Three words — 'you're bone-on-bone' — have ended more active lives, caused more unnecessary surgery, and produced more fear-driven disability than almost any other phrase in modern medicine.
Robert is sixty. Retired builder. He has been living with knee pain for eight years. He's had two arthroscopies, a cortisone injection every six months, and has been on the waiting list for a knee replacement for the past two years. He walks with a cane. He hasn't been further than the letterbox in three years. He has gained twenty-two kilos. His quadriceps have atrophied to the point where getting out of a chair requires his arms. He is, by every functional measure, disabled.
And the moment that set this entire cascade in motion was not an injury, not a fall, not a degenerative event. It was three words from his GP, spoken while looking at an X-ray: "You're bone-on-bone."
Those three words rewired Robert's brain. They installed a belief — a prediction — that his knees were structurally destroyed, that movement would accelerate the destruction, that a knee replacement was inevitable, and that the only rational response was to protect what remained by doing as little as possible.
Every clinical decision Robert made after that moment — the reduced activity, the reliance on the cane, the weight gain, the progressive deconditioning — was a logical consequence of the belief those three words created. Robert wasn't irrational. He was responding rationally to catastrophic information. The problem was that the information was misleading.
"Bone-on-bone" is one of the most commonly used, most poorly contextualised, and most clinically damaging phrases in orthopaedic medicine. It sounds terminal. It sounds structural. It sounds like the joint is grinding itself to dust with every step. It sounds like the only possible outcome is surgical replacement.
The evidence tells a very different story.
When a clinician says "bone-on-bone," they are describing a radiographic finding: the apparent loss of joint space on an X-ray, suggesting that the articular cartilage — the smooth, shock-absorbing surface that lines the ends of bones — has thinned or worn away to the point where the bones appear to be in contact.
This finding is graded using the Kellgren-Lawrence (KL) classification system, a four-point scale developed in 1957:
Grade 0: Normal. No features of osteoarthritis. Grade 1: Doubtful. Possible osteophytic lipping. Grade 2: Minimal. Definite osteophytes, possible joint space narrowing. Grade 3: Moderate. Multiple osteophytes, definite joint space narrowing, some sclerosis, possible deformity. Grade 4: Severe. Large osteophytes, marked joint space narrowing, severe sclerosis, definite deformity.
"Bone-on-bone" typically corresponds to KL Grade 4 — the most severe radiographic classification. It sounds catastrophic. The image it conjures — two bones grinding against each other like mortar and pestle — is visceral and terrifying.
But here is what the phrase obscures:
First, X-rays are two-dimensional images of a three-dimensional joint. The apparent "loss of joint space" depends on the angle of the X-ray beam, the position of the leg, and the degree of weight-bearing at the time of imaging. A weight-bearing X-ray taken with the knee in slight flexion can show significantly different joint space than a non-weight-bearing image of the same knee taken moments later. The "bone-on-bone" appearance can be partly an artefact of imaging technique.
Second, cartilage is not visible on X-ray. X-rays show bone. They infer cartilage loss from the reduction in the gap between bones. But the gap between bones on an X-ray is affected by meniscal integrity, synovial fluid volume, and joint positioning — not just cartilage thickness. An MRI — which does visualise cartilage directly — sometimes reveals more cartilage remaining than the X-ray suggested.
Third, and most critically: the correlation between radiographic severity and pain is remarkably poor.
This is where the phrase "bone-on-bone" begins to crumble under scientific scrutiny.
Bedson and Croft (2008) conducted a systematic review examining the association between radiographic knee osteoarthritis and knee pain. They found that while there was a statistical association between radiographic severity and the presence of pain, the relationship was weak and inconsistent. Many individuals with severe radiographic OA had minimal or no pain. Many individuals with significant pain had mild or no radiographic changes. The X-ray, in isolation, was a poor predictor of the clinical picture.
Hannan et al. (2000), using data from the Framingham Osteoarthritis Study, found that only 47% of individuals with radiographic knee OA (KL Grade 2 or above) had knee pain. Conversely, many individuals with knee pain had normal or near-normal radiographs. The concordance between what the X-ray showed and what the patient felt was barely better than a coin toss.
Neogi et al. (2009) examined the relationship between structural severity and pain severity in 696 participants with symptomatic knee OA. They found that while radiographic severity was associated with the likelihood of pain, it explained only a small proportion of the variance in pain intensity. Other factors — psychological distress, fatigue, widespread pain, comorbid conditions — were stronger predictors of pain severity than the structural findings themselves.
Perhaps most striking: Finan et al. (2013) demonstrated that central sensitisation — the amplification of pain processing in the central nervous system — was a stronger predictor of clinical pain severity in knee OA than the degree of structural joint damage. In other words, the state of the nervous system was more important than the state of the joint.
The collective message from this evidence is unambiguous: structural severity on imaging and clinical pain severity are poorly correlated. People with severe X-ray findings can have minimal pain. People with mild X-ray findings can have severe pain. The X-ray shows you the joint. It does not reliably tell you about the pain.
If the structural findings don't reliably predict pain, then why does telling a patient they are "bone-on-bone" so reliably make them worse?
Because of the nocebo effect.
Nocebo — the opposite of placebo — is a negative health outcome produced by negative expectation. When a patient hears "bone-on-bone," their brain processes a threat signal of extraordinary magnitude. The implied message is: your joint is destroyed. Movement will grind it further. Damage is ongoing and accelerating. The only solution is surgical replacement. Until then, protect what remains.
Every component of that implied message is a DIM — a Danger In Me signal. Each DIM fills the patient's threat cup. When the cup overflows, the brain produces pain. Not because the bones are grinding — but because the prediction says they are.
This is not speculation. Nocebo effects in clinical communication have been extensively documented. Benedetti et al. (2007) demonstrated that negative verbal suggestions about a procedure increase pain, anxiety, and physiological stress responses. Darlow et al. (2013) found that patients who received negative information about their back condition showed increased pain, disability, and fear-avoidance compared to those who received neutral or positive information. Words are not neutral. In healthcare, words are pharmacologically active.
Robert's presentation at assessment offered a clear map of where his pain was actually coming from. Yes, his knees showed osteoarthritic changes — that was real. But the factors amplifying his pain and disability extended far beyond the X-ray.
His Hardware contribution was perhaps 30%. Osteoarthritic changes in both knees. Real, but not uniquely severe for his age.
His Software contribution was approximately 25%. Eight years of being told his knees were "bone-on-bone" had installed a fear-avoidance pattern so deep that any knee sensation triggered catastrophic predictions. His Pain Catastrophising Scale score was 38. His Tampa Scale of Kinesiophobia (fear of movement) was 42. His nervous system had learned to amplify every signal from his knees into an alarm about impending destruction.
His Energy Plant contribution was 45%. His cardiovascular fitness was catastrophic — he could not sustain walking at 3 km/h for more than 90 seconds. His quadriceps strength was 40% of age-matched norms. He had gained 22 kg. His sleep was fragmented. He was socially isolated. He had essentially ceased to have an infrastructure for movement.
The three-bucket analysis revealed something that an X-ray-only assessment would have missed entirely: Robert's pain was not primarily a Hardware problem. It was an Energy Plant problem wearing a Hardware mask, amplified by Software dysregulation. Treating the Hardware with surgery would leave the majority of his pain drivers untouched.
Robert's treatment plan had three simultaneous streams.
Software rewriting. The first session was pain education. Not a lecture — a conversation. He learned what his X-ray actually showed and did not show. He learned the Bedson and Hannan findings. He learned that his quadriceps weakness was a more reliable predictor of his pain than his cartilage thinning. He learned the nocebo effect. He learned that "bone-on-bone" was a radiographic description, not a biomechanical reality.
Energy Plant rebuilding. Three minutes on a stationary bike. That was his Week 1 prescription. Not because three minutes is a meaningful dose of exercise — but because it was the dose his body could complete without triggering a flare. We then progressed by 10% per week, time-contingent not pain-contingent. By Week 8, he was doing 15 minutes. By Month 4, 30 minutes. Parallel strength training — quadriceps, glutes, calves — started with bodyweight only and progressed to resistance bands, then weights.
Hardware support. Manual therapy to address the genuine joint restrictions that had developed from years of disuse. Clinical Pilates to teach him how to load his joints safely. Appropriate use of paracetamol for flare management — not as a primary treatment but as a bridge to allow participation in the other streams.
Let me be clear: knee replacement surgery is a genuinely effective intervention for the right patient. Roughly 80-85% of patients who undergo total knee arthroplasty report satisfaction with the outcome. It can be life-changing. For the right patient, at the right time, it can be transformative.
But "the right patient" is not determined by the X-ray alone. The evidence consistently shows that outcomes from knee replacement are strongly influenced by pre-operative psychological factors, expectations, physical fitness, and the presence of central sensitisation.
Vissers et al. (2012) found that pre-operative catastrophising was a significant predictor of persistent pain after total knee arthroplasty. Patients who went into surgery with high catastrophising scores were more likely to have poor outcomes — not because the surgery was technically inadequate, but because the sensitised nervous system continued to produce pain even after the structural problem was addressed.
Wylde et al. (2011) reported that approximately 20% of patients who undergo knee replacement continue to experience chronic pain two years after surgery. The surgery was successful by every structural criterion — the joint was replaced, the alignment was corrected, the implant was well-positioned — but the patient still hurt. In many of these cases, the pain was being driven by central sensitisation, fear-avoidance, deconditioning, and psychosocial factors that the surgery could not and did not address.
This is not an argument against surgery. It is an argument for comprehensive pre-operative assessment and optimisation — addressing Software and Energy Plant before reaching for the scalpel, so that when surgery is performed, the nervous system is in a state that allows it to benefit fully from the structural correction.
And for many patients — like Robert — comprehensive conservative management renders surgery unnecessary altogether.
When Robert presented for assessment, his questionnaires told a clear story. His quadriceps were profoundly weak — eight years of disuse atrophy. His cardiovascular fitness was so low that walking to the kitchen left him breathless. His BMI had risen from 27 to 34. His sleep was broken. His mood was low. His contribution map: Hardware 30%, Software 25%, Energy Plant 45%.
Robert's knees were genuinely arthritic. Nobody was denying that. But the twenty-two kilos of weight gain, the profound deconditioning, the atrophied quadriceps, the fear of movement, and the belief that his knees were "bone-on-bone" and therefore unfixable — these were amplifying his pain far beyond what the joint changes alone would produce.
Treatment started with three minutes on a stationary bike. Three minutes was his maximum. By week eight, it was fifteen minutes. By month four, he had put the cane in the garage. By month six, he walked to the shops for the first time in five years. He cried on the walk home.
Robert walked off the knee replacement waiting list. His surgeon was surprised but supportive. His Hardware hasn't changed — he still has osteoarthritis. But his Energy Plant was rebuilt from the ground up, his Software was rewritten, and his life no longer revolves around three words on a radiology report.
"Rest didn't give me a chance to heal," he said at his last appointment. "It gave my pain permission to take over."
Reframe the phrase. If you've been told you're "bone-on-bone," try replacing it with: "I have age-related wear in my joint that is contributing to my pain alongside other factors — my muscle strength, my fitness, my sleep, my beliefs about my body, and my nervous system's sensitivity." This is not denial. This is accuracy. And accuracy is a safety signal.
Ask one question at your next appointment. "How well does the severity of my X-ray findings correlate with my level of pain?" If your clinician can discuss the evidence on concordance (or lack thereof) between imaging findings and clinical symptoms, you've found someone who understands the complexity. If they simply point at the X-ray and say "that's why you hurt," you've found someone working from an incomplete model.
Test your quadriceps. Can you stand up from a standard-height chair without using your arms? Can you do it ten times? Twenty? If you struggle with five, your muscular support around the joint is depleted — and rebuilding it is one of the most powerful interventions available for knee OA, regardless of what the X-ray shows. A physiotherapist can design a graded strengthening program that starts where you are, not where you think you should be.
Look at the whole picture. Pain in a joint with osteoarthritis is never just about the joint. It is about the joint plus the muscles around it, plus the fitness of the body that carries it, plus the sleep that restores it, plus the beliefs that either protect or terrify you, plus the nervous system that decides how much alarm to produce. The X-ray shows one piece. The pain comes from all of them.
Three words changed Robert's life for eight years. Understanding can change it back.
Absolutely not. Osteoarthritis is a genuine, well-characterised condition involving cartilage degradation, subchondral bone changes, osteophyte formation, and synovial inflammation. It is real, it produces genuine symptoms, and in some cases it requires surgical intervention. What the evidence shows is that the radiographic severity of OA is a poor predictor of the clinical pain experience — and that factors beyond the structural damage (muscle strength, fitness, sleep, beliefs, nervous system sensitivity) often contribute more to the patient's pain and disability than the joint changes themselves.
No. Imaging is a valuable clinical tool for ruling out serious pathology (fractures, infections, malignancy), assessing structural damage when surgery is being considered, and monitoring disease progression. The issue is not with imaging itself but with how the results are communicated and interpreted. Ask your clinician to interpret your results in context, not in isolation.
Yes. The evidence for exercise in knee osteoarthritis is strong across all radiographic grades. A Cochrane review by Fransen et al. (2015) concluded that land-based therapeutic exercise provides short-term benefit in terms of reduced knee pain and improved physical function among people with knee OA, and that these benefits are comparable to those reported for non-steroidal anti-inflammatory drugs. Exercise works regardless of radiographic severity.
Knee replacement is appropriate when a patient has severe, disabling knee pain that has not responded to comprehensive conservative management — including structured exercise, weight management, pain education, sleep optimisation, and psychological support — over a sustained period. The decision should be based on the patient's functional limitations, quality of life, and response to conservative care, not on the X-ray alone. Pre-operative optimisation (strengthening, weight loss, pain education, psychological preparation) improves surgical outcomes. At Upwell Health Collective in Camberwell, our multidisciplinary team works with patients to optimise all three buckets — Hardware, Software, and Energy Plant. Call (03) 8849 9096 or book online.