The 2 million people in the United States living with limb loss are expected to double by the year 2050.1 Hospital costs associated with limb amputation exceed 8.3 billion dollars annually,2 whereas patient demand for more expensive advanced prosthetic technologies is increasing. With rising costs and increasing scrutiny from third-party payers, quantitative outcome measures that are psychometrically sound are needed for acceptance across the various stakeholders to guide clinical decision making involved in prosthetic care. Although common in research, standardized outcome measures are poorly understood and not consistently adopted in prosthetic clinics, where these have potential to guide device design, monitor rehabilitation progress, and provide justification for recommended and prescribed components. Evidence derived from standardized measures supporting the effectiveness of prosthetic interventions for specific patients is necessary for evidence-based clinical practice (EBCP)3 and comparative effectiveness research (CER).
The International Classification of Functioning, Disability and Health (ICF) describes health-related domains that affect persons with disabilities in terms of multidimensional concepts including body function/structure, activity, and participation.4 The ICF component and chapter most relevant to lower-limb prosthetic use are activity and mobility, respectively. The ICF domains of mobility include changing and maintaining body positions; carrying, moving, and handling objects; walking and moving; moving around using transportation; or other.4 For the purposes of this study, the term mobility will be used to refer to the activity domains of transferring oneself, walking, and moving around.
The only system consistently used and required by third-party payers to evaluate mobility in persons with lower-limb loss is the Medicare Functional Classification Level (MFCL) or the K-level system. The K-level system has published guidelines with descriptors for classifying patients into five activity levels (K0–K4). For the purposes of this study, the term activity level will be used to reference the K-level. This system is used by third-party payers to limit patient access to prosthetic technology based on their potential activity level. Patient history, comorbidities, residual limb condition, motivation to ambulate, and the K-level guideline descriptors (i.e., variable cadence, capability to clear environmental barriers) are factors gathered through patient interview and used by the clinical team to estimate and assign patient potential K-level.5 Patient self-report is shown to be less reliable than standardized outcome measures.6 The K-level classification system places reliance on patient self-report and individual prosthetist throughout perceptions when patient interview is the only method used to gather information about patient mobility and to assign a K-level. Much controversy surrounds ambiguity in applying the K-level guidelines to classify patients in real-world clinical settings. The validity of the K-level system as a tool to limit patient access to prosthetic technology has recently been brought into question, citing a lack of scientific evidence and failure to account for factors not listed in the short K-level descriptors.7 Standardized methods for evaluating patient mobility and appropriateness of prosthetic technology are available, with the potential to benefit both patient and payer source.
Several outcome measures have been developed to objectively assess mobility in persons with lower-limb loss. In using outcome measures for clinical research, psychometric properties are important including reliability, validity, and responsiveness. In the application of outcome measures in routine clinical practice, “clinical sensibility”8 and usefulness may be equally important. The Prosthesis Evaluation Questionnaire (PEQ) measures prosthesis-related quality of life and prosthesis-related function.9 Specific to ambulation, the Mobility Subscale (MS) from the PEQ is a 12-point patient questionnaire common in research that evaluates the patient-perceived potential for mobility using prosthetic devices during the past 4 weeks.10 Normative PEQ-MS data have recently been reported for traumatic and dysvascular populations.11 Another outcome measure, the Amputee Mobility Predictor (AMP), is a test used to assess functional ambulatory potential in persons with limb loss either with (AMPPRO) or without (AMPnoPRO) a prosthesis. It consists of 20 goal-oriented balance and ambulation tasks performed by the patient and rated by an observer. The AMP has demonstrated the capability of discriminating K-levels,5 and research has shown that AMP scores are a strong predictor of performance in a 6-minute timed walk test.12 The minimal detectable change (MDC) is a psychometric property that defines the minimum change in an outcome measure score that can be confidently attributed to true change rather than statistical error. The MDC reported for the AMP is 3.4, and the MDC for the PEQ-MS is 0.8.13 Both the PEQ-MS and the AMP were identified as having potential for quantifying outcomes related to the ICF mobility domain during prosthetic rehabilitation.14 Evidence of validity and reliability of the AMP and the PEQ-MS has been reported for persons with lower-limb loss.5,10 The AMP and the PEQ-MS produce quantitative data on patient mobility and have clinical utility for use in cross-disciplinary rehabilitation of persons with limb loss.
The relationship between patient activity level and mobility has been a topic of recent research using standardized outcome measures. The AMPnoPRO has shown overlapping scores of functional ambulatory potential between adjacent K-levels and no finite separation of K-level groups.5 Recent research has shown a benefit from prosthetic feet typically reserved for K3 activity level patients when provided to K2 patients with appropriate physical therapy7 Even improvement in activity level classification of some K2 patients has been reported when they were provided access to K3-level microprocessor knee units.15 These details suggest that patient activity level spans a spectrum, and patients are not easily classified into fixed groups based on K-level descriptors. Further study has been recommended to validate the use of the AMP as a standardized instrument for making K-level assessment.14
To keep pace with the growing health care cost and advanced devices, elevation of the quality of mobility assessment, K-level assignment, and documentation of prosthetic rehabilitation should follow. “Orthotics and Prosthetics could be one of the first areas of healthcare to employ these techniques, take advantage of the new opportunities, and establish common product definitions and comparable outcome measures, en route to rebalancing the cost-quality equation.”16 The goals of incorporating outcome measures into a routine practice protocol include ensuring that each patient is assessed on the basis of standardized quantitative data and reducing reliance on patient self-report and individual prosthetist throughout perceptions to assigning K-level. This will result in providing the most appropriate prosthetic interventions to individual patients and improving overall health care cost-effectiveness.
To achieve this goal, the PEQ-MS and the AMP were adopted as a standardized outcome measure protocol to measure mobility across 11 Ability Prosthetics and Orthotics, Inc (Exton, PA, USA), clinical practices. The PEQ-MS was selected to measure patients’ perception of their mobility outside the controlled clinical environment, and the AMP was selected to measure capability with a functional test in the controlled clinical environment. To our knowledge, this may have been the first time an outcome measure protocol was implemented as routine standard practice to evaluate mobility throughout a multiregion prosthetic clinical setting.
The specific purpose of this retrospective chart review was to look back at a large number of clinical encounters of persons with lower-limb loss and investigate whether standardized outcome measures AMP and PEQ-MS were able to discriminate amputation level and K-level. It was hypothesized that mean AMPPRO and PEQ-MS scores would differ significantly between transtibial and transfemoral amputation levels and across K-levels. The results would provide prosthetist throughout a better understanding of how amputation level and K-level relate to measures of perceived potential for mobility (PEQ-MS) and functional ambulatory potential (AMP). The results could further validate the AMP and the PEQ-MS as standardized instruments for assessing mobility and assigning K-level in the clinical setting.
In adopting outcome measures as a routine standard practice, steps were taken to train staff in the clinics involved in this study. Initially, a single clinic started administering and documenting the AMPPRO and PEQ-MS scores of each patient receiving a prosthesis. A standard way of recording the results in the electronic medical record (EMR) was chosen for future querying and reviewing. Prosthetists in the other clinics were informed of the plan to adopt the outcome measures and were then given articles describing the development and instructions for administering the outcome measures. The prosthetist throughout then participated in a company-wide continuing education course, available from the Online Learning Center through the American Academy of Orthotists and Prosthetists.17 This comprehensive education on clinically relevant outcome measures provided a foundation for the prosthetist throughout to properly administer and interpret outcome measure results. Subsequent education was provided during company-wide biweekly telephone conferences and in-person quarterly meetings to ensure that the outcome measures were being correctly administered. Patient care coordinators and insurance administration personnel from the clinics were also educated on the standardized outcome measure protocol, and these personnel helped keep prosthetist throughout compliant with the newly adopted practice standard. Additional benefits from adopting this protocol were understanding the mobility limitations facing individual patients, targeting interventions to meet specific patient needs, providing continuous quality improvement, and communicating prosthesis-related mobility to other health care partners and third-party payers.
This study reviewed 6 months of clinical outcome data, which included 120 patient charts, between June 1 and December 31, 2012. This period represented the first 6 months after the adoption of the AMP and the PEQ-MS. Data were reviewed at that time to provide timely feedback to the prosthetist throughout about the mobility of the patient populations they served. In all, 12 American Board of Certification (ABC) credentialed prosthetists across 11 different clinics recorded data. Patient charts were accessed through a secure EMR system called OPIE Software. The EMR system was used to create deidentified reports of patient data, which were exported to Microsoft Excel. Inclusion criteria consisted of persons with unilateral transfemoral or transtibial limb loss between 18 and 95 years of age who were given a new prosthetic socket, foot, knee, or whole prosthesis during the chart review date range. Data obtained from the patient charts included age, amputation level, K-level, and AMPPRO and PEQ-MS outcome measure scores. The patient K-level recorded in the charts was assigned by the treating prosthetist on the basis of patient self-report and prosthetist throughout assessment using the published K-level guidelines.
The data were analyzed by a statistician using JMP statistical analysis package (SAS Institute, Cary, NC, USA) to compare AMPPRO and PEQ-MS scores among amputation level and across K-level groups. A critical value of a = 0.05 was used as a limit for accepting a statistically significant result. The Student t-test was used to test the null hypothesis that the AMPPRO and the PEQ-MS did not differ across the amputation level groups. The data set was found to have possible violations of the conditions of equal variance among the groups and normal residuals required for the hypothesis test to yield valid results. The Wilcoxon rank sum test was used allowing for violation of the conditions of equal variance and normal distribution. Analysis of variance (ANOVA) followed by the Fisher protected least significant difference test were used to test the null hypotheses that AMPPRO and PEQ-MS scores did not differ across K-level groups. The data set was also found to have possible violations of the conditions of equal variance among the groups and normal residuals required for the hypothesis test to yield valid results. The Kruskal-Wallis test was used allowing for violation of the conditions of equal variance among the groups and normal distribution. To protect against an inflated overall type I error rate, the Tukey-Kramer honestly significant difference test was also used.
Outcome measure scores were recorded in 67% of the patient charts reviewed. The age and K-level breakdown from the 120 patient charts reviewed is depicted in Table 1. There were no records from K1-level patients found in the charts reviewed. Not all clinical charts had complete sets of outcome scores, with either the AMPPRO or PEQ-MS scores missing from the clinical records. The number of K3 patients was much higher than any other K-level in the study, and transtibial amputation level was more frequent than transfemoral among all the K-levels. An ANOVA of patient age yielded significant variation across the K-level groups (p < 0.05). A post hoc Tukey-Kramer test showed that age of all K-level groups differed significantly (p < 0.05), with the K4 patients being the youngest and the K3 patients being younger than the K2 patients (see Table 1).
The data comparing AMPPRO and PEQ-MS scores for transtibial and transfemoral amputation level are summarized in Table 2. The result of the hypothesis test for AMPPRO with both Student t-test and Wilcoxon rank sum test was failure to reject the null hypothesis, suggesting that there was no significant difference in the mean AMPPRO between the two amputation level groups. The result of the hypothesis test for PEQ-MS with both Student t-test and Wilcoxon rank sum test was also failure to reject the null hypothesis, suggesting that there is no significant difference in the mean PEQ-MS between the two amputation level groups.
The result of the hypothesis test for AMPPRO across K-level with both ANOVA and Kruskal-Wallis test was rejection of the null hypothesis, suggesting that there were significant differences (p < 0.05) in the mean AMPPRO among the three K-level groups. The result of the Fisher protected least significant difference test was that groups K4 and K3 had significantly higher (p < 0.05) AMPPRO means than those of group K2. To protect against an inflated overall type I error rate, the Tukey-Kramer honestly significant difference test was also used, and the results were exactly the same; groups K4 and K3 had significantly higher (p < 0.05) AMPPRO means than those of group K2. The result of the hypothesis test for PEQ-MS across K-level with both ANOVA and Kruskal-Wallis test was rejection of the null hypothesis, suggesting that there were significant differences (p < 0.05) in the mean PEQ-MS among the three K-level groups. The result of the Fisher protected least significant difference test was that groups K4 and K3 had significantly higher (p < 0.05) PEQ-MS means than those of group K2. To protect against an inflated overall type I error rate, the Tukey-Kramer honestly significant difference test was also used, and the results were exactly the same; groups K4 and K3 had significantly higher (p < 0.05) PEQ-MS means than those of group K2.The data comparing the AMPPRO and PEQ-MS scores for different K-levels are listed in Table 3. Histograms showing the frequency of AMPPRO and PEQ-MS scores are found in Figures 1 and 2, respectively. These graphs depict the distribution or spread of patient scores across the population studied.
In review of the number of patients in each K-level group from this study, the K3 patients were the most prevalent and made up 75% of all patients. There were no K1 patients during the period reviewed. Patients at the K3 level were also more frequent in this retrospective study compared with a previous study that included patients across all K-levels (40% K3).5 The prevalence of patients with K3 activity level assignment may have been higher in the group of clinics involved in this study than the general population of persons with lower-limb loss. In addition, transtibial amputation level was three times more frequent than transfemoral amputation level in the charts reviewed. Age differences between the K-level groups were found to be statistically significant, with younger patients belonging to higher K-level groups. These findings correspond to clinical experience and suggest that younger patients more commonly achieve higher mobility.
The histograms of overall AMPPRO and PEQ-MS scores (Figures 1, 2) depict the distribution of functional ambulatory potential and patient-perceived potential for mobility from the entire patient population studied. The histogram of AMPPRO scores (Figure 1) shows the distribution to be left-skewed, with the scores centered on a median score of 41. This histogram depicts the functional ambulatory potential of patients from this study to be centered on a high score, with some of the patients distributed across midlevel scores and one outlier with an AMPPRO score of 11. The histogram of PEQ-MS scores (Figure 2) is again left-skewed. The median and mean PEQ-MS scores for the entire patient data set were similar (31 and 30.6, respectively). The overall SD for PEQ-MS scores (SD, 10.7) was larger than that of AMPPRO scores (SD, 8.1). Examining the shape and spread of both histograms, the patients from this study had a higher functional ambulatory potential but had a larger distribution and lower overall perceived potential for mobility. These observations are representative of mobility of the patient population at the clinics involved in the study, as assessed with the AMPPRO and the PEQ-MS.
There was no statistically significant difference found in AMPPRO or PEQ-MS scores between transfemoral and transtibial amputation level in this study. In addition, the differences in mean AMPPRO and PEQ-MS scores did not reach the MDC for the two measures (3.4 and 0.8, respectively13). This finding differed from expectations and previously reported normative data, in which a statistically significant difference in PEQ-MS scores between transtibial and transfemoral amputation level was found in both dysvascular and traumatic patient populations.11 The relatively small number of patients with transfemoral amputation level in this chart review may have limited the statistical power of this analysis.
When compared with published normative data for PEQ-MS scores of the subjects with transtibial amputation level, the population in this study (mean, 31.9) had lower perceived mobility potential than that of the normative traumatic transtibial group (mean, 37) and slightly higher scores than that of the normative dysvascular transtibial group (mean, 30.5).11 Similar comparison can be seen with the PEQ-MS scores of the subjects with transfemoral amputation level in this study (mean, 27.88), with the perceived mobility potential slightly higher than that of the normative dysvascular transfemoral group (mean, 26.3) but lower than that of the normative traumatic transfemoral group (mean, 34) from published data.11 The perceived mobility potential as measured with the PEQ-MS from both transtibial and transfemoral patient groups in this study was closer to normative data from dysvascular patients. The SD in the overall mean PEQ-MS scores from this study (SD, 10.7) was similar to that of reported normative data (SD, 10.3),11 which demonstrates that the variance in PEQ-MS scores in the population studied was typical of other published results from this patient questionnaire. The patient population in this study was not separated on the basis of etiology, and therefore, a more in-depth comparison with published normative data is not possible.
The AMPPRO and PEQ-MS scores of the K2 patients did differ significantly from that of the K3 and K4 patients in this study. In addition, the difference in the mean scores of the K3 and K4 groups differed from that of the K2 group by more than the MDC for both AMPPRO and PEQ-MS. This finding supports the use of both outcome measures in routine clinical practice to differentiate K2 from K3 and K4 patients. These outcome measures did not find a significant difference between K3 and K4 patients. This may be a result of possible ceiling effects of the AMP and PEQ-MS outcome measures. Recently, a variation of the AMP was developed to address patients with higher mobility and higher functioning, called the Comprehensive High-Level Activity Mobility Predictor (CHAMP).18 This outcome measure can provide prosthetist throughout with an outcome measure that is more sensitive to the differences between higher-functioning K3 and K4 patients.
To provide perspective on functional ambulatory potential of the patient population investigated in this study, AMPPRO data can be compared with previously published data.5 The mean AMPPRO score was lower and the SD was higher across the K2, K3, and K4 groups in this study compared with previously published data (K2 mean [SD], 34.7 [6.5]; K3, 40.5 [3.9]; K4, 44.7 [1.8]).5 The subjects from that study were recruited from hospitals, rehabilitation centers, extended-care facilities, and patient support groups for persons with limb loss. In addition, K-level of each subject in that study was assigned by either a prosthetist or a physician. In this retrospective chart review study, the K-level was assigned by the treating prosthetist from the 11 outpatient clinics. The prior experience and perceptions of the different prosthetist throughout groups may have affected the K-level assignment used to separate the subject groups. The task of assigning a potential K-level was not an object of this investigation; however, the lower mean AMPPRO scores and the increased variance found in this retrospective chart review compared with previously reported data suggest that K-level assignment in the clinical setting may be less consistent and may be more skewed to higher activity level classifications than when activity level is assigned in a controlled research protocol.
A limitation of this study is the small sample sizes of transfemoral versus transtibial amputation levels and the high frequency of K3 patients. A future study that collects data from more patients can be expected to have a larger transfemoral level sample size and more patients in other K-levels. In addition, the etiology of the limb loss was not a parameter analyzed in this study but is known to be a significant factor in the mobility and activity level of persons with lower-limb loss. The current EMR system used by the multiregion clinical practice did not allow querying of amputation etiology, and this feature should be developed for future studies. Missing data were also an issue in this retrospective study, with a few patient charts having no AMPPRO or PEQ-MS data recorded and some charts with only one of the measures recorded. Administering the outcome measures and recording the results in the EMR were an adopted standard practice but may not have always been feasible in the busy clinical settings. Further education on the value of this collective data can motivate better compliance among prosthetist throughout to administer standardized outcome measures. Future research using these standardized outcome measures should include larger patient populations for increased statistical power.
The outcome measure protocol adopted and the data analyzed in this retrospective chart review included a patient report questionnaire of perceived potential for mobility (PEQ-MS) and a functional ambulatory potential test (AMP). These outcome measures were readily implemented as routine clinical practice to make assessment of mobility in persons with lower-limb loss more standardized. The PEQ-MS and the AMPPRO were not found to discriminate between amputation levels in this study. The study did find a significant difference in AMPPRO and PEQ-MS scores across the K-level groups, with the K2 patients having lower scores than the K3 and K4 patients. These findings were especially important because of the controversy surrounding K-level classification between K2 and K3 and its implications for patient access to prosthetic technologies. These results support the adoption of the PEQ-MS and the AMPPRO at Ability Prosthetics and Orthotics, Inc, for assessing mobility and assigning K-level in persons with lower-limb loss.
Incorporating an outcome measure protocol has the potential to address concerns with the K-level guidelines by allowing clinical decisions to be based on standardized measures of patient mobility. Comparative effectiveness research on different prosthetic technologies and their impact on mobility can be investigated alongside clinical practice with the PEQ-MS and the AMPPRO.
The author thanks Jeff Brandt, CPO, and all the prosthetist throughout at Ability Prosthetics and Orthotics, Inc, for administering the standardized outcome measures and supporting the study.
1. Ziegler-Graham K, MacKenzie EJ, Ephraim PL, et al. Estimating the prevalence of limb loss in the United States: 2005 to 2050. Arch Phys Med Rehabil 2008; 89 (3): 422–429.
3. Stevens PM. Barriers to the implementation of evidence-based practice in orthotics and prosthetics. J Prosthet Orthot 2011; 23 (1): 34–39.
4. World Health Organization. International Classification of Functioning, Disability and Health. Geneva, Switzerland: World Health Organization; 2001.
5. Gailey RS, Roach KE, Applegate EB, et al. The Amputee Mobility Predictor
: an instrument to assess determinants of the lower-limb amputee
’s ability to ambulate. Arch Phys Med Rehabil 2002; 83 (5): 613–627.
6. Stepien JM, Cavenett S, Taylor L, Crotty M. Activity levels among lower-limb amputees: self-report versus step activity monitor. Arch Phys Med Rehabil 2007; 88 (7): 896–900.
7. Agrawal V, Gailey R, Gaunaurd I, et al. Influence of gait training and prosthetic foot category on functional outcomes of vascular transtibial amputees—implications for foot prescription. In: Journal of the Proceedings of 38th Academy Annual Meeting and Scientific Symposium. Atlanta, GA; 2012.
8. Feinstein AR, Josephy BR, Wells CK. Scientific and clinical problems in indexes of functional disability. Ann Intern Med 1986; 105 (3): 413–420.
9. Legro MW, Reiber GD, Smith DG, et al. Prosthesis Evaluation Questionnaire
for persons with lower limb amputations: assessing prosthesis-related quality of life. Arch Phys Med Rehabil 1998; 79 (8): 931–938.
10. Franchignoni F, Giordano A, Ferriero G, et al. Measuring mobility in people with lower limb amputation: Rasch analysis of the mobility section of the Prosthesis Evaluation Questionnaire
. J Rehabil Med 2007; 39 (2): 138–144.
11. Hafner B, Amtmann D, Abrahamson D, et al. Normative PEQ-MS and ABC scores among persons with lower limb loss. In: Journal of the Proceedings of the 39th Academy Annual Meeting and Scientific Symposium. Orlando, FL: American Academy of Orthotists & Prosthetists; 2013.
12. Gailey RS. Predictive outcome measures versus functional outcome measures in the lower limb amputee
. J Prosthet Orthot 2006; 18 (6): P51–P60.
13. Resnik L, Borgia M. Reliability of outcome measures for people with lower-limb amputations: distinguishing true change from statistical error. Phys Ther 2011; 91 (4): 555–565.
14. Deathe AB, Wolfe DL, Devlin M, et al. Selection of outcome measures in lower extremity amputation rehabilitation: ICF activities. Disabil Rehabil 2009; 31 (18): 1455–1473.
15. Kahle JT, Highsmith MJ, Hubbard SL. Comparison of nonmicroprocessor knee mechanism versus C-Leg on Prosthesis Evaluation Questionnaire
, stumbles, falls, walking tests, stair descent, and knee preference. J Rehabil Res Dev 2008; 45 (1): 1.
16. Heinemann AW, Fisher WP Jr, Gershon R. Improving health care quality with outcomes management. J Prosthet Orthot 2006; 18 (6); P46–P50.
18. Gaunaurd IA. The Comprehensive High-Level Activity Mobility Predictor (CHAMP): A Performance-Based Assessment Instrument to Quantify High-Level Mobility in Service Members With Traumatic Lower Limb Loss. Coral Gables, FL, University of Miami, 2012.