Evaluation of Assisted Reproductive Technology Health Insurance Coverage for Multiple Pregnancies and Births in Korea

Key Points Question Was the 2017 health insurance mandate for assisted reproduction associated with increases in multiple pregnancies and births in Korea? Findings This cohort study of 1 474 484 women found that the possibility of multiple pregnancies and multiple births increased significantly after coverage for assisted reproductive technology treatment. The number of total births per pregnant woman was on a decreasing trend before the intervention but began to increase after the intervention. Meaning These findings suggest that policies that provide financial support for infertility treatment may contribute to addressing Korea’s low birth rates.


Introduction
Over the past decades, the average total fertility rate (TFR) has decreased dramatically around the world. 1,2 The Organization for Economic Co-operation and Development defines TFR as the number of children that a woman would give birth to if she were to live to the end of her childbearing years and give birth to children in alignment with the prevailing fertility rates for each age group. 3 Assuming no migration and no change in mortality, a TFR of 2.1 children per woman ensures a largely stable population. 3 As of 2021, the TFR of Organization for Economic Co-operation and Development countries was 1.63. 4 However, Korea's TFR is only 0.81, which has resulted in it ranking at the bottom for decades. 5 The United Nations defines countries with a TFR of 2.1 or less as low-fertility countries and those with a TFR of 1.3 or less as ultra-low-fertility countries. 6 Accordingly, Korea was already classified as a low-fertility country in 1983 and as an ultra-low-fertility country in 2002. 5 It is one of the fastest-aging nations in the world. 7 Since 2020, there has been a reversal in the number of deaths and births, leading to a decline in the population. 5 South Korea is facing serious problems related to low fertility and an aging population. Various policies are being tried to prevent further population decline and increase the TFR, but the effect has not been significant. To address the complex issues associated with low fertility rates, the Korean government recognized the medical necessity of infertility care and introduced policies of financial support and an assisted reproductive technology (ART) insurance mandate for couples experiencing infertility. 8,9 The ART health insurance mandate, which has been in effect since October 2017, covers 70% of ART treatment costs to all couples experiencing infertility regardless of income. In the past, people who had incomes lower than the median were given government subsidies for ART treatment, whereas this is the first coverage policy for those with incomes above the median.
However, it has not been proven whether government policies supporting ART treatment have contributed to the increase in multiple birth rates and the national TFR.
A previous study in the US found that comprehensive insurance mandates are associated with greater use of ART treatment. 10 It has also been suggested that health insurance coverage for infertility treatment significantly increases the first birth rate for women older than 35 years. 11 Several studies have reported on the associations of ART treatment with multiple pregnancies 12,13 and multiple births. [14][15][16] Although the cost of ART treatment varies according to each country's support and coverage policies, 17 it is still a costly treatment for patients. 18 Since Korea began to provide coverage for ART in 2017, few studies have evaluated the association of covering ART with patients' economic burden, health outcomes, and fertility rates. Therefore, this study aimed to evaluate whether the ART insurance mandate is associated with the increase in multiple pregnancies and births in Korea. Our hypothesis was that the probability of having multiple pregnancies and births would rise, as well as the total number of live births per pregnant woman, after the intervention.

Data and Study Population
The NHIS delivery cohort database we obtained included information on all women who have experienced childbirth at least once at Korean medical institutions between July 1, 2015, and December 31, 2019. Delivery was defined using any record of inpatient hospitalization that included pregnancy-related diagnosis codes or procedure codes for vaginal or cesarean deliveries. 21 A total of 1 474 484 individuals constituted the study population after excluding those with no birth records at Korean medical institutions and those with missing data.
The study protocol was approved by the institutional review board of Eulji University (No. EU22-27). Because the NHIS database we used did not contain identifiable information, the requirement for informed consent was waived. In conducting this observational study, we followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

Variables
The main variable of interest was the time of introduction of the ART health insurance coverage policy, which was divided into the periods before and after the intervention. Because the NHIS began covering ART on October 1, 2017, and the last follow-up date for data was December 31, 2019, we determined the follow-up periods before and after the intervention to be 27 months each. Therefore, the preintervention period was from July 1, 2015, to September 30, 2017 and the postintervention period was from October 1, 2017, to December 31, 2019. Study participants were classified as preintervention or postintervention based on the time of their delivery. Whether or not participants received ART could only be identified during the postintervention period because ART has only been recorded in claims data since October 2017, when the ART insurance mandates were implemented. Therefore, after the intervention, ART could be defined as infertility treatment (ICD-10 code N97) using all ART procedures, including intrauterine insemination (IUI) (ICD-10 code Z31.1) and in vitro fertilization and embryo transfer (IVF-ET) (ICD-10 code Z31.2). 21,22 There were 3 dependent variables. First, multiple pregnancies were cases in which 2 or more fetuses were conceived and were identified with ICD-10 code O30. Second, multiple births were cases in which 2 or more babies were born and were determined by ICD-10 code O84. Third, total births were defined as the total number of babies born to each pregnant woman during the follow-up period and were classified into categories from 1 to 5.
As covariates, we included sociodemographic and obstetric factors. Sociodemographic factors were maternal age (<20, 20-29, 30-39, and Ն40 years), region (Seoul, metropolitan areas, small cities, and rural areas), income level (quartile 1 [lowest] to quartile 4 [highest]), and type of insurance (regionally insured, workplace insured, and medical aid). Obstetric factors were parity (nulliparous and multiparous) and maternal comorbidities (0 or Ն1). 23 log(μ i ) = β 0 + β 1 × time t + β 2 × intervention t + β 3 × time after intervention t + e t In this model, the intercept β 0 estimates the baseline level of the outcome; β 1 estimates the baseline trend of the outcome; β 2 estimates the level change after the intervention, which indicates the immediate effect size of the intervention; β 3 estimates the change in trend after the intervention; and the sum of β 1 and β 3 is the slope after the intervention, indicating the follow-up outcome trend. 25 Parameter estimates, SEs, and P values are presented as key results.

Statistical Analysis
Subsequently, a modified Poisson regression analysis with sandwich error estimation was performed to examine the association between ART treatment and outcomes after the intervention using adjusted risk ratios and 95% CIs. SAS, version 9.4 statistical software (SAS Institute Inc) was used for all analyses. A 2-sided P < .05 was considered statistically significant. the study population, those who experienced multiple pregnancies and multiple births were 1.60% and 1.10%, respectively. During the entire follow-up period, 73.54% of the women gave birth to only 1 child, and it was extremely rare for any woman to give birth to more than 3 children (0.02%). The total number of mothers after the intervention (n = 653 386) decreased compared with before the intervention (n = 821 098). However, the proportion of multiple pregnancies and multiple births increased after the intervention to 1.75% and 1.15%, respectively. Similarly, the proportion of women having 2 or more children increased during the postintervention period (28.44% vs 24.87% preintervention). eTable 1 in Supplement 1 shows the demographic characteristics of the study population by preintervention and postintervention periods. Table 2 presents the results of the segmented regression analysis used to estimate the probabilities of the 3 outcome variables, adjusting for all covariates. The likelihood of having multiple pregnancies after ART insurance coverage was estimated to increase by 0.7% compared with before coverage (estimate, 1.007; 95% CI, 1.004-1.011; P < .001). Multiple births during the preintervention period showed a decreasing trend (estimate, 0.997; 95% CI, 0.994-0.999; P = .02) but changed to an increasing trend after the intervention (estimate, 1.008; 95% CI, 1.005-1.011; P < .001). The probability of having multiple births after the intervention increased by 1.2% compared with before the intervention (estimate, 1.012; 95% CI, 1.007-1.016; P < .001). In addition, the total number of births per pregnant woman increased by 0.5% (estimate, 1.005; 95% CI, 1.005-1.005; P < .001) with ART coverage. Before the intervention, there was a downward trend in the number of births per woman (estimate, 0.998; 95% CI, 0.998-0.998; P < .001), but after the intervention, there was a significant increase (estimate, 1.003; 95% CI, 1.003-1.003; P < .001). Table 2, the size of the level and trend changes could be quantitatively confirmed by calculating the parameter estimates, whereas as seen in the Figure, it was possible to intuitively confirm the outcome trend after the intervention. The ITS results showed that multiple pregnancies, multiple births, and total births increased after ART health insurance coverage was introduced.

As shown in
As Table 3 shows, subgroup analyses stratified by income level were conducted considering that the benefits of ART treatment subsidies differed according to income level. Similarly, all remaining covariates other than income level were statistically adjusted to calculate the estimates.
Notably, the groups with income levels in quartiles 3 and 4 showed a decreasing trend in multiple births and total births before ART coverage that increased significantly after the intervention (quartile 3: trend change after intervention, 1.013; 95% CI, 1.007-1.020 [P < .001] and 1.005; 95% CI,  for those who received IVF-ET treatment than for mothers who did not receive any ART treatment.

Discussion
This cohort study evaluated the demographic policy outcomes of ART treatment coverage introduced as a solution to Korea's low fertility rate, which is the lowest in the world. 5 There were 3 principal findings. First, since October 2017, 70% of the total expenses of ART treatment were covered by health insurance for all couples experiencing infertility regardless of income level, and the possibility of multiple pregnancies and multiple births increased significantly. Second, the number of total births per pregnant woman showed a decreasing trend before the intervention but increased after the intervention. Before ART coverage, planning additional births through ART procedures would be challenging due to the high cost of infertility treatment. However, it is anticipated that after coverage, the cost burden has lessened, leading to a relative increase in the total number of births.
Third, in high-income groups that were not eligible for financial support for ART treatment before the intervention, the introduction of ART insurance mandates had a more noticeable association with multiple births and total births. We believe that health insurance coverage for ART positively contributed to multiple pregnancies, multiple births, and total births. Although it was not a complete coverage and the out-of-pocket expense was 30% of the total treatment cost, the outcomes of introducing the policy were remarkable. According to NHIS claims data, the proportion of newborns conceived through ART procedures in Korea has actually increased over the past 5 years since the implementation of ART health insurance coverage, reaching 12.3% in 2022. 26 As the age of marriage and first childbirth is delayed, the proportion of couples experiencing infertility and the use of ART treatment are expected to increase.
Most previous studies that examined the outcomes of ART insurance mandates focused on economic burden, 17,18 use of ART treatment, 27 and pregnancy outcomes. 11,28 According to their findings, implementing ART coverage policies reduces patient costs and is positively associated with ART treatment use and pregnancy outcomes. There was also evidence that ART contributes to improving fertility rates and demographic changes. 29 The findings of these studies are consistent with the implications of our study: ART should be actively supported as a national strategy to address low fertility rates. On the other hand, several studies have reported adverse associations of ART with maternal health outcomes. For example, a higher risk of severe maternal morbidity has been confirmed in mothers treated with ART than in those not treated with ART. 21 The biggest difference between previous research and our study is that we not only evaluated the outcomes of ART treatment but also the outcomes of the ART treatment coverage policy.
Considering that ART treatment imposes a very large economic burden on couples experiencing infertility, 18   first to cover people with incomes above the median. Therefore, in this study, differences in intervention outcomes according to income level were observed. This policy also contributed to an increase in total births per woman, which had previously been on a downward trend. Although Korea's TFR continues to decline, our findings suggest that the decline in TFR may be slowed slightly by the ART health insurance coverage policy.

Strengths and Limitations
This study has several strengths. A major strength is that the NHIS database we analyzed contains nationwide cohort data, which ensures its applicability to the study when assessing the outcomes of medical practice and health. In addition, we used the ITS design, which is a strong quasiexperimental approach for evaluating the longitudinal outcomes of interventions. 30 The main advantage of this design is that it makes full use of the longitudinal nature of the data and accounts for preintervention trends. 24 Previous studies have used a difference-in-differences study design that compares only 2 time points to investigate the net policy association with the outcomes 31 or segmented regression with fewer than 10 time points. 32 However, there were too few time points to capture baseline trends and changes. This study used 54 time points (27 time points each before and after the intervention) to capture trend changes more robustly.
This study also had several limitations. First, the NHIS delivery cohort data we analyzed included only information on women who had experienced childbirth during the study period and did not include information on all women of childbearing potential in Korea; therefore, it was not possible to explore the outcomes of changes in the TFR trend with all Korean women of childbearing age as the denominator. Second, the inherent limitations of administrative claims data have been pointed out.
Korea's NHIS cohort data set is not a medical record; rather, it provides data for requesting the insurer to pay a part of patients' total medical expenses. 20 Thus, the recorded disease codes in the NHIS may not represent the actual disease status of patients. 20 Moreover, it was not possible to verify whether ART treatment received an uninsured benefit. Third, we used the ICD-10 codes to select study participants and identify their specific outcomes, but these codes inherently have some problems. [33][34][35] The ICD-10 codes are primarily used for administrative purposes and may not provide detailed clinical information about the patient. There is also the concern of incomplete coding, which could misclassify or underestimate the outcomes. Fourth, we could not control for all of the other informal benefits and interventions provided to couples experiencing infertility around the same time as the ART health insurance coverage policy was introduced. Thus, although there may have been an overestimation of the associations, ART insurance coverage was the most comprehensive and official intervention promoted at that time. Fifth, the follow-up period is insufficient for assessing the actual outcomes of the policy because certain patients may have undergone IVF before the intervention and transferred embryos at different times, leading to variability in the follow-up period, which may not truly reflect the policy's impact. Sixth, we tried to adjust for potential confounders associated with childbirth, such as parity and maternal comorbidities, but residual confounding from unmeasured variables could not be ruled out.

Conclusions
The findings of this population-based cohort study suggest that ART health insurance coverage is associated with an increase in multiple pregnancies and multiple births and may also contribute to improving the birth rate. Considering the demographic structure of Korea, which has a very low fertility rate, policies that can support couples experiencing infertility should be further developed and implemented.