A literature search was performed to determine whether non-steroidal anti-inflammatory drugs (NSAIDs) adversely affect the healing of stress fractures. Evidence exists from laboratory studies and animal subjects that NSAIDs can affect fracture healing. This link has not been proved or disproved in human subjects, particularly for stress fractures. In view of the high usage of NSAIDs in treating musculoskeletal disorders, research is required to investigate whether the healing of stress fractures is affected by these drugs.
- non-steroidal anti-inflammatory drugs
- stress fracture
Statistics from Altmetric.com
Non-steroidal anti-inflammatory drugs (NSAIDs) are often recommended for alleviation of pain and inflammation from stress fractures, and their use has been described in case reports of stress fractures.1–3 However, NSAIDs affect prostaglandin synthesis, and, as prostaglandins have been shown to be essential for normal bone turnover and fracture healing, the use of NSAIDS may ultimately affect bone healing by inhibiting maturation of the callus. Animal studies involving NSAIDs have shown a delay in healing time and non-union of fractures. However, human studies have produced inconclusive results, but sources such as Bennell and Bruckner4 are advising against the use of NSAIDs in the treatment of stress fractures.
NSAIDs are a chemically diverse group of medicines available for many different routes of administration. In the United Kingdom, a few NSAIDs are on restricted sale to the public, but most have a prescription only medicine licence. NSAIDs are responsible for more than 2000 deaths a year in the United Kingdom and for nearly one quarter of all adverse drug reactions reported to the Committee on the Safety of Medicines.5
NSAIDs have both central and peripheral analgesic effects. They work peripherally by inhibiting the cyclo-oxygenase pathway, thereby reducing prostaglandin production which would otherwise potentiate the effect of additional mediators of inflammation such as histamine and bradykinin. NSAIDs are therefore most effective in alleviating types of pain that cause amplification of basic pain responses by prostaglandins, such as inflammation.5 Their methods of action centrally are less well understood, as certain NSAIDs will affect pain that is unrelated to inflammation.5
NSAIDS are currently licensed for use in all musculoskeletal disorders, and the British National Formulary6 list of the recognised side effects includes gastrointestinal discomfort, nausea, diarrhoea, and occasionally bleeding and ulceration. Other side effects include hypersensitivity, angioedema, and bronchospasm, but effects on bone metabolism or impairment of fracture healing are not currently listed.
Stress fractures were first described by Burrows6a in 1956 in a series of five ballet dancers. Most stress fractures affect the lower legs or metatarsals, but an estimated 1% occur at the femoral neck.7 It is generally accepted that certain stress fractures are at risk of progression to complete fractures. Studies have shown that stress fractures start with periosteal oedema, which can be visualised by magnetic resonance imaging, followed by marrow involvement, then ultimately a cortical stress fracture.1 Various treatments have been suggested for non-healing stress fractures, including controlled rest, braces, and surgery.4,8 Importance is now placed on maintaining fitness during the recovery process, through cross training and the modification of any risk factors before a protected return to sport occurs.4 Although many stress fractures will heal spontaneously, many factors, including continued mechanical stress at the fracture site, smoking, and possibly NSAIDs, may impair this healing.4,7
For fracture healing to progress, mesenchymal cells must differentiate to form chondrocytes and osteoblast precursors. Previous studies have shown that the cyclo-oxygenase pathway has a vital role during both osteogenesis and chondrogenesis by producing prostaglandin E2.9,10 Cyclo-oxygenase is present in many tissues as a constitutive isoform (COX-1). However, at sites of inflammation, cytokines stimulate the induction of COX-2, a second isoform. Work in 2002 by Zhang et al11 suggests that COX-1 has a role in bone homoeostasis, but that COX-2 is a stress response gene which acts in inflammation by regulating the induction of several genes (cbfa-l and osterix) to mediate normal skeletal repair by facilitating osteogenesis.
Studies initially performed in the 1970s and 1980s showed that NSAIDs may interfere with fracture healing. However, an article by Penman in the New Scientist in 2002,12 which was widely reported in the lay and medical press, highlighted this subject and put it once more into the public domain. On the basis of previously published animal and human research, Penman discussed whether NSAIDs could interfere with fracture healing.
This review sets out to investigate the evidence base for NSAIDs affecting stress fracture healing.
Literature searches for original publications and reviews were conducted using Cochrane, Embase, Medline, SPORTDiscus, and Your. Articles were then pooled for appraisal. Dates searched covered the entire history of these databases. Keywords used were: anti-inflammatory; bone; COX; cyclo-oxygenase; fracture ± stress; ibuprofen; indomethacin; healing; NSAIDs.
Specific journals were also individually searched using the appropriate websites. These included the British Journal of Sports Medicine and the BMJ. Issues of the publications Bandolier, Drugs & Therapeutics Bulletin, and MeReC Bulletin were all searched from 1993 to November 2003, as was the current edition of the BMJ’s publication Clinical Evidence. The British National Formulary and several widely available pharmacology texts were also used for reference.
Articles were filtered to English language and included animal studies. Articles were discarded by the first author (PW) if they were not deemed to assist in answering the underlying question.
Levels of evidence were graded according to standard criteria of quality: level A, meta-analysis or randomised controlled trials; level B, other scientific studies; level C, consensus or expert opinions.13
No publications were found that dealt specifically with stress-type fractures and NSAIDs, so the literature search was extended to include all fracture types. Conclusions were drawn from these as appropriate.
Studies investigating the effects of NSAIDs on bone healing
There remains a paucity of research on this topic, despite several authors highlighting the theoretical risks that NSAIDs may have on wound and bone healing. Work has been carried out on animal subjects and different NSAIDs, and research has been performed investigating the risk in human subjects with complete fractures or non-union after surgical procedures. It is mostly from these studies that any information on the healing of stress fractures and NSAIDs must be extrapolated.
NSAIDs are often used after soft tissue and bony injury and in the period after orthopaedic and other surgical procedures.14 Some authors believe that COX-2 inhibitors may be safer alternatives to traditional NSAIDs because one study involving rabbit models showed no impairment of healing after spinal surgery. However, we are reminded that NSAIDs have been shown to affect bone osteogenesis during fracture healing, and very limited work has been published comparing NSAIDs and different COX-2 inhibitors.15
Animal studies (table 1) have shown that NSAIDs, including indomethacin16–21, aspirin,16 and ibuprofen,17 can all affect healing of a variety of fractures, including those affecting the rat forelimb,16 vertebra,18 or femur.19 All doses of indomethacin appear to cause delayed bone healing, but aspirin only delays healing at levels approaching toxicity.16 The healing of partial osteotomies was impaired to a lesser extent than healing of complete osteotomies. This is thought to be due to increased mechanical stability of the fracture.19
Prostaglandin appears to be an essential component for bone healing, with delays in healing resulting from a delay in maturation of the callus rather than direct effects on collagen synthesis.16,17,21 This delay is reversible on the cessation of indomethacin but not ibuprofen.17,18
The processes of growth and development and those of bone repair appear to use different pathways, as indomethacin appears to have no effect on ordered bone growth and remodelling of rat femurs.22
Although two of the studies examining the impact of NSAIDs on bone healing after fractures in humans were prospective,23,24 the majority were retrospective (table 2). Studies on patients treated after sustaining a variety of traumatic fractures, including Colles-type,23 tibia,25 clavicle,26 non-union of the femoral diaphysis,2 and acetabular,24 produced inconsistent results, with one study showing no difference and the remainder showing delayed healing or an increased risk of non-union. Several other large, high quality studies of healing of traumatic fractures did not mention NSAIDs.27–29
Of the two prospective studies in humans, the double blinded study by Adolphson et al23 on recovery and bone density after Colles fracture showed no effect of NSAIDs, unlike the study by Burd et al,24 which looked at patients at risk of heterotopic ossification after surgical fixation for acetabular fractures. These were randomly allocated to prophylaxis, using either radiation treatment or indomethacin, or to a control group. Patients who received indomethacin had a greater risk of non-union of the long bones than those who received either radiotherapy or no prophylaxis. The retrospective studies of healing and recovery after fractures all showed an adverse effect on bone healing, either delayed bone healing or an increased rate of non-union, by the administration of NSAIDs.
Clough3 highlights the case of a 55 year old man who had suffered a stress fracture of the femoral neck. Diagnosis was delayed by 10 weeks, during which time he was given several courses of NSAIDs for analgesia. By the time it was diagnosed, the stress fracture had displaced and required closed reduction and stabilisation using a dynamic hip screw. Although the author made no connection between the administration of the NSAIDs and the non-union of the stress fracture, it highlights how often NSAIDs are used.
Several studies looked at the effects of NSAIDs on bone healing after spinal surgery.14,30 Non-union was found to be up to five times more likely in patients who received intramuscular ketorolac after the operation,30 but no difference was found in those receiving rofecoxib compared with controls.14 The fivefold increase in the study by Glassman et al30 was greater than that attributed to other risk factors, including smoking, but did involve large doses of intramuscular ketorolac, which has different pharmacokinetics from oral doses.31
NSAIDs appear to have no overall effect on the risk of fractures. Van Staa et al9 examined the fracture risks of more than 200 000 regular NSAID users, a similar number of incidental NSAID users, and matched controls, all within the primary care setting. Differing from previously published smaller studies, this study did not support an overall effect of NSAIDs on bone metabolism, and there were no significant differences in fracture risk between regular and incidental NSAID users. It did indicate that regular NSAID users had a higher risk of fracture than non-NSAID controls, which appeared to decrease as duration of treatment increased, but did not fall below baseline levels. This study showed no consistent differences in fracture risk between different NSAIDs, which is different from previously published studies. Although many confounding factors were taken into account, other factors such as previous bone density, diet, and level of exercise could not be taken into account.
In 2001, Bandolier (www.ebandolier.com) published a review32 of the study by Van Staa,9 which was followed up by an extended review published as Bandolier Extra.31 These are non-peer reviewed, independent publications concerned with evidence based health care which look at published literature located by a combination of a search of PubMed using unspecified free text terms and a review of the Cochrane Library. No specific information was readily available detailing inclusion and exclusion criteria for the reviews, and formal weighting was not given to evidence. These two reviews discuss laboratory experiments that show that NSAIDs have an effect on bone metabolism in animal subjects and appraise the research by Van Staa et al,9 which contradicts this evidence in humans. The second review examines other studies, including a Cochrane review examining heterotopic bone formation after hip replacement. This review does not mention failure of bone healing after use of NSAIDs, nor do the articles by Glassman et al30 and Reuben et al14 on the effects of NSAIDs and COX-2 inhibitors after spinal surgery, or that by Adolphson et al23 on the effects of NSAIDs after Colles fractures. It also discusses several studies on bone healing that make no mention of NSAIDs, including those of Perlman and Thordarson,29 Alho et al,27 and Karladani et al.28 The review concludes that there is good clinical evidence showing inhibition of fracture healing from animal studies, but that this may not be applicable to clinical situations involving humans. The reviews conclude that there is only reliable evidence for NSAIDs affecting heterotopic bone formation, but not fracture healing, and suggest that any significant effect on fracture healing in humans is unlikely to have been missed.
In a recently published sports medicine textbook, Bennell and Brukner4 discuss the management of stress fractures, and conclude, after a personal communication, that there is a theoretical risk that some NSAIDs may slow or prevent repair of stress fractures, but accept that there are no high quality formal trials from which to draw conclusions.
Penman12 reintroduced this subject to the public’s awareness in the New Scientist in 2002. This examined evidence from several trials with animal and human subjects, but no mention was made of how trials were identified, and no inclusion or exclusion criteria for consideration were specified. The article comments that a trial of rofecoxib in rats inhibited fracture healing, but did not give specific figures detailing risk. The article states that NSAIDs, including ibuprofen and indomethacin, delay fracture healing by one or two weeks in rats, which is the equivalent to slowing healing by 20–25%, but the trials from which these figures were gleaned are not readily identified. The article concludes with the suggestion that people with healing fractures should avoid NSAIDs and that further investigation is required.
NSAIDs can either reversibly or irreversibly block the cyclo-oxygenase pathway, thereby inhibiting prostaglandin synthesis. From the available evidence it is still unclear whether this inhibition significantly impairs healing of stress fractures in humans.
Animal studies have shown that NSAIDs delay healing in complete fractures. One study showed that NSAIDs impaired healing to a greater extent in a complete osteotomy group than in the partial osteotomy group. This was thought to be due to increased mechanical stability, adding another confounding variable to the issue. Many of the animal studies, however, used drugs, doses, and route of administration (typically indomethacin at 1–4 mg/kg/day intravenously in rats) that may not represent equivalent pharmacokinetics and pharmacodynamics to those used in human subjects. The equivalent dose in an 80 kg man would be 80–320 mg/day via a different route of administration. This would exceed the licensed dose of 50–200 mg/day in oral divided doses for analgesia.6
In studies on human subjects, there is a lack of high quality evidence. NSAIDs are often used in a wide range of musculoskeletal conditions, so it is to be expected that they are often mentioned in case reports involving stress fractures. There were no trials that looked directly at stress fracture healing with NSAID use, and, as the animal studies show that the healing processes may be impaired by NSAIDs to a different extent for different types of fracture, the interpretation of results from trials involving complete frank fractures may be misleading.
Many of the trials that showed a clinically significant effect on complete fracture healing in humans were retrospective with potential flaws in their methodology. Overall the evidence in humans is inconclusive, and, as there are many studies on fracture healing that would have been expected to show an effect of NSAIDs but did not, it is possible that the available evidence showing a link may represent publication bias.
What is already known on this topic
There is evidence from animal studies that NSAIDs can adversely affect fracture healing.
What this study adds
There is no conclusive evidence to document any effect of NSAIDs on stress fracture healing in humans. Until such evidence is produced, it is prudent to limit their use in patients with proven stress fracture.
In view of the easy availability and common use of NSAIDs to treat a variety of soft tissue and bony injuries, and the failure to answer the original question, formal studies investigating the possible impairment of stress fracture healing with the use of NSAIDs are required. Further studies are also required to assess whether COX-2 inhibitors affect bone healing differently from non-specific NSAIDs in a clinical setting. Until the results of this research are available, the option of least risk is to avoid the use of NSAIDs in the treatment of stress fractures wherever possible.
Competing interests: none declared
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.