PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Biol Blood Marrow Transplant. Author manuscript; available in PMC 2010 August 1.
Published in final edited form as:
PMCID: PMC2911964
NIHMSID: NIHMS218200

Methodological and Logistical Considerations to Study Design and Data Collection in Racial/Ethnic Minority Populations Evaluating Outcome Disparity in Hematopoietic Cell Transplantation

Abstract

Outcome disparity associated with race or ethnicity in the United States has been observed in hematopoietic cell transplantation (HCT). The underlying reasons for such disparity are not known. In the United States, an optimal study of healthcare disparity by race or ethnicity involves consideration of both biological and psychosocial determinants which requires an adequately powered, prospective cohort study design. In order to better characterize the nature and quantify the magnitude of the many impediments relevant to conducting a successful prospective study involving racial or ethnic minorities in HCT, we conducted a feasibility study to help guide planning of a larger scale, outcome and disparity study in HCT. The primary questions to be addressed in the study were: 1) can we establish a racially or ethnically diverse patient sample who will respond to a survey focused on socio-demographic, economic, health insurance, cultural, spiritual and religious well-being, and social support information?; 2) what is the retention rate in the study over time?; and 3) what is the quality of the data collected from the patients over time? The challenges we faced in conducting this multicenter feasibility study are summarized in this report. Despite the difficulty in conducting disparity studies in racial and ethnic minorities, such studies are essential to ensure that people of all ethnic and racial backgrounds have the best chance possible of benefiting from HCT.

Evidence of Outcome Disparity by Race or Ethnicity in HCT

Outcome disparity associated with race or ethnicity in the United States is a phenomenon many investigators have observed across various diseases and medical and surgical procedures. An association between African American or Hispanic minority status and outcomes has also been seen among patients with hematologic malignancies undergoing hematopoietic cell transplantation (HCT). (1-4) Retrospective analyses of large multi-center data from the Center for International Blood and Marrow Transplant Research (CIBMTR) showed that in the setting of HLA-matched sibling transplantation for acute and chronic leukemia, Hispanics had a higher risk of treatment-failure and overall mortality when compared to Caucasians. (1-2) A separate analysis of patients treated at Fred Hutchinson Cancer Research Center demonstrated a higher risk of mortality for African-Americans when compared to Caucasians. (3) In unrelated HCT, African-Americans and/or Hispanics had higher overall mortality when compared to Caucasians. (3-4). However, outcome disparities by race or ethnicity are not seen in autologous transplantation. (3,5) The reasons for the differential findings in autologous and allogeneic procedures and the underlying factors causing outcome disparity in allogeneic recipients, are not known.

Potential Causes of Outcome Disparity by Race or Ethnicity in HCT

Allogeneic HCT is a complex high-risk procedure that is usually offered with curative intent. Its effectiveness in the treatment of leukemia and lymphoma is believed to be due to a graft-versus-tumor effect. (6) Variables which can influence outcomes include both biological and non-biological factors. A key biological determinant in outcomes for unrelated allogeneic HCT is the degree of human leukocyte antigen (HLA)-matching between donor and recipient. Ongoing improvements in the characterization of HLA biology have decreased the risk of graft-versus-host disease and contributed to improved outcomes in unrelated donor transplantation in recent years. (7-10) However, the degree to which HLA diversity in different races affects outcomes is less clear because most HLA studies include few minorities. Many clinical factors associated with race also have prognostic importance, such as age at disease onset, cancer cell biology, disease stage at diagnosis, and co-morbid medical conditions.

Multiple non-biologic factors may also directly or indirectly affect the clinical outcomes after allogeneic HCT, with some associated with race or ethnicity. These factors include: patient socioeconomic status (SES), healthcare access (insurance coverage, geographic barriers, time to acquiring treatment, follow-up care) and delivery (compliance, treatment options received), and psychosocial and cultural factors (11-14). Ethnic minorities in the United States are more likely than Caucasians to have lower SES and have inadequate medical insurance coverage leading to poor outcomes (15). Lower SES is associated with lower overall healthcare usage, fewer surveillance tests, and lower quality ambulatory and hospital care (16). A significantly higher percentage of Hispanics lack adequate medical insurance compared to Whites, Blacks, and Asians (17). Other barriers to access and delivery of healthcare observed among minority groups, such as skepticism toward efficacy of treatment (18), lower satisfaction with healthcare received (19), and language barriers (20) could also potentially affect survival. Previous studies show that minority children, including Blacks and Hispanics, make fewer physician visits compared to white children (21). Little is known about how follow-up care varies by race or ethnicity among recipients of complicated treatments, such as HCT, and how this may affect outcomes. Ethnic differences in reacting to and coping with a new state of health and well-being are not well understood in HCT, but are linked with survival among patients with breast cancer (22). All the above factors may be accentuated in a procedure such as allogeneic HCT where a significant amount of recovery occurs in the outpatient setting and risks of treatment-related complications are extremely high and enduring beyond the procedure itself.

Formulating the Study Question and Appropriate Study Design

Based on past studies in HCT and in the general medical literature, there are three areas that need to be examined in studies of health disparity by race or ethnicity to establish causality: 1) the race / ethnic group distinction to use, 2) the biological determinants to consider, and 3) the psycho-social determinants to explore including the role of SES. There are no consistent standard definitions of race or ethnicity in the context of health-related studies. Definitions are usually study specific and can either be viewed as an index of biological distinction or as an index of social grouping. For the most part, racial or ethnic distinctions are usually designated by the patients themselves. This manner of designation, although acceptable, is truly imprecise and is a combination of both biological and social definition.

In the United States, any study exploring causes of healthcare disparity by race or ethnicity requires consideration of both biological and psychosocial determinants and demands an adequately powered, prospective cohort study design. While retrospective studies are essential in moving this area of research, only prospective studies will allow for a comprehensive assessment of all relevant factors that may explain disparate outcomes. These factors should include both biological and non-biological entities, such as behavioral characteristics and individual decision-making, and should cover factors before, during and after HCT. Many of the non-biologic variables can not be captured retrospectively or through administrative databases. The goal of prospective research is to identify areas where interventions may be targeted. This also allows for the magnitude of the association between race/ethnicity and outcomes to be estimated while controlling for biologic factors or understanding the complex relationships among many factors. This also allows for temporality and plausibility of the relationship to support ‘cause and effect’, something retrospective studies are not able to support with certainty.

Feasibility of Studying Racial and Ethnic Minority Patients

Table 1 summarizes the impediments encountered in the execution of a prospective cohort study designed to explore the reason for disparate outcomes by race or ethnicity.

Table 1
Study Impediments to Investigate Disparity Study by Race or Ethnicity in HCT

First is the issue of ability to accrue adequate numbers of racial or ethnic minorities. In a study published in 2005, we estimated that for 47% of 116 HCT centers in the US, racial or ethnic minorities represented at least 20% of patients undergoing allogeneic transplantation. These were considered as high-minority HCT centers. Additionally, Hispanics represent 10% or greater of the patient population at only 24% of US centers. The high-minority centers also transplanted significantly fewer patients than low-minority centers (median 40/year versus 66/year). (23) This suggests that to acquire an adequate sample of racial or ethnic minorities, a multicenter study targeting high-minority HCT centers is essential, but that center differences may confound the analysis.

Second, racial or ethnic minorities are difficult to recruit. Studies have shown that racial or ethnic minorities are less likely to participate in clinical studies in general and to HCT studies in particular. (24, 25) A third barrier is a confluence of issues related to study performance of racial or ethnic minorities over time. Some of these issues include loss to follow-up, potential language barriers related to consent procedure, Institutional Review Board (IRB) approval, survey questionnaire design, completion rates of various psychometric instruments, and considerations for statistical analyses due to the nature of sampling bias. While most biological factors are routinely collected as part of the clinical and research activities performed by transplant centers, data from patients pertaining to psychosocial factors are not. To date, even large registries such as the CIBMTR, have not been successful in obtaining SES and behavioral variables. The ability to collect this information from racial and ethnic minorities is also not well characterized.

Considering the magnitude of the many potential impediments relevant to conducting a successful prospective study involving racial or ethnic minorities in HCT, we conducted a feasibility study to guide the planning of a larger scale, outcome and disparity study in HCT. Conducting a feasibility study was crucial because of increasing concern that racial or ethnic minorities may not participate. Additionally, a feasibility study would allow the researchers to convince funding agencies that they have a grasp of the potential barriers to be encountered in a prospective study focused on modifiable causes of disparity in outcomes. The primary questions of the study to be addressed were: 1) can we establish a racially or ethnically diverse patient sample who will respond to a survey focused on socio-demographic, economic, health insurance, cultural, spiritual and religious well-being, and social support information?; 2) what is the retention rate in the study over time?; and 3) what is the quality of the data collected from the patients over time? The challenges we faced in conducting this multicenter feasibility study are summarized below.

Lessons Learned

A. Study Funding

While in general, well-designed studies of disparity by race or ethnicity are interesting and of sufficient importance to warrant funding at the national level, the funding of feasibility studies is not as attractive. After two attempts at revising the proposed study within a period of 18 months, the study principal investigator (FRL) secured a $50,000 grant from the Medical College of Wisconsin through institutional funds awarded by the American Cancer Society which made the study possible.

B. Center Recruitment and Sample Population

The study group met in person during the 2003 Annual Meeting of the American Society for Blood and Marrow Transplant at Keystone, CO to discuss the final study plan and identify a contact person at each center. Five high-minority centers with at least 20% of total transplants performed in racial or ethnic minorities were identified. Two centers from Texas (MD Anderson Cancer Center in Houston and South Texas Veterans Health Care System/University of Texas Health Science Center at San Antonio) were invited because of a high potential recruitment for Hispanic patients, one center in Chicago (University of Chicago) and another in Milwaukee (Medical College of Wisconsin) were recruited for their potential for high African-American recruitment, and one center in San Francisco (University of California San Francisco) was recruited for its potential for high Asian-American recruitment. All centers performed at least 100 HCT annually. We felt convenience sampling, as opposed to randomized selection of HCT centers, was acceptable since based on the number of minorities transplanted annually this is the likely approach we would adapt in a larger study. The centers also had investigators who are part of the study group and had the support staff to conduct the actual recruitment and follow-up of patients.

The feasibility study was designed to recruit patients at least 18 years of age who were undergoing related or unrelated allogeneic HCT. The enrollment period was 6 months with a plan to follow patients for 1 year. One Caucasian patient was recruited for every 4 racial or ethnic minorities consented. Of the 5 centers identified, only 3 were able to open the study for enrollment. Both centers which failed to open the study had coordination issues when personnel who were trained in the recruitment of patients changed prior to start of the study. Such changes in study coordination staff are a common occurrence, and this possibility should be anticipated when designing study processes. Tables 2a and b show the actual number of patients recruited for this study, and the projected number of patients who can potentially be enrolled if the study were extended to more centers and conducted over a longer study period. These numbers do not include potential loss to follow-up over time.

While the participation rate is remarkably high, 100% at baseline, information obtained from patients at 6 months and 1 year dramatically decreased to less than 60%. While most patients continued to be seen by the transplant centers and information regarding post-transplant events were recorded for clinical and research purposes, obtaining longitudinal information on survey instruments from patients proved challenging. This is partly due to different follow-up practices used by centers and the level of coordination between the clinical and research coordinators. A potential solution to this problem would utilize a centrally located coordinating center in charge of sending follow-up questionnaires. However, the IRBs of the centers participating in this study did not allow their patients to be contacted in person by a third party given the current laws regarding patient confidentiality. A strong case for having a central coordinating center responsible for processing and following up survey questionnaires, in retrospect, should had been made. Investigators should devote enough time to consult with IRBs of participating centers to identify how best to utilize a central coordinating center while maintaining utmost patient confidentiality. This approach may have prevented two issues that caused decreased data collection over time: 1) inattention from clinical coordinators who are busy taking care of the medical issues, and 2) failure to collect and process data in a timely fashion, preventing implementation of corrective actions in real time. However, centralization of data processing should be combined with close communication between the coordinator from the recruiting center and the patients so that the close relationship between the treating center and patient is harnessed into participation and study retention.

C. Data Collection

Some of the logistical difficulties with obtaining data over time were discussed in the previous section. This section will discuss the feasibility of collecting diverse types of information that are not normally collected for clinical or research purposes.

One of the most commonly hypothesized reasons for disparity in outcome by race and ethnicity is disparate income. While census-based income approximation inferred from residential ZIP codes or address level analysis are commonly utilized with varying accuracy, but these methods do not accurately reflect the actual total household income of patients and is therefore subject to limitations. Patient income is not usually collected by transplant centers, and if collected, is not incorporated into the medical records. Direct ascertainment of income from the patient may be the most accurate way to obtain these data. As shown in Table 3, approximately 20% of patients in the study did not report information regarding their income, either as a definite amount in dollars or as a range. If accurate and complete data cannot be collected, this has profound implications in addressing the role of socio-economic status on outcomes of racial or ethnic minorities. Information about health care insurance plan details of the patients also proved difficult to gather. While patients are able to report the type of insurance they have, 47% of the patients in the study were not able to provide details about their co-payments, or amount of monthly premiums. If these data are critical to an analysis, an alternative plan on how to obtain them is warranted. For instance, data on income may best be collected during a routine psychosocial interview by a social worker prior to transplant when financial needs are discussed. Insurance information may best be obtained from people in charge of securing insurance clearance prior to transplant, or review of the plan language. The expectation for missing data should therefore be considered in the statistical analyses, including power calculations. Plans for multivariate analyses should specify how missing data will be handled and the potential of these missing data to contribute bias to the results. Plans for sensitivity analyses may also need to be specified.

Table 3
List of data collected and their respective completion rates in percent

Surprisingly, while income and insurance data were hard to collect, other potentially sensitive socio-demographic information collected at baseline was more complete. Questions addressing socio-demographic characteristics, socio-cultural aspects of medical decision making, belief systems, level of religiosity and spirituality, and other psychometric instruments measuring social support, patient satisfaction and general adherence to medical regimens were obtained with an acceptable 5-10% missing data rate.

D. IRB Issues

Any study focused on racial or ethnic minorities appears to trigger increased scrutiny from IRBs. Increased sensitivity regarding how racial or ethnic minorities have been treated in historic clinical trials may explain some of this discrepancy, as does their role as a “special population”. However, this appears to have led to a surprising degree of inconsistency in the recommendations arising from different IRBs regarding the study conduct. For example, the original protocol stipulated that the questionnaire would be translated into Spanish with the hope of increasing Hispanic participation. It was interesting to discover that this can be viewed as preferential treatment of one racial or ethnic group over another and was perceived as a violation of equity. While this concern did not prevent the study from opening, additional protocol changes were required. In a multicenter setting where a harmonized protocol is desired, the need to individualize protocols for each center adds time and coordinator costs. IRBs also varied greatly in their activation speed, with the interval between submission and approval ranging from 2 months to 6 months. In retrospect, it is interesting to note that achieving IRB approval to open a multicenter clinical trial of an experimental drug may be easier than to execute an observational study dealing with disparity according to race and ethnicity. Many institutions and IRBs appeared unfamiliar with disparity studies and applied different criteria in their reviews. Familiarity with these criteria during the study design phase might have addressed issues that delayed the timely start and completion of the study.

E. Personnel Issues

The funding for this feasibility study allowed us to invite all study coordinators from the 5 participating centers to come to a central location for training. Topics included review of the protocol, the subject recruitment and consent procedures, the sampling framework for the study, the use of necessary tracking tables, and the data forms. The discussion allowed questions to be clarified and facilitated camaraderie and relationship development among the participating centers. The funding did not however cover any salary support for any personnel involved in the different centers. It took another 6 months from the initial meeting of the study coordinators to finally get the study started in 1 center after securing IRB approvals. Centers 2 and Center 3 followed after another 6 months. Patients were followed for 12 months, and data were obtained and processed for a total of 12 months. After 5 years of meetings, securing funding, study coordination, IRB approvals, patient recruitment, patient follow-up, and data form processing, we completed this feasibility study.

Within those 5 years, the main PI changed institutions and relegated the IRB related- and data management tasks to another PI, who also changed institutions halfway through the study. A study coordinator in one center who trained for the study and was responsible for recruiting patients had to take a medical leave. This prevented the study from even starting at that center as there were no other substitute personnel who could perform the needed tasks. The PI of another institution changed affiliations prior to start of the study, and therefore the infrastructure that allowed recruitment of patients at that center was lost.

Comments

This feasibility study was able to recruit a diverse group of racial or ethnic minorities in the setting of allogeneic HCT, with a high initial participation rate. While it is possible to study disparity between whites and non-whites (a grouped category of all minorities) and its causes, it may not be possible to study specific racial or ethnic groups, for instance African-Americans, Hispanics or Asians compared with Caucasians because of limited numbers. A study focused on one racial or ethnic group is likely to suffer from inadequate statistical power, unless the differences between groups are large. One possible solution is to use other outcomes that are known to differ widely by race or ethnicity (at least in other disease models), like medical service utilization, or perhaps re-hospitalization. The ability to track these alternative outcomes needs to be defined well to prevent ascertainment bias prior to conducting a large scale multicenter study. Mechanisms to optimize continued participation also appear valuable

This feasibility study also showed that socio-economic data may need to be obtained from multiple sources. A considerable number of patients may not volunteer information on income, at least in the context of research, but may be willing to divulge this information in the context of psychosocial evaluation by a social worker. Considering the amount of potential missing information on income, this would have profound implications in the analyses of studies, since income has been implicated rather consistently as potential contributor to disparity in outcomes. Other types of data (psychological, cultural, religiosity, spirituality) can be obtained in this test population with a high degree of completion. The potential for IRB-related delays in study initiation must be considered when planning a large study on disparity by race or ethnicity in HCT. A main coordinating center should help integrate all IRB requirements as part of study planning and achieve a second critical objective of better data tracking and processing. This may improve the ability to collect patient-derived information during the follow-up phase of the study.

Finally, the goal of racial and ethnic studies is to identify modifiable factors causing differences in outcome. Once identified, successful interventions to address these factors may ultimately improve survival of minorities and narrow the disparity gap. This perspective focuses less on biologic determinants of outcome, for example, disease stage and HLA-matching which might not be improved, and instead tries to sort out complex factors tied to race and ethnicity that affect the success of HCT. Despite the difficulty in conducting disparity studies in racial and ethnic minorities, such studies are essential to ensure that people of all ethnic and racial backgrounds have the best chance possible of benefiting from HCT.

Acknowledgments

Funding: This work was supported by the Medical College of Wisconsin Cancer Center Institutional Research Fund

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

1. Serna D, Zhang MJ, Baker KS, et al. Trends in Survival Rates after Allogeneic Hematopoietic Stem Cell Transplantation For acute and Chronic leukemia by Race in The United States and Canada. J Clin Onc. 2003 Oct 15;21(20):3754–3760. [PubMed]
2. Baker KS, Loberiza FR, Yu Hongmei, et al. Outcome of Ethnic Minorities with Acute or Chronic Leukemia treated with Hematopoietic Stem Cell Transplantation in the United States. J Clin Onc. 2005 oct 1;:7032–7042. [PubMed]
3. Mielcarek M, Gooley T, Martin PJ, et al. Effects of race on survival after stem cell transplantation. Biol Blood Marrow Transplant. 2005 Mar;11(3):231–9. [PubMed]
4. Baker KS, Hassebroek A, Ballen K, et al. Impact of ethnicity and socioeconomic status on outcome of unrelated hematopoietic cell transplantation. Blood. 2007 Nov 16;110(11):901.
5. Hari P, Majhail NS, Hassebroek A, et al. Similar outcomes among African-Americans and whites after autologous hematopoietic cell transplantation. Blood. 2008 Nov 16;112(11):274.
6. Horowitz MM, Gale RP, Sondel PM, et al. Graft-versus-leukemia reactions after bone marrow transplantation. Blood. 1990 Feb 1;75(3):555–62. [PubMed]
7. Szydlo R, Goldman JM, Klein JP, et al. Results of allogeneic bone marrow transplants for leukemia using donors other than HLA-identical siblings. J Clin Oncol. 1997 May;15(5):1767–77. [PubMed]
8. Flomenberg N, Baxger-Lowe LA, Confer D, et al. Impact of HLA class I and class II high-resolution matching on outcomes of unrelated donor bone marrow transplantation : HLA-C mismatching is associated with a strong adverse effect on transplantation outcome. Blood. 2004 Oct 1;104(7):1923–30. [PubMed]
9. Oh H, Loberiza FR, Zhang MJ, et al. Comparison of Graft-versus-Host Disease and Survival After HLA-Identical Sibling Bone Marrow Transplantation in Different Ethnic Populations. Blood. 2005 Feb 15;105(4):1408–16. [PubMed]
10. Lee SJ, Klein J, Haagenson M, et al. High-resolution donor recipient HLA-matching contributes to the success of unrelated donor marrow transplantation. Blood. 2007 Dec 15;110(13):4576–83. [PubMed]
11. Mandelblatt JS, Yabroff KR, Kerner JF. Equitable access to cancer services. Cancer. 1999;86:2378–2390. [PubMed]
12. Fiscella K. Is lower income associated with greater biopsychosocial morbidity? J Fam Pract. 1999;48:372–377. [PubMed]
13. Gornick ME, Eggers PW, Reilly TW, et al. Effects of race and income on mortality and use of services among Medicare beneficiaries. N Engl J Med. 1996;335:791–799. [PubMed]
14. Ulcickas Yood M, Cole Johnson C, Blount A, et al. Race and differences in breast cancer survival in a managed care population. J Natl Cancer Inst. 1999;91:1487–1491. [PubMed]
15. Wong MD, Shapiro MF, Boscardin WJ, Ettner SL. Contribution of major diseases to disparities in mortality. N Engl J Med. 2002;347:1585–1592. [PubMed]
16. Fiscella K, Franks P, Gold MR, Chaney CM. Inequality in care: addressing socioeconomic, racial, and ethnic disparities in health care. JAMA. 2000;283:2579–2584. [PubMed]
17. Mills RJ. Washington DC: United States Census Bureau; 2001. Health Insurance Coverage: 2000. Available at: http://www.census.gov/prod/2001pubs/p60-215.pdf.
18. Fiscella K, Franks P, Clancy CM. Skepticism toward medical care and health care utilization. Med Care. 1998;36:180–189. [PubMed]
19. The Henry J. Kaiser Family Foundation. Race, Ethnicity & Family Care: A Survey of Public Perceptions and Experiences. The Henry J. Kaiser Family Foundation; Menlo Park CA: 1999. pp. 1–32.
20. Guendelman S, Wagner T. Hispanics' experience within the health care system. In: Hogue CJR, Hargraves MA, Scott Collins K, editors. Minority Health in America: Findings and Policy Implications from the Commonwealth Fund Minority Health Survey. The Johns Hopkins University Press; Baltimore MD: 2000. pp. 19–46.
21. Flores G, Bauchner H, Feinstein AR, Nguyen UDT. The impact of ethnicity, family income, and parental education on children's health and use of health services. Am J Public Health. 1999;89:1066–1071. [PubMed]
22. Reynolds P, Hurley S, Torres M, et al. Use of coping strategies and breast cancer survival: results from the black/white cancer survival study. Am J Epidemiol. 2000;152:940–949. [PubMed]
23. Schwake C, Eapen M, Lee SJ, et al. Differences in Characteristics of United States Hematopoietic Stem Cell Transplantation Centers by Proportion of Ethnic Minorities. Biol Blood Marrow Transplant. 2005 Dec;11(12):988–998. [PubMed]
24. Ford JG, Howerton MW, Lai GY, et al. Barriers to recruiting underrepresented populations to cancer clinical trials: a systematic review. Cancer. 2008 Jan 15;112(2):228–241. [PubMed]
25. Lee SJ, Fairclough D, Parsons SK, et al. Recovery after stem cell transplantation for hematologic diseases. J Clin Oncol. 2001;19:243–252. [PubMed]