Healing & Recovery
Do NSAIDs Slow Tendon and Bone Healing?
The short answer
In rat surgical models, both selective and non-selective NSAIDs consistently and significantly weakened healing tendon-to-bone repairs — in one 180-rat study, both indomethacin and celecoxib produced significantly lower failure loads than untreated controls at every time point measured, and 5 of 120 NSAID-treated tendons failed to heal at all. But the human evidence on bone is much weaker: a systematic review of 16 human studies covering nearly 13,000 patients found the data too inconsistent in quality and design to support a clear recommendation, and a large 309,330-fracture database study tied elevated nonunion risk specifically to combined NSAID-plus-opioid use, not NSAID use alone. No human trial has directly tested whether NSAIDs weaken a healing tendon the way they do in rats.
Quick answers
- What is it?
- This asks whether NSAID pain relievers (like ibuprofen-class drugs) slow down the healing of tendons and bone.
- Why are people interested in it?
- NSAIDs are among the most commonly used painkillers after injury or surgery, so people want to know if taking them for pain could interfere with their own healing.
- What has actually been studied?
- Rat surgical models of tendon-to-bone repair comparing NSAID-treated to untreated animals, plus human studies including a 309,330-fracture insurance-claims database, a systematic review of 16 human studies on fracture and spinal-fusion nonunion, and one small randomized trial.
- How strong is the evidence?
- Strong and consistent in rats, weak and inconsistent in humans: rat studies reliably show impaired tendon healing with NSAIDs, but human bone-healing studies are too inconsistent in quality and design to support a clear recommendation, and no human trial has directly tested NSAIDs on a healing tendon.
- What don't we know yet?
- Whether NSAIDs impair tendon healing in humans the way they do in rats, whether a safe duration or dose exists, whether NSAID effects can be separated from opioid co-use in the human fracture data, and why the human bone-healing signal is so much weaker than the rat tendon signal.
- Where can I read the original research?
- Sources at the bottom of this page: Cohen DB, Kawamura S, Ehteshami JR, Rodeo SA, The American Journal of Sports Medicine, 2005; and Bektaş E, Çelebi ME, Kurtulmuş T, Yılmaz F, Acta Orthopaedica et Traumatologica Turcica, 2025.
The short answer: In rats, the evidence is consistent: both non-selective NSAIDs (like indomethacin) and COX-2-selective NSAIDs (like celecoxib) measurably weaken tendon-to-bone healing after surgical repair, reducing biomechanical strength and impairing collagen organization at every time point studied. In humans, the picture is much less settled. A systematic review covering nearly 13,000 patients across 16 studies — only 3 of them randomized — found the data too inconsistent to support a clear recommendation on fracture or spinal fusion healing, and a large database study tied elevated nonunion risk to combined NSAID-and-opioid use rather than NSAID use in isolation. No human trial has directly tested NSAID effects on a healing tendon.
What the research shows
Rat tendon-to-bone healing: a consistent, statistically strong signal
The most direct evidence connecting NSAIDs to impaired tendon healing comes from a 2005 study in the American Journal of Sports Medicine. Researchers performed surgical rotator cuff repairs on 180 Sprague-Dawley rats, then gave 60 rats celecoxib, 60 indomethacin, and 60 standard chow for 14 days after surgery. At 2, 4, and 8 weeks, biomechanical testing found significantly lower failure loads in both NSAID groups compared to controls (P<.001 at every time point), with no meaningful difference between the two drugs. Collagen organization and maturation, assessed with polarized light microscopy, was also significantly worse in both NSAID groups at 4 and 8 weeks (P<.001). Five tendon repairs failed to heal completely — four in the celecoxib group, one in the indomethacin group — versus none among the untreated controls. The authors concluded the effect appeared linked to COX-2 inhibition specifically, since both a COX-2-selective and a non-selective drug produced the same impairment. DOI
A more recent 2025 study extended this comparison to include acetaminophen as a non-NSAID control. In 56 rats with surgically repaired supraspinatus tendons, both naproxen and celecoxib produced significantly worse histological healing scores (modified Bonar score, P=.01) than acetaminophen or untreated controls at 28 days. The biomechanical differences in this study — maximum tensile strength and stiffness — trended in the same direction but did not reach statistical significance (P=.28 and P=.40, respectively). The authors concluded that "early postoperative administration of COX-2 selective and nonselective NSAIDs may compromise early tendon-bone healing compared to acetaminophen," while noting the biomechanical signal was weaker than the histological one at this stage. DOI
Human bone healing: a much murkier picture, despite huge sample sizes
The human evidence has been studied mostly through fracture and spinal fusion nonunion — not tendon repair directly — and the picture is far less consistent than the rat data. A 2016 study in JAMA Surgery analyzed insurance claims data covering 309,330 fractures across 18 bones in a database of roughly 90.1 million patients. The overall nonunion rate was 4.9%. In multivariate analysis, the combined use of NSAIDs plus opioids was associated with an elevated odds ratio for nonunion of 1.84 (95% CI, 1.73-1.95) — but this specific variable measured concurrent NSAID-and-opioid use, not NSAID use in isolation, and sat alongside many other risk factors including smoking (OR 1.20), diabetes (OR 1.40), and osteoporosis (OR 1.24). DOI
A 2018 systematic review in the Journal of Clinical Anesthesia took a more direct look at the question, searching 38 years of literature for human studies specifically measuring NSAID exposure against fracture or spinal fusion nonunion with radiological confirmation. It identified 16 total human studies — 3 prospective randomized controlled trials and 13 retrospective cohort studies — covering 12,895 patients combined. The review's conclusion was blunt: published human trial results "did not show strong evidence" that NSAIDs used for postoperative pain increase nonunion rates, but the underlying studies were so inconsistent in drug choice, dosage, exposure duration, and nonunion assessment method that they could not be pooled into a meta-analysis, and overall study quality was rated low by both Jadad and Oxford Levels of Evidence criteria. The reviewers concluded "no clinical recommendation can be made" from the existing human data. DOI
One of the three randomized trials referenced in that review was a 2015 study of 80 patients undergoing lumbar spinal fusion, comparing a celecoxib-containing multimodal pain regimen against standard morphine. Pain and disability scores were significantly better with the multimodal regimen, and the study reported no difference in nonunion rate between groups — though this trial was designed primarily to test pain control, not fusion healing, and was not powered specifically to detect a nonunion difference. DOI
What we don't know yet
Whether NSAIDs impair tendon healing in humans the way they do in rats. Every tendon-to-bone healing study identified in this search was conducted in rats. No human trial has directly measured NSAID effects on a healing tendon repair using biomechanical or structural outcomes.
Whether a "safe" duration or dose exists. The rat studies used continuous dosing for 14 days; human data on fracture nonunion covers highly variable, often unspecified exposure patterns, and no threshold has been established in a controlled human trial.
Whether NSAID effects can be separated from opioid co-use in the human fracture data. The largest human dataset (Zura et al., 2016) measured combined NSAID-and-opioid exposure as a single variable, not NSAID use alone.
Why the human bone-healing signal is so much weaker than the rat tendon signal. No published work directly reconciles the strong, consistent rat effect with the inconsistent, low-quality human evidence.
Whether COX-2 selectivity matters clinically. The rat data suggests both selective and non-selective NSAIDs impair healing similarly, but this comparison has not been tested in a human tendon-healing trial.
Sources
- Cohen DB, Kawamura S, Ehteshami JR, Rodeo SA. Indomethacin and celecoxib impair rotator cuff tendon-to-bone healing. The American Journal of Sports Medicine, 2005;34(3):362-369. DOI
- Bektaş E, Çelebi ME, Kurtulmuş T, Yılmaz F. Comparative effects of cyclooxygenase-2 selective and nonselective nonsteroidal anti-inflammatory drugs and acetaminophen on rotator cuff tendon-bone healing in a rat model. Acta Orthopaedica et Traumatologica Turcica, 2025;59(5):245-252. DOI
- Zura R, Xiong Z, Einhorn T, et al. Epidemiology of Fracture Nonunion in 18 Human Bones. JAMA Surgery, 2016;151(11):e162775. DOI
- Borgeat A, Ofner C, Saporito A, Farshad M, Aguirre J. The effect of nonsteroidal anti-inflammatory drugs on bone healing in humans: A qualitative, systematic review. Journal of Clinical Anesthesia, 2018;49:92-100. DOI
- Kim SI, Ha KY, Oh IS. Preemptive multimodal analgesia for postoperative pain management after lumbar fusion surgery: a randomized controlled trial. European Spine Journal, 2015;25(5):1614-1619. DOI
Article records maintained via PubMed. Peptide Proof reports what studies found. It does not provide medical advice, dosing guidance, or treatment recommendations.
Topic: Recovery & Tissue Research
Sources
- Indomethacin and celecoxib impair rotator cuff tendon-to-bone healing
- The American Journal of Sports Medicine, 2005 · n=180 · DOI
- Comparative effects of cyclooxygenase-2 selective and nonselective nonsteroidal anti-inflammatory drugs and acetaminophen on rotator cuff tendon-bone healing in a rat model
- Acta Orthopaedica et Traumatologica Turcica, 2025 · n=56 · DOI
- Epidemiology of Fracture Nonunion in 18 Human Bones
- JAMA Surgery, 2016 · n=309330 · DOI
- The effect of nonsteroidal anti-inflammatory drugs on bone healing in humans: A qualitative, systematic review
- Journal of Clinical Anesthesia, 2018 · n=12895 · DOI
- Preemptive multimodal analgesia for postoperative pain management after lumbar fusion surgery: a randomized controlled trial
- European Spine Journal, 2015 · n=80 · DOI
Get the evidence, not the hype.
New research briefs, plainly written and fully sourced. No dosing advice, ever.