Date: January 22nd, 2021

Guest Skeptic: Dr. Steve Joseph. Steve completed his Sport Medicine fellowship training with the Fowler Kennedy Sport Medicine Clinic in 2017.  He served with the Canadian Forces as a Medical Officer and Flight Surgeon. Steve is currently an Assistant Professor in the Department of Family Medicine at Western University (London, Ontario) working at the Fowler Clinic and the Roth McFarlane Hand and Upper Limb Centre.

Reference: George et al. Risk of Nonunion with Nonselective NSAIDs, COX-2 Inhibitors, and Opioids. J Bone Joint Surg Am. 2020

Case: A healthy 55-year-old woman was out for a walk and had a FOOSH (fall on outstretched hand) of her dominant arm. The X-ray demonstrates a fracture of the distal radius that is in an acceptable position and does not require a reduction. You immobilize her in a below elbow splint which provides significant pain relief and refer her to the local orthopedic fracture clinic. Upon discharge she asks what she should take for pain because she read somewhere that anti-inflammatory drugs like ibuprofen can prevent bone healing. She currently takes thyroid replacement therapy and has no known drug allergies.

Background: There are conflicting studies about fracture healing and the use of non-steroidal anti-inflammatories (NSAIDs) in humans. It remains a controversial topic in the orthopaedic specialty.

When bones break, they usually heal with either surgical or non-surgical management. Sometimes the healing process can take longer than usual (delayed union), does not heal (non-union) or in poor alignment (malunion). Non-union is defined as “a failure of the fracture-healing process” and occurs in up to 1 in 10 fractures.

Several risk factors have been associated with increased risk of delayed or non-union. These factors include: Use of tobacco products, older age, severe anemia, alcohol intake, diabetes, low vitamin D levels, hypothyroidism, poor nutrition, infection, open fracture and certain medications (ex. steroids). The top risk factors for non-union according to a study by Santolini et al were open method of fracture reduction, open fracture, presence of post-surgical fracture gap, smoking, infection, wedge or comminuted types of fracture, high degree of initial fracture displacement, lack of adequate mechanical stability provided by the implant used, fracture location in the poor zone of vascularity of the affected bone, and a fractured tibia [1].

One class of medication that has been implicated in negatively impacting bone healing is NSAIDs. Non-selective NSAIDs block cyclooxygenase (COX)-1 and 2 while selective NSAIDs only inhibit COX-2. There have been multiple studies investigating this issue with mixed results.

Clinical Question: Is there increased risk for fracture non-union with certain classes of NSAIDs?

Reference: George et al. Risk of Nonunion with Nonselective NSAIDs, COX-2 Inhibitors, and Opioids. J Bone Joint Surg Am. 2020

  • Population: Adults (18 years and older) inpatient or outpatients with a diagnosis of certain long bone fractures (neck of femur/tibia/fibula/tibia and fibula/radius/ulna/humerus/clavicle) based on ICD-9 codes.
    • Excluded: Patients less than 18 years of age, multiple fractures, metastatic disease, history of malunion fracture in the year prior or within 90 days
  • Exposure: Filled prescription for a non-selective NSAIDs, selective COX-2 inhibitor and/or opioid within 30 days of the fracture
  • Comparison: Not filling a prescription for a non-selective NSAIDs, selective COX-2 inhibitor and/or opioid within 30 days of the fracture
  • Outcomes:
    • Primary Outcome: Diagnosis of non-union within the 91 to 365 days post fracture. This was based on two definitions. The primary definition used ICD-9 code for nonunion with a procedure to treat nonunion within 30 days of the nonunion diagnosis. The secondary definition was an inpatient or outpatient diagnosis of nonunion.

Authors’ Conclusion:COX-2 inhibitors, but not non-selective NSAIDs, were associated with a greater risk of non-union after fracture. Opioids were also associated with non-union risk, although patients filling prescriptions for opioids may have had more severe fractures.”

Quality Checklist for Observational Study:

  1. Did the study address a clearly focused issue? Yes
  2. Did the authors use an appropriate method to answer their question? Yes
  3. Was the cohort recruited in an acceptable way? Unsure
  4. Was the exposure accurately measured to minimize bias? No
  5. Was the outcome accurately measured to minimize bias? Unsure
  6. Have the authors identified all-important confounding factors? Unsure
  7. Was the follow up of subjects complete enough? Yes
  8. How precise are the results? Fairly precise
  9. Do you believe the results? Yes
  10. Can the results be applied to the local population? Yes
  11. Do the results of this study fit with other available evidence? Yes

Key Results: The final cohort consisted of 339,864 patients identified in over 15 years. Less than 1% were diagnosed with a nonunion (2,996/339,864). The mean age was in the 50’s and around 60% were female. The most common fractures were radius, neck of the femur and humerus.

Patients who filled prescriptions for selective COX-2 inhibitors and opioids but not non-selective NSAIDs were associated with an increased risk of nonunion.

  • Primary Outcome: Nonunion
    • COX-2- inhibitor prescriptions: Adjusted odds ratio (aOR) 1.84 (95% CI; 1.38 to 2.46)
    • Opioid prescriptions: aOR 1.69 (95% CI; 1.53 to 1.86),
    • Non-selective NSAID prescriptions: aOR 1.07 (95% CI; 0.93 to 1.23)

1) Question: They answered the question about an association between filling a prescription for various analgesics and the risk of nonunion. However, the question we want answered is whether or not any of these medications has a causal relationship with nonunion. Stronger evidence would be an RCT of these medications with the outcome of nonunion confirmed clinically and not using ICD-9 codes. It would be difficult to get ethics approval for a placebo-controlled trial. It could also be a challenge to maintain blinding due to the side effect profile of opioids.

2) Cohort Recruitment: We were unsure if the cohort was recruited in an acceptable way to minimize bias. They used ICD-9 codes to identify the individuals with a fracture. There was no reference to support that this is a validated method.

3) Exposure: The exposure was only if the patient filled a prescription, not if they took the prescription. Some prescriptions may have been filled and missed. They did not describe the process well. Filling a prescription does not confirm if they took the medication or even how much and for how long. Non-selective NSAIDS are widely available over the counter medications which also could confound the results. We will talk more about confounders in nerdy point #5.

4) Outcome: Their outcome was based on two non-union definitions. Both relied on ICD-9 codes. They did not provide a reference validating the first definition. The second definition did have a reference. It was a small, single centre study with a positive predictive value of only 89% [2]. The authors of the cited study for the definition acknowledge these weaknesses in their discussion of limitations. This included the issue of PPV being based on prevalence of disease.

5) Confounders: They identified many confounders in their analysis like diabetes, alcohol, steroid use, and others. However, there are other confounders (severe anemia, low vitamin D levels, hypothyroidism, poor nutrition, etc) which have associated with nonunion that were not identified. This could have impacted the magnitude of the point estimate and the 95% confidence interval (precision of the results).

Comment on Authors’ Conclusion Compared to SGEM Conclusion: We generally agree with the authors’ conclusions but would have added a qualification that other unmeasured confounders could have impacted the results.

SGEM Bottom Line: There is no high-quality evidence to support the claim that non-selective NSAIDS cause an increased risk of nonunion.

Case Resolution: You suggest using acetaminophen or ibuprofen for pain.

Dr. Steven Joseph

Clinical Application: This will not change my practice. I will prescribe according to the patient’s preferences, my clinical judgment and available evidence. This will mean usually suggesting non-selective NSAIDs or acetaminophen. Sometimes I will prescribe a short course of opioids. I do not use COX-2 inhibitors because there is no high-quality evidence of better efficacy or safety profile and this study suggests an increased associated risk of nonunion.

What Do I Tell My Patient? Broken bones can be very painful. Immobilization often provides a great deal of pain relief. You can also use acetaminophen or ibuprofen to help with the pain. The goal is not to get to zero pain but to minimize suffering. There is no high-quality evidence that anti-inflammatory drugs like ibuprofen will prevent bone healing.  If the pain is getting worse, you are getting numbness, loosing function, the hand is going cold/pale or you are otherwise worried, please come back to the emergency department for reassessment.

Keener Kontest: Last weeks’ winner was Garrision Lin. He is a nursing student at Western University and is currently doing a placement at South Huron Hospital Association. He knew the first Nurse Practitioner program in the US was in 1965 by Loretta Ford and Dr. Henry K. Silver from the University of Colorado.

Listen to the SGEM podcast to hear this weeks’ question. Send your answer to with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.


Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.


  1. Santolini E, West R, Giannoudis PV. Risk factors for long bone fracture non-union: a stratification approach based on the level of the existing scientific evidence. Injury. 2015 Dec;46 Suppl 8:S8-S19. doi: 10.1016/S0020-1383(15)30049-8. PMID: 26747924.
  2. Boudreau DM, Yu O, Spangler L, Do TP, Fujii M, Ott SM, Critchlow CW, Scholes D. Accuracy of ICD-9 codes to identify nonunion and malunion and developing algorithms to improve case-finding of nonunion and malunion. Bone. 2013 Feb;52(2):596-601. doi: 10.1016/j.bone.2012.11.013. Epub 2012 Nov 19. PMID: 23174214.