MAAT: Providing Understanding and Order to the Construct of Prosthetic Mobility

Home > Articles > MAAT: Providing Understanding and Order to the Construct of Prosthetic Mobility
By Phil Stevens, MEd, CPO, FAAOP

In its original application, maat refers to the ancient Egyptian concept of truth, balance, and order. It was also the name of an Egyptian goddess who personified these virtues and brought order from chaos. However, as an acronym in the field of amputee rehabilitation, it has come to represent one of the largest national analyses of mobility among users of lower-limb prostheses. The Mobility Analysis of AmpuTees (MAAT) series comprises a number of multicenter publications released by Hanger Clinic's Department of Clinical and Scientific Affairs to clarify the concept of prosthetic mobility as it relates to such considerations as satisfaction, quality of life, comorbid health conditions, and prosthetic component choices. This article reviews the findings of this publication series.

The Instruments

For all of the papers in the MAAT series, retrospective analyses of outcomes data collected from patients during the course of their routine prosthetic care were analyzed. The primary outcome measures included the 12-item Prosthetic Limb Users Survey of Mobility (PLUS-M)1 and the well-being subsection of the Prosthesis Evaluation Questionnaire (PEQ-WB).2

The PLUS-M is a patient-reported outcome measure developed by the University of Washington under the guidance of Brian Hafner, PhD. The instrument asks users of lower-limb prostheses to rate the relative level of difficulty of 12 mobility tasks, ranging from (1) Unable to do, to (5) Without any difficulty. Each response is scored (scale 1-5), with the sum of the individual scores converted to a standardized T-score.1 This latter step facilitates comparison to a reference population where a T-score of 50 is consistent with the mean population performance and scores of 60 and 40 represent a single standard deviation above and below that mean.

The PEQ-WB was originally released as a subsection of the larger PEQ. While the comprehensive PEQ is arguably too extensive for routine clinical applications, the PEQ-WB consists of only two questions, asking respondents to reflect and report on their satisfaction and quality of life over the past four weeks.2

In addition to the patient-reported outcomes described, clinical staff augmented the outcome measures with information on prosthetic design and an assessment of the functional comorbidity index (FCI).3 In contrast to other indices of comorbidity that emphasize the potential impacts on mortality, the FCI was developed to investigate the impact of comorbid health conditions on patient function.

MAAT I: Correlating Mobility to Satisfaction and Quality of Life

In the inaugural MAAT publication, published in Prosthetic and Orthotics International, a convenience sample of 509 individuals from across the country were examined with respect to their mobility (PLUS-M) and reported satisfaction and quality of life (PEQ-WB).4 While a relationship between these constructs might be reasonably assumed, such relationships had never been demonstrated in a large cohort of prosthesis users. Unsurprisingly, positive correlations were found between all examined constructs, meaning that increased mobility scores correlated with increases in both quality of life and general satisfaction (r values of 0.511 and 0.475 respectively).4 These findings support the idea that in the holistic care of individuals who have undergone lower-limb amputations, improvements in quality of life and satisfaction may be obtained by maximizing mobility.

MAAT II: Comorbidities and Prosthetic Mobility

The relationship between comorbid health conditions and prosthetic mobility has been a subject of some uncertainty. While some comorbidities, such as hemiplegia following stroke, have been consistently associated with decreased prosthetic performance, other comorbidities, such as diabetes, have been assumed as possible correlates to decreased function with inconsistent findings to support that suggestion. Because these largely unfounded assumptions have begun to negatively influence prosthetic access among some policymakers, the relationship between comorbid health and prosthetic mobility required additional examination.

To address this area of uncertainty, the second MAAT publication, published in the American Journal of Physical Medicine and Rehabilitation, reported on a convenience sample of 596 patients using lower-limb prostheses.5 Mobility was assessed using the PLUS-M, while comorbid health was evaluated by collecting the presence of 18 possible comorbid health conditions reported by each participant. Some of the more frequently reported comorbidities included arthritis, chronic obstructive pulmonary disease (COPD), congestive heart failure, stroke, peripheral vascular disease (PVD), diabetes, upper gastrointestinal disease (such as an ulcer, hernia, or reflux), depression, anxiety or panic disorder, substantial visual impairment, degenerative disc disease, and obesity.5

The diverse patient sample ranged in age from 19 to 95 years old with a mean age of 58. Amputation etiologies included vascular disease (n = 267), traumatic injury (n = 138), non-diabetic infection (n = 65), cancer or tumor (n = 24), and congenital limb deficiency (n = 23). Unilateral transtibial amputations (n = 393) were predominant over unilateral transfemoral amputations (n = 141) and bilateral amputations (n = 62).5

Following regression analysis, the only conditions found to be predictive of decreased prosthetic performance were age, stroke, PVD, and anxiety/panic disorder. However, while a statistical relationship was found, patients over the age of 65 and patients with PVD retained group mobility levels within the minimal detectable change values of the PLUS-M relative to those without these predictive comorbid health states.5 Restated, while PVD and age might be statistically significant, they did not appear to be clinically significant. Further, while stroke and anxiety/panic disorder appeared predictive of compromised prosthetic mobility, conditions such as arthritis, COPD, congestive heart failure, and diabetes were not.5

In a secondary analysis, the total number of reported comorbidities was aggregated for each subject, with a mean PLUS-M T-score determined for each FCI value. While slight declines were observed with increasing numbers of comorbidities, after controlling for the noted predictors, there were no significant differences in PLUS-M values between those without comorbid health conditions and those with seven or more comorbid health conditions. Ultimately, the premise that comorbid health conditions compromised prosthetic mobility in a meaningful way was not supported by these findings.5

MAAT III: The Impact of MPK Technology

While there is a growing body of evidence expanding the support for the benefits associated with microprocessor knees (MPKs), this research has largely been confined to smaller study groups, often in heavily controlled environments that don't reflect a patient's real-world surroundings. In the next MAAT analysis, accepted for publication in the peer-reviewed journal Assistive Technology, prosthetic mobility was analyzed across a large national convenience sample of individuals with MPKs, non-MPKs (NMPKs), and transtibial amputations.6 Initially, 1,285 patients were identified and defined according to amputation level and current knee technology (MPK or NMPK). These individuals were then considered with respect to their FCI score, or the sum of the number of comorbid health conditions reported. To remove outliers and facilitate matched cohorts, the healthiest 150 users of NMPKs were identified and their mean FCI score determined. These individuals were then matched to MPK users and those with transtibial amputations to identify cohorts of 150 individuals with matched mean FCI scores to that of the NMPK users. The mean age of the three cohorts of 150 individuals was comparable, reported at 57.6, 56.5, and 58.4 years among NMPK users, MPK users, and transtibial prosthetic users respectively.6

As would be expected, the greatest mobility was reported by users of transtibial prostheses, with a mean PLUS-M T-score of 52. By contrast, users of NMPKs reported a mean PLUS-M score of 43. Notably, after controlling for comorbid health, the mean PLUS-M score of the MPK users was 48, 10 percent higher than their peers with NMPKs.6 The marked improvement in mobility observed with MPK use was ultimately found to split the functional mobility deficit between transfemoral and transtibial amputation in half. Even with subsequent analyses to remove the confounding effects of age, BMI, amputation etiology, and FCI, these differences persisted.

MAAT IV: The CART Analysis

MAAT IV, accepted for publication in the peer-reviewed journal Disability and Rehabilitation Assistive Technology, represents the series' first movement into predictive analytics utilizing numerous data points to generate complex predictive models. Classification and regression tree (CART) analysis is an increasingly popular approach in healthcare to translate large datasets into useful predictive resources.7 The approach begins with a large dataset with a number of predictor variables. For example, gender, age, amputation etiology, level of amputation, etc. are all potential predictor variables entered into the analysis. The analysis then works through the variables to determine which variables and which values for those variables provide the strongest likelihood of sorting a target population into either of two conditions being predicted.

In the MAAT IV CART analysis, the goal was to predict which legacy users of lower-limb prostheses were likely to function as limited community/household (K1-K2) versus unlimited community (K3-K4) ambulators. The range of variables entered as predictors into the models included gender, height, weight, amputation etiology, smoking history, BMI, FCI, age, level of amputation, PLUS-M score, and satisfaction and quality of life scores.

Using an initial test sample of 554 subjects, a predictive CART model was built. Patient weight, amputation etiology, age, and PLUS-M score variables were ultimately retained in the model. The resultant classification tree presents as a series of binary splits into two branches, each of which leads to additional branches or terminates in a "leaf" with a final probability as to which classification (i.e., K1/K2 or K3/K4) is most likely. This analysis was then tested for its predictive capacity with data from an additional 2,216 patients.7

Each patient begins at the base of the classification tree. Moving from that base, they encounter a series of binary splits. At the first split, was their PLUS-M T-score above or below 36.75? A yes answer directs the patient's case to the next decision node, is the patient above or below 58.8 years old? A no answer directs the patient's case to an alternate decision node, is the patient above or below 59.4 years old? These decision nodes continue until each pathway culminates in a terminal "leaf" issuing a probability for a given ambulation state.7

The rigor of the model is found in its ability to correctly predict community ambulation in 90 percent of those who reached this level of mobility, and household ambulation in 77 percent of those functioning at this level.7 While such predictive models would not be used to define K-level or anticipated functional levels in isolation, when coupled with a patient's individualized evaluation, they can be used to support anticipated K-level assignment.

MAAT V: The Relationship Between Foot Type and Mobility in Those With Dysvascular Amputation

Just as MAAT III contributed to the general understanding between prosthetic knee type and mobility, MAAT V sought to explore correlations between foot type and mobility, with a specific interest in those patients with diabetic or dysvascular amputations.8 In this effort, accepted for publication in The Journal of Rehabilitation and Assistive Technologies Engineering, a convenience sample of 738 patients who used a range of prosthetic foot/ankle technologies including L-5980 (n = 123), L-5981 (n = 342), L-5968 (n = 90), L-5987 (n = 155), and microprocessor feet (MPF, n = 28) was analyzed. After controlling for the covariates of age, body mass index, comorbid health status, time since amputation, and amputation level, significant differences between foot categories were observed, with the lowest mobility associated with L-5980 and L-5981 feet, followed by L-5968 feet, L-5987 feet, and MPFs.8 This constitutes the largest analysis of prosthetic foot type and mobility with implications on foot selection when enhanced mobility is a primary objective in prosthetic rehabilitation.


The MAAT series represents the field's first attempt at leveraging large convenience samples of patient-reported data to better understand the relationships between mobility and such variables as quality of life, satisfaction, comorbid health conditions, and component choices. This series has supported some legacy assumptions, such as the relationship between mobility and quality of life, while challenging others, such as the relationship between comorbid health conditions and inherent mobility restrictions. The series demonstrates the value of implementing outcome measures as a standard of care across multiple clinics and retrospective analyses of those findings.

Phil Stevens, MEd, CPO, FAAOP, is a director with Hanger Clinic's Department of Clinical and Scientific Affairs. He can be contacted at


1.  Hafner, B., I. Gaunaurd, and S. Morgan, et al. 2017. Construct validity of the Prosthetic Limb Users Survey of Mobility (PLUS-M) in adults with lower limb amputation. Archives of Physical Medicine and Rehabilitation 98(2):277–285.

2.  Legro, M., G. Reiber, and D. Smith, et al. 1998. Prosthesis evaluation questionnaire for persons with lower limb amputations: assessing prosthesis-related quality of life. Archives of Physical Medicine and Rehabilitation 79(8):931–938.

3.  Groll, D., T. To, and C. Bombardier, et al. 2005. The development of a comorbidity index with physical function as the outcome. Journal of Clinical Epidemiology 58:595–602.

4.  Wurdeman, S. R., P. Stevens, and J. H. Campbell. 2018. Mobility Analysis of AmpuTees (MAAT I): Quality of life and satisfaction are strongly related to mobility for patients with a lower limb prosthesis. Prosthetics and Orthotics International 42(5): 498–503.

5.  Wurdeman, S. R., P. Stevens, and J. H. Campbell. 2018. Mobility Analysis of AmpuTees II: Comorbidities and mobility in lower limb prosthesis users. American Journal of Physical Medicine and Rehabilitation 97:782-8.

6.  Wurdeman, S. R., P. Stevens, and J. H. Campbell. Mobility Analysis of AmpuTees (MAAT 3): Matching individuals based on comorbid health reveals improved function for above-knee prosthesis users with modern knee technology. Assistive Technology Accepted, pending publication.

7.  Wurdeman, S. R., P. Stevens, and J. H. Campbell. Mobility Analysis of AmpuTees (MAAT 4): Classification tree analysis for probability of lower limb prosthesis user functional potential. Disability and Rehabilitation Assistive Technology Accepted, pending publication.

8.  Wurdeman, S. R., P. M. Stevens, and J. H. Campbell. Mobility Analysis of AmpuTees (MAAT 5): Impact of five common prosthetic ankle-foot categories for individuals with diabetic/dysvascular amputation. Journal of Rehabilitation and Assistive Technologies Engineering Accepted, pending publication.