By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Please be advised that we experienced an unexpected issue that occurred on Saturday and Sunday January 20th and 21st that caused the site to be down for an extended period of time and affected the ability of users to access content on Wiley Online Library. This issue has now been fully resolved. We apologize for any inconvenience this may have caused and are working to ensure that we can alert you immediately of any unplanned periods of downtime or disruption in the future.
Hip fracture is the most fatal, debilitating, and expensive consequence of osteoporosis. Cooper and colleagues1, 2 estimated that the number of hip fractures world-wide would increase from 1.66 million in 1990 to 6.26 million in 2050, about half of which would occur in Asia. Those projections assumed that the age-specific rates of hip fractures around the world would not change, but the rates of hip fracture may be declining in North America3, 4 and Europe5–8 while rising in Asia.9, 10 Inaccurate reporting of hip fractures, however, can produce misleading trends in hip fracture rates. For example, in a previous study we found that Beijing public health records of hip fractures in 1990 to 1992 underestimated the actual number of hip fractures by 65% because discharge diagnoses were miscoded.11 Improvements in the accuracy of coding and patterns of care for hip fracture would produce misleading estimates about temporal changes in rates of hip fracture. To determine whether age-specific rates of hip fracture rates are increasing in Beijing, China, we systematically confirmed cases of hip fracture and compared the age incidence of hip fracture in Beijing from 2002 to 2006 to rates from a similar study from 1990 to 1992 that also validated hip fracture cases.
Materials and Methods
Municipal hospitals in the greater municipal Beijing area report hospital discharge data to the Beijing Department of Public Health. We obtained those data for the period from January 1, 2002, to December 31, 2006 for all 124 hospitals in the Beijing municipal area that reported any hip fractures. We also obtained discharge data from the People's Liberation Army Health Information Center for all 15 military hospitals that treat both military and civilian patients.
Municipal hospitals used the 10th Revision of the International Classification of Diseases (ICD-10) system for coding discharge diagnoses; we included all cases coded as S72.0 (femoral neck fracture), S72.1 (intertrochanteric fracture), or T93.1 (hip fractures diagnosed within the past 12 months). Military hospitals used the 9th Revision (ICD-9) and we included cases coded as 820 (fracture of neck of femur). We also recorded the patient gender, age or date of birth, place of residence. We excluded data from 2003 because the severe acute respiratory symdrom (SARS) epidemic closed several hospitals and may have caused incomplete recording of discharge diagnoses.
Estimating the accuracy of the reported rates
We considered several potential sources of error in the hip fracture data: reporting fractures from patients who were not residents of metropolitan Beijing, incomplete reporting of hip fractures from discharge records to the Department of Public Health, miscoding other fractures as hip fractures in discharge records, and miscoding of hip fractures as a different type of fracture.
As it was not feasible to confirm all cases in all Beijing hospitals we did so in a sample of the hospitals. As in the prior study, we included one orthopedic hospital that reported 1970 cases, more than twice as many as the next largest hospital. We then randomly sampled the remaining public hospitals until we included hospitals with least 10% of the 19,003 remaining hip fracture cases (a total of 2255 [11.9%]). It was not feasible to conduct confirmatory studies in the military hospitals. Investigators' team visited all hospitals in the sample and compared all of the cases in the lists from the Department of Public Health against the original discharge logs. To estimate the proportion of cases that had been overreported, we counted the number of cases who did not reside in Beijing, could not be found in the discharge logs, or who did not actually have a hip fracture diagnosis. To confirm the diagnosis for the remaining cases in the discharge logs, we reviewed the operative and radiograph reports in the original medical records for a 20% random sample of all discharged cases. To estimate the number cases that were missed by the public records, we obtained and reviewed the medical records with operative and radiographic reports containing discharged with ICD-10 codes that might include hip fractures, such as those miscoded as femoral shaft fractures (S72.3, 821) or found among multiple fractures (S32.7, S72.7, T07.5, T07.6, T02.7, T02.8, T02.9).
We estimated the number of hip fractures in Beijing as the product of the total reported cases by public and military records multiplied by the ratio of the (confirmed cases + missed cases) ÷ all reported cases.
In the 1990 to 1992 study we also systematically confirmed hip fracture cases that were reported to the Beijing Department of Public Health. Similarly to the current study, we drew a random sample of Beijing hospitals and then compared cases reported in public records to original discharge diagnoses, and operative and radiographic reports in medical records and then adjusted the overall rates by the proportion of underreporting of cases.11
Demographic data for the calculation of sex- and age-specific rates were obtained for the Beijing Metropolitan Area from the website of the Beijing Municipal Bureau of Statistics from the 2004 national census and the 1990 census. Demographic data for standardization to the population of the United States in 2004 were obtained from the Website of the US Census Bureau12 (http://www.census.gov/popest/national/asrh/NC-EST2005/NC-EST2005-01.xls).
Incidence rates of hip fracture per 100,000 persons were calculated for each sex and 5-year age group by dividing the corrected numbers of hip fractures by the number of Beijing residents in 2004.
To estimate confidence intervals for the numbers of hip fractures, we first calculated a confidence interval for the ratio of the confirmed plus missed cases to all reported cases, based on the size (N = 4225) of the validation sample, using the normal approximation for the binomial sampling distribution. We then multiplied the overall numbers of reported cases by the resulting confidence bounds, 0.808 and 0.832. To calculate confidence intervals for rate ratios, we used a Bonferroni procedure, calculating the lower confidence bound as the ratio of the lower bound for the numerator to the upper bound for the denominator, and the upper bound as the ratio of the upper bound for the numerator to the lower bound for the denominator. In calculating rate ratios with respect to 1990 to 1992, we assumed that those rates were estimated with the same relative precision as the 2002 to 2006 rates.
Because the age structure of Beijing may have changed between 1990 and 2006, the rates used in calculating the ratios were standardized to the 2004 US Census.12 To compare the overall rates of hip fractures in women and men in Beijing to the rates in other countries we also standardized the rates for other selected countries to the 2004 US Census where reports provided sufficient data.9, 10, 12–19
The public health and military information centers reported 20,973 hip fractures in Beijing in 2002, 2004 to 2006 among persons aged 50 years and older, including 4225 hip fractures from the sample of hospitals. Of those 4225, 714 (17%) were excluded because they occurred in nonresidents of Beijing: 517 (26%) of the 1970 cases from the large orthopedic hospital and 197 (9%) of the 2255 cases from the other hospitals in the confirmation sample. We used this 9% rate to adjust the overall incidence for all Beijing hospitals. Additionally, 72 (1.7%) of the cases in the Department of Public Health records were found in discharge logs but did not have a hip fracture diagnosis or could not be found in the discharge records. We found 34 cases that had been missed because they were mixed into categories of multiple fractures or miscoded as femoral fractures. In sum, we found 3473 confirmed fractures in Beijing residents in the sample of hospitals, 18% fewer than reported from the Department of Public Health. Therefore, we estimated a total of 16,038 (95% confidence interval [CI], 15,804–16,272) hip fractures in those aged 50 years or older occurred during that period, 9693 (95% CI, 9558–9827) in women and 6345 (95% CI, 6253–6437) in men.
Temporal changes in age-specific rates
The age-specific incidence of hip fracture rose dramatically between 1990 and 1992 and 2002 and 2006, particularly among those over age 70 years (Fig. 1). Adjusted to the age distribution of the 2004 US population, the rates among those age 50 years and older increased 2.76-fold (95% CI, 2.68–2.84) for women and 1.61-fold (95% CI, 1.56– 1.66) for men. Among just those 70 years or older, the rate increased 3.37-fold (95% CI, 3.28–3.47) for women and 2.01-fold (95% CI, 1.95–2.07) for men.
The hip fracture rates for those over age 50 years also increased substantially comparing 2002 with 2006: 1.58-fold (95% CI, 1.54–1.63) for women and 1.49-fold (95% CI, 1.45–1.53) for men (Fig. 2).
Comparisons of hip fracture rates with other regions
Age-adjusted incidence rates for hip fracture, standardized to the 2004 US population, were 129 per 100,000 for men and 229 per 100,000 for women. Despite the increased rates of hip fracture in Beijing, the age-standardized rates of hip fracture for women in Beijing in 2002 to 2006 remained less than half of the recent rates observed in Western countries or Japan (Table 1).
Table 1. Rates of Hip Fracture Among Women and Men Aged 50 Years or More from Selected Published Studiesa
The age-specific rates of hip fracture in Beijing, China, especially among those age 70 and older, increased dramatically between 1990 and 1992 and 2002 and 2006. The rate of hip fractures continued to rise very rapidly from 2002 to 2006, about 10% per year. These estimates account for the changes in the accuracy of reporting of hip fracture cases in and referrals of cases from outside of the city in both periods.
The increased risk of hip fracture is probably because of changes in lifestyle with urbanization. Physical activity, primarily walking, is a strong protective risk factor for hip fracture, and urbanization of Beijing, and other Asian cities, is characterized by increased reliance on cars and buses instead of walking or biking. Indeed, the number of adults owning cars increased from 4 per 100 adults in 1992 to 18 per 100 adults in 2004.20, 21 Squatting—an exercise that strengthens leg and lower back muscles and may improve balance22—was a nearly universal feature of daily life before urbanization when there were fewer chair or couches and squat toilets were commonplace. During the past 10 to 20 years, squatting has become uncommon as most residents of Beijing have moved into Western style apartments with chairs and couches and Western-style sit toilets. As people have moved from single-unit housing with outdoor courtyards to multistory buildings, a decrease in sun exposure might have increased vitamin D deficiency.
The increases in rates were greater for the oldest women and men. The reason is not clear. Decreased physical activity may have a greater adverse effect in the oldest people who are more prone to hip fracture because of lower bone density. Life expectancy has increased in China23 and increased rates of hip fracture might reflect improved medical care resulting in longer survival of frail elders who are at increased risk of hip fractures.24 However, paradoxically, studies indicate that the health status of elderly in China improved between 1990 and 2002.23 The increased risk was also greater for women than men, now more closely resembling the greater ratio of women to men observed in Western countries; the reason for this pattern of change is not clear.
There are no data about changes in bone density during this period; however, it seems unlikely that these temporal changes could all be attributed to decreased bone density; because the more than threefold increase in risk of fracture observed in elderly women in Beijing would require more than a one standard deviation (more than −1 T score) decrease in femoral neck bone density.25 Obesity is expected to reduce the risk of hip fracture—and increase bone density. The increase in the rates of hip fracture have occurred despite an increase in obesity in people of all ages in China that has been more pronounced in Beijing.26
Our validation studies raise the possibility that some of the increases reported for developing cities or countries may be because of changes in the accuracy of recording and reporting cases of fractures of the proximal femur. We estimate that improvement in the accuracy of coding hip fractures during that period would have substantially overestimated the temporal increase in the rate of hip fracture based on public health data. In 1990 to 1992, the data from public records underestimated the rates of hip fracture by about 75% largely because of miscoding of intertrochanteric fractures. We also estimated that the official rates for 2002 to 2006 overestimated the number of hip fractures in Beijing by about 20% largely because of referral of cases from outlying areas to hospitals in Beijing. In the earlier study we found few patients admitted to Beijing hospitals from outside of the city limits. The combination of these trends would have substantially overestimated the temporal increase in risk of hip fracture.
We also considered other spurious reasons for increased age-specific rates of hip fracture. Rates might be misleadingly low in 1990 to 1992 if Beijing residents received care for hip fractures in hospitals outside of the Beijing metropolitan area; however, we surveyed all hospitals in districts adjoining Beijing in 1992 and found no such cases.11 The previous rates might be low if many people in Beijing were treated for hip fractures outside of hospitals. However, our 1992 study included a population-based survey of a random sample of 2113 women over age 50 with a 97% response rate, finding that 96% of previous fractures and all previous hip fractures had been treated by hospitals.11
The population of Beijing increased substantially between 1990 and 2004 because of as the Beijing city limits expanded to include suburbs that had also become highly urbanized. Along with the greater age-specific rates of hip fracture, the average annual number of hip fractures in people age 50 or older cared for by the Beijing Ministry of Health increased from 479 to 2423 in women and from 441 to 1586 in men from 1990 to 1992 to 2002 to 2004.
This study has several limitations. Our random sample of hospitals did not include military hospitals; however, they accounted for only 7% of reported hip fracture cases. The rates of hip fracture in Beijing may not be representative of rates in other Chinese cities and are probably higher than rates in rural areas. The accuracy of the Beijing census might have changed between 1990 and 2004 because of patterns of immigration into Beijing; however, in both periods we excluded, and then adjusted for hip fracture cases in patients who did not have a Beijing address. Our data derive from two surveys and, therefore, we are unable to determine how much of the change is because of a cohort effect: decreases in rates in successive cohorts or decreases because of changes in conditions between the two periods of time.24
Assuming no changes in the age-specific incidence of hip fractures around the world, Cooper2 estimated that 51% of the world's hip fractures would occur in Asia by 2050. However, the age-specific rates of hip fracture are decreasing in northern Europe and North America.1, 3–7 Our results, and a few others from Asia1, 9, 10 indicate these projections have substantially underestimated the number and proportion of hip fractures that will occur in Asia.
We conclude that hip fracture rate in Beijing is rising rapidly and the burden of treatment for hip fractures and importance of preventing them may be rapidly shifting from Western countries to urbanizing areas of Asia.
All authors state that they have no conflicts of interest.
This study was supported by a grant from The Ministry of Science and Technology of the People's Republic of China (National Public Welfare Research Program 2005DIB1J085 and National Key Technology R&D Program 2006BAI03B03). The Beijing Public Health Information Center collected data from all of the hospitals in the random sample hospital. The authors also appreciate the statistical assistance of Professor Han Shaomei and Mr. Xu Tao, Department of Epidemiology and Statistics, Peking Union Medical College.
Authors' roles: All listed authors have each made substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; participated in drafting the manuscript or revising it critically for content; and have approved the final version of the submitted manuscript. W-BX, LX, and SRC accept responsibility for the integrity of the data analysis.