Racial/ethnic differences in the use of biologic disease-modifying antirheumatic drugs among California Medicaid rheumatoid arthritis patients




To assess racial/ethnic differences in the use of biologic disease-modifying antirheumatic drugs (DMARDs) among California Medicaid (Medi-Cal) rheumatoid arthritis (RA) patients.


Medi-Cal patient level data for 5,385 DMARD recipients between ages 18 and 100 years with at least 1 diagnosis of RA (International Classification of Disease, Ninth Revision, Clinical Modification code 714.xx) and the use of 1 DMARD between January 1, 1998 and December 31, 2005 were collected. The outcome of interest was the choice of either standard DMARDs (methotrexate, lefluonomide, hydroxychloroquine, and sulfasalazine) or biologic DMARDs (adalimumab, etanercept, anakinra, and infliximab). A univariate analysis and logistic regression model were applied to examine the association of the choice of DMARD among different racial/ethnic groups.


In the univariate analysis, biologic DMARD use was significantly associated with race/ethnicity (P < 0.001). In the multivariate logistic regression model, after adjusting for age, sex, insurance coverage, 12 comorbid conditions, RA-related drug prescription, RA-related inpatient stay, and rehabilitation visits, African Americans had 53% lower odds of receiving biologic DMARDs as compared to whites, whereas Hispanics had 36% increased odds of receiving biologic DMARDs as compared to whites.


In this Medi-Cal population, with its racial diversity and relatively homogenous socioeconomic status and health care benefits, racial/ethnic differences were found in RA patients receiving biologic DMARDs.


Racial/ethnic disparities have been shown to exist in the health status of patients with rheumatoid arthritis (RA) and the delivery of health services in RA (1, 2). In 2012, the American College of Rheumatology updated its recommendations for the treatment of RA through a systematic review of the literature and scientific evidence, and an aggressive therapeutic approach to RA was championed (3). As compared to monotherapy with methotrexate, a synthetic disease-modifying antirheumatic drug (DMARD), the use of biologic DMARDs or the combination of biologic DMARDs plus methotrexate leads to better radiographic outcomes, greater functional independence, and higher rates of remission (4). Moreover, the improvement of health-related quality of life years among patients receiving biologic DMARDs has also been reported (5).

Medicaid is the primary health insurance program for low-income and high-need Americans. The Medicaid program in California, known as Medi-Cal, covers 7.7 million people (1 of every 5 Californians) and insures a racially diverse population (53% Hispanic/Latino, 20% white, 10% Asian/Pacific Islander, 10% African American, and 6% other) (6, 7). This unique diversity makes Medi-Cal a rich resource to study racial disparities in health care.

In this study, we explored racial differences in the use of biologic DMARDs among Medi-Cal RA patients. Our research quantifies the variation of RA treatment among different racial/ethnic groups and underscores the importance of stratifying on or adjusting for racial/ethnic effects in epidemiologic or comparative effectiveness research.

Significance & Innovations

  • The use of California Medicaid data allowed this study to avoid a confounding effect from disparities in financial sources as well as differences in benefit designs across insurance carriers, and it provided insight into disease-modifying antirheumatic drug (DMARD) utilization by the minority population.

  • As compared to whites, African Americans were found to be significantly associated with lower odds of biologic DMARD use, while Hispanics were significantly associated with higher odds.

Patients and methods

Data source.

Data were collected from Medi-Cal paid claims and eligibility files for Medi-Cal enrollees from January 1, 1998 through December 31, 2005. Medi-Cal covers outpatient care, inpatient care, and prescription drugs for low-income and disabled California residents. The Medi-Cal paid claims files included information from institutional claims, professional service claims, and pharmacy claims. The service claims included the date of service, type of service, place of service, paid amount, billed amount, units (days) of service, provider identification, and primary and secondary diagnosis codes. The pharmacy claims included the national drug code, fill date, days of supply, paid amount, billed amount, and quantity. The eligibility files included the enrollment status of each month, in addition to the enrollees' demographic information.

Study population.

The study population included RA patients enrolled in the Medi-Cal fee-for-service program. Any enrollee older than 18 years with a diagnosis code for RA and filling a prescription for biologic or synthetic DMARDs was eligible (8). The diagnosis of RA was identified using International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes for RA (714.0–714.8). Synthetic DMARDs included methotrexate, hydroxychloroquine, leflunomide, and sulfasalazine; biologic DMARDs included adalimumab, anakinra, etanercept, and infliximab. Other nonbiologic (e.g., azathioprine) and biologic (e.g., rituximab) DMARDs were prescribed, but their utilization was not sufficient to permit an accurate statistical analysis and was therefore not included in this study.

Patients were excluded if they had ≥1 medical claims with a listed ICD-9-CM diagnosis code of Crohn's disease (555.xx), ulcerative colitis (556.xx), psoriasis (696.1), psoriatic arthritis (696.0), or ankylosing spondylitis (720.0) during their prestudy history. These exclusionary criteria were implemented to avoid including patients who might have received the treatment of interest for another disease.

Study design.

This study was a retrospective cohort study. The patients included in this study were adults with RA who were incident DMARD users based on their prescription medication usage with a 1-year baseline period. For each patient, the index date was defined as the first date that the RA patients filled any synthetic or biologic DMARD prescription. The baseline/washout period was defined as a 1-year time period prior to the index date during which 1-year continuous enrollment with no observed active DMARD treatment was required. The covariates defined below were ascertained during the baseline period for each subject.


Information on the patients' date of birth and race/ethnicity was collected at the time of their application to Medi-Cal. The applicant was first asked to select “Hispanic: yes or no,” which was then followed by a choice of race, including “American Indian,” “black or African American,” “Asian,” “Native Hawaiian or Pacific Islander,” or “white.” Only 1 selection was allowed for race. Those who selected yes to identifying themselves as Hispanic were classified as Hispanics regardless of their race. Subjects with an unknown or missing ethnicity (4.6%) were excluded from the study. American Indians and Native Hawaiian or Pacific Islanders (0.4%) were also excluded because of an insufficient sample size for analysis. The final race/ethnicity categories for this study were mutually exclusive and included Hispanic, African American, Asian, and white.

Outcome and covariates.

The outcome of interest in this study was whether a patient received a biologic DMARD or not. If a patient received any of the 4 biologic DMARDs (adalimumab, anakinra, etanercept, and infliximab) anytime during the study period, he or she was defined as a biologic DMARD user; otherwise, he or she was defined as a synthetic DMARD user. The demographics-related covariates included age at treatment date, sex, self-reported race/ethnicity, residency in an exclusive fee-for-service county, and dual eligibility (Medi-Cal and Medicare). The disease activity–related covariates collected from the baseline period included 12 comorbid conditions based on the ICD-9-CM coding algorithms (9), RA-related prescription by drug class (nonsteroidal antiinflammatory drugs [NSAIDs], corticosteroids, and analgesics) (10), inpatient stay with RA as the primary diagnosis, and rehabilitation visit (11).

Statistical analysis.

The continuous variables listed in Table 1 and the percentage of DMARD use were reported as the mean ± SD and median, as appropriate. The significance tests on continuous variables and categorical variables were the Kruskal-Wallis and chi-square test, respectively. A multiple logistic regression model was specified to measure the association between biologic DMARD use and the sociodemographic characteristics of the study subjects. All analyses were conducted using SAS, version 9.2.

Table 1. Descriptive characteristics of California Medicaid RA patients*
 AsianAfrican AmericanHispanicWhiteTotalP
  • *

    Values are the number (percentage) unless otherwise indicated. RA = rheumatoid arthritis; IQR = interquartile range; DMARDs = disease-modifying antirheumatic drugs; NSAIDs = nonsteroidal antiinflammatory drugs.

Total874 (16.2)1,053 (20.1)1,130 (21.0)2,328 (43.2)5,385 
Age, mean ± SD years62.5 ± 13.7053.9 ± 13.5755.6 ± 15.0757.7 ± 14.9757.3 ± 14.77< 0.0001
Age, median (IQR) years65 (55–72)53 (45–62)57 (45–67)58 (47–69)58 (47–68)< 0.0001
Age, years     < 0.0001
 18–3437 (4.2)59 (5.6)106 (9.4)160 (6.9)362 (6.7) 
 35–4466 (7.6)184 (17.5)167 (14.8)308 (13.2)725 (13.5) 
 45–54114 (13)353 (33.5)237 (21)479 (20.6)1,183 (22) 
 55–64199 (22.8)236 (22.4)252 (22.3)606 (26)1,293 (24) 
 ≥65458 (52.4)221 (21)368 (32.6)775 (33.3)1,822 (33.8) 
Sex     < 0.0001
 Men170 (19.5)299 (28.4)184 (16.3)463 (19.9)1,116 (20.7) 
 Women704 (80.5)754 (71.6)946 (83.7)1,865 (80.1)4,269 (79.3) 
Fee-for-service only county     < 0.0001
 Yes68 (7.8)29 (2.8)215 (19)531 (22.8)843 (15.7) 
 No806 (92.2)1,024 (97.2)915 (81)1,797 (77.2)4,542 (84.3) 
Dual eligibility     < 0.0001
 Yes599 (68.5)535 (50.8)621 (55)1,484 (63.7)3,239 (60.1) 
 No275 (31.5)518 (49.2)509 (45)844 (36.3)2,146 (39.9) 
Type of DMARDs     < 0.0001
 Synthetic744 (85.1)959 (91.1)905 (80.1)1,951 (83.8)4,559 (84.7) 
 Biologic130 (14.9)94 (8.9)225 (19.9)377 (16.2)826 (15.3) 
RA-related prescription by drug class      
 NSAIDs600 (68.6)675 (64.1)685 (60.6)1,426 (61.3)3,386 (62.9)0.0004
 Corticosteroids278 (31.8)242 (23.0)363 (32.1)731 (31.4)1,614 (30)< 0.0001
 Analgesics397 (45.4)686 (65.1)546 (48.3)1,252 (53.8)2,881 (53.5)< 0.0001
Inpatient stay with RA as primary diagnosis17 (1.9)11 (1)23 (2)50 (2.1)101 (1.9)0.1669
Rehabilitation visits8 (0.9)20 (1.9)12 (1.1)32 (1.4)72 (1.3)0.2237


Cohort characteristics.

During the 8-year study period (January 1, 1998 to December 31, 2005), 5,385 qualified subjects between the ages of 18 and 100 years were identified (Table 1). The median age of the study cohort was 58 years. Among all of the racial/ethnic groups, Asians were the oldest (median age 65 years) and African Americans were the youngest (median age 53 years). In terms of sex distribution, the proportion of African American men was at least 8% higher than the proportion of men in the other racial/ethnic groups. Among all the racial/ethnic groups, African Americans had the lowest proportion living in counties with exclusive fee-for-service reimbursement (2.8%) and the lowest percentage having dual eligibility for both Medi-Cal and Medicare (50.8%). There was 6–11% lower biologic DMARD use in the African American group as compared to the other racial/ethnic groups in this cohort. Age; sex; dual eligibility; residing in counties with exclusive fee-for-service reimbursement; and prior therapy with NSAIDs, corticosteroids, and analgesics were significantly different across the 4 racial/ethnic groups. Nevertheless, both the proportion of subjects having at least 1 inpatient stay with RA as the primary diagnosis and the proportion of subjects having at least 1 rehabilitation visit were not different across the groups.

Adjusted odds ratio (OR).

After adjusting for age, sex, insurance coverage, comorbidities, RA-related prescription drug use, RA-related inpatient stay, and rehabilitation visit, African Americans had 53% lower odds of receiving biologic DMARDs as compared to whites, whereas Hispanics had 36% increased odds of receiving biologic DMARDs as compared to whites (Table 2).

Table 2. Adjusted ORs for biologic DMARD use in California Medicaid RA patients*
 OR (95% CI)P
  • *

    The adjusted odds ratios (ORs) were derived from multiple logistic regression modeling biologic disease-modifying antirheumatic drug (DMARD) use with age, sex, race/ethnicity, fee-for-service county, dual eligibility, 12 comorbidities, rheumatoid arthritis (RA)–related prescription by drug class, inpatient stay with RA as primary diagnosis, and rehabilitation visit. All predictors were entered into the model simultaneously. 95% CI = 95% confidence interval; NSAIDs = nonsteroidal antiinflammatory drugs.

  • The input value was the number of fills of each of the specified drugs in the baseline period.

  • Whether a subject had at least 1 inpatient stay with RA as the primary diagnosis in the baseline period.

  • §

    Whether a subject had at least 1 rehabilitation visit, physical therapy or occupational therapy, in the baseline period.

Age category, years  
 18–343.25 (2.30–4.58)< 0.0001
 35–442.44 (1.83–3.25)< 0.0001
 45–541.95 (1.51–2.51)< 0.0001
 55–641.73 (1.39–2.17)< 0.0001
 Women1.11 (0.91–1.37)0.31
 Asian0.98 (0.77–1.23)0.83
 African American0.47 (0.36–0.61)< 0.0001
 Hispanic1.36 (1.11–1.65)0.00
Fee-for-service county0.69 (0.55–0.87)0.00
Dual eligibility1.76 (1.45–2.14)< 0.0001
 Myocardial infarction0.70 (0.30–1.64)0.41
 Congestive heart failure0.96 (0.69–1.35)0.83
 Peripheral vascular disease0.71 (0.47–1.09)0.12
 Cerebrovascular disease0.93 (0.63–1.38)0.72
 Chronic pulmonary disease1.00 (0.81–1.22)0.97
 Diabetes mellitus1.09 (0.88–1.34)0.44
 Peptic ulcer disease0.83 (0.49–1.39)0.47
 Paraplegia and hemiplegia0.41 (0.19–0.91)0.03
 Renal disease0.74 (0.42–1.29)0.29
 Liver disease1.09 (0.76–1.55)0.65
 Hypertension0.93 (0.77–1.12)0.46
 Cardiac arrhythmia1.06 (0.74–1.51)0.77
RA-related prescription by drug class  
 NSAIDs1.26 (1.20–1.33)< 0.0001
 Corticosteroids1.43 (1.34–1.52)< 0.0001
 Analgesics1.12 (1.07–1.17)< 0.0001
Inpatient stay with RA as primary diagnosis2.14 (1.34–3.40)0.001
Rehabilitation visit§1.33 (0.72–2.45)0.36

The odds of receiving biologic DMARDs significantly decreased with increasing age. The probability of patients younger than 45 years being prescribed biologic DMARDs was more than twice the probability of those who were at least 65 years old. The OR of biologic DMARD use in patients between the ages of 45 and 54 years was 1.95 (95% confidence interval [95% CI] 1.51–2.51). There was no association between sex and use of biologic DMARDs. Patients residing in counties with exclusive fee-for-service reimbursement had 31% lower odds of receiving biologic DMARDs; for patients with dual eligibility, the odds increased by 76%. The 3 RA therapy–related drug classes that reflect the active status of RA were significantly associated with the use of biologic DMARDs, with increased odds of 26%, 43%, and 12% for NSAIDs, corticosteroids, and analgesics, respectively. RA-related inpatient stay significantly increased the odds of receiving biologic DMARDs (OR 2.14 [95% CI 1.34–3.40]).


Our study examined 5,385 RA patients over an 8-year study period (from 1998 to 2005). By studying the Medi-Cal population, with its racial diversity yet relatively homogenous socioeconomic status and health care benefits, we were able to avoid the confounding effect from disparities in financial sources and/or differences in benefit designs across insurance carriers in the analysis, and we therefore were better able to determine the racial/ethnic differences in biologic DMARD use.

From the adjusted logistic regression model, a distinctive pattern of biologic DMARD use in different racial/ethnic groups was revealed. The significantly lower likelihood of African Americans receiving biologic DMARDs as compared to whites was consistent with the findings from studies by Constantinescu et al, which were conducted with limited samples in an interview-based approach (2, 12). Constantinescu et al concluded that when considering treatment choices for RA, African American patients were more concerned about medication risk and adverse effects, whereas white patients were more concerned about medication effectiveness and the slowing of radiographic progression. In addition to patient preference, a patient's trust in their physician is another important factor, since this is reported to have a greater effect on patient confidence in a decision to start DMARD therapy than their DMARD-specific knowledge, disease-related factors, or demographic characteristics (13). However, because we were limited by the retrospective observational design of our study, our analysis could not further examine the impact from either patient preference or a patient's trust in their physician.

Hispanics compose the dominant racial group in the Medi-Cal population (53%) (7); however, Hispanics accounted for only 21% of our study cohort. Our results indicate that Hispanics had 36% higher odds of receiving biologic DMARDs compared to whites. Since biologic DMARDs are regarded by many practitioners as a second-line and aggressive treatment, one possible explanation is that the disease severity was higher and/or physicians suspected that the disease progression was more rapid in Hispanics. We are not aware of any studies that delineate the factors that contribute to the treatment decisions for Hispanic RA patients; however, Yazici et al found poor functional scores from the Multidimensional Health Assessment Questionnaire, higher psychological distress, and more morning stiffness to be factors using an early RA treatment evaluation registry (n =118 patients) (14). In the Rheumatoid Arthritis DMARD Intervention and Utilization Study, Hispanics were found to have more active disease on the basis of the Health Assessment Questionnaire and patient global assessments when compared to whites (15). The National Health Interview Survey, which reported a higher prevalence of arthritis in Hispanics as compared to whites, found that the prevalence of activity limitation, work limitation, and severe joint pain was significantly higher among Hispanics than among whites (16). These pieces of evidence point toward a more severe clinical manifestation as well as a debilitating effect in the Hispanic RA population, and therefore more aggressive treatment in this population is required.

Our results also indicate there exists an inverse relationship between receipt of biologic DMARDs with increasing age. One study from the UK showed that the choice of DMARD was related to the mode of administration, and younger patients were significantly more confident about self-administering treatment (17). Unlike synthetic DMARDs, which can be taken orally, adalimumab, etanercept, and anakinra are administered subcutaneously. Our results are consistent with the finding from a study using data from the Consortium of Rheumatology Researchers of North America (CORRONA) database (18). Tutuncu et al reported that late-onset RA patients received combination therapy with DMARDs and biologic treatments less frequently than younger-onset patients, despite both groups having comparable disease severity, activity, and duration.

To address the concern that the paucity of African American subjects residing in counties with exclusive fee-for-service programs may have markedly skewed the results, a sensitivity test was conducted in a county that had a large size of Medi-Cal enrollees and had no exclusive fee-for-service plan. The result from this subgroup analysis was consistent with our original findings.

One critical factor for deciding on therapy with biologic DMARDs is disease severity. Due to the nature of claims-based studies, it is unlikely to collect this information retrospectively. However, a recent report has attempted to proxy the severity of RA by developing an algorithm with the inclusion of rheumatologist visits, rehabilitation visits, medication count, tests for inflammatory markers, number of chemistry panels, and platelet counts ordered (11). In our study, with the available data components from Medi-Cal, we characterized the number of rheumatologist visits, rehabilitation visits, RA-related inpatient stays, and the number of prescription fills for NSAID, corticosteroid, and analgesic medications as covariates in the regression model to control the effect of RA severity.

Experts have not yet reached a consensus regarding the comparative effectiveness of different drug treatment approaches for RA. Our study was not designed as a cost-effectiveness study, but our findings present strong evidence of racial/ethnic differences in biologic DMARD use. These differences are independent of insurance coverage, geographic variation, socioeconomic status, and even disease severity. The underlying mechanism may be partly explained by patient preference and/or patient trust of their physician's recommendations.

As the Baby Boomer generation ages, diverse populations continue to grow, and Medicaid programs expand during health care reform, understanding the racial/ethnic differences in RA treatment could be meaningful when customizing intervention programs for different racial groups. Our findings provide a strong case for health care providers and researchers to focus their attention on engaging in identifying the causes for the wide variation in RA treatments.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kawatkar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Chu, Kawatkar.

Acquisition of data. Kawatkar, Nichol.

Analysis and interpretation of data. Chu, Portugal, Kawatkar, Stohl, Nichol.