The high prevalence of knee osteoarthritis in a rural Chinese population: The Wuchuan osteoarthritis study




To estimate the prevalence of radiographic and symptomatic knee osteoarthritis (OA) in a remote rural region of northern China, and compare them with those reported in Beijing and with data from the Framingham (Massachusetts) cohort.


A population-based cross-sectional survey was conducted among 1,030 residents of Wuchuan County, Inner Mongolia, age ≥50 years. Survey participants, mostly farmers reporting heavy physical occupational activity, completed an interviewer-based questionnaire, and bilateral weight-bearing posteroanterior semiflexed knee radiographs were obtained.


Whereas the overall prevalence of radiographic knee OA was similar to that demonstrated in the Beijing OA study, men in Wuchuan had approximately double the prevalence of severe radiographic knee OA (prevalence ratio [PR] 2.5, 95% confidence interval [95% CI] 1.6–3.8) and symptomatic knee OA (PR 1.9, 95% CI 1.3–2.9). Women in Wuchuan also had a higher prevalence of both severe radiographic (PR 1.4, 95% CI 1.0–2.0) and symptomatic knee OA (PR 1.6, 95% CI 1.2–2.1) compared with their Beijing counterparts. The prevalence of bilateral OA and lateral compartment disease were 2–3 times higher in both Chinese cohorts compared with estimates from the Framingham OA study.


The prevalence of symptomatic knee OA in rural areas of China is much higher than reported from urban regions of China or in the Framingham cohort. The higher representation of bilateral and lateral compartment disease in China suggests a unique phenotype to OA. These findings will be useful for guiding the distribution of future health care resources and preventive strategies.

This article was published online on April 29, 2009. The incorrect figure was published for Figure 1. This notice is included in the online and print versions to indicate that both have been corrected.


Musculoskeletal conditions are the most common cause of severe chronic pain and physical disability among older people, directly affecting the health of millions worldwide (1). Knee pain from osteoarthritis (OA) is the most frequently reported peripheral joint symptom (1–4). Most large, population-based observational studies evaluating musculoskeletal disease prevalence have been conducted in North America or Europe, with scant information from less developed regions of the world. However, it has been estimated that by 2050 almost four-fifths of the world's older population (age ≥65 years) will be living in these less developed regions (5).

In 1981, the World Health Organization launched a program of increased research into the causes and consequences of rheumatic diseases. During the past 25 years, the Community Oriented Program for the Control of Rheumatic Diseases initiative completed several small, community-based surveys in the Asia-Pacific region and reported that current knee pain was present in 5–13% of adult populations (age ≥16 years) (6–9).

Other small, community-based Asian studies have reported knee pain rates ranging from 2% in Pakistan (10) to 15–18% in Omani men and women (11) and 39% in a suburban village in Beijing (12). The wide range of knee pain prevalence found may be attributed to differences in genetic, environmental, psychosocial, or cultural factors. However, differences in methods of prevalence ascertainment, and the specific questions used to solicit information on knee symptoms, were not clearly described or standardized in most of these surveys, which limited the validity of the comparisons.

A recent population-based survey conducted in central Beijing (13) provided initial evidence that radiographic or symptomatic knee OA prevalence or disease presentation observed in North America or Europe may not be directly extrapolated to less developed regions of the world. The Beijing OA study, which used the same survey questionnaire and radiographic protocol as the Framingham OA study in the US, found that the prevalence of radiographic and symptomatic knee OA was higher in older women living in Beijing compared with age-matched white women in the Framingham study, and was very similar when comparing men in both studies.

Rural–urban differences exist for most chronic diseases due to varying social, environmental, and occupational influences. Therefore, the prevalence of knee OA found in Beijing may not be applicable to the majority of people in China (and most developing countries) who live in rural communities. The aim of the Wuchuan OA study was to determine the prevalence of radiographic and symptomatic knee OA among older people in a remote rural region of northern China. To allow valid comparisons with the Beijing OA study and the Framingham cohort, the same survey instruments evaluating knee pain and comparable radiographic protocols were used.


The target population for this cross-sectional survey was people age ≥50 years living in rural communities in Northern China. Five communities were randomly selected from a total of 8 rural communities in Wuchuan County, Inner Mongolia. This location was considered representative of rural communities in Northern China and allowed feasible access to a hospital-based radiography facility.

Within these 5 randomly selected communities, a compact segment sampling method (14, 15) was used to identify clusters, each containing 6–8 villages. Clusters were then selected with a probability proportional to the population size at the last census. Then a sketch map was drawn of each selected cluster, showing the dwellings present. The selected clusters were each split into a small number of segments, such that the number of dwellings per segment was always roughly the same. One segment was then chosen at random from each cluster, and all households in the segment were included in the survey. Specific ethnic groups were neither targeted nor excluded. The study was approved by the Peking University Health Science Center Ethics Committee and informed consent was obtained from all study participants.


Trained health professionals administered the survey questionnaires because it was anticipated that many study participants would be illiterate. All interviewers, clinical examiners, and radiograph technicians were trained under the supervision of the study's chief investigators (XK, JL, and JC).

The trained interviewers went door-to-door to enumerate and interview all men and women, mostly farmers age ≥50 years, within the selected households who were self-described residents of Wuchuan County. People who self-reported rheumatoid arthritis (RA), cerebrovascular disease, or a history of lower extremity surgery were excluded from further participation. Participants with self-reported cerebrovascular disease were excluded because lower extremity disability, usually associated with stroke, may confound future analyses of associations between knee OA and disability.

Participants were interviewed at their home or work place. The survey questionnaire focused on joint symptoms and possible risk factors for knee OA. At the end of the interview, all study participants were invited to a central examination site at Wuchuan Hospital for a clinical examination and knee radiographs on the same day. Transportation to the hospital was provided.

Clinical examination and knee radiographs

Height was measured with a wall-mounted stadiometer, using the average of 2 measurements taken. Body weight was assessed using a balance beam scale with 0.1-kg precision. A posteroanterior, weight-bearing, semiflexed, metatarsophalangeal-view radiograph was taken of both knees strictly according to a validated acquisition protocol (16). Radiographs were read by the study's chief investigator (XK) using the Osteoarthritis Research Society International atlas, and Kellgren/Lawrence (K/L) grades (range 0–4) were also assigned. Each knee was also graded at 4 sites (the lateral femur, medial femur, lateral tibia, and medial tibia) for individual radiographic features.

To ensure measurement reliability and comparability of reading with the Beijing OA study, the chief investigator (XK) was sent to Boston University to undergo training using radiographs from the Beijing OA study. Prior to evaluating radiographs from the Wuchuan OA study, the chief investigator reassessed a batch of radiographs from the Beijing OA study. Preliminary reading of batches of randomly selected OA and non-OA films continued until XK reached a high level of interrater agreement using the readings from the primary reader of the Beijing OA study (Dr. Piran Aliabadi, an academically-based musculoskeletal radiologist) as the gold standard. For each batch (∼50 radiographs), 4 previously read knee radiographs from the Wuchuan OA study were fed back to the reader to test intrarater reliability. For K/L grading, the weighted kappa statistic for interrater reliability was 0.80 (95% confidence interval [95% CI] 0.72–0.88) and the intrarater reliability was 0.92 (95% CI 0.86–0.99). The small number of disagreements did not occur in any particular direction, suggesting no likelihood of estimate bias.

We used the same definitions to define cases of radiographic and symptomatic knee OA as were used in the Beijing and Framingham studies (4, 13). Radiographic knee OA was defined as a K/L grade ≥2 in one or both knees. Similarly, severe radiographic knee OA was defined as a K/L grade ≥3 in one or both knees. Symptomatic knee OA was defined as having at least one knee with both a K/L grade ≥2 and a positive response to the question, “In the past 12 months, have you had knee pain lasting most days of at least a month?” We defined medial or lateral radiographic OA, respectively, as a knee with a K/L grade ≥2 and a medial or lateral joint space narrowing score ≥1.

Statistical analysis

Survey participants were divided into 3 age groups: 50–59 years, 60–69 years, and ≥70 years. The age-specific prevalence was calculated separately for men and women. For comparisons between the Wuchuan OA study, the Beijing OA study, and the Framingham cohort, we limited the analysis to subjects ages 59–84 years, which is the overlapping age range of the 3 cohorts. When comparing the prevalence of radiographic and symptomatic knee OA in the Wuchuan OA study with that found in the Beijing OA study, subjects were categorized into 4 age groups: 59–64 years, 65–69 years, 70–74 years, and 75–84 years. We calculated the age-standardized prevalence ratios (PRs) and 95% CIs (17) for subjects in the Wuchuan OA study compared with subjects in the Beijing OA study using the age distribution in the Wuchuan cohort as the standard. The statistical program R, version 2.6.1 (18) was used for the statistical analysis.


A total of 1,165 individuals reporting to be age ≥50 years were identified in 762 randomly selected households in Wuchuan County, Inner Mongolia (Figure 1). Of these subjects, 27 were excluded from further study participation; 7 participants who had RA, 11 who had cerebrovascular disease, and 9 with a history of lower extremity surgery. The remaining 1,138 residents were approached, and 1,030 (91%) consented to participate in the study and completed the home interview in October and November 2005. Those who declined to participate were older compared with the study participants (mean ± SD age 64 ± 7 years versus 58 ± 8 years). Only 4 consenting participants did not attend the radiographic examination.

Figure 1.

Study flow chart.

The characteristics of the participants are presented in Table 1. At the time of data analysis, it was discovered that 24 men and 26 women were actually younger than 50 years (either 48 or 49 years). We retained these 50 participants (7%) in the 50–59-year age category. Under this classification, ∼65% of men and ∼73% of women were ages 50–59 years.

Table 1. Characteristics of the Wuchuan County study participants*
 Men (n = 505)Women (n = 520)
  • *

    Values are the percentage unless otherwise indicated.

  • Includes 50 participants age 48 or 49 years at the time of the survey.

Age, years  
Biometrics, mean ± SD  
 Height, cm166.2 ± 6.3155.2 ± 6.1
 Weight, kg59.7 ± 8.956.2 ± 9.5
 Body mass index, kg/m221.6 ± 2.723.3 ± 3.5
Years of education  
 ≥10 6 2
Main occupation farming9488
Still working9081

The mean body mass index (BMI) was 22 kg/m2, with only 63 participants (6%) considered to be obese (BMI ≥28 kg/m2) and 219 (21%) considered to be overweight (BMI 24–27 kg/m2) according to specific Asian population criteria (19). Approximately one-quarter of men and half of the women had not received any formal education. Almost all participants (91%) were farmers or had been engaged in farming as their main occupation, with the remaining 9% reporting their main occupation as businessman or shopkeeper. At the time of the survey, 85% of all participants were still working (Table 1). Most participants (91%) reported that the occupation they had held longest involved heavy physical work.

Prevalence of either radiographic or symptomatic knee OA increased markedly between the fifth and sixth decades for both men and women, with further increases in the seventh decade for men (Table 2). Compared with men, women in Wuchuan had a much higher prevalence of both radiographic knee OA (PR 2.2, 95% CI 1.6–3.0) and symptomatic knee OA (PR 2.4, 95% CI 1.7–3.5).

Table 2. Prevalence of tibiofemoral knee osteoarthritis in Wuchuan County, Inner Mongolia*
Age, yearsMenWomen
  • *

    Values are the number (percentage) unless otherwise indicated.

  • Includes 50 participants age 48 or 49 years at the time of the survey.

50–5932713 (4)9 (3)38148 (13)33 (9)
60–6911118 (16)14 (13)9036 (40)26 (29)
≥706722 (33)12 (18)4918 (37)15 (31)
All50553 (10)35 (7)520102 (20)74 (14)

Although the overall prevalence of radiographic knee OA in Wuchuan County was similar to that reported in the Beijing OA study, both men and women from the rural Wuchuan community showed a higher prevalence of both symptomatic knee OA and severe radiographic disease (K/L grade ≥3) compared with their Beijing counterparts (Table 3). Men in Wuchuan had approximately double the prevalence of both severe radiographic disease (PR 2.5, 95% CI 1.6–3.8) and symptomatic knee OA (PR 1.9, 95% CI 1.3–2.9) compared with men in the Beijing cohort. Women in Wuchuan also had a higher prevalence of both severe radiographic disease (PR 1.4, 95% CI 1.0–2.0) and symptomatic knee OA (PR 1.6, 95% CI 1.2–2.1) compared with their counterparts in the Beijing cohort. Similar to the findings from the Beijing OA study, bilateral radiographic knee OA was more common than unilateral in both men and women (Table 3).

Table 3. Comparison of crude/age-adjusted prevalence and age-standardized PR of knee OA among rural (Wuchuan) and urban (Beijing) Chinese participants ages 59–84 years*
 Crude prevalenceAge-adjusted prevalenceAge-standardized PR (95% CI)
  • *

    Values are the percentage unless otherwise indicated. Radiographic osteoarthritis (OA) defined by a Kellgren/Lawrence (K/L) grade ≥2, severe radiographic OA defined by a K/L grade ≥3. Data for Beijing reproduced with approval from the Boston University Arthritis Center. PR = prevalence ratio; 95% CI = 95% confidence interval.

Men, n196995197995 
 Radiographic OA202220201.00 (0.74–1.34)
  Unilateral99981.06 (0.64–1.74)
  Bilateral121312120.96 (0.63–1.44)
 Severe radiographic OA1461462.48 (1.62–3.78)
 Symptomatic OA1371371.91 (1.25–2.91)
Women, n1581,4911611,491 
 Radiographic OA364236410.88 (0.70–1.09)
  Unilateral121212111.05 (0.67–1.65)
  Bilateral243024300.81 (0.61–1.08)
 Severe radiographic OA211521151.42 (1.03–1.97)
 Symptomatic OA271827181.55 (1.17–2.05)

Both Chinese cohorts demonstrated a 2–3-fold higher prevalence of medial tibiofemoral compartment disease compared with the lateral compartment. In contrast, the Framingham OA study reported a 5–8-fold higher prevalence of medial compartment disease compared with lateral compartment disease (Table 4).

Table 4. Comparison of the prevalence of medial and lateral knee OA among rural (Wuchuan) and urban (Beijing) Chinese participants ages 59–84 years and the Framingham OA study*
  • *

    Values are the number of patients/total number of knees (percentage). Medial and lateral osteoarthritis (OA) were defined as a Kellgren/Lawrence grade ≥2 and a medial or lateral joint space narrowing score ≥1. Data for Beijing and Framingham reproduced with approval from Boston University Arthritis Center.

 Medial66/316 (21)590/2,977 (20)153/1,117 (14)
 Lateral22/316 (7)271/2,977 (9)30/1,093 (3)
 Medial44/392 (11)179/1,988 (9)93/626 (15)
 Lateral17/392 (4)95/1,988 (5)12/621 (2)


Although the overall prevalence of radiographic knee OA in Wuchuan was similar to that reported in Beijing, the prevalence of both symptomatic knee OA and severe radiographic disease in these rural farming communities in Wuchuan was much more common than that reported in the urban-based Beijing OA study. Similar to findings from other parts of the world, knee OA was more prevalent among women compared with men, and prevalence increased with age.

The Wuchuan OA study also confirmed 2 other unexpected findings from the Beijing OA study. Unilateral and bilateral disease is fairly evenly represented in the Framingham OA study (4), but bilateral disease was 2–3 times more prevalent than unilateral disease among participants in each of the 2 Chinese studies. This finding was unexpected because bilateral disease has been associated with obesity in developed countries (20). In fact, participants in the rural Wuchuan OA study had an even lower mean BMI (22–23 kg/m2) than participants in the Beijing OA study (25–26 kg/m2). In addition, the ratio of medial tibiofemoral compartment versus lateral tibiofemoral compartment disease was similar in both Chinese cohorts (∼2.8:1). This finding is in sharp contrast to the Framingham OA study and other surveys from developed regions of the world, where knee OA mostly presents in the medial tibiofemoral compartment compared with the lateral tibiofemoral compartment (∼6.3:1). Both findings of OA phenotype proportions support the hypothesis of markedly different risk profiles for the development of knee OA between these regions of the world. Further research will be required to explore whether genetic or environmental factors, or a combination of both, are more frequently implicated. For example, there may be more valgus alignment in the lower extremity in China, and this may account for the increased predilection for lateral compartment disease there.

The strongest risk factors identified to date for incident radiographic and symptomatic knee OA in developed countries include aging, female sex, obesity, knee joint injury, knee surgery, and/or employment in an occupation requiring heavy physical work (21). Specifically, knee OA has been associated with occupations involving knee bending, squatting, kneeling, and heavy lifting, with obesity adding considerably to the risk (21–23). The comparisons between the Chinese cohorts have been stratified for age and sex. Furthermore, obesity is almost negligible, and only 5% and 7% of participants in the Beijing and Wuchuan cohorts, respectively, reported having injured their knee in the past sufficiently to limit walking for at least a week. Similar to the Beijing OA study participants, the majority of the Wuchuan OA study participants were Han Chinese. Therefore, the much higher prevalence of symptomatic knee OA and severe radiographic disease in this rural community compared with the urban Beijing cohort is likely to be mostly attributable to a lifetime of heavy physical work associated with farming as an occupation. Approximately 91% of participants (men and women) in the Wuchuan OA study reported having been engaged in farming for most of their adult lives, with almost all (91%) reporting that the job they had held the longest involved heavy physical work. In contrast, only 34–39% of women and men in the Beijing OA study reported that the job they had held the longest involved moderate to heavy labor.

Symptomatic knee OA was twice as prevalent in the rural Wuchuan cohort as it was in Beijing. The increased likelihood of developing chronic knee pain with increasing severity of radiologic changes in the knee joint is well established. Although the overall prevalence of radiographic OA was similar in Wuchuan and Beijing, the percentage of those with radiographic OA who had knee pain was much higher in Wuchuan. We ascribe this, in part, to the greater severity of radiographic disease in Wuchuan, but we also believe that the greater physical demands on persons in the rural Wuchuan environment are likely to produce more knee pain in those who already have OA. We should note that, unlike recent studies in North America, no one in Wuchuan was excluded because they had had a knee replacement, despite the high prevalence of severe radiographic disease that was symptomatic. Of further interest is that only 1 participant in the Beijing OA cohort had a knee replacement, and only 8 subjects in the Framingham cohort (1983–1985) had either bilateral or unilateral knee replacements. These small numbers of excluded participants would not influence prevalence comparisons.

Several characteristics of the Wuchuan OA study should be noted. A rigorous sampling strategy was utilized and the response rate was excellent (91%). In this farming community, the most common reason for declining participation was lack of time. In Wuchuan, this phrase is often used to avoid saying no and it tells very little about the actual cause, although this phrase was mostly used by the younger age group who were still working. Among participants age <65 years, 93% reported that they were currently working, compared with 54% of the older participants.

The radiographic and symptomatic knee OA criteria were standardized to those used in the Beijing OA study and the Framingham OA study (K/L grade ≥2, and reported knee pain most days of 1 month during the past year). Furthermore, the chief investigator of the study (XK) was trained in radiographic measurement on scored radiographs from the Beijing OA study.

A possible limitation of the Wuchuan OA study is the use of semiflexed knee radiographs, as opposed to the traditional, fully extended knee radiographs mostly used by the Beijing OA study. This decision was made because the semiflexed radiograph provides a better measurement of the functional joint space width compared with the fully extended knee radiograph, and has demonstrated superior long-term test–retest measurement reliability (16). We should note, however, that the last 700 subjects in the Beijing OA study had knee films obtained using the same semiflexed method as was used here, and prevalence estimates using this approach versus the fully extended approach, also obtained in these subjects, were the same. Therefore, the semiflexed protocol should not have changed the radiographic prevalence estimates reported.

It should also be noted that the knee OA prevalence estimates presented for these cohorts are conservative because radiographs did not include the patellofemoral joint. Furthermore, the prevalence demonstrated in this survey can only be generalized to similar farming communities in Northern China, not to the population of Inner Mongolia in general. Participants in the Wuchuan OA study were almost exclusively Han Chinese, but this ethnic group makes up only 79% of the Inner Mongolian population, with Mongolians accounting for most of the remaining 21%. The lifestyles of these 2 ethnic groups are very different, and therefore their prevalence of knee OA may also be very dissimilar.

In conclusion, we found a high prevalence of symptomatic and severe radiographic knee OA among older people in a rural farming region of northern China compared with their ethnic peers living in Beijing. The findings of the Wuchuan study suggest that having an occupation requiring heavy physical work is associated with knee OA, even in the absence of obesity. The actual magnitude of the rural–urban differential for this important chronic musculoskeletal disease in other regions of China or in other developing Asian countries would need to be confirmed in further studies. Interestingly, unilateral versus bilateral disease and medial compartment versus lateral compartment disease presentation demonstrated marked similarities between the 2 Chinese cohorts, which were in marked contrast to the disease presentation demonstrated in the Framingham OA study. In a similar fashion to other population-based surveys evaluating disease prevalence across geographic regions in the world, our study has generated hypotheses to important modifiable risk factors for OA. These factors have likely not yet been fully recognized in this specific population, and are worthy of further exploration in prospective studies. Heavy physical work, for example, is a risk factor that may be modifiable by access to more sophisticated farming equipment and possibly by changes in work postures and repetitive movements achieved through self-management education programs.

Given the high prevalence of symptomatic knee OA in Wuchuan, we suggest that knee OA represents a major public health concern in rural China. It is more prevalent there than in other reported population surveys. With the availability of knee replacements severely limited and occupational demands persisting into middle and older years, knee OA may be a major source of disability among Chinese adults in rural areas where most Chinese people still live.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Lin had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Kang, Fransen, Zhang, Li, Ke, Lu, Su, Song, Guo, Chen, Niu, Felson, Lin.

Acquisition of data. Kang, Fransen, Zhang, Li, Ke, Lu, Su, Song, Guo, Chen, Niu, Felson, Lin.

Analysis and interpretation of data. Kang, Fransen, Zhang, Li, Ke, Lu, Su, Song, Guo, Chen, Niu, Felson, Lin.


We would like to thank Dr. Bateer Er, Dr. Yuan Zhang, Dr. Yucheng Wang, Mr. Jixiang Li, Mr. Xiaoyong Guo, Ms Wenying Du, Ms Wei Wang, and Mr. Jinyou Wan for their great support in this study. Also, we wish to thank Dr. Piran Aliabadi for his help in teaching us how to score the radiographs.