Lower body mass index, lower spine bone mineral density, lower levels of the bone formation marker procollagen type 1 N-terminal propeptide, no history of falls, and no history of cancer were among the factors associated with a lower risk of incident hip or knee replacement in a population-based Australian cohort study. Several of these associations, however, did not hold across all sensitivity analyses, and the findings as a whole require careful interpretation — a point the researchers themselves emphasize.
"In interpreting these findings, it is essential to recognize that [joint replacement] is not solely a biological endpoint but also reflects healthcare access, patient preference, and clinical decision-making," wrote corresponding author Aminu Suleiman, PhD, of the School of Medicine at Deakin University, Victoria, Australia, and colleagues. "This underscores the need for caution when linking observed associations to disease mechanisms alone."
The study, published in Arthroplasty, included 2,882 adults from the Geelong Osteoporosis Study who had no prior hip or knee replacement at baseline. Participants were followed for a median of 16.7 years. During follow-up, 223 participants underwent a first hip or knee replacement, including 131 hip replacements and 92 knee replacements. Women accounted for 61% of the joint replacements.
Joint replacement was identified through the Barwon Joint Registry, Medicare records, medical records, self-report, and dual-energy x-ray absorptiometry images when available. Only first incident hip or knee replacements were included; revision procedures were excluded. The researchers could not determine whether each surgery was performed for osteoarthritis, fracture, or another indication.
Researchers used time-dependent Cox proportional hazards models, with age as the primary time scale and sex as a stratification variable. Covariates were updated at scheduled follow-up assessments and carried forward until the next assessment. Women contributed data from baseline and 10- and 15-year follow-up visits, whereas men contributed data from baseline and 5- and 15-year follow-up visits.
Key Findings
In the main adjusted model, each 1 kg/m² decrease in BMI was associated with a 4% lower risk of joint replacement in the pooled analysis. Lower spine T-score was associated with a 16% lower risk, and lower procollagen type 1 N-terminal propeptide (P1NP) level was associated with a 31% lower risk. Participants with no history of falls were 26% less likely to undergo joint replacement, and those without a history of cancer were 34% less likely.
Several of these main adjusted findings require qualification before clinical translation, as detailed in the sections below. In particular, the cancer and dietary calcium associations did not persist across all sensitivity analyses, the spine BMD finding is counterintuitive in direction, and the BMI association differed by age group.
A note on BMI and age: The direction of the BMI association differed meaningfully by age group. Among participants aged 40 to 59 years, lower BMI was associated with reduced joint replacement risk — consistent with the role of mechanical load in osteoarthritis progression. Among those aged 60 years or older, the association reversed: lower BMI was associated with increased risk. The researchers suggested this older-adult pattern may reflect frailty, sarcopenia, or unintentional weight loss due to comorbidities rather than protection from mechanical loading. Both estimates were borderline significant, and the findings are exploratory, but the age-dependent divergence is clinically relevant.
On the spine BMD finding: The association between lower spine bone mineral density and reduced joint replacement risk is counterintuitive and should not be interpreted as biologically protective. The authors explain that higher spine BMD in this context likely reflects degenerative structural changes — osteophytes, subchondral sclerosis, and other hallmarks of established osteoarthritis — that can artificially elevate DXA-derived measurements. Lower spine BMD at baseline may therefore indicate fewer pre-existing degenerative changes rather than an independent protective mechanism.
On P1NP: Although P1NP is primarily a marker of bone formation, elevated concentrations indicate increased bone turnover. The authors suggest that elevated P1NP may also reflect active subchondral bone remodeling in early-to-mid-stage osteoarthritis — a process involving increased turnover and sclerosis that precedes cartilage degeneration and may contribute to disease progression. The P1NP association was directionally consistent across analytical approaches, though it was attenuated to borderline significance when osteoarthritis was added to the multivariable model, which the researchers attributed to shared variance between OA progression and bone remodeling activity. The authors suggest P1NP warrants further study as a possible biomarker, and note the association may reflect not only disease biology but also healthcare access and clinical decision-making.
Findings That Did Not Hold Across All Analyses
Cancer: The cancer finding was not robust in all analyses. In a Fine-Gray competing-risk model accounting for death as a competing event, history of cancer was no longer statistically associated with joint replacement. The authors suggest this attenuation reflects the higher competing mortality among participants with cancer, which reduces their probability of undergoing elective joint replacement, rather than any clear biological relationship.
Dietary calcium: Lower dietary calcium intake was associated with lower joint replacement risk in the main analysis, but the association was attenuated and no longer statistically significant after researchers adjusted for calcium supplement use. The researchers attributed this pattern to reverse causality or confounding by indication: participants with early musculoskeletal concerns may have been more likely to increase calcium intake or use supplements in response to clinical guidance. The adjusted sensitivity analysis did not support an independent association between lower dietary calcium intake and reduced joint replacement risk after accounting for supplement behavior.
Falls: Although non-fallers had a 26% lower risk of joint replacement, the study explicitly notes that this association "should be interpreted as observational rather than as evidence of a modifiable pathway." The relationship between falls and osteoarthritis is likely bidirectional — while falls can contribute to joint damage and subsequent joint replacement, individuals with osteoarthritis may also be more prone to falls due to pain, stiffness, and impaired mobility.
Socioeconomic Status
In a finding the researchers attribute to healthcare access rather than disease burden, participants in the middle and high socioeconomic groups had higher joint replacement rates than those in the lowest group. In the main model, the association reached statistical significance for the middle socioeconomic group; for the highest group, the association was in the same direction but did not reach significance in the primary model.
In a secondary analysis evaluating sex-by-socioeconomic status interaction, both middle and high socioeconomic status were associated with higher joint replacement risk, with a stronger association among men than women.
Absolute joint replacement rates ranged from 2.8 procedures per 1,000 person-years among men with low socioeconomic status to 6.0 procedures per 1,000 person-years among women with middle socioeconomic status.
The researchers attribute this gradient to differential healthcare access: data from the 2025 Australian Orthopaedic Association National Joint Replacement Registry, cited in the paper, indicate that 60 to 70% of joint replacements in Australia are performed in private hospitals, disproportionately among higher socioeconomic status individuals. The authors argue this implies significant unmet surgical need among lower socioeconomic status populations, where disease prevalence may be similar but access to surgery is limited.
Hip vs. Knee: Distinct Pathways
Exploratory analyses suggested that risk factors may differ by joint site, though these are subgroup findings with reduced statistical power and should be considered hypothesis-generating rather than definitive — particularly because joint replacement indications were unavailable.
Higher BMI was more strongly associated with knee replacement than hip replacement, consistent with the predominance of mechanical loading in knee osteoarthritis. By contrast, P1NP, prior fracture, and falls appeared more relevant to hip replacement, suggesting that systemic and bone-related factors may play a larger role in hip disease. The researchers describe this differential as plausible but not definitive given the outcome classification limitations.
Limitations
The researchers noted several limitations. The study could not classify joint replacements by indication or laterality, and occupational exposures, prior joint injuries, and joint pain severity were unavailable. Lifestyle factors and some comorbidities were self-reported. Falls were reported within the past year, but the temporal relationship between falls and osteoarthritis onset could not be established, limiting causal interpretation. The number of joint replacement events was relatively small, particularly in subgroup analyses, which may have limited statistical power and helps explain why some associations significant in the pooled model appeared only borderline in stratified analyses. BMI did not distinguish fat mass from lean mass. The model was intended to identify population-level associations rather than to predict individual surgical risk. Because the cohort was based in southeastern Australia, findings may not generalize to populations with different demographics or healthcare systems.
"The findings are exploratory and warrant further investigation before drawing practical implications," the researchers wrote. "They also highlight the importance of considering joint site and age when interpreting risk factors and designing preventive strategies for [joint replacement]."
The Geelong Osteoporosis Study was funded by the Victorian Health Promotion Foundation and the National Health and Medical Research Council of Australia. Dr. Suleiman was supported by a Deakin University Postgraduate Research Scholarship, and Lana Williams was supported by a National Health and Medical Research Council Emerging Leader Fellowship. The researchers reported no competing interests.
Source: Arthroplasty