Purpose: To evaluate the relationship of numeracy with satisfaction with risk communication about Disease Modifying Antirheumatic Drugs (DMARD). Methods:   The RA DMARD Knowledge Profile (DKP) is a 19 item index testing patient knowledge necessary to participate with providers in the decision to select a new DMARD.  Four items relate to interpreting numeric information. A mail survey of 900 RA patients assessed patient demographics, DKP numeracy and content knowledge, Combined Outcome Measure for Risk Communication (COMRADE) sub-score for satisfaction with risk communication (sat-risk-comm), Health Assessment Questionnaire II (HAQ II).  Statistical analysis: t test for between group differences; correlation and multiple linear regression for associations between independent and dependent variables.  Results:  Response rate = 71%.  Sample: mean age 57.7 years, female 72.9%, early RA 22.1%, mean HAQ II .901 (sd .578), anti-TNF therapy 47.9%, white 88.7%, minority 6.8%, depression 21%, education < high school graduation 14.6%,  Medicaid 11.1%.  Mean sat-risk-comm with recent DMARD was 42.00 (sd 9.88) (maximum possible score = 50).  Mean numeracy score was 1.84 (sd=1.20). Percent correct response for the 4 numeracy items by education: .

% correct responses	Overall	<9	9-12	HS Grad	College Grad	Grad School
Interpret frequency	57.0	38.5	59.2	64.1	57.1	64.3
Interpret probability	50.0	38.5	38.2	46.4	62.3	81.0
Multiply probability	47.3	23.1	38.2	45.8	62.3	78.6
Convert probability to percentage	29.5	15.4	26.3	31.4	32.5	31.0

       

DKP numeracy correlated moderately with DKP content knowledge (R=.463, P<.001) but weakly with sat-risk-com related to DMARD selection (R=.134, P<.002) and education (R=.191, P<.001).  When patients with below and above mean scores on the sat-risk-com were compared, there was significant difference in DKP numeracy scores 1.85 and 2.08 respectively (P<.015). A multiple linear regression model predicting sat-risk-com using purposeful stepwise selection of covariates is  sat-risk-com = 39.27 – 5.87 low income– 2.987 HAQii -3.76 +1.06 RA duration – 3.524 minority.  Standardized regression coefficients:  -.172, -.191, .099, and -.092. Overall model R = .374, R² =.140 (p< .001).  Controlling for major covariates and content knowledge, numeracy score did not add significantly to the predictive model.  Conclusion:  Numeracy was poor across education. However satisfaction with anti-rheumatic risk communication was overall high in this population.  Satisfaction with risk communication about anti-rheumatic drugs was not well explained by patient risk related numeracy, income, disease experience, or minority status.