Activity and Participation, Bimanual Function, and Prosthesis Satisfaction are Strong Predictors of General Well-Being Among Upper Limb Prosthesis Users

Abstract

Objective

To determine the role of multiple factors on general well-being for upper limb prosthesis users.

Design

Retrospective cross-sectional observational design.

Setting

Prosthetic clinics across the United States.

Participants

At the time of analysis, the database consisted of 250 patients with unilateral upper limb amputation seen between July 2016 and July 2021.

Intervention

Not applicable.

Main Outcomes Measures

Dependent variable: well-being (Prosthesis Evaluation Questionnaire- Well-Being). Independent variables included in analysis: activity and participation (Patient Reported Outcomes Measurement Information System [PROMIS] Ability to Participate in Social Roles and Activities), bimanual function (PROMIS-9 UE), prosthesis satisfaction (Trinity Amputation and Prosthesis Experience Scales-Revised; TAPES-R), PROMIS pain interference, age, gender, average daily hours worn, time since amputation, and amputation level.

Results

A multivariate linear regression model using a forward enter method was applied. The model included 1 dependent variable (well-being) and 9 independent variables. Within the multiple linear regression model, the strongest predictors of well-being were activity and participation (β=0.303, P<.0001), followed by prosthesis satisfaction (β=0.257, P<.0001), pain interference (β=-0.187, P=.001), and bimanual function (β=0.182, P=.004). Age (β=-0.036, P=.458), gender (β=-0.051, P=.295), time since amputation (β=0.031, P=.530), amputation level (β=0.042, P=.385), and hours worn (β=-0.025, P=.632) were not significant predictors of well-being.

Conclusion

Reducing pain interference and improving clinical factors such as prosthesis satisfaction and bimanual function with their associated effects on activity and participation will positively affect the well-being of individuals living with upper limb amputation/congenital deficiency.

Acquired amputation and congenital upper limb deficiency (ULA) are associated with a number of disabling characteristics. In addition to the functional challenges associated with the absence of the affected extremity, these individuals contend with a spectrum of pain experiences and social stigma, and in the case of acquired amputation, a dramatic alteration in self-image and vocation.

General well-being has been described as a blended construct combining the elements of quality-of-life and satisfaction.¹ In simplest terms, a person with good well-being is likely to have a positive outlook and affect, noted to generally feel good.^2,3 The restoration or improvement of well-being among individuals with ULA can reasonably be considered one of the primary objectives of rehabilitation. However, restoring or improving general well-being is a difficult objective given the large number of factors that may influence one's well-being.¹ Furthermore, while many factors can influence one's well-being, the effect of these factors is likely to vary among sub-populations. For instance, a recent study among lower limb prosthesis users found their level of mobility was strongly correlated to their general well-being.⁴ Other factors that would be more influential upon the upper limb prosthesis user's well-being are unclear in the literature.

Individuals who have undergone upper limb amputation have reported reduced scores with respect to various physical elements of quality-of-life compared with both population norms⁵ and individuals with lower limb amputation.⁶ Gallagher et al identified frequently encountered broad activity limitations for this population. These included getting dressed (52.9%), taking care of household responsibilities (52.9%), and participating in day-to-day work/school activities (40.0%).⁷ Additionally, in consideration of restrictions to participation, the most frequently identified restrictions have been suggested in employment or job seeking (91.7%), family life (41.2%), leisure/cultural activities (41.2%), sports or physical recreation (38.5%), shopping (35.3%), living with dignity (35.3%), and socializing (23.5%).⁷ Similar findings have been observed with respect to the related construct of satisfaction with life even when controlling for a range of potentially confounding variables such as age and gender.⁸

Beyond the potential increased struggles with activity and participation, bimanual physical function is affected with persons with ULA. A previous study found greater difficulty associated with bimanual activities such as washing and drying dishes and food preparation, and lesser difficulty associated with tasks that lend themselves to 1-handed performance such as driving, brushing one's teeth, and opening and closing doors.⁹ Subsequently, the difficulty and potential negative effect of ULA on bimanual function is likely to have a strong effect on predicting a person's well-being.⁹

While it is possible that constrained participation in activities and social roles, as well as reduced bimanual capacity, may heavily influence well-being in the ULA population, this population also contends with a number of often overlapping pain experiences. Pain is strongly correlated to a person's quality-of-life and well-being. For ULA, pain experiences include phantom limb pain, residual limb pain, and overuse pains experienced in the sound side extremity or torso.^5,10 One or more of these pain experiences are reported by between 70% to 90% of those with upper limb amputation, with most reporting multiple, overlapping pain experiences.^5,10 While phantom limb pain and residual limb pain appear to be more prevalent, overuse pains have been reported as both more severe and disruptive.¹⁰

Subsequently, the purpose of this retrospective analysis was to better understand the relations observed between well-being and multiple factors among a sample of individuals with unilateral ULA who have an upper limb prosthesis. The factors assessed included activity and participation, bimanual upper limb function, satisfaction with a prosthesis, prosthesis wear times, pain interference, age, gender, time since amputation, and amputation level.

Methods

Study design

A retrospective cross-sectional analysis was performed for patient reported outcomes collected within routine care from July 2016 to July 2021. Outcomes included for analysis were limited to those collected at follow-up appointments (ie, any appointment after receiving an upper limb prosthesis). For individuals with multiple follow-up appointments, the appointment with the highest well-being score was selected to represent a state closest to the individual's maximum well-being. Clinics were included from locations across all regions of the continental United States. This analysis was approved by the Western Copernicus Group Institutional Review Board (protocol number: 20170059), and reporting guideline conforms to the Strengthening Reporting of Observational Studies in Epidemiology (STROBE).¹¹ The IRB deemed this study exempt from informed consent.

Subjects

Outcomes were only included for patients with wrist disarticulation, transradial, elbow disarticulation, and transhumeral level ULA. Additionally, outcomes were only used from the unilateral database. There were no restrictions based on type of device the patient used, but all individuals had a prosthesis.

Endpoint measures

Within the standard of care of the involved clinics, clinical outcomes were routinely assessed as part of the prosthetic care provided to patients after upper limb amputation. Specifically, the areas of general well-being, satisfaction with activity and participation, and bimanual physical function were assessed. Well-being was assessed using the validated Prosthesis Evaluation Questionnaire-Well-Being (PEQ-WB) subsection.¹² Note the PEQ-WB was originally administered as part of the larger Prosthesis Evaluation Questionnaire (PEQ) but has been validated and administered independent of the PEQ in its entirety.^4,12 This amputation specific questionnaire asks questions pertaining to satisfaction with amputation status and quality of life. In its original form, the PEQ-WB was administered with continuous visual analog response scales. For ease of scoring, the instrument has recently been administered using discrete scales.^4,13,14 The current analysis used the discrete scale administration with response options from 1 to 10.

Activity and participation was measured using the Patient Reported Outcomes Measurement Information System (PROMIS) instrument: Ability to Participate in Social Roles and Activities (APSRA).¹⁵ The instrument asked questions specific to an individual's reported trouble in participating in activities specific to leisure and work with friends and family. The clinics involved in this analysis used the short form (v.2.0 Short Form 4a).¹⁶ The 5 response options to questions were “Always”, “Usually”, “Sometimes”, “Rarely”, and “Never”. Each response was scored a coded weight value from 1 to 5, respectively. Raw values were summed and then converted to the normalized T-score value according to PROMIS scoring instructions. T-scores were used for all subsequent analysis.

Bimanual physical function was assessed using a custom 9-item PROMIS questionnaire for assessing upper extremity function (PROMIS-9 UE). The PROMIS-9 UE consists of 9 questions from the PROMIS upper extremity item bank (v2.0). Items deemed to require bimanual engagement were previously selected by a panel of subject matter experts, and subsequently validated for physical function assessment in individuals with upper extremity amputation.¹⁷ Similar to the APSRA, the PROMIS-9 UE has 5 response options “Unable to do”, “With much difficulty”, “With some difficulty”, “With a little difficulty”, and “Without any difficulty”. Responses are again scored from 1 to 5, respectively. Raw values were summed across all responses, and then the score converted to a normalized T-score value per PROMIS methods. T-scores were used for all subsequent analysis.

In addition to bimanual physical function, prosthesis satisfaction and pain interference were captured as potential predictors in each of the models. Prosthesis satisfaction was measured using the single response question from the Trinity Amputation and Prosthesis Experience Scales-Revised (TAPES-R) questionnaire¹⁸ which asked individuals to describe satisfaction with prosthesis on a discrete scale from 0 to 10, where 0 corresponded to “Not at all Satisfied” and 10 was noted to be “Very Satisfied”. A higher score indicated greater satisfaction. Pain interference was measured using the single question from the PROMIS Pain Interference item bank v1.1 question (PAININ9), which stated, “In the past 7 days how much did pain interfere with your day-to-day activities?”.¹⁹ Responses were scored from 1 to 5 according to whether individuals respond “Not at all”, “A little bit”, “Somewhat”, “Quite a bit”, or “Very much”. The pain interference question is reverse scored whereby a higher score indicated a worse situation where pain interferes very much with day-to-day activities.

Finally, the model was enriched by including 4 descriptive variables considered anecdotally in the clinical community as potential important predictors of well-being. These included how many months since the person had an amputation, how many hours the person wore their prosthesis each day, the person's age in years, gender, and amputation level. Gender (0=women, 1=men) and amputation level (2=wrist disarticulation and transradial, 3=elbow disarticulation and transhumeral) were coded as separate variables for analysis.

Analysis

To address the hypothesis, a multivariate linear regression model was run using a forward enter method. The model assessed prediction of patient well-being with the following predictors: APSRA, PROMIS-9 UE, PAININ9, TAPES-R, time since amputation, amputation level, average daily wear time, age, and gender. In addition to the multivariate linear regression model, each variable included in the multivariate model was individually run through univariate linear regression to assess individual effects. In cases of missing data, Little's MCAR test was run to confirm data were missing at random and then imputed using the mean.

Results

There were 250 individuals with upper limb amputation who had an outcome assessment on file for analysis (Table 1), all 250 patients’ outcomes were retained in the final analysis. Of the 250 individuals, 73.2% had wrist disarticulation or transradial amputations. Just under half the sample (46.0%) reported having an electronic arm. The most prevalent amputation etiology was injury, which constituted 66.4% of the sample for which amputation etiology was recorded, or 38.8% of the entire sample (104 individuals did not report amputation etiology).

Table 1.

Patients’ demographic and health characteristics at the time of survey administration (n=250)

	n (%)	Mean±SD
Age (y)	226 (90.4)	50.2±16.2
Height (cm)	133 (53.2)	175.3±9.7
Weight (kg)	124 (49.6)	85.4±18.3
Time since amputation (months)	250 (100)	48.0±104.8
Hours worn daily	221 (88.4)	9.3±5.3
Gender
Women	55 (22.0)	_
Men	175 (70.0)	_
Not reported	20 (8.0)	_
Amputation level
Wrist disarticulation/transradial	183 (73.2)	_
Elbow disarticulation/transhumeral	67 (26.8)	_
Cause of amputation
Vascular	2 (0.8)	_
Injury	97 (38.8)	_
Infection	7 (2.8)	_
Congenital	31 (12.4)	_
Cancer/tumor	9 (3.6)	_
Other/not reported	104 (41.6)	_
Device type
Body powered	135 (54.0)	_
Electronic arm	115 (46.0)	_
Bimanual Physical Function (PROMIS-9 UE)	245 (98.0)	32.8±11.8
Activity and Participation (APSRA)	243 (97.2)	49.0±8.9
Pain Interference	244 (97.6)	2.2±1.2
Well-Being (PEQ-WB)	246 (98.4)	7.5±2.1
Prosthetic Satisfaction (TAPES-R)	246 (98.4)	7.8±2.5

For predictor and dependent variables, Little's MCAR test found missing data points were to be considered missing completely at random and subsequently imputed with the mean (χ²(93, N=250)=112.86, P=.079).

Multivariate model: well-being

The overall fitted multiple linear regression model was statistically significant (R=0.676, F_(9,240)=22.428, P<.001; Table 2). This is considered a large effect size.²⁰ Within the model, four predictors were statistically significant (APSRA T-score, PROMIS-9 UE T-score, pain interference, and prosthesis satisfaction). The largest predictor of greater well-being within the multivariate model was APSRA T-score.

Table 2.

Multivariate linear regression (dependent: well-being)

			Adjusted Model
	B	SE	β	t	P
Constant	2.144	1.034		2.074	0.039
Activity and Participation: APSRA T-score	0.070	0.015	0.303	4.637	<0.0001
Physical Function: PROMIS-9 UE T-score	0.032	0.011	0.182	2.911	0.004
Prosthesis Satisfaction: TAPES-R	0.213	0.043	0.257	4.964	<0.0001
Pain interference	−0.324	0.097	−0.187	−3.335	0.001
Hours worn daily	−0.010	0.021	−0.025	−0.479	0.632
Time since amputation (months)	0.001	0.001	0.031	0.629	0.530
Age (years)	−0.005	0.006	−0.036	−0.743	0.458
Gender	−0.255	0.243	−0.051	−1.050	0.295
Amputation level	0.195	0.224	0.042	0.871	0.385

Univariate model: well-being

None of the individual univariate models (Table 3) were as effective at predicting well-being as the combined multivariate model. The univariate regression models using predictor variables of APSRA T-score (R=0.577, F_(1,247)=123.386, P<.001), PROMIS-9 UE T-score (R=0.489, F_(1,247)=77.746, P<.001), pain interference (R=0.456, F_(1,247)=64.838, P<.001), prosthesis satisfaction (R=0.422, F_(1,248)=53.786, P<.001), and average daily wear time (R=0.172, F_(1,246)=7.536, P=.006) were statistically significant. The univariate regression models using predictor variables of time since amputation (R=0.099, F_(1,248)=2.477, P=.117), age (R=0.023, F_(1,248)=0.129, P=.720), gender (R=0.019, F_(1,248)=0.087, P=.768), and amputation level (R=0.031, F_(1,248)=0.231, P=.631) failed to reach statistical significance.

Table 3.

Univariate linear regression models (dependent: well-being)

			Unadjusted Models
	B	SE	β	t	P
Constant	0.960	0.601		1.598	0.111
Activity and Participation: APSRA T-score	0.134	0.012	0.577	11.108	<0.0001
Constant	4.708	0.339		13.867	<0.0001
Physical Function: PROMIS-9 UE T-score	0.086	0.010	0.489	8.817	<0.0001
Constant	4.823	0.387		12.455	<0.001
Prosthesis Satisfaction: TAPES-R	0.349	0.048	0.422	7.334	<0.0001
Constant	9.238	0.242		38.193	<0.0001
Pain interference	−0.791	0.098	−0.456	−8.052	<0.0001
Constant	6.865	0.274		25.057	<0.0001
Hours worn daily	0.071	0.026	0.172	2.745	0.006
Constant	7.435	0.142		52.323	<0.0001
Time since amputation (months)	0.002	0.001	0.099	1.574	0.117
Constant	7.681	0.444		17.304	<0.0001
Age (y)	−0.003	0.008	−0.023	−0.359	0.720
Constant	7.600	0.274		27.737	<0.0001
Gender	−0.094	0.317	−0.019	−0.295	0.768
Constant	7.848	0.677		11.595	<0.001
Amputation level	−0.141	0.293	−0.031	−0.481	0.631

Discussion

The absence of one's upper limb can be highly effectful to a person's life. Minimizing the effect to ensure positive life experiences is reliant on a clear understanding of the relevant factors influencing the general well-being of persons living with limb loss/deficiency. Therefore, this study sought to predict well-being using factors that are clinically relevant to the care pathway of patients with ULA. Of the nine factors used in the multivariate regression model, prosthesis satisfaction and activity and participation (APSRA) were notably the strongest predictors of patients’ well-being, followed by pain interference and bimanual physical function.

Prior studies have shown a strong relation between social participation and both quality of life and health.^21,22 Although these studies, among individuals with physical disabilities, were not specific to individuals with limb loss, their findings provide general insights that increased social involvement will lead to better quality of life and general well-being. This is consistent with the current study's finding of a strong relation between both patient reported activity and participation and well-being among patients living with upper limb loss. Rehabilitation professionals, including prosthetists, can use targeted interventions to improve patients’ participation in social roles and activities. A properly designed upper limb prosthesis tailored to an individual's priorities with respect to their activity and participation may allow patients to better engage in bi-manual activities and social roles that they otherwise could not perform. For example, an individual who enjoys biking may find increased ability to participate in biking clubs with a terminal device designed specifically to aid in grasping handlebars, thereby improve bi-manual function for the activity and facilitating greater social participation. Future studies should examine the effect of prosthesis design type on social roles.

In the current study, the average APSRA T-score for the entire sample was 49.0. This is only 1 point below the normative sample of the U.S. population (T-score=50.0). This would suggest that patients with either a body-powered or electronic arm perceived their participation in social roles and activities similar to the general population. This is also greater than previously reported values for individuals with amputation after outpatient discharge (APSRA T-score: 45.3±6.1),²³ although that was on a smaller sample size (ie, <30) and at an early discharge period.

Satisfaction with a prosthesis appears to be very user-dependent, taking into consideration such elements as appearance, weight, and reliability. Available evidence from previous studies suggests that different upper limb prosthesis types (eg, body-powered, myoelectric, and aesthetic) appear to address different areas of satisfaction with no uniformly preferred device type.²⁴ Rather, prosthesis satisfaction appears most influenced by amputation level, with greater satisfaction associated with the receipt of appropriate training.²⁴ Optimizing prosthesis satisfaction is likely a product of matching device characteristics with user priorities. For example, prior studies have cited the device characteristics of the myoelectric prosthesis as having greater precision for grasping large or small diameter objects, reduction in phantom limb pain, enhanced cosmesis, and improved social roles.²⁵ While body-powered devices, on the other hand, have been cited to be more durable, require less prosthesis training, and more acceptable for heavy work.²⁵ Given the current findings of prosthesis satisfaction as a strong predictor of well-being, future efforts should target methods to continue improving individualized prosthesis design to enhance users’ satisfaction.

Bimanual physical function, evaluated with the PROMIS-9 UE instrument, was also a significant predictor of well-being among persons living with ULA. Notably, the items within the PROMIS-9 UE custom short form consisted of bi-manual upper limb activities likely requiring a prosthesis to complete, indicating that patients’ perceived capacity to complete these tasks was significantly related to well-being. It is unsurprising that physical function is a strong predictor of well-being among patients with ULA, as previous studies have shown that physical function is positively associated with quality of life among older individuals without amputation.^26,27

Additionally, pain interference was a significant predictor of well-being. This relation is important to highlight for persons living with upper limb amputation. Given that prior studies noted that 90% of patients with ULA report some issue with pain post-amputation surgery,¹⁰ targeting this complex domain of pain and pain interference may improve well-being, reduce prosthesis rejection, and possibly improve adjustment to a prosthesis.

Lastly, patient reported prosthesis wear times failed to correlate with well-being. This finding, combined with the overall study results, indicates that well-being is better predicted by steps to ensure high levels of prosthesis satisfaction and both the ability to perform bimanual activities and participate in social roles and activities than simply how long a person may choose to wear a prosthesis through the day.

Study limitations

There are a few limitations to consider for this study. First, this study is a cross-sectional retrospective analysis which can limit interpretation of causality. Additionally, there is possibility for sampling bias with outcomes administration as persons who rejected the use of a prosthesis are not likely to return to their prosthetist for care. So while the sample may have included individuals who may subsequently stop using a prosthesis, it is unlikely to represent those who have already ceased prosthesis use. Next, our analyses were limited to persons seen at a follow-up appointment. Some of these patients may have been seen during an appointment in which they were seeking adjustments and it is likely the outcomes did not reflect a best state situation. In addition, although amputation etiology and time from first appointment to follow-up were absent from the analysis, the time since amputation variable may have contributed some elements of timing to the model. This methodology also limits visibility into those that have never had a prosthesis. Future study should consider their effect. While our sample included a blend of patients with either acquired or congenital etiology, it is possible that factors may vary based on etiology.

Conclusions

Optimizing factors such as activity and participation, physical function, and prosthesis satisfaction can significantly aid in predicting optimal well-being of upper limb prosthesis users. The modification of these factors should be considered beneficial for improving patient outcomes among upper limb prosthesis users.