Knee osteoarthritis linked to higher osteoporosis risk in 406,810-person cohort

Knee osteoarthritis was associated with a higher risk of subsequently developing osteoporosis in a prospective study of 406,810 UK Biobank participants, according to researchers who followed the cohort for a median 13 years. Osteoporosis occurred in 5.2% of participants with knee osteoarthritis at baseline, compared with 3.9% of those without osteoarthritis. After adjustment for demographic, lifestyle, socioeconomic and clinical factors, the fully adjusted model estimated a 22% higher hazard of osteoporosis among people with knee osteoarthritis, with a hazard ratio of 1.22 and a 95% confidence interval of 1.16–1.29.

The findings point to an association rather than proving that knee osteoarthritis causes systemic bone loss. The researchers described knee osteoarthritis as an independent risk indicator for osteoporosis, but the observational design cannot establish a causal relationship. An independent clinical cohort of 65 participants also found higher concentrations of two circulating markers of bone resorption in people with knee osteoarthritis, supplying limited biochemical evidence consistent with the population result. The authors presented both components as a prospective epidemiological analysis with a small translational validation study, not a clinical trial of an intervention.

The UK Biobank analysis began with 502,143 volunteers recruited between 2006 and 2010. The researchers excluded 89,386 participants with osteoarthritis outside the knee, non-specified knee osteoarthritis or inflammatory or metabolic arthropathies, including rheumatoid arthritis and gout. They also excluded 5,947 people who had a recorded osteoporosis diagnosis before baseline, leaving 406,810 participants: 29,718 classified as having knee osteoarthritis and 377,092 without an osteoarthritis diagnosis. Knee osteoarthritis was identified from the ICD-10 code M17 in linked hospital inpatient records, while incident osteoporosis was defined by codes M80, M81 or M82. Participants without an osteoporosis event were censored at death, loss to follow-up or the administrative censoring date.

The groups differed substantially at baseline, a limitation that motivated extensive statistical adjustment. People with knee osteoarthritis had a median age of 61 years, compared with 58 in the reference group, and a higher median body mass index of 29.1 versus 26.6 kilograms per square metre. Sedentary behavior was reported by 24.5% of the osteoarthritis group and 20.3% of the comparison group. Hypertension was more common among participants with knee osteoarthritis, affecting 52.7% compared with 33.8%, and diabetes affected 14.4% rather than 9.0%. The osteoarthritis group also had a greater overall burden of comorbidity. The researchers accounted for age, sex, education, area-level deprivation measured by the Townsend Deprivation Index, smoking, alcohol use, physical activity, sedentary time, medication use, calcium or vitamin D supplementation, cardiometabolic and other comorbidities, prior malignancies and a polygenic risk score for estimated bone mineral density and fracture risk. Missing covariate data were addressed through multiple imputation by chained equations.

Time-to-event analysis used the interval from baseline assessment until the first recorded osteoporosis diagnosis, death, loss to follow-up or administrative censoring. Kaplan–Meier curves showed early separation between the groups and a progressively widening difference over time; a log-rank test yielded p<0.001. In unadjusted Cox regression, knee osteoarthritis was associated with an osteoporosis hazard ratio of 1.29, with a 95% confidence interval of 1.22–1.36. The estimate decreased to 1.22 after the most extensive adjustment, indicating that the association was not fully explained by the measured baseline differences. Because the study was not randomized and several potential pathways remained unmeasured, however, the adjusted estimate cannot eliminate residual confounding or prove that the joint disease itself produces the extra risk. The researchers evaluated proportional-hazards assumptions with Schoenfeld residuals and assessed multicollinearity before including covariates in the models.

Several measured characteristics were also associated with osteoporosis in the fully adjusted model. Advancing age, female sex, greater socioeconomic deprivation, current smoking, physical inactivity, sedentary behavior and a history of malignancies were associated with higher hazard, as were hypertension and diabetes. Greater grip strength, higher body mass index and being physically active were associated with lower hazard. The polygenic risk score for estimated bone mineral density and fracture risk was also associated with osteoporosis. A history of vitamin D supplementation was linked to higher hazard in the model, while the relationship with alcohol consumption differed between current and former drinkers compared with never-drinkers. These results show that the observed relationship between knee osteoarthritis and osteoporosis existed within a multifactorial risk profile rather than in isolation, while also underscoring the observational nature of the findings. They do not show that any single lifestyle or clinical factor mediates the association.

In the independent validation cohort, 35 patients with clinically diagnosed knee osteoarthritis were compared with 30 age- and sex-matched controls who had no joint pain or osteoarthritis history. The osteoarthritis group had radiographic Kellgren–Lawrence grades of at least 2 and symptoms lasting more than six months. People with metabolic bone disease, inflammatory arthritis, secondary osteoporosis, active malignancy, severe renal or hepatic insufficiency, recent use of systemic corticosteroids or bone-modulating treatment were excluded. Morning fasting blood samples were collected under standardized conditions, with all draws performed between 08:00 and 10:00 after an overnight fast of more than 10 hours. Serum was separated, stored at −80 °C and assayed in duplicate using enzyme-linked immunosorbent assay kits. The researchers measured serum beta-cross-linked C-telopeptide, or β-CTX, and tartrate-resistant acid phosphatase 5b, or TRACP-5b.

The knee osteoarthritis group had mean serum β-CTX of 0.512±0.118 nanograms per millilitre, compared with 0.334±0.082 in controls, and mean TRACP-5b of 3.85±0.78 units per litre, compared with 2.68±0.54. Both differences were reported as statistically significant at p<0.001. β-CTX was used as an indicator of type I collagen degradation during bone-matrix breakdown, while TRACP-5b was used as an indicator of osteoclast number and activity. The result is consistent with greater bone resorption activity in the osteoarthritis group, but the small biomarker comparison did not demonstrate that participants had osteoporosis or that the elevated markers predicted future fracture. It also did not establish whether the circulating changes preceded, accompanied or followed the participants’ joint disease.

Several biological mechanisms could connect a diseased knee joint with skeletal fragility elsewhere in the body, but the study did not directly test them. The authors cited chronic low-grade inflammation, with mediators such as interleukin-1 beta, interleukin-6 and tumor necrosis factor alpha, as potentially contributing to increased osteoclast formation and reduced osteoblast activity. Pain may also contribute indirectly: reduced mobility can diminish weight-bearing activity, weaken muscles and reduce mechanical stimulation of bone. Chronic pain may additionally alter autonomic and neuro-endocrine signaling, including sympathetic pathways linked to bone remodeling. Sensory neuropeptides such as calcitonin gene-related peptide and substance P were also discussed as possible participants in bone remodeling and inflammatory amplification. These remain potential explanations rather than demonstrated mediators, particularly because the registry analysis did not formally model psychological distress, pain or neuro-endocrine changes as longitudinal pathways.

The study has several important limitations beyond its observational nature. Diagnoses came from linked hospital inpatient records rather than universal dual-energy X-ray absorptiometry or imaging-based assessment of every participant. Such coding is likely to capture more clinically severe or symptomatic disease and may miss milder cases managed in primary care. UK Biobank volunteers may also be healthier and more health-conscious than the general population. Residual confounding cannot be ruled out despite extensive adjustment. Information on osteoarthritis severity, pain intensity and analgesic use was unavailable, preventing evaluation of whether the association varied with disease severity. Physical activity and sedentary behavior were self-reported at baseline and may not reflect lifestyle changes over the multi-year follow-up. The authors also noted that competing death can complicate estimates of absolute osteoporosis incidence, with standard cumulative-incidence curves potentially overestimating absolute probabilities in populations with higher death rates. The validation cohort was small and cross-sectional, so it provided biochemical support but did not resolve the causal and prognostic uncertainties of the large prospective analysis. The results should therefore be read as a strong population-level association accompanied by preliminary human biomarker evidence, not as proof that knee osteoarthritis directly causes osteoporosis.

Subject of Research: Association between knee osteoarthritis and subsequent osteoporosis risk in a large prospective cohort with a small biomarker validation cohort

Article Title: Bridging joint degeneration and systemic skeletal fragility: knee osteoarthritis as an independent risk factor for osteoporosis

Article References: Sun, W., Ma, J., He, J., Liu, Y., Zhang, W., & Zhang, L. (2026). Bridging joint degeneration and systemic skeletal fragility: knee osteoarthritis as an independent risk factor for osteoporosis. Frontiers in Public Health, 14, Article 1964979. https://doi.org/10.3389/fpubh.2026.1964979

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1964979

Keywords: knee osteoarthritis, osteoporosis, bone resorption, UK Biobank, prospective cohort, hazard ratio, β-CTX, TRACP-5b, skeletal fragility, musculoskeletal health, Bridging, joint