There are several studies that have evaluated subtrochanteric and femoral shaft (ST/FS) by the International Classification of Disease (ICD) codes , which have allowed large-scale studies of numerous data sources. These data can be used to determine the potential impact of these fractures in terms of providing an upper limit of the number of atypical fractures to expect, since only a small fraction of ST/FS fractures are atypical. However, with this type of database study, it is impossible to determine if the fractures are atypical since there is no radiographic adjudication. In addition, when these databases are used, the results can be impacted by the source of the data, whether there was a restriction on age, exclusion of trauma or other diseases, as well as which ICD codes were used in the analysis. For example, a comparison of 3 studies [8–10] used to evaluate trends in hip fracture rates and ST/FS fractures is found in Table 6.1. Ng used a subanalysis to determine that ST/FS fractures are increasing in women age 60 and older who sustained a fracture with moderate trauma, and this includes falls from standing height [10]. There are differences noted in terms of the database evaluated, age studied, ICD-9 codes utilized, and the exclusions applied as well as the results. An additional problem in studies that use ICD-9 codes for finding fracture cases is that the fracture location could be misclassified. In fact, only 36 % of femoral shaft fractures were correctly identified for their location [11]. However, despite these limitations, many studies have found that the rate of ST/FS fractures have increased recently [10, 12, 13].

Several studies have found that subtrochanteric and diaphyseal femur fractures share the epidemiology of classic osteoporosis-related hip fractures in terms of increasing incidence with age and higher rates in females than males [8, 14].

Danish administrative data [15] were queried to investigate the relationship between bisphosphonate use and subtrochanteric or femoral shaft fractures among a large number of alendronate users (n = 39,567) and nonusers (n = 158,268). Alendronate users were more likely to have hip fractures than nonusers (HR, 1.5; 95 % CI, 1.4–1.5). In addition, alendronate users were more likely to have subtrochanteric or femoral shaft fractures (HR, 2.0; CI, 1.8–2.3); however, the risk did not vary with length of bisphosphonate use. Of note, these were not AFF since there was no radiographic review. A similar lack of association between BP use and ST/FS fractures was found in another Danish study using untreated controls and in both a US study and one based on data from Taiwan both of which used other osteoporosis drug users as controls [16–18]. A nested case -control study in Canada did find that BP use for more than 5 years as compared to transient use was associated with ST/FS fracture (OR = 2.7, 95 % CI 1.25, 6.02), although as expected there was a concomitant reduced odds of classic hip fractures [19]. Again, there was no radiographic review, so these were not AFFs.

Several clinical and radiographic features were identified from numerous case reports and series [20–29] which have helped to differentiate atypical fractures as being distinct from osteoporosis-related, prosthesis-related, or major trauma-related fractures. Asian women appear to represent a greater proportion of persons with AFF as compared to the population proportion [30, 31]. One of these features is now included in the minor features of AFF prodromal pain in the thigh or leg for weeks or months prior to the fracture [20, 24, 25, 32, 33], lack of trauma precipitating a fracture [20, 21, 24, 25], bilaterality in approximately 30 % (either simultaneous or sequential) [21, 25, 27, 28] and typically at the same location in the contralateral leg [34], transverse fractures [24], cortical hypertrophy or thickness [25], stress reaction on the affected and/or unaffected side [20, 22, 24, 25, 27], and poor fracture healing [25, 26]. Other features identified with AFF include use of proton pump inhibitors , steroid therapy , in addition to the bisphosphonate [22, 25, 28, 35–38], and adherence to bisphosphonates [39] and other antiresorptive agents [38, 40–42], although in some cases there was prior bisphosphonate use. Normal or low bone mass but not osteoporosis in the hip region [28, 43–45] has also been reported. Comorbid conditions (e.g., vitamin D deficiency, rheumatoid arthritis, hypophosphatasia) have also been related to AFF [45, 46]. A recent study of 11 AFFs compared to those with typical femoral fractures (n = 58) identified hypocalcemia due to latent hypoparathyroidism as primary risk factor for AFF; younger age, higher BMI, early menopause, and less compromised BMD may also influence the development of AFF [47]. Structural features and hip geometry [34, 48] may also be important, and in fact 12 elderly women had AFF associated with bowing deformity, and only 6 of 12 had prior bisphosphonate use [49]. This is further described in Chap. 7. Although cortical thickening was thought to relate to long-term BP use, this is not true, and increased subtrochanteric femoral cortical thickening may instead be a risk factor. In fact, many AFFs occur at the point of greatest cortical thickening [50, 51]. A registry for AFF would help to identify more commonly underlying features sinc e additional case series are unlikely to continue to be published.

The ASBMR Task Force [1] had extensively reviewed the epidemiologic studies concerning AFF, and these studies are found in Table 6.2 [30, 33, 52–61]. Many of these studies found an association between bisphosphonate use and AFF [30, 33, 52–58]. The benefits of bisphosphonates have been estimated to be 100-fold greater than the risk of AFF [38].