Participants were patients of the Durham VA Medical Center who completed an in-person survey between March 2002 and January 2003. Potential participants were initially identified on the basis of an ICD-9 code indicating OA (715), using VA electronic medical records. We also identified individuals with an upcoming clinic appointment, so that the in-person survey could be conducted in conjunction with a regularly scheduled clinic visit. This was done because many VA patients travel long distances for their clinic appointments. Patients were recruited by telephone. Among 359 veterans who were contacted, 32 indicated they did not have OA, and 29 stated their upcoming VA appointment had been cancelled or rescheduled. Of the remaining 298 eligible patients, 25 refused participation. Another 68 individuals initially agreed to participate but then did not show up for the interview. This is likely due to skipping or rescheduling clinic appointments. An additional three veterans did not provide self-reported race on the survey. Since the current analyses focus on racial differences, they were also excluded. The final sample for these analyses included 202 African-American and Caucasian veterans.


Participants were asked to complete a survey including questions about arthritis symptoms, and perceptions about these treatments.

Specific medication-related variables examined in these analyses included the following: any use of prescription or over-the-counter analgesic/anti-inflammatory drugs, use of specific drug categories (acetaminophen, traditional NSAIDs, COX-2 selective NSAIDs, opioid analgesics), and rating of helpfulness of current drugs. Participants were asked to bring all current prescription and over-the-counter medications they were using for OA to the interview. We asked participants to rate the helpfulness of each analgesic/anti-inflammatory drug in relieving arthritis symptoms on a single-item scale of 1-10 (1 = not at all helpful, 10 = very helpful). This item was developed in the context of recommendations from both physicians and health services researchers. Because patients rated each current medication separately, multiple observations were included for patients who were using more than one medication. Statistical analyses were used to account for multiple observations among these subjects. canadian cialis

The following survey variables were included as covariates in multivariate models since they may also be related to perceptions of medication efficacy: arthritis severity, age, general health (1 = poor to 5 = excellent), education level (no college vs. some college or more), years of arthritis symptoms, number of joints affected by OA, number of arthritis drugs currently being taken (one, two, or three), and the class(es) of the drug(s) being taken. Arthritis severity was measured using the Western Ontario McMaster Universities Osteoarthritis Index (WOMAC), a validated and reliable scale (Cronbach’s alpha >0.80), designed to assess pain and function in OA. The WOMAC score can range from 0-96, with higher scores indicating greater symptom severity. Drugs were grouped into the following classes: acetaminophen, COX-2 inhibitors, NSAIDs, and opioid analgesics.

Statistical Analysis

Three sets of analyses were conducted. First, demographic and clinical characteristics were compared according to race (African-American versus Caucasian), using t-tests for continuous variables and Chi-squared test for dichotomous variables. Second, use of each drug class was compared across racial groups in a bivariate analysis and a multivariable analysis (including other patient demographic and clinical characteristics). Since each patient could have taken drugs from each class, the binary drug use outcome was structured in two levels: the between-patient level (i.e., race and other patient descriptions) and the within-patient level (i.e., drug class). PROC GENMOD (SAS Inc., Cary, NC) was used to analyze the binary drug use outcome for both the bivariate and multivariable analyses. Third, for our primary analyses, we examined the relationships of the following variables with patients’ ratings of medication helpfulness: patient characteristics (race, and the other demographic and clinical variables), the number of drugs taken, and the class(es) of the drug(s) taken. A two-level analysis of variance (ANOVA) was used for this analysis, using SAS PROC GLM. Patient characteristics, a well as the number of arthritis drugs taken, describe the patient and were therefore located at the between-patient level of the data. The remaining factor, class of the drugs taken, was located at the within-patient level, since patients could have rated the helpfulness of medications in more than one drug class. For all analyses, statistical significance was evaluated at the p<0.05 level.
generic cialis 20mg