Patients with fresh fracture who used LIPUS had a 96% HR (Table 1), whereas the HR from the literature averages 93% [15-20]. Thus, LIPUS may have reduced the nonunion rate by up to 40%, with respect to that literature [15-20]. Age did not have a significant impact on fracture HR, and the HR for patients aged 30 to 79 years was equivalent (Figure 1). However, HR is confounded by other risk factors for nonunion, such as elevated BMI (Figure 2). Fracture HR varies as a function of the bone broken (Table 4), but the most compelling risks for treatment failure are patient age, open fracture, treatment delay, and use of medications (Table 5). HR can be reduced by medical comorbidities (Table 6) and by medications such as prescription NSAIDs, anticoagulants, steroids, and calcium channel blockers (Table 7). Unlike prescription NSAIDs, non-prescription NSAIDs do not have a significant effect on HR (Table 7).
Evaluation of LIPUS-treated patients over age 30 did not show the decline in HR (Figure 1) often described in the literature [15-20]. However, patients who failed to heal with LIPUS (Table 5) were significantly older than patients who did heal. We surmise that this apparent discrepancy is explained by LIPUS-treated patients under age 30, who healed well and reduced the average age of healed patients (Table 5).
There is strong evidence in the literature that increasing age is associated with impaired fracture healing [2,21,22]. Femoral neck HR decreased with age, both in displaced (p < 0.001) and non-displaced (p < 0.003) fractures [2]. Among patients under age 60, just 7 of 106 had fracture nonunion (HR = 93.4%); among patients age 70 to 80 years, 84 of 337 had nonunion (HR = 75.1%) [2]. In a separate study, patients with displaced femoral neck fracture who developed nonunion were significantly older than patients with normal healing [21]. Nevertheless, in a third study, age was not identified as a predictor of nonunion in 202 patients with intracapsular femoral neck fracture [22].
Patient age has an impact on tibial healing in the absence of LIPUS [23-26]. Among 42 patients treated surgically for tibial plateau fracture, the failure rate was just 7% in young patients, but increased to 79% in patients older than age 60 [23]. Age was a risk factor for total knee arthroplasty among 8,426 patients (p < 0.0001) with tibial plateau fracture [24]. Increasing patient age was associated with worse surgical outcomes after tibial surgery [25,26]. Yet age was not noted as a nonunion risk factor in studies of 211 patients with long bone fracture [20] or 416 patients with tibial shaft fracture [27].
Age can also have an impact on healing in humeral fracture. A retrospective review of nonunions of the midshaft humerus concluded that advanced age accounted for 57% of nonunions [28]. A prospective study of 110 elderly patients with humeral diaphyseal fracture found that nonunion was predicted by patient age (p < 0.05), prior stroke (p < 0.001), or prior nonunion (p < 0.001) [29]. In a review of 37 patients with humeral fracture, the nonunion rate was 10.8% and patient age was the only significant predictor of humeral nonunion [30]. Finally, age is a risk for nonunion in clavicular fracture. Among 337 patients with clavicular fracture, patient age was an important predictor of nonunion [31], and older patients with clavicular fracture typically had a higher risk of nonunion [32,33].
Our data suggest that patients who are obese generally do not heal as well as people of ideal weight (Figure 2). Obesity has been recognized as a risk factor for fracture nonunion for decades [34], and BMI is significantly higher in patients with failed reduction of the distal tibia [35]. Obesity (BMI ≥30) is also a significant risk factor (p < 0.01) for required revision of total hip replacement after femoral neck fracture [36], and BMI is known to interact with risk factors for nonunion, including smoking [37].
A major strength of the registry database is the size of the cohort, which we believe to be the largest ever reported. This enabled us to gain insight into complex relationships. For example, BMI interacts with patient age in a complex way (Figure 2); obese patients generally do not heal as well as non-obese patients. Patients older than age 60 who are obese have a substantially lower HR than either young obese patients or elderly (≥60 years of age) patients of ideal weight. This could be because patients who are elderly and obese can have a range of other health problems, including diabetes, hypertension, and vascular insufficiency (Table 6), and may be taking various medications (Table 7). Conversely, patients under age 20, or underweight patients, have a higher HR than other patient groups (Figure 2).
A second strength of this study is that patients enrolled in the registry did not differ in obvious ways from a random sample of fracture patients. Roughly 16.2% of registry patients were smokers in 1994 to 1998 (Table 6), which is consistent with the 18.1% of people who are currently smokers in the United States [38]. Roughly 5.6% of registry patients had diabetes, which is comparable to the 6.6% of Americans age 45 to 64 years who had diabetes in 1996 [39], the median year in the registry. Roughly 4.7% of registry patients had hypertension, somewhat less than the 10.5% of Americans (age 18–44 years) who had a diagnosis of hypertension between 2005 and 2008 [40].
A limitation of this study is that blinding was impossible, and patients may have been motivated to perceive a treatment benefit, even if none existed. Because the registry collated outcomes from treatment, it is a therapeutic study, and therapeutic studies are stronger, when blinded [41]. Results reported herein are a retrospective analysis of standardized, prospectively-collected treatment data, so they provide Level III evidence [41]. Yet the large number of patients (N = 4,190; see Table 1) and the relatively high rate of retention (73%; see Table 1), may overcome some of the limitations inherent to a registry study [42].
A second limitation of this study is that the data are as much as 20 years old. It has been argued that fracture management has not changed substantially since the registry opened [43]. Intramedullary nailing was accepted as treatment for long bone fracture prior to when the bulk of the registry patients were enrolled [44]. Therefore, many of the registry patients appropriate for nailing may have received that treatment. We cannot exclude the notion that changes in patient management may have had an impact on HR of patients in the registry, but we believe that such changes were probably evolutionary, not revolutionary.
A third limitation of our work is that registry studies typically capture a limited dataset based on clinical convenience, sacrificing data granularity for breadth of clinical capture. While it would be useful to have more detailed information about the precise location of each fracture within the bone, or the extent of associated soft tissue injury, or even why the physician elected to treat a fracture with LIPUS, these fields were not in the database.
A final limitation of this study is that the calculated number-needed-to-treat (NNT) for patients is currently 33, in order to avoid one nonunion. However, this calculation is misleading. If fresh fracture patients are selected for treatment based on the clinical understanding of risk factors contemporary with registry enrollment, then NNT would be 33. But a combination of advanced age and obesity decreases the HR substantially (Figure 2), and patients who are aged, obese, and diabetic may show a further reduction of HR. Therefore, the NNT for an aged, obese, and diabetic patient may be lower than 33, although a multivariate analysis will be required to test this hypothesis.
It is very important that fracture treatment in the elderly be effective, because there can be severe consequences of even brief physical inactivity. Sarcopenia is the loss of lean muscle mass that occurs even with exercise during healthy aging [45]. Elderly adults lose muscle mass and lean body tissue far more rapidly than do young adults during prolonged physical inactivity [46]. Ten days of experimental bed rest in otherwise-healthy 67-year-old adults resulted in a 14% loss of power, a 13% loss of strength, and a 12% loss of aerobic capacity [47]. Though physical performance was not impaired by bed rest among healthy adults [47], one might expect physical performance to be diminished in adults with fracture. Elderly patients hospitalized with acute illness experience a rapid functional decline [48]; among 71 patients (average age = 74 years), two-thirds experienced a functional decline by the second day in hospital. Assessments of physical mobility, toileting, incontinence, feeding, grooming, and mental status showed a continued decline during hospitalization in 10% of patients, and most patients had no improvement in functional ability until after release from the hospital [48]. A more recent study involving 2,293 patients (age ≥70) showed that one-third of patients declined in activities of daily living between hospitalization and discharge [49]. The frequency of functional decline increased markedly with age: 23% of patients aged 70–74 declined in function, but 63% of patients >90 years old experienced such a decline [49]. Femoral neck fracture in the elderly is associated with a 5-fold elevation in risk of mortality in 1 year, with deficits in mobility, respiratory and renal function, cognition, and endocrine function [50]. If LIPUS can be used to mitigate age as a risk factor for healing, then the severe impact of physical inactivity post-fracture can potentially be minimized.