October 2002 Bulletin

Use of COX-2 inhibitors in patients with fractures

Is there a trade off betweeen pain relief and healing?

By Thomas A. Einhorn, MD

In June 2002, two scientific articles triggered widespread controversy over the use of cyclooxygenase-2 (COX-2) inhibitors during fracture repair.1,2 Although both studies reported impaired bone healing in the absence of COX-2 activity in animals, physicians were left wondering about the clinical indications for these drugs in their patients.

Non-steroidal anti-inflammatory drugs (NSAIDs) and COX-2 inhibitors are among the most common drugs prescribed by orthopaedic surgeons. Their action to inhibit cyclooxygenase results in a reduction in prostaglandin synthesis and hence diminished inflammation. NSAIDs, agents that have been available for over 20 years, inhibit cyclooxygenase-1 as well as -2 and, as such, interfere with certain homeostatic functions such as the role of prostaglandins in maintaining normal gastric mucosa and the role of thromboxanes in regulating clotting mechanisms. COX-2 inhibitors specifically target cyclooxygenase-2, the isoform responsible for prostaglandin-induced inflammation. The development of COX-2 specific inhibitors is considered a major advance because levels of anti-inflammatory activity can be achieved without producing significant dyspepsia or increasing bleeding time. This favorable safety profile has led to their use at higher doses that render them effective as post-operative and post-fracture analgesics. The availability of a non-addictive oral medication with an analgesic profile comparable to that of a narcotic has obvious advantages.

Over the past two decades, several studies using animal models of fracture healing have been reported showing inhibitory effects on skeletal repair of various NSAIDs. Although only two papers have been published in the peer-reviewed literature regarding the effects of COX-2 specific inhibitors on fracture healing, both of these showed significant inhibitory effects in animals. Moreover, while not yet published, nearly every study that has asked this or a similar question—and that has been presented at a national or international meeting—has reported similar findings.

Generally speaking, when a body of data from animal studies is so overwhelmingly consistent from investigation to investigation, it is expected that similar effects would be observed in human clinical trials. However, before translating these animal data to our management of patients, several points must be considered. First, most of the studies testing these drugs in animal models of bone repair have evaluated healing at fairly early time points. In fact, recent data from our laboratory has shown that, if high doses of a specific COX-2 inhibitor are administered after fracture and healing is assessed at three weeks, significant inhibitory effects are observed. However, if drug administration is discontinued any time up to three weeks after fracture, and healing is then assessed at five weeks (a time when normal healing would be expected to occur in a rat), control and COX-2 inhibitor-treated animals show no differences in healing. This suggests that the inhibitory effects are quickly reversed by resumption of normal prostaglandin production. If this scenario were to be translated to a clinical setting, one could imagine that, if a patient were to take an NSAID or COX-2 inhibitor for one to two weeks (the typical amount of time one would expect to require pharmacological analgesia after a fracture or skeletal surgery) and then stop using the drug after that time, healing could occur quite normally.

Unfortunately, there are few data available from well-controlled studies regarding the use of NSAIDs or COX-2 inhibitors in patients undergoing a bone repair process. One NSAID, ketorolac, was studied in a retrospective analysis of 288 cases of spinal fusion performed at a single center.3 This drug was administered as a 15 mg intramuscular loading dose followed by 30 mgs, every six hours as needed. One hundred and sixty-seven patients received ketorolac and no NSAID was given to 121. Nonunion occurred in 5 of 121 (4%) nontreated controls and 29 of 167 (17%) patients receiving ketorolac (odds ratio 4:9). There was a dose-dependent relationship between nonunion rate and ketorolac dose. By contrast, Reuben examined nonunion rates in 106 patients undergoing posterior spinal fusion and receiving analgesic doses of rofecoxib (Vioxx).4 No effect on nonunion rate was noted.

Although several well-controlled clinical studies have shown that NSAIDs inhibit heterotopic ossification after total hip arthroplasty or pelvic fracture, similar clinical data do not exist for their effects on bone healing. Interestingly, in those instances where NSAIDs have been used during healing of a pelvic fracture, heterotopic bone was inhibited but fracture healing occurred. Although it is unknown if the healing of these fractures was in any way delayed, clinical observations did not suggest a problem. Moreover, when one considers all of the thousands of patients during the past two decades who may have used NSAIDs during a period of bone healing, the lack of a clinically obvious effect may suggest the lack of a clinically important concern.

So where does this leave us? In the absence of strong clinical evidence supporting or refuting an inhibitory effect of NSAIDs or COX-2 inhibitors on fracture healing, we can only rely on data from animal studies. A careful analysis of those data show that high doses of drugs that inhibit cyclooxygenase-2 impair bone healing during the time those drugs are administered. Once the drugs are discontinued, animals quickly recover their prostaglandin production and rescue the inhibitory effects. To extrapolate this scenario to a clinical setting, one might imagine that a patient who sustains a fracture or undergoes an operation requiring bone healing might safely be able to take an NSAID or COX-2 inhibitor for one to two weeks and then discontinue the drug. In this case, normal, timely healing might be expected to occur. However, there may be a concern that these drugs would interfere with bone healing when comorbid conditions create settings that are not optimal. For example, are they safe for use in patients who smoke or take glucocorticoids, who have diabetes, or in whom fracture stability or operative fixation is not ideal?

In the absence of randomized, controlled, prospective clinical trials there are very few questions that can be answered with confidence. The role of COX-2 inhibition during fracture healing is no exception. For the present, while I would not dismiss the use of these drugs in the management of post-fracture or post-operative pain in patients requiring bone healing, I would advise that physicians and their patients be familiar with the data and make their decisions accordingly. I would advise that if these drugs were to be used in exchange for narcotic analgesics, their administration should be for fairly short periods of time, probably not to exceed 10 to 14 days. I would be less inclined to consider using these drugs if a patient has any comorbid condition that might itself prevent or delay healing. It is hoped that with proper experimental design and consideration and careful attention to the ethical concerns of research, clinical trials can be conducted to help resolve these issues.


  1. Simon AM, Manigrasso MB, and O’Connor JP: Cyclo-Oxygenase 2 function is essential for bone fracture healing. J Bone Miner. Res., 2002:17:963-976.
  2. Zhang X, Schwarz EM, Young DA, Puzas JE, Rosier RN, and O’Keefe RJ: Cycooxygeanse-2 regulates mesenchymal cell differentiation into the osteoblast lineage and is critically involved in bone repair. J. Clin. Invest., 2002:109:1405-1414.
  3. Glassman SD, Rose SM, Dimar JR, Puno RM, Campbell MJ, and Johnson J: The effect of postoperative nonsteroidal anti-inflammatory drug administration on spinal fusion. Spine 1998:23:834-838.
  4. Reuben SS: Considerations in the use of COX-2 inhibitors in spinal fusion surgery. Anesthesia & Analgesia, 2001:93:798-804.

Thomas A. Einhorn, MD, is chairman of the Department of Orthopaedic Surgery and professor of Orthopaedic Surgery and Biochemistry at Boston University School of Medicine. He serves on the Board of Trustees of the Orthopaedic Research and Education Foundation and on the Board of Trustees of the National Osteoporosis Foundation. He is deputy editor for Current Concepts Reviews for The Journal of Bone and Joint Surgery, and serves on the editorial boards of The Journal of Bone and Mineral Research and Bone. He can be reached at (617) 638-8435.

Home Previous Page