The risks of birth defects and childhood cancer with conception by assisted reproductive technology

Abstract

STUDY QUESTION

Is there an association between fertility status, method of conception and the risks of birth defects and childhood cancer?

SUMMARY ANSWER

The risk of childhood cancer had two independent components: (i) method of conception and (ii) presence, type and number of birth defects.

WHAT IS KNOWN ALREADY

The rarity of the co-occurrence of birth defects, cancer and ART makes studying their association challenging. Prior studies have indicated that infertility and ART are associated with an increased risk of birth defects or cancer but have been limited by small sample size and inadequate statistical power, failure to adjust for or include plurality, differences in definitions and/or methods of ascertainment, lack of information on ART treatment parameters or study periods spanning decades resulting in a substantial historical bias as ART techniques have improved.

STUDY DESIGN, SIZE, DURATION

This was a population-based cohort study linking ART cycles reported to the Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) from 1 January 2004 to 31 December 2017 that resulted in live births in 2004–2018 in Massachusetts and North Carolina and live births in 2004–2017 in Texas and New York. A 10:1 sample of non-ART births were chosen within the same time period as the ART birth. Non-ART siblings were identified through the ART mother’s information. Children from non-ART births were classified as being born to women who conceived with ovulation induction or IUI (OI/IUI) when there was an indication of infertility treatment on the birth certificate, and the woman did not link to the SART CORS; all others were classified as being naturally conceived.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The study population included 165 125 ART children, 31 524 non-ART siblings, 12 451 children born to OI/IUI-treated women and 1 353 440 naturally conceived children. All study children were linked to their respective State birth defect registries to identify major defects diagnosed within the first year of life. We classified children with major defects as either chromosomal (i.e. presence of a chromosomal defect with or without any other major defect) or nonchromosomal (i.e. presence of a major defect but having no chromosomal defect), or all major defects (chromosomal and nonchromosomal), and calculated rates per 1000 children. Logistic regression models were used to generate adjusted odds ratios (AORs) and 95% CIs of the risk of birth defects by conception group (OI/IUI, non-ART sibling and ART by oocyte source and embryo state) with naturally conceived children as the reference, adjusted for paternal and maternal ages; maternal race and ethnicity, education, BMI, parity, diabetes, hypertension; and for plurality, infant sex and State and year of birth. All study children were also linked to their respective State cancer registries. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs of cancer by birth defect status (including presence of a defect, type and number of defects), and conception group.

MAIN RESULTS AND THE ROLE OF CHANCE

A total of 29 571 singleton children (2.0%) and 3753 twin children (3.5%) had a major birth defect (chromosomal or nonchromosomal). Children conceived with ART from autologous oocytes had increased risks for nonchromosomal defects, including blastogenesis, cardiovascular, gastrointestinal and, for males only, genitourinary defects, with AORs ranging from 1.22 to 1.85; children in the autologous-fresh group also had increased risks for musculoskeletal (AOR 1.28, 95% CI 1.13, 1.45) and orofacial defects (AOR 1.40, 95% CI 1.17, 1.68). Within the donor oocyte group, the children conceived from fresh embryos did not have increased risks in any birth defect category, whereas children conceived from thawed embryos had increased risks for nonchromosomal defects (AOR 1.20, 95% CI 1.03, 1.40) and blastogenesis defects (AOR 1.74, 95% CI 1.14, 2.65). The risk of cancer was increased among ART children in the autologous-fresh group (HR 1.31, 95% CI 1.08, 1.59) and non-ART siblings (1.34, 95% CI 1.02, 1.76). The risk of leukemia was increased among children in the OI/IUI group (HR 2.15, 95% CI 1.04, 4.47) and non-ART siblings (HR 1.63, 95% CI 1.02, 2.61). The risk of central nervous system tumors was increased among ART children in the autologous-fresh group (HR 1.68, 95% CI 1.14, 2.48), donor-fresh group (HR 2.57, 95% CI 1.04, 6.32) and non-ART siblings (HR 1.84, 95% CI 1.12, 3.03). ART children in the autologous-fresh group were also at increased risk for solid tumors (HR 1.39, 95% CI 1.09, 1.77). A total of 127 children had both major birth defects and cancer, of which 53 children (42%) had leukemia. The risk of cancer had two independent components: (i) method of conception (described above) and (ii) presence, type and number of birth defects. The presence of nonchromosomal defects increased the cancer risk, greater for two or more defects versus one defect, for all cancers and each type evaluated. The presence of chromosomal defects was strongly associated with cancer risk (HR 8.70 for all cancers and HR 21.90 for leukemia), further elevated in the presence of both chromosomal and nonchromosomal defects (HR 21.29 for all cancers, HR 64.83 for leukemia and HR 4.71 for embryonal tumors). Among the 83 946 children born from ART in the USA in 2019 compared to their naturally conceived counterparts, these risks translate into an estimated excess of 761 children with major birth defects, 31 children with cancer and 11 children with both major birth defects and cancer.

LIMITATIONS, REASONS FOR CAUTION

In the SART CORS database, it was not possible to differentiate method of embryo freezing (slow freezing versus vitrification), and data on ICSI were only available in the fresh embryo ART group. In the OI/IUI group, it was not possible to differentiate type of non-ART treatment utilized, and in both the ART and OI/IUI groups, data were unavailable on duration of infertility. Since OI/IUI is underreported on the birth certificate, some OI/IUI children were likely included among the naturally conceived children, which will decrease the difference between all the groups and the naturally conceived children.

WIDER IMPLICATIONS OF THE FINDINGS

The use of ART is associated with increased risks of major nonchromosomal birth defects. The presence of birth defects is associated with greater risks for cancer, which adds to the baseline risk in the ART group. Although this study does not show causality, these findings indicate that children conceived with ART, non-ART siblings, and all children with birth defects should be monitored more closely for the subsequent development of cancer.

STUDY FUNDING/COMPETING INTEREST(S)

This project was supported by grant R01 HD084377 from the National Institute of Child Health and Human Development. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Child Health and Human Development, or the National Institutes of Health, nor any of the State Departments of Health which contributed data. M.L.E. reports consultancy for Ro, Hannah, Dadi, Sandstone and Underdog; presidency of SSMR; and SMRU board member. The remaining authors report no conflict of interest.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

Births conceived with ART, the ex vivo manipulation of both sexes’ gametes to achieve conception, accounted for 2.2% of all births in the USA in 2019, a proportion which has more than doubled since 2000 (Martin et al., 2002, 2021; Toner et al., 2016; Zeger-Hochschild et al., 2017; Centers for Disease Control and Prevention, 2021). It is well-established that infertility is associated with compromised maternal and infant outcomes, including higher risks for birth defects, which is the leading cause of death in the USA among children during the first year of life, and the fifth leading cause of death among children younger than age 20 (Hansen et al., 2002, 2012, 2013; Halliday et al., 2010; Davies et al., 2012b; Luke et al., 2021; Xu et al., 2021; Centers for Disease Control and Prevention, National Center for Health Statistics, 2022). A growing body of literature indicates that ART-conceived children may also have a greater risk of childhood cancer. Some studies have reported increased risks of cancer overall (Hargreave et al., 2019), while other studies have found no overall increased risk, but increased risks of certain cancer types (Williams et al., 2013, 2018; Spector et al., 2019); not all studies confirm these associations (Gilboa et al., 2019; Pettersson et al., 2022). Since birth defects and childhood cancer are strongly associated (Johnson et al., 2017; Lupo et al., 2019), we sought to evaluate the risk of childhood cancer as a function of birth defect status and method of conception, expanding on our initial analyses (Spector et al., 2019; Luke et al., 2020, 2021). We report the results of the linkage of the Society for Assisted Reproductive Technology Clinic Outcome Reporting System (SART CORS) to birth certificates, birth defects registries, and cancer registries in four States to create the most contemporary population-based study in the USA of the association of ART and subfertility with birth defects and childhood cancer.

Materials and methods

This study linked data from birth certificates to data from birth defects registries, cancer registries, and the national ART database, the SART CORS, in four States (New York, Texas, Massachusetts and North Carolina). Data from birth certificates (2004–2013) were collected in a prior grant [NIH R01 CA151973] on the risk of childhood cancer and ART (Spector et al., 2019). Additional birth data, birth defects data and updated childhood cancer data were obtained in the current grant [NIH R01 HD084377]. New York, Texas, Massachusetts and North Carolina were chosen for the current study because they are large, and racially and ethnically diverse, with birth defect registries utilizing similar case definitions and data collected. These four States ranked #2 #3, #6 and #15 in the highest number of annual ART births in the USA, respectively, in 2018, accounting for 3.5%, 1.7%, 5.4% and 1.5% of all births in each State (Martin et al., 2019; Sunderam et al., 2022).

SART CORS data

The SART CORS contains comprehensive information on ART procedures from 86% of all clinics providing ART and more than 92% of all ART cycles in the USA. Data are collected and verified by SART and reported to the Centers for Disease Control and Prevention in compliance with the Fertility Clinic Success Rate and Certification Act of 1992 (Public Law 102-493). The Society makes data available for research purposes to entities that have agreed to comply with SART research guidelines. Patients undergoing ART at SART member clinics sign clinical consent forms that include permission to use their data for research with appropriate provisions for safeguarding confidentiality. Data are submitted by individual clinics and verified by the medical director of each clinic. Approximately 10% of clinics are audited each year to validate the accuracy of reported data (Centers for Disease Control and Prevention, 2021). During each audit visit, data reported by the clinic are compared with information recorded in the medical record; most data fields have discrepancy rates <2%.

Fertility treatment data

ART represents only a small portion of all infertility treatments used in the USA. The National Survey of Family Growth reported that infertility services included medical advice (29%), infertility testing (27%), ovulation drugs (20%), artificial insemination (7.4%), surgery or treatment for blocked Fallopian tubes (3.2%) and ART (3.1%) (Chandra et al., 2014). Identifying non-ART treatments is challenging, as there is no national registry for these therapies in the USA. In the 2003 revision of the US Standard Certificate of Live Birth, a checkbox was added to indicate that the pregnancy resulted from infertility treatment (worded as: if yes, check all that apply): (A) Fertility-enhancing drugs, artificial insemination, or intrauterine insemination; (B) Assisted reproductive technology (e.g. ART (in vitro fertilization), GIFT (gamete intrafallopian transfer)). Of the four States in this study, Massachusetts has collected data on infertility treatment on its birth certificate since 1996 and adopted the other items in the 2003 revision in 2012; Texas adopted the revision in 2005; New York State in 2004 and New York City in 2008 (New York City maintains a separate birth registry); and North Carolina in 2010. Births which linked to the SART CORS cycles were categorized as ART; births with an indication that they resulted from infertility treatment (via any infertility checkbox on the birth certificate) but that did not link to an ART cycle in the SART CORS were categorized as ovulation induction/IUI (OI/IUI); the remaining births were categorized as naturally conceived. Since <1% of births were checked as OI/IUI, all births prior to implementation of the checkbox on each State’s birth certificate were labeled as naturally conceived.

Linkage procedure

This study linked ART cycles reported to the SART CORS from 1 January 2004 to 31 December 2017 that resulted in live births in 2004–2018 in Massachusetts and North Carolina and in 2004–2017 in New York and Texas, to the birth certificates, birth defects registries and cancer registries in these four States. Each State received a SART CORS file with identifiers for women with ART cycles resulting in a live birth who were residents of that State during the study time period. These identifiers included the woman’s first, middle and last names, her date of birth, social security number, State and zip code of residency, and for the index live birth, the date of delivery, plurality and birthweight(s) and sex(es) of the infant(s). The States linked the SART CORS data to birth certificate data to identify the ART-conceived births; 91% of the mothers were linked to birth certificates in their State of residence. (Based on data from SART CORS ∼12.5% of the clinics for ART-conceived births are out of the mother’s state of residence which may be associated with the likelihood of an out-of-state delivery.) Only in-state deliveries are included in this study. The States then matched ART mothers to all study years to identify siblings of the ART birth reported to the SART CORS (ART sibling group); mothers may have more than one birth in the SART CORS; all live births would be classified as ART. Supplementary Fig. S1 shows the details of the linking procedure.

For each ART-conceived delivery, we requested that the subsequent 10 deliveries (all liveborn infants from a pregnancy) be selected as the non-ART comparison group. Each study child (naturally conceived, OI/IUI-conceived, non-ART siblings and ART-conceived) was then linked to their respective State’s birth defects registry and cancer registry. The linked files were de-identified before being sent to the investigators. We then linked the de-identified files to ART treatment parameters from the SART CORS by the use of unique research identifiers to create the final analytic file.

This study was approved by the Institutional Review Boards (IRB) at Michigan State University, the University of Michigan and the New York, Texas and Massachusetts State Departments of Health; the North Carolina Department of Health and Human Services only required that IRB approval be given by Michigan State University and the University of Michigan. The Michigan State University IRB determined that this research did not involve human subjects, as defined in 45 CFR 46.102 (f), in review dated 13 November 2015.

Birth defects

The four States participating in this project are current or former Centers for Disease Control (CDC) Centers for Birth Defects Research and Prevention. As such, they conduct enhanced birth defects surveillance in terms of scope and quality of data. Each State conducts active or a combination of active and passive population-based surveillance that includes the major birth defects. These States employ standard case definitions as defined by the National Birth Defects Prevention Study and National Birth Defects Prevention Network (NBDPN) and code birth defects using the CDC coding system adapted from the British Pediatric Association codes, which is more specific for birth defects than ICD-9 or ICD-10 coding (Supplementary Table SI) (National Birth Defects Prevention Network (NBDPN), 2004). For this study, we analyzed selected major birth defects diagnosed within the first year of life, as defined in Supplementary Table SI. We then classified individuals with major birth defects as either ‘chromosomal’ (i.e. presence of a chromosomal defect with or without any other major defect) or ‘nonchromosomal’ (i.e. presence of a major defect but having no chromosomal defect). Of the families that were identified, only 0.5% of the families of children with birth defects had more than one delivery where a birth defect was reported; 75% of these occurrences would be expected randomly based on the frequency of birth defects; therefore, no adjustment for repeat deliveries was necessary.

Blastogenesis defects

We chose also to include birth defects classified as a group by Halliday et al. (2010) as blastogenesis defects, defined on the basis of pathologic development rather than by organ system. This allowed us to define defects that were expected to originate within the first 4 weeks of gestation, excluding cardiac defects, which we evaluated separately. Disorders of blastogenesis in the current study were defined as the presence of one or more of the following: abdominal wall defects, vertebral segmentation defects, tracheoesophageal fistula, diaphragmatic defects, neural tube defects, anal atresia, renal agenesis, caudal regression sequence, laterality defects, sirenomelia, sacrococcyeal teratoma, holoprosencephaly, acro-renal field defect and ammelia.

Defects in twin pairs

We examined the incidence of birth defects in twins and found that the rate of twins both having a major birth defect or both having the same type of birth defect (except central nervous system (CNS)) was greater than expected at random. For example, there were 174 pairs of twins where both had cardiovascular system defects. Based on the frequency of the defect, only six such pairs were expected. Therefore, when modeling the probability of a birth defect, we excluded the second twin if both twins had the same type of birth defect (or when analyzing major defects, both had a major defect.) Excluded twin children constituted <1.1% of each defect. We compared the estimates of the odds ratios using all the data with the odds ratios after excluding the second twin when both had a defect: except for plurality, the differences in the estimates of the odds ratios were in the 2nd or 3rd decimal place; the estimate of the effect of plurality is now slightly underestimated since some pairs of defects would randomly occur (using all the data, the effect of plurality is overestimated since too many pairs of defects are included in the modeling).

Cancer

Each of the four study States maintains a high-quality cancer registry dating back at least as far as 2004, our earliest year of cancer linkage. All of the study States’ cancer registries are part of the CDC-funded National Program of Cancer Registries, and were certified at the highest levels for data during the study years. Cancers were categorized according to the International Classification of Childhood Cancers, third edition^, (ICCC-3) updated in 2017 (Supplementary Table SII) (Steliarova-Foucher et al., 2022). Examined as a single tumor class were embryonal cancers [including neuroblastoma, retinoblastoma, nephroblastoma (Wilms tumor), hepatoblastoma, embryonal rhabdomyosarcoma, pulmonary and pleuropulmonary blastoma, medulloblastoma, primitive neuroectodermal tumors, medulloepithelioma and atypical teratoid rhabdoid tumors] and solid tumors (all cancers excluding leukemia and lymphoma). Children born in 2004–2013 were initially linked to State cancer registries through December 2013 (prior grant, R01 CA151973). All the children were relinked in 2020–2021 (the final year of the current study). Although 79 children had more than one cancer diagnosis, only in seven children did the second diagnosis differ from the first. In these cases, since one diagnosis preceded the other in age at diagnosis, it was considered the primary cancer and was the only cancer included in the analysis. There were only four cases where both twins had cancer and only one family where cancer occurred in more than one delivery; therefore, we included all of these cancers in our analysis. All study children were also linked to State death records.

Conception groups

As described above, four groups of births were defined based on the absence/presence of subfertility/infertility and the method of conception. The non-ART comparison group births were categorized as naturally conceived or OI/IUI. The other two groups were ART and their non-ART siblings. The ART births were further divided into four subgroups depending on the combination of oocyte source (autologous or donor) and embryo state (fresh or thawed), based on our prior analyses indicating associations of these combinations with adverse perinatal outcomes, including birth defects (Luke et al., 2016b, 2017, 2020, 2021; Liberman et al., 2017).

Independent variables

Independent variables were selected a priori for inclusion in the models based on established associations with birth defects, cancer and/or ART. These included paternal age at delivery (grouped as 18–34, 35–40 and ≥41 years); maternal age at delivery (18–29, 30–34, 35–37, 38–40, 41–43 and ≥44 years), race (White, Black, Asian, other or missing), Hispanic ethnicity, education (less than high school graduate, high school graduate or general educational development, some college or associate degree, bachelor’s degree, post-graduate education, or missing), parity (nulliparous, primiparous or multiparous prior to the index pregnancy), BMI (weight/height²) (≤24, underweight or normal weight; 25–29, overweight; and ≥30, obese or missing) calculated from height and pre-pregnancy weight reported on the birth certificate, diabetes (pregestational and/or gestational), hypertension (chronic/pregestational and/or gestational and/or eclampsia) and infant sex, plurality (singleton or twin) as well as State and year of birth. Birthweight z-score was calculated as [(actual weight—reference weight)/standard deviation for the reference population], as recommended by Land (2006), using sex-specific national standards (Talge et al., 2014). Infants with z-scores of ≤−1.28 were categorized as small-for-gestation and infants with z-scores of ≥1.28 were categorized as large-for-gestation. ART factors and treatment parameters included infertility diagnoses (male factor, endometriosis, ovulation disorders, diminished ovarian reserve, tubal ligation, other tubal factors, uterine factor, unexplained, other (immunologic, chromosomal or other serious disease) and other-noninfertile (single woman or same-sex partners)); the use of ICSI (which was only available for fresh IVF cycles); sperm source was limited to partner. Singleton and twin births were analyzed together, with a variable of plurality in the models, indicating the risk of twins compared to singletons. Triplets and higher-order multiples were excluded. Descriptive data are shown in Tables I–III; birth defects data are shown in Table IV; cancer data are shown in Table V; and cancer risks by birth defect status and conception group are shown in Table VI.

Table I.

Characteristics of the study population: singleton births.

		Naturally conceived	OI/IUI conceived	Non-ART siblings	ART by oocyte source and embryo state
					Autologous		Donor
					Fresh	Thawed	Fresh	Thawed
	All children, N	1 315 497	8875	30 036	55 494	33 835	7162	4828
Maternal Race	White	74.7	85.2	85.7	82.9	78.4	83.5	81.7
(%)	Black	16.7	5.3	4.9	6.0	6.7	6.9	7.8
	Asian	8.6	9.5	9.5	11.0	14.9	9.6	10.5
	Other/missing	7.4	2.7	2.9	2.6	3.1	3.2	2.9
Maternal Ethnicity (%)	Hispanic	28.5	9.5	10.9	11.5	9.8	11.4	10.9
Maternal age	Mean years (SD)	29.1 ± 5.8	33.7 ± 5.0	33.4 ± 4.8	34.9 ± 4.2	35.2 ± 4.2	42.2 ± 4.8	42.8 ± 5.3
(%)	18–29	52.1	20.3	19.8	10.4	8.8	1.3	1.6
	30–34	29.0	37.3	37.6	35.9	35.1	6.5	6.1
	35–37	11.2	20.3	22.6	25.2	26.5	8.0	8.3
	38–40	5.5	13.1	13.9	18.8	19.0	15.2	12.8
	41–43	1.8	6.5	4.8	8.8	8.8	24.9	21.0
	≥44	0.3	2.5	1.1	0.9	1.8	44.1	50.2
Paternal age	Mean years (SD)	31.9 ± 6.8	35.7 ± 6.2	35.6 ± 5.7	37.3 ± 5.9	37.7 ± 5.9	42.8 ± 6.6	43.5 ± 7.0
(%)	18–34	66.9	46.2	43.9	34.1	30.5	9.9	9.1
	35–40	23.0	34.8	38.6	40.8	42.9	26.9	25.4
	≥41	10.1	19.1	17.4	25.1	26.6	63.2	65.5
	Missing	10.9	5.7	2.0	0.7	2.2	1.2	8.5
Maternal Education (%)	<High school	15.1	1.7	1.9	1.3	1.8	1.5	1.4
	High school graduate	24.2	8.0	8.0	7.6	5.7	6.5	5.9
	Some college/Assoc. Degree	27.5	21.5	17.4	19.9	18.4	19.1	20.4
	Bachelor’s degree	20.2	34.7	38.3	38.2	38.6	38.1	36.7
	Post-graduate degree	13.0	34.2	34.5	33.1	35.5	34.8	35.7
	Missing	1.3	0.3	1.5	1.3	1.0	1.4	1.2
Parity (%)	0	40.6	64.6	44.1	69.4	55.2	69.5	52.3
	1	32.7	26.9	32.8	23.4	32.7	21.9	32.9
	≥2	26.7	8.6	23.0	7.2	12.2	8.7	14.7
BMI (kg/m2)	Mean (SD)	26.3 ± 6.3	26.6 ± 6.5	24.4 ± 5.1	25.2 ± 5.5	25 ± 5.4	25.2 ± 5.3	25.5 ± 5.6
(%)	12–24	51.4	51.1	65.6	59.5	61.0	58.7	56.6
	25–29	25.8	24.2	21.4	23.4	22.9	24.8	25.6
	30–59	22.8	24.7	13.0	17.1	16.1	16.4	17.7
	Missing	36.8	36.4	42.6	44.9	20.8	42.7	28.3
Diabetes (%)	Pre-gestational or gestational	5.7	10.0	5.4	7.3	8.2	8.9	10.0
Hypertension (%)	Pre-gestational or gestational	5.8	9.7	5.1	6.7	9.1	13.0	14.0
Length of gestation	Mean weeks (SD)	38.7 ± 1.9	38.5 ± 2.3	38.6 ± 2	38.4 ± 2.2	38.5 ± 2.2	38.2 ± 2.3	37.9 ± 2.4
(%)	<28 weeks	0.5	0.9	0.8	0.7	0.7	0.7	1.1
	28–32 weeks	1.1	1.8	0.9	1.7	1.4	2.2	2.5
	33–36 weeks	6.1	7.4	6.3	8.5	8.0	12.1	12.4
	≥37 weeks	92.4	89.8	92.0	89.0	89.8	85.0	84.0
Birthweight	Mean g (SD)	3310 ± 554	3285 ± 623	3339 ± 565	3235 ± 601	3369 ± 602	3233 ± 634	3221 ± 652
(%)	300–999 g	0.5	0.9	0.7	0.8	0.7	0.7	1.1
	1000–1499 g	0.5	1.0	0.4	0.9	0.7	1.4	1.0
	1500–2499 g	5.0	6.2	4.5	7.3	5.2	8.5	9.2
	≥2500 g	94.0	91.9	94.4	91.0	93.5	89.4	88.7
Birthweight Z-score	Mean (SD)	−0.02 ± 0.96	0.01 ± 0.97	0.09 ± 0.96	−0.07 ± 0.96	0.19 ± 0.97	0.05 ± 0.99	0.11 ± 0.99
(%)	SGA (Z-score ≤−1.28)	8.3	8.6	6.5	9.6	5.4	7.9	7.0
	LGA (Z-score ≥1.28)	8.9	9.7	10.6	8.1	12.8	9.9	11.3
Infant sex (%)	Males	51.2	51.9	51.6	51.3	52.4	51.4	51.2
Infant death (%)	By 1 year	0.4	0.4	0.9	0.4	0.3	0.3	0.4

OI/IUI, ovulation induction/IUI; SGA, small for gestational age; LGA, large for gestational age. Values in italics are the percent missing.

Table II.

Characteristics of the study population: twin births.

		Naturally conceived	OI/IUI conceived	Non-ART siblings	ART by oocyte source and embryo state
					Autologous		Donor
					Fresh	Thawed	Fresh	Thawed
Variables*	Children, N	37 943	3576	1488	40 119	13 881	7570	2236
	Pregnancies with live births, N	23 363	2038	787	20 313	7014	3831	1130
Maternal Race	White	72.7	83.9	86.8	84.1	77.4	83.9	81.6
(%)	Black	21.1	6.7	6.0	5.8	9.3	7.2	9.2
	Asian	6.2	9.4	7.3	10.2	13.3	9.0	9.2
	Other/missing	6.2	3.1	4.0	3.0	3.6	2.9	3.4
Maternal Ethnicity (%)	Hispanic	24.0	10.0	13.5	12.6	12.6	13.1	11.7
Maternal age	Mean years (SD)	30.0 ± 5.6	34.0 ± 5.0	33.3 ± 5.2	33.9 ± 4.0	34.1 ± 4.1	41.6 ± 4.9	42.1 ± 5.3
(%)	18–29	45.0	17.0	23.5	14.0	13.2	1.7	2.2
	30–34	32.1	40.1	38.6	41.4	41.1	7.9	6.7
	35–37	13.9	20.4	19.5	25.2	24.9	9.6	9.7
	38–40	6.5	13.3	10.5	15.0	15.1	15.7	14.8
	41–43	1.9	5.6	3.9	4.2	4.7	26.2	23.0
	≥44	0.5	3.7	4.0	0.2	0.9	39.0	43.5
Paternal age	Mean years (SD)	32.8 ± 6.7	36.1 ± 6.2	35.8 ± 6.4	36.4 ± 5.7	36.7 ± 5.8	42.0 ± 6.6	42.7 ± 7.0
(%)	18–34	62.3	43.5	46.8	39.8	37.1	11.9	10.2
	35–40	26.1	34.7	30.2	39.5	38.9	29.6	25.8
	≥41	11.6	17.9	19.8	20.1	21.8	57.3	55.8
	Missing	12.4	3.9	3.2	0.6	2.2	1.2	8.1
Maternal Education (%)	<High school	12.6	2.2	1.0	1.4	1.9	1.5	1.8
	High school graduate	22.3	7.7	8.0	8.3	7.0	7.6	6.3
	Some college/Assoc. degree	28.3	21.3	18.2	20.6	22.8	19.0	21.7
	Bachelor’s degree	22.0	34.7	37.9	38.9	37.8	38.2	39.0
	Post-graduate degree	14.8	34.1	34.9	30.9	30.5	33.6	31.2
	Missing	1.5	0.6	3.0	1.3	1.2	1.7	2.4
Parity (%)	0	23.9	41.5	29.9	42.1	37.0	41.0	35.7
	1	35.1	41.4	40.7	42.2	41.6	41.2	40.6
	≥2	41.0	17.0	29.4	15.7	21.4	17.8	23.7
BMI (kg/m2)	Mean (SD)	27.2 ± 6.7	26.2 ± 6.5	24.9 ± 5.6	25.2 ± 5.4	25.7 ± 5.8	25.2 ± 5.2	26.3 ± 5.8
(%)	12–24	46.2	53.7	60.5	59.5	56.0	59.0	49.8
	25–29	25.8	22.4	25.1	23.6	24.5	24.5	29.1
	30–59	28.0	23.8	14.4	16.8	19.6	16.5	21.1
	Missing	39.1	42.0	45.8	47.1	24.3	44.5	31.0
Diabetes (%)	Pre-gestational or gestational	7.0	12.0	6.0	8.4	10.1	10.5	12.2
Hypertension (%)	Pre-gestational or gestational	11.4	16.5	10.6	12.1	17.5	23.3	24.7
Length of gestation	Mean weeks (SD)	35.3 ± 3.1	35.2 ± 3.2	33.8 ± 4.7	35.3 ± 3.0	35.2 ± 3.0	35.1 ± 2.9	34.9 ± 3.0
(%)	<28 weeks	3.7	4.3	13.3	3.2	3.7	2.9	3.4
	28–32 weeks	10.3	10.8	11.7	10.4	10.2	11.1	13.0
	33–36 weeks	43.6	43.0	41.7	44.8	45.5	50.4	49.4
	≥37 weeks	42.4	41.8	33.2	41.6	40.6	35.6	34.3
Birthweight	Mean g (SD)	2357 ± 609	2341 ± 626	2170 ± 795	2354 ± 594	2418 ± 609	2339 ± 587	2317 ± 605
(%)	300–999 g	3.8	4.2	12.7	3.3	3.3	3.0	3.3
	1000–1499 g	5.4	6.3	5.9	5.5	5.0	5.5	6.8
	1500–2499 g	46.2	44.4	40.1	47.1	42.0	49.6	48.3
	≥2500 g	44.6	45.1	41.3	44.0	49.6	42.0	41.5
Birthweight Z-score	Mean (SD)	−0.64 ± 0.87	−0.63 ± 0.83	−0.50 ± 0.88	−0.64 ± 0.87	−0.44 ± 0.85	−0.57 ± 0.88	−0.54 ± 0.89
(%)	SGA (Z-score ≤−1.28)	22.3	21.1	17.5	22.0	15.3	19.9	19.6
	LGA (Z-score ≥1.28)	1.5	1.0	1.5	1.4	2.2	1.7	1.5
Infant sex (%)	Males	49.9	51.3	51.6	50.7	51.2	51.4	52.1
Infant death (%)	By 1 year	1.7	1.8	9.5	1.2	1.1	1.0	1.1

^* Percents and means for all variables from maternal race to length of gestation calculated by pregnancy; percents and means for all variables from birthweight to infant death calculated by child. OI/IUI, ovulation induction/IUI; SGA, small for gestational age; LGA, large for gestational age. Values in italics are the percent missing.

Table III.

Characteristics of the ART groups by oocyte source, embryo state and plurality at birth.

		Singleton births				Twin births
	Oocyte source	Autologous		Donor		Autologous		Donor
	Embryo state	Fresh	Thawed	Fresh	Thawed	Fresh	Thawed	Fresh	Thawed
	Children, N	55 494	33 835	7162	4828	40 119	13 881	7570	2236
	Pregnancies/treatment cycle	55 494	33 835	7162	4828	20 312	7014	3831	1130
Factor*	Categories
Diagnoses (%)	Male factor	37.7	34.4	17.0	15.5	39.0	36.6	17.5	17.7
	Endometriosis	10.8	8.7	6.2	5.6	11.6	10.0	7.0	6.5
	Ovulation disorders	15.4	15.9	3.3	3.6	18.0	19.6	4.6	3.6
	Diminished Ovarian Res.	16.1	13.9	78.4	74.7	11.5	9.5	77.4	73.1
	Tubal ligation	1.7	1.3	0.9	0.9	2.2	2.1	1.6	0.9
	Tubal-hydrosalpinx	1.4	1.3	0.8	0.5	1.5	1.6	0.6	0.8
	Tubal-other	16.5	14.4	7.5	7.3	17.1	17.4	8.6	8.4
	Uterine factor	4.3	6.2	6.1	6.9	4.2	7.4	6.5	6.1
	Unexplained	17.1	16.0	4.7	5.0	16.7	15.8	4.4	6.1
	Other-RFA**	9.2	18.7	15.4	20.4	8.5	12.9	14.7	17.3
	Other-noninfertile	0.3	0.8	0.3	1.5	0.2	0.7	0.3	1.3
	Other-PGD	1.4	10.2	1.2	4.0	0.9	5.7	0.9	2.2
# Diagnoses (%)	One	76.0	74.4	68.3	68.9	76.4	73.6	66.8	67.9
	Two or more	24.0	25.6	31.7	31.1	23.6	26.4	33.2	32.1
	Missing	0.8	2.3	0.1	0.4	0.6	1.9	0.3	0.2
ICSI	None	31.3	—	26.1	—	29.9	—	28.3	—
	Some oocytes	5.9	—	5.8	—	6.6	—	6.8	—
	All oocytes	62.8	—	68.0	—	63.5	—	64.8	—
	Missing	0.3	—	0.0	—	0.2	—	0.1	—

^*Percents calculated by pregnancy/treatment cycle.

^**Other-RFA is reason for ART, including immunologic, chromosomal or other serious disease.

Values in italics are the percent missing.

Table IV.

Rates and risks of birth defects by method of conception.*

			Naturally	OI/IUI	Non-ART	ART by oocyte source and embryo state				Plurality	Sex
		Plurality	Conceived	Conceived	Siblings	Autologous		Donor		Twins vs singletons	Males vs females
						Fresh	Thawed	Fresh	Thawed
	Children, N	Singletons	1 315 497	8875	30 036	55 494	33 835	7162	4828
		Twins	37 943	3576	1488	40119	13881	7570	2236
		All children	1 353 440	12 451	31 524	95 613	47 716	14 732	7064
Number of birth defects	Nonchromosomal	All children	25 470	315	645	2573	1376	405	214
	Chromosomal	All children	1966	21	64	199	64	7	5
Type of defect
Major defect** (Nonchromosomal)	Rate****	Singletons	18.4	21.9	20.0	23.2	25.3	23.3	27.3
		Twins	34.1	33.8	30.2	32.0	37.5	31.4	36.7
	AOR		1	1.21	1.20	1.24	1.23	1.05	1.20	1.52	1.50
	95% CI		Reference	1.08, 1.36	1.11, 1.31	1.18, 1.30	1.15, 1.30	0.93, 1.18	1.03, 1.40	1.45, 1.58	1.47, 1.54
Blastogenesis	Rate	Singletons	2.2	1.9	2.1	2.8	2.5	2.8	4.8
		Twins	3.1	4.5	5.4	3.3	4.2	3.2	2.2
	AOR		1	1.28	1.22	1.41	1.37	1.28	1.74	1.41	1.16
	95% CI		Reference	0.90, 1.82	0.95, 1.55	1.22, 1.63	1.14, 1.66	0.90, 1.83	1.14, 2.65	1.23, 1.60	1.08, 1.24
Cardiovascular	Rate	Singletons	10.4	9.8	11.4	13.1	14.6	13.1	12.6
		Twins	22.3	17.6	17.5	19.4	23.6	19.6	23.7
	AOR		1	1.04	1.25	1.23	1.26	1.05	1.08	1.73	0.95
	95% CI		Reference	0.88, 1.23	1.12, 1.40	1.15, 1.31	1.17, 1.37	0.90, 1.23	0.87, 1.33	1.63, 1.83	0.92, 0.98
Central nervous system	Rate	Singletons	0.4	0.2	0.3	0.3	0.3	0.4	0.0
		Twins	0.7	0.8	0.0	0.6	0.8	0.4	0.9
	AOR		1	1.05	0.91	1.02	1.11	0.89	0.64	1.76	1.06
	95% CI		Reference	0.43, 2.57	0.47, 1.78	0.70, 1.48	0.69, 1.76	0.35, 2.32	0.15, 2.84	1.31, 2.37	0.91, 1.23
Gastrointestinal	Rate	Singletons	0.9	1.6	1.0	1.8	1.5	1.3	2.3
		Twins	1.4	2.8	2.0	2.0	2.2	2.4	0.9
	AOR		1	2.01	1.28	1.85	1.79	1.51	1.58	1.40	1.02
	95% CI		Reference	1.32, 3.07	0.89, 1.83	1.52, 2.26	1.39, 2.30	0.94, 2.42	0.85, 2.92	1.16, 1.68	0.92, 1.13
Musculoskeletal	Rate	Singletons	2.9	3.3	3.0	3.6	2.9	3.8	3.5
		Twins	4.3	5.0	4.7	4.6	5.0	3.7	5.4
	AOR		1	1.26	1.20	1.28	1.10	1.28	1.48	1.46	1.45
	95% CI		Reference	0.94, 1.69	0.98, 1.48	1.13, 1.45	0.93, 1.30	0.93, 1.74	0.99, 2.22	1.30, 1.63	1.37, 1.54
Genitourinary (males only)	Children, N	Singletons	673 831	4606	15 485	28 459	17 734	3683	2472
		Twins	18 941	1834	768	20 355	7113	3889	1166
	Rate	Singletons	7.3	11.5	8.9	10.4	11.7	10.6	14.2
		Twins	11.6	17.4	14.3	12.9	12.7	10.5	13.7
	AOR		1	1.40	1.15	1.22	1.22	1.00	1.37	1.26	—
	95% CI		Reference	1.12, 1.75	0.97, 1.36	1.10, 1.36	1.07, 1.39	0.77, 1.30	1.00, 1.87	1.15, 1.39	—
Orofacial	Rate	Singletons	1.4	2.4	1.5	2.0	1.7	1.1	2.1
		Twins	1.7	0.8	0.7	1.7	1.9	1.5	0.9
	AOR		1	1.42	1.01	1.40	1.26	0.83	1.04	0.93	1.20
	95% CI		Reference	0.94, 2.14	0.75, 1.38	1.17, 1.68	1.00, 1.60	0.49, 1.39	0.56, 1.94	0.78, 1.11	1.11, 1.31
Chromosomal	Rate	Singletons	1.5	1.9	2.0	1.9	1.2	0.4	0.4
		Twins	1.0	1.1	2.0	2.3	1.6	0.5	1.3
	AOR		1	0.81	1.08	1.02	0.61	0.06	0.08	1.00	0.96
	95% CI		Reference	0.52, 1.26	0.83, 1.39	0.86, 1.20	0.47, 0.79	0.03, 0.13	0.03, 0.21	0.84, 1.19	0.88, 1.04
All major*** (Chromosomal and Nonchromosomal)	Rate	Singletons	19.8	23.8	22.0	25.1	26.5	23.7	27.8
		Twins	35.1	35.0	32.3	34.3	39.0	32.0	38.0
	AOR		1	1.17	1.19	1.21	1.16	0.80	0.92	1.48	1.45
	95% CI		Reference	1.05, 1.31	1.10, 1.29	1.15, 1.27	1.09, 1.23	0.71, 0.90	0.79, 1.06	1.42, 1.55	1.42, 1.48

^*Models included conception groups, IVF groups, paternal and maternal ages, maternal race and ethnicity, education, BMI, parity, diabetes, hypertension, plurality, infant sex and State and year of birth.

^**Major defects are limited to nonchromosomal only as defined by the National Birth Defects Prevention Network (see Supplementary Table SI).

^***All major includes both chromosomal and nonchromosomal as defined by the National Birth Defects Prevention Network (see Supplementary Table SI).

^****Rate is per 1000 children. Bolded values are significant at P < 0.05. OI/IUI, ovulation induction/IUI; AOR, adjusted odds ratio.

Table V.

Rates* and risks of childhood cancers by method of conception.**

		Naturally conceived	OI/IUI conceived	Non-ART siblings	ART by oocyte source and embryo state				Plurality	Sex
					Autologous		Donor		Twins vs singletons	Males vs females
					Fresh	Thawed	Fresh	Thawed
Children	All children, N	1 353 440	12 451	31 524	95 613	47 716	14 732	7064
	Children with cancer, N	1469	19	59	165	50	16	9
Age at first cancer	Years (mean ± SD)	3.2 ± 3.0	4.5 ± 2.9	3.9 ± 3.2	3.3 ± 3.0	2.0 ± 2.1	2.2 ± 2.3	2.4 ± 3.4
	%, <1 year	22.9	5.3	15.3	23.0	38.0	31.3	55.6
	1–4 years	54.0	47.4	52.5	52.1	50.0	43.8	22.2
	5–14 years	23.1	47.4	32.2	24.8	12.0	25.0	22.2
Follow-up	Mean years (±SD)	6.1 ± 3.8	5.6 ± 3.8	7.4 ± 3.8	7.2 ± 3.6	3.9 ± 3.4	7.0 ± 3.6	4.7 ± 3.6
	Median years (IQR)	5.7 (2.7, 9.2)	5.1 (2.4, 8.4)	7.5 (4.5, 10.4)	7.2 (4.4, 10.0)	2.9 (1.4, 5.7)	7.0 (4.1, 10.0)	3.9 (1.8, 7.1)
	Person-years	8 211 752	69 939	234 141	684 186	187 953	103 294	33 399
	%, <1 year	9.3	11.3	5.9	4.4	19.7	4.8	13.6
	1–4 years	34.8	37.4	21.8	26.1	50.4	28.0	46.5
	5–9 years	36.3	34.8	44.2	44.1	21.9	42.3	28.5
	10–14 years	19.5	16.4	28.2	25.4	8.0	24.9	11.3
ALL CANCERS	Rate	1.09	1.53	1.87	1.73	1.05	1.09	1.27
	HR	1	1.42	1.34	1.31	1.28	0.77	1.23	0.97	1.18
	95% CI	Reference	0.89, 2.27	1.02, 1.76	1.08, 1.59	0.95, 1.73	0.43, 1.37	0.59, 2.57	0.79, 1.18	1.07, 1.29
LEUKEMIA	Rate	0.34	0.64	0.63	0.45	0.29	0.20	0.28
	HR	1	2.15	1.63	1.17	1.28	0.54	1.08	1.01	1.19
	95% CI	Reference	1.04, 4.47	1.02, 2.61	0.81, 1.71	0.73, 2.26	0.15, 1.93	0.24, 4.99	0.70, 1.46	1.00, 1.42
CNS TUMORS	Rate	0.23	0.08	0.57	0.44	0.15	0.54	0.00
	HR	1	0.37	1.84	1.68	0.98	2.57	—	0.75	1.21
	95% CI	Reference	0.05, 2.74	1.12, 3.03	1.14, 2.48	0.45, 2.16	1.04, 6.32	—	0.48, 1.18	0.99, 1.49
EMBRYONAL TUMORS	Rate	0.34	0.40	0.38	0.46	0.46	0.61	0.28
	HR	1	1.07	0.89	1.01	1.47	1.16	0.72	1.30	1.05
	95% CI	Reference	0.43, 2.67	0.49, 1.60	0.70, 1.46	0.92, 2.35	0.51, 2.64	0.16, 3.22	0.93, 1.82	0.88, 1.24
SOLID TUMORS	Rate	0.65	0.80	1.11	1.13	0.67	0.88	0.85
	HR	1	1.19	1.30	1.39	1.30	1.06	1.38	0.97	1.07
	95% CI	Reference	0.62, 2.26	0.91, 1.84	1.09, 1.77	0.89, 1.90	0.55, 2.04	0.56, 3.38	0.75, 1.25	0.94, 1.20

^*Rates are per 1000 children.

^**Models included conception groups, IVF groups, paternal and maternal ages, maternal race and ethnicity, education, BMI, parity, diabetes, hypertension, plurality, infant sex and State and year of birth. IQR, interquartile range; OI/IUI, ovulation induction/IUI; HR, hazard ratio.

Embryonal tumors include neuroblastoma, retinoblastoma, nephroblastoma (Wilms tumor), hepatoblastoma, embryonal rhabdomyosarcoma, pulmonary and pleuropulmonary blastoma, medulloblastoma, primitive neuroectodermal tumors, medulloepithelioma and atypical teratoid rhabdoid tumors.

Solid tumors include all cancers excluding leukemia and lymphoma.

Bolded values are significant at P < 0.05.

Table VI.

Rates* and hazard ratios (HR) and 95% CI** of childhood cancer*** by birth defect status, number and type**** and conception group.

					ART				Birth defect status, number, type****
		Naturally conceived	OI/IUI conceived	Non-ART siblings	Autologous		Donor			One	≥Two	C	Both
					Fresh	Thawed	Fresh	Thawed	None	NC	NC	Only	C & NC
All children, N		1 353 440	12 451	31 524	95 613	47 716	14 732	7064	1 529 216	865	28 842	2156	1461
Children with cancer, N		1469	19	59	165	50	16	9	1660	8	82	9	28
Cancer group or
Tumor class***
All cancers	Rate*	1.09	1.53	1.87	1.73	1.05	1.09	1.27	1.09	9.25	2.84	4.17	19.16
	HR	1	1.41	1.34	1.30	1.29	0.83	1.31	1	2.75	4.48	8.70	21.29
	95% CI	Reference	0.88, 2.25	1.02, 1.76	1.07, 1.58	0.95, 1.74	0.47, 1.47	0.63, 2.72	Reference	2.19, 3.45	2.30, 8.75	4.27, 17.7	14.48, 31.28
Leukemia	Rate*	0.34	0.64	0.63	0.45	0.29	0.20	0.28	0.32	6.94	0.66	1.39	17.11
	HR	1	2.13	1.63	1.17	1.33	0.67	1.29	1	2.14	5.11	21.9	64.83
	95% CI	Reference	1.03, 4.41	1.02, 2.61	0.80, 1.70	0.75, 2.34	0.19, 2.36	0.29, 5.85	Reference	1.34, 3.42	1.60, 16.28	9.59, 50.02	42.57, 98.71
CNS	Rate*	0.23	0.08	0.57	0.44	0.15	0.54	—	0.25	1.16	0.35	1.39	—
	HR	1	0.37	1.84	1.67	0.98	2.57	—	1	1.49	6.67	4.94	—
	95% CI	Reference	0.05, 2.73	1.11, 3.03	1.13, 2.47	0.44, 2.14	1.04, 6.32	—	Reference	0.78, 2.83	2.09, 21.29	0.67, 36.70	—
Embryonal tumors	Rate*	0.34	0.40	0.38	0.46	0.46	0.61	0.28	0.33	1.16	1.21	1.86	1.37
	HR	1	1.06	0.88	1.00	1.46	1.18	0.72	1	3.76	6.27	3.53	4.71
	95% CI	Reference	0.43, 2.64	0.48, 1.59	0.69, 1.44	0.91, 2.32	0.52, 2.66	0.16, 3.23	Reference	2.65, 5.34	2.29, 17.13	0.48, 26.22	1.14, 19.48
Solid tumors	Rate*	0.66	0.80	1.11	1.13	0.67	0.88	0.85	0.67	2.31	1.91	2.78	1.37
	HR	1	1.18	1.29	1.38	1.29	1.06	1.37	1	2.98	4.79	3.54	2.44
	95% CI	Reference	0.62, 2.25	0.91, 1.83	1.08, 1.75	0.88, 1.88	0.55, 2.04	0.56, 3.37	Reference	2.26, 3.94	2.11, 10.86	0.86, 14.62	0.59, 10.09

^*Rates per 1000 children.

^**Models included conception groups, IVF groups, birth defect status (including number of defects and type), paternal and maternal ages, maternal race and ethnicity, education, BMI, parity, diabetes, hypertension, plurality, infant sex and State and year of birth.

^***Cancer groups as defined in the International Classification of Childhood Cancer, 3rd edition, update 2017, as shown in Supplementary Table SII.

^****Birth defects: NC, nonchromosomal, C, chromosomal. CNS, central nervous system.

Embryonal tumors include neuroblastoma, retinoblastoma, nephroblastoma (Wilms tumor), hepatoblastoma, embryonal rhabdomyosarcoma, pulmonary and pleuropulmonary blastoma, medulloblastoma, primitive neuroectodermal tumors, medulloepithelioma, and atypical teratoid rhabdoid tumors.

Solid tumors include all cancers excluding leukemia and lymphoma.

Bolded values are significant at P < 0.05.

Statistical analysis

Data from each State were processed to generate a common dataset. We excluded children whose mother or father was younger than 18 years of age, unknown sex of child or implausible values (gestational age <22 weeks or birthweight <300 g even if indicated as a live birth). Because most independent variables were categorized, missing values were included as a separate category for maternal BMI, education and race, and father’s age. There were no missing data for other variables. Based on expected birth defects rates per 1000 live births averaged across the four study States, we expected our naturally conceived and ART study populations to provide 90% power to detect an effect size of 6–8% with a two-sided α of 0.05 for major defects and cardiovascular defects, and an effect size of 15–30% for blastogenesis, genitourinary, orofacial and gastrointestinal defects. This power analysis is appropriate for the larger groups; OI/IUI conceived, and the ART donor groups would need larger effect sizes. We used logistic regression to model the risk of a major birth defect, a major nonchromosomal birth defect (i.e. major defect not accompanied by a chromosomal defect), blastogenesis, cardiovascular, orofacial, gastrointestinal, musculoskeletal and chromosomal defects, and genitourinary defects in male children, with naturally conceived children as the reference group. Within the ART group, risks were modeled by oocyte source-embryo state combinations. As described above, if both twins had the same birth defect, only one was included in the analysis. Exposure time for study children was from their date of birth until the diagnosis of cancer, death or 31 December 2017 (Massachusetts and Texas) or 31 December 2018 (New York and North Carolina). We used Cox proportional hazards regression to calculate hazard ratios (HRs) and 95% CIs for childhood cancer risk by method of conception, and as a function of birth defect status by method of conception, with naturally conceived children without birth defects as the reference group. All analyses were performed using SAS Version 9.4 software.

Results

Characteristics of the study population

The study population included 165 125 ART-conceived children, 31 524 non-ART siblings, 12 451 children born to OI/IUI-treated women and 1 353 440 naturally conceived children. The characteristics of the study population are shown in Table I for singletons and Table II for children from twin deliveries. Women and their male partners were youngest in the naturally conceived group; ages shifted upwards in the OI/IUI and non-ART siblings group; further increasing in the ART autologous group; and were oldest in the ART donor group. Women in the naturally conceived group were more likely to be Black, of Hispanic ethnicity, have higher parity, and were less likely to be college graduates compared to the other conception groups. Women in the naturally conceived and non-ART sibling groups were about half as likely to have pre-gestational or gestational diabetes or hypertension compared to women in the other groups. Approximately 8% of the non-ART siblings were conceived with OI/IUI, although this may be underestimated due to under-reporting (Schieve et al., 2009; Thoma et al., 2014). The characteristics of the ART groups are shown in Table III for both pluralities. More than one-third of births from cycles using autologous oocytes included the diagnosis of male factor infertility, and one-sixth had a diagnosis of unexplained infertility. Three-fourths of births from cycles using donor oocytes included the diagnosis of diminished ovarian reserve. The use of ICSI was only available in the SART CORS for cycles using fresh embryos, and averaged more than 60% for deliveries from autologous and donor oocytes in both singleton and twin births.

Risk of birth defects by mode of conception

The rates of birth defects and the results of the logistic regression models are shown in Table IV. A total of 29 571 singleton children (2.0%) and 3753 twin children (3.5%) had a major birth defect (chromosomal or nonchromosomal). Compared to naturally conceived children, risks for defects were increased for all other groups for nonchromosomal (adjusted odds ratios (AORs) ranging from 1.20–1.24, except for donor-fresh), blastogenesis (AORs 1.22–1.74), cardiovascular (AORs 1.04–1.26), gastrointestinal (AORs 1.28–2.01), musculoskeletal (AORs 1.10–1.48) and among male children, genitourinary (AORs 1.15–1.40, except for donor-fresh). Among children with a blastogenesis defect, 3.7% also had a chromosomal defect. Orofacial defects were increased in the OI/IUI and autologous-fresh and autologous-thawed groups (AORs 1.26–1.42). As expected, chromosomal defects were lower in the donor-fresh and donor-thawed groups (AORs 0.06–0.08), but also in the autologous-thawed group (AOR 0.61). Approximately 81% of children with a chromosomal defect had Down syndrome. Twins had greater risks compared to singletons in every birth defect category except orofacial and chromosomal, with AORs ranging from 1.26 to 1.76. Males had greater risks than females for nonchromosomal, blastogenesis, musculoskeletal and orofacial defects, with AORs ranging from 1.16 to 1.50 and lower risks for cardiovascular defects.

Risk of childhood cancer by mode of conception

The rates of all cancers, leukemia, CNS tumors, embryonal tumors and solid tumors, and the results of the Cox proportional hazards regression models are shown in Table V. The naturally conceived group had 1469 of the 1789 cancers, the ART autologous groups 165 and 50 cancers, the non-ART siblings 59 cancers and the three smallest groups had 19, 16 and 9 cancers; the number of cancers decrease proportionately to the rate when specific cancer types are modeled (Table V). The risk of any cancer was increased among ART autologous-fresh and non-ART siblings (HR 1.31 and 1.34, respectively), with a similar but not significant increase in children in the OI/IUI conceived children, and autologous-thawed groups (HR 1.42 and 1.28). The risk of leukemia was increased in the OI/IUI and ART autologous-fresh groups (HRs 2.15 and 1.63). The risk of CNS tumors was increased among children in the ART autologous-fresh and donor-fresh and non-ART sibling groups (HRs ranging from 1.68 to 2.57). The risk of solid tumors was increased in the ART autologous-fresh group (HR 1.39). As a sensitivity analysis, the analysis was repeated excluding all children with a major birth defect (see Supplementary Table SIII); the HRs were of a similar magnitude, although some did not achieve significance due to the >40% reduction in sample size.

Risk of cancer as a function of birth defect status

A total of 127 children had both birth defects and cancer, of which 53 (42%) had leukemia. The Cox proportional hazards regression model (Table VI) showed that the risk of cancer had two components: 1) method of conception (see results above), and 2) presence, type and number of birth defects. The interaction between conception group and birth defect status was not significant. The presence of nonchromosomal defects increased the cancer risk, greater for two or more defects versus one defect, for all cancers and each type evaluated. The presence of chromosomal defects was strongly associated with cancer risk (HR 8.70 for all cancers and HR 21.90 for leukemia), further elevated in the presence of both chromosomal and nonchromosomal defects (HR 21.29 for all cancers, HR 64.83 for leukemia and HR 4.71 for embryonal tumors). The two components of the risk of cancer are independent and therefore, on the average, their coefficients are multiplicative.

Discussion

This study presents contemporary, population-based findings in three important areas of child health: birth defects, childhood cancer, and their co-occurrence among children who were conceived naturally, with OI/IUI, and with ART by oocyte source and embryo state, building upon findings from our prior analyses in these areas (Spector et al., 2019; Luke et al., 2020, 2021). Among the 83 946 children born from ART in the USA in 2019, compared to their naturally conceived counterparts, these risks translate into an estimated excess of 761 children with major birth defects, 31 children with cancer and 11 children with both major birth defects and cancer.

These results add to the literature on the co-occurrence of birth defects and childhood cancer (Johnson et al., 2017; Norwood et al., 2017; Von Behren et al., 2017; Lupo et al., 2019; Kampitsi et al., 2022), and extend our preliminary findings among ART-conceived children (Luke et al., 2020). Prior research has shown that even when analyses excluded well-established, known birth defect-cancer associations (such as Down syndrome and leukemia), the risk of cancer was still elevated in the presence of birth defects, particularly among children younger than 5 years of age (Dawson et al., 2015). In their population-based study from Australia, Dawson et al. (2015) reported three significant associations after these known birth defect-cancer exclusions: cardiovascular defects with cancer (HR 1.76, 95% CI 1.04, 2.99), birth defects with hepatic tumors (HR 7.54, 95% CI 2.36.4.03) and leukemias other than acute lymphocytic leukemia and acute myeloid leukemia (HR 4.30, 95% CI 1.23, 15.09). Although not feasible to ascertain in the current study because of small numbers of children with both birth defects and cancers, prior studies have reported elevated risks of anomalies and malignancies within the same organ system, including neurological defects and CNS tumors (Wong-Siegel et al., 2017; Lupo et al., 2019); congenital anomalies of the kidney and urinary tract and urinary tract cancer (Calderon-Margalit et al., 2021); eye defects and retinoblastoma, and gastrointestinal defects and hepatoblastoma (Lupo et al., 2019).

ART and birth defects and cancer

Fertility is strongly related to age, for both male and female partners, which can be seen in the differences in the conception groups in this study (Hassan and Killick, 2003; American Society for Reproductive Medicine, 2014). The increased birth defect risk in the OI/IUI, non-ART siblings, and ART autologous-fresh and thawed groups suggests an association with underlying parental subfertility, while the increased risk in the donor-thawed versus the donor-fresh group may be associated with the process of cryopreservation. The pattern of elevated cancer risk for both non-ART siblings and children in the ART autologous-fresh group suggests common genetic and/or environmental factors. Blastogenesis defects and embryonal tumors have particular relevance to ART. Our analysis indicated that, compared to the naturally conceived group, the risks of blastogenesis defects and cardiovascular defects were increased in all of the other six conception groups. Embryonal tumors have been hypothesized to be associated with developmental disruptions, thereby sharing pathophysiologic features with birth defects (Narod et al., 1997; Botto et al., 2013). Our analyses indicate that the risk of embryonal tumors increases in the presence of nonchromosomal defects, also reflecting developmental disruption in the periconceptual period, when the epigenome is most vulnerable (Hattori et al., 2019; Mani et al., 2020; Yeung et al., 2021).

Sibling risks

The choice of an appropriate comparison group in infertility research poses a special challenge. Although most studies compare women treated for infertility to fertile women, this approach has several potential disadvantages, including differences in age, socioeconomic status, education and reproductive history. Comparisons within families, as repeat pregnancies to the same woman, have the advantage of eliminating the fixed effects of the parents (mainly the genetic contribution), with adjustments possible for her change in age, parity and, if appropriate, method of conception. In our prior studies of siblings in Massachusetts, declining fertility status, with or without ART treatment, was associated with increasing risks for adverse outcomes, greatest for women whose fertility status declined the most between the two pregnancies (Luke et al., 2016b). In an additional sibling analysis to their Australian study (Davies et al., 2012b), Davies et al. (2012a) reported an increased risk of birth defects among ART-conceived siblings compared to naturally conceived siblings (crude odds ratio, 1.50, 95% CI, 1.08, 2.09), CIs which overlap our findings for naturally conceived ART siblings (AOR 1.20, 95% CI 1.11, 1.31).

Strengths

This study has a number of strengths, including a large sample size, population-based design, and contemporary time period. The four study States include racially and ethnically diverse populations, with high linkage rates to the SART CORS, vital records, birth defects and cancer registries, utilizing similar case definitions and data collected. Our live birth prevalence rates of birth defects are in accord with both US and European rates (Mai et al., 2019; EUROCAT prevalence rates, 2022), as well as our prior research in Massachusetts (Luke et al., 2021), and research by other groups in Oklahoma (Janitz et al., 2016) and Arkansas (Patel et al., 2020). Our findings of higher birth defects rates among twins compared to singletons are also in accord with prior studies (Hansen et al., 2013; Dawson et al., 2016). The data on infertility, birth outcomes, cancer and birth defects were independently collected, minimizing the risk of ascertainment bias.

Limitations

This study must be considered in light of certain limitations. In the SART CORS database, it was not possible to differentiate method of embryo freezing (slow-freezing versus vitrification); data on ICSI was only available in the fresh embryo ART group; and data were unavailable on duration of infertility, which has been reported to be related to birth defect risk (Ghazi et al., 1991; Zhu et al., 2006). Data on preimplantation testing were not available, other than the infertility diagnosis of preimplantation genetic diagnosis. For the OI/IUI group, it was not possible to differentiate type of non-ART treatment utilized. Data on birth defects were not available on miscarriages, terminations, or stillbirths, only on live births; this limitation is also noted in other population-based studies (Källén et al., 2005, 2010), which for legal reasons could not be included in the linkages or analyses. This limited our ability to study conditions that are more likely to be terminated after prenatal detection. In addition, data were unavailable on imprinting disorders. Because of the lack of a national registry for non-ART infertility procedures, the OI/IUI group is likely underrepresented, with some treated women included in the naturally conceived group, as well as a higher percentage than identified in the non-ART sibling group (Luke et al., 2016a). This underrepresentation would tend to bias toward finding less of a difference between conception groups. We were also not able to determine if and when children moved out of the four study States, with the result that person-time in our study may have been slightly overestimated, which would lead to an under-estimation of absolute cancer rates. This under-estimation should not have biased the HRs, though, as long as the denominator error did not differ by group. We also did not identify pregnancies to the same woman when there was a change in paternity.

Inherent in any observational study is the common potential issue of unmeasured confounding. As shown in Tables I and II, women who conceived with OI/IUI or ART were more likely to be White, non-Hispanic, more educated, and older than women who conceived naturally. Differences in these factors may reflect unmeasured confounding in areas such as income, housing and neighborhood environments, diet and nutrition and prenatal care, which may affect fetal development and subsequent risk of both birth defects and childhood cancer. Although we included a range of factors in the models, it is not possible to estimate the effect of unmeasured confounding on the HRs. In addition, for some factors obtained from birth certificate data, such as diabetes and hypertension, a missing response could not be distinguished from the absence of the condition.

The primary goal of this grant (R01 HD084377) was to compare the rate of birth defects in the ART births to that of the naturally conceived births. Since the rate of cancers is <1/10 that of birth defects, only three groups (the two ART autologous groups and the non-ART siblings) had sufficient cancers to be compared to the naturally conceived births; however, even for these three groups, the power is less than the power estimated for birth defects.

Another limitation is the lack of pathology data on the umbilical cord and placenta, which may be critical factors in the development of birth defects (Ebbing et al., 2013, 2020; Choux et al., 2015; Sacha et al., 2020, 2022; Stern et al., 2022). Among women with the same infertility diagnosis, those whose pregnancies were conceived with versus without ART treatment have been shown to be at greater risk of placental complications (Stern et al., 2022). There is a greater incidence of anatomic, inflammatory, and vascular placental pathology in ART-conceived pregnancies from fresh embryos versus natural conceptions, and among ART-conceived pregnancies, among those from thawed versus fresh embryos (Sacha et al., 2020, 2022). Population-based analyses from Norway have shown increased risks of velamentous or marginal umbilical cord insertions associated with ART conception, and a 31% greater risk of single umbilical artery, which was strongly associated with birth defects in the gastrointestinal tract (atresia or stenosis), renal agenesis and cardiovascular defects (Ebbing et al., 2013, 2020).

Conclusions

Compared to naturally conceived children, we found a 20–24% increased risk of nonchromosomal birth defects among children conceived with OI/IUI, naturally conceived ART siblings, and ART children born from autologous-fresh, autologous-thawed and donor-thawed cycles, and a 50–52% increased risk among males versus females, and twins versus singletons. The risk of cancer was increased by 34% and 31%, respectively, among non-ART siblings and ART children born from autologous-fresh cycles, and 18% among males versus females. The risk of cancer had two components: (i) method of conception and (ii) presence, type and number of birth defects. Among both naturally conceived and ART-conceived children, the presence of birth defects was associated with a greater risk of cancer. This information regarding birth defects and future cancer risk should be included when counseling patients about the risks and benefits of ART, including the growing segment of the fertile population who elect to use ART as a family planning option (Rinaudo and Adeleye, 2018), as well as women treated with OI/IUI (Barnhart, 2013; Wilkinson et al., 2020). These findings have important implications for all children with birth defects. Recent research from the Nordic countries indicates that the increased risk of cancer among individuals with birth defects persisted into adulthood, for both nonchromosomal and chromosomal defects (Daltveit et al., 2020). These findings indicate that children conceived with ART, non-ART siblings, and all children with birth defects should be monitored more closely for the subsequent development of cancer.

Data Availability

The data used in this analysis were obtained from private (SART CORS) and public (vital records, birth defects registries and cancer registries) sources, under data use agreements and confidentiality pledges assuring that the data would not be shared or distributed, and therefore are not available to other investigators.