Sample Bias Among Women With Retained DNA Samples for Future Genetic Studies

Aagaard-Tillery, Kjersti MD, PhD; Sibai, Baha MD; Spong, Catherine Y. MD; Momirova, Valerija MS; Wendel, George Jr MD; Wenstrom, Katharine MD; Samuels, Philip MD; Cotroneo, Margaret RN; Moawad, Atef MD; Sorokin, Yoram MD; Miodovnik, Menachem MD; Meis, Paul MD; O’Sullivan, Mary J. MD; Conway, Deborah MD; Wapner, Ronald J. MD; for the National Institute of Child Health and Human Development Maternal–Fetal Medicine Units Network

Obstetrics & Gynecology:
doi: 10.1097/01.AOG.0000241536.19539.14
Original Research

OBJECTIVE: To evaluate whether women who agree to future use of their biologic specimens for genetic studies reflect the larger study population from which they are derived.

METHODS: Women were questioned as to the future disposition of their maternal and fetal DNA samples upon enrollment in a multicenter, observational study originally designed to identify factor V Leiden mutation carriers and prospectively ascertain the estimated rate of pregnancy-related venous thromboembolism and adverse pregnancy outcome. Univariate and multivariate analyses was carried out on the 5,003 of 5,188 enrolled women who indicated their desire regarding future disposition of their DNA samples.

RESULTS: Among these 5,003 women, 20.1% desired that their samples be discarded and not available for future genetic studies. Multivariate analysis demonstrated that women who agreed to subsequent use of samples were less likely African-American (odds ratio [OR] 0.6, 95% confidence interval [CI] 0.4–0.7) or Hispanic (OR 0.4, 95% CI 0.3–0.5), and more likely to use tobacco (OR 1.2, 95% CI 1.0–1.6) than those who desired that their samples be discarded.

CONCLUSION: Genetic samples from women agreeing to their use in a sample repository may not be representative of the index study cohort. This should be considered in their subsequent interpretation and generalizability.


In Brief

Genetic samples from women agreeing to their use within the context of larger studies may not be representative of the study cohort.

Author Information

From the 1Department of Obstetrics and Gynecology, University of Utah School of Medicine, Salt Lake City, Utah; 2University of Tennessee, Memphis, Tennessee; 3National Institute for Child Health and Human Development; Rockville, Maryland, 4Biostatistics Coordinating Center, George Washington University, Washington, DC; and the Departments of Obstetrics and Gynecology, 5University of Texas—Southwestern, Dallas, Texas, 6University of Alabama at Birmingham, Birmingham, Alabama; 7Ohio State University, Columbus, Ohio, 8Magee Women’s Hospital/University of Pittsburgh, Pittsburgh, Pennsylvania, 9University of Chicago, Chicago, Illinois, 10Wayne State University, Detroit, Michigan, 11University of Cincinnati, Cincinnati, Ohio, 12Wake Forest University, Winston-Salem, North Carolina, 13University of Miami, Coral Gables, Florida, 14University of Texas—San Antonio, San Antonio, Texas, and 15Thomas Jefferson University, Philadelphia, Pennsylvania.

*For members and institutions participating in the Maternal–Fetal Medicine Units Network, see the Appendix.

Supported by the National Institute of Child Health and Human Development through the Maternal–Fetal Medicine Units Network (U10-HD27869, U10-HD21414, U01-HD36801, U10-HD34208, U10-HD27860, U10-HD34116, U10-HD34136, U10-HD27861, U10-HD34122, U10-HD21410, U10-HD27915, U10-HD34210, U10-HD27905, and U10-HD27917) and the National Institutes of Health’s Office of Research on Women’s Health.

The authors thank Dr. Donna Dizon-Townson for her initial conception and execution of the index study with the Maternal–Fetal Medicine Units Network. In addition, the following subcommittee members participated: protocol development and coordination between clinical research centers, Julia McCampbell, RN; protocol and data management and statistical analysis, Elizabeth Thom, PhD; and protocol development and oversight, Michael Varner, MD.

Corresponding author: Kjersti Aagaard-Tillery, MD, PhD, University of Utah Health Sciences Center, Department of Obstetrics & Gynecology, 30 North 1900 East, Room 2B200, Salt Lake City, Utah 84132; e-mail:

Article Outline

Recruitment and retention of subjects representative of the entirety of the study population are integral to the success of any clinical trial. Although the Human Genome Project identified the commonality of the primary genomic sequence by virtue of race, recent advances in our understanding of human health and disease has underscored the role of racial and environmental influence on genomic expression.1,2 As the scientific community attempts to understand the complex array of gene–environment interactions that likely govern disease, a key aspect to the determination of genetic variance involves prospectively acquired samples that consider racial and ethnic composition, sociodemographic factors, and environmental influence. The applicability across racial, ethnic, and socioeconomic groups will be limited if the population from which the biologic material is derived is biased, particularly if this bias is unrecognized.

Although previous research has examined the theoretical willingness of the general population to participate in genetic research, empiric evidence regarding potential sociodemographic bias with respect to consent for storage of DNA for future genetic research has only recently been considered.3 In an analysis of trends of consenting individuals, investigators analyzed data from National Health and Nutrition Examination Surveys, and demonstrated that only non-Hispanic black race or ethnicity was a significant independent predictor of refusal to allow future genetic research.3 These findings were limited by two factors. First, fewer than 1,700 of the 4,880 eligible study participants were young, reproductive-aged individuals, arguably the cohort most likely to be labeled as “healthy,” although they may harbor any number of genetic conditions that have yet to present clinically. Second, because the primary aim of the study did not involve biologic sample collection, those declining participation in the genetic repository might represent a cohort that either opposes genetic research per se or has concerns regarding participation in studies involving biologic sample collection. To date, there are no published reports regarding bias of a range of potential influential factors (ethnicity, parity, maternal habits and habitus, maternal age, and socioeconomic factors) in a population of pregnant women being asked to allow unrestricted research access to their biologic samples for future genetic studies (MEDLINE and PubMed search, 1966–2006; search terms “genetic samples,” “genetic repository,” “genetic consent,” “biologic samples,” “sample bank,” “biosample,” “DNA repository,” “sample repository,” alongside the terms “pregnancy,” “gestation,” “maternal,” or “fetal”).

For these reasons, we sought to assess whether a study that was initially designed to determine the rate of pregnancy-related thromboembolic events and adverse pregnancy outcomes with an a priori unknown genetic condition (factor V Leiden mutation) might demonstrate a bias in subjects consenting to participate in a genetic repository.4,5 Specifically, we aimed to assess population bias among women that participated in a multicenter, prospective observational study and consented for use of their DNA samples in future genetic studies.

Back to Top | Article Outline


The index study was a prospective, observational multicenter study conducted within the National Institute of Child Health and Human Development Maternal–Fetal Medicine Units Network.4 Consenting women with a singleton pregnancy before 14 weeks of gestation were eligible unless one of the following exclusion criteria were present: current anticoagulation therapy, known factor V Leiden status, diagnosis of antiphospholipid syndrome, previous venous thromboembolism, known fetal demise, planned pregnancy termination, uncertain gestational dates, intention to deliver outside of a Maternal–Fetal Medicine Unit Network site, or participation in another research study that might influence the primary or secondary outcomes. The primary outcome was the occurrence of venous thromboembolism. institutional review board approval was obtained at each of the 13 clinical centers and at the data coordinating center.

Written informed consent for the index study was obtained from each of 5,188 enrolled subjects from April 2000 to August 2001. Upon enrollment, women were questioned as to the future disposition of their biologic samples. Specifically, three options were given 1) consent to utilize samples in a nonidentifiable fashion in future studies (“unrestricted permission”), 2) consent to recontact for permission before future use (“ask permission”), 3) instructions to discard samples after the index study (“discard”). Subsequent statistical analysis was carried out among these 5,003 women. At each of the 13 enrollment sites, consent attainment was exclusively performed by trained Maternal–Fetal Medicine Unit Network research nurses for both the index study as well as for desired future disposition of samples in a uniform fashion.

Information on race or ethnicity was based on self-identification, and categorized as African American, white, Hispanic, Asian, and Native American; individuals who did not self-select into these categories were classified as “other.” Marital status was classified as “Married” if the woman was married or living with a partner. Total years of schooling were reported numerical values up to a maximum of 16 years. Patients were queried as to tobacco use with the question “Have you smoked cigarettes at any time during this pregnancy?” Body mass index was calculated from reported prepregnancy weight and the recorded value for maternal height at the first prenatal visit. Each woman had a venous blood sample drawn and submitted to a central reference laboratory (DNA Diagnostic Laboratory, University of Utah) for batch analysis for the presence of the factor V Leiden mutation as previously described.4

Baseline, demographic, and outcome data were forwarded to the data coordinating center as were all of the laboratory test results of both the women and their infants. No interventions were undertaken as a part of the index study. In addition to maternal demographic information, fetal birth weight and length, mode of delivery (cesarean, spontaneous vaginal delivery, mid, low, or outlet vacuum or forceps, or vaginal breech) indication for cesarean or assisted-vaginal delivery, stillbirth, fetal gender, and neonatal intensive care unit admission was similarly gathered and used in the secondary analysis.

Univariate analysis was employed to assess patient desire regarding subsequent use of their biologic samples at each of the three levels (“unrestricted permission,” “ask permission,” and “discard”). To assess for differences between those agreeing to or declining future use of their samples in a logistic regression model, acceptance at two levels was assessed by collapsing “unrestricted permission” and “ask permission” into the single category “use samples,” with comparison to “discard.” In the multivariate regression model, acceptance by race or ethnicity, tobacco use, marital status, parity, eventual mode of delivery, and fetal outcome were assessed. Consent rates, 95% confidence intervals, and multivariate logistic regression analysis were computed with SAS 8.2 (SAS Institute, Inc., Cary, NC). Comparisons were performed using Fisher exact test or χ2 analysis for discrete variables and Wilcoxon or Kruskal-Wallis test for the continuous variables. A nominal P<.05 was considered significant. No adjustment was made for multiple statistical comparisons.

Back to Top | Article Outline


Among the 5,003 of 5,188 eligible women enrolled in the index study, 96.4% indicated their desire regarding future disposition of their biologic samples in a genetic repository. The mean maternal age was 25.1 years (± 5.6 years), and 1,544 (30.9%) of the women were nulliparous. The index study population was equally distributed with respect to ethnicity, and comprised 34.5% African American, 31.9% Hispanic, 31.5% white, 0.8% Asian, 0.2% Native American, and 1.1% “Other.”

Although the majority (n 3,626, or 72.5%) of participants enrolled in the index genetic study were willing to grant unrestricted permission, a significant minority (n of 1,008, or 20.1%) directed that their samples be discarded, whereas 7.4% (n 369) requested that investigators first ask permission.

Univariate analysis assessing acceptance for consent at each of the three levels (“discard,” “ask permission,” and “unrestricted permission”) revealed differences of samples available for unrestricted future genetic testing by virtue of a number of maternal factors (Table 1). Specifically, maternal race, maternal age, tobacco use, and attained level of education were significantly associated with an increased likelihood to elect to have samples discarded (Table 1). The lowest consent rates for unrestricted future use was among Hispanics (67.6%) compared with white and other patients (75.7%), who had the highest rate. Conversely, the most likely to elect to discard samples after the index study were African Americans (19.9%) and Hispanics (28.3%), P<.001 for an overall difference between the three race or ethnicity categories. Similarly, patients who smoked were more likely to discard their samples (20.9% of nonsmokers compared with 15.0% of smokers electing to discard, P<.001, as were multiparous women, 21.3% compared with 17.6% discard, P=.011).

When the variables maternal age and attained level of education were assessed for significance (Table 1), the women who elected for permission to be requested before using the samples were older (26.3±5.7, overall P<.001) and more educated (median and interquartile range of attained years of education 13 [12, 16], overall P<.001).

To determine which factors independently contributed toward willingness to consent to future use, multivariate analysis in a logistic regression model was employed (Table 2). Of the potential confounders, our analysis recognized African-American race (OR 0.6, 95% CI 0.4–0.7), and Hispanic race (OR 0.4, 95% CI 0.3–0.5) to be associated with the lack of desire for samples to be used.

Given that this study was conducted at multiple sites, with varying regional preferences and racial or ethnic backgrounds, we controlled for this factor to determine whether the observed differences would persist. Multivariate analysis, including primary recruitment site (for those enrolling more than 200 patients, n=7) revealed differences regarding consent dependent upon enrollment site and ethnicity (Table 3). African Americans were significantly less likely to allow subsequent use of their samples (OR 0.7, 95% CI 0.5–0.9; Table 3). This finding persisted when we repeated this model controlling for the four sites with the highest recruitment (which enrolled more than 70% of the patients) [(OR 0.70, 95% CI 0.6–0.9)].

Back to Top | Article Outline


In this study, we sought to assess whether reproductive-aged women initially enrolling in a genetic study would demonstrate a bias in their willingness to participate in a genetic repository. Our analysis of the data indeed recognizes inclusion bias of samples available for future genetic testing with respect to maternal race, age, marital status, and attained education. Moreover, nearly one fifth of index study participants elected to have their samples discarded. These findings suggest that women might not grant unrestricted permission for future genetic testing. Moreover, our analyses suggest that those allowing future testing are not necessarily reflective of the entire study cohort.

As a result of the far-reaching ethical issues surrounding genetic research, a consensus statement was generated nearly a decade ago addressing the need for specific consent for genetic studies.6 In the interval since, only one other study (National Health and Nutrition Examination Survey) has attempted to discern whether there exists a bias in the available biologic material in prospectively acquired samples.3 National Health and Nutrition Examination Survey is an ongoing cross-sectional national survey conducted by the National Center for Health Statistics and Centers for Disease Control and Prevention that provides statistics on the health and nutritional status of the civilian U.S. population through household interviews and a standardized physical examination in special mobile examination centers.3 Since 1999, the National Health and Nutrition Examination Survey has also enabled the establishment of a national repository of material for future genetic research; participants in National Health and Nutrition Examination Survey are allowed the option of consenting for sample collection after completion of the initial survey. Importantly, pregnancy was a contraindication to participation in the National Health and Nutrition Examination Survey studies. Clearly it is important that any such genetic repository be an unbiased representation of the population at-large if the information gleaned is to be generalizable, clinically relevant, and biologically pertinent.

Concerns for ethnic and sociodemographic bias in human studies are not unique to genetic research. Such concerns have led to the development of a national U.S. mandate for the inclusion of women and minorities in study populations.7–13 Despite such a mandate, bias persists in multiple aspects of clinical research.14–17 The National Health and Nutrition Examination Survey trial was the first to suggest a similar bias trend exists in genetic repositories, and our results presented herein substantiate and expand these authors’ findings. Recent advances in human genome research, pharmacogenetics, medical genetics, and the evolving field of epigenetics have furthered our understanding of the interactions of the heritable genome and chromatin structure in the causal pathways employed in the development of human disease.18–23 To further our understanding of the role of such phenomena as single gene polymorphisms and allelic frequencies, genomic imprinting, and gene regulation achieved by transcriptional silencing, prospectively acquired samples need to be available and accessible for use in translational research.

The strengths of our study are several. First, we enrolled more than 5,100 pregnant women in a prospective, observational study from 13 sites representative of diverse geographic and sociodemographic regions across the United States. Our design therefore avoided bias by avoiding the limitations inherent to retrospective analyses of regional and racial or ethnic influences. Second, consent for subsequent use was given before enrollees and their providers would have any potential knowledge of mutation status, and therefore eliminated the possibility of bias based on discovery of a genetic condition. Third, our secondary analysis explored bias in consent according to both known maternal factors, as well as ad hoc factors unknown at enrollment. Given the lack of evident bias according to eventual pregnancy outcome, our findings cannot be explained by random event or chance alone. Finally, by virtue of the fact that the index study enrolled women who were willing to participate in a prospective analysis of the role of a specific gene mutation on clinical outcome, our observed inclusion bias regarding use of their samples for subsequent genetic analyses cannot be explained by a general reluctance to participate in genetic research. Thus, in a diverse population with broad-based acceptance of genetic research, there still existed a significant fraction of participants who did not wish future unrestricted use of their samples for similar purposes.

Others exploring the issues of recruitment and retention of underrepresented and underserved subjects in research trials have suggested that logistic issues such as remuneration and complexity of the research protocol are influential factors.14 Of note, recent analyses examining impediments to participation in medical research studies among African Americans found that racial differences in the willingness to participate are primarily due to the lower level of trust of medical research among minority populations.14,24–26 Our data support this notion, because enrollees were willing to participate in the index genetic study, but were unwilling to grant unrestricted permission for future use.

Our secondary analysis recognizes inclusion bias among samples available for future genetic testing according to maternal race and smoking status. The implications of our findings are two-fold. First, researchers cannot assume consent for future genetic research among index study participants, because one in five study participants directed that their samples be discarded. Second, future studies in which permission to retain biologic samples from pregnant women is requested could benefit from efforts aimed at enrolling and consenting individuals who are less likely to grant such permission.

Back to Top | Article Outline


1. Bjornsson HT, Fallin MD, Feinberg AP. An integrated epigenetic and genetic approach to common human disease. Trends Genet 2004;20:350–8.
2. Egger G, Liang G, Aparicio A, Jones PA. Epigenetics in human disease and prospects for epigenetic therapy. Nature 2004;429: 457–63.
3. McQuillan GM, Porter KS, Agelli M, Kington R. Consent for genetic research in a general population: The NHANES experience. Genet Med 2003;5:35–42.
4. Dizon-Townson D, Miller C, Sibai B, Spong CY, Thom E, Wendel G Jr, et al. The relationship of the factor V Leiden mutation and pregnancy outcomes for mother and fetus. Obstet Gynecol 2005;106:517–24.
5. Dizon-Townson DS, Nelson LM, Jang H, Varner MW, Ward K. The incidence of the factor V Leiden mutation in an obstetric population and its relationship to deep vein thrombosis. Am J Obstet Gynecol 1997;176:883–6
6. Clayton EW, Steinberg KK, Khoury MJ, Thomson E, Andrews L et al. Informed consent for genetic research on stored tissue samples. JAMA 1995;274:1786–92.
7. Berger JT. Culture and ethnicity in clinical care. Arch Intern Med 1998;158:2085–90.
8. Hood RG. Inaugural address: realizing the promise of health care equality: our leadership mandate for the new millennium. J Natl Med Assoc 2000;92:417–23.
9. Kreiger N, Ashbury F, Cotterchio M, Macey J. A qualitative study of subject recruitment for familial cancer research. Ann Epidemiol 2001;11:219–24.
10. Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race-, sex-, and age-based disparities. JAMA 2004;291:2720–6.
11. Swanson GM, Ward AJ. Recruiting minorities into clinical trials: toward a participant friendly system. J Natl Cancer Inst 1995;87:1747–59.
12. Britton A, McKee M, Black N, McPherson K, Sanderson C, Bain C. Threats to applicability of randomised trials: exclusions and selective participation. J Health Serv Res Policy 1999;4:112–21.
13. U.S. Department of Health and Human Services, Public Health Service, NIH. Inclusion of women and minorities in study populations: instructions and information. Washington (DC): U.S. Government Printing Office; 1990.
14. Gorelick PB, Harris Y, Burnett B, Bonecutter FJ. The recruitment triangle: reasons why African Americans enroll, refuse to enroll, or voluntarily withdraw from a clinical trial. An interim report from the African-American Antiplatelet Stroke Prevention Study (AAASPS). J Natl Med Assoc 1998;90:141–5.
15. Vidaver RM, Lafleur B, Tong C, Bradshaw R, Marts SA. Women subjects in NIH-funded clinical research literature: lack of progress in both representation and analysis by sex [published erratum appears in J Womens Health Gend Based Med 2000;9:1041–3]. J Womens Health Gend Based Med 2000; 9:495–504.
16. McDermott MM, Lefevre F, Feinglass J, Reifler D, Dolan N, et al. Changes in study design, gender issues, and other characteristics of clinical research published in three major medical journals from 1971 to 1991. J Gen Intern Med 1995;10:13–8.
17. Hoyo C, Reid ML, Godley PA, Parrish T, Smith L, Gammon M. Barriers and strategies for sustained participation of African-American men in cohort studies. Ethn Dis 2003;13:470–6.
18. Willett WC. Balancing life-style and genomics research for disease prevention. Science 2002;296:695–8.
19. Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer. Nature Rev Genet 2002;3:415–28.
20. Suter CM, Martin DI, Ward RL. Germline epimutation of MLH1 in individuals with multiple cancers. Nat Genet 2004;36:497–501.
21. Stoughton RB, Friend SH. How molecular profiling could revolutionize drug discovery. Nat Rev Drug Discov 2005;4: 345–50.
22. Gerhardt A, Scharf RE, Beckmann MW, Struve S, Bender HG, Pillny M, et al. Prothrombin and factor V mutations in women with a history of thrombosis during pregnancy and the puerperium. N Engl J Med 2000;342:374–80.
23. Sankar P, Cho MK, Condit CM, Hunt LM, Koenig B, Marshall P, et al. Genetic research and health disparities. JAMA 2004;291:2985–9.
24. Corbie-Smith G, Thomas SB, Williams MV, Moody-Ayers S. Attitudes and beliefs of African Americans toward participation in medical research. J Gen Intern Med 1999;14:537–46.
25. Roberson NL. Clinical trial participation. Viewpoints from racial/ethnic groups. Cancer 1994;74 suppl:2687–91.
26. Sengupta S, Strauss RP, DeVellis R, Quinn SC, DeVellis B, Ware WB. Factors affecting African-American participation in AIDS research. J Acquir Immune Defic Syndr 2000;24: 275–84.
Back to Top | Article Outline

Members and institutions participating in the Maternal–Fetal Medicine Units Network: University of Utah: D. Dizon-Townson, M. Varner, K. Anderson, K. Jolley, A. Guzman; University of Alabama at Birmingham: D. Rouse, A. Northen, K. Bailey; George Washington University Biostatistics Center: E. Thom, A. Arrieta, L. Leuchtenburg; National Institute of Child Health and Human Development: C. Spong, M. Klebanoff, S. Pagliaro, D. McNellis, K. Howell; University of Chicago: A. Moawad, P. Jones, G. Mallett; University of Pittsburgh Magee-Womens Hospital: S. Caritis, K. Lain, T. Kamon; University of Texas Southwestern Medical Center: K. Leveno, J. McCampbell, S. Williams; Wayne State University: M. Dombrowski, G. Norman, P. Lockhart, C. Sudz; University of Cincinnati: M. Miodovnik, T. Siddiqi, H. How, N. Elder, W. Knox; Wake Forest University: M. Harper, M. Swain, K. Lanier; Ohio State University: J. Iams, F. Johnson, C. Latimer; University of Miami: M. J. O’Sullivan, F. Doyle; University of Tennessee: B. Sibai, W. Mabie, R. Ramsey; University of Texas at San Antonio: O. Langer, D. Dudley, S. Barker, D. Skiver; Thomas Jefferson University: R. Wapner, A. Sciscione, M. Talucci, M. DiVito Cited Here...

Cited By:

This article has been cited 2 time(s).

Obstetrics & Gynecology
Patient Attitudes Toward Genotyping in an Urban Women’s Health Clinic
Haas, DM; Renbarger, JL; Meslin, EM; Drabiak, K; Flockhart, D
Obstetrics & Gynecology, 112(5): 1023-1028.
PDF (186) | CrossRef
Obstetrics & Gynecology
Pharmacogenomics of Maternal Tobacco Use: Metabolic Gene Polymorphisms and Risk of Adverse Pregnancy Outcomes
for the Eunice Kennedy Shriver National Institute of Child Health, Human Development (NICHD) Maternal–Fetal Medicine Units Network (MFMU), ; Aagaard-Tillery, K; Spong, CY; Thom, E; Sibai, B; Wendel, G; Wenstrom, K; Samuels, P; Simhan, H; Sorokin, Y; Miodovnik, M; Meis, P; O'Sullivan, MJ; Conway, D; Wapner, RJ
Obstetrics & Gynecology, 115(3): 568-577.
PDF (305) | CrossRef
Back to Top | Article Outline
© 2006 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.