Table 2.

Multinomial logistic regression model examining factors associated with biologic DMARD use in Medicare beneficiaries with rheumatoid arthritis using DMARDs.

	RRR (95% CI)
	Biologic DMARD use (vs. non-biologic DMARD use)
	Part B Biologic DMARD	Part D Biologic DMARD
Age
65–72	1	1
73–78	0.99 (0.85, 1.16)	0.73 (0.60, 0.88)
79+	0.73 (0.62, 0.86)	0.49 (0.40, 0.60)
Gender
Male	0.77 (0.64, 0.92)	0.85 (0.69, 1.06)
Female	1	1
Race/ethnicity
Black	0.56 (0.41, 0.77)	0.66 (0.49, 0.88)
Other	0.81 (0.58, 1.14)	1.07 (0.81, 1.43)
White	1	1
Charlson comorbidity score
1	0.90 (0.72, 1.12)	0.78 (0.61, 0.99)
2	0.80 (0.62, 1.03)	0.80 (0.61, 1.06)
3+	1	1
Subsidy Status
Non-LIS	1	1
LIS	0.58 (0.48, 0.69)	2.98 (2.50, 3.56)
Area-level SES
Quintile 1 (lowest SES)	0.91 (0.73, 1.13)	0.77 (0.61, 0.98)
Quintile 2	0.85 (0.69, 1.05)	0.56 (0.43, 0.73)
Quintile 3	0.80 (0.65, 0.98)	0.57 (0.44, 0.74)
Quintile 4	0.96 (0.79, 1.18)	0.61 (0.47, 0.79)
Quintile 5 (highest SES)	1	1
Number of hospitalizations
0	0.78 (0.60, 1.02)	0.74 (0.55, 0.99)
1	0.94 (0.70, 1.26)	0.80 (0.56, 1.12)
2+	1	1

DMARD=disease-modifying anti-rheumatic drug. LIS=low-income subsidy. SES=socioeconomic status. RRR=relative risk ratio.

The outcome in this multinomial logisitic regression model is biologic DMARD use, categorized by whether the biologic DMARD was obtained in Part B versus Part D. The RRRs therefore represent a comparison between biologic DMARD users in each program to non-biologic DMARD users with rheumatoid arthritis.