Joint hypermobility is not positively associated with prevalent multiple joint osteoarthritis: a cross-sectional study of older adults

Osteoarthritis (OA) is a common and debilitating disease with a large public health burden in the United States and globally [1]. OA development and progression can be influenced by biomechanical factors (e.g., joint injury, obesity) that change joint structures, alignment, motion, and loading, Joint hypermobility, broadly defined as range of motion of the joint that is greater than normal, may also influence OA development and progression [2]. The laxity of ligaments can contribute to increased range of motion, potentially leading to a hypermobile joint. Abnormalities in collagen and elastin may contribute to ligamentous laxity, with less severe defects occurring in isolated forms of joint hypermobility syndrome compared to heritable collagen diseases, such as Ehlers-Danlos syndrome [3]. Hypermobility may contribute to joint injury [4] (microtraumas over time from stresses to joint structures at extremes of range of motion or an increased susceptibility to a single major traumatic event), pain [5], and damage to atypical contact areas of cartilage tissue [6]. Due to this propensity for altered biomechanics and injury, joint hypermobility may be a unique risk factor for OA [7, 8].

Joint hypermobility, which occurs in 10–25% of adults depending on the population and hypermobility definition [9‐12], is associated with female sex and young age, with the degree of joint range of motion decreasing as an individual ages [13]. Due to conflicting evidence in the literature to date, the relationship between hypermobility and OA is unclear. Multiple studies have found joint hypermobility to be related to arthralgia and OA of joints in the upper and lower body [2, 13‐15], although these associations are not always consistent [16, 17]. More recent work has found a wide range of relationships between hypermobility and various joint sites affected by OA [18‐21].

In the individual with OA, more than one joint site may be affected. Thus, it is appropriate to evaluate the association of generalized hypermobility (as measured by the Beighton score) not solely with single joint OA, but with the occurrence of multiple joint OA (MJOA) as well. Currently, the possible role of joint hypermobility in MJOA is not known. It is evident, however, that MJOA may represent a distinct etiology from mono-articular OA [22‐27] and is associated with increased disease burden in the individual patient [28‐35]. Accordingly, it should be considered separately when assessing for associated conditions, including joint hypermobility. Of note, MJOA is relatively understudied and lacks a precise definition in the literature. To account for this, a systematic review was conducted in conjunction with this study to arrive at a list of representative definitions to classify the condition and aid in data analysis [36]. In this study, we aimed to examine the association of literature-based MJOA definitions and joint hypermobility in a large community-based sample of adults: the Johnston County Osteoarthritis (JoCo OA) Project.

There were 1697 participants with Beighton and any OA data available for analyses. Participants with missing values for age (n = 2), BMI (n = 2), Beighton (n = 1) or MJOA (n = 16) were excluded (total n = 20), leaving 1677 participants for these analyses (Fig. 1). The mean age of the overall cohort was 68.6 (SD 9.1) years, mean BMI was 31.5 (SD 7.2) kg/m², with 68% women and 31% African Americans. Joint hypermobility was relatively infrequent, with 3.9% of participants (n = 65) with a Beighton score ≥ 4 and 8.9% (n = 150) with a score ≥ 3. Frequencies of the 4 MJOA definitions within the overall JoCo cohort ranged from 13 to 49% (Table 1). Among those meeting at least one definition of radiographic MJOA (n = 1064), about a third met criteria for one (n = 303; 28.5%) or two (n = 319; 30.0%) definitions, 16.8% (n = 179) for 3 definitions, and nearly a quarter met criteria for all four definitions (n = 263; 24.7%, Fig. 2).

Descriptive characteristics of the participants included in the hypermobility analyses by radiographic MJOA status are detailed in Table 2. Overall, compared to those without MJOA, those with MJOA did not differ by BMI but were older, more often women than men and more often white than African American. The only exception to this was in MJOA-4 analyses, wherein there was no statistically significant difference in disease frequency based on sex or race.

The aOR and corresponding 95% CIs comparing the odds of meeting different MJOA definitions by Beighton criteria are shown in Tables 3 (rOA) and 4 (sxOA). With some exceptions, hypermobility as defined by varying Beighton cutoffs was generally associated with lower odds of both radiographic and symptomatic MJOA; not all associations were statistically significant.

Participants with joint hypermobility were almost 75% less likely to have MJOA-1 (aOR 0.26, 95%CI 0.13–0.52, Table 3). However, no statistically significant associations were identified for the other MJOA definitions. The sxOA definitions followed a similar trend, such that those with joint hypermobility were nearly 60% less likely to have MJOA-1 (aOR 0.42, 95%CI 0.20–0.87, Table 4).

We also assessed for relationships between hypermobility at individual sites (knee, trunk, elbow, fifth finger, thumb) and each MJOA definition (shown in the lower rows of Tables 3 and 4). Four of these subgroups—trunk, elbow, knee, and thumb—had small sample sizes (n < 100). Many more participants had fifth finger hypermobility (n = 1108) than had hypermobility at any other individual joint site. Overall, the pattern of association of MJOA and individual site hypermobility was consistent with findings for generalized hypermobility. Namely, although most of the associations did not attain statistical significance, hypermobility at individual joints was associated with lower odds of radiographic MJOA across definitions. An exception to this trend was that for rOA, elbow hypermobility was statistically significantly associated (aOR 0.34, 95%CI 0.19–0.62) with decreased odds of MJOA-1 (> 1 IP node and > 2 other sites out of hip, knee, and spine).

For sxOA and site-specific hypermobility, trunk hypermobility was associated with decreased odds of symptomatic MJOA-4 before adjustment (OR 0.56, 95%CI 0.33–0.94), but the association attenuated slightly after adjustment (aOR 0.61, 95% CI 0.35–1.04). Hypermobility of the fifth finger was associated with 25–34% lower odds of symptomatic MJOA-1, − 2, − 4 (Table 4).

In this community-based cohort of middle-aged to older adults, hypermobility was generally associated with decreased odds of MJOA across most definitions, although not all relationships were statistically significant. Notably, this relationship was found for both radiographic and symptomatic MJOA definitions. For MJOA-1, generalized joint hypermobility was statistically significantly associated with lower odds of radiographic and symptomatic MJOA. Hypermobility at individual joint sites followed a similar pattern but with inconclusive results likely due in part to small cell sizes because of the low prevalence of joint hypermobility. Joint hypermobility was less common in our cohort (< 10%) than reported in other studies, which may be due to the high frequency of OA among our participants, particularly in the joints that are tested by the Beighton criteria. A joint that might have once been hypermobile could present as not-hypermobile on the Beighton criteria when OA is present.

Our group is the first, to our knowledge, to assess literature-based definitions of MJOA for associations with hypermobility, so we cannot directly compare findings for MJOA across studies. Other published cross-sectional studies have suggested an inverse relationship between joint hypermobility and OA. In the large family-based CARRIAGE study (African American and Native American heritage, n = 280), hand and knee hypermobility were significantly associated with a lower prevalence of clinical OA in these joints [17]. Kraus et al. found decreased odds of PIP rOA among those with generalized hypermobility in the GOGO cohort (n = 1043, [41]). In our recent work examining the JoCo OA, GO, and GOGO cohorts, participants who completed the trunk flexion maneuver, suggestive of a flexible spine and hamstrings, were less likely to have rOA of the lumbar spine or facet joints; the directionality of this association remains unclear [19].

In contrast, a number of studies reported an increased risk of OA in association with hypermobility. One clinic-based study of female patients (n = 100) reported a statistically significant association between sxOA in at least 3 sites and generalized or site-specific hypermobility as defined by Beighton [14]. A 2016 clinical study with 503 Turkish participants found both generalized hypermobility and knee hypermobility to be associated with knee rOA [42]. Lastly, a clinical study in an Icelandic cohort by Jonsson et al. produced mixed findings; generalized hypermobility was associated with 1st CMC clinical OA, yet was inversely associated with hand IP clinical OA (n = 200) [15].

These conflicting findings could be due to a few factors. First, multiple outcomes were assessed across studies, including both radiographic and symptomatic OA, and joint symptoms, reducing comparability of results. Second, samples ranged widely in size and source, further reducing generalizability across studies. From a genetic standpoint, joint hypermobility and OA appear to be linked, potentially by abnormalities in collagen contributing to both conditions, as observed in reports showing a higher likelihood of OA among individuals with collagen diseases, like Ehlers- Danlos syndrome [43]. From a biomechanical standpoint, hypotheses exist in the literature for both possibilities of a positive and negative correlation of hypermobility with MJOA. Hypermobility has been hypothesized to be positively associated with MJOA through the mechanism of joint malalignment and injury due to abnormal forces at the joint [7, 8]. Two studies showed that greater loading of foot structures (i.e., greater midfoot peak pressure and maximum force values at the midfoot and hallux) was associated with more severe joint hypermobility [44, 45]. Conversely, it has been proposed that hypermobility may cause pain- or joint instability-induced moderation of activity that may in fact reduce OA risk [17].

A strength of our study design was our use of a large, community-based sample including African American and white men and women. Our consideration of multiple joint sites in data analysis was unique and was supported by a formal systematic review conducted prior to data analysis to derive multiple evidence-based MJOA definitions [36]. Finally, consideration of both rOA and sxOA allowed a greater breadth of understanding of the condition and added to the clinical relevance of the findings.

Limitations of this study include its cross-sectional nature, such that our analyses were restricted to MJOA frequency, without consideration of incidence, progression, or causal associations. The included JoCo participants were all above 45 years old with a mean age of 69 years. Individuals in this cohort who had joint hypermobility when they were younger may be less likely to present with hypermobile joints at the time of the research clinic visit because of aging-related changes to the joint or the development of OA, and thus, their joint hypermobility status may be misclassified. Further, compared to participants without MJOA, those with MJOA were generally older, and for MJOA-1 through − 3 were more likely to be women and white. Thus, we cannot generalize our results to other age groups men, or other racial/ethnic groups. The Beighton score for hypermobility is itself a limitation. As the Beighton score only assesses 5 joint sites, this limits comparison between hypermobility and MJOA measures. For instance, some of our MJOA definitions included a requirement for IP involvement, which, aside from the tests for 1st and 5th digit hypermobility, is not well-represented in the Beighton criteria. Also, hip hypermobility is assessed as part of the trunk maneuver—along with spine mobility and hamstring flexibility—thus preventing independent assessment of this joint. This, along with the small sample sizes for those with hypermobility in individual joints, limited our ability to draw meaningful conclusions from the relevant data at each joint site. The associations overall were modest and should be considered in the context of the overall sample size and number of comparisons presented.

Validation of the MJOA definitions used in these analyses in other cohorts is needed to determine generalizability. Additionally, a better appreciation of hypermobility at individual joint sites, including hip and specific IP measures, could improve understanding of site-specific relationships between MJOA and hypermobility. Lastly, more research into the biomechanical effects of hypermobility on joint physiology would aid our understanding of a potential mechanistic link between hypermobility and MJOA and guide further study of these conditions.

In summary, in this large community-based cohort of adults, joint hypermobility was inversely associated with at least one definition of MJOA and was not positively associated with either radiographic or symptomatic MJOA by any definition. To assess their validity, it would be of value for MJOA definitions used here to be applied to other large cohorts. Longitudinal studies are needed to determine the contribution of hypermobility to the incidence and progression of MJOA outcomes over the life course in cohorts that include younger adults.