Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

VA Health Systems Research

Go to the VA ORD website
Go to the QUERI website

IIR 11-113 – HSR Study

IIR 11-113
Discontinuing NSAIDs in Veterans with Knee Osteoarthritis
Liana Fraenkel, MD MPH FRCPC
VA Connecticut Healthcare System West Haven Campus, West Haven, CT
West Haven, CT
Funding Period: September 2013 - August 2018
Background/Rationale: 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. While short-term studies have demonstrated that NSAIDs are more effective than placebo and acetaminophen, there are no long-term data supporting their use and the decrease in pain conferred by NSAIDs does not meet patient-defined thresholds indicating clinically significant improvement. Moreover, long-term use of NSAIDs is associated with significant morbidity and mortality. This is especially true for Veterans, a population at particularly high risk for NSAID-induced toxicity.

The objectives of this proposal were to determine whether 1) placebo was as effective as continued NSAIDs, and 2) cognitive behavioral therapy (CBT) was a viable alternative to continued NSAID use.

Participants participated in a 2-week run-in period where all subjects replaced their current NSAID with the study drug (meloxicam 15mg per day). Those who remained eligible at the end of the run-in period participated in a 4-week, double-blind, placebo-controlled, non-inferiority randomized withdrawal trial (RWT). After 4 weeks, subjects in the meloxicam arm continued the study drug. Subjects in the placebo arm stopped the placebo and participated in a 12-week CBT program
Participants were drawn from Veterans with knee OA enrolled in the Connecticut, Boston, Providence, and Gainesville VA Medical Centers. Veterans with knee pain despite NSAID use and/or Veterans with one or more risk factors for NSAID-induced nephrotoxicity, gastrointestinal toxicity, and /or cardiovascular toxicity were eligible to participate. Veterans who had contraindications to NSAIDs, were on opioids and/or a Cox-2 inhibitor other than meloxicam, were using an NSAID (not including ASA) for a painful condition in addition to knee OA, having received hyaluronic acid knee injections within 6 months or corticosteroid knee injections within 3 months or were scheduled to have these injections, arthroscopy or knee surgery were excluded. We also exclude those with co-morbid conditions that could limit outcome assessment.
The primary endpoint was between-group differences in the WOMAC pain score measured at 4 weeks. Secondary outcomes were area under the curve of the WOMAC pain scale score, lower extremity disability, global impression of change, and use of co-therapies.
The range of the WOMAC pain subscale is 0-20 and the minimum clinically important difference (MCID) is 2.1. We set the equivalence margin to 1.0 which is less than 50% of the MCID. A sample size of 434 (217 per arm) achieves 90% power to detect non-inferiority using a 1-sided, 2-sample t-test. A 1-sided test is recommended for non-inferiority trials. We used a linear regression model to test the non-inferiority of placebo to meloxicam in terms of WOMAC knee pain intensity at 4 weeks (scaled from 0 to 20, with higher values indicating more pain). The margin of non-inferiority was prespecified as 1. The regression model included treatment group (placebo vs. meloxicam), baseline WOMAC knee pain and site. The analysis was intent-to-treat.

490 participants underwent the run-in. Out of these, 364 were randomized (180 placebo, 184 meloxicam), and N=351 (173 placebo, 178 meloxicam) subjects had the outcome available for the primary analysis. At baseline, the mean pain score was 5.39 (SD=3.76) in the placebo group and 5.86 (SD=3.87) in the meloxicam group. At week 4 post-baseline, the mean pain increased to 7.85 (SD=3.84) in the placebo group and 6.67 (SD=3.56) in the meloxicam group. After adjusting for baseline pain and site, the mean difference between placebo and meloxicam in pain at 4 weeks was estimated as 1.46 (95% CI 0.79, 2.13). The upper limit of the CI (2.13) exceeded the non-inferiority margin of 1, so we cannot conclude that placebo is non-inferior to meloxicam. The p-value for the test of non-inferiority was 0.91. Since all the values in the CI are positive, we can conclude that the mean pain was higher with placebo than with meloxicam (i.e. meloxicam better than placebo, with a benefit that could plausibly be as low as 0.79 but no more than 2.13).
Secondary analyses were conducted using linear regression models adjusted for the stratification variables (baseline WOMAC knee pain and site). The number of subjects who completed the end of treatment (end of phase 2) survey was N=336 (160 placebo, 176 meloxicam). Out of these, 72% completed the survey at Week 14 from baseline, and 27% during Weeks 15 to 17. The difference in pain scores between placebo+CBT and meloxicam was estimated as 0.47 (95% CI -0.24 to 1.19), with upper limit of CI greater than 1, so we cannot conclude non-inferiority of placebo+CBT to meloxicam at the end of treatment (p-value non-inferiority=0.07). In terms of area under the curve (AUC) across the first 4 weeks (phase 1), meloxicam was superior to placebo by 1.14 (95% CI 0.57 to 1.71), p=0.0001. In terms of AUC across all weeks in the trial, meloxicam was superior to placebo+CBT by 1.05 (95% CI 0.53 to 1.57), p<0.0001. There was no evidence of a difference between placebo+CBT and meloxicam in either Global Impression of Change (p=0.15) or WOMAC Lower Extremity Disability (p=0.45). The CBT arm accumulated $87,417 in costs with 7.873 QALYs gained; the meloxicam arm accumulated $144,733 in costs with 7.520 QALYs gained. Thus, the meloxicam strategy was dominated compared to CBT (it cost more and produce fewer QALYs).

According to our preliminary analyses, Veterans discontinuing meloxicam experience slightly more pain compared to those who continue meloxicam. There is however no difference in function or impression of change in veterans who continue meloxicam versus those who stop meloxicam and engage in a CBT program. CBT costs less and results in a greater number of quality adjusted life years than meloxicam. The latter is driven by greater costs accrued for mental health services in the meloxicam group.

External Links for this Project

NIH Reporter

Grant Number: I01HX000646-01A2

Dimensions for VA

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

Learn more about Dimensions for VA.

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.
    Search Dimensions for this project


Journal Articles

  1. Goulet JL, Buta E, Brennan M, Heapy A, Fraenkel L. Discontinuing a non-steroidal anti-inflammatory drug (NSAID) in patients with knee osteoarthritis: Design and protocol of a placebo-controlled, noninferiority, randomized withdrawal trial. Contemporary clinical trials. 2018 Feb 1; 65:1-7. [view]

DRA: Aging, Older Veterans' Health and Care
DRE: Treatment - Efficacy/Effectiveness Clinical Trial
Keywords: none
MeSH Terms: none

Questions about the HSR website? Email the Web Team

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.