Racial Discrimination, Disease Activity, and Organ Damage: The Black Women's Experiences Living With Lupus (BeWELL) Study

David H. Chae; Connor D. Martz; Thomas E. Fuller-Rowell; Erica C. Spears; Tianqi Tenchi Gao Smith; Evelyn A. Hunter; Cristina Drenkard; S. SamLim


Am J Epidemiol. 2019;188(8):1434-1443. 

In This Article


Sample and Procedures

The cross-sectional, observational data used in this study are from the Black Women's Experiences Living with Lupus (BeWELL) Study. Participants were recruited from the Georgians Organized Against Lupus (GOAL) cohort, which drew primarily from the Georgia Lupus Registry.[64] The Georgia Lupus Registry is a population-based registry funded by the Centers for Disease Control and Prevention and designed to estimate the prevalence and incidence of SLE in metropolitan Atlanta, Georgia.[6] The Georgia Lupus Registry includes a full spectrum of patients, from mild to severe cases of SLE, and from all levels of socioeconomic strata. To maximize ascertainment of potential cases, a broad range of case-finding sources was used, including hospitals, health care providers (i.e., rheumatologists, dermatologists, nephrologists), commercial laboratories, and population databases. Hospital-based laboratories and regional pathology laboratories were also queried for results to identify patients with potential SLE. Data from larger commercial laboratories and the Centers for Medicare and Medicaid Services End-Stage Renal Disease database were also screened. Other unique databases, such as from the Veterans Administration, Medicaid claims, other state databases (e.g., hospital discharge), and electronic medical record systems, were analyzed. The result was one of the largest, population-based lupus epidemiology registries ever in the United States, with more than 1,500 people with validated lupus diagnoses meeting the American College of Rheumatology classification criteria for SLE (≥4 criteria) or 3 criteria with a diagnosis of SLE by a board-certified rheumatologist.[65] GOAL is further enhanced through recruitment of participants from the Lupus Clinic of Grady Memorial Hospital, a large public hospital in Atlanta, as well as from diverse community rheumatologist practices.

Eligibility criteria for the BeWELL Study were as follows: consent given to participate in the GOAL cohort, self-identification as black or African American; between 18 and 79 years of age; living in metropolitan Atlanta; and able to read, write, and understand English and respond to questions on a computer. We attempted to contact a total of 710 potentially eligible women who were enlisted in GOAL during the recruitment period, from April 2015 to May 2017. Attempted contact occurred initially through mail, which included study information and a request to contact study staff by telephone or by returning an interest reply form in a prepaid envelope. For those who did not respond in 2 weeks, study staff followed up through telephone calls. We were unable to reach 102 women. Of the remaining 608 participants, 12 did not meet eligibility criteria and 55 refused to participate (refusal rate = 9.2%); 103 women who were contacted could not be scheduled despite repeated attempts. This left a total sample size of 438 participants.

We compared the 260 black women in GOAL who were believed to be eligible but did not participate with the women who participated in BeWELL. Examining responses provided in the GOAL survey, we found that BeWELL participants were younger at the time of recruitment in the GOAL cohort (mean = 46.1 (standard deviation (SD), 12.3) years vs. 47.9 (SD, 12.9) years; P < 0.001), were diagnosed with SLE at a younger age (mean = 31.3 (SD, 11.0) years vs. 34.6 (12.0) years; P < 0.001), and had higher levels of disease activity (mean Systemic Lupus Activity Questionnaire score = 17.6 (SD, 9.1) vs. 15.8 (9.1); P < 0.001). Furthermore, examining the GOAL survey, BeWELL participants were more likely to be poor (50.1% vs. 40.7%; P = 0.03). There were no significant differences between the groups in terms of other SLE characteristics (i.e., disease duration, organ damage) or sociodemographic variables (i.e., marital status, education, employment, insurance).

Respondents were assessed primarily on-site at the Division of Rheumatology of the Emory University School of Medicine; 20 respondents participated through home visits. Trained lay interviewers assessed demographic characteristics and measures of organ damage and disease activity. More sensitive questions, including those assessing racial discrimination, were self-administered via computer-assisted software. Signed informed consent was obtained from all study participants. The median duration of study visit was 2.2 hours. All protocols and procedures were approved by the Institutional Review Board of Emory University.


SLE outcomes. SLE activity was measured using the Systemic Lupus Activity Questionnaire, a validated, patient-reported measure developed to track disease activity.[66,67] The questionnaire includes 24 items related to disease activity in the past 3 months, such as fatigue, fever, oral ulcers, rashes, vasculitis, myalgias, and joint swelling. Items were grouped and weighted, with possible scores ranging from 0 to 44. Higher scores indicate greater disease activity.

SLE organ damage was measured using the Brief Index of Lupus Damage (BILD), a validated, patient-reported measure of damage due to SLE in 12 organ systems; the index is used in clinical research studies.[68–70] Cumulative organ damage is an important outcome and predicts death, physical function, quality of life, and disability. The BILD enables researchers to assess major irreversible damage to an organ or system since the onset of SLE and present for at least 6 months. Items are endorsed as present or absent, with possible scores ranging from 0 to 30. Higher scores indicate greater organ damage.

Racial discrimination. Racial discrimination was measured using the Experiences of Discrimination measure, which asks participants, "Have you ever experienced discrimination, been prevented from doing something, or been hassled or made to feel inferior in any of the following situations because of your race, ethnicity or color," followed by 9 specific domains: at school; getting a job; at work; getting housing; medical care; service at a store or restaurant; obtaining credit or a loan; on the street or in a public setting; and from the police or in the courts.[71–73] Response choices were "no," "once," "two or three times," and "four or more times." We examined 2 scoring methods: 1) the situation version, which is a count of the number of items endorsed at least once, and which ranges from 0 to 9; and 2) the frequency version, which is calculated as the mean score of items with the following values assigned to response choices: 0 = no; 1 = once; 2.5 = 2 or 3 times; and 5 = 4 or more times.

Covariates. Age in years was measured on the basis of date of birth. Years since diagnosis were calculated on the basis of response to 1 of the following: the number of years and months since being diagnosed; the month and year of diagnosis; or the age at diagnosis. Relationship status was categorized as married or in a marriage-like relationship; romantic relationship; divorced, separated, or widowed; or single. Socioeconomic covariates were as follows: education (less than high school, high school, some college, or college graduate or advanced degree), work status (full-time; part-time; out of labor force, including retired, homemaker, or student; or not working, including those unemployed, laid-off, or unable to work due to health or disability), insurance status (private, public, or none), and ratio of household income to the poverty threshold. Household income in the past month was reported in categories of $500 increments, from which we took the midpoint of the response category and multiplied by 12. For those who volunteered past-year household income, responses were recorded in categories of $5,000 increments, from which we took the category midpoint to represent annual household income. A follow-up question assessed whether the income reported was before or after taxes; for those reporting that it was after taxes, we calculated the pre-tax amount on the basis of Georgia income tax rates for participant interview year.[74] We calculated the ratio of household income to the federal poverty threshold on the basis of the number of adults and children in the household.[75]

Health-related covariates were body mass index, examined continuously from height and weight, which were measured using standardized protocols; self-reported days of exercise per week in the past year; self-reported current smoking status (yes vs. no); and information on current SLE medication use from lists brought by participants to the interview, in addition to a checklist of lupus medications that interviewers went through with each participant. In the current study, SLE medication used was coded as yes versus no for the following: steroids (e.g., prednisone, medrol, methylprednisolone), antimalarials (e.g., hydroxychloroquine sulfate), and other immunosuppressant drugs (e.g., methotrexate, cyclophosphamide, cyclosporine, mycophenolate, dapsone, azathioprine, belimumab, rituximab).

Analysis Plan

Eleven participants (2.5%) with missing data for any of the variables being investigated were excluded from analyses, leaving a total analytic sample size of 427 participants. We specified multivariable linear regression models examining SLE activity. We also used multivariable linear regression to examine log-transformed BILD, given its right-skewed distribution. Results were substantively similar to models examining BILD continuously. We also compared models using the situation versus frequency scoring methods for the Experiences of Discrimination measure, which did not lead to different conclusions. Here, we present results using the frequency version of the Experiences of Discrimination measure. Results from models using the situation version to score the Experiences of Discrimination measure are available upon request. Nested models were specified, controlling for demographic, socioeconomic, and health-related characteristics entered in block groups.