The Geographic Incidence and Treatment Variation of Common Fractures of Elderly Patients Abstract Fractures of the hip, wrist, proximal humerus, and ankle frequently are observed among the elderly patient population in the United States. The Medicare patient population has shown dramatic geographic variation in the rates of these common fractures, with an increased incidence observed throughout the Southeast. Treatment (surgical versus nonsurgical) is also highly variable and dependent on the geographic location but not necessarily on the type of injury. Whereas regional variation in medical treatment may be attributed to variations in practice patterns, the etiology behind the dramatic variations in fractures is less well-defined and is likely multifactorial, related to environmental, occupational, genetic, or nutritional factors. O lder patients (>65 years) are known to be at risk for fracture because of their increased incidence of osteoporotic bone, poor balance, compromised vision, and delay- ed reaction times. 1-4 Additionally, many older patients have atrophic soft tissues, such that forces may be more easily transmitted to the un- derlying bony structures. 4 As a re- sult, this elderly group represents a substantial proportion of the pa- tients with fractures of the hip, wrist, proximal humerus, and an- kle. 5 Despite the well-known fre- quency of fractures in older individ- uals, little is known about the geographic differences between frac- ture rates and the geographic varia- tion in treatment of these common injuries. Authors have shown varying fracture patterns among elderly in- dividuals throughout the United States. In these limited patient pop- ulations, patients residing in the South were more likely than those in the Northeast to sustain a hip fracture. 6-10 However, researchers know little about the underlying eti- ology of this discrepancy and whether the discrepancy persists when patients change geographic lo- cation. 11 Additionally, the regional variation in treatment of these frac- tures is not well understood. Using Medicare data, Weinstein in the Dartmouth Atlas of Muscu- loskeletal Health Care (DAMHC) showed large geographic variations among the rates of the four most common orthopaedic fractures in the elderly (ie, hip, wrist, proximal humerus, ankle) despite similar ac- cess to medical care. 5 Additionally, geographic residence among patients sustaining these injuries strongly correlates with the type of fracture management (ie, surgical versus nonsurgical). 12 Scott M. Sporer, MD, MS James N. Weinstein, DO, MS Kenneth J. Koval, MD Dr. Sporer is Assistant Professor, Department of Orthopaedic Surgery, Rush Medical College, Winfield, IL. Dr. Weinstein is Chairman and Professor, Department of Orthopaedic Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH, and Professor, Dartmouth Medical School and the Center for the Evaluative Clinical Sciences, Dartmouth College, Hanover, NH. Dr. Koval is Professor, Department of Orthopaedic Surgery, Dartmouth-Hitchcock Medical Center. None of the following authors or the departments with which they are affiliated has received anything of value from or owns stock in a commercial company or institution related directly or indirectly to the subject of this article: Dr. Sporer, Dr. Weinstein, and Dr. Koval. Supported in part by NIAMS #U01- AR45444-01A1, the Robert Wood Johnson Foundation, the American Academy of Orthopaedic Surgeons, and the American Hospital Association. Reprint requests: Dr. Sporer, Rush Medical College, 25 N Winfield Road, Winfield, IL 60190. J Am Acad Orthop Surg 2006;14:246- 255 Copyright 2006 by the American Academy of Orthopaedic Surgeons. 246 Journal of the American Academy of Orthopaedic Surgeons Variation in the Rates of Fracture The entire methodology of the DAMHC may be found in its Appen- dix on Methods. Briefly, Medicare data from 1996 and 1997 were used to evaluate regional variation in the rates of fracture and surgical treat- ment for fractures involving the hip, wrist, proximal humerus, and ankle. Databases provided through the Health Care Financing Administra- tion were analyzed to determine the number of possible Medicare benefi- ciaries in a designated region, as well as demographic data for these indi- viduals (age, sex, and race). The Medicare Provider Analysis and Re- view (MEDPAR) File (hospital claims data) and Medicare Part B data were analyzed to determine the rates of utilization for fracture care of the hip, wrist, proximal humerus, and ankle. Hospital referral regions were defined according to the 1996 to 1999 DAMHC guidelines. 12 These hospital referral regions represent tertiary care facilities in which there is delivery of specific cardiovascular and neurosurgical procedures. The incidence of fracture of the hip, wrist, proximal humerus, and ankle among Medicare enrollees was determined from the Medicare Part B file utilizing physician-generated Current Procedural Terminology (CPT) codes (Table 1). These codes allowed all patients to be categorized into surgical and nonsurgical treat- ment groups. Procedures and diag- nostic tests were adjusted for patient age, race, and sex. The regional rates of fracture for the wrist, proximal humerus, hip, and ankle were then calculated, and maps of the ratio be- tween observed versus expected re- sults were generated. The expected number of fractures was calculated by multiplying the average fracture rate with the number of people in each Health Services Administra- tion population. In 1996 and 1997, Medicare enroll- ees sustained >450,000 fractures. The hip, wrist, ankle, and proximal hu- merus account for >400,000 of these fractures and represent approxi- mately 85% of all injuries observed in this patient population 12 (Figure 1). Fractures about the hip (intertro- chanteric and femoral neck) ac- counted for >213,000 injuries and were the most commonly observed types of fracture within the Medi- care population for that period. The incidence of hip fractures varied by a factor of 2, from 4.9 (Honolulu, HI) to 10.7 (Rome, GA) per 1,000 Medi- care enrollees (Figure 2). Four hospi- tal referral regions had rates of hip fracture at least 30% greater than the national average, whereas 6 re- gions had rates at least 25% below the national average. 12 Wrist fractures were the second most common fracture observed in the Medicare patient population, ac- counting for >96,000 injuries, 85% of them observed in women. The in- cidence of wrist fractures varied by a factor of 4, from 1.5 (Everett, WA) to 5.7 (Huntsville, AL) per 1,000 Medi- care enrollees (Figure 3). Twenty- nine referral regions had rates of wrist fracture at least 30% greater than the national average, whereas 61 regions had rates at least 25% be- low the national average. 12 Ankle fractures accounted for >46,000 injuries that occurred in the Medicare patient population in 1996 and 1997. The incidence of ankle fracture was far more varied throughout the United States, with certain regions of the country show- ing a more than sixfold variation in incidence, from 0.5 per 1,000 (Hono- lulu, HI) to 3.1 per 1,000 (Danville, PA) (Figure 4). Forty-six hospital re- ferral regions had rates of ankle frac- ture of at least 30% greater than the national average, whereas 79 regions had rates at least 25% below the na- tional average. Proximal humerus fractures ac- counted for 43,500 injuries in the Medicare patient population. Simi- lar to ankle fractures, the incidence of these injuries showed dramatic re- gional variation, from 0.3 per 1,000 enrollees (Jackson, TN) to 3.4 per 1,000 enrollees (Covington, KY). Fifty-six hospital referral regions had rates of proximal humerus fracture at least 30% greater than the nation- al average, whereas 111 regions had rates at least 25% below the nation- al average. 12 Treatment Variation Fractures can be treated either surgi- cally or nonsurgically. The decision to proceed with surgery is highly de- pendent on a number of factors, in- cluding the anatomic location, asso- ciated soft-tissue injuries, and the number of associated fractures. The great majority of patients with frac- tures of the hip undergo surgical in- tervention. Conversely, the majority of patients with fractures of the proximal humerus undergo nonsur- gical management. Other fractures, such as those of the ankle and wrist, often are treatable with either surgi- cal stabilization or cast immobiliza- tion, depending on the severity of in- jury (Figure 5). More than 98% of the Medicare patients who sustained a hip fracture in 1996 and 1997 were treated with surgical stabilization. Mobilization is difficult without surgery; therefore, nonsurgical treatment is generally re- served for patients with multiple co- morbidities and for injuries that pose a substantial surgical risk. W einstein reported that the mortality rates for patients undergoing surgery were 7% at 30 days and 25% at 1 year. 5 In con- trast, patients treated nonsurgically had a 17% mortality rate at 30 days and a 39% rate at 1 year. Wrist fractures were most com- monly managed with closed reduc- tion and cast immobilization. De- spite this trend, different regions of the United States were far more likely to incorporate surgical inter- vention in their treatment with ei- ther pins, screws, or external fixa- tion. The proportion of wrist fractures treated surgically varied by Scott M. Sporer, MD, MS, et al Volume 14, Number 4, April 2006 247 a factor of 10 from 5.1% (Greenville, NC) to 50.7% (Olympia, WA). In gen- eral, patients in the Northwest were more likely than patients in the Southeast to have surgical interven- tion (Figure 6). In seven regions, 40% of wrist fractures received surgical treatment; in 30 regions, only 10% of them received surgical treatment. 5 Similar to wrist fractures, most ankle fractures sustained by the Medicare population were treated nonsurgically. Again, patients in cer- tain regions of the country were more likely than others to undergo Table 1 CPT Codes for the Four Most Common Orthopaedic Fractures in the Elderly Hip Fracture CPT Codes 27235 Percutaneous skeletal fixation, femoral fracture, proximal, neck 27236 Open treatment, femoral fracture, proximal, neck, internal fixation/prosthetic 27244 Open treatment, inter/per/subtrochanteric femoral fracture, with plate/screw type implant 27245 Open treatment, inter/per/subtrochanteric femoral fracture; with intramedullary implant 27230 Closed treatment, femoral fracture, proximal end, neck; without manipulation 27232 Closed treatment, femoral fracture, proximal end, neck; with manipulation 27238 Closed treatment, inter/per/subtrochanteric femoral fracture; without manipulation 27240 Closed treatment, inter/per/subtrochanteric femoral fracture; with manipulation Ankle Fracture CPT Codes 27766 Open treatment, medial malleolus fracture, with/without internal/external fixation 27792 Open treatment, distal fibular fracture, with/without internal/external fixation 27814 Open treatment, bimalleolar ankle fracture, with/without internal/external fixation 27822 Open treatment, trimalleolar ankle fracture, medial/lateral malleolus; without fixation 27823 Open treatment, trimalleolar ankle fracture, medial/lateral malleolus with fixation 27826 Open treatment, fracture, weight bearing articular surface, distal tibia, with fixation; fibula 27827 Open treatment, fracture, weight bearing articular surface/portion, distal tibia, with fixation; tibia 27828 Open treatment, fracture, weight bearing articular surface, distal tibia, with fixation; fibula and tibia 27829 Open treatment, distal tibiofibular joint disruption, with/without internal/external fixation 27760 Closed treatment, medial malleolus fracture; without manipulation 27762 Closed treatment, medial malleolus fracture; with manipulation, with/without skin/skeletal traction 27786 Closed treatment, distal fibular fracture (lateral malleolus); with/without manipulation 27788 Closed treatment, distal fibular fracture (lateral malleolus); with manipulation 27808 Closed treatment, bimalleolar ankle fracture, without manipulation 27810 Closed treatment, bimalleolar ankle fracture, with manipulation 27816 Closed treatment, trimalleolar ankle fracture, without manipulation 27818 Closed treatment, trimalleolar ankle fracture, with manipulation 27824 Closed treatment, fracture, weight bearing articular portion, distal tibia without manipulation 27825 Closed treatment, fracture, weight bearing articular portion, distal tibia with skeletal traction Proximal Humerus Fracture CPT Codes 23615 Open treatment, proximal humeral fracture, with/without internal/external fixation/tuberosity repair 23630 Open treatment, greater humeral tuberosity fracture with/without internal/external fixation 23670 Open treatment, shoulder dislocation w/fracture, greater tuberosity, with/without external rotation 23680 Open treatment, shoulder dislocation, w/surgical/anatomical neck fixator 23600 Closed treatment, proximal humeral fracture; without manipulation 23605 Closed treatment, proximal humeral fracture; with manipulation 23620 Closed treatment, greater humeral tuberosity fracture; without manipulation 23625 Closed treatment, greater humeral tuberosity fracture; with manipulation 23665 Closed treatment, shoulder dislocation with fracture, greater tuberosity, with manipulation 23675 Closed treatment, shoulder dislocation, with humoral neck fracture, with manipulation Wrist Fracture CPT Codes 25611 Percutaneous skeletal fixation, distal radial fracture/epiphyseal separation, with manipulation 25620 Open treatment, distal radial fracture/epiphyseal separation 25600 Closed treatment, distal radial fracture; without manipulation 25605 Closed treatment, distal radial fracture; with manipulation The Geographic Incidence and Treatment Variation of Common Fractures of Elderly Patients 248 Journal of the American Academy of Orthopaedic Surgeons surgical fixation. The percentage of ankle fractures treated surgically varied by a factor of nearly 4 from 20.8% (Altoona, PA) to 77.1% (Chi- co, CA). On average, patients in the Northwest were more likely than patients in the Southeast to receive surgical intervention. In 32 regions, at least 60% of ankle fractures were surgically treated, whereas in 50 re- gions, 30% of such fractures were surgically treated. 5 Most proximal humerus fractures represent low-energy injuries and can be treated nonsurgically with a sling and swath for immobilization. Surgical intervention was initiated, on average, 14.3% of the time in the United States. 5 Large variations in the percentages of surgical interven- tion were observed, from 6.4% (Takoma Park, MD) of all proximal humerus fractures to 60.0% (Taco- ma, WA) (Figure 7). In 8 regions, at least 40% of proximal humerus frac- tures were treated surgically; in 35 regions, less than 10% were treated surgically. 5 Figure 1 Fractures among Medicare enrollees during 1996. Fractures of the hip, wrist, ankle, and proximal humerus were the most common fractures observed. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 96.) Figure 2 The geographic variation of hip fracture rates within the United States in 1996 and 1997, adjusting for age, race, and sex. No- tice the increased prevalence of hip fractures throughout the southern states. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 101.) Scott M. Sporer, MD, MS, et al Volume 14, Number 4, April 2006 249 Figure 3 The geographic variation of wrist fracture rates within the United States in 1996 and 1997, adjusting for age, race, and sex. Notice the increased prevalence of fractures throughout the eastern states. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 123.) Figure 4 The geographic variation of ankle fracture rates within the United States between 1996 and 1997, adjusting for age, race and sex. Notice the increased prevalence of fractures throughout the eastern states. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 109.) The Geographic Incidence and Treatment Variation of Common Fractures of Elderly Patients 250 Journal of the American Academy of Orthopaedic Surgeons Geographic Variation of Fracture Rates Hip As the population ages, the prev- alence of fracture in the United States unquestionably will increase. The population aged ≥80 years is the fastest growing cohort in the United States. 13 More than 450,000 fractures of the hip, wrist, ankle, and proximal humerus were identified in the Medicare patient population aged >65 years in 1996 and 1997. 5 Numer- ous authors have demonstrated that increased age, female gender, smok- ing, and osteoporosis are risk factors for sustaining these injuries. 14 Addi- tionally, poor vision, decreased reac- tion times, nutritional status, and a smaller soft-tissue envelope pose other risks for fracture. 2-4,15-17 Although studies have shown dra- matic differences in regional variation of certain surgical procedures, such as radical prostatectomy and coronary artery bypass, little information was Figure 5 The proportion of fractures treated surgically among Medicare enrollees in 199 6 and 1997. Notice that nearly all hip fractures are treated surgically, whereas most proximal humerus fractures are treated nonsurgically. Other fracture patterns demonstrate marked regional variability in their preferred method of treatment. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 97.) Figure 6 The proportion of wrist fractures treated surgically in 1996 and 1997. Note that most fractures are treated nonsurgically. However, several areas treat >40% of wrist fractures with surgery. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000, p 125.) Scott M. Sporer, MD, MS, et al Volume 14, Number 4, April 2006 251 known about the regional variation of common fractures in the Medicare population. 12 Our results suggest that, despite adjustment for age, sex, and race, specific populations in various regions of the country are at increased risk of fracture. 5 Fractures of the prox- imal humerus, ankle, wrist, and hip showed a tenfold, sixfold, fourfold, and twofold variation in fracture rate, respectively. 5 The southeastern states, on aver- age, had greater risk of fracture, whereas northern states were rela- tively protected. The underlying eti- ology for these dramatic differences has yet to be determined. It is un- likely that the differences observed in these multiple studies 5 are a result of chance alone. Although there may be a systemic sampling bias secondary to variable physician coding, the ob- served differences are too large to be related to this variable alone. These trends also have been observed over very large geographic regions rather than specific locations. Additionally , other authors have reached similar conclusions with regard to the geo- graphic variation of hip fractures. 9 One potential hypothesis is that peo- ple living in the southern regions are exposed to environmental factors that place them at an increased risk of fracture. 11 Potential environmental risk fac- tors may be directly related to the re- gion, such as air quality, degree of sunlight, or the water quality. Alter- natively, environmental risk factors may be associated with specific ar- eas of the country related to diet, poverty, or medical practice pat- terns. Another hypothesis is that pa- tients with different genetic suscep- tibility to fractures live in the southern regions. 14,18 Geographic variation in the rate of hip fractures was initially described by Bacon et al 19 during a review of the 1979 to 1985 National Hospital Discharge Survey. 20 The rates of hos- pitalization for fracture of the hip were 45% higher in the Northeast compared with the South. This study was limited in its ability to provide only regional hospitalization rates because of the sampling from the National Hospital Discharge Sur- vey. Additionally, this survey exam- ined only procedures that were per- formed within a specific region, and it did not provide any patient demo- graphic information. The authors hy- pothesized that the increased risk of hip fracture within the northern re- gion may be secondary to a relatively high proportion of residents of Scan- dinavian ancestry in this area. Previ- ous studies that evaluated national hospital discharge data demonstrated higher fracture rates throughout northern European countries. 14,18 The increased risk for fracture is con- sidered to be secondary to a higher prevalence of osteoporosis through- out this region. Stroup et al 6 used data from the 1985 Medicare Provider Analysis and Review file (MEDPAR) to deter- mine the relative rate of hip fracture within the United States. The inci- dence of hip fractures was shown to follow a north-to-south gradient, Figure 7 The proportion of proximal humerus fractures treated surgically in 1996 and 1997. Note that most fractures are treated nonsurgically. However, several areas treat >40% of proximal humerus fractures with surgery. (Reproduced with permission from Weinstein JN, Birkmeyer JD [eds]: The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospit al Publishing, 2000, p 115.) The Geographic Incidence and Treatment Variation of Common Fractures of Elderly Patients 252 Journal of the American Academy of Orthopaedic Surgeons with higher fracture rates among the southern states. The geographic vari- ation in fracture rates was consistent among both men and women and among both “white” and “other than white” age groups. The authors did not think that the delivery of health care or practice patterns could account for these differences because the trend was consistent among both men and women. Several authors have demonstrat- ed an increased relative risk of frac- ture among patients living in the southeastern and Appalachian re- gions. Hinton et al 7 reviewed Medi- care date from 1984 to 1987 (687,850 hip fractures). They concluded that rates of hip fracture were greater for women than men and were higher in the southern region of the United States. There was an approximate twofold variation throughout the United States, from 0.88 (New Jer- sey) to 1.25 (Mississippi) per 100,000 Medicare enrollees. This study also examined the location of the hip fracture as either cervical, trochan- teric, or subtrochanteric. The rates of both cervical and trochanteric fractures remained higher in the southern regions for white women than in the northern regions. How- ever, this variation was dispropor- tionate because of the higher rates of cervical fractures. The significance of this finding is unclear, yet the au- thors concluded that the cervical re- gion of the hip may be more sensi- tive to the effects of nutritional, socioeconomic, or environmental factors. This study also demonstrat- ed that the risk of a hip fracture dou- bles each successive 5 years and that, among women, the relative risk of cervical to trochanteric frac- tures varies inversely by age. 21 Karagas et al 9 reviewed 39,599 Medicare hip fractures between 1986 and 1990 and found results similar to those of Hinton et al. 7 The overall rate of both femoral neck and tro- chanteric fractures was highest among white women, whereas the ratio of trochanteric to femoral neck fractures increased with advancing age. However, these trends were not observed among men or among black patients. The authors hypoth- esized that the observed hip fracture rates could be related to localized differences in bone density at specif- ic sites of the proximal femur. Mel- ton et al 21 had previously shown that once a patient’s bone mineral densi- ty drops below 0.60 g/cm 2 , that pa- tient had an increased incidence of trochanteric fracture. Jacobsen et al 22 reported similar north-to-south regional variation among hip fractures using data from the HCFA and the Department of Veterans Affairs. Additionally, data from the Bureau of Health Profes- sions Area Resource File was used to examine potential environmental risk factors. A regression analysis was performed that confirmed “a positive association between hip fracture incidence and the percent of the 65-year and older population be- low the poverty level and the per- cent of land in farms.” 22 The authors also found a weak association be- tween soft and fluoridated water and reduced sunlight exposure with an increased risk of hip fracture. The effect of fluoridated water also has been evaluated by Karagas et al. 8 Using a 5% sample of the Medi- care population, a correlation be- tween fluoridated drinking water and the risk of hip or ankle fracture was not observed. However, within the study population, a north-to-south geographic gradient for increased in- cidence of hip fracture persisted. The potential of reduced sunlight as a risk factor for hip fracture was in- directly examined by Jacobsen et al 10 when they reported on the seasonal variation in the incidence of hip frac- ture. Using HCFA data from 1984 to 1987, the seasonal variation among 621,387 hip fractures demonstrated a definite seasonal pattern among both male and female patients, with a peak in December and February and a nadir between July and August. This seasonal trend persisted among all five latitude groups. The authors hypothesized that, given that the for- mation of vitamin D is dependent on sunlight, and given that there are fewer hours of sunlight during the winter months, the degree of osteo- malacia may increase in elderly pa- tients during this time and can place them at a higher risk for fracture. Lauderdale et al 11 examined the impact of former residence on the rate of hip fracture in the Medicare patient population. The authors con- cluded that the risk for fracture was dependent on region of residence early in life rather than later in life. 11 These results suggest that strategies designed to improve peak bone mass during early childhood would be more effective than mini- mizing risk factors later in life. The etiology of the variable rates of hip fracture throughout the Unit- ed States is unknown. However, it is unlikely that these large variations are the result of chance alone. Rath- er, the fracture variability probably is multifactorial and includes exter- nal factors related to environmental exposure and internal factors related to genetic predisposition. Shoulder, Wrist, and Ankle Fractures Whereas the geographic variation among hip fractures has been gener- ally well-described, a paucity of in- formation is available about the re- gional variation among other common fractures (ie, proximal hu- merus, distal forearm, ankle) in the elderly population. Karagas et al 8 were the first to describe geographic trends among fractures of the proxi- mal humerus, distal forearm, and ankle. Using Medicare data, they showed that fractures of the proxi- mal humerus and distal forearm oc- cur in a geographic pattern that is distinct from that observed with hip fractures. The risk of fracture in- creased from west to east rather than from north to south (a trend also ob- served with hip fractures). The data presented from the DAMHC sup- Scott M. Sporer, MD, MS, et al Volume 14, Number 4, April 2006 253 port the finding of an increased risk of proximal humerus, distal forearm, and ankle fractures among eastern regions of the United States. 5 One hypothesis for these observations is that the risk factors for hip fractures are different from the risk factors for proximal humerus, distal forearm, and wrist and ankle fractures. The exact etiology of the geo- graphic variation among fracture rates within the United States remains un- clear. It is likely multifactorial, a re- sult of varying degrees of certain en- vironmental, occupational, genetic, and nutritional risk factors that exist between regions of the country. One of our hypotheses is that fractures of the hip are related to relatively sed- entary individuals, whereas fractures of the ankle, distal forearm, or prox- imal humerus are related to individ- uals actively participating in the workforce or in recreational activities. A second hypothesis of ours is that nutritional factors preferentially af- fect the bone metabolism of the fem- oral neck. We think that further ep- idemiologic research should be performed in this area to help reduce the cost, societal burden, and loss of independence related to these debil- itating fractures within the elderly pa- tient population. Geographic Variation in Treatment There also are dramatic differences in the treatment strategies used for the most common fractures in Medicare patients. Some fractures, such as those of the hip, have been shown to be treated best with surgical inter- vention; thus they show little geo- graphic variation in the proportion of fractures treated surgically. This was observed in the DAMHC data, with more than 98% of patients receiving surgical intervention for a hip frac- ture. 5 Conversely, certain fractures can be treated either surgically or nonsurgically. In general, nondis- placed fractures can be treated with cast immobilization, whereas dis- placed fractures require surgical re- duction and fixation. We observed large variations in the proportion of wrist, ankle, proximal humerus, and distal forearm fractures treated sur- gically throughout the United States. 5 In general, the northwestern regions were more likely to initiate surgical treatment despite their rel- ative lower incidence of fracture. There are several possible expla- nations for the observed differences in the proportion of patients receiv- ing surgical treatment. It is possible that the fractures encountered in the Northwest are more displaced or open or have associated injuries; the severity of injury is unable to be de- termined from the Medicare data. However, to our knowledge, no data support the concept that more se- vere injuries are more likely to occur in these regions. Diagnostic intensity also has been shown to influence the rates of surgical intervention. Regions with more aggressive diagnostic imaging tend to have higher surgical rates for specific conditions. 23 However, pa- tients who sustain fractures of the hip, wrist, ankle, and proximal hu- merus are in significant discomfor t, and plain radiographs are sufficient to make a diagnosis. Therefore, in- creased diagnostic testing is unlike- ly a plausible explanation. The varying incidence of surgical intervention also may be related to the population density of practicing orthopaedic surgeons in a particular area of the country. Keller at el 24 de- scribed regional variation in the pro- cedural rate among several major or- thopaedic conditions, a variation that may be partially attributed to the number of practicing ortho- paedic surgeons in an area. However, other authors have failed to show a similar relationship between the uti- lization of certain orthopaedic proce- dures and the population density of orthopaedic surgeons in an area. 25,26 It has been shown that surgeons possess varying thresholds to recom- mend and initiate surgical interven- tion, referred to as the local aggres- siveness phenomena. 12 As a result, regions throughout the country tend to have so-called surgical signa- tures, reflecting the practices regard- ing surgical treatment of the ortho- paedic surgeons in that area. A region’s surgical signature may be a result of variability in orthopaedic training in different parts of the country or a result of a paucity of data regarding the optimal treatment of a particular fracture. Weinstein 27 has shown that the rate of surgical intervention depends on the vari- ability in clinical decision making as well as patient-perceived risks and benefits. For low variability proce- dures, such as hip fractures, the sur- gical rate is relatively constant throughout the United States. In contrast, disk herniations have nu- merous treatment options, less sci- entific uniformity, and greater po- tential risks. Consequently, regional variability is far greater than that seen with hip fractures. Patient expectations also may contribute to the geographic vari- ability among surgical rates. It is possible that the perceived benefit of surgery is regionally dependent, and that patients in the northwestern portion of the United States think that the likelihood of returning to their preinjury status is greater with surgery. Consequently, patients and surgeons may be more likely to ini- tiate surgical intervention for a par- ticular fracture. Patients are the ones ultimately affected by the decision to proceed with surgical intervention. Thus, pa- tients need to be actively involved in the decision-making process in order to make an informed choice. 7 When patients are properly educated about surgical alternatives, they make choices that are most appropriate for their specific situations. Summary The United States’ population is ag- ing. As a result, orthopaedic sur- The Geographic Incidence and Treatment Variation of Common Fractures of Elderly Patients 254 Journal of the American Academy of Orthopaedic Surgeons geons are likely to experience a greater number of fractures in the elderly population. Throughout the country, the variability in the inci- dence of the most common fractures is marked. The underlying etiology is unclear but likely is multifactori- al, including social, environmental, nutritional, and genetic characteris- tics. There also is a wide range of hy- potheses; one possibility is that the risk factors for fractures in different locations vary, another that certain fractures are related to the activity level of the individual, and another that nutrient factors preferentially affects bone metabolism in different regions. There also is marked vari- ability in treatment among these common fractures. This surgical variability likely represents the sur- geon’s preference and his or her threshold for initiating surgical treatment. Hypothetically , this vari- ability also may exist because some fractures encountered in one region are different from those in another or it may be a result of the population density of orthopaedic surgeons practicing in a region. Additional studies are needed to elucidate un- derlying patient preferences and whether the decision to proceed with surgery is driven by the physi- cian, the patient, or both jointly. References Citation numbers printed in bold type indicate references published within the past 5 years. 1. Baron JA, Barrett JA, Karagas MR: The epidemiology of peripheral fractures. Bone 1996;18(suppl 3):209S-213S. 2. Felson DT, Anderson JJ, Hannan MT, Milton RC, Wilson PW, Kiel DP: Im- paired vision and hip fracture: The Framingham Study. J Am Geriatr Soc 1989;37:495-500. 3. Kelsey JL, Browner WS, Seeley DG, Nevitt MC, Cummings SR: Risk fac- tors for fractures of the distal forearm and proximal humerus: The Study of Osteoporotic Fractures Research Group. Am J Epidemiol 1992;135: 477-489. 4. Kreiger N, Kelsey JL, Holford TR, O’Connor T: An epidemiologic study of hip fracture in postmenopausal women. Am J Epidemiol 1982;116: 141-148. 5. Weinstein JN, BirkmeyerJD(eds): The Dartmouth Atlas of Musculoskeletal Health Care. Chicago, IL: American Hospital Publishing, 2000. 6. Stroup NE, Freni-Titulaer LW, Schwartz JJ: Unexpected geographic variation in rates of hospitalization for patients who have fracture of the hip: Medicare enrollees in the United States. J Bone Joint Surg Am 1990;72: 1294-1298. 7. Hinton RY, Lennox DW, Ebert FR, Ja- cobsen SJ, Smith GS: Relative rates of fracture of the hip in the United States: Geographic, sex, and age vari- ations. J Bone Joint Surg Am 1995;77: 695-702. 8. Karagas MR, Baron JA, Barrett JA, Ja- cobsen SJ: Patterns of fracture among the United States elderly: Geographic and fluoride effects. Ann Epidemiol 1996;6:209-216. 9. Karagas MR, Lu-Yao GL, Barrett JA, Beach ML, Baron JA: Heterogeneity of hip fracture: Age, race, sex, and geo- graphic patterns of femoral neck and trochanteric fractures among the US elderly. Am J Epidemiol 1996;143: 677-682. 10. Jacobsen SJ, Goldberg J, Miles TP, Bro- dy JA, Stiers W, Rimm AA: Seasonal variation in the incidence of hip frac- ture among white persons aged 65 years and older in the United States, 1984-1987. Am J Epidemiol 1991; 133:996-1004. 11. Lauderdale DS, Thisted RA, Goldberg J: Is geographic variation in hip frac- ture rates related to current or former region of residence? Epidemiology 1998;9:574-577. 12. Wennberg JC, Cooper MM: The Dar t- mouth Atlas of Health Care. Chicago, IL: American Hospital Publishing, 1997. 13. Trends in aging United States and worldwide. MMWR Morb Mortal Wkly Rep 2003;52:101-106. 14. Lips P: Epidemiology and predictorsof fractures associated with osteoporo- sis. Am J Med 1997;103:3S-8S. 15. Cummings SR, Kelsey JL, Nevitt MC, O’Dowd KJ: Epidemiology of os- teoporosis and osteoporotic fractures. Epidemiol Rev 1985;7:178-208. 16. Lewinnek GE, Kelsey J, White AA III, Kreiger NJ: The significance and a comparative analysis of the epidemi- ology of hip fractures. Clin Orthop Relat Res 1980;152:35-43. 17. Felson DT, Kiel DP, Anderson JJ, Kan- nel WB: Alcohol consumption and hip fractures: The Framingham Study. Am J Epidemiol 1988;128:1102-1110. 18. Bacon WE, Maggi S,Looker A, etal: In- ternational comparison of hip frac- ture rates in 1988-89. Osteoporos Int 1996;6:69-75. 19. Bacon WE, Smith GS, Baker SP: Geo- graphic variation in the occurrence of hip fractures among the elderly white US population. Am J Public Health 1989;79:1556-1558. 20. US Department of Health and Human Services: Utilization of short-stay hospitals: Annual summary, in Statis- tics NCFH. US Department of Health and Human Services Publication (PHS)87-1752, series 2, no 39. Wash- ington, DC: Government Printing Of- fice, 1987. 21. Melton LJ III, Wahner HW, Richelson LS, O’Fallon WM, Riggs BL: Os- teoporosis and the risk of hip fracture. Am J Epidemiol 1986;124:254-261. 22. Jacobsen SJ, Goldberg J, Miles TP, Bro- dy JA, Stiers W, Rimm AA: Regional variation in the incidence of hip frac- ture: US white women aged 65 years and older. JAMA 1990;264:500-502. 23. Birkmeyer JD, Sharp SM, Finlayson SR, Fisher ES, Wennberg JE: Variation profiles of common surgical proce- dures. Surgery 1998;124:917-923. 24. Keller RB, Soule DN, Wennberg JE, Hanley DF: Dealing with geographic variations in the use of hospitals: The experience of the Maine Medical As- sessment Foundation Orthopaedic Study Group. J Bone Joint Surg Am 1990;72:1286-1293. 25. Vitale MG, Krant JJ, Gelijns AC, et al: Geographic variations in the rates of operative procedures involving the shoulder, including total shoulder re- placement, humeral head replace- ment, and rotator cuff repair. J Bone Joint Surg Am 1999;81:763-772. 26. Peterson MG, Hollenberg JP, Sza- trowski TP, Johanson NA, Mancuso CA, Charlson ME: Geographic varia- tions in the rates of elective total hip and knee arthroplasties among Medi- care beneficiaries in the United States. J Bone Joint Surg Am 1992;74: 1530-1539. 27. Weinstein JN: The missing piece: Em- bracing shared decision making to re- form health care. Spine 2000;25:1-4. Scott M. Sporer, MD, MS, et al Volume 14, Number 4, April 2006 255