The database used in the current analysis was created as part of an ongoing longitudinal project measuring health-related characteristics of patients with arthritis. Patients seen during an outpatient visit in the rheumatology clinic at the University of North Carolina Hospitals (UNC) or 13 selected private rheumatology practices in North Carolina were asked to participate. Patients who agreed to participate completed a consent form and a baseline self-report questionnaire on demographic and health-related characteristics; diagnosis and disease onset were provided by the patient's physician. Those individuals with RA, osteoarthritis (OA), and fibromyalgia who completed this process and agreed to further contact were mailed the survey described below (n = 2,075).
The individuals who agreed to further contact were mailed 2 survey booklets. The first booklet asked about health, health beliefs, and use of health care. The second booklet was a detailed review of 8 categories of CAM use. After 3 weeks, nonrespondents were contacted again with a second survey, and then again by phone if neither mail survey elicited a response. A total of 51% of the patients responded (n = 1,063). All procedures and surveys were approved by the UNC Institutional Review Board.
The following demographic data were gathered on all respondents: sex, race (African American or white), age, educational level (dichotomized as less than high school, or high school or more), and marital status. Physician diagnoses of OA, RA, or fibromyalgia were measured as dichotomous variables. Physicians provided multiple diagnoses for patients. However, patients with more than 1 type of arthritis were classified in the following order of priority: 1) RA, 2) fibromyalgia, and 3) OA, and only 1 diagnosis was included in the database. This classification method has been used by prior researchers (11).
The Health Assessment Questionnaire (HAQ) (12) was also administered. The HAQ is based on a 4-point scale rating difficulty in performing 20 activities of daily living (0 = without any difficulty and 3 = unable to do). A summary score was obtained by taking the unweighted mean of the responses. The HAQ has acceptable reliability and validity and is widely used to measure physical limitations due to rheumatic diseases (12–15).
Participatory decision-making style was measured using the 3-item scale developed by Kaplan et al (3). The items were 1) “If there were a choice between treatments, would your doctor ask you to help him/her make the decision? (definitely yes, yes, unsure, no, definitely no)”; 2) “How often does your doctor make an effort to give you some control over your treatment? (very often, often, sometimes, hardly at all, not at all)”; and 3) “How often does your doctor ask you to take some of the responsibility for your treatment? (very often, often, sometimes, hardly at all, not at all).” A summary score ranging from 0 to 100 was calculated by summing the responses, dividing by 15, and then multiplying by 100. Higher scores reflected patients rating their physicians as being participatory. Participatory decision-making style was then recoded into a dichotomous variable (score <70 or ≥70). A cutoff point of 70 has been used in previous studies of physician use of a participatory decision-making style (3, 16).
Medical skepticism can be defined as “doubts about the ability of conventional medical care to appreciably alter health status” (17). Medical skepticism was included in our analyses because one could hypothesize that patients who are more skeptical of conventional medical care might rate their physicians as being less participatory. Medical skepticism was measured using 4 items with a 5-point scale that assessed attitudes toward medical care (17).
Use of CAM for arthritis was queried by asking separately about the following 8 areas: 1) use of alternative health providers or therapists; 2) special diets or food plans; 3) vitamins or minerals; 4) herbs, mixtures, or other supplements taken by mouth; 5) rubs, lotions, liniments, creams, or oils; 6) copper bracelets or magnets or other body treatments used; 7) movement activity; and 8) spiritual, relaxation, or mind-body activities. These data were used to create a continuous variable that reflected the number of categories of CAM the patient used (possible range 0–8), and a dichotomous variable (no use/any use).
Whether patients reported telling their physicians about their CAM use was measured as a dichotomous variable (yes/no). Patients who stated that they did not tell their physician were asked why they did not mention it and were given the following options: “I don't believe he/she knows enough about the strategies to advise me,” “There is no reason to tell him/her,” “I don't think he/she would want to know,” “It's none of his/her business,” and “Did not think he/she would approve.” Patients who stated that they did tell their physician were asked why they did mention it and were given the following options: “I want to know more about the strategy,” “I will only take or do something with his/her approval,” “I prefer to have his/her approval (but his/her approval won't necessarily stop me),” “I am concerned about drug interactions (mixing drug/herbal therapies),” “I am concerned the strategy may jeopardize my health,” and “I want my doctor to be fully informed about my health.” The categories were not mutually exclusive; patients could check several reasons for why they told or did not tell their rheumatologist about their use of CAM.
All analyses were performed using STATA software, version 8.0 (StataCorp, College Station, TX). As discussed above, 1,063 patients responded to the survey and were included in the current analysis. However, only 786 patients were included in the multivariable analyses predicting patients' rating of their physician's participatory decision-making style, because 277 respondents had missing data on 1 or more variables included in the generalized estimating equation (GEE) analyses. A total of 153 patients were excluded from the multivariable analyses because of missing data on 1 or more items of the physician style scale, an additional 79 were excluded due to missing data on race, another 21 were excluded for missing data on educational level, and an additional 24 were excluded due to missing data on medical skepticism, age, marital status, or HAQ scores. A GEE analysis was conducted to examine whether patients' rating of their physician's participatory decision-making style was independently associated with patient age, sex, race, educational level, HAQ score, marital status, OA, RA, fibromyalgia, number of CAM types used, and medical skepticism. GEE was used to adjust the standard errors of the regression coefficients to account for the fact that patients are nested within physicians (18, 19).
Next we examined only those patients who reported ever using CAM for their arthritis (n = 979). GEE was conducted to examine the relationship between the independent variables noted above and whether patients reported ever discussing CAM use with their physician. Whether patients rated their physician as participatory was also included as an independent variable. Only 741 patients were included in the GEE analysis because of missing data on 1 or more of the independent variables.