Participants
Participants were recruited from Temple University Hospital's Department of Orthopedics inner-city outpatient office between December 2009 and March 2011. The participants were all under the care of one of 3 board-certified orthopedic physicians for knee and/or hip OA. The OA diagnosis was based on x-ray (Kellgren-Lawrence score ≥ 1) [9] as well as knee or hip clinical symptoms (e.g., pain, stiffness, swelling, impaired function). Exclusion criteria were any other rheumatic disease (e.g., rheumatoid arthritis, lupus, gout) or diagnosis of complex regional pain syndrome. A total of 184 patients were queried for their participation in the study of which 22 patients were excluded and/or declined to participate. Characteristics of those not willing to participate are unknown. Institutional Review Board approval by Temple University's Institutional Review Board as well as written informed consent were obtained prior to data collection.
Research design
The study design was cross-sectional. Four interview-based questionnaires were used to address the specific aims and research questions of the study: Demographic and Health History Questionnaire, Health History Follow-up Questionnaire, Pharmacological and Supplement Use Assessment Questionnaire, and Pharmacological and Supplement Use Follow-up Questionnaire (see Additional file 1). A priority of 8 specific aims with several specific research questions were identified and approached as an exploratory analysis in this study. Professionally-guided care was defined as a prescription medication, OTC medication, or dietary supplement recommended by a physician. Self-guided care was defined as any OTC medication, prescription drug not used for its original purpose, or dietary supplement taken without the recommendation from a physician.
Instrumentation
Four interview-based questionnaires were used to obtain information about the participant's health history and pharmacological and supplemental use: two initial questionnaires and two follow-up questionnaires. The questionnaires were developed by the research team and reviewed by two statisticians: one specializing in psychoeducational assessment research and the other in community-based epidemiology. All of the questionnaires were interview based. The questionnaires were an item-by-item design, where each question could be scored individually. The interviewer practiced the procedures and data coding on 18 non-study patients with knee OA prior to starting the study.
Demographic and health history questionnaire
The Demographic and Health History Questionnaire was used to obtain participant general health as well as knee and/or hip joint specific information. It consisted of participant identity information and questions concerning his/her demographics, education level, physical activity level, knee and/or hip injury and treatments (e.g., medication use), and general medical history. In addition, each participant was asked to list all medications he/she was currently taking. The Demographic and Health History Questionnaire consisted of "yes" or "no", Likert-type 5-point scale, as well as open- and close-ended questions.
Health history follow-up questionnaire
The Health History Follow-up Questionnaire was used to discern additional information in follow-up to the Demographic and Health History Questionnaire. Follow-up questions were prompted if a participant responded "yes" to any questions regarding previous injury, surgery, other arthritic joint, other treatments used for OA, and sport participation history.
Pharmacological and supplement use assessment questionnaire
The Pharmacological and Supplement Use Assessment Questionnaire was used to identify the specific medications commonly used by the participant as part of his/her knee and/or hip OA joint symptom management. Drug categories were: NSAIDs, herbal remedies, nutritional supplements, injections, acetaminophen, pain relieving medications, and any other drugs for joints. In addition, each participant provided information about when he/she had used each drug: in the last 24 hours, greater than 24 hours but less than 2 weeks ago, 2 to less than 4 weeks ago, 4 weeks to less than 3 months ago, or more than 3 months ago.
Pharmacological and supplement use follow-up questionnaire
The Pharmacological and Supplement Use Follow-up Questionnaire was used to discern additional details concerning the participant's specific medication use. Follow-up questions addressed the drug dose; OTC versus prescription drug; if the drug was taken as directed and consistently; if more than one drug was being taken; and if the drug was no longer being taken, and, if not, why it was stopped.
Procedures
Participants were recruited from Temple University Hospital's Department of Orthopedics and informed of the study's purpose. Those who agreed to participate read and signed the study compliance forms. Next, the participant was interviewed by one of the coauthors (SB) using the Demographic and Health History Questionnaire as well as Pharmacological and Supplement Use Assessment Questionnaire in sequential order in a private examination room. The interview had no time limit, but the average interview took between 15 and 20 minutes to complete. Based on the responses to each of the initial questionnaires, participants were then interviewed by the same co-author using the appropriate follow-up questionnaire(s).
Statistical analysis
Data analyses included descriptive statistics using the SAS 9.2 (SAS Institute, Cary, NC) statistical package. Items were analyzed with mean, standard deviation, frequency and/or percentage statistics. Sampling error was ± 7.84%.
Logistic regressions were performed to predict 1) same day dual-medication use, 2) participants that only used self-guided recommendations compared to those that only used professionally-guided recommendations, 3) use of medication as directed, and 4) reporting a medication as ineffective or requiring a stronger medication. Potential predictors were ethnicity, sex, obesity, age, reporting multiple OA joints, and education. Models were calculated with SPSS 20 (IBM Corporation, Armonk, NY) and considered significant if the overall p-value was ≤ 0.05. The sensitivity and specificity of significant models was calculated with Diagnostic Utility Statistics [10].
Two Fisher's Exact Tests were performed in SAS 9.2 to further assess potential confounders for patients reporting their medication as ineffective or requiring a stronger medication. One test assessed the association between participants reporting their medication as ineffective and not always using medication as directed. The other test evaluated the association between participants reporting their medication as ineffective and not always using medication consistently. Fishers exact tests were considered significant if the p-value was ≤ 0.05.