Obesity Profiles with Knee Osteoarthritis: Correlation with Pain, Disability, Disease Progression


  • Ray Marks

    Corresponding author
    1. Department of Health and Behavior Studies, Columbia University, Teachers College, New York, and Department of Health/Physical Education & Gerontological Studies/Services, York College, The City University of New York, New York, New York.
    Search for more papers by this author

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Department of Health and Behavior Studies, Teachers College, Box 114, 525W 120th Street, New York, NY 10027. E-mail: rm226@columbia.edu


Objectives: To identify the prevalence of overweight among community-dwelling adults diagnosed as having knee osteoarthritis (OA) and the relationship between the weight status of these individuals, selected disease-related outcomes, and disease progression.

Research Methods and Procedures: The BMIs of 82 women and 18 men with unilateral or bilateral knee OA were examined on a single occasion along with data on physical comorbidities, pain, and function and subjected to correlation analyses. BMIs from two additional samples, one that included 16 women with and without knee OA and one that included 24 women and 6 men with knee joint OA that required surgery for the subsequent onset of hip OA, were also assessed.

Results: At least 80% of all present cohorts were overweight or obese. Those with higher BMIs reported more pain than those with lower BMIs (p < 0.05) and pain was related to perceived physical exertion (p < 0.05). Body mass indices were not significantly correlated with generic gait measures, but an inverse trend toward the time spent in the gait cycle (r = −0.63; p = 0.097) that may impact the disease process was identified. Those with comorbidities had the same body mass, on average, as those with no comorbidities, and those with bilateral disease were heavier than those with unilateral disease.

Discussion: A high body mass is present in most adults with knee OA. Moreover, being overweight may affect knee joint impact rates and pain incrementally. Having high body weights may heighten the risk for bilateral knee joint, as well as hip joint, OA.


Knee joint osteoarthritis (OA),1 a common disabling condition of multifactorial origin (1), may be influenced by factors other than aging, such as pain (2). In addition, a variety of comorbid conditions may be implicated in the pathogenesis of knee joint OA.

Moreover, attempts to ameliorate this painful condition are likely to be impacted negatively by accompanying comorbid states such as obesity.

However, even though being overweight is associated with increases in the amount of force across a weight-bearing joint (1), and the development of incident OA at all joints studied (3) and at the knee in particular (4, 5), little comparative documentation exists about the degree to which obesity prevails in North American men and women with unilateral and bilateral knee joint OA, especially those in the older age ranges. There is sparse documentation as well concerning the relative effect of sex, age, and comorbidities and being overweight or obese among adults with knee OA and whether having a high body mass impacts selected physical and psychological correlates of this condition and disease progression.

To reduce this gap, this analysis examined the prevalence of overweight and obesity among ambulatory individuals living in the community who were diagnosed as having mild-to-moderate radiographic and symptomatic OA of one or both knee joints. It also examined the relationship between the BMIs of the subjects and a proxy measure of their functional ability, and how body mass varied by disease distribution, sex, disease severity, and age range. The relationship between BMI, having one or more comorbid conditions, and pain also was assessed.

The information was sought in an ongoing effort to improve our knowledge concerning the complex mechanisms that interact to produce or accentuate knee OA disability. Using previously collected data, the specific goal was to identify trends that could help establish what prevention/intervention strategies, other than those currently considered standard practice, might be efficacious for ameliorating the magnitude of this disabling condition. Because knee pain in younger subjects is significantly associated with obesity (6), and knee structural alterations are found to be higher in those who are obese (1), it was felt those individuals with knee OA who were obese would be apt to experience more subjective pain than those with lower body weights as identified for a cohort of Japanese women with knee pain by Aoyagi et al. (7) and Spanish patients by de Miguel Mendieta et al. (8).

Research Methods and Procedures

The data for this case series were extracted from the available records of three different samples of non-institutionalized community-dwelling individuals with clinical and radiologic evidence of idiopathic or primary knee joint OA deemed eligible to participate in either an exercise or strength-testing study specifically designed for adults with knee OA. Individuals with any evidence of secondary OA, inflammatory arthritis, and those with neurologic conditions were excluded. All subjects had to undergo a physician-led clinical and radiologic examination to verify their eligibility, and criteria for inclusion were derived from those outlined by Altman (9). These criteria for classification of symptomatic OA have been developed by a subcommittee of the American College of Rheumatology and include knee pain, joint space narrowing, and morning stiffness of less than 30-minute duration. All data were collected before study participation on a single occasion using standard validated procedures. All subjects provided informed consent using an ethics board-approved protocol.

Common variables assessed were age, sex, body weight, height, and number of involved knees. Other data retrieved included reported comorbidities and subjective pain levels as assessed over the past week using two well-established measures, the Huskisson visual analog scale (VAS) for pain and the Arthritis Impact Measurement Scale (AIMS) pain subscale, that have both been found to have satisfactory test–retest reliability and validity. To obtain a proxy measure of obesity, a BMI profile was calculated from the available height (in meters) and weight (in kilograms) data recorded using a standard laboratory scale and categorized according to national standards in kilograms per meter squared, where being underweight is denoted by a BMI <18.5 kg/m2; normal-weight ranges from 18.5 to 24.9 kg/m2; overweight is 25 to 29.9 kg/m2; Obese Class 1 is denoted by BMIs ranging from 30 to 34.9 kg/m2; Obese Class 2 ranges from 35 to 39.9 kg/m2; Obese Class 3 equals or exceeds a BMI of 40 kg/m2 (10). Function was assessed according to distance walked indoors at self-paced speed in a 6-minute time interval, a fast self-paced walk test on level ground, a subjective assessment of perceived exertion recorded on an ordinal 0 to 15 point scale when walking for 6 minutes, and using measures of cadence, step length, and gait cycle duration.

The descriptive and inferential statistics used to assess the presence of overweight and other variables of interest, plus the univariate correlations between these and body mass, were analyzed using the SPSS statistical software package, version 10.0 (SPSS, Inc., Chicago, IL).


Sample 1

Sample 1 was made up of 58 cases with primary unilateral knee OA and 42 cases with bilateral knee OA (mean age, 70.07 ± 9.55 years; mean BMI, 29.3 ± 5.8 kg/m2, indicating that, on average, the cohort was overweight).

As shown in Figure 1, there were no cases that could be classified as underweight in this sample. Moreover, the percentage of cases with normal weight was low, more than 80% were overweight, and 9 cases were classified as having a BMI greater than 40 kg/m2, or being in the Class 3 category of obesity.

Figure 1.

: BMI distribution across 100 cases with knee OA showing high percentage of cases who are overweight or obese.

As shown in Figure 2, those with a confirmed diagnosis of bilateral knee OA, whose average BMI was 30.3 ± 5.7, had higher BMIs in both the overweight and the obesity categories, on average, than those with unilateral knee OA (average BMI, 28.8 ± 5.4 kg/m2), and a two-tailed t test revealed that there was a statistically significant difference in their mean BMIs (p = 0.000).

Figure 2.

: Comparison of BMI distribution in cases of unilateral and bilateral knee OA.

As shown in Figure 3, men (N = 18) and women (N = 82) in this cohort had similar BMIs of 29.4 ± 4.4 and 29.4 ± 5.8 kg/m2, respectively. In addition, the BMIs of those 39 cases with no comorbid health conditions and the remaining cases with one or more comorbid health conditions were 29.04 ± 5.0 and 29 ± 5.8 kg/m2, respectively, and were comparable.

Figure 3.

: BMIs across disease classification and sex among 100 cases of knee OA.

In terms of sex, 15.8% of men were in the normal BMI range compared with 21% of the women (Table 1). Men were also more likely to be categorized in the obese range of BMIs of 30 to 44 kg/m2 than women, but women were more likely to be categorized in the Class 3 obese range.

Table 1. . Comparisons across body weight categories among 18 men and 82 women with knee OA from a similar community, expressed as percentages
Category of body weightMenWomen
  1. OA, osteoarthritis.

Class 3 obesity5.39.5

In terms of age, our data revealed that, after age 70, cohort members were more likely to have bilateral rather than unilateral disease than members in the 50- to 59-year-old age range, and the BMIs were generally higher for cohort members over 70 years of age with bilateral vs. unilateral disease (Figure 4). The BMIs (35.2 ± 7.3 kg/m2) were higher for the cases with unilateral disease between 50 to 69 years of age than they were for bilateral disease cases (30.1 ± 7.4 kg/m2) and were highest, on average, in the 40- to 49-year-old category (36.5 ± 3.6 kg/m2). No cases of bilateral knee OA were presently observed.

Figure 4.

: Distribution of body mass across disease classification and age showing high mean BMIs and unilateral joint disease only in younger age ranges across 100 cases.

Significant two-tailed positive Pearson correlations of note were those recorded between pain as measured on a VAS of r = 0.265 (p = 0.007), pain as measured on the AIMS of r = 0.352 (p = 0.000), and BMI. There were no significant correlations between BMI and perceived exertion on a 6-minute walk test (r = 0.012, p = 0.905), distance walked in 6 minutes (r = 0.040, p = 0.689), and fast walking velocity (r = −0.003, p = 0.975). However, age affected walking distance (r = −0.41; p = 0.000), and perceived exertion was related to pain (p < 0.05). Having one or more comorbid conditions seemed to impact walking ability adversely and independently of obesity.

Other Findings

The mean overall pain score as recorded on a VAS was 5.01 ± 2.13 and it was 4.89 ± 2.02 on the AIMS. Men rated their pain levels on these scales as 4.11 ± 1.65 and 4.33 ± 2.02, respectively. Women rated themselves as having scores of 5.23 ± 2.18 and 5.03 ± 3.07 on these two pain scales, respectively, although men's and women's BMIs were similar on average (Figure 3). The correlation between body weight and BMI among this group was 0.83 (p = 0.000), indicating weight is a significant predictor of BMI in this group, and overall BMI was inversely correlated with age (r = −0.56; p = 0.000). There was an inverse trend exhibited among this knee OA cohort between time spent in the gait cycle when walking at a self-determined fast walking pace and their BMI (r = −0.63; p = 0.097). A longer gait cycle duration was also more evident among those with unilateral knee OA than those with bilateral disease (p = 0.037).

Sample 2

The second convenience sample was made up of 16 women with clinical and radiographic knee OA with a mean age of 60.9 ± 10.6 years (range, 34 to 74 years) and a mean BMI of 30.1 ± 4.2 kg/m2. Again, no case was deemed underweight, 12.5% were normal weight, 37.5% were overweight, and 50% were in the obese category. The relationship between their body weight and their BMIs was r = 0.896. Two cases had had OA for more than 35 years, and 6 cases were less than 60 years of age. Although the sample size was small, when data from 12 of these women (mean age, 60.83 ± 11.89 years) were compared with available data from a group of 12 women with no knee joint disease (mean age, 61.16 ± 13.43 years), the women without knee OA were found to have a mean BMI of 23.5 ± 1.5 kg/m2, and 58% were of normal weight. Moreover, although 52% were overweight, only 1 individual without the disease was classified as obese. When compared statistically using a two-tailed t test for independent samples, the BMIs of 31.1 ± 4.5 kg/m2 of these women who were age-matched within 5 years were clearly higher for those with knee OA who were either overweight or obese (p = 0.001; Tables 2 and 3).

Table 2. . Summary of selected studies linking knee OA and excess body weight
  1. OA, osteoarthritis.

Al-Arfaj, 2002 (37)Saudi ArabiaCross-sectional studyExcess weight was strongly associated with knee OA in women, with a weaker link for men
Davis et al., 1990 (4)United StatesCross-sectional and follow-up survey of adults ages 45 to 74, 1971 to 1975Obesity was associated with unilateral and bilateral knee OA, regardless of age, sex, and health status
Dawson et al., 2003 (14)United KingdomCase-control studyBeing overweight before 40 increased risk of knee OA in women
Holmberg et al., 2005 (5)SwedenPopulation-based case control studyOverweight at any time was related to knee OA
Manek et al., 2003 (15)United KingdomCross-sectional twin studyThere was a strong association between high body mass and knee OA, which may be environmentally impacted
Raynauld et al., 2005 (38)CanadaProspective study of 107 cases over 24 monthsHigh BMI was a significant predictor of fast cartilage volume loss
Table 3. . Descriptives for 12 control women with no knee OA and 12 women with knee OA age-matched within 5 years showing their ages and BMIs
Age (yrs)BMI (kg/m2)Age (yrs)BMI (kg/m2)
  • OA, osteoarthritis.

  • *

    p = 0.001.

Mean23.3 31.1*
Standard deviation1.5 5.0

Sample 3

Sample 3 was a cohort of 30 cases who all had histories of knee joint OA warranting arthroplasty surgery in the past 4 to 5 years or more and were currently awaiting surgical intervention for late onset of hip OA. Among this cohort, 18 had unilateral knee joint OA, 6 were men, and 4 were less than 60 years of age. The 24 women, 17 with unilateral and 7 with bilateral knee OA, had a mean age of 71.2 ± 10.34 years and an average BMI of 28.0 ± 5.1 kg/m2, and the 6 men, 1 with unilateral knee joint OA and 5 with bilateral knee OA, were 66.83 ± 10.52 years of age, on average, with a mean BMI of 33.8 ± 11.6 kg/m2. The distribution of BMIs in this cohort again showed 80% were in the overweight or obese range, and the prevalence of overweight was highest in the 60- to 69-year-old age range, followed by the 40- to 49-year-old age range (Figures 5 and 6).

Figure 5.

: Distribution of BMIs across 30 cases with prior histories of knee OA warranting replacement surgery, presently scheduled for hip replacement surgery.

Figure 6.

: Distribution of mean BMIs of 30 cases with knee OA warranting knee replacement surgery showing highest body mass averages in the 60- to 69-year-old age range.


In an attempt to better understand the etiology and pathogenesis of OA, given its enormous economic and social impact, much emphasis has recently been placed on the physical correlate of body mass as a potential mediator of OA disability (11). Indeed, as outlined in Table 2, previous work has strongly implicated excess body mass in the pathogenesis of knee OA and its associated disability. However, very few studies have specifically examined this variable in a sample of North American community-dwelling men and women with varying degrees of clinical and radiologically defined knee joint disease. Also, very little focus has been placed on establishing the degree to which body mass can impact OA disability and pain.

To examine whether being overweight is a key risk factor for knee OA disability, as suggested by some prior studies listed in Table 2, this study examined the prevalence of being underweight, normal-weight, overweight, and obese as defined by the proxy measure of BMI. It examined separate cohorts of men and women diagnosed with unilateral or bilateral knee OA, as well as the belated onset of hip OA. It also examined the extent to which body mass impacts the disability correlates of function and pain.

In general, among these groups, the prevalence of overweight of 80% was consistent and was considerably higher than that found in United States adults where rates are between 20% and 30% (12). Even though very few healthy individuals could be located as controls, those who were healthy were much less likely to be overweight than those with knee OA, regardless of age. Moreover, in those with no surgical history, no subject with knee OA could be classified as being underweight. The prevalence of overweight among these cohorts was also higher than that reported by Ding et al. (1), where 60% of 372 men and women who were adult children of subjects who had knee replacement surgery for primary knee OA or who were randomly selected controls were overweight. However, not all of the subjects in that study had radiographic knee OA, and, thus, the sample may not have been truly representative of people with radiographic knee OA.

Our data also show slightly higher BMIs than those reported by Coggon et al. (13) (among 525 cases where the median BMI was 28.1 kg/m2), but it was unclear whether the diagnostic criteria were similar to ours and whether a proportion of their patients had bilateral disease. Although our data are subject to limitations because of the inherent nature of self-report, and we did not study a large healthy disease-free group in detail, our data are supported by Dawson et al. (14) and Manek et al. (15), who found that a high body mass was strongly associated with knee OA in women, and by Okoro et al. (16), who studied adults ≥45 years of age for self-reported arthritis.

Similar to findings of Cimmino et al. (17), who noted a significant relationship between pain and BMI ≥30 kg/m2 (odds ratio = 1.52; 95% confidence interval, 1.42 to 1.61) among people with different forms of OA, and Creamer et al. (18), who found BMI and pain measured on a VAS among a cohort of 68 outpatients with knee OA, we also found BMIs and pain were related, such that having a higher body mass was associated with more pain. Although the cause–effect relationship of this association cannot be teased out in a cross-sectional study, and both the pain scales we used determined self-reported pain only, it has been observed that load exerted on the knee per step during daily activities, which could hasten its destruction and aggravate pain, is likely to be greater in adults with a high body mass compared with a lower body mass (19). Indeed, subjects with high BMIs who are categorized in the highest waist circumference tertile have a higher risk of incurring knee OA compared with those in the lowest waist circumference tertile (p < 0.05) (20). In addition, Manek et al. (15) found a strong association between a high BMI and the presence of radiographic knee OA in a United Kingdom sample.

Although we could not establish a relationship between function and body mass in this analysis, and this may reflect limitations caused by the performance-based methods used (21), the fast self-paced walking and 6-minute walk tests are considered reliable for making decisions at the individual patient level in people with knee OA (22). Other data reveal that the natural course of knee pain and function in knee OA may be associated with body mass (23, 24). This may reflect a higher presence of body fat, which is associated with more mechanical stress and derangement (25), as well as hormonal products that cause pain or act as inflammatory mediators that hasten cartilage degradation (26). Greater infiltration of the thigh muscles by fat could also impact lower extremity performance negatively (27), because of its association with reduced muscle strength and quality (28), thereby exposing the aging knee joint to greater impact and, hence, to greater knee pain (29), than if the muscle bulk is well preserved.

We also noted an inverse trend exhibited among the first sample of knee OA individuals between the time spent in each gait cycle when walking at a self-determined fast walking pace and their BMI (r = −0.63; p = 0.097), suggesting that more impacts on the knee joint would occur for comparable walking distances in those individuals with a higher body mass because of the higher stepping rate. A longer gait cycle duration was also more evident among those with unilateral knee OA than those with bilateral disease (p = 0.037), suggesting that once the disease is more widespread, the impact on the affected knees for similar walking distances may increase accordingly. Walking with a shorter step cycle may not only increase knee joint impacts at heel strike and, thus, increase pain caused by the heel strike, but also it may explain, in part, why those in pain perceive more exertion when walking than those who have less pain.

As outlined in Figure 7, there are numerous ways for excess body weight to amplify knee joint damage directly or indirectly. Because joint pain is the cardinal clinical complaint of people with OA, and because radiographic and laboratory testing do not correlate well with the disease (30), attempts to unravel the source of the presently observed positive correlation between the level of obesity and pain are likely to prove useful. Moreover, although this basic finding seems quite intuitive, it may be important to note because even though Van Baar et al. (31) found that, after controlling for other characteristics, kinesiologic and psychological characteristics in patients with OA are associated with disability, their association with pain is less clear. Also noteworthy in this context is the finding that the contribution of overweight to the pain experience in knee OA may be greater for women than men with the same diagnosis, thus explaining the higher prevalence of this painful condition among women.

Figure 7.

: Schematic of potential explanation of relationship between obesity and knee OA disease progression.

A further fact supporting a possible independent role for obesity in mediating knee OA, which could occur because of immobility as a result of other health conditions, is that the mean BMI of those with and without comorbid conditions in this study was comparable. Indeed, even though our study does not allow us to predict whether body weight increased before or after the onset of the knee OA condition, those with bilateral joint disease were heavier, on average, than those with unilateral disease, in all categories of overweight, suggesting a disease-associated BMI dose-response effect. In addition, the low prevalence of being underweight or normal-weight, the high prevalence of overweight in this sample (Figures 1 and 5), and the manifestation of OA in people less than 60 years of age who are overweight and who have no other identifiable cause for their condition support the view of a close link between body mass and symptomatic knee joint OA. Moreover, because we studied only non-institutionalized, functionally able community dwellers, our data could under-represent the role of overweight in mediating this condition in those who are less functional.

We recognize this descriptive analysis of convenience samples has several limitations. These include the use of a retrospective cross-sectional study approach, potential biases imposed by self-report, the use of BMI as a proxy measure of fat mass, and the limited sample size and make-up. However, we observed body weight to be a strong predictor of BMI and found similar trends in three different samples of cases with varying degrees of knee OA. Moreover, although our research design does not allow for any true hypothesis testing, our very small comparative sample of women with and without knee OA suggests there is a difference in their BMIs that might be worth studying further. As Issa and Sharma (32) have stated, the anticipated increase in OA prevalence in the ensuing decades has made the need to identify and clarify risk factors for incident OA, OA progression, OA-associated physical function decline, and disability an especially high priority. Pain seems to be difficult to manage in individuals with end-stage lower limb OA and warrants more proactive management by health professionals (33), including the importance of weight loss, as this may impact the disease process (19). It is in this context that we believe our findings are especially salient. Indeed, if our findings can be supported by further prospective studies of larger samples, with more sophisticated methods, such as multi-frequency bioelectrical impedance measures to assess body composition, we feel this could markedly impact attempts to prevent and intervene in this condition and the physical outcomes of knee OA and its related disability as outlined by Maly et al. (34) and Okoro et al. (16). In particular, because pain is a predictor of greater functional decline (32), we believe professionals interested in improving the outlook for people with knee OA can use these results and those of previous researchers to design efficacious interventions to reduce this widespread problem, as outlined by Focht et al. (35) and Messier et al. (19). In addition, although the present knee OA cases with and without comorbid conditions were of comparable body mass, efforts to help adults in general to maintain a healthy weight are advocated because being overweight can increase the risk of developing one or more comorbid conditions, which can impact disease severity (16).

In summary, taken in their entirety, we feel our observations and their interpretation have considerable practical implications for the management of knee OA and for heightening our ability to prevent this condition or retard its severity. As outlined by Powell et al. (36), if obesity is the main preventable risk factor for the onset and progression of OA, it is through prevention of obesity that the quality of life for many across the globe may be markedly increased (16). Future research to support this premise is indicated.


This study was supported in part by the Osteoarthritis Research Foundation, Canada.


  • 1

    Nonstandard abbreviations: OA, osteoarthritis; VAS, visual analog scale; AIMS, Arthritis Impact Measurement Scale.