Association of race, ethnicity, and gender to disparities in functional recovery and social health following major lower limb amputation: a cross-sectional pilot study.

Abstract

Objective:

To assess if evidence of disparities exists in functional recovery and social health post-lower limb amputation.

Design:

Race-ethnicity, gender, and income-based group comparisons of functioning and social health in a convenience sample of lower limb prosthetic users.

Setting:

Prosthetic clinics in four states.

Participants:

A geographically diverse cohort of 56 English and Spanish speaking community-dwelling individuals with dysvascular lower limb amputation, between 18–80 years old.

Interventions:

None.

Main outcomes measures:

Primary outcomes included two physical performance measures, the Timed Up and Go test and 2-minute walk test, and thirdly, the Prosthetic Limb Users Survey of Mobility^™. The PROMIS^® Ability to Participate in Social Roles and Activities survey measured social health.

Results:

Of the study participants, 45% identified as Persons of Color, and 39% were women (mean (SD) age, 61.6 (9.8) years). People identifying as non-Hispanic White men exhibited better physical performance than Men of Color, White women, and Woman of Color by −7.86 (95% CI, −16.26 to 0.53, p=0.07), −10.34 (95% CI, −19.23 to –1.45, p=0.02), and −11.63 (95% CI, −21.61 to –1.66, p=0.02) seconds, respectively, on the TUG, and by 22.6 (95% CI, −2.31 to 47.50, p=0.09), 38.92 (95% CI, 12.53 to 65.30, p<0.01), 47.53 (95% CI, 17.93 to 77.13, p<0.01) meters, respectively, on the 2-minute walk test. Income level explained 14% and 11% of the variance in perceived mobility and social health measures, respectively.

Conclusions:

Study results suggest that sociodemographic factors of race-ethnicity, gender, and income level are associated with functioning and social health post-lower limb amputation. The clinical impact of this new knowledge lies in what it offers to healthcare practitioners who treat this patient population, in recognizing potential barriers to optimal recovery and quality of life. More work is required to assess lived experiences after amputation and provide better understanding of amputation-related health disparities.

Substantial evidence suggests that certain U.S. health disparity populations, including Persons of Color (i.e., African American/Black race, Hispanic ethnicity, Native American peoples) and individuals from disadvantaged backgrounds (e.g., low income households, rural dwelling),¹ are at disproportionately greater risk of undergoing major lower limb amputation (LLA).^2–5 Issues surrounding elevated rates of chronic conditions like diabetes mellitus and peripheral artery disease in these historically underserved populations⁶ are believed to be the primary cause of the disproportionate rates of dysvascular LLA. However, contemporary research is acknowledging the complex interactions of sociodemographic factors and occurrence of major LLA in these at-risk individuals. ³ Causes of disparities in dysvascular amputation surgical rates are multifactorial, with low health literacy,⁷ geography,⁸ access to healthcare, and low income,^{3, 9, 10} named as primary contributors. Additionally, evidence exists that women and socioeconomically disadvantaged persons are at greater risk of complications, like reamputation, after dysvascular LLA.¹⁰ Rates of prosthetic prescription are variable and may depend on amputation level, etiology, or geographic location,^11–13 but use of a prosthetic limb is typically viewed as an essential component of optimal functional recovery after LLA. Furthermore, despite purported equivalent access to prosthetic services within the VA healthcare system, Black and women Veterans are prescribed prosthetic legs at lower rates.^{11, 14} Though disparities in amputation rates and prosthetic prescription are well-established, there is currently no evidence on whether disparities in functional recovery, including use of a prosthetic leg, and quality of life exist post-amputation.

Lower limb amputation is one of the costliest^{15, 16} and most debilitating chronic conditions affecting society,¹⁷ with long-term disablement being a significant concern.¹⁸ In the U.S., it is estimated within the next 30 years there will be 3.6 million people living with LLA,¹⁹ with dysvascular amputation rates increasing in an ever-younger population due to increasing rates of diabetes and vascular diseases. ^20,21 Historically underserved racial-ethnic minority and low income communities exhibit higher incidence of these limb-threatening diseases^{22, 23} and subsequent amputation,⁵ creating a scenario of life with multiple chronic conditions. Yet, research involving purposive sampling for sociodemographic diversity post-LLA to examine differences in functional recovery is lacking, and existing data regarding amputation disparities is exclusively retrospective. Furthermore, women are historically underrepresented in dysvascular amputation related research, leaving gaps in the evidence with respect to any gender-based differences in recovery of this population. Therefore, the objective of this study was to utilize a patient-involved approach to investigate if sociodemographics (race, ethnicity, gender, income) of people with dysvascular LLA are associated with poorer functional recovery and social health post-amputation.

Methods

Study design and Participants

We conducted a cross-sectional observational study of a convenience sample of English and Spanish-speaking community-dwelling (not living in any type of assisted living environment) prosthetic limb users with dysvascular LLA from varying income levels. We chose to enroll only prosthetic limb users with dysvascular LLA as the aim of this pilot study was to investigate functional recovery with a prosthetic limb. The study was approved by the Institutional Review Board at Florida International University, and all participants provided written informed consent. Data were collected from November 2020 through April 2022 at multiple prosthetic clinic sites in Florida, Nevada, Arizona and Utah to facilitate a geographically, economically, racial-ethnic, and linguistically diverse cohort. Additionally, we employed purposive sampling of women and Persons of Color. Inclusion criteria were age 18–80 years with major (above ankle level) LLA of at least one leg due to vascular causes (e.g., diabetes, peripheral artery disease), ability to walk 30 feet or more without assistance of another person, and use of a prosthesis for at least three months. People were excluded if they presented with open wounds on either lower limb or did not use a prosthesis for daily ambulation. All study procedures were performed or supervised by research personnel who were licensed physical therapists. Participants received a stipend for their time and effort in completing the study.

Sociodemographic Measures

Our primary explanatory measures were three self-reported social identity metrics. These metrics were, 1) race-ethnicity (self-identifying as non-Hispanic White, or Person of Color (identifying as American Indian or Alaska Native, Black, Hawaiian Native or other Pacific Islander, or Hispanic, alone or in combination with another racial or ethnic category)), 2) gender (self-identifying as a male, female, or other), and 3) income level (family income less than $20,000/year; between $20–50,000/year; over $50,000/year). Race and ethnicity were combined into a collective variable of race-ethnicity to embody well-established racial and ethnic disparities in the prevalence of LLA and to reflect our interest in reporting differences among people identifying as non-Hispanic White and those from historically underrepresented populations.

Function and Social Health Outcome Measures

To establish this study as seminal health disparities research in people post-LLA and improve clinical translation, common rehabilitation measures were utilized to assess the primary outcomes of interest. To assess functional recovery, our dependent variables were three common measures of functional mobility with a prosthetic leg. Two are measures of physical capacity with a prosthesis that have been validated for use in the LLA population. The third is a valid and reliable patient-reported measure of perceived mobility with a prosthetic limb. The Timed-Up-and-Go (TUG)^{24, 25} is a well-known physical performance test used to assess basic mobility tasks of stand/sit transitions, walking and negotiation of obstacles over a three-meter distance. A lower time to perform the test equates to better basic daily mobility and decreased risk of falls.^{26, 27} The 2-minute walk test is a physical performance test that is indicative of walking capacity.^{28, 29} The greater the distance ambulated indicates better mobility and endurance. The final measure of functioning with a prosthesis was the Prosthetic Limb Users Survey of Mobility (PLUS-M^™).³⁰ This 12-item patient-report survey records the participants perceived ability for functioning with a prosthetic leg. The PLUS-M underwent formal Spanish translation for our study, following the Ermanco et al. guidelines.³¹ Our final primary measure, the Patient-Reported Outcome Measure Information System (PROMIS^®) Ability to Participate in Social Roles and Activities,³² was administered to assess social health (representing an aspect of quality of life). PROMIS measures are supported by extensive psychometrics and offer standardized survey administration and interpretation of outcomes across populations.³³

As secondary variables of interest, participants were asked to rate their perception to barriers in their natural and built environments and in access to insurance or rehabilitation services. To measure participants perceived environmental barriers and community accessibility, four discrete items were chosen from the Environmental Factors Item Bank.³⁴ Availability of insurance and access to rehabilitative services was assessed dichotomously with yes/no questions.

Sample size calculations

Power calculations were performed with PASS 2019 (PASS 2019 Power Analysis and Sample Size Software (2019). NCSS, LLC. Kaysville, Utah, USA). It is important to note that this is a pilot study, and while it is powered to detect large effects, the more significant point is that it provides valuable estimates that will be used in the future to adequately power larger clinical studies with the capability to detect smaller effects. A sample size of 56 achieves 80% power, α=0 .05, to detect group differences with large effects (Cohen’s d=0.8); and medium to large effects (f² =0.26) using multiple linear regression modeling with a maximum of 5 predictors. The number of predictors is limited given the sample, and we considered inclusion into the multivariate model based on univariate findings.

Statistical Analysis

Descriptive statistics were used to examine participant characteristics. Categorical variables, perceived environmental barriers, and access to insurance and rehabilitation services were compared using Fisher’s exact test stratified by race-ethnicity and gender. Primary outcome measures were examined stratified by gender and race-ethnicity and compared using two-sample t-test. Cohen’s d measure of effect size was used to provide more information regarding the magnitude and clinical relevance of observed differences (Cohen’s d≥0.2 is larger than small effect) between the cohorts.^{35, 36} Analysis of variance (ANOVA) was applied to compare primary outcomes based on income level. Two-way ANOVA was used to examine primary measure results based on identity metrics of gender, race-ethnicity, and potential interactions of gender and race-ethnicity. Least squares means and Tukey adjustments were applied to control for Type I error. Linear regression modeling was used to explore associations between the three social identity metrics and the primary outcome measures. The three income levels were examined for associations individually, with the lowest income level chosen as the reference. Additionally, age and amputation level frequently influence prosthetic mobility outcomes^{25, 37} and are relevant metrics in clinical practice. Hence, they were examined as potential covariates for regression modeling. Transfemoral amputation level was chosen as the referent level, as research has shown that individuals with transfemoral amputation exhibit poorer mobility than those with unilateral or bilateral transtibial amputation.³⁸ Independent variables were considered for the final multivariable backward stepwise regression model if they were a significant contributor to the dependent variable at α ≤ 0.1. All analyses were conducted using SAS statistical software version 9.4, and significance was considered at α ≤ 0.05 (2-tailed) in context with the rest of the statistical evidence such as strength and magnitude of differences and associations, effect sizes, variability, and levels of confidence.

Results

Eighty-eight people with LLA were screened, and 56 individuals met inclusion criteria and completed the study (Figure 1). The mean (SD) age of participants was 61.6 (9.8) years, 25 (45%) were Persons of Color, and 22 (39%) were women (Table 1). Mean (SD) time since amputation was 6.5 (6.4) years. All participants with bilateral LLA were at the transtibial level. No differences existed between non-Hispanic White (hereafter, referred to as White) persons and Persons of Color, or genders, with respect to amputation level, time since amputation, number of comorbidities per the Functional Comorbidities Index,³⁹ income level, perceived environmental barriers or access to rehabilitation. A statistical difference in age existed, with White persons averaging 5.3 years older (95% CI, 0.22 to 10.43, p=0.04). Though not statistically different, it was noted that 44% of Persons of Color reported family income less than $20,000 per year compared to 23% of White persons, and participants categorized in the lowest income level were the youngest group.

Figure 1.

Flow diagram of study procedures.

Table 1.

Demographic characteristics of study participants (n=56)

Sociodemographic variable	No. (%)
Gender
Men	34 (61%)
White	18
Men of color	16
Women	22 (39%)
White	13
Women of color	9
Other	0
Race-ethnicity
Non-Hispanic White	31 (55%)
Persons of Color	25 (45%)
African American/Black	14
Asian	3
Native American	1
White Hispanic	7
Yearly family income
Less than $20,000	18 (33%)
$20,000-$50,000	17 (30%)
Greater than $50,000	17 (30%)
Refused to report	4 (7%)
Amputation level
Transtibial (below knee)	37 (66%)
Transfemoral (above knee)	9 (16%)
Bilateral	10 (18%)

Group comparisons

In all four primary outcome measures, participants identifying as White exhibited higher scores than those identifying as Persons of Color (Table 2), with medium effect of race-ethnicity associated with the two performance-based measures. However, effect sizes for the patient-reported measures were small (Cohen’s d ≤ 0.2). Furthermore, men exhibited significantly better scores on our three functional mobility measures than women, with associated medium to large effects (Cohen’s d=0.6–1.1) for gender.

Table 2.

Race-ethnicity, and gender comparisons on primary outcomes

Variable	Non-Hispanic Whites (n=31) Mean (SD)	Persons of Color (n=25) Mean (SD)	P value	Cohen’s d
Age (years)	63.9 (9.2)	58.6 (9.8)	0.04	0.6
Timed Up and Go (sec)	17.9 (8.5)	22.8 (11.3)	0.04	0.5
2-minute walk test distance (m)	79.5 (28.2)	64.2 (35.3)	0.07	0.5
PLUS-M	47.9 (10.8)	47.2 (11.7)	0.82	0.1
PROMIS Ability to Participate	46.7 (11.5)	44.7 (9.0)	0.48	0.2
	Men (n=34) Mean (SD)	Women (n=22) Mean (SD)
Age (years)	62.1 (9.2)	60.7 (10.8)	0.6	0.1
Timed Up and Go (sec)	17.3 (8.5)	24.4 (11.1)	0.008	0.7
2-minute walk test distance (m)	85.2 (29.7)	53.4 (26.2)	<.001	1.1
PLUS-M	50.1 (10.5)	43.6 (11.2)	0.03	0.6
PROMIS Ability to Participate	45.7 (9.6)	46.0 (11.8)	0.9	0.03

Abbreviations: SD=standard deviation; PLUS-M, Prosthetic Limb Users Survey of Mobility^™; PROMIS^®, Patient-reported Outcome Measure Information System.

In examining the subgroup cohorts by race-ethnicity and gender, no differences existed in mean age (White men = 63.6 (8.4) years; White women = 64.4 (10.6); Men of Color = 60.5 (10); Women of Color = 55.3 (9.0), p=0.12). Compared to other participants, those who self-identified as White men exhibited better scores on all four primary outcome measures (Table 3). Notably, on physical performance measures, White men outperformed Men of Color, White women and Women of Color by −7.86 (95% CI, −16.26 to 0.53, p=0.07), −10.34 (95% CI, −19.23 to –1.45, p=0.02), and −11.63 (95% CI, −21.61 to –1.66, p=0.02) seconds, respectively, on the TUG, and by 22.6 (95% CI, −2.31 to 47.50, p=0.09), 38.92 (95% CI, 12.53 to 65.30, p<0.01), 47.53 (95% CI, 17.93 to 77.13, p<0.01) meters, respectively, on the 2-minute walk test. No significant interaction of gender and race-ethnicity was observed on any of the measures.

Table 3.

Sub-group comparisons by race-ethnicity, gender, and interaction of race-ethnicity and gender

Timed Up and Go test (sec); Interaction gender x race-ethnicity (F = 1.67, p = 0.20)
Group	Score Mean (SD)	Between group difference, mean (95% CI)	P value
White men	13.6 (3.0)	NA
MOC	21.4 (10.6)	−7.86 (−16.26 to 0.53)	0.07
WW	23.9 (10.2)	−10.34 (−19.23 to –1.45)	0.02
WOC	25.2 (12.9)	−11.63 (−21.61 to –1.66)	0.02
Men of color	21.4 (10.6)	NA
WW	23.9 (10.2)	−2.48 (−11.60 to 6.65)	0.89
WOC	25.2 (12.9)	−3.77 (−13.95 to 6.41)	0.76
White women	23.9 (10.2)	NA
WOC	25.2 (12.9)	−1.29 (−11.89 to 9.30)	0.99
2-minute walk test (meters); Interaction gender x race-ethnicity (F = 0.86, p = 0.36)
Group	Score Mean (SD)	Between group difference, mean (95% CI)	P value
White men	95.8 (18.2)	NA
MOC	73.2 (35.8)	22.6 (−2.31 to 47.50)	0.09
WW	56.9 (23.7)	38.92 (12.53 to 65.30)	0.002
WOC	48.3 (30.0)	47.53 (17.93 to 77.13)	0.002
Men of color	73.2 (35.8)	NA
WW	56.9 (23.7)	16.32 (−10.75 to 43.39)	0.39
WOC	48.3 (30.0)	24.94 (−5.26 to 55.14)	0.89
White women	56.9 (23.7)	NA
WOC	48.3 (30.0)	8.62 (−22.82 to 40.05)	0.90
PLUS-M™; Interaction gender x race-ethnicity (F = 0.12, p = 0.73)
Group	Score Mean (SD)	Between group difference, mean (95% CI)	P value
White men	51.0 (10.7)	NA
MOC	49.1 (10.5)	1.90 (−8.07 to 11.88)	0.96
WW	43.5 (9.8)	7.47 (−3.09 to 18.05)	0.25
WOC	43.7 (13.6)	7.26 (−4.59 to19.11)	0.37
Men of color	49.1 (10.5)	NA
WW	43.5 (9.8)	5.58 (−5.26 to 16.42)	0.53
WOC	43.7 (13.6)	5.36 (−6.74 to 17.46)	0.64
White women	43.5 (9.8)	NA
WOC	43.7 (13.6)	−0.22 (−12.81 to 12.37)	1.00
PROMIS® Ability to Participate; Interaction gender x race-ethnicity (F = 0.41, p = 0.52)
Group	Score Mean (SD)	Between group difference, mean (95% CI)	P value
White men	47.3 (10.7)	NA
MOC	43.8 (8.1)	3.43 (−6.24 to 13.10)	0.78
WW	45.8 (12.9)	1.44 (−8.80 to 11.69)	0.98
WOC	46.2 (19.8)	1.12 (−10.39 to 12.60)	0.99
Men of color	43.8 (8.1)	NA
WW	45.8 (12.9)	−2.00 (−12.50 to 8.52)	0.96
WOC	46.2 (19.8)	−2.32 (−14.05 to 9.40)	0.95
White women	45.8 (12.9)	NA
WOC	46.2 (19.8)	−0.33 (−12.54 to 11.87)	1.00

Abbreviations: NA, not applicable; SD, standard deviation; CI, confidence interval; MOC, Men of Color; WW, White Women; WOC, Women of Color; PLUS-M, Prosthetic Limb Users Survey of Mobility^™; PROMIS^®, Patient-reported Outcome Measure Information System.

We recategorized four participants who refused to report their income to the median income level reported by all study participants ($20–50K/year) and then secondarily justified the categorization using the Social Vulnerability Index⁴⁰ of their home address. No significant differences were exhibited between the three income groups on the two physical performance measures. However, participants in the highest income level scored significantly better than the lowest income level on their perceived mobility by an additional 11.42 (95% CI, 2.99 to 19.86, p<0.01) points on the PLUS-M, and in social health by an additional 8.68 (95% CI, 0.54 to 16.83, p=0.04) on the PROMIS Ability to Participate.

Regression analyses

Multivariable backward regression was used to determine associations of the three social identity metrics (race-ethnicity, gender, income) with the four primary outcome measures (Table 4). In the TUG and 2-minutewalk test models, both race-ethnicity and gender were the significant contributors, with race-ethnicity explaining 6% and 7%, and gender explaining 10% and 23% of the variance, respectively. We deviated, slightly, from our power analysis by using six explanatory variables in our TUG model, giving it one additional degree of freedom. Incidentally, when income levels one and two were combined into a single variable, our results were unchanged.

Table 4.

Associations of explanatory variables to primary outcome measures

Independent variable	Parameter estimate	SE	Squared semi-partial correlation	F	P value
Timed Up and Go model: F = 6.28, R2 = 0.19, p = 0.004
TT amputation	−3.74	2.67	0.03	1.96	0.17
Bilateral TT amputation	−2.57	4.65	0.02	0.31	0.58
Income level 2	−2.78	3.17	<0.01	0.77	0.39
Income level 3	−1.84	2.75	0.02	0.45	0.51
Race-ethnicity	5.33	2.49	0.06	4.57	0.04
Gender	7.50	2.54	0.10	8.72	0.005
2-minute walk test model: F = 11.83, R2 = 0.31, p < 0.001
Age	−0.005	0.03	<0.01	0.03	0.87
Income level 2	2.32	7.84	<0.01	0.09	0.77
Income level 3	4.11	9.41	<0.01	0.19	0.66
Race-ethnicity	−17.20	7.35	0.07	5.49	0.02
Gender	−32.87	7.48	0.23	19.32	<0.001
PLUS-M model: F = 7.28, R2 = 0.30, p < 0.001
TT Amputation level	6.30	2.71	0.08	5.38	0.02
Bilateral Amputation level	1.48	4.8	<0.01	0.09	0.76
Income level 2	3.96	3.22	0.01	1.51	0.22
Income level 3	8.59	2.80	0.14	9.42	0.003
Gender	−6.92	2.63	0.08	6.92	0.01
PROMIS Ability to participate model: F = 5.31, R2 = 0.09, p = 0.03
Income level 2	4.91	3.20	<0.01	2.35	0.13
Income level 3	8.68	3.38	0.11	6.61	0.01
Race-ethnicity	−0.01	2.83	0.01	0.0	0.97
Gender	1.40	2.86	<0.01	0.24	0.62

TF=transfemoral, TT=transtibial. PLUS-M, Prosthetic Limb Users Survey of Mobility^™; PROMIS^®, Patient-reported Outcome Measure Information System. Income level 1= less than $20,000; 2=$20,000-$50,000; 3=more than $50,000. Parameter estimates for gender, and race-ethnicity are associated with being a woman, and a Person of Color, respectively. The lowest income level, and TF amputation level were used as reference levels for ordinal variables.

In the PLUS-M model, gender, transtibial amputation level, and the highest income level were significant contributors to perceived functional mobility, with gender and amputation level explaining 8%, and income explaining 14% of the variance. Race-ethnicity was not associated with the PLUS-M in linear regression (p=0.82), therefore was not included in the final PLUS-M model for the sake of parsimony. Only the highest income level was significantly associated with our measure of social health, explaining 11% of the variance in the PROMIS survey.

Discussion

Our social identity metrics (race-ethnicity, gender, income) were shown to be significantly associated with recovery several years post-LLA. Because this study is the first, to our knowledge, to purposively investigate post-LLA functioning and social health in a racially and economically diverse cohort, there is little context on which to contrast many of our findings. Looking to other patient populations, racial and sex disparities exist in recovery following stroke, with identifying as White or male associated with better functional recovery .⁴¹ The authors report that income was significantly lower in Black versus White participants who had suffered a stroke. Additionally, being female was associated with greater motor function deficits following stroke, with depression cited as a potential influencing variable. Though falling short of statistical significance in our study, 44% of Persons of Color reported being in the lowest income level compared to 23% of White participants, as did 45% of women versus 24% of men in our study. Furthermore, 41% of women versus 18% of men reported depression as a comorbidity, which has been shown to result in lower prosthetic prescription rates in women with dysvascular LLA.⁴² In a larger study of functional recovery, these variables may reach significance and provide further insight into disparities after LLA.

Our study utilized a novel patient-involved approach to identify racial-ethnic disparities in functional recovery and social health after dysvascular LLA. Unique to our post-amputation study, is that race and ethnicity were self-reported by participants, avoiding issues of bias on part of the researchers, or of misclassification which is a limitation of retrospective database research.⁴³ Our results indicate that well-established racial disparities in LLA surgical rates appear to persist into recovery post-amputation, as race, ethnicity, and gender were associated with meaningfully poorer physical performance with a prosthetic leg. Furthermore, these disparities in recovery extend to women with LLA and particularly Women of Color. These results advance the evidence that race-ethnicity not only impacts amputation rates but remains a factor affecting recovery in this vulnerable population.

To our knowledge, no previous disparity-related studies in amputation have employed physical performance measures to assess functional recovery. This is significant, in that only moderate correlations exist between physical performance and patient-reported measures of mobility in the lower limb amputation population.⁴⁴ Capturing actual physical abilities provides a more complete assessment of someone with LLA and fills an important gap in amputation-related disparity research. Prior research has identified gender-based differences in physical functioning after LLA, specifically that women walk 20 meters less distance than men during the 2-minute walk test.²⁹ However, our study results indicate that subgroup differences between White men and both groups of women with dysvascular LLA are exceeding the minimal detectable change (MDC) of 34.3 meters⁴⁵ and the minimal clinically important difference (MCID) of 37.2 meters on the 2-minute walk test.²⁸ And White men are outperforming Men of Color and all women on the TUG, exceeding the previously established MDC of 3.6 seconds for the test.⁴⁵ The breaching of established psychometrics for these measures indicates that there are clinically meaningful differences between White men and the other cohorts in our study that are not due to chance. Furthermore, our regression results suggest that a Woman of Color would walk 50 meters less on the 2-minute walk test and 13 seconds slower on the TUG than a White man, despite women of color being the youngest cohort in our study. With little research inclusive of women with dysvascular LLA and no previous post-amputation research purposively inclusive of Women of Color, hypothesizing why such dramatic differences are being exhibited provides opportunities to further explore potential barriers to recovery after LLA.

Age and level of amputation are frequently associated with functioning after LLA,²⁵ however neither were associated with the two physical performance measures in our study. Age may not have appeared significant due to the homogeneity of our cohort having dysvascular amputation, which is more prevalent in an older, less healthy population. Transtibial amputation was associated with better perceived mobility on the PLUS-M, but not with physical performance on the TUG or 2-minute walk test. This may be attributed to only moderate correlations between physical performance and patient-reported prosthetic mobility measures.⁴⁴ The highest income level was associated with better PLUS-M score and identifying as a woman was related to poorer perceived mobility, with both social identity metrics individually exceeding the established minimal detectable change (MDC) of 4.5 points for the PLUS-M.⁴⁶ This is note-worthy in that only one other study has reported significant gender differences in perceived mobility.⁴⁷ However, unlike our results, values did not exceed the MDC. Furthermore, yearly income greater than $50,000/year was associated with significantly better social health in our study, with differences exceeding the minimal important change score for the PROMIS measure.³² Previous research reports that experiencing financial difficulty after amputation is associated with reduced activity participation, including employment.⁴⁸ Because lack of full-time employment is associated with poorer physical functioning after LLA,⁴⁹ a complex interaction of factors could contribute to lower income and increased risk for suboptimal recovery. The evident impact of sociodemographic factors on recovery after LLA indicates that more research into patient perceptions, attitudes and lived experiences after LLA is needed to provide further insights.

Limitations and future work

Our main study limitation is our sample size. To that point, in efforts to provide more robust statistical results, we categorized all non-White and Hispanic identifying participants as persons of color, which could mask important heterogeneity among these individuals. Despite this, we possibly were underpowered to detect interactions of gender and race-ethnicity which may become more apparent with larger sample sizes. We acknowledge that some group comparisons resulted in p values slightly exceeding p ≥ 0.05, but believe that moderate, or greater, effect sizes ³⁶ and the exceeding of previously established MDCs/MCIDs are meaningful indicators of the potential effects race-ethnicity, and gender are having on outcomes. As this was a pilot study in an unchartered area of research, generalizability of our results is premature. However, this study provides a foundation for reflection on previously performed disparity research and offers a basis for larger, longitudinal studies to further investigate health disparities surrounding amputation. Furthermore, our results offer insight not only into optimal patient care post-amputation, but support evidence previously established for underserved at-risk populations. Another limitation may be our exclusion of non-prosthetic limb users with LLA, as including this subgroup will likely offer further insight into disparities. However, our aim to assess functional mobility with a prosthetic limb precluded us from enrolling this potentially more vulnerable population and may affect the generalizability of our results. An additional limitation is that our regression models do not explain much of the variance in our primary measures. For the sake of parsimony, variables known to contribute to prosthetic mobility, like self-efficacy,⁴⁹ were not considered for the models as our focus was on the three social identity metrics. Future studies with larger sample sizes may provide opportunities to more closely examine which variables are predictive of optimal recovery after LLA. Lastly, we acknowledge that significant progress has been made in clarification of gender descriptors, making our surveyed gender choices of “male,” and “female” inappropriate. Future research will ensure correct terms for gender descriptors.

This study incorporated a novel patient-involved approach to examining amputation-related health disparities. The current evidence base relies heavily on secondary analyses of data to explore disparities in dysvascular amputation rates, with no studies going beyond the final surgical outcome. Our study goes a step further, investigating if disparities persist into recovery post-amputation, potentially influencing the well-being of millions of people. Despite being a convenience sample, our purposively sampled cohort of individuals with dysvascular LLA exceeds the racial-ethnic diversity of the larger populations retrospectively studied in previous amputation-related disparities research^{2, 3} and is a fair representation for pilot work in this area. Our results provide important preliminary evidence for further post-amputation disparity research.

Conclusions

Our results offer novel insight into sociodemographic factors that are associated with function and aspects of quality of life of individuals with dysvascular LLA, indicating that barriers may exist to optimal recovery in certain patient populations. These results can contribute to a rethinking in rehabilitation medicine, by supporting targeted clinical policies that directly address sociodemographic factors. Clinicians should recognize that multi-marginalized populations are at risk of poorer outcomes after LLA, potentially because of structural racism, unconscious bias, and sexism historically encountered during healthcare experiences.^{50, 51} By engaging these at-risk patients regarding perceived barriers and facilitators to their recovery, clinicians can better gauge expectations and mitigate obstacles to improved function after LLA.

Abbreviations:

LLA

lower limb amputation

PLUS-M

Prosthetic limb users survey of mobility

PROMIS

Patient-Reported Outcome Measure Information System

TUG

timed up and go test