Skip Navigation


American Journal of Epidemiology Advance Access originally published online on July 10, 2007
American Journal of Epidemiology 2007 166(7):852-856; doi:10.1093/aje/kwm143
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
166/7/852    most recent
kwm143v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Spector, L. G.
Right arrow Articles by Bunin, G. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Spector, L. G.
Right arrow Articles by Bunin, G. R.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology © The Author 2007. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.

PRACTICE OF EPIDEMIOLOGY

Feasibility of Nationwide Birth Registry Control Selection in the United States

Logan G. Spector1,2, Julie A. Ross1,2, Susan E. Puumala1, Michelle Roesler1, Andrew F. Olshan3 and Greta R. Bunin4

1 Division of Epidemiology/Clinical Research, Department of Pediatrics, University of Minnesota, Minneapolis, MN
2 Cancer Center, University of Minnesota, Minneapolis, MN
3 Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, NC
4 Children's Hospital of Philadelphia, Philadelphia, PA

Correspondence to Dr. Logan G. Spector, Division of Epidemiology/Clinical Research, Department of Pediatrics, University of Minnesota, 420 Delaware Street, SE, MMC 715, Minneapolis, MN 55455 (e-mail: spector{at}epi.umn.edu).

Received for publication March 7, 2007. Accepted for publication April 9, 2007.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
As random digit dialing becomes increasingly unfeasible for many types of studies, alternative methods for control selection are needed, especially for studies of childhood cancer. US birth registries are an appealing source of young control children because they are population based, provide demographic and pregnancy data for comparison of participants with the study base, and maintain data that enable matching on birth characteristics. Here the authors describe the ability of US birth registries to release information sufficient to locate potential control subjects for two ongoing case-control studies of hepatoblastoma and infant leukemia. The birth registries of 32 states, in which 75.8% of US children 0–5 years of age resided in 2004, agreed to participate in control selection. Data sufficient to track participants and to characterize nonrespondents were available from a majority of registries. These results suggest that birth registries may be used to select controls for studies of rare childhood diseases on a national scale.

case-control studies; methods; neoplasms; pediatrics; sampling studies

Abbreviations: RDD, random digit dialing


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Cancer is the leading cause of death due to disease in US children 1–19 years of age (1) but is rare in absolute terms (2), necessitating that most investigations use the case-control design. Although studies of the most common types of childhood cancer have achieved adequate sample sizes in limited areas (3–5), cases of the less frequent malignancies must be recruited nationwide. Random digit dialing (RDD) enabled identification of widely dispersed study participants and so was preferred for control selection in epidemiology studies of childhood cancer in the United States. However, we recently documented a 2.5 percent annual decline from 1982 to 2002 in the RDD response rates of eight methodologically similar Children's Oncology Group case-control studies (6). Epidemiologists working with the Children's Oncology Group have therefore begun to use alternatives to RDD (7).

Birth registries are an appealing option for recruitment of young US children as controls for several reasons. Registries are population based, with an estimated 99 percent of all births occurring in the United States in 2002 enumerated by state registrars (8), and collect data that facilitate matching and allow characterization of nonparticipants. Some disadvantages particular to this method are difficulty in tracing and administrative burden. Several investigations have used rosters generated by birth registries to recruit subjects (9–13), demonstrating that multiple states will participate in such research. However, since some projects originate from within state health departments that may be entitled by statute to access birth registry data, it is unclear whether researchers unaffiliated with government would meet with similar success.

We are currently conducting case-control studies of hepatoblastoma (designated COG-AEPI04C1) and infant leukemia (designated COG-AE24) within the Children's Oncology Group, the North American cooperative clinical trials group whose 200-plus members treat a large majority of children diagnosed with cancer at 0–5 years of age (14, 15). These two studies of pediatric malignancies that affect less than 150 children each in the United States (2) represent the first to attempt to elicit cooperation with all US birth registries for recruitment of controls. Here we report details of registry participation, approval, and availability of data. We also describe the ability of the limited number of participating registries to describe the pediatric population of the entire United States by use of census data (16). We anticipate that this report will be of use to investigators studying other rare childhood diseases.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
We contacted the registrars of the 50 states, the District of Columbia, and New York City (which maintains a separate registry) between December 2004 and June 2005 to determine the potential for cooperation in control selection for the study of hepatoblastoma. The first phase of the infant leukemia study used RDD for control selection (17), while the second phase was intended to extend the case series for assessment of gene-environment interaction (18). Birth registry control selection was later added to the protocol as a supplement; because of limited funds, we contacted only the 15 states that contributed the largest number of cases in the study's first phase for control selection in the second.

From each registry, we requested approval of a control recruitment procedure in which we would receive rosters of potential controls with accompanying demographic data, track subjects to their present address, send a letter of invitation, and follow up with a recruitment phone call. Data collection for both studies includes an hour-long maternal telephone interview and optional buccal cell sample collection from mother and child. We allowed 6 months for registries to respond to our request before considering it passively rejected. The University of Minnesota Institutional Review Board approved both studies.

A simple random sample of births between 2002 and 2006 was sought for the infant leukemia study. In the hepatoblastoma study, births between 1994 and 2005 were initially selected within frequency matching strata of sex, age, birth weight, and year of birth. We requested parents' and children's names, parents' Social Security numbers (if allowed), and residential address at birth for tracking purposes. In addition, we requested parents' ages at the birth of the child, parents' years of education, parents' race and ethnicity, and child's sex, date of birth, birth weight, and gestational age.

Registry participation, reasons for nonparticipation, and operational details of interest to epidemiologists are reported as counts and percentages. We also compared participating and nonparticipating registries using data on households with children under 6 years of age from the US Census Bureau's year 2005 American Community Survey Public Use Microdata Sample (16). The variables examined included number of children under the age of 6 years, race, ethnicity, educational attainment of adults, foreign origin, household language, poverty level, and median household income. We calculated weighted estimates using either the household or individual weights, depending on the variable under consideration, and standard errors using replicate weights. All analyses were performed using SAS, version 9.1, software (SAS Institute, Inc., Cary, North Carolina). Values of census characteristics are presented for qualitative assessment of difference, since the large sample size resulted in statistical significance of all comparisons at an alpha level of 0.05.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Thirty-two states, in which 75.8 percent of children 0–5 years of age resided in 2004, approved the recruitment of controls through their birth registries. The reasons cited for nonparticipation of the remaining birth registries were statute (n = 9), departmental policy (n = 4), nonresponse (n = 4), and lack of resources (n = 3). The latter three registries indicated their willingness to make first contact on behalf of the studies but lacked the personnel necessary to track and telephone potential controls. Details of the participating registries are shown in table 1.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Administrative characteristics and data release among US birth registries participating in control selection for case-control studies of hepatoblastoma and infant leukemia, 2004–2005

 
Three registries required that we modify our contact procedures. One registry required that written consent for contact be returned to us by mail before phoning potential controls. Another registry approved an opt-out contact procedure, in which the registry mailed a letter of invitation and return "no contact" postcard to the birth addresses of potential controls. Contact and demographic information were released if the postcard was not returned within 3 weeks. The third registry was willing to make first contact on behalf of the studies, was experienced in tracking subjects, and agreed to release anonymous demographic data of nonparticipants. Potential subjects had to return written consent for contact to the registry before data could be released to us.

Data sufficient to track and characterize potential controls were released by most registries, although parents' Social Security numbers were frequently unavailable. Father's name was not made available by three registries, which reflected concerns about revealing marital status at the time of birth.

Census characteristics of participating and nonparticipating registries are shown in table 2. Households with children aged 0–5 years in participating registries were qualitatively similar to those in the United States as a whole with respect to income, education, immigration, and race/ethnicity. Population density appeared substantially higher in participating registries, due to nonparticipation among several sparsely populated states.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Comparison of US census characteristics for 2005* among individuals residing in households with children 0–5 years of age by registry participation in control selection

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
As the usefulness of RDD declines, alternative methods of control selection are becoming necessary. In etiologic studies of childhood cancer, participants must frequently be recruited from across North America. Hospital-, neighborhood-, and school-based control selection would be impractical to implement in this context, while sibling or cousin controls elicit concern about overmatching (7). Family-based study designs may be used in a nationwide study but do not allow assessment of environmental main effects (19). Thus, we previously concluded that birth registries may offer the best alternative to RDD for studies of childhood cancer (7).

Several advantages of birth registry controls over RDD are readily apparent. Most children selected by birth registries will be eligible, since many that would not meet typical eligibility criteria in Children's Oncology Group studies (17, 20) may be screened out a priori. For instance, birth registries may easily select children by year of birth and also will screen out adoptees. In addition, most participating registries can release parental age, education, and race/ethnicity—three basic parameters with which prevalence of many risk factors relevant to offspring health varies (21–23); although it is possible that additional data available on US standard birth certificates (24) could be released, our studies did not request any. Many studies have evaluated agreement between birth certificate data and those of medical records (25). In general, parents' demographic data, the child's sex, and birth weight appear highly reliable, while pregnancy conditions, labor, and substance use do not. Recognizing that some error does occur in birth records, we can compare participating controls with nonparticipants to assess selection bias, in at least a rudimentary way. Correction for selection bias may also be possible if probabilities of exposure can be assigned to strata of maternal age, education, and race/ethnicity (26). This method of control selection also facilitates matching on infrequent birth characteristics. In the hepatoblastoma study, for example, it was of interest to match on very low birth weight (<1,500 g) (27–29), which comprises only 1.5 percent of births (30).

Our analysis of census characteristics demonstrated that children born in areas covered by participating registries should suffice to represent all US children, although perhaps not for some risk factors of interest. For example, the geographic distribution of agriculture may be such that the participating areas are not representative of the country as a whole. However, in general, it appears reasonable to compare cases born anywhere in the United States with controls selected from the subset of participating registries. Sensitivity analysis, with cases limited to those born in covered areas, may be performed to gauge the extent of bias.

Registries record the address at birth but do not track children, making the location of potential controls at current addresses a challenge of this methodology. Although it is too early to evaluate our studies' success in tracking participants nationwide, the experience in several states is instructive. The Pregnancy Risk and Monitoring System, which contacts new mothers between 2 and 6 months after giving birth in 23 states, achieved a contact rate of 82 percent in 2001 (12). Contact rates in case-control, rather than cohort, studies have been evaluated in three reports. The Northern California Childhood Leukemia Study enrolled birth registry controls up to 14 years after birth and reported a contact rate of 80.5 percent (10). Another study of childhood leukemia in New York collected data by mailed questionnaire from mothers of control children up to 30 years after birth (11). Eighty-eight percent of potential controls were locatable, insofar as they had valid addresses. Finally, a case-control study of birth defects based in seven Texas counties, which sought to interview mothers by telephone 3–4 years after the births of their children, achieved a contact rate of only 30 percent among potential controls; young maternal age and Black race decreased the likelihood of locating mothers (9). The cooperation rates among subjects who could be located were 91 (12), 78 (10), 55 (11), and 31 (9) percent, respectively, in the studies described above. These reports suggest that contact and cooperation of study subjects recruited through birth registries may vary by region, data collection modality, and time since birth.

Engaging with multiple birth registries carries a substantial administrative burden not encountered with control selection by RDD. For instance, we applied to 19 institutional review boards in order to access birth data. As a result, the start-up phase for control recruitment was prolonged, and we now have reduced flexibility to enact protocol changes that require institutional review board approval. Three states required contact protocol modifications, which may affect success in recruitment. Study material modifications were requested by over a third of registries, necessitating special effort to ensure that the correct letters and consent forms are sent to each potential participant. However, no registry requested changes to the interview instrument. Birth registries provide data, by varied formats and coding schemes, that must then be made consistent before uploading to study databases. Finally, most registries do not make data available for public release until 9 months after the end of a calendar year.

Our experience from two ongoing case-control studies of childhood cancer demonstrates that it is possible to select controls through a preponderance of US birth registries, which account for a large majority of young children and which adequately represent the characteristics of the nation as a whole. Cooperation of birth registries with other investigators may depend on study details (e.g., seriousness of the condition and participant burden), changes in statute, or bureaucratic caprice. The utility of this method of control selection will ultimately depend on the successful location and enrollment of subjects. Therefore, in addition to our ongoing studies, we are conducting a methodological study (National Institutes of Health grant R01CA108934; G. Bunin, Principal Investigator) that will systematically evaluate the validity of birth registry controls, up to 14 years of age, with an emphasis on minority participation. Birth registry control selection may be preferable to RDD, if field response rates equal those recently achieved by the latter method, given the availability of key demographic data regarding nonparticipants.


   ACKNOWLEDGMENTS
 
This research was funded by grants RO1CA111355, RO1CA079940, and U10CA98543 and by the Children's Cancer Research Fund, Minneapolis, Minnesota. A. F. O. was supported in part by a grant from the National Institute of Environmental Health Sciences (P30ES10126).

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Anderson RN, Smith BL. Deaths: leading causes for 2002. Natl Vital Stat Rep (2005) 53:1–89.[Medline]
  2. Ries LAG, Harkins D, Krapcho M, et al. SEER cancer statistics review, 1975–2003 (2006) Bethesda, MD: National Cancer Institute.
  3. Chang JS, Selvin S, Metayer C, et al. Parental smoking and the risk of childhood leukemia. Am J Epidemiol (2006) 163:1091–100.[Abstract/Free Full Text]
  4. Rosenbaum PF, Buck GM, Brecher ML. Allergy and infectious disease histories and the risk of childhood acute lymphoblastic leukaemia. Paediatr Perinat Epidemiol (2005) 19:152–64.[ISI][Medline]
  5. Preston-Martin S, Gurney JG, Pogoda JM, et al. Brain tumor risk in children in relation to use of electric blankets and water bed heaters. Results from the United States West Coast Childhood Brain Tumor Study. Am J Epidemiol (1996) 143:1116–22.[Abstract/Free Full Text]
  6. Bunin GR, Spector LG, Olshan AF, et al. Secular trends in response rates for controls selected by random-digit dialing in childhood cancer studies: a report from the Children's Oncology Group. Am J Epidemiol (2007) 166:109–16.[Abstract/Free Full Text]
  7. Ross JA, Spector LG, Olshan AF, et al. Invited commentary: birth certificates—a best control scenario? Am J Epidemiol (2004) 159:922–4. discussion 925.
  8. Technical appendix from vital statistics of the United States, 2002, natality (2003) Hyattsville, MD: National Center for Health Statistics. (http://wwwtest.cdc.gov/nchs/data/techap02.pdf).
  9. Gilboa SM, Mendola P, Olshan AF, et al. Characteristics that predict locating and interviewing mothers identified by a state birth defects registry and vital records. Birth Defects Res A Clin Mol Teratol (2006) 76:60–5.[CrossRef][ISI][Medline]
  10. Ma X, Buffler PA, Layefsky M, et al. Control selection strategies in case-control studies of childhood diseases. Am J Epidemiol (2004) 159:915–21.[Abstract/Free Full Text]
  11. Rosenbaum PF, Buck GM, Brecher ML. Early child-care and preschool experiences and the risk of childhood acute lymphoblastic leukemia. Am J Epidemiol (2000) 152:1136–44.[Abstract/Free Full Text]
  12. Shulman HB, Gilbert BC, Msphbrenda CG, et al. The Pregnancy Risk Assessment Monitoring System (PRAMS): current methods and evaluation of 2001 response rates. Public Health Rep (2006) 121:74–83.[ISI][Medline]
  13. Yoon PW, Rasmussen SA, Lynberg MC, et al. The National Birth Defects Prevention Study. Public Health Rep (2001) 116(suppl 1):32–40.
  14. Liu L, Krailo M, Reaman GH, et al. Childhood cancer patients' access to cooperative group cancer programs: a population-based study. Cancer (2003) 97:1339–45.[CrossRef][ISI][Medline]
  15. Ross JA, Severson RK, Pollock BH, et al. Childhood cancer in the United States. A geographical analysis of cases from the Pediatric Cooperative Clinical Trials groups. Cancer (1996) 77:201–7.[CrossRef][ISI][Medline]
  16. American Community Survey (ACS) public use microdata sample (PUMS), 2005 (2006) Washington, DC: US Census Bureau. (http://factfinder.census.gov/home/en/acs_pums_2005.html).
  17. Spector LG, Davies SM, Robison LL, et al. Birth characteristics, maternal reproductive history, and the risk of infant leukemia: a report from the Children's Oncology Group. Cancer Epidemiol Biomarkers Prev (2007) 16:128–34.[Abstract/Free Full Text]
  18. Khoury MJ, Flanders WD. Nontraditional epidemiologic approaches in the analysis of gene-environment interaction: case-control studies with no controls! Am J Epidemiol. (1996) 144:207–13.
  19. Weinberg CR, Umbach DM. Choosing a retrospective design to assess joint genetic and environmental contributions to risk. Am J Epidemiol (2000) 152:197–203.[Abstract/Free Full Text]
  20. Shankar S, Davies S, Giller R, et al. In utero exposure to female hormones and germ cell tumors in children. Cancer (2006) 106:1169–77.[CrossRef][ISI][Medline]
  21. Alcohol consumption among women who are pregnant or who might become pregnant—United States 2002. MMWR Morb Mortal Wkly Rep (2004) 53:1178–81.[Medline]
  22. Folate status in women of childbearing age, by race/ethnicity—United States, 1999 –2000, 2001–2002, and 2003–2004. MMWR Morb Mortal Wkly Rep (2007) 55:1377–80.[Medline]
  23. Staras SA, Dollard SC, Radford KW, et al. Seroprevalence of cytomegalovirus infection in the United States, 1988 –1994. Clin Infect Dis (2006) 43:1143–51.[CrossRef][ISI][Medline]
  24. Kirby RS, Salihu HM. Back to the future? A critical commentary on the 2003 U.S. national standard certificate of live birth. Birth (2006) 33:238–44.[CrossRef][ISI][Medline]
  25. Northam S, Knapp TR. The reliability and validity of birth certificates. J Obstet Gynecol Neonatal Nurs (2006) 35:3–12.[CrossRef][ISI][Medline]
  26. Greenland S. Basic methods for sensitivity analysis and external adjustment. In: Modern epidemiology—Rothman KJ, Greenland S, eds. (1998) Philadelphia, PA: Lippincott-Raven. 343–58.
  27. McLaughlin CC, Baptiste MS, Schymura MJ, et al. Maternal and infant birth characteristics and hepatoblastoma. Am J Epidemiol (2006) 163:818–28.[Abstract/Free Full Text]
  28. Reynolds P, Urayama KY, Von Behren J, et al. Birth characteristics and hepatoblastoma risk in young children. Cancer (2004) 100:1070–6.[CrossRef][ISI][Medline]
  29. Ansell P, Mitchell CD, Roman E, et al. Relationships between perinatal and maternal characteristics and hepatoblastoma: a report from the UKCCS. Eur J Cancer (2005) 41:741–8.[CrossRef][ISI][Medline]
  30. Martin JA, Hamilton BE, Sutton PD, et al. Births: final data for 2004. Natl Vital Stat Rep (2006) 55:1–101.[Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
166/7/852    most recent
kwm143v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Spector, L. G.
Right arrow Articles by Bunin, G. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Spector, L. G.
Right arrow Articles by Bunin, G. R.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?