Prevalence, Incidence, and Epidemic Trends of Osteoarthritis Among Adults in Beijing, China, From 2008 to 2017

Objective To estimate the prevalence, incidence, and epidemic trend of diagnosed osteoarthritis from 2008 to 2017 in Beijing, China. Methods This was a retrospective serial cross-sectional study. We used health-insurance claims of 17.7 million adults for 2008–2017 to identify people over 18 years-old with osteoarthritis. The population data for Beijing from China’s 6th national census were applied to standardize the results. Trends in prevalence and incidence were analyzed using joinpoint regression. Results We identied 2,793,467 people with osteoarthritis between 2008 and 2017, 60% of which were women. The 10-year average age-standardized prevalence and incidence of osteoarthritis in Beijing was, respectively, 5.2% and 28.5 per 1000 person-years. Prevalence increased with age, surging after 55 years-old. The average crude prevalence for this decade was 14.9% for people over 55 years-old: 10.7% for males and 20.4% for females. The prevalence showed an increasing trend from 2008 to 2017, including a period of rapid rise, with an annual percentage change of 44.3% from 2008 to 2011 (P (cid:0) 0.05); the increase in prevalence was greatest in people under 35 years-old, with the an average annual percentage change of 36.1% (P < 0.05). We observed that the average age-standardized prevalence of diagnosed osteoarthritis in Beijing per decade was at a low-to-medium level compared to worldwide levels. Annual prevalence increased signicantly from 2008 to 2017, with osteoarthritis being diagnosed at younger ages. More public health efforts are needed to prevent osteoarthritis in China.


Introduction
Osteoarthritis is a common chronic degenerative joint disease, which mainly manifests as pain, stiffness, and functional limitation of the affected joints, and can lead to disability [1]. The joints most vulnerable to osteoarthritis include the knees, hips, elbows, ankles, the lumbar spine, cervical spine, and hand joints.
Osteoarthritis affects 240 million people worldwide [2], and it is estimated that 9.6% of men and 18.0% of women over 60 years-old have symptomatic osteoarthritis, worldwide [3]. The years lived with disability (YLDs) caused by osteoarthritis ranked tenth in China in 2016 [4], and the growth rate of YLDs due to osteoarthritis ranked fth in Nordic countries [5]. Therefore, the disease burden caused by osteoarthritis is huge.
Many western countries have conducted epidemiological studies of osteoarthritis, including large-scale cross-sectional studies and longitudinal cohort studies [6][7][8]. However, in China, which is the largest developing country in the world, most epidemiological studies of osteoarthritis have had limited resources and used relatively small sample sizes [9,10]. Thus, they could not adequately assess the prevalence and incidence of osteoarthritis or its epidemic trends in recent years.
Beijing, the capital of China, is a metropolis of more than 20 million people, while aging, sedentary lifestyle, obesity, and other risk factors for osteoarthritis have been rapidly increasing [11,12]. What is the prevalence and incidence of osteoarthritis in the this city in recent years? Is it increasing year by year? We do not currently know the answer to these questions. Beijing has long-term detailed medical data, giving us the opportunity to observe a panoramic view of the This was a retrospective serial cross-sectional study. All bene ciaries aged ≥ 18 years-old in the BMCDE from 1 January 2008 to 31 December 2017 were included in the study. Bene ciaries with lack of personal information or unclear diagnosis of disease were excluded. Information from the same individual could be linked anonymously using an encrypted number.

Statistical analysis
Prevalence was calculated by dividing the number of people with osteoarthritis each year by the average number of bene ciaries that year (bene ciaries are calculated by taking the average number of bene ciaries on January 1 and December 31 of each year). Incidence was calculated by dividing the number of people newly diagnosed with osteoarthritis per year by the total person-years of all bene ciaries that year (those who already have osteoarthritis have been subtracted from the bene ciaries). The annual prevalence and incidence rates were standardized through age adjustment, using population data for Beijing from the sixth national census.
We used joinpoint regression to analyze trends in prevalence and incidence rates [15]. Joinpoint regression is a common method to investigate epidemiological time trends that has been used in many studies [16]. This model aims to establish piecewise regression according to the time characteristics of disease distribution, dividing time into different intervals (or phases) through several connection points, and conducting trend tting and optimization for each interval, so as to evaluate the disease change characteristics of different intervals within the whole time range in more detail. The term joinpoint in the model refers to a signi cant change in the direction or size of the linear trend found by permutation tests [16]. A log-linear regression model was used in each phase: where y refers to the prevalence or incidence at year x. Annual percent Change (APC), Average Annual percent Change (AAPC), and 95%CIs are the main results of joinpoint models, representing the percentage change in prevalence or incidence in a given year compared to the previous year within each interval and over the whole study period, respectively. APC was calculated based on the slope in each phase, using the following formula: The parameter calculation method of AAPC was used to calculate the regression coe cient of each phase by weighting the span width w of the segmented interval [17], using the formula: We analyzed differences using t-tests for numerical variables and chi-square tests for categorical variables. The Z-test was used to analyze the statistical signi cance of APC and AAPC, with non-signi cant changes indicating stable trends [17]. A two-tailed P<0.05 was considered statistically signi cant. All statistical analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC) and the Joinpoint Regression Program (Version 4.8.0.1) from the Surveillance Research Program of the US National Cancer Institute.

Patient characteristics
Of the 17.7 million participants, 2,793,467 (16%) people with osteoarthritis were identi ed between 2008 and 2017. Most patients were women (60%), 55-64 years-old (27%), retired (58%), diagnosed in tertiary hospitals (53%), and lived in the city (75%). The characteristics of patients with osteoarthritis are shown in Table 1. Note: * Age group by age at diagnosis; ‡ P-value for sex differences in the characteristics of people with osteoarthritis.
Prevalence and incidence of osteoarthritis Table 2 shows the annual age-standardized prevalence and incidence of diagnosed osteoarthritis in Beijing from 2008 to 2017. 1.8%, and the highest point appeared in 2015, which was 7.3%. The annual age-standardized prevalence of women was higher than that of men between 2010 and 2017. The prevalence of osteoarthritis gradually increased with age, and the rate of increase became greater after 55 years-old (Fig. 1). The average crude prevalence of the decade was 14.9% for people over 55 years-old; 10.7% for males and 20.4% for females. Trends in the prevalence and incidence of osteoarthritis Table 3 shows the trends in the age-standardized prevalence and incidence of diagnosed osteoarthritis in Beijing from 2008 to 2017. The prevalence increased signi cantly with an AAPC of 16.4% (P < 0.05) during this decade. There were two different trends in prevalence during this period and one joinpoint in 2011. The period from 2008 to 2011 was a phase of rapid increase, with an APC of 44.3% (P < 0.05). This increasing trend in prevalence was observed among all age groups, but the AAPC was greater in people under 35 years-old (36.1%, P < 0.05). The annual crude prevalence of osteoarthritis in people under 35 yearsold from 2008 to 2017 is shown in Fig. 2.  Note: 95% CI, 95% con dence interval. *P < 0.05 was considered statistically signi cant.
(Please place the table under the paragraph about "Trends in the prevalence and incidence of osteoarthritis") The age-standardized incidence of osteoarthritis remained stable from 2008 to 2017, with a non-signi cant AAPC of 1.2% (P > 0.05), and no signi cant joinpoint. Joinpoint regression showed that the AAPC in the incidence of patients aged 18-34 years-old was greater than that of patients aged 35-44 yearsold (22.7% vs. 18.7%). Moreover, there was a joinpoint in 2012 for both these age groups. The period from 2008 to 2012 was a phase of rapid increase, with an APC of 53.8% (P < 0.05) in the 18-34 age group and 54.9% (P < 0.05) in the 35-44 age group.

Discussion
The present study is the rst study to our knowledge to describe trends in the prevalence and incidence of osteoarthritis in China. Using data from the healthinsurance claims of nearly 18 million people, we found that the 10-year average age-standardized prevalence and incidence rates of osteoarthritis in Beijing were, respectively, 5.21% and 28.54 per 1000 person-years. Furthermore, prevalence increased signi cantly within a decade. In addition, we also found that the AAPC in the prevalence and incidence during the decade was the largest in the low-age group (18-34 years), indicating that osteoarthritis was diagnosed at younger ages.
Our results showed that the age-standardized prevalence of osteoarthritis in people over 18 years-old was 5.21%, which is close to the ndings of the Australian scholar Minaur (5.5%) [18]. The highest prevalence of osteoarthritis was 20.7% [19] and the lowest was 1.6% [20] in relevant and comparable studies. Compared with related studies in other countries, the prevalence of osteoarthritis in Beijing is at a low-medium level. We attribute the difference in prevalence among studies to differences in race, life and work styles, and other variables. Furthermore, the prevalence observed in the present study increased signi cantly within a decade, which is similar to the change in prevalence observed in other countries, such as the United States and the United Kingdom [6,21]. The reason for this may be the aging of the population and increased risk factors for osteoarthritis (such as low physical activity and high body mass index) in recent years [22][23][24]. In addition, the present study found the AAPC in the prevalence of osteoarthritis was greatest in the age group under 35 yearsold. We should be alert to the phenomenon that the rate of osteoarthritis is increasing among young people, not only among older adults. More efforts are needed in the future to prevent osteoarthritis in young people.
Previous studies have shown that the prevalence and incidence of osteoarthritis are higher in women than in men [25][26][27], which is consistent with the results of the present study. This phenomenon may be attributed to differences in hormone levels, muscle strength, and health-seeking behavior between the genders [28][29][30]. Estrogen levels in postmenopausal women are signi cantly lower than premenopausal and male levels, which may affect cartilage metabolism and change the mechanical environment of joints [31]. Moreover, men and women have different sensitivity and tolerance to disease, and women are more likely to seek timely medical treatment than men are [32]. This suggests that men may be less likely to be diagnosed with osteoarthritis, which is re ected in the lower prevalence and incidence among men in this study.
It is known that the prevalence and incidence of osteoarthritis increases with age. Our study showed that the prevalence and incidence of osteoarthritis increased signi cantly after 55 years-old, and that the average crude prevalence from 2008 to 2017 of people over 55 years-old was 14.86%: 10.65% in males and 20.36% in females. These results are consistent with the results of other studies [33][34][35]. The reasons may include the following [30,36,37]: (i) the cell functions and properties of articular cartilage change with age and it responds differently to cytokines and growth factors; (ii) the articular cartilage secretes less synovial uid with age, which reduces lubrication of the joints; (iii) muscle strength is reduced with age, so it is di cult to support the surrounding articular cartilage, thereby accelerating cartilage wear; and (iv) after menopause, changes in hormone levels can cause bone hyperplasia and accelerate the onset and progression of osteoarthritis. However, in our study, the prevalence and incidence of osteoarthritis in people over 85 years-old were lower than they were in the 55-64 and 65-74 age groups. The explanation for this may be related to an increased comorbidity rate and a decreased rate in medical visits due to osteoarthritis in this group of people. Another possible explanation for this phenomenon is survivor bias, which allows relatively healthy people to survive to the oldest age group.
This study has certain strengths. Currently, there is a lack of epidemiological studies on large samples of people with osteoarthritis in China. We used data from the health-insurance claims of nearly 18 million people to estimate the prevalence and incidence of osteoarthritis in Beijing in the past 10 years, and determined the prevalence trend for osteoarthritis. These ndings about osteoarthritis could provide valuable evidence for other developing cities in China, and even in cities in other countries in the future. Moreover, the estimates of prevalence and incidence were based on a dynamic population, which is closer to realworld population changes.
The limitations of this study include the following. Since the data were based on hospital visits, we could not obtain information about people who suffered from osteoarthritis but did not see a doctor. This may result in an underestimation of prevalence and incidence rates. In addition, although the BMCDE covers more than 80% of the resident population of Beijing, we could not obtain information about some immigrants in this city, which may cause selection bias.
Finally, because the BMCDE includes only claims data, information on socioeconomic status and health behaviors was not available. Therefore, we could not do a more detailed analysis, such as the identi cation of risk factors.

Conclusion
Using data from the BMCDE, we observed that the average age-standardized prevalence of diagnosed osteoarthritis in Beijing over a decade was at a low-tomedium level relative to world levels. Furthermore, our analyses showed that the annual prevalence rate increased signi cantly from 2008 to 2017, and that osteoarthritis was being diagnosed at younger ages. Therefore, we should pay attention to the younger trend of osteoarthritis.

Con icts of Interest
The authors declare that they have no con ict of interest.