Prevalence of Knee Radiographic Symptomatic Osteoarthritis and Ultrasound Features Across Sex and Body Mass Index Groups: The Johnston County Health Study

Participants and Methods

The Johnston County Health Study (JoCoHS) is a prospective, community-based study that enrolled 902 adults residing in Johnston County, North Carolina, between 2019 and 2024. Participants were randomly recruited from household listings, regardless of OA status or symptoms. Eligible residents (ages 35–70, fluent in English or Spanish, and identifying as Black/African American, White, or Hispanic) provided informed consent and completed a single ~3-hour research clinic visit. No symptomatic, structural, or diagnostic criteria were required for inclusion in this population-based cohort. The JoCoHS is similar, but distinct from the prior Johnston County Osteoarthritis Project (JoCoOA); both enrolled community participants using a population-based survey strategy, but the earlier JoCoOA (1991–2018) included only Black or White men and women aged 45 years and older.¹¹

Baseline assessments included questionnaires collecting data on demographics, comorbidities, clinical features and symptoms of KOA, along with knee US and radiographs of the knee using standardized acquisition and scoring, in all participants. Knee radiographs were obtained and graded for KLG by an experienced musculoskeletal radiologist with high reliability (inter-rater reliability κ=0.86; intra-rater κ=0.89).¹² Standardized US images of each knee were also collected by a trained sonographer following a published protocol.⁹ All images were scored by two expert readers following a published atlas, with adjudication and final scoring by a third reader in cases of disagreement.⁹ More detailed methodology is detailed in our prior publication.¹⁰ The study was approved and monitored by the University of North Carolina at Chapel Hill Institutional Review Board (#18–0438).

Results

Unweighted participant characteristics among 902 participants aged 35–74 years (mean [SD] 55.0 [9.6] years), indicated that most were female (67.2%) and non-Hispanic white (65.5%), with 24.1% identifying as non-Hispanic Black and 10.4% as Hispanic (Table 1). Compared with the 2020 Johnston County population, younger adults (aged 35–44 years) and males were underrepresented, whereas older adults and females were overrepresented. The mean BMI was 33.3 ± 8.3 kg/m², 42.4% had obesity and 19.7% had severe obesity. A history of knee injury was reported by 27.9% of participants (Table 1). Of 902 participants, 861 individuals had US imaging (images were missing due to bilateral TKR [n=39, the contralateral knee was included for those with unilateral TKR], difficulty positioning/image quality [n=2]). No major demographic differences were noted between the group with US imaging and the overall group.

Weighted prevalence estimates of symptom and radiographic outcomes by sex and BMI categories are shown in Table 2. The overall weighted prevalence of Ksx, rKOA, severe rKOA, and sxKOA, were 61%, 36%, 21%, and 27%, respectively, and were generally higher among females and participants with overweight/obesity. Overall, 62.1% of females and 58.9% of males reported knee symptoms. rKOA was present in 37.1% of females and 35.2% of males, severe rKOA in 22.7% and 19.9%, and sxKOA in 27.8% and 26.0%, respectively (Table 2).

Across both sexes, numerically higher prevalence of Ksx, rKOA, severe rKOA, and sxKOA was observed with higher BMI categories (Table 2). Among females, the prevalence of sxKOA was similar among those who were normal (7.3%) or overweight (11.8%), but higher with non-overlapping CIs for both obesity (32.3% (95% CI 25.1–39.5)) and severe obesity (54.7% (95% CI 44.8–64.6%)) categories. A similar, but not statisitcally significant gradient was observed among males, with sxKOA prevalence rising from 20.1% (95% CI 5.9–34.4%) in the normal weighted group to 37.7 (95% CI 22.9–52.6%) in the severe obesity group. Among those with severe obesity, the prevalence of Ksx was higher (with non-overlapping CIs) among females than males. Among both sexes, rKOA, sxKOA, and severe rKOA were more prevalent among those with severe obesity compared to those with obesity (Table 2).

Among knees with rKOA compared to those without (in a supplemental knee-level analysis), osteophytes, meniscal extrusion, calcium crystals, and moderate to severe cartilage damage were much more prevalent (Supplemental Table A). Moderate to severe effusion/synovitis (grades 2–3) was observed in about 32% of rKOA knees versus 19% of non-rKOA knees. Large or medium medial tibiofemoral osteophytes (grade 2–3) were present in 43–51% of rKOA knees compared to 2–4% of non-rKOA knees (Supplemental Table A).

Weighted, person-level prevalence estimates of knee US features by sex in the JoCoHS cohort are shown in Table 3. Mild (grade 1) US features were seen in one-third to one-half of the population in at least one knee. Greyscale (GS) effusion and synovitis were numerically more prevalent among males compared with females across all grades, with non-overlapping CIs for moderate to severe GS effusion/synovitis and GS synovitis and presence of GS effusion. In contrast, US findings of medial and lateral cartilage damage were more prevalent in females than males, although some CIs overlap (Figure 1). Osteophyte prevalence was generally similar between sexes, with overlapping 95% confidence intervals across grades (Table 3).

Table 4 shows weighted, person-level US prevalence estimates by BMI categories. GS effusion and Grade 1 PD decreased slightly with greater BMI, from 67.1 and 50.7%, respectively, in normal BMI participants to 43.3 and 26.6% in those with severe obesity (Table 4 and Figure 2). In contrast, the prevalence and severity of most other structural features was higher in higher BMI categories. Medial tibiofemoral osteophytes of grade 2 or 3, medial femoral cartilage thinning, and meniscal extrusion were more prevalent in higher categories of BMI (Table 4, Figure 2).

Discussion

We identified differences in the prevalence of US features of KOA, Ksx, rKOA, sxKOA, and severe rKOA particularly among females with severe obesity. Overall, there is a high prevalence of knee symptoms and radiographic OA in this community-based cohort and prevalence is higher with greater BMI, particularly among females. When compared to the baseline examination of JoCoOA, the prevalence of Ksx, rKOA, sxKOA, and severe rKOA has increased substantially across each category, although there are a few differences in inclusion criteria and definitions between the two cohorts.¹¹ At JoCoOA baseline (1991–1997), the prevalence of Ksx, rKOA, sxKOA, and severe rKOA was 43%, 28%, 16% and 8% respectively,¹¹ compared with 61%, 36%, 27%, and 21%, respectively, in the current study (JoCoHS, 2019–2024). To further enhance comparability, we also restricted JoCoHS data to those 45 years of age and excluded TKR, resulting in estimates of 62%, 42%, 31%, and 23%, respectively, further supporting this increase. Consistent with the baseline examination, prevalence in both populations was numerically greater in females.

These estimates are consistent with other literature that demonstrates a large and increasing burden of rKOA, likely secondary to an aging population and increasing obesity across populations.^{4, 5, 13, 14} The Framingham Osteoarthritis study (1987) reported an rKOA prevalence among adults aged ≥45 years to be 19.2%.¹⁵ In comparison, JoCoOA (1991–1997) reported a higher rKOA prevalence of 27.8% in the same age group.¹⁵ In the subsequent 2002–2005 Framingham cohort this prevalence rose to 27.7%,¹⁴ which parallels the increase in prevalence over time seen in the JoCoOA. This increase in prevalence across time parallel trends seen for sxKOA (knee pain on most days in the past month in radiographic KOA in the same knee) in the Framingham Study, where sxKOA approximately doubled in females and tripled in males over 20 years.¹⁴ Similarly, estimates from the Global Burden of Disease study reports that the overall prevalence of KOA which is defined as both symptoms and radiographic evidence is about 23% which is consistent with the recent Johnston County findings.¹³

Population-based prevalence estimates of US features are a unique contribution of this analysis. Our study revealed that males had higher prevalence of inflammatory features including effusion/synovitis, while females had a higher prevalence of structural cartilage changes. Jiang et al similarly found that males had higher prevalence of synovial hypertrophy and effusion.¹⁶ Other studies have found that females have lower cartilage thickness than males at baseline, and also lose articular cartilage at a greater rate.¹⁷ These findings might be partially explained by the fact that males normally have thicker synovium and more synovial fluid than females.¹⁷ Additionally, while synovial hypertrophy and effusion can be intermittent, cartilage damage is more likely to reflect persistent damage. Sex differences in OA have been attributed to a variety of etiologies including hormone-determined interactions, molecular based including IGF, anatomical differences, and biomechanical differences, and are likely multifactorial.¹⁸ Osteophytes by US were common in the population overall with a similar prevalence in males and females.

Prior prevalence studies of KOA have not distinguished between obesity and severe obesity, due to the relatively small proportion of the populations under study within the severe obesity category.¹⁹ Systematic reviews have demonstrated that higher BMI is associated with worsening knee OA symptoms, structural progression, and higher rates of total knee replacements.^{20, 21} These systematic reviews also suggest that obesity-related metabolic and inflammatory changes, in addition to increased mechanical load can be potential contributing factors to the pathogenesis and progression of KOA.^{20, 21} In this study, BMI related gradients were observed in both clinical and imaging outcomes. Weighted prevalence estimates demonstrated non-overlapping 95% CI (and thereby statistically significant differences) for Ksx, rKOA, sxKOA, and severe rKOA in females with obesity compared with females with severe obesity. This is consistent with a prior systematic review of 27 studies that revealed that a high BMI confers a greater relative risk for incident KOA in females than in males.²² This differentiation could lead to the development of more effective prevention and treatment strategies, as well as a better understanding of the distinct mechanisms driving KOA in individuals with severe obesity. Studies have demonstrated that osteophyte formation and cartilage thinning are associated with increasing BMI.^{23, 24} Notably, US structural features including osteophytes and cartilage thinning also followed this BMI gradient, while GS effusion and PD were exceptions to this rule. Of note, subcutaneous fat distribution, which could affect effusion detection, was not directly assessed and does not directly correlate with BMI.²⁵ This distinction between features could also relate to attenuation from overlying soft tissue particularly affecting effusion and PD. These findings can reflect how mechanical and metabolic contributions of obesity could potentially contribute to OA pathogenesis and phenotype.^{26, 27} Additionally, structural osteoarthritis features may reflect the cumulative and metabolic effects of obesity. However, the observation that synovial inflammation might fluctuate independently of BMI is in contrast to prior work showing that knee effusion-synovitis progression was slowed by weight loss and that worsening effusion-synovitis mediated the association between BMI and KOA progression.^{28, 29} This study has several strengths including the use of a large, community-based cohort with sample weighting to allow for generalization to Johnston County, NC. Hence, these estimates from this cohort are likely generalizable to the southeastern United States. The combined use of radiographs with a standardized reliability-tested US scoring protocol allowed for comprehensive characterization of KOA, even in the absence of MRI. Significant differences were conservatively interpreted based on whether their confidence intervals did not overlap; inferential hypothesis testing was not performed, as the purpose of this work was to quantify prevalence rather than to test for differences. While this population is a large community-based cohort, the data are cross-sectional, limiting our ability to assess temporal or causal associations between risk factors and outcomes. The use of standardized static views, particularly for effusion detection, may have limited our ability to detect small effusions, which combined with variable attenuation, may have differential effects in obesity³⁰.

These findings highlight the impact of obesity, particularly severe obesity, on the burden of KOA in community populations. Clinically, recognizing severe obesity as a modifiable risk factor is crucial in prevention of KOA. As a modifiable risk factor, weight management intervention including lifestyle modification and anti-obesity pharmacotherapy should be considered in patients with obesity for KOA prevention and management. Ongoing trials, such as the Osteoarthritis Prevention Study (TOPS), are evaluating whether long-term weight loss and exercise interventions can prevent the incidence of structural KOA in females with obesity, reflecting a developing focus on targeted sex-specific prevention strategies.³¹ Furthermore, US may have an expanding role in clinical care, offering ability to assess features and potentially phenotyping the disease. Future studies can further evaluate how US can integrate into clinical care to improve phenotyping and personalized management of KOA.

Conclusion

This study reveals a significant burden of Ksx, rKOA, sxKOA, and severe rKOA, particularly among females and individuals with higher BMI in a contemporary population. Males had a higher prevalence of inflammatory features by US including effusion/synovitis, while females had a higher prevalence of structural cartilage changes. However, effusion/synovitis was not associated with increasing BMI. The differentiation between obesity and severe obesity revealed meaningful gradients in both symptomatic and structural features, demonstrating an important potential role for point-of-care US in KOA assessment. Additional longitudinal population-based studies are needed to confirm our findings, assess relationships with long-term outcomes, and identify new treatment targets in this chronic condition.

Supplementary Material