Arthritis & Rheumatism, Volume 63,
November 2011 Abstract Supplement

Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Chicago, Illinois November 4-9, 2011.

DMARD Underuse in the Medicare Current Beneficiary Survey: Evidence for Socioeconomic Disparities.

Solomon1,  Daniel H., Ayanian2,  John Z., Lu3,  Bing, Brookhart4,  M. Alan, Schneeweiss5,  Sebastian, Shaykevich1,  Tamara, Katz1,  Jeffrey N.

Brigham & Womens Hospital, Boston, MA
Brigham and Women's Hospital
Brigham and Women's Hospital, Boston, MA
University of North Carolina
Harvard University, Boston, MA


There has been an increased appreciation for the importance of DMARDs for virtually all patients with RA. However, numerous studies have documented that many patients with RA do not appear to receive DMARDs. Our prior work has documented that there racial disparities in who receives DMARDs, but these studies have not examined the role of insurance, income, and education as possible correlates of racial disparities. The goal of these analyses was to determine if racial disparities in DMARD prescribing can be explained by insurance, income, and/or educational differences.


We examined DMARD use in the Medicare Current Beneficiary Survey (MCBS), a four-year longitudinal US national survey of randomly selected Medicare beneficiaries. This study included data from years 2001–2006. Participants in MCBS with at least one Medicare claim for RA plus a self-report of RA were included in the analyses. DMARD use was based on an in-home assessment of all medications. Variables considered as potential correlates of DMARD use in regression models included race/ethnicity, insurance, income, education, rheumatology visit, region, age, gender, comorbidity index, and calendar year. Because of the repeated measures of DMARD use, each year of follow-up was considered as a separate observation with adjustment for within subject correlation using generalized linear model applying a Generalized Estimating Equation approach.


The cohort consisted of 513 MCBS participants with a mean age of 70 years, of which 72% were female. During follow-up, 42% received a DMARD. In fully adjusted analyses among the whole cohort, the strongest predictor of DMARD use was seeing a rheumatologist (see Table). In the total cohort, lower annual income was associated with a reduced probability of DMARD use. This was not observed in the cohort seen by rheumatologists. Among individual not seen by rheumatologists, Black non-Hispanics were much less likely to receive a DMARD (OR 0.16, 95% CI 0.03–0.98) compared to white, non-Hispanics; this was not the case when these individuals were seen by rheumatologists (OR 0.89, 95% CI 0.12–6.62). Older age was associated with a reduced probability of DMARD use, especially in the cohort not seen by rheumatologists.

Table .: Regression models predicting DMARD use in MCBS

VariablesTotal CohortAmong Subjects Without Rheumatology VisitsAmong Subjects With Rheumatology Visits
 Odds Ratio (95% confidence interval)
Age, years 
   < 751.001.001.00
   75–840.58 (0.37–0.92)0.36 (0.18–0.73)0.83 (0.46–1.51)
   85+0.09 (0.02–0.31)0.05 (0.01–0.26)0.15 (0.02–0.89)
Gender, female1.04 (0.65–1.67)0.91 (0.49–1.71)1.34 (0.71–2.53)
   White, non-Hispanic1.01.01.0
   Black, non-Hispanic0.49 (0.14–1.76)0.16 (0.03–0.98)0.89 (0.12–6.62)
   Hispanic0.99 (0.30–3.22)0.40 (0.08–2.01)1.28 (0.22–7.60)
   Other, non-Hispanic1.23 (0.41–3.67)0.68 (0.17–2.77)1.56 (0.29–8.22)
   30–50K0.89 (0.33–2.40)0.44 (0.14–1.32)1.15 (0.40–3.34)
   20–30K0.59 (0.20–1.63)0.27 (0.08–0.84)0.74 (0.25–2.20)
   15–20K0.58 (0.20–1.71)0.28 (0.08–0.95)0.75 (0.20–2.61)
   < 15K0.57 (0.21–1.53)0.17 (0.05–0.55)1.08 (0.36–3.29)
Rheumatology care7.74 (5.37–11.1)NANA
Notes: Bolded findings are statistically significant.NA, not applicable; Models adjusted for variables in the table plus education, region of residence (South, Midwest, Northeast, and West), insurance status (private, HMO, and drug insurance), comorbidity index, as well as year of observation.


We found that less than half of participants diagnosed with RA in MCBS used DMARDs. Among individuals not seeing rheumatologists, Black non-Hispanic ethnicity appears associated with a reduced probability of DMARD use, independent of income, insurance, and education. As has been the results of several other cohort studies, seeing a rheumatologist is the strongest predictor of DMARD use.

To cite this abstract, please use the following information:
Solomon, Daniel H., Ayanian, John Z., Lu, Bing, Brookhart, M. Alan, Schneeweiss, Sebastian, Shaykevich, Tamara, et al; DMARD Underuse in the Medicare Current Beneficiary Survey: Evidence for Socioeconomic Disparities. [abstract]. Arthritis Rheum 2011;63 Suppl 10 :123

Abstract Supplement

Meeting Menu