidance for or compliance with radio-diagnostic imaging protection in nonpregnant women, and the risk assessment remains unclear. Studies from extensive cohort databases are necessary to clarify the potential risk of ionization from medical imaging on stillbirth. In this study, we aim to investigate the potential risk of IR from CT regarding stillbirths and their subsequent pregnancies for prepregnant women by analyzing the Taiwan National Health Insurance Databases (NHIRD) and the Birth Reporting Database (BRD).

Results

This nationwide population-based cohort study analyzed 13,631 stillbirth mothers and 1,179,443 live birth mothers from the Taiwan National Health Insurance Research Database. Baseline characteristics demonstrated significant demographic and clinical differences between groups (Table 1). Compared with the control group, mothers experiencing stillbirth were significantly more likely to be of advanced maternal age (≥ 35 years: 36.0% vs. 24.3%), have multiple births (9.2% vs. 3.4%), and present with a history of pregnancy loss (18.3% vs. 14.5%). Maternal comorbidities demonstrated substantially higher prevalence rates in the stillbirth group compared with controls, with all conditions achieving statistical significance. The most pronounced differences included fetal chromosomal abnormalities (17.8% vs. 0.4%), infertility (23.1% vs. 18.7%), and placenta previa and abruptio placentae (15.8% vs. 12.2%). Maternal chronic medical conditions were consistently more prevalent among stillbirth mothers, including hypertension (3.6% vs. 1.3%), diabetes mellitus (2.3% vs. 1.0%), and obesity (0.7% vs. 0.4%). Maternal mental health disorders also showed significant associations, including depression (3.4% vs. 2.9%), anxiety disorders (3.9% vs. 3.5%), and schizophrenia (0.3% vs. 0.2%).

Table 1.

Demographic characteristics of stillborn birth mother and live birth mother.

Maternal data	Stillborn birth mother (N = 13,631)		Live birth mother (N = 1,179,443)
	n	%	n	%
Age at pregnancy, year-old
<20	402	2.9	19,330	1.6
20–24	1006	7.4	103,283	8.8
25–29	2551	18.7	297,717	25.2
30–34	4759	34.9	472,153	40.0
35–39	3873	28.4	248,193	21.0
40–44	978	7.2	37,207	3.2
≥45	62	0.5	1560	0.1
Urbanization level
1 (City)	2869	21.0	256,726	21.8
2	7381	54.1	641,759	54.4
3	2260	16.6	185,968	15.8
4 (Villages)	1121	8.2	94,990	8.1
Income, New Taiwan dollar/month
0	38	0.3	3658	0.3
1–15,840	1700	12.5	128,401	10.9
15,841–25,000	5272	38.7	449,611	38.1
>25,000	6621	48.6	597,773	50.7
Plurality
Singleton	12,371	90.8	1,139,049	96.6
Multiple birth	1260	9.2	40,394	3.4
Parity
Primipara	7363	54.0	626,687	53.1
Multipara	6268	46.0	552,756	46.9
History of pregnancy loss
Yes	2498	18.3	171,578	14.5
No	11,133	81.7	1,007,865	85.5
Maternal comorbidities
Hypertension	496	3.6	14,944	1.3
Diabetes Mellitus	307	2.3	12,116	1.0
Obesity	97	0.7	4837	0.4
Infertility	3150	23.1	221,139	18.7
Epilepsy	50	0.4	3138	0.3
Anxiety	537	3.9	41,498	3.5
Depression	459	3.4	33,919	2.9
Schizophrenia	41	0.3	1802	0.2
Placenta previa and abruptio placentae	2154	15.8	143,999	12.2
Intrauterine growth restriction	412	3.0	32,493	2.8
Chromosomal abnormality in fetus	2420	17.8	4775	0.4
Children data
Gender
Male	6873	50.4	611,155	51.8
Female	6395	46.9	568,288	48.2
Unknown	363	2.7	0	0.0
Gestational age at delivery, mean (SD), week	24.9 (5.6)		38.2 (1.7)
Birth weight, mean(SD), gram	827.2 (777.0)		3038.5 (460.1)

Unadjusted analyses revealed consistently higher rates of preconceptional CT exposure among stillbirth mothers across all time windows examined (Tables 2 and 3). The proportion of mothers receiving CT scans was significantly elevated in the stillbirth group for exposure within 1 year (2.5% vs. 2.1%; p = 0.0004), 3 years (6.1% vs. 5.4%; p < 0.0001), and 5 years (9.0% vs. 8.1%; p < 0.0001) before conception. CT exposure within 90 days approached no statistical significance.

Table 2.

Comparisons of CT exposure within different preconceptional windows between stillborn birth mother and live birth mother groups.

Preconception CT exposure window	Stillborn birth mother (N = 13,631)		Live birth mother (N = 1,179,443)		p value
	n	%	n	%
90 days					0.1113
No	13,546	99.4	1,173,258	99.5
Yes	85	0.6	6185	0.5
1 year					0.0004
No	13,292	97.5	1,155,230	97.9
Yes	339	2.5	24,213	2.1
3 years					< 0.0001
No	12,793	93.9	1,116,327	94.6
Yes	838	6.1	63,116	5.4
5 years					< 0.0001
No	12,398	91.0	1,084,154	91.9
Yes	1233	9.0	95,289	8.1

Table 3.

Comparisons of number of CT scan(s) within different preconceptional windows between stillborn birth mother and live birth mother groups.

Number of scan(s) within different preconceptional windows	Stillborn birth mother (N = 13,631)		Live birth mother (N = 1,179,443)		p value
	n	%	n	%
within 90 days					0.2814
0	13,546	99.4	1,173,258	99.5
1	81	0.6	5892	0.5
≥2	4	0.0	293	0.0
within 1 year					0.0004
0	13,292	97.5	1,155,230	97.9
1	301	2.2	22,106	1.9
≥2	38	0.3	2107	0.2
within 3 years					< 0.0001
0	12,793	93.9	1,116,327	94.6
1	709	5.2	54,443	4.6
≥2	129	0.9	8673	0.7
within 5 years					< 0.0001
0	12,398	91.0	1,084,154	91.9
1	988	7.2	79,164	6.7
≥2	245	1.8	16,125	1.4

Table 5.

Logistic regression analysis for number of CT scan(s) within different preconceptional windows between stillborn birth mother and live birth mother groups.

Number of scan(s) within different preconceptional windows	Crude				Adjusteda
	OR	95% CI		p value	OR	95% CI		p value
within 90 days
0	0.80	0.64	1.01	0.0610	0.89	0.69	1.13	0.3261
1	1.00	Reference			1.00	Reference
≥2	0.92	0.32	2.63	0.8795	0.82	0.26	2.63	0.7412
within 1 year
0	0.83	0.74	0.94	0.0029	0.90	0.79	1.02	0.1016
1	1.00	Reference			1.00	Reference
≥2	1.28	0.89	1.84	0.1835	1.27	0.87	1.85	0.2095
within 3 years
0	0.88	0.81	0.95	0.0013	0.94	0.87	1.02	0.1550
1	1.00	Reference			1.00	Reference
≥2	1.13	0.92	1.38	0.2457	1.13	0.91	1.39	0.2667
within 5 years
0	0.91	0.85	0.98	0.0107	0.98	0.91	1.05	0.5414
1	1.00	Reference			1.00	Reference
≥2	1.22	1.05	1.41	0.0095	1.20	1.03	1.40	0.0218

^aAdjusted for maternal age at pregnancy, urbanization level, income, plurality, parity, history of pregnancy loss, hypertension, diabetes mellitus, epilepsy, schizophrenia, placenta previa and abruptio placentae, intrauterine growth restriction, chromosomal abnormality in fetus by using stepwise model selection.

Analysis of scan frequency demonstrated that multiple CT examinations (≥ 2 scans) within 5 years were more common among stillbirth mothers (1.8% vs. 1.4%; p < 0.0001). Similar patterns were observed for shorter time intervals, with multiple scans within 3 years (0.9% vs. 0.7%) and 1 year (0.3% vs. 0.2%) showing statistically significant differences (both p < 0.0001).

After adjusted for maternal age at pregnancy, urbanization level, income, plurality, parity, history of pregnancy loss, hypertension, diabetes mellitus, epilepsy, schizophrenia, placenta previa and abruptio placentae, intrauterine growth restriction, chromosomal abnormality in fetus by using stepwise model selection, multivariable logistic regression analysis demonstrated significant associations between preconceptional CT exposure and stillbirth risk (Table 4). CT exposure within 1 year (adjusted OR = 1.14, 95% CI: 1.01–1.28, p = 0.0325) and 3 years (adjusted OR = 1.08, 95% CI: 1.00–1.17, p = 0.0482), while 5-year exposure showed a borderline association (adjusted OR = 1.06, 95% CI: 0.99–1.13, p = 0.0917). The 90-day window was not significant (adjusted OR = 1.12, 95% CI: 0.88–1.42, p = 0.355). Analysis of scan frequency effects revealed evidence of a dose-response relationship (Table 5). Among mothers receiving multiple CT examinations within 5 years before conception, those with ≥ 2 scans demonstrated a significantly elevated stillbirth risk compared to those receiving only one scan (adjusted OR = 1.20, 95% CI: 1.03–1.40, p = 0.0218). This effect was specific to the 5-year exposure window, as multiple scans within shorter intervals (90 days, 1 year, or 3 years) did not achieve statistical significance.

Discussion

In this large nationwide, population-based cohort including both term and preterm live births as the control group, we found only a small and at most modest association between preconception CT exposure and stillbirth. Across the different exposure windows, adjusted odds ratios ranged from 1.06 to 1.14, suggesting that if a causal link exists, the effect size is limited. Specifically, CT exposure within 1 year (aOR = 1.14, 95% CI: 1.01–1.28) and 3 years (aOR = 1.08, 95% CI: 1.00–1.17) before conception showed statistically significant associations with stillbirth, whereas the 90-day (aOR = 1.12, 95% CI: 0.88–1.42) and 5-year (aOR = 1.06, 95% CI: 0.99–1.13) intervals were not significant. Given the large sample size, these findings underscore the importance of focusing on the magnitude and consistency of the estimates rather than statistical significance alone.

The lack of statistical significance and the smaller effect estimate for the 90-day window compared with the 1-year and 3-year intervals is noteworthy. One possible explanation is that CT scans performed near conception are more likely to be prompted by acute, transient illnesses or trauma that may not be mechanistically related to stillbirth risk, whereas scans performed further before pregnancy could be markers of chronic or systemic conditions with long-term reproductive impacts. Alternatively, this discrepancy may simply reflect chance variation or unmeasured confounding factors. In keeping with prior literature, our results do not demonstrate a dose–response pattern for shorter intervals, yet we observed that multiple CT scans within 5 years were associated with slightly higher odds of stillbirth than a single scan, suggesting a possible cumulative exposure effect worth further exploration.

Exposure to IR can increase teratogenic, carcinogenic, or mutagenic risks depending on the dosage and duration of exposure²¹. Low-dose (< 1 Gy) IR exposure often causes nontargeted effects, inducing systemic cellular responses, including radiation-induced bystander effects, radioadaptive responses, and radiation-induced genomic instabilities²². The Biological Effects of Ionizing Radiation (BEIR) VII report of the US National Academy of Science defines low doses of IR as up to 100 mSv (= 100 mGy)²³ including IR from diagnostic imaging (such as X-ray and CT scan), radiotherapy, and occupational/residential exposure²¹. In recent decades, the increasing use of CT exams has led to higher levels of radiation exposure. The cumulative dose of IR exposure from CT scans is associated with an increased subsequent risk of cancer in adults^3,24 and children²⁵. Patients undergoing repeated CT exams accumulate higher organ radiation doses and have higher cancer mortality rates²⁶. In addition to risks of malignancy, IR from medical imaging could play a role in causing damage to tissues and organs²³. The ovaries contain limited resting nonrenewable oocytes, and gonads are highly sensitive to radiation damage²⁷. Radiation exposure to ovaries could lead to loss of the follicle reservoir and reduced ovarian function, and the damage depends on the IR dose received and the stage of oogenesis/folliculogenesis²⁸. Acute ovarian failure is characterized by the cessation of estrogen production, which can occur within five years after radiation exposure²⁹. Exposure to IR through radiotherapy has been found to damage and negatively impact the fertility of adolescent girls and women of childbearing age who are receiving cancer treatment. Abdominal irradiation altered uterine vascularization, reduced myometrial and cervical elasticity, and endometrial atrophy/insufficiency, which were associated with complications during pregnancy (such as placental disorders, fetal malposition, preterm delivery, low birth weight, and higher risk of uterine rupture). Cranial irradiation influences hypothalamic-pituitary-gonadal axis dysfunction and anterior pituitary hormone deficiency³⁰. In a study monitoring women nuclear industry workers, exposure to low doses of IR before conception was associated with an increased risk of early (< 13 weeks of gestation) miscarriage (OR = 1.3; 95% CI: 1.0-1.6) and stillbirth (OR = 2.2; 95% CI: 1.0-4.6)³¹. Adverse birth outcomes such as low birth weight, miscarriage, stillbirth, and preterm delivery, were observed in women who received radiotherapeutic IR before pregnancy²¹. Although most current perspectives on medical imaging suggest that the risk of radiation exposure from CT scans is negligible, we cannot completely disregard the potential for IR to cause harm to reproductive organs.

This study’s strength lies in its analysis of extensive, unbiased data from the NHIRD, which covers more than 99% of Taiwan’s population. The NHIRD includes time-stamped medical records and examinations, enabling us to track CT scan records before patients’ pregnancies. Previous studies have mainly focused on comparing the risk of radiation exposure during pregnancy with current radiation protection guidelines geared toward pregnant women. However, the risk of prepregnant women receiving radiation imaging has not been thoroughly explored. To our knowledge, this study is the first to assess the risk of preconceptional CT and its subsequent effect on pregnancy loss (stillbirth). Raising awareness and concern regarding radiation imaging’s impact on underlying illnesses is crucial for patients and health care providers. However, our study has several weaknesses. First, the cohort was limited to primarily Asian women, lacking diversity of ethnicities or demographic differences. Second, other modalities of radio-diagnostic imaging, such as X-ray and mammography, were not included in the analysis, so the risks of other radiation exposures could not be evaluated. Third, a primary limitation inherent in using administrative claims data is the potential for confounding by indication. The clinical reasons for ordering a CT scan (e.g., trauma, acute abdominal pain, or screening for other conditions) were not available in our database. Although we performed comprehensive adjustments for a wide array of diagnosed maternal comorbidities, obstetric histories, and pregnancy-related complications, we cannot completely rule out the possibility that the underlying pathology that necessitated the CT scan, rather than the radiation exposure itself, is the primary driver of the observed association with stillbirth. Despite our extensive adjustments, there may be residual confounding from unmeasured variables. Factors such as maternal infections, anemia, malnutrition, substance use, or environmental exposures, which are also known risk factors for stillbirth, are not reliably captured in the NHIRD and thus could not be controlled for in our models. In addition, the body portions of CT scans performed, and the scan parameters were not included in the database; thus, further evaluations regarding the radiation doses and the impacts on reproductive organs by CT scans of different parts and parameters could not be evaluated. Finally, the causes of stillbirths are not addressed. Although the correlation between radiation exposure and the risk of stillbirth is significant, we cannot assume that medical diagnostic imaging is the only reason for stillbirths, as it may cause maternal or fetal damage.

In conclusion, in this large nationwide cohort, preconception CT exposure within 1 to 3 years before pregnancy was associated with a very modest increased risk of stillbirth, while exposures within 90 days or 5 years showed no significant associations. The small magnitude, limited consistency, and potential for residual confounding suggest that any causal relationship, if present, is weak. Further research should aim to clarify dose–response relationships, account for clinical indications of imaging, and integrate more granular exposure assessment to better inform guidelines on radiation protection for women of reproductive age.

Methods

Data sources

The National Health Insurance Research Database (NHIRD)

The Taiwan National Health Insurance (NHI) reimbursement system, launched in 1995, is a compulsory social insurance program. It provides social insurance coverage, and the NHIRD is derived from it. The NHIRD comprises deidentified medical claims from 99% of Taiwan’s population, approximately 23 million people. The NHIRD collects public health information, including administrative and health claims data, and is confidentially maintained by the Personal Electronic Data Protection Law. It is designed for epidemiological research on prescription use, disease diagnoses, and the cost of hospitalizations³².

The Birth Reporting Database (BRD)

The BRD is a nationwide dataset maintained by Taiwan’s Ministry of Health and Welfare to record all live births and stillbirths (≥ 20 weeks of gestation or ≥ 500 g). It contains 34 fields covering maternal background, pregnancy risk factors, delivery details, newborn information, and congenital anomalies. Data are de-identified and updated annually, supporting public health monitoring, policy-making, and research on maternal and child health trends across the country.

The Catastrophic Illness Certificate Database (CICD)

The CICD is used to identify severe illness patients according to the International Classification of Diseases, Ninth Revision (ICD-9-CM, before 2016) codes, or Tenth revision (ICD-10-CM, from 2016) codes. All patients are clinically, histologically, or cytologically confirmed before a catastrophic illness certificate is issued.

Maternal data and covariates

We extracted comprehensive data on maternal demographics, obstetric history, and clinical conditions from the databases to control for potential confounding. Demographic variables included maternal age, socioeconomic status (approximated by urbanization level and monthly income), and plurality. Obstetric history was defined by parity and any prior history of pregnancy loss identified from procedural codes. A panel of pre-existing maternal comorbidities and pregnancy-related complications was identified using ICD diagnostic codes, validated by requiring at least one inpatient or multiple outpatient visits for most chronic conditions. These included hypertension, diabetes mellitus, obesity, infertility, epilepsy, and mental health disorders (anxiety, depression, schizophrenia). We also accounted for key pregnancy-related complications such as placenta previa/abruptio placentae, intrauterine growth restriction, and fetal chromosomal abnormalities. All of these variables were included as covariates in the multivariable regression models.

Procedures

Birth data, including live and stillborn and their mothers, were retrieved from the BRD from 2013 to 2018. Mothers who were unable to link to NHIRD, as well as who had diseases including malignancy, congenital deficiency of clotting factors, hereditary/acquired/aplastic anemia, autoimmune diseases, congenital metabolic diseases, congenital anomalies, congenital immune deficiency, and other rare diseases listed in CICD that could cause stillbirth and increasing CT exposure as confounding factors, were excluded for further analysis. Mothers who experienced stillbirth were identified as the study group, while mothers with live births served as the control group. The process of subject inclusion and exclusion, resulting in a final cohort of 13,631 stillbirths and 1,179,443 live births, is depicted in Fig. 1.This approach was chosen to minimize potential confounding, as stillbirth and preterm birth are both adverse pregnancy outcomes that share numerous underlying risk factors, which may also be associated with the likelihood of receiving medical imaging. Finally, quantitative comparisons between CT exposure with 90 days, 1 year, 3 years, and 5 years preconceptional exposure window (radiation exposure from CT scan) between the two groups were performed.

Fig. 1.

Flow chart of selecting stillborn birth mother and live birth mother groups. Note: NHIRD = National Health Insurance Research Database.

Statistical analysis

Differences between stillborn birth mother and live birth mother groups were assessed by Pearson’s chi-square test for categorical variables and the t test for continuous variables. A multivariable logistic regression model was conducted to investigate the CT exposure associated with stillbirth. For women who could contribute multiple births or deliveries, we used the generalized estimating equation with an exchangeable correlation structure to evaluate the effects of model parameters. The final adjusted model (Adjusted^a model in Tables 4 and 5) was developed using a stepwise model selection procedure to control for significant confounding variables, which included maternal age, urbanization, income, plurality, parity, history of pregnancy loss, and the maternal comorbidities listed above. P < 0.05 indicates statistical significance for all statistical tests and regression models, and the analyses were carried out with SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA).

Table 4.

Logistic regression analysis for CT exposure between stillborn birth mother and live birth mother groups.

Preconception CT exposure window	Crude				Adjusteda
	OR	95% CI		p value	OR	95% CI		p value
90 days
No	1.00	Reference			1.00	Reference
Yes	1.24	0.99	1.55	0.0594	1.12	0.88	1.42	0.3550
1 year
No	1.00	Reference			1.00	Reference
Yes	1.23	1.10	1.38	0.0003	1.14	1.01	1.28	0.0325
3 years
No	1.00	Reference			1.00	Reference
Yes	1.16	1.08	1.25	< 0.0001	1.08	1.00	1.17	0.0482
5 years
No	1.00	Reference			1.00	Reference
Yes	1.13	1.07	1.21	< 0.0001	1.06	0.99	1.13	0.0917

^aAdjusted for maternal age at pregnancy, urbanization level, income, plurality, parity, history of pregnancy loss, hypertension, diabetes mellitus, epilepsy, schizophrenia, placenta previa and abruptio placentae, intrauterine growth restriction, chromosomal abnormality in fetus by using stepwise model selection.

Author contributions

Conceptualization, I-Gung Li and Yuan-Hsiung Tsai; Data curation, Ming-De Huang; Formal analysis, Yao-Hsu Yang; Funding acquisition, I-Gung Li, Yao-Hsu Yang and Yuan-Hsiung Tsai; Investigation, Sheng-Wei Chang; Methodology, Yao-Hsu Yang; Project administration, Chuan-Pin Lee and Ko-Jung Chen; Resources, Yu-Li Lee; Software, Ko-Jung Chen; Supervision, I-Gung Li and Yuan-Hsiung Tsai; Validation, Chuan-Pin Lee; Writing – original draft, I-Gung Li and Sheng-Wei Chang; Writing – review & editing, Yu-Li Lee. All authors reviewed the manuscript.

Funding

The research was supported by grant CFRPG6K0071 from Chang Gung Medical Foundation, Chiayi Chang Gung Memorial Hospital. Declaration of interest None.

Data availability

The National Health Insurance Research Database provided the data used in this study, and it has since been transferred to the Health and Welfare Data Science Center (HWDC). To obtain the data, researchers must submit a formal proposal to the HWDC, Department of Statistics, Ministry of Health and Welfare, Taiwan ([https://dep.mohw.gov.tw/DOS/cp-2516-59203-113.html] (https:/dep.mohw.gov.tw/DOS/cp-2516-59203-113.html)), along with IRB approval for research purposes. If you wish to access the data, please contact the HWDC.

Declarations

Competing interests

The authors declare no competing interests.

Informed consent statement

Informed consent was waived by the Institutional Review Board (IRB) of Chang Gung Medical Foundation.

Ethical approval

This study was approved by the Institutional Review Board (IRB) of Chang Gung Medical Foundation (No. 202001474B0D001).

Compliance

This study was performed in accordance with the Declaration of Helsinki.