What the Research Says About Risk Factors for Knee Osteoarthritis

We only cite studies published in peer-reviewed journals. We summarize findings without overstating conclusions.

This article summarizes a systematic review and meta-analysis published on August 29, 2023, in the open-access Journal of Orthopaedic Surgery and Research (BioMed Central/Springer Nature), titled “Evidence on Risk Factors for Knee Osteoarthritis in Middle-Older Aged: A Systematic Review and Meta-Analysis.” The study was authored by Yawei Dong, Yan Yan, and Jun Zhou (Beijing University of Chinese Medicine, Beijing, China), Qiujun Zhou (Department of First Clinical Medical College, Zhejiang Chinese Medical University, and affiliated clinically with a hospital in the Chaoyang District of Beijing), and Hongyu Wei (corresponding author, Department of Orthopaedic Surgery, China-Japan Friendship Hospital, Beijing, China). The review was registered on PROSPERO under identifier CRD42022329710 and conducted in compliance with PRISMA reporting guidelines. It synthesized evidence from 29 observational studies – 17 case-control studies and 12 cohort studies – encompassing 60,354 participants from 11 countries, all focused specifically on adults aged 40 and older. The review examined 14 potential risk factors and protective factors for knee osteoarthritis (KOA), producing statistically significant pooled estimates for 7 of them. The full text is freely available at PubMed Central (PMC10464102).

The authors declare no competing interests. No external funding was reported for this research.

Background: Why Identifying Modifiable Risk Factors Matters

Knee osteoarthritis is one of the most prevalent chronic conditions worldwide. Estimates cited by the authors place approximately 32.5 million American adults as affected, with 14 percent of the U.S. population experiencing symptomatic knee OA between 2008 and 2014. Epidemiological surveys from Korea, the United Kingdom, Spain, and other regions find that the incidence of KOA in adults over age 40 has reached more than 15 percent in multiple study populations. With aging populations globally, these numbers are rising.

Current treatment for established knee OA is fundamentally symptomatic: analgesics, NSAIDs, corticosteroid injections, sodium hyaluronate (hyaluronic acid) injections, physical therapy, and traditional medicines can reduce pain and improve function to varying degrees, but none of these interventions reverses the underlying structural changes or stops disease progression. The trajectory for most patients with moderate-to-severe KOA ends at joint replacement surgery – a procedure that carries its own risks, costs, and recovery burden, and which may eventually require revision. Long-term symptomatic treatment imposes substantial financial costs on patients and healthcare systems alike, while the disease itself significantly diminishes quality of life.

Given that no disease-modifying treatment for knee OA exists, prevention – or at minimum, delay of onset – becomes a clinical priority of the first order. Prevention requires knowledge of which factors genuinely increase risk. Some KOA risk factors are non-modifiable: a person cannot change their age, their sex, or the fact that they sustained a knee injury in the past. But others are modifiable – weight, physical activity habits, and certain occupational exposures can all be influenced by individual behavior and public health policy. Identifying which modifiable factors carry the strongest evidence behind them, and how large their effects actually are, is the central purpose of this 2023 review.

The authors note that the most directly comparable predecessor review was published in 2015 (Silverwood et al., Osteoarthritis and Cartilage), which identified overweight, obesity, female sex, and previous knee injury as the main risk factors in adults aged 50 and older – but excluded case-control studies and was thus based on a narrower evidence base. The 2023 Dong et al. review extended eligibility to age 40 and above (capturing the earlier-onset population, particularly relevant for women, among whom KOA has been observed to begin approximately a decade earlier than in men), incorporated both cohort and case-control designs to maximize available evidence, and included literature from Chinese-language databases not captured by English-only searches. The result is the largest and most geographically diverse meta-analysis of KOA risk factors published to that point.

Why the Starting Age of 40 Matters

A key methodological decision in this review is the choice to include studies with mean participant ages of 40 or older, rather than limiting to age 50-plus as prior reviews had done. The authors justify this explicitly: clinical observation and epidemiological data suggest that knee OA onset in women tends to precede onset in men by approximately one decade, making 40 a more appropriate lower bound for capturing the full risk-factor landscape in both sexes. Setting the threshold at 50 would systematically exclude the earlier-onset female population and potentially miss risk factor associations that are most evident in that group.

The authors also note that a peak in KOA incidence occurs around age 50 in women, particularly in the context of obesity. A Japanese cohort study included in the review found that early KOA prevalence was highest among middle-aged women with obesity, confirming that the 40-to-50 age decade is not a dormant pre-disease period but an active risk window where preventive action would have the greatest impact.

Methodology: How the Review Was Conducted

Search Strategy and Study Types

Seven databases were searched from their inception through July 2023: PubMed, Web of Science, Ovid Technologies, the China National Knowledge Infrastructure (CNKI), the Chinese Science and Technology Periodical Database (VIP), the Wanfang Database, and SinoMed. Search terms combined MeSH-standardized terminology for knee osteoarthritis with terms for all potential risk factors of interest. The initial search returned 3,597 records. After title and abstract screening, 16 studies underwent full-text review, and 29 ultimately met all inclusion criteria. Two reviewers independently screened all records, with a third reviewer resolving disagreements.

Eligible study designs were cohort studies and case-control studies – the two observational designs best suited to evaluating risk factors for chronic diseases. Unlike randomized controlled trials, which test interventions, these observational designs follow real-world populations over time (cohort) or compare people with and without a disease and look back at their exposure histories (case-control). Randomized trials of risk factors for chronic disease are largely impossible to conduct (you cannot randomly assign people to be obese or to sustain a knee injury), making high-quality observational evidence the best available data.

Included studies had to use either the American College of Rheumatology (ACR) clinical criteria for knee OA diagnosis (in case-control studies) or the Kellgren-Lawrence radiographic grading system (in cohort studies) – the two internationally recognized, validated diagnostic standards. This requirement excluded studies with vague or non-standard case definitions, improving diagnostic homogeneity across the included literature.

Quality Assessment

Study quality was assessed using two standard tools. Case-control and cohort studies were evaluated with the Newcastle-Ottawa Quality Assessment Scale (NOS), which awards “stars” across three domains: selection of study groups, comparability of groups, and quality of outcome assessment. The mean NOS score across the 17 included case-control studies was 5.7 out of 9, and across the 12 cohort studies was 7 out of 9 – indicating moderate to good quality overall, with the cohort studies rated somewhat more favorably. The most common methodological shortcomings were inadequate selection of controls (in case-control studies) and incomplete reporting of non-response rates. A GRADE assessment (Grading of Recommendations Assessment, Development and Evaluation) was also performed to rate the overall quality of the evidence for each risk factor.

Statistical Analysis

Meta-analysis was performed using RevMan 5.3. Adjusted odds ratios (ORs) and their 95% confidence intervals were extracted from each study; when both unadjusted and multivariable-adjusted estimates were available, the adjusted estimates were used to minimize confounding. The I² statistic measured heterogeneity, with values exceeding 50% indicating moderate-to-high between-study variation. Given the mixed study design (cohort and case-control), the authors applied a random-effects model throughout – including when I² fell below 50% – reflecting appropriately conservative uncertainty about whether the same underlying effect was being measured across study types. For risk factors with high heterogeneity, sensitivity analyses were conducted by sequentially excluding individual studies to identify potential outliers driving the variation. Funnel plots were constructed to assess publication bias. Subgroup analyses compared results separately in cohort studies versus case-control studies to test consistency across designs.

The Included Studies: Scope and Geography

The 29 included studies were published between 2005 and 2023 and were conducted across 11 countries including China, Japan, Finland, Germany, the United Kingdom, Lebanon, Morocco, Iran, Pakistan, Armenia, and the Netherlands. This geographic breadth is a notable strength compared to prior reviews, which often drew primarily from Western European or American populations. The inclusion of multiple Chinese-language studies from mainland China provides data from populations with different occupational patterns, dietary backgrounds, BMI distributions, and healthcare contexts – making the findings more generalizable globally.

Of the 60,354 total participants, the 17 case-control studies and 12 cohort studies examined the following 14 candidate risk or protective factors: trauma history in the knee, BMI at or above 24 kg/m², female sex, age, exercise, high school education, university education, junior high school education, living environment (damp, cold, or dark conditions), agricultural labor, manual labor, family history, drinking history, and smoking history.

Results: The Seven Statistically Significant Findings

Risk Factor 1: Knee Trauma History

Six studies examined prior knee injury as a risk factor for KOA onset. The pooled odds ratio was 1.37 (95% CI: 1.03–1.82; p = 0.030), indicating that individuals with a history of knee trauma have 37 percent higher odds of developing knee OA compared to those without such a history. Notably, heterogeneity was zero (I² = 0%), meaning the six contributing studies were entirely consistent with one another in their estimates – the most statistically clean result in the entire meta-analysis.

Subgroup analyses revealed an important nuance, however: when cohort studies were analyzed separately from case-control studies, the trauma history association did not reach statistical significance in either design group alone. This paradox – significant in the combined analysis, not significant in either subgroup – can arise when the combined dataset has power that the individual subgroups lack, but it also means the estimate should be interpreted with some caution. The underlying biology, however, is unambiguous: knee injury – anterior cruciate ligament (ACL) tears, meniscal damage, fractures involving the joint surface – initiates a cascade of cartilage inflammation, altered biomechanics, and disrupted joint stability that accelerates the degenerative process. Post-traumatic OA is a well-established clinical entity that develops years to decades after acute joint injury even when the original injury appeared to heal satisfactorily.

The practical implication is straightforward: protecting knee joints from injury – through appropriate protective equipment in contact sports, strength training to stabilize the joint, and prompt treatment of acute injuries – is a genuine preventive strategy for KOA, not merely common sense.

Risk Factor 2: Elevated BMI (≥ 24 kg/m²)

Eleven studies examined the relationship between BMI and KOA risk, the largest evidence base for any factor in this review. The pooled odds ratio was 1.30 (95% CI: 1.09–1.56; p = 0.004) for BMI at or above 24 kg/m², with zero heterogeneity (I² = 0%). This is a very clean and consistent finding: across all 11 contributing studies from multiple countries, elevated BMI was associated with a 30 percent higher odds of developing knee OA compared to those with lower BMI.

It is important to note the BMI threshold used. The authors applied the Asian-Pacific BMI classification standard, under which BMI ≥ 24 kg/m² is defined as overweight, rather than the Western standard of ≥ 25 kg/m². This reflects the demographic of many included studies (particularly the large Chinese cohort) and means the threshold is slightly lower than what Western readers may be accustomed to. The broad category includes overweight individuals alongside those with full obesity, making the effect size of 1.30 a conservative estimate of what a more severely obese population might show.

Subgroup analysis confirmed this finding specifically in cohort studies (OR 1.29; 95% CI: 1.02–1.63; p = 0.030), lending it the more robust prospective design’s support. The association did not reach significance in case-control studies alone, likely due to smaller sample sizes in that subgroup rather than a genuine difference in the underlying relationship.

The mechanisms connecting excess body weight to knee OA are multiple. The most direct is mechanical: the knee joint bears approximately four to six times body weight during walking, meaning that every kilogram of excess body mass translates to multiple additional kilograms of force through the joint with every step. Over years and decades, this added mechanical stress accelerates cartilage wear, subchondral bone remodeling, and meniscal degeneration. But the mechanism is not purely mechanical: adipose tissue is metabolically active, secreting inflammatory adipokines (including leptin, adiponectin, resistin, and visfatin) that enter systemic circulation and contribute to the low-grade systemic inflammation that drives cartilage degradation in OA. The inflammatory contribution of excess fat is supported by the observation that OA also affects non-weight-bearing joints (such as the hands) in obese individuals at higher rates than in lean individuals – a distribution pattern that cannot be explained by mechanical loading alone.

A Japanese prospective cohort study included in the review (Ito et al.) found that maintaining a lighter body weight over a 10-year period reduced the risk of KOA-related knee pain by 27.5% – providing direct evidence that sustained weight management is protective, and that the protective benefit of weight loss is not merely theoretical.

Risk Factor 3: Female Sex

Seventeen studies – the most for any single factor in this review – examined sex as a determinant of KOA risk. The pooled odds ratio was 1.04 (95% CI: 1.00–1.09; p = 0.030; I² = 20%), indicating that women have approximately 4 percent higher odds of developing knee OA than men in this population. The confidence interval barely includes 1.00 at its lower boundary, meaning the result is statistically significant but the magnitude of the sex difference within this specific age group (40 and older) is modest.

As with trauma history, the subgroup analysis produced a nuanced picture: female sex did not independently reach statistical significance as a KOA risk factor in either the cohort study subgroup or the case-control subgroup when analyzed separately. This suggests the pooled result may partly reflect the larger statistical power of combining both designs. The real-world significance of sex as a KOA risk factor in the specific 40-and-older age group requires cautious interpretation.

The broader literature on sex differences in OA is more definitive: women have a higher lifetime prevalence and greater severity of knee OA than men, a disparity that accelerates markedly after menopause. The leading hypotheses for this sex difference include estrogen’s role in maintaining cartilage integrity (declining estrogen after menopause removes this protective effect), sex differences in knee alignment and quadriceps muscle function, and the interaction between female sex and obesity in risk amplification. Research cited by the authors shows that early KOA onset peaks in middle-aged women around age 50, and that the risk elevation is most pronounced when female sex combines with obesity. The independent effect of sex in the 40-plus population, controlling for age and BMI, may indeed be smaller than the raw unadjusted sex difference observed across all ages.

Risk Factor 4: Age ≥ 40

Eleven studies contributed to the meta-analysis of age as a continuous risk factor. The pooled odds ratio was 1.02 (95% CI: 1.01–1.03; p = 0.007; I² = 37%), meaning that for each additional year of age beyond 40, the odds of KOA increase by approximately 2 percent. This may sound small, but the cumulative effect over decades is substantial: someone at age 70 compared to age 40 would have had 30 years of additional 2-percent annual risk compounding – corresponding to a very significant elevation in absolute risk over that timespan.

The subgroup analysis confirmed this specifically in cohort studies, where the age effect was larger and more precise: OR 1.04 (95% CI: 1.02–1.06; p < 0.001) – a 4 percent per-year increase per additional year of age. In case-control studies, age did not reach significance independently, likely because case-control designs are less well-suited to capturing gradual continuous-exposure effects than the longitudinal cohort design.

The biological basis for age-related OA risk is well-established. Aging affects virtually every tissue in the joint: chondrocyte function declines, the capacity for cartilage matrix repair diminishes, mitochondrial dysfunction increases oxidative stress in joint cells, inflammatory signaling pathways become more readily activated (a phenomenon called inflammaging), and the neuromuscular control of joint loading becomes less precise. Epigenetic changes accumulate that shift chondrocyte gene expression toward a more catabolic, pro-inflammatory state. These are not simply mechanical wear effects – they reflect fundamental biological deterioration of the joint’s ability to maintain itself.

Age and sex are non-modifiable risk factors, and the authors address this directly. Their argument is that while these factors cannot be changed, knowing that you are in a higher-risk category – middle-aged, female, approaching or past age 50 – should heighten attention to the modifiable risk factors. In practice, this means that a 48-year-old woman who is overweight and does not exercise regularly faces a compounded risk from multiple factors simultaneously, and that addressing the modifiable components (weight and physical activity) at this stage would have greater impact than waiting until symptoms develop.

Protective Factor 1: Regular Exercise

Four studies (contributing five data points, as one study analyzed professional and low-intensity exercise separately) examined exercise as a protective factor against KOA. The pooled odds ratio was 0.75 (95% CI: 0.62–0.91; p = 0.003), meaning that those who engaged in regular exercise had 25 percent lower odds of developing knee OA compared to those who did not.

A clinically important within-study finding deserves emphasis: one of the four studies (Xiang Xiangsong et al.) distinguished between low-intensity exercise and professional-level or high-intensity exercise. Low-intensity exercise was a significant protective factor (OR = 0.72; 95% CI: 0.58–0.89), whereas professional exercise (high-intensity, competitive, or occupationally demanding physical activity) was not protective and may carry its own joint-stress risks. This finding is entirely consistent with the broader OA and exercise literature: moderate, appropriately-dosed physical activity is protective, while excessive or high-impact loading without adequate recovery is itself a risk factor for joint damage.

How does exercise protect the knee? Multiple mechanisms converge. Regular moderate exercise maintains muscle strength around the joint – particularly the quadriceps – which absorbs loading forces that would otherwise pass directly through the articular cartilage and subchondral bone. Stronger periarticular musculature provides better joint stabilization, reducing the abnormal motion and impact forces that initiate cartilage damage. Exercise also promotes cartilage health at the biological level: the intermittent compression and decompression of cyclic loading is the primary mechanism by which nutrients and oxygen are distributed through articular cartilage, which has no blood supply of its own. Regular activity helps maintain normal cartilage metabolism. Exercise additionally reduces adiposity and systemic inflammatory markers, addressing the BMI and inflammaging mechanisms simultaneously. The 2019 guidelines from both the American College of Rheumatology (ACR) and the Osteoarthritis Research Society International (OARSI) recommend exercise as a core element of KOA prevention and management, strongly supported by this meta-analytic evidence.

Protective Factors 2 and 3: Higher Educational Attainment

Two case-control studies examined educational attainment as a determinant of KOA risk, with results stratified by level: junior high school education, high school education, and university education. High school completion was associated with a nearly 50 percent reduction in KOA odds compared to lower education levels (OR 0.49; 95% CI: 0.30–0.79; p = 0.003). University education was associated with an 80 percent reduction (OR 0.22; 95% CI: 0.06–0.86; p = 0.030). Neither junior high school education alone nor the individual subgroup analyses reached statistical significance.

These are striking effect sizes – among the largest in this entire meta-analysis – but they rest on only two contributing studies, which limits the reliability of the estimates substantially. The confidence intervals are wide (particularly for university education: 0.06–0.86 spans more than a tenfold range), reflecting the small evidence base. These findings should be treated as hypothesis-generating rather than conclusive.

That said, the direction and plausibility of this association are well-grounded. Multiple pathways could explain a protective effect of education on KOA risk. Higher educational attainment is associated with reduced engagement in manual and agricultural labor – occupations involving repetitive kneeling, squatting, heavy lifting, and prolonged standing that place chronic mechanical stress on the knee joint. Education is also associated with higher health literacy – better understanding of the importance of weight management, exercise, and injury avoidance. It correlates with higher socioeconomic status, which provides better access to healthy food, safe recreational spaces, and preventive healthcare. One of the contributing studies found that higher education predicted a lower risk of KOA specifically through its association with reduced physical labor demanding high knee joint loads. The authors draw this conclusion explicitly in their discussion: “higher education is associated with a lower risk of KOA, possibly due to the reduced physical labor and lower risk of knee joint damage associated with higher education.”

The Seven Non-Significant Factors

Of the 14 factors examined in meta-analysis, 7 did not achieve statistical significance. These were: living environment (damp, cold, or dark conditions), agricultural labor, manual labor, family history of OA, drinking history, smoking history, and junior high school education. The authors do not claim these factors are definitely unrelated to KOA risk – only that the current pooled evidence from available studies was insufficient to produce a statistically significant estimate for each of them.

Several of these non-significant factors have biological plausibility worth noting. Family history of OA carries a genuine genetic component – genome-wide association studies have identified multiple genetic loci associated with OA susceptibility – but the included studies may have been too few, or their family history definitions too heterogeneous, to detect a consistent signal. Manual and agricultural labor involve prolonged kneeling, squatting, and heavy mechanical loading that are biologically plausible risk factors, and some individual studies do find elevated KOA rates in these occupational groups. The meta-analytic non-significance may reflect variation in how these exposures were defined, how thoroughly they were documented, and what confounders (particularly BMI) were controlled for across the contributing studies.

Smoking’s relationship with OA is biologically complex and contested in the literature. Some studies find reduced OA risk in smokers (possibly because nicotine has anabolic effects on cartilage or because smoking is associated with lower body weight in some populations), while others find increased risk (through inflammatory and microvascular mechanisms). The null finding here is consistent with this mixed literature. Alcohol’s relationship with OA is similarly unresolved.

The Limits of Observational Evidence: What This Review Cannot Establish

The study’s authors are appropriately measured in acknowledging its limitations, which reflect the inherent constraints of observational epidemiology.

The fundamental limitation of all observational evidence is that association does not equal causation. A risk factor that consistently correlates with KOA development across many studies is most likely causal – but could also be a marker of another underlying cause, or could be confounded by unmeasured variables that drive both the risk factor and the disease. The authors used adjusted odds ratios from multivariable analyses to reduce this concern, but residual confounding is impossible to fully eliminate from observational data. The exception is age, where the association with OA is so biologically supported and consistent across contexts that causality is not seriously disputed.

The evidence base for several factors – particularly education and exercise – rests on a small number of contributing studies (two for education, four for exercise). The confidence intervals for these estimates reflect this uncertainty, and large shifts in the pooled estimate are possible as additional studies accumulate. The education finding especially – with an OR of 0.22 for university education based on two studies – should be treated with much more caution than, say, the BMI finding with its 11 contributing studies and I² of zero.

The meta-analysis also combines cohort studies and case-control studies, which have different strengths and vulnerabilities. Cohort studies follow participants forward in time, allow direct measurement of incidence, and are less susceptible to recall bias (people reporting exposures from memory after a disease diagnosis). Case-control studies are more efficient for studying rare diseases but are vulnerable to recall bias, where patients with KOA may remember or report risk factor exposures differently than healthy controls. The finding that several factors that were significant in the combined analysis were not significant in either design subgroup alone is a genuine methodological concern that the authors acknowledge.

The review examined only whether factors were associated with KOA onset – not their impact on KOA progression, severity, or response to treatment. A factor that is risk-neutral for onset may still matter for how quickly the disease advances once established. Conversely, a factor that reduces incidence may not modify disease course in those who develop OA anyway.

Studies were heterogeneous in their definitions and measurement of risk factors – particularly for exercise (which ranged from self-reported activity levels to structured questionnaires), living environment, and occupational exposures. Standardization of exposure definitions would improve future meta-analyses substantially.

Practical Implications: What People Can Actually Do

The authors frame their conclusions explicitly in terms of what middle-aged and older individuals can do to reduce their personal risk of developing knee OA. The modifiable factors with the strongest evidence – and therefore the most actionable prevention messages – are body weight and physical activity.

Maintaining a healthy body weight through diet and exercise is the single intervention supported by the most studies (eleven for BMI), the most consistent evidence (I² = 0%), and the best understanding of biological mechanism. The 27.5 percent risk reduction associated with sustained lighter body weight over 10 years, from the Ito et al. cohort study, illustrates the magnitude of the achievable benefit. For individuals already managing OA, weight reduction is also recommended by current clinical guidelines as one of the most effective symptom-management strategies available.

Regular, moderate-intensity exercise – of the kind that builds and maintains muscle strength around the knee, promotes healthy cartilage nutrition, and manages body weight – is the second most robustly supported modifiable protective factor (25 percent lower odds of KOA in exercisers, from four studies). Both the ACR and OARSI guidelines endorse exercise as a core recommendation for both prevention and management of knee OA, and this meta-analysis adds prospective cohort-supported evidence for its preventive role specifically in the 40-and-older population.

The warning from the exercise analysis is also practically important: not all exercise is equally protective. Low-to-moderate intensity activity (walking, swimming, cycling, gentle strength training) appears protective; high-intensity, high-impact activity without adequate recovery may increase joint stress, particularly in people who are already overweight or who have prior joint injuries. The clinical message is that appropriate exercise is beneficial, not that maximal exercise is best.

The non-modifiable factors – age and sex – cannot be changed, but they identify who should be most attentive to modifiable factors. Women approaching and passing through menopause, and all adults as they enter and progress through their 40s and 50s, have accumulating biological susceptibility that makes weight management and active lifestyle choices increasingly important rather than less.

Summary of Key Takeaways

• This 2023 systematic review and meta-analysis is the largest dedicated synthesis of knee OA risk factors in adults aged 40 and older, drawing on 29 studies (17 case-control, 12 cohort) and 60,354 participants from 11 countries, published between 2005 and 2023.
• Meta-analysis was performed for 14 candidate risk and protective factors. Seven achieved statistical significance (p < 0.05). The four statistically confirmed risk factors were: history of knee trauma (OR 1.37; 95% CI: 1.03–1.82), elevated BMI ≥ 24 kg/m² (OR 1.30; 95% CI: 1.09–1.56), female sex (OR 1.04; 95% CI: 1.00–1.09), and each additional year of age from 40 onward (OR 1.02 per year; 95% CI: 1.01–1.03).
• The BMI finding was supported by the most studies (eleven), had the cleanest statistical picture (I² = 0%), and was confirmed in cohort study subgroup analysis – making it the most robustly established modifiable risk factor. A 30 percent elevation in KOA odds per unit of BMI elevation above 24 kg/m² aligns with multiple prior meta-analyses, and a 10-year sustained lighter body weight was associated with a 27.5 percent reduction in knee pain risk in one contributing cohort.
• Three statistically significant protective factors were identified: regular exercise (OR 0.75; 95% CI: 0.62–0.91; 25 percent lower odds of KOA), high school education (OR 0.49; 95% CI: 0.30–0.79; 51 percent lower odds), and university education (OR 0.22; 95% CI: 0.06–0.86; 78 percent lower odds). The education findings rest on only two studies with wide confidence intervals and should be interpreted cautiously; the exercise finding is based on four studies and is consistent with broad clinical guideline support for exercise as a preventive measure.
• Within the exercise analysis, low-intensity exercise was specifically identified as protective (OR 0.72; 95% CI: 0.58–0.89), while professional or high-intensity exercise was not. This is clinically important: moderate, appropriate physical activity is recommended; excessive loading is not.
• Seven of the fourteen examined factors – living environment, agricultural labor, manual labor, family history, smoking, drinking, and junior high school education – did not produce statistically significant pooled estimates. Non-significance does not mean these factors are unrelated to KOA; it means current available evidence was insufficient to confirm an association, and several remain plausible candidates for future research.
• Subgroup analyses revealed an important methodological caveat: several factors (trauma history, female sex, and age) that reached significance in the combined cohort-plus-case-control analysis did not reach significance when cohort and case-control designs were analyzed separately. This suggests some results may benefit from interpretation as directional signals that require larger design-specific evidence bases to confirm definitively.
• The study’s primary practical message for the general public is that two modifiable factors – maintaining a healthy body weight and engaging in regular, appropriately intense exercise – have the strongest combined evidence base for reducing knee OA risk in middle-aged and older adults, and that both are also recommended by current ACR, OARSI, and other major international clinical guidelines for exactly this purpose.

Dong, Yawei, Yan Yan, Jun Zhou, Qiujun Zhou, and Hongyu Wei. “Evidence on Risk Factors for Knee Osteoarthritis in Middle-Older Aged: A Systematic Review and Meta-Analysis.” Journal of Orthopaedic Surgery and Research, vol. 18, 2023, article 634. https://doi.org/10.1186/s13018-023-04089-6. Full text available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC10464102/.