Outcome Measures in Lower Limb Prosthetics

Brian Kaluf CP, FAAOP

Sara J. Morgan CPO, PhD

Brian Kaluf or an immediate family member serves as a paid consultant to or is an employee of Ability Prosthetics and Orthotics; has stock or stock options held in Ability Prosthetics and Orthotics; has received research or institutional support from Ottobock and Parker Hannifin; and serves as a board member, owner, officer, or committee member of American Academy of Orthotists and Prosthetists. Dr. Morgan or an immediate family member serves as a board member, owner, officer, or committee member of American Academy of Orthotists and Prosthetists.

This chapter is adapted from Stevens PM: Outcome measures in lower limb prosthetics, in Krajbich JI, Pinzur MS, Potter BK, Stevens PM, eds: Atlas of Amputations and Limb Deficiencies: Surgical, Prosthetic, and Rehabilitation Principles, ed 4. American Academy of Orthopaedic Surgeons, 2016, pp 645-662.

ABSTRACT

The administration, interpretation, and communication of clinical outcome measures have emerged as vital competencies for prosthetists and other healthcare clinicians. Several patient-reported and performance-based outcome measures have been designed for or tested with people who have lower limb loss and are deemed well-suited for clinical applications. It is helpful to consider the benefits of outcome measurement and to have useful tips for incorporating these measures into routine clinical practice.

Introduction

The use of standardized outcome measures has become increasingly important in the provision and management of prosthetic care. When used regularly, outcome measures can improve quality of care and help maximize the overall value of prosthetic rehabilitation. Consistent use of relevant outcome measures can help clinicians predict patient benefit from a prosthetic intervention, document patient improvement, and inform clinical decision-making.¹ Proper selection of high-quality outcome measures, appropriate training in their use, familiarity with their measurement properties, and development of strategies to standardize assessments are equally necessary steps if a practitioner aims to realize the benefits of a measurement program.² These steps ensure that routine administration of outcome measures is accompanied by correct interpretation of scores, meaningful discussions about outcomes with patients, family members, and healthcare clinicians, and documentation of clinical changes over time.³

It is important to define measurement terms and properties, describe examples of outcome measures that are well-suited to assessment of mobility, pain, and health-related quality of life in people with lower limb loss, and to discuss strategies for incorporating routine outcome measurement into a clinical setting. It is also important to focus on the clinical use of outcome measures, but this information may also be useful for the incorporation of standardized outcome measures in research settings.

Measurement Terms and Properties

Measurement Terms

Clinical measurement is defined as assigning numerals to variables to better understand, evaluate, or compare patients on a specific attribute.⁴ Clinicians often measure variables that can be directly observed and are based on physical or physiological properties (eg, height, weight, or temperature). However, many important clinical variables are abstract and require shared definitions and indirect forms of measurement to understand a patient’s current state and change over time.⁴ These abstract variables (eg, mobility, pain, quality of life) are called constructs, and make up many of the clinical outcomes that are important to patients with limb loss and their practitioners.

Outcome measures are standardized instruments that are developed to quantify constructs of interest. Outcome measures are usually developed for the purposes of discrimination, evaluation, or prediction⁴ and can be categorized as patient-reported outcome measures or performance-based
outcome measures. Patient-reported outcome measures are questionnaires completed by a patient or their caregiver (proxy-reported outcome measures). Patient-reported outcome measures assess a patient’s perspective or experience outside the controlled clinical environment. Performance-based outcome measures are assessed by a clinical professional based on observations of patients completing a task or activity, and the scores are not influenced by patient perspectives.⁵^,⁶ Performance-based outcome measures are typically administered in a clinical environment, which may not reflect the variety of settings encountered in a patient’s daily life. Patient-reported and performance-based outcome measures provide unique and complementary information about a patient’s health outcomes, and thus both are recommended for clinical use.

Measurement Properties

An understanding of psychometric properties, such as validity, reliability, and responsiveness to change, is useful to guide the selection of outcome measures.⁷ Evidence of a measure’s psychometric properties can help clinicians distinguish between measures that are more or less suitable for assessment of a specific patient cohort, treatment, or clinical circumstance.¹^,⁸^,⁹ Validity, reliability, and responsiveness are frequently reported in outcome measurement research and can be examined with statistical methods.⁹

Validity is the extent to which a measure assesses the construct of interest.⁴^,⁷^,⁹ A measure’s validity can be described as evidence of content validity, criterion validity, and construct validity.⁹ Evidence of content validity demonstrates that items within a measure adequately assess all aspects of the construct of interest, and is often based on expert opinion, earlier literature, and other research activities that guided the development of the measure. Evidence of criterion validity demonstrates a relationship between a measure’s score and scores on a benchmark measure.⁴^,⁹ Similarly, evidence of construct validity demonstrates that a measure’s scores correlate with those of another measure of the same construct (convergent validity) and are unrelated to scores from a measure of a different construct (discriminant validity). Evidence of construct validity can also be demonstrated with differences in scores across diverse patient groups (known groups validity).⁹ Evidence of validity is often reported in terms of strength of association using correlation coefficients (eg, Pearson product-moment coefficient [r], Spearman rho [ρ]), and with statistical comparison tests for known groups validity.⁴ There are no statistical indexes for content validity.⁴

Reliability is the extent to which scores from an outcome measure are consistent and free from error.⁴ When a measure is sufficiently reliable, differences between scores can be attributed to real differences in the construct rather than measurement error or noise.⁷^,⁹ A measure’s reliability can be described as test-retest reliability, rater reliability, and internal consistency.⁹ Test-retest reliability is the consistency of scores across timepoints, and rater reliability is consistency of scores across (interrater) or within (intrarater) observers. Intraclass correlation coefficients or kappa statistics are calculated to assess evidence of test-retest and rater reliability, depending on the type of data.³^,⁴^,⁹ Clinical measures are often considered sufficiently reliable if reliability coefficients are 0.8 or higher,⁴ though some suggest that reliability coefficients of 0.9 or higher are needed for the assessment of individual patients over time.⁹ Internal consistency is the extent to which items within a measure are related. Cronbach’s alpha (α) is used to evaluate internal consistency, and values of 0.7 to 0.9 are considered to be strong.⁴

Responsiveness is the extent to which a measure can detect a change that is clinically important.⁷^,⁹ Minimal detectable change (MDC) and minimal clinically important difference (MCID) are estimated values in the units of a particular measure that help clinicians interpret changes in an individual’s scores over time. MDC represents the smallest change that exceeds measurement error and can be attributed to a real change in the construct of interest.⁴ MDC is related to the reliability of the outcome measure; measures with high reliability have small estimates of MDC. MDC estimates are commonly reported at 90% or 95% confidence intervals (MDC90 and MDC95, respectively).⁴^,¹⁰ MCID represents the minimal change in scores that represents a true and meaningful improvement or worsening in the construct of interest.⁴ Estimates of MDC and MCID aid in interpretation of change over time, and facilitate clinical decision-making based on changes in scores.

Ceiling and floor effects are an important consideration when choosing an outcome measure. Measures with ceiling effects are unable to detect meaningful changes in the construct at the highest extremes of the score, and conversely, measures with floor effects are unable to detect changes at the lowest extremes.³^,⁹^,¹¹ These effects are examined by plotting the frequency of scores and examining the shape of the distribution.¹⁰^,¹² If 15% or more of the scores are at one extreme of the scale, the measure is likely exhibiting a ceiling or floor effect.¹³

Another important concept in measurement is the availability of normative or reference data and criterion scores. Scores on an outcome measure may be challenging to interpret without information about scores that are typical for the general population, or for people who have similar clinical characteristics to the patient being measured. Normative or reference data are data collected from a large sample that provide typical scores as well as information about the spread of data. Normative scores can help clinicians and patients understand what a score means, and how a patient’s score is related to scores within a general or representative group. Criterion scores identify a level of the construct that is clinically meaningful and might indicate that the patient is at higher or lower risk for a clinical event (eg, falls).⁴

An outcome measure can never be assumed to be free from error or perfectly accurate, but evidence of acceptable reliability, validity, and responsiveness can increase a clinician’s
confidence that the measure will provide stable scores and meaningful information throughout a patient’s course of care.³^,⁴^,⁹ It is important to note that the measurement properties of existing outcome measures are often continually researched and refined, and thus there may be a range of values available for each measure under consideration.³ Comparisons of relative measurement properties between measures can inform selection of outcome measures best-suited for evaluating and guiding clinical decisions for individuals and small patient populations.³^,¹⁰

Mobility, Physical Function, and Balance Measures

Amputee Mobility Predictor (AMP)

The Amputee Mobility Predictor AMP is a performance-based outcome measure developed specifically for individuals with lower limb loss to assess functional capabilities and mobility.¹⁴ It rates performance in 21 tasks involving sitting balance, transfers, standing balance, gait, stair ascent and descent, and use of an assistive device. The AMP can be administered without a prosthesis (AMPnoPRO) or with a prosthesis (AMPPRO). Examples of these tasks are depicted in Figure 1. Scores on the AMPnoPRO range from 0 to 43 points, and scores on the AMPPRO range from 0 to 47 points. The AMP has demonstrated excellent interrater and intrarater reliability¹⁴ and a reported MDC90 of 3.4 points.¹⁰ Concurrent validity was assessed by comparing AMPPRO and AMPnoPRO scores with the Amputee Activity Scale and the Six-Minute Walk Test (6mWT).¹⁴ A known groups method was used to establish evidence of construct validity, with AMP scores distinguishing between Medicare Functional Classification Level (MFCL) groups.¹⁴ The MFCL (ie, K-levels) provides guidelines for clinicians to categorize patients into functional levels ranging from K0 through K4 based on rehabilitation potential and ambulatory ability.¹⁵ A recent study found evidence of predictive validity of the AMPnoPRO before initial fitting with future performance on the Two Minute Walk Test (2mWT), the Timed Up and Go (TUG), and MFCL assignment at the end of prosthetic rehabilitation.¹⁶ Reference data are included in the initial article that described the development of the measure,¹⁴ as well as other studies.¹⁷^,¹⁸

The AMP was developed to help predict and discriminate between MFCL.¹⁴ The relationship of the AMP score with the MFCL and the 6mWT is strong,¹⁴ and knowledge of AMP score in combination with other patient-related factors (ie, age, amputation level, comorbidities, ability to balance on one leg, and manual muscle testing) can predict the MFCL and potential to ambulate with a prosthesis. However, the AMP developers have cautioned that no clear cut-off scores exist, and AMP scores were overlapping when subjects were grouped based on professional-rated MFCL.¹⁴ Other authors have cautioned against applying AMP cut-off scores for assigning MFCL without assessing other factors.¹⁶^,¹⁹ Because the initial study evaluated only current prosthesis users, it may not be generalizable to persons with lower limb loss who are still recovering and undergoing rehabilitation following surgery.¹¹

FIGURE 1 Clinical photograph of Amputee Mobility Predictor (AMP) item # 8 single limb standing balance for 30 seconds. (Reproduced with permission from Umbrell C: Meaningful Measures. O&P Almanac. 2021, vol 70, No. 7, p 22. https://issuu.com/americanoandp/docs/august_2021_issue/s/13248391.)

FIGURE 2 Clinical photograph of 10-Meter Walk Test (10mWT) setup with cones for acceleration and deceleration zones. (Reproduced with permission from Umbrell C: Meaningful Measures. O&P Almanac. 2021, vol 70, No. 7, p 22. https://issuu.com/americanoandp/docs/august_2021_issue/s/13248391.)

Timed Walk Tests (TWTs)

TWTs encompass a variety of performance-based outcome measures that assess readiness to ambulate and the capacity for exercise.¹⁴^,²⁰ Initially introduced as the Twelve-Minute Walk Test, more frequently reported TWTs for individuals with lower limb loss are shorter in duration (ie, 2mWT and 6mWT). Required equipment includes a stopwatch and a measuring tape or distance measuring wheel. A TWT should be administered in a quiet area such as a hallway, the walking speed is stipulated by the rater with standardized verbal instructions, the rater should walk behind the patient, and rest should be permitted during the test, if needed¹¹^,²¹^,²² (Figure 2). When the test is administered in a hallway, setting up cones 30 m apart and measuring the distance (to the nearest 0.10 m) from the last cone to where the patient stops
can facilitate scoring of the measure.²² Shorter standardized walking courses are permitted, but they negatively influence distance travelled and reference values are no longer applicable. The score is reported as distance traveled (m), or average walking speed (m/s) can be calculated.²¹

Six-Minute Walk Test (6mWT)

Typically viewed as a benchmark for assessing walking ability, the 6mWT instructs patients to walk as far as possible in 6 minutes. The 6mWTscores have been shown to differentiate amputation levels²³ as well as MFCL groups (K0-K4).¹⁴ Strong evidence of convergent validity with the Timed Up and Go (TUG) test also has been demonstrated.²⁴ Evidence of the excellent reliability of the 6mWT has been established in persons with lower limb loss,²³^,²⁴ with a reported MDC90 of 45 m.¹⁰ Reference average scores for the different MFCL groups have been reported.¹⁴ In individuals with lower limb loss, a large amount of the variance found in the 6mWT was predicted by age, muscle strength, balance, time since amputation, cause of amputation, and level of amputation.²⁵ A walking distance of 191 m at the time of discharge from rehabilitation was found to be predictive of sustained prosthesis use 12 months after discharge.²⁶ While not established separately for persons with lower limb loss, the cut-off for community ambulation is reported as 0.8 m/s²⁷ or 300 m on the 6mWT,²⁸^,²⁹ and an accepted threshold for a substantial change in walking speed is 0.1 m/s or 50 m on the 6mWT.³⁰

Only gold members can continue reading. Log In or Register to continue