Knee osteoarthritis (OA) is now recognized as a major health problem. It is the number one cause of lower extremity disability and has significant deleterious effects on quality of life. While there are numerous therapies available for knee OA, most have limited efficacy. Of particular concern, is the widespread use of nonsteroidal anti-inflammatory drugs (NSAIDs) for this disorder. Veterans, as a group, are at high risk for both gastrointestinal and cardiovascular NSAID-induced complications. In this study the investigators propose to examine whether replacing NSAIDs with cognitive behavioral therapy delivered by telephone is an effective strategy for Veterans with knee OA. Telephone-administered therapy is particularly appealing since Veterans with knee OA are more likely to have limited mobility. If successful, this program may result in significant cost-savings for both Veterans (decreased co-pays and transportation costs) and the VA (decreased hospitalizations due to NSAID induced toxicity).

Knee osteoarthritis (OA) is a major cause of disability among Veterans and is the primary indication of knee replacement in the VA. Nonsteroidal anti-inflammatory drugs (NSAIDs) are the most commonly prescribed medications for knee OA. Though short-term studies have demonstrated that NSAIDs are more effective than placebo and acetaminophen, there are no long-term data supporting their use. This is of concern, because long-term use of NSAIDs is associated with significant morbidity and mortality. Guidelines have been published to improve safe use of NSAIDs; however, adherence to these evidence-based recommendations is low in the VA. There is therefore an important need to determine if the long-term use of NSAIDs offers any incremental benefit over safer alternatives. This is especially true for Veterans, a population at high risk for NSAID-induced toxicity.

Cognitive behavioral therapy (CBT) is an effective and safe treatment alternative for OA. This modality is becoming increasingly available in the VA for treatment of chronic pain as well as other chronic disorders such as depression, post-traumatic stress disorder and insomnia. CBT can be successfully administered over the telephone and thus stands to benefit Veterans living in more remote areas with limited access to hospital or community-based outpatient clinics.

In this study, the investigators propose to conduct a 2-phase randomized withdrawal trial (RWT). The trial will focus on recruiting Veterans with knee OA who have been using NSAIDs for at least 3 months.

In the first phase of the study, 544 Veterans with knee OA will be randomized to continue NSAIDs or to placebo for 4 weeks. This double-blind phase will enable us to infer whether placebo is non-inferior to continued NSAID use. In the second phase, subjects in the NSAIDs group will continue NSAIDs and those on placebo will stop taking the placebo and participate in a 10-week CBT program. The second, single-blind, phase will allow us to infer whether CBT is non-inferior to NSAIDs. All study data will be collected over the telephone thus enabling Veterans who have difficulty arranging transportation to the VA to participate.

The investigators will test for between-group differences in knee pain measured using the well-validated Western Ontario and McMaster Universities Osteoarthritis Index (primary outcome) at 4 and 14 weeks. The investigators will also test for between group differences in lower extremity disability, subjects' global impression of change and use of co-therapies (secondary outcomes). As recommended for non-inferiority trials, the investigators will perform both an intent to treat and per protocol analysis. Lastly, the investigators will estimate the potential cost-effectiveness of the CBT protocol compared with continued NSAID use.

Though it would be ideal for subjects randomized to the active study drug to continue their current NSAID, having the VA pharmacy formulate multiple different active drugs and maintaining the blind is not possible. Therefore, the investigators will include a 2-week run-in period where study subjects will replace their NSAID with meloxicam. Meloxicam was chosen as the study drug because it is the most commonly prescribed at the investigators' center and has a favorable safety profile compared to other NSAIDS.

If successful, the trial will improve the quality of care delivered to Veterans with chronic knee pain due to OA. The proposed strategy is particularly appealing because it replaces the widespread use of NSAIDs with a safer alternative, enables delivery of care to Veterans with limited access, and is likely to be cost saving.

Inclusion Criteria:

Subjects will include those for whom a discontinuation trial of NSAIDs is most appropriate: 1) Veterans with knee pain despite NSAID use and/or 2) Veterans at relatively higher risk of NSAID toxicity 55-59 as ascertained by meeting 1 or more of the following 4 criteria:

• Answer affirmatively to the question: "Do you have some knee pain on most days over the past 3 months?"
• Have 1 or more risk factors for NSAID-induced nephrotoxicity (age greater than 60 years, atherosclerotic cardiovascular disease, current diuretic use, chronic renal insufficiency, congestive heart failure (New York Heart Association class I-II. Note, Class III and IV are excluded).
• Have 1 or more risk factors for NSAID-induced gastrointestinal toxicity (history of peptic ulcer disease, age > 65 years, concurrent use of daily ASA or corticosteroids), and are currently on a gastro-protective agent.
• Have 1 or more risk factors for NSAID-induced cardiovascular toxicity (prevalent cardiovascular disease, hypertension, hypercholesterolemia, diabetes, smoking, family history of early heart disease or age greater than 55 years for women).

In addition, subjects must:

• Be age 20 years or older. While the usual cut off for knee OA is approximately 40 years, the investigators chose to lower the age cutoff as younger Veterans have a higher than expected risk of OA (see B.1).
• Have radiographic evidence of knee OA reported in the VistA electronic system.
• Be using an NSAID (other than daily ASA) for knee pain on most days of the month for at least the past 3 months.
• Be able to understand and speak English and have a telephone.
• Be willing to engage in a CBT program, to discontinue (or replace) their NSAID, and to restrict co-therapies to acetaminophen for 14 weeks.

Exclusion Criteria:

• Subjects desiring escalation of analgesics for their current level of knee pain as determined by endorsement of the following statement: "Is your knee pain bad enough that you want to talk to your doctor about taking stronger pain medications?"
• Current use of opioids and/or Celebrex.
• Current use of an NSAID (not including ASA) for a painful condition in addition to knee OA.
• Contraindications to chronic NSAID use: current use of warfarin or antiplatelet agent other than ASA, allergy to any NSAID, active upper gastrointestinal ulceration in the previous 30 days, upper gastrointestinal bleeding in the past year, history of gastroduodenal perforation or obstruction, cardiovascular event within the past 6 months (myocardial infarction, cerebrovascular event, coronary-artery bypass graft, invasive coronary revascularisation, or new-onset angina), severe congestive heart failure (New York Heart Association class III-IV), evidence of serious anemia, hepatic, renal (including nephrotic syndrome), or blood coagulation disorders, and pregnancy.

***Though the investigators are proposing a RWT - and thus will not be initiating NSAID therapy - it would not be appropriate to continue NSAIDs (even when prescribed) in high-risk patients. The investigators acknowledge that these exclusion criteria limit generalizability, but the investigators feel justified to ensure subjects' safety.***

• Previous hyaluronic acid knee injections (within 6 months) or corticosteroid knee injections (within 3 months).
• Scheduled knee hyaluronic acid or corticosteroid injections, arthroscopy, or knee surgery.
• Co-morbid conditions that include the following: known other causes of arthritis (infectious arthritis, rheumatoid arthritis, connective tissue disease, or psoriatic arthritis), gout or pseudogout attack within the last 12 months, peripheral neuropathy or cardiopulmonary disease that limits walking more than knee pain, bone metastases or Paget's disease involving the lower extremities, and history of drug or alcohol abuse within the past 2 years, bilateral knee replacements or knee pain in the replaced knee only.
• Current involvement in litigation or receiving workmen's compensation.
• Hearing, cognitive impairment or mental illness, as determined by chart review that would preclude participation in a CBT program.
• For Women of Childbearing Age: Must not currently be pregnant, agree to avoid getting pregnant during the course of the study and should inform the study team if pregnancy occurs at any time during study participation.
• Previous meloxicam use discontinued due to lack of effective symptom relief
• Contraindications to prolonged NSAID use, per PI discretion.

490 participants were enrolled and underwent the run-in; 364 (74%) participants remained eligible at the end of the run-in period and were randomized: 180 to placebo followed by CBT and 184 to meloxicam.

Adverse event data were collected through study completion, an average of 14 weeks. After 2-week run-in period where all subjects replaced their current NSAID with the study drug (meloxicam 15mg per day), remaining eligible subjects participated in 4-week, double-blind, placebo-controlled, non-inferiority randomized withdrawal trial (RWT). After 4 weeks, subjects in the meloxicam arm continued study drug. Subjects in the placebo arm stopped the placebo and participated in a 10-week CBT program.

All related adverse events were identified by study coordinator or study therapist during the time when subject participates in the study. Any serious adverse events and unanticipated problems involving risks to subjects and others were reported immediately to the study PI and within 24 hrs to the VA Central Institutional Review Board (IRB). The reporting processes, as listed in the VA Central IRB Table of Reporting Requirements, was followed.