Presented at the Fourth Biennial Cancer Survivorship Research Conference entitled “Cancer Survivorship Research: Mapping the New Challenges,” Atlanta, Georgia, June 18-20, 2008.
Adult childhood cancer survivors are at higher risk for developing late medical effects related to their cancer treatments. Health-promoting behaviors may reduce the risk of some late effects and the severity of others. This article describes the design and recruitment of the Chicago Healthy Living Study (CHLS), an on-going study designed to examine the health behaviors and body mass index of minority adult childhood cancer survivors compared with nonminority survivors and noncancer controls.
Survivors are identified by the hospital cancer registries at 5 treating institutions in the Chicago area; then, a multilevel recruitment plan will be implemented with the objective of enrolling 450 adult survivors of childhood cancer (150 in each racial/ethnic group). Simultaneously, 375 adult African-American, Hispanic, and non-Hispanic white noncancer controls (125 in each racial/ethnic group) living in the Chicago area will be recruited by using listed, targeted digit dialing. All participants will complete a 2-hour interview of questionnaires related to diet, physical activity, smoking, and associated mediators. Height and weight also will be measured.
In 2009, approximately 12,000 children will be diagnosed with cancer in the United States, and 80% will likely survive disease-free for >5 years.1 Fortunately, advances in treatment have led to dramatic improvements in survival, such that an estimated 1 in every 640 adults between ages 16 years and 44 years will be a survivor of childhood cancer.2, 3 Despite improved treatments, many pediatric cancer survivors develop adverse medical effects of treatment or “late effects” that include osteoporosis, cardiovascular disease (CVD), and secondary cancers.4-8 These late effects may be worsened by high-risk health behaviors.9 Adherence to health-promoting practices, such as consuming a healthy diet, engaging in regular physical activity, maintaining a healthy weight, and avoiding smoking, may be particularly beneficial to adult survivors of childhood cancers.
Health Behaviors in Adult Childhood Cancer Survivors
To our knowledge to date, approximately 18 published studies have examined the weight status and health behaviors of adult childhood cancer survivors.4, 10-26 Overall, the results suggest that, despite their childhood illness, this population has body habitus and health behavior profiles similar to those in the general population. Many consume diets that are high in fat, low in fruits and vegetables and do not include sufficient calcium.4, 11, 13, 20, 22, 25 Many also are physically inactive, with a minority meeting the Centers for Disease Control and Prevention (CDC) recommendation for regular physical activity.4, 11-14, 17, 18, 20, 22, 24, 25 These behaviors, along with the effects of some treatments, lead many survivors, particularly those who have survived acute lymphoblastic leukemia (ALL), to become overweight or obese.27-29 Smoking is 1 area that does not mirror the general population. Childhood cancer survivors appear to smoke either less or at similar rates compared with their siblings or age-matched peers.10, 15, 19, 21 However, any smoking is of concern given the higher risk for secondary cancers among many survivors because of their treatment exposure. Although the current literature provides a general understanding of the health behaviors of adult childhood cancer survivors, the methodological weaknesses of many studies limit their impact. Limitations include nonrepresentative or single-site convenience samples, lack of noncancer comparison groups, lack of theoretical frameworks to guide methodology, and/or use of subjective and/or nonstandardized measures. An additional and critical limitation is that studies to date have included an over-representation of non-Hispanic white survivors.
Addressing the Needs of Minority Childhood Cancer Survivors
The limitations of current studies have resulted in sparse knowledge regarding the health behaviors of minority childhood cancer survivors, interfered with the ability to recognize intervention needs, and prevented the development of appropriate interventions. There is only 1 published article that focuses on the health behaviors of minorities, although the study was not designed with that intent.12 Results indicated that African Americans and Hispanics were less likely to smoke than their non-Hispanic white counterparts, although >50% were physically inactive, a rate similar to that among non-Hispanic whites.12 A second study of physical activity among adult ALL survivors considered race/ethnicity in its analyses and reported that being a racial/ethnic minority was associated with a higher likelihood of not meeting CDC guidelines for physical activity.18 Neither study considered cognitive or cultural mediators, and only Florin et al18 included a noncancer comparison group. These methodological considerations are important, because research has demonstrated that minorities in the general population are less likely to engage in health-promoting activities.30-32 Some purport that this is caused in part by cultural attitudes, health beliefs, and limited knowledge regarding health promotion.33-35 Results from Florin et al18 suggest that their sample of ALL survivors was more likely to be inactive than participants in the 2003 Behavioral Risk Factor Survey Study. Further data are needed to understand whether and how the health behaviors of African-American and Hispanic survivors may differ from their noncancer-affected peers. This will allow for the development of appropriate interventions that potentially may enhance the health and quality of life in these underserved communities. In this report, we describe the design and recruitment strategy of a study aimed at increasing our knowledge of health behaviors among African-American, Hispanic, and non-Hispanic white childhood cancer survivors.
MATERIALS AND METHODS
Research Design and Sample
The Chicago Healthy Living Study (CHLS) is an observational, cross-sectional study with 2 primary objectives: 1) to describe and compare the influence of race/ethnicity (African American, Hispanic, or non-Hispanic white) and survivor status (cancer survivor or not) on individual health behaviors (diet, physical activity, smoking) and body mass index (BMI) and 2) to identify differences in the mediators of each health behavior and BMI between these groups. The study involves the collaboration of 5 Chicago-area healthcare institutions that treat childhood cancers. Recruitment is underway with the objective of enrolling 150 adult childhood cancer survivors within each of 3 ethnic/racial category (African American, Hispanic, and non-Hispanic white) equally mixed by sex. In addition, 125 noncancer controls will be recruited within each racial/ethnic category. Eligibility for all participants require that they 1) self-identify as African American/black, Hispanic/Latino, or non-Hispanic white/Caucasian; 2) are aged ≥18 years; 3) are not pregnant or <3 months postpartum; 4) have no significant cognitive delays or unmedicated mental illness; and 5) agree to provide informed consent and complete the study interview. Survivors also must have been diagnosed with a childhood malignancy before age 21 years and ≥5 years before study participation, must currently be cancer free, and must live/work within 2 hours' driving distance from the study site or be willing to come to the study site for the interview. Additional eligibility criteria for controls require that they have no prior diagnosis of any cancer with the exception of basal cell skin cancer nor any diagnosis of any chronic illness that would have affected their diet and/or exercise patterns before age 18 years (ie, juvenile diabetes, arthritis, diabetes). They also must be willing to come to the study site or to a community location for the interview. The study includes a 1-time interview and has been approved by the Institutional Review Board at the University of Illinois Chicago and all collaborating institutions.
The conceptual framework for the proposed study is based on an integration of Kornblith's model for psychosocial adaptation to cancer,36, 37 the Health Belief Model (HBM)38-40 and Social Learning Theory (SLT) (Fig. 1).41 Tenets of cultural competency theory also are incorporated.42
Kornblith's model of psychosocial adaptation to cancer holds that physical health and functioning are mediated by individual characteristics, social support, economic resources, and other stressors to determine levels of adaptation to cancer.36, 37 Health behaviors may be considered aspects of long-term adaptation to cancer. Figure 1 illustrates how the independent variables of cancer survivor status and ethnicity are mediated by health status, healthcare utilization, cognitive processes, social support, and sociocultural factors to determine the health behaviors and BMI of study participants.
Perceived vulnerability to illness and perceived importance of health promotion are mediators that come from the HBM,. The HBM is 1 of several theories used to understand why individuals engage in particular behaviors when confronted with illness or health problems38-40 According to the HBM, survivors of childhood cancers are more likely to practice a health-promoting behavior if they believe they are vulnerable, for example, to secondary cancers; if they believe the risk is serious; and if they believe that there are benefits to taking action that outweigh the costs.
Rationale for including attitudes, knowledge, and self-efficacy as mediators is based on SLT.43 SLT proposes that behavior and behavior change occur as a result of the dynamic interaction between behavior, cognition (attitudes, knowledge, self-efficacy), and the environment (social support).
Finally, the potential mediating effects of ethnic identity, spirituality, and cultural beliefs/practices also are being considered.42 Assessment of these variables will help to identify the inherent diversity that exists in the beliefs and practices of individuals from different ethnic and racial backgrounds and will be important to consider in developing interventions.
Adult Childhood Cancer Survivors
Survivors are being recruited in a staggered process through 5 institutions that treat approximately 90% of the children who are diagnosed with cancer in the Chicago area. Recruitment goals for each race/ethnicity are calculated for each site based on the approximate number of new cases diagnosed each year.
The multistep recruitment process begins with the pediatric cancer registry at each institution generating a list of patients who meet the eligibility criteria. This list is checked for accuracy of inclusion criteria and then serves as the potential research pool for recruitment. Recruitment begins with the least intensive effort and proceeds through 2 increasingly intensive steps. Invitation letters are printed on letterhead and signed by the physicians and nursing staff at each treating institution. A self-addressed, stamped response card (with an envelope to maintain confidentiality) is included. Survivors can decline to participate by returning the response card or by calling study staff. In such cases, the survivor is considered a refusal, and no further contact is made. If survivors send back the response card or call study staff indicating that they are interested, then a CHLS interviewer contacts them to screen for eligibility and, if they are eligible, to set up an interview appointment. Survivors who do not send back the response card or call within 2 weeks from its postage date are called by CHLS study staff to gauge their level of interest in participating. In most cases, the contact information contained in the pediatric cancer registries is incorrect. If the initial recruitment letter comes back as undeliverable and/or the telephone contact is incorrect, then recruitment efforts proceed to a third, more intensive level: “unable to locate” (UTL) searches (see Fig. 2). Some cases begin at the UTL level; those for which there was no address or an incomplete address and telephone number from the hospital pediatric registry. If the initial UTL search fails to provide accurate contact information, then a second search is conducted 6 months to 8 months later. When updated information is obtained, the recruitment cycle begins anew with the letter if an address is identified or with a telephone call if only a telephone number is identified. If no information can be obtained after 2 UTL searches, then the case is considered “lost.” Recruitment goals are set at each site, and, once a racial/ethnic group goal is met, recruitment ceases for that group at that particular site.
Outside of the multistep recruitment plan, recruitment also may occur in person at the treating institutions. A small number of adult childhood cancer survivors continue to see their oncology team for healthcare and are told about the study by their physician and/or nurse. Survivors also may self-refer if they hear about the study from other survivors or from media publicity about the study.
Control participants are being recruited through listed, targeted digit dialing by the Survey Research Laboratory (SRL) at the University of Illinois at Chicago based on a stratified sampling design with the goal of accumulating a specific number of controls stratified by age, sex, and race to ensure comparability with the survivor group. Recruitment is being done using a listed telephone sample for which the age and sex of a household member is known. SRL purchased 2 categories of sample files from Marketing Systems Genesys: sample files based on Hispanic surnames and sample files to target African Americans and non-Hispanic whites. To do this, the files exclude Asian, Indian, and Armenian surnames. This second sample file type is controlled further by the proportion of African Americans within particular zip codes. Before screening, these 2 file types are combined so that any individual who may not be eligible in 1 group (eg, an African American in the Hispanic surname file) can be used to fill 1 of the other target groups. Using the sample in this manner is more efficient than screening each group separately from each sample file.
Once contact is made by SRL, staff informs the respondent about the study. If the respondent is interested, then the SRL caller conducts a brief interview to determine eligibility and to gather information regarding age, sex, and ethnicity. This information is communicated to a CHLS interviewer who then contacts the potential participant to schedule the interview.
Participants complete a 1-time interview that is preceded by the process of informed consent. Survivors are given several location choices for completing the interview, including the study site, their treating institution (survivors only), their home or work (survivors only), or a community site (ie, public library). Controls are asked first to come to the study site; then, if they are unable or unwilling, we offer a community location. The interview requires approximately 2 hours and consists of questionnaires related to demographics, health behaviors, and mediating variables. Height and weight also are measured. Survivor participants are asked to sign a Health Insurance Portability and Accountability Act (HIPAA) waiver to access their medical record from their childhood cancer treating institution; however, this is not a requirement of participation. At the end of the interview, participants receive $50.00 for their time. Interviews are conducted by trained interviewers with extensive training and experience in collecting sensitive data. For quality-control purposes 10% of all interviews are recorded and reviewed for consistency.
USDA, US Department of Agriculture; CCSS, Childhood Cancer Survivor Study; SF-36, Medical Outcomes Study 36-item Short Form Health Survey; BMI, body mass index.
Single item in demographics questionnaire (yes/no)
Self-report by racial/ethnic group
American Lung Association Smoking Knowledge
CCSS knowledge related to cancer diagnosis and treatment
Perceived vulnerability to illness and importance of health promoting behaviors
Eating Habits and Exercise Confidence Surveys
Confidence questionnaire related to smoking behavior
Social support (friends/family)
Social Support and Eating Habits Survey
Social Support for Exercise Survey
National Health Interview Survey Health Care Utilization Items
Late Effects and Comorbid Conditions Rating Scale
Health Survey (SF-36)
Cultural Beliefs Questionnaire
Systems of Belief Inventory
Multiethnic identity measure
Height and weight
Brief Block 98 Food Frequency
Modified Activity Questionnaire
“Luxury Model” Smoking Behavior Questionnaire
Data are collected on sex, date of birth, self-identified ethnic group, survivor status, marital status, reproductive history, education, occupational status, and annual income.
The Short Acculturation Scale (SAS) for Hispanics is a 5-item self-report measure of acculturation.44 Responses are provided on a 5-point scale. The SAS is available in English and in an equivalent Spanish version. Concurrent validity of the SAS is supported with the Bi-directional Acculturation Scale.45 Only Hispanic participants complete this measure, and data are used descriptively within the group.
Diagnosis and treatment history are abstracted from medical records after HIPAA authorization is obtained from the participant. Data are collected on a structured abstraction form that queries concerning disease, stage at diagnosis, and detailed treatment history. Medical data will be used for descriptive purposes.
Mediator: Sociocultural Factors
The Multigroup Ethnic Identity Measure46 contains 15 items with 2 factors: 1) ethnic identity and affirmation and 2) belonging and commitment. This measure has been used in many studies and consistently has good reliability (α>.80) across a wide range of ethnic groups and ages.
This questionnaire contains 29 items that query about traditional and nontraditional cultural beliefs and practices related to cancer. This measure is based on an extensive interview that was developed by Lannin et al in a study of cultural factors in African-American and non-Hispanic white patients with breast cancer47 and was further informed by studies of Hispanic cultural beliefs and practices.48-50
Religious and Spiritual Beliefs and Practices
The Systems of Belief Inventory51 was developed for cancer patients/survivors and consists of 15 items and 2 subscales: religious/spiritual beliefs and devotional practices and social support obtained from religious peers and leaders.
Mediator: Cognitive Processes
Ten true/false items related to smoking were developed for this study based on information from the American Lung Association and the American College of Chest Surgeons.
This measure contains 7 items that query about knowledge related to healthy eating. Questions were selected from those included in the Prototype Notebook for the Food Stamp Nutrition Education program that was developed by the US Department of Agriculture's Economic Research Service in collaboration with Mathematica Policy Research, Inc.52
Physical Activity Knowledge
This questionnaire has 8 items that ask about the recommended mode, frequency, duration, and intensity related to physical activity. Questions for this measure were drawn from the interview developed by Morrow et al53 on knowledge related to the Surgeon General's report on physical activity.
Knowledge Related to Cancer Diagnosis and Treatment (Survivors Only)
This measure contains 7 questions that assess the survivor's knowledge of their cancer diagnosis, treatment, and potential late effects.54 These items come from the Childhood Cancer Survivors Study (CCSS), a large multi-institutional study of a cohort of over 10,000 adult childhood cancer survivors.55 These data will be compared with the information obtained from medical abstraction to evaluate the accuracy of survivors' responses.
This measure contains 2 items and was based on a measure that was developed specifically for adult survivors of pediatric cancers by Eiser et al.56 The first item asks respondents to rank, on a scale from 1 (a lot less important) to 5 (a lot more important), their perception of the importance of practicing certain health behaviors (ie, eat a healthy diet) compared with other individuals their age. Similarly, the second item queries regarding their perceived level of risk for developing several health conditions (ie, cancer, lung problems) compared with their peers.
Self-Efficacy for Eating and Exercise Behaviors
Ten items from Sallis et al's measure of self-efficacy for eating and exercise behavior were selected, after consultation with the author.57 This measure has strong psychometric properties.58 In our abbreviated version, 7 items ask about healthy eating self-efficacy (eating low-fat foods and healthy portions despite high-fat temptations), and 3 items assess exercise self-efficacy (ie, beliefs about one's ability to exercise 5 times a week in the face of barriers). In response to each item, participants rate how confident they are that they could really motivate themselves to do things like these consistently for at least 6 months. Ratings are made on a 5-point scale with responses ranging from “Sure I could not do it,” to “Sure I could do it.”
Mediator: Social Support
Social Support for Eating and Exercise
This measure is an abbreviated version of Sallis et al's social support for eating and exercise questionnaire.58 After consulting with the authors, 7 items were selected to assess social support for healthy eating, and 3 items were selected to measure social support for exercise. Respondents rate, on a 5-point scale from 1 (none) to 5 (very often), the frequency that friends and family have done or said certain things related to the respondents' efforts to change dietary or exercise habits.
Mediator: Health Status
Health-Related Quality of Life
The 36-item Short-Form Health Survey (SF-36)59 was constructed to survey health status in the Medical Outcomes Study. The SF-36 was designed for use in clinical practice and research, health policy evaluations, and general population surveys. The SF-36 includes 1 multi-item scale that assesses 8 health concepts: 1) limitation in physical activities because of health problems, 2) limitations in social activities because of physical or emotional problems, 3) limitations in usual role activities because of physical health problems, 4) bodily pain, 5) general mental health (psychological distress and well being), 6) limitations in usual role activities because of emotional problems, 7) vitality (energy and fatigue), and 8) general health perceptions. This instrument has been used widely with diverse healthy and clinical populations and has good reliability and validity.60
Late Effects and Comorbid Conditions Rating Scale
The Modified Cumulative Illness Rating Scale61, 62 is administered to gain information on comorbid conditions and late effects. This measure classifies comorbidities by 14 organ systems that may be affected and rates them according to severity from 0 to 4. This measure can generate 4 ratings, including total score, number of categories endorsed, severity index (total score/number of categories endorsed), and number of categories at Level 3.
Mediator: Healthcare Utilization
Access to healthcare, healthcare utilization, and health insurance status are being assessed by items drawn from the 2004 National Health Interview Survey.63
Outcomes: Body Mass Index and Health Behaviors
Body Mass Index
BMI is calculated as weight (kg)/height (m)2. Height is assessed using a portable stadiometer, and weight is assessed using a digital scale with shoes and outer clothing removed. To ensure quality control, 10% of all individuals are weighed and measured a second time by a second research assistant.
The Brief Block 98 Food Frequency Questionnaire (FFQ)64 asks study participants to report on the frequencies and amounts of 60 different food items. The FFQ was developed from National Health and Nutrition Examination Survey III food-intake data and includes a food list that was derived separately for African Americans, non-Hispanic whites, and Hispanics. Reliability and validity have been established for the measure in a wide range of age, sex, income, and ethnic groups.65, 66
The Modified Activity Questionnaire (MAQ)67 assesses leisure and occupational activity, television viewing, and inactivity because of disability. For leisure activity, respondents view a list of 29 popular activities (ie, walking, jogging, gardening) and select those that they performed on at least 10 different occasions in the last year. Respondents then provide information on average frequency and duration for each activity. For occupational activity, respondents provide information on common activities performed at work and transportation to/from work. The MAQ has been used in several large studies with diverse samples, including cancer survivors,68 and has well established reliability and validity.67
Historic and current use of tobacco is being assessed by the “Baseline Luxury Model” measure recommended by the Tobacco Working Group for use in the Behavior Change Consortium studies funded by the National Institutes of Health.69 This measure contains 14 items concerning current smoking, past smoking, quitting history, and other tobacco use.
Given the reliance on self-report measures that could invite socially desirable responses, the Marlowe-Crown Measure of Social Desirability70 is administered.
Power and Sample Size
Estimating power for survey data that subsequently will be analyzed using complex regression models is a difficult task. For any given analysis, power will depend on not only the effect size of the variable(s) in question, but also on the number and effect size for the covariates in the model. For our purposes, a conservative approach is to think of the design as a simple, 2-way factorial with race/ethnicity as 1 factor and cancer survivor status (yes/no) as the other, resulting in a 3 × 2 factorial. Although we are unlikely to achieve an exactly orthogonal result, because the departure from that should not be large, equal allocation of cases to non-Hispanic white, African-American, and Hispanic will assure maximal power for between-group comparisons. For a continuous outcome, using Cohen's71 definitions of effect sizes, we estimated that the projected sample size of roughly 750 cases would allow us to detect effect sizes of approximately ≥.2 both for the main effects and for the race/ethnic-by-survivor status interaction. Table 2 is based on calculations using PASS software. Results are shown in the table.
Table 2. Results of Calculations Using PASS Software
Fixed-Effects ANOVA Power Analysis
ANOVA indicates analysis of variance; DF, degree(s) of freedom.
With the addition of covariates to the model (eg, age, sex, socioeconomic status), we expect the error sum of squares to decrease and the power to increase. Most outcomes will be continuous, but a few (eg, marital status) will be categorical and will require logistic regression. For dichotomous outcomes, power depends on the baseline probability; for example, it is easier to detect a between-group difference of .4 versus .5 than a difference of .1 versus .2. For the latter comparison, which corresponds to an odds ratio of 2.2, we have 80% power for a dichotomous predictor, correlated with other independent variables at .20.
The CHLS is designed to examine the health behaviors of adult African-American and Hispanic childhood cancer survivors and compare these with the health behaviors of their non-Hispanic white counterparts and their noncancer-affected peers. The study seeks to recruit a sample that is mostly non-Hispanic white. Unlike the CHLS, samples of previous childhood cancer survivor studies have been predominantly non-Hispanic white.4, 10, 11, 13-17, 19-26, 55 For example, the CCSS, perhaps the largest and most well known study of adult survivors, has a sample that is 87% non-Hispanic whites, 2% African Americans, and 5% Hispanics.55
Whereas the CCSS includes noncancer-affected siblings as a comparison group, the CHLS includes noncancer-affected peers. This is a major contribution to the literature, because only 1 of the 2 published studies12, 18 that addresses health behaviors in African-American and Hispanic survivors included a noncancer-affected comparison group.18 By using a listed, targeted digit dial approach, the CHLS will recruit 325 racially/ethnically matched noncancer-affected controls. Inclusion of the control sample will allow CHLS to provide important insight into the similarities and differences in health behaviors and their mediators between minority cancer survivors and their peers who have not been affected by such an experience.
The CHLS is set apart further from previous studies of health behaviors in adult childhood cancer survivors by its use of validated measures to assess diet, smoking, and physical activity. In addition, the BMI is calculated based on a standardized measurement for height and weight, in contrast to much of the published work to date, which relies on self-reported weight and nonstandardized measures of health behaviors.11, 12, 14, 20, 22, 25, 27, 29
A final strength of CHLS is the underlying goal of collecting data that will inform future interventions. This goal guided much of the study methodology. To our knowledge to date, the majority of research on the diet, smoking, and physical activity patterns of childhood cancer survivors has been descriptive.10-12, 14, 18, 19 Few studies have examined factors that potentially influence health behaviors and that require attention for interventions to be successful in promoting behavior change.16, 17, 22, 25 The CHLS assesses several mediators, the selection of which was guided by the principals of well established health behavior theories. The CHLS also includes sociocultural mediators. Understanding the cultural beliefs and attitudes that are part of an individual's “health culture” will be key in developing interventions that serve the needs of minority survivors.
In conclusion, studies of non-Hispanic white survivors of childhood cancer report diet, physical activity, and smoking behaviors similar to those in the general population. Few studies have considered the status of these behaviors in minority survivors of childhood cancers. Several lines of research support the rationale for the current study's focus on minority survivors: 1) Health disparities are prevalent in minority populations; 2) up to 75% of adult survivors of childhood cancer experience adverse late effects of treatment; and 3) obesity, CVD, and secondary cancers are late effects that may be amenable to preventive health behaviors. In the general population, minorities are more likely to exhibit high BMIs and fewer health-promotion behaviors, have inadequate knowledge of preventive lifestyles, and have less healthful attitudes than non-Hispanic whites. Thus, minority survivors of childhood cancer potentially are at greater risk than nonminorities for developing the late effects of obesity, CVD, and secondary cancers. Currently, there are insufficient data to support or refute this hypothesis. CHLS will provide critical data with which this hypothesis can be considered and upon which appropriate interventions can be developed.
Conflict of Interest Disclosures
Sponsored by National Cancer Institute Grant R01CA116750.
Cosponsored by the National Cancer Institute's Office of Cancer Survivorship, the Office of Cancer Survivorship of the Centers for Disease Control and Prevention, and the American Cancer Society's Behavioral Research Center.