Bisphosphonate Use and Atypical Fractures of the Femoral Shaft

T h e n e w e ngl a nd j o u r na l o f m e dic i n e original article Bisphosphonate Use and Atypical Fractures of the Femoral Shaft Jörg Schilcher, M.D., Karl Michaëlsson, M.D., Ph.D., and Per Aspenberg, M.D., Ph.D. A bs tr ac t From the Department of Experimental and Clinical Medicine, Faculty of Health Science, Linköping University, Linköping (J.S., P.A.); and the Department of Surgical Sciences, Section of Orthopedics, and Uppsala Clinical Research Center, Uppsala University, Uppsala (K.M.) both in Sweden. Address reprint requests to Dr. Aspenberg at the Department of Experimental and Clinical Medicine, Faculty of Health Sciences, Linköping University, SE-581 85 Linköping, Sweden, or at per.aspenberg@liu.se. Drs. Schilcher and Michaëlsson contributed equally to this article. This article (10.1056/NEJMoa1010650) was last updated on August 9, 2012, at NEJM.org. N Engl J Med 2011;364:1728-37. Copyright 2011 Massachusetts Medical Society. Background Studies show conflicting results regarding the possible excess risk of atypical fractures of the femoral shaft associated with bisphosphonate use. Methods In Sweden, 12,777 women 55 years of age or older sustained a fracture of the femur in 2008. We reviewed radiographs of 1234 of the 1271 women who had a subtrochanteric or shaft fracture and identified 59 patients with atypical fractures. Data on medications and coexisting conditions were obtained from national registries. The relative and absolute risk of atypical fractures associated with bisphosphonate use was estimated by means of a nationwide cohort analysis. The 59 case patients were also compared with 263 control patients who had ordinary subtrochanteric or shaft fractures. Results The age-adjusted relative risk of atypical fracture was 47.3 (95% confidence interval [CI], 25.6 to 87.3) in the cohort analysis. The increase in absolute risk was 5 cases per 10,000 patient-years (95% CI, 4 to 7). A total of 78% of the case patients and 10% of the controls had received bisphosphonates, corresponding to a multivariableadjusted odds ratio of 33.3 (95% CI, 14.3 to 77.8). The risk was independent of coexisting conditions and of concurrent use of other drugs with known effects on bone. The duration of use influenced the risk (odds ratio per 100 daily doses, 1.3; 95% CI, 1.1 to 1.6). After drug withdrawal, the risk diminished by 70% per year since the last use (odds ratio, 0.28; 95% CI, 0.21 to 0.38). Conclusions These population-based nationwide analyses may be reassuring for patients who receive bisphosphonates. Although there was a high prevalence of current bisphosphonate use among patients with atypical fractures, the absolute risk was small. (Funded by the Swedish Research Council.) 1728 n engl j med 364;18 nejm.org may 5, 2011

Bisphosphonate Use and Atypical Fractures Bisphosphonates reduce the overall risk of fracture among patients with osteoporosis, with a long-lasting beneficial effect. 1 However, since bisphosphonates reduce bone remodeling, they might freeze the skeleton, allowing accumulation of microcracks over time, leading to fatigue fractures (also called stress fractures). 2 Fatigue fractures are well known in mechanical engineering, and they occur with age and overload in many materials. They result from the slow propagation of cracks, resulting in a peculiar appearance. In bone, their appearance is characterized by a straight fracture line running perpendicularly to tractional forces. Such fractures are not uncommon in athletes, and if they are not displaced, they heal slowly with minor external bone formation, often appearing as cortical thickening. Stress fractures in compact bone occur at sites with high tensional stress, such as the lateral cortex of the proximal femoral shaft; this site corresponds to the locations of reported bisphosphonate-associated atypical fractures. 2-11 In published case reports that classified femoral fractures as stress fractures or nonstress fractures according to their radiographic appearance, among patients admitted to tertiary centers, such stress fractures (commonly called atypical femoral fractures, the term used here) were more common in patients who received bisphosphonates than in those who did not. 3,7 A large proportion of patients who were receiving bisphosphonate treatment and who had atypical fractures did, however, also receive other medications, especially systemic glucocorticoids and proton-pump inhibitors. This finding has prompted the suggestion that these two types of drugs, in addition to being markers of coexisting conditions, are important contributors to the risk of the development of atypical femoral fractures. 9,10 In contrast, an association between atypical fractures and bisphosphonate use was not shown in nationwide cohort analyses based on registry identification of atypical femoral fractures or in randomized trials in which the classification of fractures was based on radiologic reports. 12-14 It might be that either the case series were biased or the registry studies and the randomized studies did not identify a sufficient number of atypical fractures accurately. We reviewed radiographs of all femoral subtrochanteric and shaft fractures that occurred in 2008 in the entire population of Sweden. Individual linkage of these fracture cases to national registries allowed both a nationwide cohort analysis and a population-based case-control study to elucidate the relation between atypical femoral fractures and bisphosphonate use. Me thods Study Population In 2008, according to the National Swedish Patient Register, 12,777 women 55 years of age or older sustained a fracture of the femur (Fig. 1). Of these women, 1271 had a femoral subtrochanteric or shaft fracture (International Classification of Diseases, 10th Revision [ICD-10] diagnosis code S722 or S723 with external-cause code W [i.e., excluding any type of transportation accident]). Digitized radiographs from 1234 patients were obtained from the involved hospitals. The remaining 37 case patients were excluded because of an insufficient quality of the radiographs (30 patients) or because the radiographs were not available (7 patients). The study was approved by the regional ethics committee, and patients were enrolled without consent being obtained. The 1234 radiographs were reviewed and classified according to the pattern of fractures, as described below. After the fractures were classified, data on drug use and inpatient and outpatient care were obtained from registries of the Swedish National Board of Health and Welfare. The Swedish Prescribed Drug Register contains data on all prescriptions dispensed to the entire Swedish population (approximately 9 million inhabitants) from July 2005 onward. The Swedish National Patient Register contains discharge diagnoses for hospitalizations in Sweden, with complete coverage since 1987. 15 This registry also contains data on outpatient visits since 2001. Complete linkage between the registries is possible through the use of the personal identification number provided to all Swedish citizens. Classification of Fractures Patients who were identified in the National Swedish Patient Register as having subtrochanteric or shaft fractures were categorized into four major groups. The first group included 47 patients with typical stress fractures that were transverse on the lateral side without intermediate fragments, together with thickening of the lateral cortex at n engl j med 364;18 nejm.org may 5, 2011 1729

T h e n e w e ngl a nd j o u r na l o f m e dic i n e 1,521,131 Women 55 yr or older in 2008 were eligible for the study, bisphosphonate use, 5% 12,777 Had fractures of the femur, bisphosphonate use, 11% 6254 Had femoralneck fractures 4664 Had trochanteric fractures 1351 Had subtrochanteric or shaft fractures from any cause, bisphosphonate use, 17% 435 Had distal fractures 73 Had other type of fractures 1271 Had fractures because of falls and were included in blinded radiographic classification 37 Were excluded because radiographs were not available or were of bad quality 912 Had fractures that were not relevant for comparison with stress fractures 625 Had trochanteric or condylar fractures 238 Had fractures associated with implants 16 Had pathologic fractures 24 Had femoral-neck fractures 9 Did not have femoral fractures 59 Had atypical fractures, bisphosphonate use, 78% 263 Had control fractures, bisphosphonate use, 10% Figure 1. Identification of Stress (Atypical) Fractures in the Study Population. the fracture site. The fracture did not involve the trochanteric or the condylar area. The femur did not have implants, and the fracture was not pathologic. The second group included 12 patients with suspected stress fractures. These fractures were the same as those described in the first group, but without clear thickening of the lateral cortex or with a separate intermediate fracture fragment. The third group included 263 control patients with fractures that did not appear to be stress fractures but were relevant for comparison with stress fractures. None of the shaft fractures were transverse on the lateral side, and there was no trochanteric or condylar engagement, except for minor trochanteric fracture extensions. There were no implants or pathologic fractures. The fourth group included 912 patients with fractures that were not relevant for comparison with stress fractures. Of these fractures, 625 were associated with major involvement of a trochanter or distal condyle, 238 were associated with implants, and 16 were pathologic. This last group also included 24 femoral-neck fractures and 9 fractures that were not femoral. During the revision of this article, a consensus document was published in which stress fractures without cortical thickening were defined as atypical. 16 We therefore analyzed the first two groups together (Fig. 1). After completion of our analysis, we reviewed all the routine radiologic reports. These reports mentioned a transverse fracture in 23 of the 59 case patients and a fatigue (also called stress or insufficiency) fracture or an atypical fracture pattern in only 1 case patient. The remainder mentioned only the location of fractures. 1730 n engl j med 364;18 nejm.org may 5, 2011

Bisphosphonate Use and Atypical Fractures Before the retrieval of registry information, 72 of the 1234 fractures (including 22 atypical fractures) were randomly selected for reclassification by one of the authors who was not informed about the previous categorization or the number of atypical fractures in the sample. There was complete agreement between the two classifications. Statistical Analysis Cohort Study All 1.5 million women who were 55 years of age or older and who were residents of Sweden in 2008 according to data from Statistics Sweden were included as a reference population in a cohort analysis. Of these women, 83,311 received bisphosphonates and 59 had atypical fractures. Age was divided into 5-year strata, and the age of patients who were older than 90 years of age was an additional category. Age-stratified incidence proportions of atypical fractures were calculated for women who received bisphosphonates and for women with no reported use of bisphosphonates; these calculations were used to estimate both the age-adjusted relative risk and the absolute risk of atypical fracture (with 95% confidence intervals 17 ) in the total population. The duration of bisphosphonate use and the time since the last use were considered in three categories (<1.0 year, 1.0 to 1.9 years, and 2.0 years). In addition, similar nationwide cohort analyses were performed, albeit with only registry identification of categories of femoral fractures according to ICD-10 codes, as described by Abrahamsen et al. 12 Case Control Study The case patients with atypical fractures were compared with the 263 control patients who had fractures in a similar location. To reduce the risk of selection bias, these control patients were selected because they were representative of the population of patients with underlying vulnerability to fractures. All drug exposures were analyzed in relation to the date of fracture. Bisphosphonate use before the fracture was categorized according to the type of bisphosphonate. The total duration of bisphosphonate use was estimated according to defined daily doses from July 2005 to the date of the fracture, excluding periods of nonuse. The duration of use was analyzed both as a continuous variable and in categories (<1.0 year, 1 to 1.9 years, and 2.0 years). The time since the last use was treated as a continuous variable and in categories (<1.0 year, 1.0 to 1.9 years, or 2.0 years since the end of treatment). The use or nonuse of systemic glucocorticoids, estrogen-replacement ther apy, selective estrogen-receptor modulators, antiepileptic drugs, antidepressants, or protonpump inhibitors was categorized into separate marker variables. Diseases that were diagnosed before the fractures occurred were identified from the Swedish National Patient Register (as ICD-10 codes) and defined as present or not present. From these registry data, the Charlson comorbidity index score was calculated. 18 Odds ratios with 95% confidence intervals were calculated by means of unconditional logistic regression and used as measures of association for the relative risk of atypical fracture with bisphosphonate use. Ageadjusted odds ratios (with age as a continuous variable) and odds ratios additionally adjusted according to use or nonuse of glucocorticoids and the Charlson comorbidity index score (a continuous variable) were determined. Odds ratios changed little after adjustment for the use of estrogen therapy, antiepileptic agents, antidepressants, or proton-pump inhibitors. We also performed stratified analyses according to the use or nonuse of glucocorticoids, the use or nonuse of proton-pump inhibitors, or age, with a test for homogeneity between strata. 17 R esult s Cohort Study Of all 12,777 women with a fracture of the femur, we identified 59 with atypical fractures (Fig. 1). The incidence proportion of atypical fracture among users and nonusers of bisphosphonates in the entire Swedish population is shown in Table 1. The age-adjusted relative risk of atypical fracture with any use of bisphosphonates was 47.3 (95% confidence interval [CI], 25.6 to 87.3). The difference in the risk of atypical fracture between users and nonusers of bisphosphonates was 5 cases (95% CI, 4 to 7) per 10,000 patient-years, corresponding to an average number needed to harm (i.e., the number of bisphosphonate users needed for one case of fracture to occur) of 2000 per year of use. The risk was higher with a longer duration of use. Most bisphosphonate users with atypical fractures had a history of recent use. Bisphosphonate users also had a higher relative risk of registry-identified fractures of the n engl j med 364;18 nejm.org may 5, 2011 1731

T h e n e w e ngl a nd j o u r na l o f m e dic i n e Table 1. Risk of Atypical Femoral Fracture Associated with Bisphosphonate Use during the 3 Years (2005 2008) Preceding the Fracture.* Variable Bisphosphonate use No. of Women Cases of Atypical Fracture No. of Atypical Fracture Cases Crude Incidence no./10,000 patient-yr Age-Adjusted Relative Risk (95% CI) Never 1, 437,820 13 0.09 1.0 (reference) Age-Adjusted Absolute Risk (95% CI) Ever 83,311 46 5.5 47.3 (25.6 87.3) 0.0005 (0.0004 0.0007) Duration of use <1.0 yr 15,672 3 1.9 18.4 (5.3 64.3) 0.0002 (0.0000 0.0004) 1.0 1.9 yr 21,406 4 1.9 17.0 (5.7 50.7) 0.0002 (0.0000 0.0004) 2.0 yr 46,233 39 8.4 67.0 (35.8 125.8) 0.0008 (0.0006 0.0011) Time since last use <1.0 yr 83,311 42 5.0 42.9 (22.9 80.4) 0.0005 (0.0004 0.0007) 1.0 1.9 yr 70,036 1 0.1 3.5 (1.0 11.9) <0.0001 (0.0000 0.0000) 2.0 yr 75,583 3 0.4 3.2 (1.0 10.1) <0.0001 (0.0000 0.0001) * CI denotes confidence interval. hip (relative risk, 1.32; 95% CI, 1.25 to 1.40), the subtrochanteric region alone (relative risk, 1.80; 95% CI, 1.50 to 2.17), and the femoral shaft (relative risk, 3.83; 95% CI, 3.08 to 4.78) as compared with persons who did not receive bisphosphonates (Table 2). After exclusion of irrelevant fractures and atypical fractures in the subtrochanteric and shaft regions by our review of radiographs, the relative risk of other types of fractures (i.e., in the controls in our case control study) was 1.27 (95% CI, 0.85 to 1.90) an estimate similar to that for hip fractures (Table 2). Case Control Study The characteristics of the study participants are listed in Table 3. Case patients with atypical fractures were younger than control patients. Case patients were also less frequent users of antidepressants and were less likely to have sustained a previous osteoporotic or hip fracture, even after adjustment for the age difference. A greater proportion of the bisphosphonate users (both cases and controls) had received systemic glucocorticoids and had a diagnosis of osteoporosis or inflammatory joint disease. Of the case patients, 78% had received bisphosphonates (81% of the women with stress fractures and 67% of the women with suspected stress fractures), as compared with 10% of the controls (Table 4). This finding corresponds to a multivariable-adjusted odds ratio of 33.3 (95% CI, 14.3 to 77.8), with a similar estimate for women who had received alendronate and risedronate. No atypical fractures were associated with etidronate, but few women received this bisphosphonate. At the time of the latest prescription, bisphosphonates were predominantly taken once a week (in 83% of the women) or daily. None of the patients received injections. The risk of an atypical fracture was higher with an increasing duration of bisphosphonate use, with an odds ratio of 1.3 (95% CI, 1.1 to 1.6) per 100 prescribed daily doses. This risk was approximately 10 times as high as a normal level of risk within the first 2 years of use and 50 times as high thereafter (Table 4). The data did not suggest that the risk continued to increase with a longer duration of treatment. In Sweden, patients renew their prescriptions every 3 months. Therefore, in an additional analysis, we excluded women who received prescriptions for bisphosphonates during the first 3 months after the initiation of the Swedish Prescribed Drug Register in order to specifically evaluate the association in women who had drug exposure for 2.0 to 2.9 years. Of these women, 14 of 17 were case patients with atypical fractures, corresponding to a multivariableadjusted odds ratio of 61.7 (95% CI, 14.0 to 272.6). 1732 n engl j med 364;18 nejm.org may 5, 2011

Bisphosphonate Use and Atypical Fractures Table 2. Risk of Fracture Associated with the Use of Bisphosphonates.* Variable Bisphosphonate Users (N = 83,311) No. of Fracture Cases Crude Incidence Nonusers of Bisphosphonates (N = 1,437,820) No. of Fracture Cases Crude Incidence Age-Adjusted Relative Risk (95% CI) Age-Adjusted Absolute Risk (95% CI) no./10,000 patient-yr no./10,000 patient-yr Diagnosis from registry data Any hip fracture, ICD-10 code S720, S721, or S722 1255 151 10,585 74 1.32 (1.25 1.40) 0.0036 (0.0028 0.0045) Femoral-neck fracture, ICD-10 code S720 599 72 5,655 39 1.19 (1.09 1.29) 0.0011 (0.0005 0.0017) Trochanteric fracture, ICD-10 code S721 528 63 4,136 29 1.41 (1.29 1.54) 0.0018 (0.0013 0.0024) Subtrochanteric fracture, ICD-10 code S722 128 15 794 5.5 1.80 (1.50 2.17) 0.0007 (0.0004 0.0009) Femoral-shaft fracture, ICD-10 code S723 105 13 324 2.3 3.83 (3.08 4.78) 0.0009 (0.0009 0.0012) Subtrochanteric or femoral-shaft fracture 233 28 1,118 7.8 2.34 (2.03 2.70) 0.0016 (0.0012 0.0019) After radiographic classification Subtrochanteric or femoral-shaft fracture, with atypical 26 3 237 1.6 1.27 (0.85 1.90) 0.0001 (0.0000 0.0002) fractures excluded Subtrochanteric or femoral-shaft fracture, with atypical 72 8.6 250 1.7 3.39 (2.61 4.39) 0.0006 (0.0004 0.0008) fractures included Atypical fracture 46 5.5 13 0.09 47.3 (25.6 87.3) 0.0005 (0.0004 0.0007) * ICD-10 denotes International Classification of Diseases, 10th Revision. n engl j med 364;18 nejm.org may 5, 2011 1733

T h e n e w e ngl a nd j o u r na l o f m e dic i n e Table 3. Characteristics of the Case Patients and Controls.* Variable Case Patients (N = 59) Controls (N = 263) Bisphosphonate Users (N = 46) Nonusers of Bisphosphonates (N = 13) Bisphosphonate Users (N = 26) Nonusers of Bisphosphonates (N = 237) Age yr 75.1± 9.5 76.2±11.9 80.0±8.0 83.7± 9.4 Drug use no. (%) Glucocorticoid 17 (37) 2 (15) 11 (42) 26 (11) Estrogen 12 (26) 3 (23) 5 (19) 42 (18) Selective estrogen-receptor modulator 0 0 0 0 Antidepressants 6 (13) 3 (23) 11 (42) 79 (33) Antiepileptic agents 4 (9) 1 (8) 2 (8) 19 (8) Proton-pump inhibitors 18 (39) 5 (38) 11 (42) 61 (26) Previous diagnoses no. (%) Any fracture 18 (39) 4 (31) 13 (50) 104 (44) Osteoporotic fracture 7 (15) 3 (23) 11 (42) 85 (36) Hip fracture 3 (7) 2 (15) 5 (19) 57 (24) Musculoskeletal disease 22 (48) 4 (31) 13 (50) 52 (22) Inflammatory joint disease 3 (7) 0 3 (12) 1 (<1) Osteoporosis 3 (7) 0 3 (12) 2 (1) Cardiovascular disease 15 (33) 3 (23) 15 (58) 116 (49) Ischemic heart disease 8 (17) 1 (8) 8 (31) 33 (14) Stroke 2 (4) 0 0 24 (10) Endocrine disorder 2 (4) 1 (8) 5 (19) 56 (24) Diabetes mellitus 1 (2) 0 4 (15) 25 (11) Cancer 7 (15) 1 (8) 3 (12) 28 (12) Neurologic disorder 3 (7) 0 3 (12) 40 (17) Psychiatric disorder 2 (4) 0 4 (15) 38 (16) Kidney or urinary disease 8 (17) 2 (15) 5 (19) 72 (30) Gastrointestinal disease 9 (20) 4 (31) 8 (31) 61 (26) Respiratory disease 4 (9) 1 (8) 6 (23) 37 (16) * Plus minus values are means ±SD. Most atypical fractures associated with bisphosphonate use occurred within 1 year after the last prescription. There was a 70% reduction in risk for every year since the last use (multivariableadjusted odds ratio, 0.28; 95% CI, 0.21 to 0.38). The risk reduction was similar if the bisphosphonates had been prescribed for less than 2 years (odds ratio, 0.45; 95% CI, 0.26 to 0.76) or for 2 years or more (odds ratio, 0.26; 95% CI, 0.19 to 0.35). The association between the risk of atypical fracture and bisphosphonate use was not increased among users of glucocorticoids or proton-pump inhibitors as compared with nonusers (see the Table in the Supplementary Appendix, available with the full text of this article at NEJM.org). The increase in risk associated with bisphosphonate use was similar for women younger than 85 years and those 85 years of age or older. Discussion These population-based nationwide analyses should be reassuring for bisphosphonate users. Although there was a high prevalence of current bisphosphonate use among patients with atypi- 1734 n engl j med 364;18 nejm.org may 5, 2011

Bisphosphonate Use and Atypical Fractures Table 4. Odds Ratios for Atypical Femoral Fractures Associated with Bisphosphonate Use.* Variable Case Patients (N = 59) Controls (N = 218) Odds Ratio (95% CI) Age-Adjusted Multivariable-Adjusted Bisphosphonate use Never 13 195 1.0 (reference) 1.0 (reference) Ever 46 23 24.6 (11.3 55.5) 33.2 (13.6 80.9) Type of bisphosphonate Alendronate 38 16 31.2 (13.5 72.0) 36.8 (14.6 92.7) Risedronate 6 4 15.5 (3.6 65.9) 32.4 (5.5 192.0) Etidronate 0 4 NA NA Ibandronate 2 0 NA NA Risk of fracture per 100 defined daily doses NA NA 1.4 (1.2 1.6) 1.3 (1.1 1.6) Duration of use <1.0 yr 3 6 6.0 (1.2 29.4) 9.9 (1.7 55.8) 1.0 1.9 yr 4 6 5.3 (1.2 23.6) 7.7 (1.6 36.2) 2.0 yr 39 11 47.7 (19.1 119.1) 52.0 (19.7 137.4) Time since last use <1.0 yr 42 15 37.0 (15.8 86.9) 42.8 (16.8 108.9) 1.0 1.9 yr 1 3 2.8 (0.2 31.4) 5.7 (0.5 69.3) 2.0 yr 3 5 5.7 (1.1 30.0) 9.4 (1.7 52.6) Risk of fracture per yr since last use NA NA 0.31 (0.23 0.41) 0.29 (0.22 0.40) * NA not applicable. Variables were adjusted for age (as a continuous variable), use or nonuse of a glucocorticoid, and Charlson index score (as a continuous variable). cal fractures, the magnitude of the absolute risk was small. With a correct indication, the benefits of fracture prevention with bisphosphonate use will greatly outweigh the risk of atypical femoral fracture (i.e., the numbers needed to treat 14 will be lower than the numbers needed to harm). The risk appeared to be unrelated to the use of systemic glucocorticoids and other drugs with effects on bone and was independent of coexisting conditions and age. Previous nationwide population-based studies and randomized trials have shown only modest site-specific increases in the risk of femoral fractures with bisphosphonate use. 12-14 The use of registry classification alone enabled us to confirm these results. 12,13 The increased risks of femur fracture, irrespective of fracture location, that were associated with bisphosphonate use are probably the consequence of osteoporosis. 12 The relative risk with bisphosphonate use among our controls who had radiologically defined fractures of the femoral shaft was similar to the risk of hip fractures, indicating a similar and somewhat increased base risk of femur fracture in this fracture-prone subpopulation. Earlier studies relied on registry data or hospital records, whereas the present investigation used reclassification of fractures according to radiographs. The specific radiographic classification is important, since our analysis shows that the rare atypical femoral fracture will be overshadowed by other types of fractures in registry studies, impeding the detection of their association with bisphosphonates. Our results appear to contrast with those of a recent reanalysis of three randomized trials 14 in which 10 patients with fractures that were located at the femoral shaft or subtrochanteric areas were identified from radiologists reports. No relation to bisphosphonates was shown. 14 The specific type of fracture was not reported, and our data suggest that in most cases, routine radiologic assessment does not detect the atypical fracture n engl j med 364;18 nejm.org may 5, 2011 1735

T h e n e w e ngl a nd j o u r na l o f m e dic i n e pattern. If identified, the atypical fractures in these 10 women would probably have been few, and the results are therefore not incongruent with our data. In our study, the women who sustained atypical fractures did not appear to be especially frail. They had a lower frequency of previous osteoporotic fractures than the controls. It is possible that for the women with bisphosphonate-associated atypical fractures, the drug had not been prescribed for a proper indication. However, more than one third of all patients receiving bisphosphonates in this study were receiving systemic glucocorticoids, which is a correct indication for prophylactic use of bisphosphonates. We 19 and other investigators 3 have previously presented data similar to the current results. The present study adds to the previous literature with a larger number of cases, estimates of relative and absolute risk, and strictly population-based comparisons both with patients with other types of fractures and with the general population. We also conducted analyses of the type of bisphosphonate, the duration of treatment, and how recently the drug was used, and in the analyses we considered the concomitant use of other drugs and coexisting conditions. It has been proposed that glucocorticoids and proton-pump inhibitors are likely to contribute to the risk of atypical fractures, 3,9 but our data suggest that this is not the case. The risk of atypical fractures decreased more rapidly after drug withdrawal than would be expected, given the prolonged presence of the drug in the bone. This observation and the increased risk during the first year of treatment are difficult to reconcile with the etiologic hypothesis that these fractures are a consequence of increased age of the cortical bone because of reduced remodeling. Thus, the pathogenic role of bisphosphonates probably involves faster processes. Bisphosphonates accumulate on fracture surfaces immediately after dosing. Ongoing treatment thus makes these surfaces highly resistant to resorption, whereas previous treatment does not. It is therefore conceivable that ongoing bisphosphonate treatment blocks targeted remodeling of cracks, which can grow and cause a stress fracture. This explanation, which implies that the long-lasting protective effect after bisphosphonate withdrawal 1 is not associated with a similarly lasting risk of atypical fracture, argues in favor of intermittent use. 16 Our study has several limitations. First, the observational design precludes definite causal interpretation of the results. Nonetheless, our study design precluded any bias associated with differential recall of bisphosphonate use and covariate information. Second, drug use before 2005 was not measured. We were therefore unable to estimate the risk pattern with a longer duration of use. Third, results adjusted for coexisting conditions and the use of other drugs were based on the case-control analysis only. Fourth, the study was performed in women and in a Northern European country, limiting the generalizability to men and other ethnic groups. Finally, no data were available to evaluate whether the risk of atypical fracture was dependent on bone density. We conclude that the absolute risk of atypical fracture associated with bisphosphonates for the individual patient with a high risk of osteoporotic fractures is small as compared with the beneficial effects of the drug. Dr. Aspenberg reports receiving consulting fees from Eli Lilly and Amgen and grant support to his institution, Linköping University, from Eli Lilly and Amgen, as well as holding stock in AddBIO, a company trying to commercialize a method for bisphosphonate coating of implants to be inserted in bone, and holding a patent for this method. No other potential conflict of interest relevant to this article was reported. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. References 1. Black DM, Schwartz AV, Ensrud KE, et al. Effects of continuing or stopping alendronate after 5 years of treatment: the Fracture Intervention Trial Long-term Extension (FLEX): a randomized trial. JAMA 2006;296:2927-38. 2. Neer RM. Skeletal safety of tiludronate. Bone 1995;17:Suppl:501S-503S. 3. Girgis CM, Sher D, Seibel MJ. Atypical femoral fractures and bisphosphonate use. N Engl J Med 2010;362:1848-9. 4. Ing-Lorenzini K, Desmeules J, Plachta O, Suva D, Dayer P, Peter R. Low-energy femoral fractures associated with the long-term use of bisphosphonates: a case series from a Swiss university hospital. Drug Saf 2009;32:775-85. 5. Kwek EB, Goh SK, Koh JS, Png MA, Howe TS. An emerging pattern of subtrochanteric stress fractures: a long-term complication of alendronate therapy? Injury 2008;39:224-31. 6. Lenart BA, Lorich DG, Lane JM. Atypical fractures of the femoral diaphysis in postmenopausal women taking alendronate. N Engl J Med 2008;358:1304-6. 7. Lenart BA, Neviaser AS, Lyman S, et al. Association of low-energy femoral fractures with prolonged bisphosphonate use: a case control study. Osteoporos Int 2009;20:1353-62. 8. Neviaser AS, Lane JM, Lenart BA, Edobor-Osula F, Lorich DG. Low-energy femoral shaft fractures associated with alendronate use. J Orthop Trauma 2008; 22:346-50. 9. Odvina CV, Levy S, Rao S, Zerwekh JE, Sudhaker Rao D. Unusual mid-shaft fractures during long term bisphosphonate 1736 n engl j med 364;18 nejm.org may 5, 2011

Bisphosphonate Use and Atypical Fractures therapy. Clin Endocrinol (Oxf) 2010;72: 161-8. 10. Odvina CV, Zerwekh JE, Rao DS, Maalouf N, Gottschalk FA, Pak CY. Severely suppressed bone turnover: a potential complication of alendronate therapy. J Clin Endocrinol Metab 2005;90:1294-301. 11. Visekruna M, Wilson D, McKiernan FE. Severely suppressed bone turnover and atypical skeletal fragility. J Clin Endocrinol Metab 2008;93:2948-52. 12. Abrahamsen B, Eiken P, Eastell R. Subtrochanteric and diaphyseal femur fractures in patients treated with alendronate: a register-based national cohort study. J Bone Miner Res 2009;24:1095-102. 13. Idem. Cumulative alendronate dose and the long-term absolute risk of subtrochanteric and diaphyseal femur fractures: a register-based national cohort analysis. J Clin Endocrin Metab 2010;95:5258-65. 14. Black DM, Kelly MP, Genant HK, et al. Bisphosphonates and fractures of the subtrochanteric or diaphyseal femur. N Engl J Med 2010;362:1761-71. 15. Calltorp J, Adami HO, Aström H, et al. Country profile: Sweden. Lancet 1996; 347:587-94. 16. Shane E, Burr D, Ebeling PR, et al. Atypical subtrochanteric and diaphyseal femoral fractures: report of a task force of the American Society for Bone and Mineral Research. J Bone Miner Res 2010; 25:2267-94. 17. Rothman K, Greenland S, Lash T. Modern epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins, 2008. 18. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83. 19. Schilcher J, Aspenberg P. Incidence of stress fractures of the femoral shaft in women treated with bisphosphonate. Acta Orthop 2009;80:413-5. Copyright 2011 Massachusetts Medical Society. an nejm app for iphone The NEJM Image Challenge app brings a popular online feature to the smartphone. Optimized for viewing on the iphone and ipod Touch, the Image Challenge app lets you test your diagnostic skills anytime, anywhere. The Image Challenge app randomly selects from 300 challenging clinical photos published in NEJM, with a new image added each week. View an image, choose your answer, get immediate feedback, and see how others answered. The Image Challenge app is available at the itunes App Store. n engl j med 364;18 nejm.org may 5, 2011 1737