To evaluate the recent epidemiology of hip fractures in the US.
To evaluate the recent epidemiology of hip fractures in the US.
We identified hip fracture cases from the 2008 Nationwide Emergency Department Sample, which contains more than 28 million emergency department (ED) records.
In 2008, approximately 341,000 (95% confidence interval 323,000–358,000) patients visited EDs with hip fractures. Of those, 90% were age >60 years. Between ages 60–85 years, the risk of fracture doubled for every 5- to 6-year increase in age. However, the hip fracture risk increased slowly after age 85 years. The overall trochanteric-to-cervical fracture ratio was nearly 2:1. The risk of trochanteric fracture increased faster with age compared with the risk of cervical fracture. At age 85 years, the rates of trochanteric and cervical fractures (per 100,000) were 1,300 and 700, respectively, among women and 800 and 500, respectively, among men.
The slowed growth of hip fracture risk after age 85 years suggests that the eldest old group may have a distinct hip fracture risk. Our study showed that trochanteric fractures were twice as common as cervical fractures. Because trochanteric fractures are more closely related to severe and generalized bone loss than cervical fractures, we hypothesize that the high incidence rate of trochanteric fractures in the US suggests that osteoporosis is a health problem that is linked to hip fracture. In addition to improved safety measures to reduce falls, rigorous preventive treatments of osteoporosis may be needed.
Hip fractures, frequently encountered in older adults, are associated with osteoporosis. More than 90% of hip fractures among older adults are caused by falls (1). Due to their frequency, hip fractures are one of the most challenging public health concerns, often leading to considerable consequences, including disability (2) and premature death (3, 4). Ongoing pain and limited physical mobility can severely compromise quality of life (5). Approximately 20% of US Medicare patients die within 6 months of sustaining a hip fracture (6). In addition to morbidity and mortality, hip fractures impose an enormous economic burden on the health care system (7, 8).
Hip fractures can be classified into 2 major anatomic types: trochanteric and cervical. It has been suggested that these 2 fracture types may have different risk factors (9). Low bone mineral density increases a person's likelihood of sustaining a hip fracture in a fall and is a well-established risk factor (10–14). In particular, trochanteric femur fractures are more closely associated with severe and generalized bone loss than fractures in the cervical region (15–19). Structurally, the trochanteric region has more trabecular components than the cervical region (20) and trabecular bone tends to be more sensitive to changes in estrogen level and subject to more rapid demineralization than cortical bone. As a result, later menopausal age is protective for trochanteric fractures but not for cervical fractures (21). Indeed, bone structure analysis revealed significantly lower trabecular bone volume in patients with trochanteric fracture than in patients with cervical fractures (15, 22). Patients with trochanteric fractures had significantly lower serum vitamin D levels compared with those with cervical hip fractures or the control group (23). This fact suggests that a lower serum vitamin D level may be a risk factor for trochanteric fracture but not for cervical fracture. In addition to the differences in physiopathologic processes leading to trochanteric and cervical fractures, individuals with a trochanteric fracture are hospitalized longer (17) and have higher postfracture mortality (24) compared to those with a cervical fracture.
Despite evidence that these 2 types of hip fractures have distinct underlying pathologies and consequences, the occurrences of trochanteric and cervical fractures in the US are not well understood. Understanding the incidences of these 2 types of hip fractures would benefit development of hip fracture prevention strategies. The emergency department (ED) is largely a point of entry to medical care for patients with hip fractures. Therefore, ED visits may best reflect the incidence of hip fractures in the population. To date, there are no published studies analyzing ED visits for hip fractures in the US.
The objective of this study was to estimate the incidence rate of hip fractures (trochanteric and cervical) based on representative ED records in the US. Additionally, although it is well known that the incidence of fracture exponentially increases with age (25), neither the age interval of the exponential growth nor the age-related increase in hip fracture rates is well documented. We modeled age- and sex-specific hip fracture rates. Using this updated information, we estimated the future burden of hip fractures with respect to the aging population.
This is the first study of hip fractures based on representative emergency department samples in the US.
This study estimated the incidence rates of trochanteric and cervical fracture.
Overall, the trochanteric-to-cervical fracture ratio was 2:1 in the US. Because the trochanteric region of the femur is more vulnerable to severe and generalized bone loss than the cervical region, our study suggests that osteoporosis is a health problem linked to hip fracture.
We studied the annual frequency of hip fractures in the US using the 2008 Nationwide Emergency Department Sample (NEDS) database. Our institutional review board (IRB) determined that this study did not involve human subject research and that no IRB review was needed.
The NEDS is a database of nationally representative hospital-based ED records collected annually under the auspices of the Agency for Healthcare Research and Quality (AHRQ). The objective of NEDS is to document national estimates of ED visits in the US. Details concerning sampling procedures have been published elsewhere (26). Briefly, this annual survey selects approximately 20% of a stratified sample of hospital-based EDs to produce a nationally representative sample. Stratification for the NEDS includes geographic region of the US, hospital ownership (for profit or not for profit), teaching hospital, urban-rural location, metropolitan area designation, and trauma center designation. In 2008, probability samples of 980 EDs were selected and all visits within the selected EDs were included in the database. As a result, probability samples of 28 million ED visits were selected from a total of 125 million ED visits in 2008.
We identified hip fracture cases on the basis of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 820 (fractures, femur). The individual anatomic regions of the hip fractures were further evaluated by the first subcategory of ICD-9-CM codes: 820.0 and 820.1 for cervical fractures, and 820.2 and 820.3 for trochanteric fractures. When the subcategory was not specified, we redistributed the missing values using the trochanteric-to-cervical ratio within each age.
The annual incidence rate indicates the risk for an individual in a specified population to sustain a hip fracture within a 1-year period. When the incidence (or occurrence) is divided by the number of people at risk, the incidence becomes the incidence rate (27) as follows:
In this study, the number of new events was the number of hip fracture cases that occurred during 2008. The population at risk is frequently estimated using the mid-interval population during the specified time period (27) and we used the population estimate as of July 1, 2008 (28). The base rate we used was 105, which is a frequently used base rate in published studies.
To estimate the cumulative risk of hip fractures over a lifetime, an annual age- and sex-specific risk of hip fracture was calculated first. To determine the age- and sex-specific annual incidence rates of hip fractures, we calculated the number of ED visits in 2008 for every 100,000 people in each respective demographic group on the basis of the US Census (28). The probabilistic risk of hip fractures over a lifetime is then calculated by adding an annual age- and sex-specific risk of hip fracture. To calculate the risk of hip fracture over the entire lifespan, we assumed that the age- and sex-specific fracture rates would remain the same.
We considered the sampling weight and sampling design in estimating the total number of cases and their SEs. Sampling weights provided by AHRQ were used to account for the unequal sampling probabilities and to produce estimates for all hospital admissions in the US. The SE was estimated using Taylor's linearization. As suggested by the National Center for Health Statistics, a relative SE in excess of 30% is considered to be unreliable.
To reflect the nonlinear growth of hip fracture risk with respect to increasing age, we fitted an exponential equation to the data from which the percentage increase of fracture risk was calculated. The exponential function fitted was y = α exp(βx), where y is the rate of hip fracture, x is the numerical age, and the “exp” (approximately 2.718) is the base of natural logarithms. The constant a is the vertical stretch factor, whereas the constant β is the horizontal stretch factor. The fit of the model was evaluated by the coefficient of determination and the residual plot.
In 2008, nearly 341,000 ED visits (95% confidence interval 323,000–358,000) in the US were for hip fractures. Of those, approximately 0.26% (or estimated 910 cases) were open hip fractures. Due to the small sample size, we could not further evaluate or comment on open hip fractures. Hip fractures were the second most common fracture in the lower extremity (Figure 1), constituting 22% of lower extremity fractures. Unlike other lower extremity fractures, as shown in Figure 1, hip fractures were considerably more common in women (68%) than in men (32%).
Nearly 94% of hip fracture cases (n = 321,000) occurred among those age ≥50 years. Because hip fractures were rare events for young adults, we limited further evaluation to patients age ≥50 years in order to report reliable age- and sex-specific estimates. To investigate the patient burden in EDs due to hip fractures, we first estimated the frequency of ED visits for each age without an adjustment to the population size. The vast majority of hip fracture patients in EDs were women in their 80s; more than 140,000 patients were in their 80s and 73% of those were women. As shown in Figure 2, the number of ED visits for hip fractures rose steadily in adults as age increased, with the peak occurring at age 86 years. Among hip fracture cases, nearly 44% of women and 34% of men were age ≥85 years.
To estimate the risk of fracture, we calculated age- and sex-specific fracture rates taking into account their respective US population sizes (28). The risk of hip fracture increased exponentially for women between ages 60–85 years (R2 = >0.99) and for men between ages 60–90 years (R2 = >0.99). The fitted lines shown in Figure 3 are as follows: hip fracture rate = 0.0288 × exp(0.1317 age) for women and hip fracture rate = 0.0362 × exp(0.1225 age) for men.
Among women, the hip fracture rate (per 100,000) at age 60 years was 78 (= 0.0288×exp[0.1317×60]) and the rate of fracture increased by 14% (= exp[0.1317]) for every 1-year increase in age. That is, compared with women age 60 years, the risk of sustaining a hip fracture was nearly 4 times greater for women age 70 years, 14 times greater for women age 80 years, and 27 times greater for women age 85 years.
Among men, the rate of hip fracture (per 100,000) at age 60 years was 56 and the rate of fracture increased by 13% for every 1-year increase in age. For both women and men, the risk of fracture nearly doubled for every 5- to 6-year increase in age between ages 60–85 years. At age 85 years, the rates of hip fracture for women and men were nearly 2,000 and 1,300 (per 100,000), respectively (Figure 3).
The exponential model was not adequate for describing trends for women age >85 years and for men age >90 years, due to a slower rate of increase. However, the risk of fracture increased steadily, with the highest peak occurring among those between ages 93–96 years. Nearly 3,100 and 2,400 (per 100,000) women and men had hip fractures, respectively, in this age group (Figure 3).
Until age 64 years, the cumulative risk of hip fracture was less than 1%, although the risk grew rapidly with age (Figure 4). Given an average lifespan of 80 years, unless fracture prevention programs improve, approximately 10% of women and 6% of men in the US will experience a hip fracture during their lifetime. Approximately 19% of women and 12% of men who reached age 85 years sustained hip fractures, and nearly 30% of women and 20% of men who reached age 90 years sustained hip fractures.
The overall ratio of trochanteric-to-cervical fractures was approximately 2:1. The rate of trochanteric fractures was substantially larger than that of cervical fractures for women and men of all ages (Figure 5). Age was a particularly strong risk factor for trochanteric fractures compared with cervical fractures. At age 90 years, the rates of trochanteric and cervical fractures (per 100,000) were 1,950 and 850, respectively, among women, and 1,450 and 650, respectively, among men.
To approximate the incidence of ED visits for hip fractures, we applied the 2008 age- and sex-specific fracture rates to the US population size projected for 2020, 2030, and 2040 (29). Hip fracture rates declined from 1997–2003 in the US (30, 31). However, no decreasing pattern is apparent among women since 2004 (31); it is not clear if the fracture rates have now stabilized. In an effort to calculate a conservative projection, we assumed a 20% decrease in the hip fracture rate compared to 2008 before the rate apparently stabilized. As shown in Figure 6, if the hip fracture risk is 20% lower than that in 2008, approximately 437,000 ED visits due to hip fractures can be expected in 2030. In 2040, when baby boomers are between ages 76–94 years, approximately 582,000 ED visits due to hip fractures are expected.
Our study documents the sex- and age-specific incidence rate of hip fracture in the US. Although a decline in the age-adjusted incidence rate of hip fractures was reported between 1997 and 2003 among inpatients (30, 31) and US Medicare patients (32), the rate of hip fracture in the US is still higher than in non-Scandinavian countries (33–35); this finding is consistent with a previous review study (36).
Overall, the trochanteric-to-cervical fracture ratio was 2:1 in the US. In Spain, trochanteric fractures are slightly more common than cervical fractures (54% versus 46%) (34). However, cervical fractures are more common in China (37), women in Taiwan (38), Sweden (39), Norway (40), and Finland (41). It is well understood that the trochanteric region of the femur is more vulnerable to severe and generalized bone loss than the cervical region (10, 15–19), and we hypothesize that the increase in the number of trochanteric fractures in the US suggests that osteoporosis, as a significant health problem, is linked to hip fracture. However, we also acknowledge that this phenomenon could be explained by multiple factors, including ethnicity, diet, completeness of diagnostic coding in the database, and femur geometry. The association between the proximal femur geometry and fracture type was previously reported (42, 43), yet results were conflicting.
Our study found that trochanteric fractures were more common than cervical fractures for all ages and both sexes. Two decades ago, a study based on the Medicare population reported that the ratio of trochanteric-to-cervical fractures increased with age among white women (44); our study confirms this finding. In addition, the authors reported that cervical fractures were more common among white women age <80 years (44); however, our study found that trochanteric fractures were more common in women and men of all ages. Further investigation is needed to determine whether our findings reflect a change in the occurrence of hip fractures by anatomic location compared to 2 decades ago.
The eldest old group is an understudied group, and our study suggests that this age group had a distinct hip fracture risk. In general, the risk of sustaining a hip fracture grows exponentially with age among the elderly up to age 85 years. However, the exponential function did not accurately describe trends among people age >86 years. Previously, Couris et al reported the same finding on the basis of inpatient data in France; they concluded that an exponential model was not adequate to describe the increase in the incidence rate among women age ≥85 years (45). As Couris and colleagues noted, we believe that women ages 85–95 years who are still at risk for hip fracture may represent a population with better overall health and a lower risk of hip fracture. However, the reasons for the decrease in hip fracture rates among those age >95 years are not clear. It is possible that many people in this age group are unable to get out of bed without assistance, use a wheelchair, and are generally inactive and therefore have less opportunity to fall. Given that 40% of hip fractures occur in individuals age ≥85 years, an accurate description of the risk for this age group will add to our understanding of the future burden of hip fractures in the US.
A number of prevention strategies have been proposed to slow bone loss (46–49). The age-adjusted rate of hospitalization for hip fractures has decreased between 1997 and 2003 in the US (30, 31), and researchers have speculated that the decline in fracture incidence rates could be, in part, a result of the introduction of several bisphosphonates (30, 32). Despite the reported decreases in hip fracture incidence rates in the US, our study showed that hip fractures remain a growing burden in the US primarily due to the rapidly expanding elderly population. Hip fractures occurred mainly among those age ≥75 years. Compared with 2008, by 2040 it is expected that there will be 26 million more people age ≥75 years (from 18.6 million to 44.6 million) (29). This expanding elderly population will inevitably increase the number of hip fractures and associated ED visits.
The strength of this study lies in the large sample size of the NEDS data, which constitute over 28 million visits selected with known probability. This sample was large enough to yield estimations for specific ages (not age groups) for women and men. There are, however, several limitations of this study. First, while it has been reported that whites have a 2- to 3-fold higher risk for hip fractures (50), the 2008 NEDS database does not include patients' race or ethnicity. Therefore, we were not able to stratify our results according to race or ethnic group. Second, although it is possible that an individual patient may present with multiple hip fractures during the same year, we could not estimate how many patients were returning to the ED with a second hip fracture in 2008. Third, although the ED visits may best reflect the incidence of hip fractures, we also recognize that not all cases of hip fractures are seen in an ED. For instance, if a patient who sustained a hip fracture died from concurrent trauma at the scene, that patient may not be brought to the ED. If a patient with a hip fracture was directly admitted to a hospital, this patient would not be captured in the ED database. However, we cannot confirm the extent or frequency of such occurrences. Finally, the accuracy of the ICD-9-CM codes submitted by each ED is unknown. Although large databases are prone to some degree of error based on the accuracy of data entry and diagnostic coding, we believe these databases are still valuable for insight into ED utilization for hip fractures.
In conclusion, the risk of hip fracture grew exponentially among women and men ages 60–85 years; the risk of fracture almost doubled for every 5- to 6-year increase in age. Surprisingly, the risk of trochanteric fracture, a strong indication of osteoporosis, increased significantly faster with age compared with the risk of cervical fracture. Rigorous safety measures to prevent injurious falls, and treatments aimed at preventing osteoporosis, would benefit men and women in this age group. Despite prior reports showing decreased hip fracture incidence rates in the US, as baby boomers age, an expanding elderly population will inevitably result in an increase in the number of hip fractures, the number of ED visits, and the demands for inpatient care.
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. Kim 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. Kim, Meehan, Blumenfeld, Szabo.
Acquisition of data. Kim.
Analysis and interpretation of data. Kim, Meehan, Blumenfeld, Szabo.
The authors thank Dr. Betty Guo at University of California, Davis, School of Medicine, Office of Research, for commenting on an earlier draft of the manuscript.