A comprehensive descriptive analysis of hip and knee radiographic osteoarthritis in the UK Biobank in relation to joint pain, joint site interrelationships, obesity, race and deprivation: Findings from 59,475 individuals

This study of 59,475 UK Biobank participants reveals that radiographic hip and knee osteoarthritis exhibit distinct patterns of inter-joint relationships, symptom associations, and risk factor profiles, highlighting the heterogeneity of the disease and the need for joint-specific treatment approaches.

The Gist

Imagine your body is a busy construction site with two main pillars holding up the roof: your hips and your knees. Over time, the "cement" (cartilage) in these pillars can wear down, and the edges can get rough and bumpy. This condition is called Osteoarthritis (OA).

For a long time, doctors thought of hip and knee wear-and-tear as the same problem happening in different places. But this new study, looking at data from nearly 60,000 people in the UK, suggests that hips and knees are actually like two different types of houses that happen to be next to each other. They might look similar from the outside, but they have different blueprints, different vulnerabilities, and different ways of breaking down.

Here is the breakdown of what the researchers found, using some simple analogies:

1. The "Bump" Detector (How they measured it)

Usually, to see arthritis, you need an X-ray. But this study used a special type of scan called a DXA scan (often used for bone density) and ran it through a super-smart AI robot.

• The Analogy: Think of the AI as a highly trained mechanic with a magnifying glass. It looked at the scans and counted the "bumps" (bone spurs) and measured the "gaps" (joint space) to give every joint a grade from 0 (perfect) to 4 (very worn out).
• The Result: They found that the more "bumps" and "gaps" a joint had, the more likely the person was to be in pain. It's like a car with a flat tire: the more damaged the tire, the harder it is to drive.

2. The "Mirror" Effect vs. The "Neighbor" Effect

The study looked at how arthritis spreads.

• The Mirror Effect (Bilateral): If you have bad knees on the left, you are very likely to have bad knees on the right. If you have bad hips on the left, you are likely to have bad hips on the right.
  • Analogy: It's like having two identical twins. If one twin gets a cold, the other almost certainly will too. The study found that if you have knee arthritis, your odds of having it in the other knee are 26 times higher. That's a huge mirror effect!
• The Neighbor Effect (Cross-Joint): If you have bad hips, does that mean you'll get bad knees? Not really. The link is much weaker.
  • Analogy: If your left shoe is worn out, it doesn't mean your left sock is worn out. They are different items. The study found that having hip arthritis only slightly increases your risk of getting knee arthritis. This suggests they aren't just one big "leg disease," but two separate issues.

3. The "Weight" vs. "Height" Puzzle

This is where the "different houses" idea gets really clear. The study looked at what makes these joints break down.

• Knees are the "Heavy Load" Bearers: Being overweight is a massive problem for knees.
  • Analogy: Imagine the knee is a suspension bridge. If you put too much weight on the bridge (obesity), the cables snap. The study found that being heavy makes knee arthritis much more likely.
• Hips are the "Tall" Problem: Being tall is actually linked to more hip arthritis, but being tall seems to protect your knees.
  • Analogy: Imagine the hip is a tall tree. The taller the tree, the more wind it catches at the top, making it more likely to snap. But for the knees, being tall might mean better alignment, acting like a sturdy foundation.
• The Takeaway: Losing weight might save your knees, but it might not fix your hips. You need different strategies for different joints.

4. The "Pain" Connection

• Knees: Even a little bit of wear and tear (a small scratch) in the knee often causes pain.
• Hips: The hip is tougher. It can have a lot of wear and tear (a big scratch) before it starts to hurt.
• The Multi-Joint Rule: If you have arthritis in just one joint, you might be okay. But if you have it in multiple joints (both knees, or a knee and a hip), the pain gets much worse. It's like having a leak in one room of a house is annoying, but having leaks in the kitchen, bathroom, and bedroom at the same time is a disaster.

5. Who is at Risk?

• Deprivation: Surprisingly, in this specific group of people, living in a poorer area didn't strongly predict who got arthritis. (Though the study noted that poorer people often get less treatment, which is a different problem).
• Race: The study found that people of Asian descent had a higher risk of knee arthritis, but race didn't seem to change the risk for hip arthritis.

The Big Picture: What Should We Do?

The main message of this paper is: Stop treating all arthritis as the same thing.

• For Doctors: Don't just look at the knee; look at the whole person. If a patient has knee pain, check their weight. If they have hip pain, check their height and bone structure.
• For Patients: If you have arthritis in one knee, be careful with your weight to protect the other knee. But don't panic that your hips will automatically fail just because your knees are hurting.
• For the Future: We need "tailor-made" treatments. A weight-loss drug might be a miracle for knee pain, but it might not help hip pain. We need to build specific tools for specific joints.

In short: Your hips and knees are not a team; they are two different players on the same field. They have different strengths, different weaknesses, and they need different coaches to stay in the game.

Technical Summary

1. Problem Statement

Osteoarthritis (OA) is a leading cause of global disability, yet most epidemiological research treats lower-limb OA (hip and knee) as isolated, monoarticular conditions. Emerging evidence suggests OA is a systemic, multi-joint disease with distinct genetic and biological mechanisms across different anatomical sites. However, there is a lack of large-scale, population-level data characterizing:

• The interrelationships between hip and knee OA (bilateral vs. cross-joint).
• The specific associations between radiographic severity and joint pain.
• How risk factors (BMI, height, race, deprivation) differentially affect hip versus knee OA.
• The prevalence of multi-joint involvement in a contemporary cohort.

2. Methodology

Study Population:

• Cohort: UK Biobank imaging sub-study.
• Sample Size: 59,475 individuals (mean age 65.1 years; 52.8% female).
• Inclusion: Participants with bilateral hip and knee dual-energy X-ray absorptiometry (DXA) scans.

Data Acquisition & Imaging:

• Scans: High-resolution bilateral hip and knee DXA scans (Lunar iDXA scanners).
• Automated Analysis: Validated machine-learning algorithms were applied to delineate joint shapes and quantify OA features.
  • Features Measured: Osteophyte area (mm²) and minimum joint-space width.
  • Exclusions: Subchondral sclerosis and cysts were excluded due to low frequency on DXA.
• Grading System:
  • Osteophytes were graded 0–3 based on area.
  • Joint-space narrowing was graded 0–3.
  • Composite Score: Sum of osteophyte and narrowing grades.
  • Definition of Radiographic OA (rOA): Custom grade $\ge$ 2.
    • Hip: Grade 2 = sum 2–3; Grade 3 = sum 4–6; Grade 4 = sum 7–12.
    • Knee: Grade 2 = sum >1 to $\le$3; Grade 3 = sum >3 to $\le$4.5; Grade 4 = sum >4.5.

Variables & Statistical Analysis:

• Outcomes: Radiographic OA (rHOA, rKOA), self-reported chronic joint pain (>3 months), and hospital-diagnosed OA (ICD-10).
• Covariates: Age, sex, race (White vs. non-White), height, weight, BMI, and Townsend Deprivation Index (TDI).
• Models:
  • Logistic regression to assess associations between rOA grades and pain.
  • Logistic regression to examine bilateral (left-right) and cross-joint (hip-knee) interrelationships.
  • Models adjusted for demographics and body size metrics.
• Software: Stata version 18.

3. Key Contributions

• Scale: One of the largest studies to date utilizing automated DXA-derived radiographic measures for both hip and knee OA in a single cohort.
• Multi-Joint Focus: Shifts the paradigm from single-joint analysis to a holistic view of lower-limb OA, quantifying the burden of multi-joint disease.
• Joint-Specific Heterogeneity: Provides robust evidence that hip and knee OA are distinct phenotypes with different risk factor profiles and pain correlations, challenging the notion of a single "lower-limb OA" disease process.
• Validation of DXA: Confirms that DXA-derived structural measures correlate strongly with clinical symptoms and hospital diagnoses, validating their use as a surrogate for radiographic OA in large cohorts.

4. Key Results

Prevalence & Demographics:

• rHOA Prevalence: 6.9% (Right), 8.1% (Left).
• rKOA Prevalence: 6.3% (Right), 7.1% (Left).
• Bilateral Involvement: 16.8% of rHOA cases and 36.1% of rKOA cases were bilateral.
• Multi-Joint Involvement: Among those with rOA, 26.8% had two joints affected, 3.1% had three, and 0.5% had all four. Only 14.5% of two-joint cases involved non-cognate joints (one hip + one knee).
• Sex Differences: rKOA was twice as prevalent in females; rHOA was twice as prevalent in males.

Pain Associations:

• Severity Gradient: Strong positive association between rOA grade and chronic pain. Grade 4 rOA increased odds of pain by 11–16 times (hip) and 9–15 times (knee) compared to no OA.
• Thresholds: Minimal knee OA (Grade 1) was associated with a 2-fold increase in knee pain, whereas Grade 1 hip OA showed no significant pain association.
• Cumulative Effect: Each additional affected joint increased the odds of pain. Individuals with all four joints affected had 4.29 times higher odds of pain compared to those with no OA.
• Cross-Joint Pain: rHOA showed moderate association with knee pain; rKOA showed no association with hip pain.

Joint Interrelationships:

• Bilateral Symmetry: Very strong associations.
  • Knee: >26-fold increased odds of contralateral KOA if one knee is affected.
  • Hip: >5-fold increased odds of contralateral HOA.
• Cross-Joint (Hip-Knee): Weak but present associations (OR ~1.25–1.36), indicating that while systemic factors exist, joint-specific mechanisms dominate.
• Correlation: Strong correlation in joint-space width between bilateral knees ($r=0.73$) and hips ($r=0.66$), but weak correlation between hip and knee ($r \approx 0.27-0.30$).

Risk Factor Profiles (Joint-Specific):

• BMI/Obesity: Strongly associated with Knee OA (OR ~1.57 per SD; OR ~2.45 for obesity) but weakly associated with Hip OA (OR ~1.05–1.07).
• Height:
  • Hip OA: Positive association (Taller = Higher risk; OR ~1.33).
  • Knee OA: Inverse association (Taller = Lower risk; OR ~0.82).
• Deprivation & Race: No consistent association with deprivation. Asian race was a risk factor for Knee OA (OR ~1.65) but not Hip OA.

5. Significance and Implications

Clinical Practice:

• Tailored Management: The distinct risk profiles (e.g., BMI driving knee OA but not hip OA; height driving hip OA) suggest that prevention and treatment strategies must be joint-specific. For instance, weight loss interventions (GLP-1 agonists) may be highly effective for knee OA but less so for hip OA.
• Symptom Interpretation: Clinicians should recognize that early radiographic changes in the knee are more likely to be symptomatic than early changes in the hip.
• Multi-Joint Monitoring: Patients with OA in one joint should be monitored for bilateral progression, particularly in the same joint type, rather than assuming a generalized "wear and tear" affecting all joints equally.

Research & Policy:

• Phenotyping: Supports the genetic and biological hypothesis that hip and knee OA are distinct diseases with different pathomechanisms.
• Intervention Targets: Future trials should consider "multi-joint OA" as a specific phenotype requiring cumulative management strategies rather than isolated joint treatment.
• Equity: While deprivation did not strongly correlate with radiographic prevalence in this cohort (likely due to UK Biobank's selection bias), disparities in access to joint replacement remain a critical policy issue.

Limitations:

• Cross-Sectional Design: Cannot establish causality or temporal progression.
• Imaging Modality: DXA scans are non-weight-bearing, though they correlate well with clinical outcomes.
• Diversity: The UK Biobank cohort is predominantly White and less deprived than the general UK population, limiting generalizability to diverse ethnic and socioeconomic groups.