Can We Catch Them Before They Fall?
June 2019 Issue
If you conduct a Google search using the term "amputee falls" the first return is a link to a short article on the Amputee Coalition website.1 The article, "Tips on Preventing and Handling Falls," carries as an ominous subtitle, "Lower-extremity amputees fall, right? It's not a question of if, but when." The article describes simple considerations that may pose unintended trip hazards (such as pets and throw rugs), as well as guidance on how to fall (bend your joints and spare your head). The article and its revealing subtitle underscore the reality that individuals with lower-limb amputations contend with elevated fall risks. A growing body of evidence has clarified the incidence of falls in this population, the factors that may elevate that risk, and some screening considerations that can help identify those individuals at the greatest risk of a fall. This article considers the construct of falls among individuals following lower-limb amputation with reliance on several recent publications.
Incidence of Falls Through Phases of Rehabilitation
Unfortunately, the collectively reported fall incidence among individuals with lower-limb amputations supports the earlier suggestion that it's not a question of if, but when. However, a recent systematic review on this topic that focused on fall rates during various stages of post-amputation recovery provides a more nuanced understanding.2 More specifically, the authors considered the relative fall rates associated with the acute post-operative phase of hospital admission, during inpatient rehabilitation, and after returning to community living.
Immediately following amputation, possibly misled by phantom sensations or compromised by pain medications, individuals can forget that they have undergone an amputation and fall during basic transfers. Confined to a single retrospective analysis of 370 individuals who underwent primary or revision unilateral amputation and were still in the surgical ward and had not yet been fitted with a prosthesis, a fall rate of 16.5 percent was reported.3 Prominent risk factors for falls during this early phase of rehabilitation included dysvascular etiology, transtibial amputation, and a right-side amputation.3
During inpatient rehabilitation, as patients are fitted with a prosthesis and begin to stand and walk within these new constraints, they must learn how to transfer their body weight on and off of their prosthesis and ensure toe clearance with a static prosthetic ankle mechanism. Systematic review of several publications suggests a fall rate of approximately 25 percent during this phase of recovery.2 One detailed retrospective analysis of 1,460 patients identified prominent risk factors for falls and the associated odds ratios (OR), or the relative increased likelihood of falls associated with a given characteristic. For example, an OR of 2 indicates an individual with a given risk factor is two times more likely to fall than those without that risk factor.4 Prominent among these were being prescribed opiates or benzodiazepines (OR of 5.76 and 2.22 respectively). Additional risk factors included four or more comorbid health conditions (OR = 1.93), taking two or more medications (OR = 1.81), cognitive compromise (OR = 1.68), and an age above 70 years old (OR = 1.40).4
Once individuals return to community living, they will encounter less controlled environments than they did in the inpatient rehabilitation setting. This population, informed in recent systematic review by 14 separate publications, appears to have an annual fall rate of just over 50 percent.2 One detailed analysis of 435 patients identified the prominent risk factors for falls and their associated odds ratios.5 Prominent among these were having four or more problems related to the residual limb or prosthesis and unilateral transfemoral amputation (OR of 3.09 and 2.78 respectively). These were followed by back pain (OR = 1.96) and joint pain (OR = 1.67).5
When the entirety of the available evidence was considered, the presence of multiple comorbid health conditions or other medical problems emerged as a common elevated risk factor for falls.2 Other risk factors were suggested only in single studies or observed with inconsistency across published findings. These areas of less certainty included the presence of vascular disease and amputation level.
Incidence of Fall-related Injury
In addition to fall rates, some authors have also reported on the incidence of injurious falls. These findings have been summarized by rehabilitation phase in a separate systematic review.6 Within the acute surgical recovery stage, injurious falls may be rather common, reported by one author at 61 percent.3 Fortunately, the majority of these (62 percent) were described as being of mild severity.3 By contrast, during inpatient rehabilitation, injurious falls are less frequent, reported at only 18 percent.4 Injurious falls appear to become more common among community-dwelling adults, reported between 27 and 40 percent.6
The Cost of Falls
The best available data on the medical costs associated with falls among those with limb loss comes from the Rochester Epidemiology Project (REP).7 Briefly, the REP capitalizes on the unique healthcare delivery structure observed in Olmsted County, Minnesota, where two multidisciplinary medical groups and a few independent providers have all agreed to share the medical records of consenting patients. These records include cost data, allowing for a comprehensive consideration of those healthcare costs associated with a fall that required a trip to the emergency room or hospital admission.7
Between 2000 and 2014, the REP identified 77 individuals who had either transfemoral or knee disarticulation amputations.7 Of the 46 prosthesis users, 22 individuals experienced 31 falls. Notably, to ensure that medical costs were associated with a given fall event, several constraints were put in place. Specifically, each fall required 12 months of cost data prior to the fall and six months of cost data after the fall (reflecting any cost increases relative to the baseline healthcare expenses of the individual). Further, falls occurring within 18 months of another fall were eliminated, as these falls could still have lingering cost effects from previous falls that might bias the six-month cost of the subsequent falls.7
These necessary constraints further reduced the available data to 17 falls among 16 individuals, inclusive of 11 falls among ten individuals requiring emergency room services and six falls among six individuals requiring hospital admission.7
Using statistical models beyond the scope of this review, the authors identified a median six-month fall cost for those requiring a visit to an emergency department of $18,091.7 The median six-month fall cost for those requiring hospital admissions was predictably higher at $25,652.7 Thus, the cost of injurious falls within this community are substantial.
In addition to exploring potential demographic risk factors for falls, work has been done to identify objective screening measures that may prove predictive of falls. Among these, poor vibration sense and increased gait variability have been identified as possible screening indices. For example, individuals with poor vibration sense were described as having experienced three times more falls than those with good vibration sense.6 Similarly, gait variability with respect to step length, step width, and swing time, assessed with GAITRite data, has been found predictive of elevated fall risk.6
Notably, a recent effort has examined the predictive capacity of outcome measures with greater clinical utility.8 In this cross-sectional analysis of 20 fallers and 20 nonfallers, researchers considered the performance values observed with the Amputee Mobility Predictor with Prosthesis (AMP PRO), the Functional Reach (FR) test, the Single Limb Stance (SLS) test, and the Timed Up and Go (TUG) test.8 For the purposes of this assessment, fallers were those prosthesis users reporting two or more falls in the previous year, while nonfallers had no repeated fall history.
Six of the 20 individuals self-identified as fallers had a transfemoral amputation. All nonfallers had transtibial amputations. While similar with respect to mean age (averaging 57-58 years old), they differed in their Medicare Functional Ambulation Levels. While nonfallers consisted of 13 K3-level subjects and seven K4-level subjects, the fallers cohort consisted of four K2-level subjects, 12 K3-level subjects and only four K4-level subjects. Walking aids were utilized within both cohorts, with cane use identified as the predominant assistive device among both fallers (n = 10) and nonfallers (n = 7).8 Consistent with some of the trends described earlier, differences were also identified with respect to medication use, with 14 of the fallers reporting four or more current medications, compared to only ten nonfallers reporting similar medication consumption.
Performance values on the FR, described as the maximum distance of forward reach a patient can obtain with a fixed base of support failed to discriminate between the two populations.8 The average distance demonstrated by nonfallers of 10.7 inches was not significantly longer than the 9.1 inches observed among fallers.
Similarly, performance values on the SLS, assessed by asking each participant to stand alternately on his or her sound limb and then on the prosthesis and attempt to maintain those single-limb stance positions for up to 30 seconds, also failed to discriminate between the two populations.8 For this measure, the average combined stance times of the sound and prosthesis-enabled limbs was actually higher among the fallers than the nonfallers (13.6 and 12.1 seconds respectively).8
By contrast, differences between the two cohorts were observed with the AMP PRO and TUG. With the AMP PRO, a statistically significant difference of 4.75 points was observed between the mean scores of the fallers (mean score of 36.95/47) and the nonfallers (mean scores of 41.70/47).8
The clinical utility of this information was reported in terms of a cut-off score with an associated sensitivity, specificity, and area under the curve (AUC). The AMP PRO was reported with a cut-off score of 39.5. The sensitivity of that AMP PRO score was reported at 80 percent, meaning 80 percent of those scoring below 39.5 on the AMP PRO were fallers. The related metric of specificity for that AMP PRO score was reported at 60 percent, meaning that 60 percent of those scoring above 39.5 on the AMP PRO were nonfallers. The AUC values for a score of 39.5 on the AMP PRO were reported at 0.70, meaning that the probability of correctly identifying an individual as a faller versus a nonfaller using a score of 39.5 on the AMP PRO was 70 percent.8 While shy of 100 percent, this exceeds a 50 percent probability associated with an uninformative test.
Similarly, significant differences were observed between the two cohorts with the TUG. On this measure, fallers completed the test in an average of 10.67 seconds while nonfallers required an average of 14.84 seconds.8
The cut-off score determined for the TUG test was 10.03 seconds. The sensitivity of that TUG time was reported at 70 percent, meaning 70 percent of those who required more than ten seconds to complete the TUG were fallers.8 The related metric of specificity for that TUG time was reported at 60 percent, meaning that 60 percent of those requiring less than ten seconds to complete the TUG were nonfallers.8 The AUC values for a time of ten seconds on the TUG were reported at 0.68, meaning that the probability of correctly identifying an individual as a faller versus a nonfaller using a TUG time of ten seconds was 78 percent.8 Again, while shy of 100 percent, this exceeds a 50 percent probability associated with an uninformative test.
Collectively, this effort has identified two performance measures of reasonably clinical utility that may assist the rehabilitation community in screening for those prosthesis users who might be at an elevated fall risk.
The risk of falls among individuals who have sustained a major lower-limb amputation is substantial. These risks are comparatively mild during hospitalization immediately after surgery, with progressive increases as individuals transition through inpatient rehabilitation, culminating at just over 50 percent among community-dwelling individuals. Injurious falls are both common and expensive, with mean six-month fall costs ranging from $18,000 to $25,000 for those who require visits to the emergency room or hospital admission. While compromised sensitivity to vibration and increased temporospatial gait variability have both been suggestive of increased fall risk, the AMP PRO and TUG have emerged as clinically viable measures that may help identify those at an increased risk of falls.
Phil Stevens, MEd, CPO, FAAOP, is a director with Hanger Clinic's Department of Clinical and Scientific Affairs. He can be contacted at firstname.lastname@example.org.
1. Amputee Coalition. Tips on Preventing and Handling Falls, https://www.amputee-coalition.org/resources/tips-on-preventing-falls (accessed on March 14, 2019).
2. Steinberg, N., A. Gottleib, I. Siev-Ner, and M. Plotnik. 2018. Fall incidence and associated risk factors among people with a lower limb amputation during various stages of recovery-a systematic review. Disability and Rehabilitation, doi:10.1080/09638288.2018.1449258.
3. Yu, J. C., K. Lam, and A. Nettel-Aguirre et al. 2010. Incidence and risk factors of falling in the postoperative lower limb amputee while on the surgical ward. PM&R (2):926-34.
4. Pauley, T., M. Devlin, and K. Heslin. 2006. Falls sustained during inpatient rehabilitation after lower limb amputation: prevalence and predictors. American Journal of Physical Medicine and Rehabilitation 85:521-32.
5. Miller, W. C., M. Speechley, and B. Deathe. 2001. The prevalence and risk factors of falling and fear of falling among lower extremity amputees. Archives of Physical Medicine and Rehabilitation 82:1031-37.
6. Hunter, S. W., F. Barchelor, and K. D. Hill, et al. 2017. Risk factors for falls in people with lower limb amputation: A systematic review. PM&R 9(2):170-80.
7. Mundell, B., H. M. Kremers, and S. Visscher, et al. 2017. Direct medical costs of accidental falls for adults with transfemoral amputations. Prosthetics and Orthotics International 41(6):564-70.
8. Hakim, R. M., C. M. Frey, K. E. Spadoni, and K. Meyer. 2018. Identifying fallers using clinical balance measures in community-dwelling adults with lower extremity amputation: A cross-sectional study. Journal of Developmental and Physical Disabilities 30:677-88.