Abstract
Background:
Birth defects are associated with childhood cancer, but little is known regarding pediatric carcinomas, a group of especially rare tumors.
Methods:
We used Cox proportional hazards regression to estimate the hazard ratio (HR) and 95% confidence interval (CI) for any carcinoma, as well as thyroid, hepatocellular, and renal carcinoma specifically, up to 18 years of age among children with major, non-syndromic anomalies or chromosomal/genetic syndromes, relative to unaffected children.
Results:
Our registry-linkage study included nine states and 21,933,476 children between 1990–2018: 641,827 with non-syndromic anomalies, and 49,619 with syndromes. Carcinomas were diagnosed in 833 children, including 35 with non-syndromic anomalies and eight with syndromes. The hazard of carcinoma was increased both among children with non-syndromic anomalies (HR: 1.7, CI: 1.2–2.4; N=35) and syndromes (HR: 4.7, CI: 2.3–9.5; N=7). Hepatocellular carcinoma was associated with non-syndromic anomalies (HR: 4.6, CI: 2.2–9.7; N=8) and syndromes (HR: 8.0, CI: 1.1–58.1; N<5). The hazard of renal carcinoma was markedly increased in children with tuberous sclerosis (HR 59.6, CI: 23.7–149.5; N=5), a known cause of renal cancer. Thyroid carcinoma was not associated with non-syndromic anomalies or syndromes.
Conclusion:
Birth defects are associated with hepatocellular and renal carcinoma in children.
Keywords: birth defects, congenital anomalies, carcinoma, childhood cancer, epidemiology
INTRODUCTION
Pediatric carcinomas are a heterogeneous group of rare and understudied malignancies. These tumors have been reported in the context of chromosomal or genetic syndromes, including Li-Fraumeni, Von Hippel Lindau, DICER1, Beckwith-Wiedemann, tuberous sclerosis, constitutional mismatch repair deficiency, and 22q11.2 deletion.[1] Grobner et al. reported that 50% of adrenocortical carcinoma patients harbored germline pathogenic variants, predominantly in TP53, which was the highest rate observed for any evaluated pediatric malignancy.[2] A single patient with a pathogenic TP53 variant was identified among seven cases of pediatric carcinoma in a registry-based study from Sweden.[3] However, little is known regarding the etiology of sporadic carcinomas.
Birth defects, often defined as structural, functional, or metabolic changes present at delivery, are diagnosed in 3–5% of U.S. births.[4] While there is extensive epidemiologic evidence linking birth defects to cancer in children,[5–7] little is known regarding carcinomas specifically. Because of their extremely low incidence, few population-based studies have reported on these tumors; these studies suggest increased risk of hepatic and thyroid carcinomas among children with birth defects.[8–10] Here, we sought to characterize risk of specific carcinomas in children with various types of birth defects using data from a multi-state population-based registry linkage study including >22 million livebirths.
METHODS
We used data from the Genetic Overlap Between Anomalies and Cancer in Kids (GOBACK) Registry Linkage Study to evaluate risk of carcinomas in children with congenital anomalies.[5] Table 1 lists the included birth years, which varied by state but ranged from 1990–2018. In each state, we linked data from vital records to population-based birth defects registries to identify children with major, non-syndromic congenital anomalies (‘anomalies’) or known chromosomal/genetic syndromes (‘syndromes’) and a comparison cohort of children without these conditions. We list the evaluated anomalies and syndromes in Supplemental Table S1. We then linked these data to population-based cancer registries, to identify cases diagnosed with primary malignant carcinomas at ≤18 years of age (Supplemental Table S2). Children who did not link to a record in the state cancer registry were treated as not having cancer and right-censored at 18 years of age or the end of the study period. We excluded children with missing information on sex assigned at birth, children with only minor anomalies, and children whose first primary malignant tumors were not carcinomas (e.g., leukemias, brain tumors) from our analyses.
Table 1.
Child and Maternal Characteristics by Congenital Anomaly Status, Genetic Overlap Between Anomalies and Cancer in Kids (GOBACK) Registry Linkage Cohort.
| Without birth defects | With major non-syndromic anomalies | With chromosomal or genetic syndromes | |
|---|---|---|---|
| N (%) | N (%) | N (%) | |
| TOTAL | 21,242,030 (100.0) | 641,827 (100.0) | 46,619 (100.0) |
| Sex assigned at birth | |||
| Male | 10,782,479 (50.8) | 374,482 (58.3) | 25,138 (50.7) |
| Female | 10,459,551 (49.2) | 267,345 (41.7) | 24,481 (49.3) |
| Plurality | |||
| Singleton | 20,532,333 (96.7) | 597,804 (93.1) | 47,721 (96.2) |
| Multiple | 686,631 (3.2) | 43,073 (6.7) | 1,796 (3.6) |
| Unknown | 23,066 (0.1) | 950 (0.1) | 102 (0.2) |
| Cancer diagnosis | |||
| None | 21,241,240 (100.0) | 641,792 (100.0) | 49,611 (100.0) |
| Carcinoma | 790 (0.0) | 35 (0.0) | 8 (0.0) |
| Thyroid carcinoma | 542 (68.6) | 18 (51.4) | 1 (12.5) |
| Hepatocellular carcinoma | 59 (7.5) | 8 (22.9) | 1 (12.5) |
| Renal carcinoma | 49 (6.2) | 2 (5.7) | 5 (62.5) |
| Age at carcinoma diagnosis * | 14.0 (10.8, 16.0) | 11.9 (6.14, 13.6) | 8.3 (7.0, 13.0) |
| State of birth (delivery years) | |||
| Arkansas (1995–2011) | 605,754 (2.9) | 15,793 (2.5) | 943 (1.9) |
| Florida (1998–2012) | 2,754,520 (13.0) | 124,106 (19.3) | 5,240 (10.6) |
| Massachusetts (2000–2017) | 1,319,943 (6.2) | 16,323 (2.5) | 3,114 (6.3) |
| Michigan (1992–2018) | 3,120,171 (14.7) | 114,748 (17.9) | 9,655 (19.5) |
| New Jersey (1990–2018) | 3,197,048 (15.1) | 83,642 (13.0) | 5,992 (12.1) |
| North Carolina (2003–2018) | 1,896,281 (8.9) | 45,710 (7.1) | 4,851 (9.8) |
| Oklahoma (1997–2012) | 773,406 (3.6) | 21,397 (3.3) | 1,971 (4.0) |
| South Carolina (2008–2018) | 593,488 (2.8) | 7,338 (1.1) | 779 (1.6) |
| Texas (1999–2017) | 6,981,419 (32.9) | 212,770 (33.2) | 17,074 (34.4) |
| Year of birth | |||
| 1990–1999 | 3,170,276 (14.9) | 78,052 (12.2) | 6089 (12.3) |
| 2000–2004 | 4,874,390 (22.9) | 136,761 (21.3) | 10232 (20.6) |
| 2005–2009 | 5,575,299 (26.2) | 176,242 (27.5) | 12507 (25.2) |
| 2010–2014 | 4,830,601 (22.7) | 152,040 (23.7) | 12381 (25.0) |
| 2015–2018 | 2,791,464 (13.1) | 82,939 (12.9) | 7467 (15.0) |
| Unknown | 0 (0.0) | 15,793 (2.5) | 943 (1.9) |
| Maternal age (years) | |||
| <20 | 2,128,820 (10.0) | 64,337 (10.0) | 3,399 (6.9) |
| 20–24 | 5,072,424 (23.9) | 150,492 (23.4) | 8,192 (16.5) |
| 25–29 | 5,786,257 (27.2) | 170,582 (26.6) | 9,499 (19.1) |
| 30–34 | 5,101,719 (24.0) | 152,519 (23.8) | 10,615 (21.4) |
| 35–39 | 2,459,866 (11.6) | 80,120 (12.5) | 10,724 (21.6) |
| ≥40 | 564,478 (2.7) | 20,506 (3.2) | 6,982 (14.1) |
| Unknown | 128,466 (0.6) | 3,271 (0.5) | 208 (0.4) |
| Maternal race and ethnicity | |||
| Hispanic or Latina | 5,033,369 (23.7) | 149,720 (23.3) | 13,325 (26.9) |
| American Indian/Alaska Native | 157,997 (0.7) | 4812 (0.7) | 374 (0.8) |
| Asian or Pacific Islander | 861,268 (4.1) | 19,944 (3.1) | 1,562 (3.1) |
| Black | 3,337,108 (15.7) | 107,379 (16.7) | 7,214 (14.5) |
| White | 11,403,534 (53.7) | 348,425 (54.3) | 26,276 (53.0) |
| Additional groups | 269,227 (1.3) | 8,318 (1.3) | 558 (1.1) |
| Unknown | 17,9527 (0.8) | 3,229 (0.5) | 310 (0.6) |
Presented as median (interquartile range) among 833 carcinoma cases.
We estimated the incidence rate per million person-years, along with 95% confidence intervals, among children without birth defects, children with anomalies, and children with syndromes.[11] We compared age at carcinoma diagnosis between groups using medians, interquartile ranges (IQRs), and the Mann-Whitney U test. We used Cox proportional hazards regression to estimate the hazard ratio (HR) and 95% confidence interval (CI) of any carcinoma and thyroid, hepatocellular, and renal carcinoma specifically (the three most prevalent forms in our cohort) among children with anomalies or syndromes relative to children without birth defects. We assessed the proportional hazards assumption by visualizing Schoenfeld residuals for each covariate in the Cox proportional hazards model. We also evaluated the hazard of any carcinoma separately among children with an isolated anomaly and multiple anomalies. When there were five or more exposed cases, we adjusted for sex assigned at birth, state of birth, and maternal age. For models with fewer than five exposed cases, we estimated crude HRs. We performed all statistical analyses in R version 4.2.1 (R Core Team, Vienna, Austria). We used the package ‘epiR’ (v2.0.75) to calculate incidence rates and 95% CIs, and the packages ‘survival’ (v3.3.1) and ‘survminer’ (v0.4.9) for Cox regression.
This study complied with the principles of the Declaration of Helsinki and was approved by the Institutional Review Boards of participating institutions. The requirement for written informed consent was waived as it involved secondary analysis of de-identified data.
RESULTS
Table 1 presents perinatal and maternal sociodemographic information for the cohort. In this cohort, 2.9% of children were diagnosed with major anomalies and 0.2% were diagnosed with syndromes. Relative to the comparison cohort, children with anomalies were more often male and part of a multiple birth, and people who delivered children with syndromes were more often ≥35 years of age at delivery. Median (IQR) follow-up time was 9.8 (5.0–14.8) years in the group without birth defects, 9.2 (4.5–14.3) years in the group with anomalies, and 8.7 (3.7–14.3) years in the group with syndromes. During the study period, 833 children were diagnosed with primary malignant carcinomas, including 790 without birth defects and 35 with anomalies as well as eight with syndromes, corresponding to incidence rates (95% CIs) of 3.8 (3.5–4.0), 5.8 (4.0–8.1), and 17.8 (7.7–35.1) per million person-years in each of the three groups, respectively. Among cases with carcinomas, children without birth defects had an older median age at diagnosis (14.0 years, IQR: 10.8–16.0) than children with anomalies (11.9 years, IQR: 6.1–13.6; p<0.001) or syndromes (8.3 years, IQR: 7.0–13.0 years; p=0.06).
The proportional hazard assumption was met, and the hazard of any carcinoma was increased among children with anomalies or syndromes relative to children without anomalies (Table 2). Individuals with anomalies had increased risk of hepatocellular carcinoma (HR: 4.8, CI: 2.3–10.1; N=8), and this association was stronger among children with multiple anomalies (HR: 8.2, CI: 3.0–22.5) than among children with an isolated anomaly (HR: 3.5, CI: 1.3–9.5) (both N<5); however in the latter analysis, there were fewer than five exposed cases in each group. Children who had tuberous sclerosis were at increased risk of renal carcinoma (HR: 59.6, 95% CI 23.7–149.5; N=5); the median age at renal carcinoma diagnosis was 7.7 years (IQR: 7.0–9.0 years). Finally, we found no association between birth defects and thyroid carcinoma.
Table 2.
Hazard Ratio and 95% Confidence Interval of Carcinomas Through 18 Years of Age among Individuals with Birth Defects, Genetic Overlap Between Anomalies and Cancer in Kids (GOBACK) Registry Linkage Cohort.
| Carcinoma diagnosis | ||||||||
|---|---|---|---|---|---|---|---|---|
| Any | Thyroid | Hepatocellular | Renal | |||||
| N | HR (95% CI) | N | HR (95% CI) | N | HR (95% CI) | N | HR (95% CI) | |
| Major non-syndromic anomalies * | ||||||||
| Any | 35 | 1.7 (1.2–2.4) | 18 | 1.3 (0.8–2.1) | 8 | 4.8 (2.3–10.1) | <5 | 1.5 (0.4–6.0) |
| Isolated | 24 | 1.7 (1.1–2.5) | 12 | 1.3 (0.7–2.3) | <5 | 3.5 (1.3–9.5) | <5 | 2.1 (0.5–8.5) |
| Multiple | 11 | 1.8 (1.0–3.3) | 6 | 1.5 (0.7–3.3) | <5 | 8.2 (3.0–22.5) | 0 | - |
| Chromosomal/genetic syndromes | 8 | 4.7 (2.3–9.5) | <5 | 0.9 (0.1–6.3) | <5 | 8.0 (1.1–58.1) | 5† | 59.6 (23.7–149.5) |
Abbreviations: CI=confidence interval; HR=hazard ratio.
Excludes children with chromosomal or genetic syndromes.
All cases were diagnosed with tuberous sclerosis.
N indicates the number of exposed cases. Adjusted hazard ratios from models including terms for state of birth, sex assigned at birth, and maternal age are presented for phenotypes with ≥5 exposed cases. Crude hazard ratios are presented for phenotypes with <5 exposed cases. A comparison group of children without congenital anomalies was used for all analyses.
DISCUSSION
We observed an increased hazard of hepatocellular carcinoma in children with anomalies, and of hepatocellular and renal carcinoma in children with syndromes. Thyroid carcinoma, the most common carcinoma among children in our cohort, was not associated with either anomalies or syndromes. Given the low incidence rates for these tumors in the pediatric population, it is worth noting that absolute risk remains low even among children with birth defects. Nonetheless, as little is known regarding their etiologies, population-based reports such as this may be useful in identifying potential etiologic factors.
Our findings are consistent with a recent study using data from Nordic health registries (N=240,878) that reported threefold increased odds (95% CI: 1.2–7.3, N=5) of hepatic carcinoma in children with major non-syndromic anomalies.[10] The authors did not specifically evaluate hepatocellular carcinoma among people with syndromes but did find a non-statistically significantly increased risk of hepatic tumors in this group (OR: 2.9, 95% CI: 0.4–21, N<5). They also reported an increased risk of thyroid tumors (OR: 3.4, 95% CI: 2.3–5.1, N=26) among children with major non-syndromic anomalies, whereas we observed an increased hazard for thyroid carcinoma that did not reach statistical significance. Narod et al. reported that two of 22 children with renal carcinoma in the British National Registry of Childhood Tumors were diagnosed with spinal anomalies, more than would have been expected based on the frequency of these anomalies in the study sample or the birth prevalence in a Canadian registry that served as an external comparator.[7] Renal carcinoma has been described in adults (and less frequently, children) with tuberous sclerosis, with incidence as high as 4%.[13] We report a sixty-fold increased hazard of this tumor in children with tuberous sclerosis relative to children without anomalies.
Several potential mechanisms linking birth defects to cancer in early life have been enumerated.[5, 6] These include: 1) shared genetic causes, supported by genetic overlap in the control of fetal development and cancer initiation/progression; 2) maternal (e.g., diabetes mellitus) and environmental (e.g., atrazine) exposures associated with both phenotypes; and 3) iatrogenic exposures among neonates with anomalies, such as could occur during imaging studies or in the neonatal intensive care unit.[12]
A major strength of our study is its size; our analyses included a far greater number of cases with pediatric carcinomas than previous epidemiologic studies. Whereas previous studies often analyzed these in combination with other, more common pediatric tumors (e.g., hepatocellular carcinoma with hepatoblastoma, renal carcinoma with nephroblastoma), we evaluated thyroid, hepatocellular, and renal carcinoma specifically, which should result in more accurate estimates of the associations between anomalies/syndromes and carcinomas. In addition, our study should be unaffected by selection or recall bias as we used prospectively collected data from population-based birth defects and cancer registries. These registries have essentially complete case ascertainment and are monitored to ensure timely and complete data. The chief potential limitations are that we lacked data on non-cancer related mortality and were unable to identify and censor children who emigrated from their state of birth, so we likely overestimated follow-up times among individuals without a successful linkage to cancer registries. This may have led to overestimation of survival times, biasing our risk estimates away from the null. We believe that any practical impact on our analysis would have been minor as mortality rates are low even among children with birth defects [14] and evidence suggests that residential mobility is similar for families of children with birth defects and families of children without.[15] Finally, due to sample size, we were unable to perform analyses of specific anomalies.
CONCLUSIONS
In a prospective, multi-state registry linkage study, we found that people with birth defects were at increased risk of hepatocellular and renal carcinomas during childhood. Both anomalies and syndromes were associated with hepatocellular carcinoma, and children with syndromes (specifically, tuberous sclerosis) were at increased risk of renal carcinoma. Both anomalies and syndromes were not associated with thyroid carcinoma, the most common pediatric carcinoma in our cohort. Our study adds to the sparse epidemiologic literature concerning these rare pediatric tumors.
Supplementary Material
Funding statement:
This research was supported by the Rally Foundation for Childhood Cancer Research (Career Development Award 23CND05), the National Cancer Institute (1R01CA284531), and the Department of Defense (Impact Award CA190214). This research was also supported in part by a grant from the Cancer Prevention and Research Institute of Texas (CPRIT) for the Systems Epidemiology of Cancer Training (SECT) Program (RP210037; PI: Thrift A).
Footnotes
Conflict of interest statement: The authors have no conflicts of interest to declare.
Data availability statement:
Aggregated data are available from the corresponding author upon reasonable request. Restrictions apply to individual-level data, which were used with permission of the participating states. Individual-level data are available by application to: Arkansas Reproductive Health Monitoring System, Florida Department of Health, Massachusetts Department of Health, Michigan Department of Health and Human Services, New Jersey Department of Health, North Carolina Department of Public Health, Oklahoma State Department of Health, South Carolina Department of Health and Environmental Control, and Texas Department of State Health Services.
REFERENCES
- [1].Perrino M, Cooke-Barber J, Dasgupta R, Geller JI, Genetic predisposition to cancer: Surveillance and intervention, Semin. Pediatr. Surg 28(6) (2019) 150858. [DOI] [PubMed] [Google Scholar]
- [2].Gröbner SN, Worst BC, Weischenfeldt J, Buchhalter I, Kleinheinz K, Rudneva VA, Johann PD, Balasubramanian GP, Segura-Wang M, Brabetz S, Bender S, Hutter B, Sturm D, Pfaff E, Hübschmann D, Zipprich G, Heinold M, Eils J, Lawerenz C, Erkek S, Lambo S, Waszak S, Blattmann C, Borkhardt A, Kuhlen M, Eggert A, Fulda S, Gessler M, Wegert J, Kappler R, Baumhoer D, Burdach S, Kirschner-Schwabe R, Kontny U, Kulozik AE, Lohmann D, Hettmer S, Eckert C, Bielack S, Nathrath M, Niemeyer C, Richter GH, Schulte J, Siebert R, Westermann F, Molenaar JJ, Vassal G, Witt H, Burkhardt B, Kratz CP, Witt O, van Tilburg CM, Kramm CM, Fleischhack G, Dirksen U, Rutkowski S, Frühwald M, von Hoff K, Wolf S, Klingebiel T, Koscielniak E, Landgraf P, Koster J, Resnick AC, Zhang J, Liu Y, Zhou X, Waanders AJ, Zwijnenburg DA, Raman P, Brors B, Weber UD, Northcott PA, Pajtler KW, Kool M, Piro RM, Korbel JO, Schlesner M, Eils R, Jones DTW, Lichter P, Chavez L, Zapatka M, Pfister SM, The landscape of genomic alterations across childhood cancers, Nature 555(7696) (2018) 321–327. [DOI] [PubMed] [Google Scholar]
- [3].von Stedingk K, Stjernfelt K-J, Kvist A, Wahlström C, Kristoffersson U, Stenmark-Askmalm M, Wiebe T, Hjorth L, Koster J, Olsson H, Øra I, Prevalence of germline pathogenic variants in 22 cancer susceptibility genes in Swedish pediatric cancer patients, Sci. Rep 11(1) (2021) 5307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Birth Defects are Common, Costly, and Critical, 2022. https://www.cdc.gov/ncbddd/birthdefects/infographic.html#:~:text=Birth%20defects%20affect%201%20in,translates%20into%20about%20120%2C000%20babies.&text=Each%20year%2C%20total%20hospital%20costs,care%20or%20many%20provider%20charges. (Accessed 10/03/2024.
- [5].Lupo PJ, Schraw JM, Desrosiers TA, Nembhard WN, Langlois PH, Canfield MA, Copeland G, Meyer RE, Brown AL, Chambers TM, Sok P, Danysh HE, Carozza SE, Sisoudiya SD, Hilsenbeck SG, Janitz AE, Oster ME, Scheuerle AE, Schiffman JD, Luo C, Mian A, Mueller BA, Huff CD, Rasmussen SA, Scheurer ME, Plon SE, Association Between Birth Defects and Cancer Risk Among Children and Adolescents in a Population-Based Assessment of 10 Million Live Births, JAMA oncology 5(8) (2019) 1150–1158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Mandalenakis Z, Karazisi C, Skoglund K, Rosengren A, Lappas G, Eriksson P, Dellborg M, Risk of Cancer Among Children and Young Adults With Congenital Heart Disease Compared With Healthy Controls, JAMA Netw Open 2(7) (2019) e196762. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Narod SA, Hawkins MM, Robertson CM, Stiller CA, Congenital anomalies and childhood cancer in Great Britain, Am. J. Hum. Genet 60(3) (1997) 474–85. [PMC free article] [PubMed] [Google Scholar]
- [8].Altmann AE, Halliday JL, Giles GG, Associations between congenital malformations and childhood cancer. A register-based case-control study, Br. J. Cancer 78(9) (1998) 1244–1249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Agha MM, Williams JI, Marrett L, To T, Zipursky A, Dodds L, Congenital abnormalities and childhood cancer, Cancer 103(9) (2005) 1939–48. [DOI] [PubMed] [Google Scholar]
- [10].Daltveit DS, Klungsøyr K, Engeland A, Ekbom A, Gissler M, Glimelius I, Grotmol T, Madanat-Harjuoja L, Ording AG, Sørensen HT, Troisi R, Bjørge T, Sex differences in childhood cancer risk among children with major birth defects: a Nordic population-based nested case-control study, Int. J. Epidemiol (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].ULM K, SIMPLE METHOD TO CALCULATE THE CONFIDENCE INTERVAL OF A STANDARDIZED MORTALITY RATIO (SMR), Am. J. Epidemiol 131(2) (1990) 373–375. [DOI] [PubMed] [Google Scholar]
- [12].Lai TT, Bearer CF, Iatrogenic environmental hazards in the neonatal intensive care unit, Clin. Perinatol 35(1) (2008) 163–81, ix. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Yang P, Cornejo KM, Sadow PM, Cheng L, Wang M, Xiao Y, Jiang Z, Oliva E, Jozwiak S, Nussbaum RL, Feldman AS, Paul E, Thiele EA, Yu JJ, Henske EP, Kwiatkowski DJ, Young RH, Wu CL, Renal cell carcinoma in tuberous sclerosis complex, Am. J. Surg. Pathol 38(7) (2014) 895–909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Sattolo ML, Arbour L, Bilodeau-Bertrand M, Lee GE, Nelson C, & Auger N, Association of birth defects with child mortality before age 14 years, JAMA Netw Open 5(4) (2022) e226739–e226739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Miller A, Siffel C, & Correa A. Residential mobility during pregnancy: patterns and correlates, Matern Child Health J. 14 (2010) 625–634. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Aggregated data are available from the corresponding author upon reasonable request. Restrictions apply to individual-level data, which were used with permission of the participating states. Individual-level data are available by application to: Arkansas Reproductive Health Monitoring System, Florida Department of Health, Massachusetts Department of Health, Michigan Department of Health and Human Services, New Jersey Department of Health, North Carolina Department of Public Health, Oklahoma State Department of Health, South Carolina Department of Health and Environmental Control, and Texas Department of State Health Services.