Socioeconomic Factors Predict Successful Supervised Exercise Therapy Completion

Abstract

Objective:

Supervised exercise therapy (SET) for patients with intermittent claudication (IC) can lower the risk of progression to chronic limb threatening ischemia and amputation, while preserving and restoring functional status. Despite supporting evidence, it remains underutilized and among those who initiate programs, attrition rates are extremely high. We hypothesize that socioeconomic factors may represent significant barriers to SET completion.

Methods:

Patients with IC referred to SET at a multi-hospital, single-institution healthcare system (2018 – 2022) from a prospectively maintained database were retrospectively analyzed. Our primary endpoint was SET program completion and graduation, defined as completion of 36 sessions. Our secondary endpoints were vascular intervention within one year of referral and change in ankle-brachial index (ABI). Baseline demographics were assessed using standard statistical methods. Predictors of SET graduation were analyzed using multivariable logistic regression generating adjusted odds ratios (aOR) with 95% confidence intervals (95%CI). Change in ABI was analyzed using t-test between subgroups. Reasons for attrition were tabulated. Patient Health Questionnaire-9 (PHQ-9), metabolic equivalent level, Vascular QOL, Duke Activity Status (DASI), and ABI were analyzed using paired t-tests across the entire cohort.

Results:

Fifty-two patients met inclusion criteria: mean age 67.85 ± 10.69 years, 19 (36.54%) females, mean baseline ABI 0.77 ± 0.16. The co-pays for 100% of patients were fully covered by primary and secondary insurance plans. Twenty-one (40.38%) completed SET. On multivariable analysis, residence in a zip code with median household income < $47,000 (aOR = 0.10 [95% CI, 0.01 – 0.76]; P = .03) and higher BMI (aOR = 0.81 [95% CI, 0.67 – 0.99; P = .04) were significant barriers to SET graduation. There were no differences in ABI change or vascular intervention within one year between graduates and non-graduates. Non-graduates reported transportation challenges (25.00%), lack of motivation (20.83%), illness/functional limitation (20.83%) as primary reasons for SET attrition. Metabolic Equivalent Level (P = <.01) and DASI scores (P = .04) were significantly greater after participating in a SET program.

Conclusions:

Although SET participation improves lower extremity and functionality outcomes, only 40% of referred patients completed therapy in our cohort. Our findings suggest that both socioeconomic and functional factors influence the odds of completing SET programs, indicating a need for holistic pre-referral assessment to facilitate enhanced program accessibility for these populations.

Table of Contents Summary

Only 40.38% of 52 patients completed supervised exercise therapy (SET) for intermittent claudication. Both socioeconomic and functional factors influence the odds of completing SET programs, indicating a need for holistic pre-referral assessment to facilitate enhanced program accessibility for these populations.

INTRODUCTION

Intermittent claudication (IC) is the most common clinical manifestation of peripheral artery disease (PAD)¹. Not only does IC reduce quality of life by impeding patients’ ability to work or care for themselves², but it also has been shown that 20% of patients will progress to chronic limb threatening ischemia (CLTI)³. Supervised exercise therapy (SET) is supported by level-one data and recommended by numerous clinical practice guidelines as first-line therapy^4–6 to improve symptoms in patients with IC and lower the risk of progression to CLTI and amputation^7–10. SET for PAD has been covered by the Center for Medicare and Medicaid Services (CMS) since 2017¹¹.

There are two key steps to successful utilization of SET: 1) Referral to a SET program by a vascular provider and 2) completion of the SET program by patients. Despite supporting evidence for SET, referral rates are as low as 15% and attrition rates are as high as 50–90%^{12, 13}. Furthermore, there is some evidence to suggest that early peripheral vascular intervention (PVI) within 6 months of diagnosis may initiate a domino effect of additional endovascular and open interventions, and potentially increase rates of progression from IC to CLTI. PVI is frequently employed for symptom management even before goal directed medical therapy, including a trial of SET, is achieved^{14, 15}. Cited reasons in the literature for SET attrition include cost of copay, time-consuming nature of treatment, and inconvenience of travel;¹⁶ however, these data are limited to subjective, patient-level survey data.

Therefore, the objective of the current study was to identify patient demographic and comorbid conditions that may contribute to SET attrition, in addition to outlining SET outcomes. We hypothesized that a large portion of our cohort would not complete SET and that socioeconomic factors may represent significant barriers to SET completion.

METHODS

All patients with IC who attended one of two SET programs at a multiple-institution healthcare system (January 2018 to January 2022) were analyzed (n = 52). Our dataset was prospectively maintained by the SET center and included all patients who were at the very least scheduled for one SET session. We retrospectively added datapoints as necessary and analyzed this data. We excluded patients still enrolled in SET at the time of data analysis. Collected data captured patient demographics including self-identified gender and race, primary residence and SET location zip codes, pre- and post-SET ankle brachial index (ABI), and SET graduation status. Additionally, patients were queried for reasons of SET program attrition by phone or in-person and their responses were matched to one of six categories: financial, motivation, transportation, illness/functional limitation, time, and other. SET graduation was defined as any patient that had completed at least 36 (30–60 minute) therapy sessions based on the number of sessions covered by CMS and previous randomized clinical trials¹¹. Pre-SET ABI was defined as the most recent ABI prior to SET initiation, and the lower ABI was used in patients with bilateral readings. Drive time to SET location was calculated using ArcGIS Online (Esri, Redlands, CA) between residential and SET facility addresses¹⁸. Median household income was assessed by matching the patients’ zip code to United States census data on zip code and income¹⁹. A household was categorized as low-income if the median household income < $47,000. This value was determined using the median household income of individuals age 65 years and older (approximate age of our cohort) according to U.S. census data¹⁹. The University of Pittsburgh’s human research protection office approved the present study with an exemption of informed consent (STUDY23020032).

Our primary outcome was SET completion. Secondary endpoints included reasons for SET incompletion, vascular intervention within one year of SET initiation, and change in ABI. Further endpoints included a comparison of pre- versus post-SET functional measurements using Patient Health Questionnaire-9 (PHQ-9), metabolic equivalent level, Vascular QOL, and Duke Activity Status (DASI). Due to the lack of data collected in patients who did not complete a substantial number of SET sessions, we were unable to compare changes in these functional metrics across graduates and non-graduates. Patients with vascular intervention within one year of referral were excluded in the change in ABI and functional measurement analysis.

Normality testing was conducted using the Shapiro-Wilk test. Continuous variables were reported as mean ± standard deviation (SD) or median (IQR) and were compared between cohorts using t-tests for normal variables and Wilcoxon rank sum tests for skewed variables. Categorical variables were presented as frequencies (%) and analyzed with Pearson’s Chi-squared and Fisher’s exact tests. Multivariable logistic regression was used to identify factors that predicted SET completion. Clinically relevant variables selected a priori (sex, race [white vs. non-white], pre-SET ABI ≥ 0.70, drive time to SET (continuous, in minutes) and residency in low-income zip code) and additional variables with p ≤ 0.1 on univariable analysis (body mass index [BMI], continuous) were included in the model. The model goodness-of-fit and multicollinearity were assessed using the Hosmer Lemeshow test and variance inflation factors. Postestimation adjusted SET graduation rates were calculated from logistic regression for significant predictors. Reasons for attrition were tabulated. Change in ABI was analyzed using paired t-test within and t-test between groups. Pre- and post-SET functional outcomes were analyzed using paired t-tests. A p < 0.05 was considered significant and all analyses were performed in Stata version 17 (StataCorp LP, College Station, TX).

RESULTS

Demographics and Clinical Characteristics

Among 52 patients referred to SET, 21 (40.38%) completed the program. There were no significant differences in baseline characteristics between those who completed SET and those who did not. Overall, patients were elderly (mean age 67.85 ± 10.69 years), predominantly male (63.46%), and White (65.38%). Patients were overweight (BMI 27.69 [IQR 6.88]), with higher median BMI trending toward significance in non-graduates vs. graduates (29.37 [IQR 6.56] vs. 24.75 [IQR 5.25], P = .06). Pre-SET ABI was on average 0.77 ± 0.16, which again did not differ significantly between groups. Patients lived a median 15.00 (IQR 9.00) minute drive away from the SET programs and this did not differ significantly between groups. Nineteen percent of patients lived in zip codes with median household income < $47,000, with a trend toward more low-income non-graduate patients (25.81% vs. 9.52%, P = .14). The primary or secondary insurance plans allowed all patients to receive SET without any visit co-pay (Table I). Of the patients that did not complete SET, 13 completed 0 – 5 sessions, five completed 5 – 10 sessions, seven completed 10 – 15 sessions, and four completed 20 – 35 sessions (Figure 1).

Table I.

Baseline Characteristics of Claudication Patients Undergoing Supervised Exercise Therapy

Variable	Overall (n = 52)	SET Graduates (n = 21)	SET Non-graduates (n = 31)	P-value
Age**	67.85 (10.69)	69.43 (10.50)	66.77 (10.85)	0.39
Female⧺	19 (36.54)	6 (28.57)	13 (41.94)	0.33
Non-White⧺	18 (34.62)	7 (33.33)	11 (35.48)	0.87
Married ⧺	25 (48.08)	11 (52.38)	14 (45.16)	0.61
Diabetes ⧺	20 (38.46)	6 (28.57)	14 (45.16)	0.23
Vascular Referring Provider ⧺	44 (84.62)	19 (90.48)	25 (80.65)	0.34
Current Tobacco User ⧺	19 (36.54)	5 (23.81)	14 (45.16)	0.12
Body Mass Index***	27.69 (6.88)	24.75 (5.25)	29.37 (6.56)	0.06
Pre-SET Ankle Brachial Index**	0.77 (0.16)	0.80 (0.12)	0.74 (0.19)	0.22
Required Copay ⧺	0 (0)	0 (0)	0 (0)	-
Drive Time to SET location (minutes)***	15.00 (9.00)	15.00 (10.00)	15.00 (9.00)	0.43
Residence in zip code where median household income < $47,000 ⧺	10 (19.23)	2 (9.52)	8 (25.81)	0.14

^**mean (SD)

^***median (IQR)

^⧺n (%)

Figure 1:

Histogram of number of SET sessions completed by non-graduates

Predictors of Completion

On multivariable logistic regression, residence in a zip code with median household income < $47,000 (aOR = 0.10 [95% CI, 0.01 – 0.76]; P = .03) and higher BMI (aOR = 0.81 [95% CI, 0.67 – 0.99; P = .04) were significant barriers to SET graduation. Increased drive time from SET did not reach statistical significance (aOR = 0.93 [95% CI, 0.86 – 1.01]; P = .10; Figure 2). Postestimation adjusted SET graduation rates were 48.65% (95% CI, 41.54% - 55.76%) for those with residence in a zip code with median household income ≥ $47,000, and only 20.00% (95% CI, 10.96 – 29.04%) in a low-income zip code. Postestimation adjusted SET graduation rates were 48.77% (95% CI, 36.46% - 61.07%) in non-obese (BMI < 25) patients vs. 40.40% (95% CI, 33.70% - 47.10%). Finally, postestimation adjusted SET graduation rates in patients with both residence in a zip code with income ≥ $47,000 and BMI < 25 were 59.61% (95% CI, 45.78% - 73.44%) vs. 17.36% (95% CI, 4.01% - 30.72%) in patients with a residence in a low-income zip code and BMI ≥ 25. Female gender (aOR = 0.54 [95%CI, 0.11 – 2.62]; P = .45), non-White (aOR = 0.82 [95%CI, 0.17 – 3.92]; P = .80), or pre-SET ABI ≥ 0.70 (aOR = 2.07 [95%CI, 0.47 – 9.16]; P = .34) were not independently predictive of SET graduation. A non-significant test result (P = .45) of the Hosmer Lemeshow test and low variance inflation factors (< 1.3) indicated a good model fit and that multicollinearity was not a concern in the regression analysis.

Figure 2:

Forest plots of multivariable analysis of SET completion (body mass index and drive time to SET location are continuous variables)

Post-SET Outcomes & Reasons for Incompletion

Of the 31 patients that did not complete SET, 24 provided their reason for incompletion. Non-graduates reported transportation challenges (25.00%), lack of motivation (20.83%), illness/functional limitation (20.83%), other (16.67%), time (12.50%), and financial (4.17%) as reasons for attrition (Figure 3). ABI at one-year increased non-significantly by 0.04 ± 0.10 (P = .07) for SET graduates and 0.01 ± 0.13 (P = .69) for non-graduates that did not undergo revascularization. Change in ABI (0.04 vs. 0.01, P = .43) and intervention within one year (19.1% vs. 30.0%, P = .38) between SET graduates and non-graduates did not reach statistical significance (Table II). Metabolic Equivalent Level (P = <0.01) and DASI scores (P = .04) were significantly greater than baseline after participating in a SET program. However, PHQ-9 scores, and Vascular QOL scores did not differ after participating in a SET program (Table III).

Figure 3:

Reasons for incompletion of supervised exercise therapy (n = 24)

Table II.

Post-Supervised Exercise Therapy Outcomes in SET Graduates versus SET Non-Graduates

Variable	Overall (n = 52)	SET Graduates (n = 21)	SET Non-graduates (n = 31)	P-value
Intervention within One year⧺	13 (25.49)	4 (19.05)	9 (30.00)	0.38
Change in Ankle Brachial Index**	0.02 (0.12)	0.04 (0.10)	0.01 (0.13)	0.43

^**mean (SD)

^⧺n (%)

Table III.

Supervised Exercise Therapy Functional Outcomes Pre-SET versus Post-SET

Variable	Pre-SET	Post-SET	P-value
Patient Health Questionnaire-9**	2.68 (0.66)	2 (0.63)	0.32
Metabolic Equivalent Level**	2.88 (0.18)	4.74 (0.40)	<0.01
Vascular QOL**	14 (1.26)	15 (1.23)	0.26
Duke Activity Status Index**	18.88 (3.60)	25.53 (4.69)	0.04
Ankle Brachial Index	0.79 (0.03)	0.81 (0.03)	0.14

^**mean (SD)

DICSUSSION

In our present study of 52 patients referred to SET, we identified demographic predictors of SET completion to graduation, with a focus on health disparities, which had not been previously examined. There were no differences upon univariable analysis between SET graduates and non-graduates, however, we did find that both residence in a zip code with a median income < $47,000 and a higher BMI were significant barriers to SET completion. These data highlight that socioeconomic factors pose as barriers to the implementation of SET as a first-line treatment for IC in accordance with multiple societal guidelines. There were no differences in post-SET ABI and rate of vascular intervention within one year. However, we did find that DASI and metabolic equivalent level improved across the entire cohort highlighting the efficacy of SET in improving patient functionality.

Our results regarding barriers to SET completion mirror those noted within the existing literature. Cetlin et al.¹⁶ administered a questionnaire to 516 patients with symptomatic PAD and found that 43.7% would be unable to pay the $11 copay for a hypothetical SET session, totaling almost $400 for the minimum requirement of 36 sessions¹¹. Similarly, Altin et al. surveyed 77 patients referred to SET and found that 51% declined participation, and of those who declined, 23% cited cost of copayment and 10% cited interference with work schedule as reasons. Although cost was not explicitly a common reason for attrition in our study, this is likely due to the fact that none of the patients in our cohort had a copay due to primary and secondary insurance plans. This finding is interesting, and perhaps indicates that there may be a bias against referring patients who either have no insurance or have to pay a copay, and/or that such patients do not even schedule a session upon the realization of the out-of-pocket costs required, up to $400 even with copay ($11 per session for 36 sessions as covered by CMS)¹¹. Despite none of our cohort requiring copay, household income was still a significant predictor of SET completion, with this subset of patients in low-income zip codes that had only a 20% adjusted likelihood of completion compared to 49% adjusted likelihood in higher income counterparts. This may indicate that besides actual upfront cost of each session, there are other hidden costs that are preventing SET completion. Potential examples include cost of transportation and cost of time away from work, which is significant in a program that may require up to three sessions per week with each spanning 30 to 60 minutes.

Although increased drive time from patient residents to SET was not a statistically significant predictor of SET completion (P = .10), the lack of significance may be limited by the study’s sample size. We hypothesize it is likely clinically significant given the fact that transportation is consistently cited as a reason for being unable to participate in SET in the existing survey literature. Further, the median drive time for all SET participants was only 15 minutes, which may suggest that those who had further commutes never even began SET. Lastly, transportation issues to SET was the most reported reason for SET attrition.

Our finding that an increased BMI predicts SET attrition is novel and has not been previously reported in the existing literature. Although Cetlin et al.¹⁶ and Altin et al. did collect BMI data, the former only looked at what patients would participate in a in a hypothetical SET scenario and the latter did not perform any multivariable analyses predicting declining SET involvement. Additionally, higher BMI has been shown to be associated with a lower annual household income, which indicates that there may be an interaction between these two variables; however, our model testing did not detect collinearity in our data analysis. Similar to other studies, motivation, illness/functional limitation and time were again cited by survey respondents in our study. These results, in combination with the existing literature, suggest that the use of supervised exercise therapy for individuals with lower socioeconomic status needs to be studied further to determine the specific challenges they are facing to improve SET adherence.

Our results also report no significant differences in post-SET ABI or across or within SET graduates and non-graduates. However, patients in our cohort improved with regard to their DASI and metabolic equivalent level. Given there is inherent selection bias to our cohort, only including patients who both got referred to SET and made an appointment, this may represent a cohort who was going to do well overall and thus dampen the effect size of SET on ABI reduction. However, our findings are consistent with Schieber et al.²² who found that despite SET having improved the quality of life and walking distance within their cohort, ABI did not significantly change. Altin et al.’s work also supports this finding²⁰. Several studies have also found SET improved function and symptoms in patients with IC.^{8, 9} The lack of change in ABI and improvements in patient functionality found in our investigation in addition to the previous literature is consistent with SET being hypothesized to improve collateralization as opposed to improving the inflow of vessel disease. Interestingly, we found no differences in Vascular QOL score between SET graduates and non-graduates. However, this may be attributable due to the inherent selection bias of our cohort in addition to the limited sample size.

Although prior work has demonstrated that SET-enrollees have a decreased rate of endovascular (11.9% vs. 16.9%) and surgical revascularization (2.4% vs. 5.9%) relative to those not enrolled in SET, our study demonstrates that there is no difference in the rate of PVI within one year between SET graduates and non-graduates. Though our results may seem to differ from existing data, our comparator group consists of individuals that participated in SET to varying degrees that did not meet the criteria for SET graduation. Although current guidelines state that PAD SET programs consist of 36–62 sessions across 3–6 months, it is plausible that several of the participants in the SET non-graduate cohort did indeed participate in enough sessions (< 36) to reap the benefits of SET regarding the prevention of future PVI. Additionally, we only had 13 patients (30% of non-graduates and 19.1% of graduates) undergo PVI within one year, which significantly limits the potential power of this analysis. In addition to these findings, we found that commonly reported reasons for attrition were transportation challenges and lack of motivation, which may be attenuated by a shorter SET program duration. Further, any participation in SET, even without full course completion, may be associated with a reduction in revascularization procedures and highlight the need for further evaluation of the optimal programmatic requirements while simultaneously investigating structured ways to improve accessibility and adherence.

Our study has several limitations. Firstly, the small sample of IC patients who attended SET at our institution limited the potential for multivariable analyses as we were forced to balance the inclusion of an increasing number of clinically relevant variables with the potential for model overfitting and the associated increasing error in our estimates. This further prevented us from performing multivariable analyses on our secondary outcomes due to the abundance of risk factors for ABI changes and PVI required for clinically meaningful adjustments. These data could have shed more insight into the impact of SET attrition or graduation in patients with IC. Further analysis, once our cohort expands significantly, will be necessary in order to confirm the findings of our current study and generate multivariable analyses of secondary outcomes. In addition, our population’s insurance allowed them to avoid a copay, which limits the generalizability of our findings. Variable missingness disallowed the inclusion of use of pre-SET metabolic equivalents in our multivariable modeling and may represent unmeasured confounding. The potential for residual confounding was highlighted by the fact that illness and functional limitation was reported as a reason for incompletion in the survey portion of our study. Investigating barriers to SET completion in larger sample sizes will be beneficial to further validate these findings as the number of SET referrals increases. Additionally, longer-term follow-up and analysis of PROMs could provide more comprehensive insight into the benefits of SET programs in patients with IC.

Despite its limitations, our hypothesis-generating data represents important ground-work for future study. Analysis of a larger cohort of IC patients, including those who did and did not attend SET therapy, will shed light on physician and patient-level barriers to SET referral. Additionally, and potentially most importantly, qualitative-based research methodology inductively investigating modifiable barriers in both physicians and patients is urgently needed to fill in gaps left by retrospective and survey data. Qualitative data can provide insight on both patient and provider experiences, biases, and opinions that can directly influence referral and completion of SET. Additionally, understanding how income plays a role in SET attrition despite insurance coverage will help reveal the “hidden costs” of participation. Such data is necessary to inform future implementation initiatives in order to boost both SET referral and completion.

CONCLUSION

Although SET participation has been shown to improve lower extremity and functionality outcomes, only 40% of referred patients completed therapy programs in our cohort. We found that socioeconomic and demographic factors including income, transportation, and BMI influence the adjusted odds of and willingness to complete SET programs. Our findings suggest that further research into strategies such as improved accessibility and patient outreach is necessary to mitigate existing barriers and improve SET graduation.

ARTICLE HIGHLIGHTS.

Type of Research: Single-center retrospective cohort study

Key Findings: Only 40.38% of 52 patients completed supervised exercise therapy (SET) for intermittent claudication. Barriers to completion included residence in a zip code with median household income < $47,000 and higher BMI. Metabolic Equivalent Level and DASI scores were significantly greater after participating in a SET program.

Take home Message: Both socioeconomic and functional factors influence the odds of completing SET programs, indicating a need for holistic pre-referral assessment to facilitate enhanced program accessibility for these populations.