Effect of Gamification Plus Automated Coaching to Increase Physical Activity Among Patients With Peripheral Artery Disease: The GAMEPAD Randomized Controlled Trial

Abstract

Background

Supervised exercise therapy improves walking performance in patients with peripheral artery disease (PAD), but few participate. Interventions leveraging concepts from behavioral economics increase physical activity in patients at high cardiovascular risk, but barriers to physical activity differ in patients with PAD.

Methods

In this randomized controlled trial, conducted from October 2020 through January 2024, patients with PAD were provided with a wearable fitness tracker, established a baseline daily step count, and set a step goal increase. They were randomly assigned to attention control or to gamification. The control group received feedback from the fitness tracker but no other interventions for 24 weeks. The gamification group was entered into a 16‐week game designed using insights from behavioral economics and received educational text messages. No intervention occurred during an 8‐week postintervention follow‐up period.

Results

A total of 103 patients (mean age, 70±9 years; 54 [52%] men, 74 (72%) with exertional lower extremity symptoms) were randomized to attention control (n=52) or gamification (n=51). Compared with controls, gamification participants had a greater increase in mean daily steps from baseline during the intervention period (adjusted difference, 920 [95% CI, −22 to 1861]; P=0.06) that became statistically significant during the follow‐up period (adjusted difference, 1074 [95% CI, 133–2015]; P=0.03).

Conclusions

In this randomized clinical trial, gamification increased physical activity compared with attention control over a 24‐week follow‐up. This intervention may represent a scalable approach for increasing physical activity in patients with PAD who are not able to participate in supervised exercise therapy.

Registration

URL: https://clinicaltrials.gov/study/NCT04536012; Unique Identifier: NCT04536012.

Nonstandard Abbreviations and Acronyms

GAMEPAD

Gamification‐Augmented Home‐Based Exercise for Peripheral Artery Disease

SET

supervised exercise therapy

WIQ

Walking Impairment Questionnaire

Clinical Perspective.

What Is New?

• In patients with peripheral artery disease, a fully home‐based intervention leveraging behaviorally designed gamification and automated coaching increased daily step count by nearly 2000 steps more than baseline and 1000 steps more than attention control, over the course of a 16‐week intervention and 8‐week postintervention follow‐up.

What Are the Clinical Implications?

• Supervised exercise therapy represents the standard of care for improving quality of life and functional capacity in patients with peripheral artery disease, but it is inaccessible to many patients.

• This scalable intervention may represent an attractive approach for increasing physical activity in the many patients with peripheral artery disease who are not able to participate in supervised exercise therapy.

Based on randomized clinical trials demonstrating that supervised exercise therapy (SET) improves walking performance, functional capacity, and quality of life in patients with peripheral artery disease (PAD), ¹ , ² , ³ consensus guidelines recommend SET for patients with PAD and lower extremity exertional symptoms. ⁴ , ⁵ Despite the strong evidence for benefit and a decision by the Centers for Medicare and Medicaid Services to cover SET for patients with PAD, ⁶ few patients with PAD attend SET programs due to logistical and financial barriers. ⁷ , ⁸ A home‐based walking intervention may overcome some of these barriers and increase accessibility. However, although there have been some promising tests of hybrid programs incorporating both home‐ and center‐based walking, ³ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ the utility of fully home‐based unstructured walking is uncertain. ⁴

Behavioral economics is a scientific field that uses principles from psychology and economics to understand and guide individuals’ decision‐making. Behavioral economic concepts relevant to motivation for engaging in healthy behaviors include immediacy (people are more motivated by immediate rewards than future rewards), endowment effects (people put more value in something they already have relative to something they could attain), loss‐framing (people are more motivated when a situation is framed as a loss rather than a gain), and status quo bias (people avoid initiating change). ¹⁴ , ¹⁵ , ¹⁶ Gamification programs designed using principles from behavioral economic theory have increased physical activity compared with control in several trials enrolling patients with or at risk for cardiovascular disease. ¹⁷ , ¹⁸ , ¹⁹ In these studies, each participant assigned to gamification was awarded points each week, with a fraction of those points taken away each day that they did not meet their step goal. They progressed through levels each week based on the total points they retained from that week. Applying similar principles to develop an entirely home‐based physical activity program for patients with PAD could allow more patients with PAD to access a formal exercise program. We therefore conducted a randomized controlled trial to test the efficacy of gamification plus automated coaching, compared with attention control, to increase physical activity in patients with PAD.

Methods

Study Design

GAMEPAD (Gamification‐Augmented Home‐Based Exercise for Peripheral Artery Disease) was a randomized clinical trial conducted from October 20, 2020, to January 9, 2024, consisting of a 2‐week run‐in period, a 16‐week intervention period, and an 8‐week follow‐up period. Details of the study design and protocol have been published previously. ²⁰ The trial protocol (Data S1) was approved by the University of Pennsylvania Institutional Review Board, and all patients provided informed consent for participation and use of their data. Data were not deidentified. The study was conducted using Way to Health, ²¹ a research technology platform at the University of Pennsylvania used to implement and test behavior change interventions. ¹⁸ , ¹⁹ , ²² , ²³ The data that support the findings of this study are available from the corresponding author upon reasonable request. This randomized controlled trial is reported in accordance with the Consolidated Standards of Reporting Trials 2010 statement.

Participants

Recruitment occurred from October 20, 2020, to July 25, 2023, within the University of Pennsylvania Health System. Participants were eligible for the trial if they were aged ≥18 years, owned a smartphone or tablet operating the iOS or Android operating system, and had PAD, defined as ankle–brachial index <0.90, lower extremity computed tomography scan or ultrasound consistent with PAD, angiography with ≥70% stenosis in any lower extremity artery, or a history of medical or surgical lower extremity revascularization. Participants were excluded if they were unable or unwilling to provide informed consent; had chronic limb‐threatening ischemia (defined as rest pain, ulceration, or tissue loss involving the lower extremity), planned lower extremity revascularization, or prior above‐ or below‐knee amputation; required a wheelchair or use of a walking aid other than a cane; were participating in a SET program for patients with PAD; had anticipated life expectancy <6 months; or if there was any other reason it was not feasible to complete the entire 6‐month study. Potentially eligible patients were identified using data from the health system’s clinical data warehouse (n=5053) or by direct referral from their cardiologist or vascular surgeon (n=20). They were then contacted by unsolicited email, text message, or telephone call and offered participation in the study and provided with a link to the online study platform.

On the study website, participants created an account, provided informed consent, and completed baseline surveys. Baseline surveys included self‐reported demographics (including race and ethnicity), markers of socioeconomic status and familiarity with wearable fitness trackers, PAD symptom status (the San Diego Claudication Questionnaire), ²⁴ PAD‐specific functional capacity (Walking Impairment Questionnaire [WIQ]), ²⁵ and general quality of life (Short Form‐36). Participants’ medical history was extracted from the electronic health record. Eligible participants were mailed a wrist‐worn activity tracker (Fitbit Inspire 1 and 2), which they connected to the Way to Health platform for remote data collection. Participants received $25 for enrolling, $25 for completing the 16‐week intervention and survey, and $25 for completing the 24‐week survey. Participants were also allowed to keep the Fitbit at the conclusion of the study.

Baseline Step Count

Each participant completed a 2‐week run‐in period, during which time a baseline step count was estimated using the second week of data. ¹⁷ , ¹⁹ , ²² As prespecified in the original trial protocol, participants who did not complete the run‐in phase or had baseline step counts >7500 or <1000 steps per day were excluded. Those taking >7500 steps per day were excluded on the basis of data showing that increasing step count beyond 7500 steps per day does not improve outcomes in all‐comer populations. ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ Those taking <1000 steps per day were excluded to ensure that we did not enroll frail patients with a need for more intensive exercise training or physical therapy.

Goal Setting

Each participant was informed of his or her baseline step count by text message and asked to select a goal step increase of 33%, 40%, or 50% above their baseline, or a custom goal at least 1500 steps greater than their baseline. This strategy allows participants to set their own goal, the most effective goal‐setting approach ¹⁸ while providing only achievable yet ambitious options.

Randomization

After goal setting, participants were randomized 1:1, using an electronic number generator in the Way to Health platform, to either attention control or gamification plus automated coaching, stratified by baseline step count (<2500 steps, 2500–5000 steps, >5000 steps). Treatment assignment was necessarily unblinded to participants, but participants were not told anything about the other study arm. Investigators, statisticians, and data analysts remained blinded to arm assignments until the analysis was completed.

Interventions

All patients were instructed to begin walking for 15 minutes per day at a comfortable pace that caused minimal or no pain, to stop and rest if they developed more severe leg pain before resuming their walk, and to gradually increase their walking duration to 30 to 60 minutes each day. This differs from the regimen typically prescribed for SET—that is, walk at a pace that will induce claudication after 5 to 10 minutes, rest until the pain resolves, continuing for 60 minutes ⁶ —but the simpler, less discomfort‐inducing strategy was developed to overcome concerns that patients would not be adherent to a traditional SET strategy without personalized motivation from a physical therapist or exercise physiologist.

Participants randomized to attention control received a daily text message for 24 weeks telling them whether they met their step goal the previous day.

Participants in the gamification plus automated coaching arm were entered into a game with points and levels that was run automatically (participants did not have to actively play the game, just strive for step goals) and provided a daily notification of their progress. The design of the gamification intervention was based on prior work ¹⁸ , ¹⁹ , ²² , ²³ , ³⁶ and incorporated principles from behavioral economics.

First, each participant signed a pledge to strive to achieve their daily step goal. ³⁷ Participants had a 4‐week ramp‐up toward their step goal, during which time their daily step goal was gradually increased each week until it reached their self‐selected number. After the ramp‐up, the participant’s goal remained the same for the rest of the study.

Second, at the start of each week, each participant received 70 points. Each day, the participant was informed of their step count from the previous day. If the step goal was met, the participant retained their points; if the step goal was not met, they were notified that they had lost 10 points. We chose this “loss‐framed” approach because prospect theory shows that highlighting potential losses is more effective at driving behavior change than emphasizing gains. ¹⁴

Third, at the end of each week, participants moved up or down through 5 levels on the basis of their points retained the previous week. Those with ≥40 points advanced 1 level; those with <40 points dropped down a level. Each participant began in the middle level so that they would have immediate motivation to maintain this status. ³⁸

Fourth, we leveraged the fresh‐start effect—the concept that individuals are more motivated for aspirational behavior around temporal landmarks like the start of a new week ³⁹ —by awarding points at the start of each week. Furthermore, every 8 weeks, individuals in the 2 lowest levels were moved up to the middle level and offered a chance to change their goal step count within the original range of 33% to 50% above their baseline. This prevented participants from becoming discouraged if their initial goals were too ambitious.

Finally, each participant picked a family member or friend to receive an email each week summarizing the participant’s performance. Engagement of a support partner was intended to leverage social incentives, motivation derived from social interactions, acceptance, and status. Before participants started the game, they and their support partners had a 3‐way phone call with a study team member. During the call, the study team member reviewed the rules of the game, and each participant–partner dyad identified 3 strategies they could use to motivate the participant.

The game lasted for 16 weeks. After 16 weeks, participants in the gamification arm received the same messaging as the attention control arm (a daily message telling them whether they had met their step goal the prevous day) for an 8‐week postintervention follow‐up period.

Automated coaching consisted of biweekly text messages including information about the benefits of physical activity to improve functional capacity and symptoms in people with PAD, education about strategies to make walking at home easier, and motivation to keep striving for their goals.

Outcome Measures

The primary outcome was change in daily steps from baseline through the intervention period (weeks 5–16, which excludes the 4‐week ramp‐up phase). The 4‐week ramp‐up phase was excluded from the primary outcome because step goals differed between arms during this period. Secondary outcomes included change in daily steps from baseline through the follow‐up period and change from baseline through the intervention and follow‐up periods in WIQ score. The WIQ is a validated instrument that assesses patients’ ability to walk at various speeds and distances, and to climb stairs; higher scores indicate less functional impairment. The minimum clinically important difference in WIQ (or its subscales) has not been defined, but SET increases the WIQ distance score by ≈5 points more than control. ⁴⁰

Statistical Analysis

A priori power calculations were based on physical activity data from patients with PAD, who walk ≈3900±2689 steps per day. ¹² With 100 patients (50 in each arm), the study had 80% power to detect an 1100‐step difference in change in daily step count (roughly half a mile) between the 2 arms, with a 2‐sided α=0.05. At the time the study was conceived and initiated, the minimum clinically relevant difference in daily steps for patients with PAD was unknown. In a subsequent study, a 770 steps per day increase in physical activity was associated with a small change in health‐related quality of life, and a 1211 steps per day increase was associated with a large change. ⁴¹

All randomly assigned patients were included in the intention‐to‐treat analysis.

Missing data occurred on days when participants either did not use the wearable device or failed to upload their data. For the main analysis, we used multiple imputation for step values that were missing or for values <1000 steps per day, consistent with prior work, ¹⁸ , ²³ because daily step values <1000 are unlikely to represent full data capture. ²⁷ , ⁴² Five imputations were conducted using the mice package in R (R Foundation for Statistical Computing, Vienna, Austria), which allows for participant random effects, and results were combined using Rubin’s standard rules. The following determinants of missing data were included in the imputation model: study arm, calendar month (fitted as a nominal variable), week of study, baseline daily steps, age, sex, race and ethnicity, educational level, marital status, household income level, and self‐reported health. We performed 2 sensitivity analyses using collected data without imputation, 1 including days with step counts <1000, and 1 excluding these days.

The primary analysis fit mixed models for repeated measurements to evaluate changes in physical activity and quality‐of‐life outcomes adjusting for each participant’s baseline measure, time, calendar‐month fixed effects (fitted as a nominal variable), and participant random effects to account for repeated measures. In a fully adjusted model, we additionally adjusted for age, sex, race, education, marital status, self‐reported income, baseline self‐reported health, and body mass index. For change in steps, we assumed a normal distribution and obtained the difference in steps between arms for the intervention and follow‐up periods as least squared means.

Results

Of 5073 patients contacted and offered enrollment, 184 completed the informed consent process and all baseline surveys and were mailed a Fitbit device. Of these 184, 103 completed the baseline run‐in period with a step count >1000 but <7500 and were randomized to attention control (n=52) or gamification plus automated coaching (n=51) (Figure 1). Demographics and clinical characteristics were similar between the groups (Table 1). The mean age of the cohort was 69.7±8.6 years, 47.6% were women, and 15.5% were Black individuals. On the San Diego Claudication Questionnaire, 20.4% of patients reported classic claudication, 51.4% atypical leg pain, and 28.2% no lower extremity symptoms. Mean baseline daily step count was 4367±2039, and mean goal step count increase was 1353±1541.

Figure 1. Consolidated Standards of Reporting Trials diagram.

Potentially eligible participants identified via electronic health record searches were contacted by unsolicited email or phone call and offered participation in the trial via a link to the online study platform. More than 90% of those contacted either did not click on the link at all or did not complete enrollment on the study platform. Those who completed the informed consent process and baseline questionnaires were mailed a wearable fitness tracker, which they used to track daily step counts, establish a baseline, and set a step goal increase. Participants in the control arm received regular feedback from the wearable device but no other interventions. Participants in the gamification plus automated coaching arm were entered into a game that ran automatically during the first 16 weeks and then had no intervention during an 8‐week follow‐up period. Six patients randomized to gamification plus automated coaching dropped out before completing the intervention period (1 died of myocardial infarction, 3 had worsening lower extremity discomfort, and 2 were no longer interested in participating); 1 patient randomized to control dropped out before completing the intervention period (no longer interested in participating). No patients dropped out during the follow‐up period.

Table 1.

Baseline Characteristics

	Control (n=52)	Intervention (n=51)	Overall (N=103)
Age, y, mean±SD	70±9.1	69.4±8.2	69.7±8.6
Male sex, n (%)	25 (48.1)	29 (56.9)	54 (52.4)
Race and ethnicity
White	42 (80.8)	39 (76.5)	81 (78.6)
Black	6 (11.5)	10 (19.6)	16 (15.5)
Asian	2 (3.8)	0 (0)	2 (1.9)
Hispanic	0 (0)	1 (2.0)	1 (1.0)
Other	2 (3.8)	1 (2.0)	3 (2.9)
Education
Some high school	2 (3.8)	0 (0)	2 (1.9)
High school graduate	7 (13.5)	7 (13.7)	14 (13.6)
Some college	11 (21.2)	22 (43.1)	33 (32.0)
College graduate	32 (61.5)	22 (43.1)	54 (52.4)
Marital status, n (%)
Single	8 (15.4)	11 (21.6)	19 (18.4)
Married	27 (51.9)	30 (58.8)	57 (55.3)
Other	17 (32.7)	10 (19.6)	27 (26.2)
Annual household income
<$50 000	17 (32.7)	18 (35.3)	35 (34)
$50 000–$100 000	19 (36.5)	15 (29.4)	34 (33)
>$100 000	16 (30.8)	18 (35.3)	34 (33)
Prior use of smartphone or wearable to track steps, n (%)	35 (67.3)	27 (52.9)	62 (60.2)
Body mass index, mean±SD	29.4±7.2	30.8±5.4	30.1±6.4
Current smoking, n (%)	7 (13.5)	4 (7.8)	11 (10.7)
Hypertension, n (%)	37 (71.2)	46 (90.2)	83 (80.6)
Hyperlipidemia, n (%)	37 (71.2)	36 (70.6)	73 (70.9)
Diabetes, n (%)	17 (32.7)	20 (39.2)	37 (35.9)
Prior myocardial infarction, n (%)	16 (30.8)	10 (19.6)	26 (25.2)
Stroke, n (%)	8 (15.4)	4 (7.8)	12 (11.7)
Heart failure, n (%)	7 (13.5)	5 (9.8)	12 (11.7)
Chronic obstructive pulmonary disease, n (%)	7 (13.5)	12 (23.5)	19 (18.4)
Kidney disease, n (%)	10 (19.2)	10 (19.6)	20 (19.4)
Baseline daily step count, mean±SD	4540 (2379)	4191 (1627)	4367 (2039)
Goal step increase, mean±SD	1160 (2033)	1549 (740)	1353 (1541)
San Diego Claudication Questionnaire, n (%)
Classic claudication	9 (17.3)	12 (23.5)	21 (20.4)
Atypical leg pain	26 (50.0)	27 (52.9)	53 (51.4)
No symptoms	17 (32.7)	12 (23.5)	29 (28.2)

During the intervention and follow‐up periods, step data were missing or <1000 on 11.1% of participant days, which is similar to previous physical activity interventions (Table S1). A total of 96 participants (94.1%) completed the entire 24‐week study. Overall, there were 54 adverse events in 34 participants, including 11 serious adverse events in 9 participants (Table S2). No adverse reactions related to the interventions were reported throughout the entire trial. The trial stopped enrolling new participants after reaching the prespecified enrollment target, although participants currently in the baseline period when the 100th participant was enrolled were allowed to continue the enrollment process and be randomized. The trial was completed when all enrolled participants reached the end of the postintervention follow‐up period (24 weeks after starting the intervention).

Daily Step Counts

Unadjusted mean daily step counts by week and study arm are shown in Figure 2. The gamification plus automated coaching arm took more steps per day and had a greater change from baseline than attention control throughout the entire trial.

Figure 2. Mean daily steps by week and study arm.

Mean daily steps derived from the mixed model for repeated measures for each arm using imputed data are shown; shading around each line represents ±1 standard error. Vertical lines separate the ramp‐up, main intervention, and follow‐up periods.

In the main adjusted model, participants in the gamification plus automated coaching arm had an increase from baseline of 920 steps more than those in the attention control over the main intervention period (95% CI, −21 to 1861; P=0.06) and an increase from baseline of 1074 steps more than those in attention control over the 8‐week follow‐up period (95% CI, 133–2015; P=0.03) (Table 2). Results were directionally similar but attenuated in the fully adjusted model. In sensitivity analyses that used collected data without multiple imputation, differences in mean daily steps between the control and gamification plus automated coaching arm were larger and statistically significant over both the intervention and follow‐up periods (Tables S3 and S4).

Table 2.

Daily Step Outcomes

Variable	Control	Intervention
Steps per day, baseline, mean±SD	4540±2379	4191±1627
Steps per day, weeks 5–16 (main intervention period), mean±SD	5323±3021	5960±2985
Main adjusted model
Difference vs control (95% CI)	NA	920 (−21 to 1861)
P value	NA	0.06
Fully adjusted model
Difference vs control (95% CI)	NA	774 (−248 to 1797)
P value	NA	0.13
Steps per day, weeks 17–24 (follow‐up period), mean±SD	5157±2846	6002±2907
Main adjusted model
Difference vs control (95% CI)	NA	1074 (133 to 2015)
P value	NA	0.03
Fully adjusted model
Difference vs control (95% CI)	NA	917 (−122 to 1957)
P value	NA	0.08

NA indicates not applicable.

Patient‐Reported Outcomes

Unadjusted scores on patient‐reported outcome measures are shown in Figure 3. WIQ overall scores declined by 0.1 points from baseline to 16 weeks and increased by 1.4 points from baseline to 24 weeks in the control arm (where an increase represents less impairment in walking) and increased by 2.5 points from baseline to 16 weeks and by 7.2 points from baseline to 24 weeks in the gamification plus automated coaching arm. Compared with control, we observed that patients randomized to gamification plus automated coaching had a 5.7‐point greater increase in overall WIQ score (95% CI, −3.5 to 15.0) at 24‐week follow‐up, although this difference did not achieve statistical significance. We observed greater increases for participants in the gamification plus automated coaching arms than control in the WIQ distance subscale and Short Form‐36 general health and health change subscales at 16‐ and 24‐week follow‐up, although only the difference in Short Form‐36 health change was statistically significant (Table S5).

Figure 3. Unadjusted change from baseline to 16‐ and 24‐week follow‐up on patient‐reported outcomes by study arm.

Shown are data for the overall, distance, speed, and stair‐climbing measures, along with the PROMIS pain and social function scales, and SF‐36 general health and health change scales. PROMIS indicates Patient‐Reported Outcomes Measurement Information System; SF‐36, Short Form‐36; and WIQ, Walking Improvement Questionnaire.

Discussion

In this randomized clinical trial enrolling 103 participants with PAD, gamification plus automated coaching increased physical activity by ≈1000 steps per day more than attention control over a 16‐week intervention period and 8‐week follow‐up. Concordant with this increase in physical activity, patients randomized to gamification plus automated coaching reported a clinically meaningful reduction in exertional lower extremity symptoms. The intervention tested in this trial represents the first fully home‐based, fully automated approach to increase physical activity in patients with PAD, a population that benefits greatly from increasing physical activity but faces challenges in accessing effective center‐based exercise programs.

SET improves functional capacity and quality of life in patients with PAD ¹ ; however, patients with PAD are older, have functional limitations, and struggle with transportation and logistics. ⁴³ In 1 cohort study, just 2% of patients seen in PAD specialty clinics participated in SET. ⁸ Six previous trials have tested partially home‐based exercise programs in this population. Gardner et al randomized patients with PAD to supervised treadmill exercise, home‐based exercise, or control. ⁹ Both supervised treadmill exercise and home‐based exercise increased maximal and pain‐free walking distance at 12‐week follow‐up, but the home‐based exercise program required in‐person visits every 2 weeks to meet with an exercise physiologist. In another study with similar design, Gardner et al showed similar benefits of home‐based and center‐based exercise on 6‐minute walk distance and treadmill walking time. ¹⁰ The Group Oriented Arterial Leg Study demonstrated the effectiveness of a group‐mediated cognitive behavioral theory intervention for increasing treadmill walking distance, 6‐minute walk distance, and other patient‐reported measures of quality of life. ³ Although the majority of walking was done at home in the group‐mediated cognitive behavioral therapy intervention, there were weekly in‐person group coaching and walking sessions. A second trial of a similar intervention in patients with PAD and diabetes did not demonstrate effectiveness. ¹¹ In the Home‐Based Monitored Exercise for PAD trial, an intervention that tested telephone coaching and a wearable activity monitor to promote home‐based walking did not improve 6‐minute walk distance or patient‐reported outcomes compared with usual care. Finally, the Low‐Intensity Exercise Intervention trial tested high‐ and low‐intensity home‐based exercise versus usual nonexercising control in patients with PAD and found that high‐intensity exercise (defined as exercise that induces ischemic leg symptoms) increased 6‐minute walk distance over 12‐month follow‐up compared with control, but low‐intensity exercise did not. However, neither home‐based walking program was fully home‐based, as participants visited the medical center weekly for the first 4 weeks and received telephone coaching from exercise coaches on a weekly basis for the remainder of the 12‐month study period. ¹³

By contrast, the gamification plus automated coaching intervention tested in the GAMEPAD study was fully remote, with no center visits. There was a 3‐way phone call between participants, their support partner, and study staff at the start of the study, but study staff members had no specialized training in exercise physiology or coaching, and there was no scheduled, ongoing telephone‐based coaching during the conduct of the study. Instead, the intervention relied on concepts from behavioral economics, including immediacy, loss aversion, social ranking, goal gradients, and status quo bias, ¹⁴ , ¹⁵ , ¹⁶ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ which have increased physical activity in several trials enrolling populations with or at risk for cardiovascular disease. ¹⁷ , ¹⁸ , ¹⁹ , ²² , ²³ Although the primary outcome in this trial was technically neutral, the large point estimate for the effect size, positive results on the key secondary outcome measure of change from baseline daily steps through the end of 24‐week follow‐up, and demonstrated effectiveness of similar interventions for increasing physical activity in other trials indicates the promise of this approach. In particular, the continued increase in mean daily step count in the intervention arm over the postintervention follow‐up period is consistent with a hypothesized mechanism of action in which the intervention provides a “jump‐start” to increase physical activity, starting a virtuous cycle by which increased physical activity begets the physiological and psychological changes that help participants further increase physical activity. ⁴⁸ These changes take some time to occur but are durable when the intervention is withdrawn, which may explain why the differences between the intervention and control groups continued to grow during the 8‐week postintervention follow‐up period.

The intervention was composed of 2 parts, gamification and automated coaching, and it is impossible to separate out the relative contributions of each. Interventions based in behavioral economic theory are designed to help individuals overcome barriers to behavior change driven by biases that cause misalignment of individuals’ long‐term desires and near‐term decisions. Automated coaching was included in the intervention to help participants understand why it was important for them to walk more (especially important for patients with PAD, many of whom have significant knowledge gaps related to their disease ⁴⁹ ), and gamification was included to help align short‐term decisions with the long‐term desire to walk more.

In this trial, which was conducted entirely remotely, it was not feasible to test the effect of gamification plus automated coaching on 6‐minute walk distance, the gold standard for assessment of functional capacity in patients with PAD. ⁵⁰ Though 6‐minute walk is an important measure of functional capacity and is strongly predictive of long‐term outcomes in patients with PAD, ⁵⁰ patient‐reported quality‐of‐life measures are also important to individuals with PAD. ⁵¹ Daily step count is, on its face, an important patient‐centered measure of exertional capacity, reflecting patients’ ability to engage in activities of daily living, and is associated with meaningful changes in patient‐reported quality of life. ⁴¹ In addition to changes in daily steps, intervention arm participants in GAMEPAD had increases in patient‐reported measures, with change in the WIQ overall and distance scales comparable with the difference observed in comparisons between SET and control. ⁴⁰

The increase in physical activity with gamification plus automated coaching was in comparison with attention control, which involved goal setting and daily text messages, and was itself associated with substantially increased physical activity. Compared with their own baseline, intervention arm participants in GAMEPAD increased mean daily step count by >1700 from baseline to the end of the study and WIQ distance score by 7 points—changes that would be associated with meaningful changes in quality of life. ⁵² Given the strength of the attention control arm, the primary study results may underestimate the true impact of the intervention versus usual care, and change from baseline may provide a more meaningful estimate.

Ultimately, the results from GAMEPAD suggest that gamification plus automated coaching may become an important tool in a suite of options to offer patients with PAD. Although traditional SET has the greatest strength of evidence, access challenges currently limit its availability to most patients with PAD. ⁸ More intensive home‐based programs, with occasional center visits and intensive telephone coaching from exercise physiologists reviewing accelerometer‐based data may be an option for patients seen at some well‐resourced centers. ⁴ The intervention in GAMEPAD, however, requires minimal personnel (and no specialized personnel) and has minimal costs to implement at scale, a key strength of this intervention. Ubiquity of wearable devices and digital health platforms has allowed virtual deployment of remotely managed, scalable physical activity programs enhanced by behavioral economic theory. The flexibility of the Way to Health platform would allow for widespread delivery at multiple health systems using already‐existing infrastructure. In addition to demonstrating efficacy, this study shows the feasibility of such a program in older patients with PAD, with high levels of engagement. As such, should its effectiveness versus usual care be validated, it could be offered broadly, including in communities with limited access to vascular subspecialists, even if it is not as effective as more intensive and expensive programs. A key barrier, however, will be connecting with patients with PAD from these communities, as they were not well represented in GAMEPAD and may be challenging to engage in nontraditional care pathways.

This study does have a number of limitations. First, a minority of patients contacted and offered enrollment ultimately enrolled in the trial. Although all clinical trials enroll a selected cohort, and the low consent rate in this trial partially reflects a low‐touch approach to recruitment by text message and email rather than in‐person contact, ⁵³ the results of this study are most applicable to the cohort of patients who volunteer to participate in a study designed to test ways to increase physical activity. In particular, participants enrolled in GAMEPAD were more likely to be White individuals and college‐educated than the population of patients with PAD seen in many communities, and the gamification intervention should be tested in more diverse and representative populations of patients with PAD. However, a similar gamification intervention has been tested and demonstrated effective for increasing physical activity in multiple clinical trials collectively enrolling >2500 patients, ¹⁸ , ¹⁹ , ²⁰ , ²³ including 1 clinical trial that only enrolled participants from low‐income zip codes. ¹⁸ Second, not all patients enrolled in GAMEPAD had symptomatic PAD. However, 70% of patients did report some lower extremity symptoms at baseline, and it has been noted that even patients with PAD who do not report lower extremity symptoms may have substantial unrecognized functional limitation and benefit from physical activity. ² , ³ , ⁵⁴ , ⁵⁵ Third, the dropout rate was higher in the gamification plus automated coaching arm, though the overall number of patients who dropped out was small, and the relevance of this difference is uncertain. Fourth, this study was conducted in a single health system, and findings may not generalize to the broader population of patients with PAD. It will be important for the results from this study to be validated in a multicenter study and to determine whether effects on daily step count translate to other validated PAD outcomes. Finally, it is not certain whether improvements in daily step count will also improve 6‐minute walk distance or PAD‐specific quality of life, the most validated measure of PAD functional status and the outcome that patients with PAD report is most important to them, respectively. ⁵⁰ , ⁵¹ The effect of gamification and automated coaching on these outcomes should be investigated.

Conclusions

In this randomized clinical trial, gamification plus automated coaching increased mean daily steps compared with control over a 16‐week intervention period and 8‐week follow‐up. The scalable intervention tested in this study may represent an attractive approach for increasing physical activity in the many patients with PAD who are not able to participate in SET.

Sources of Funding

GAMEPAD was funded by the National Center for Advancing Translational Sciences of the National Institutes of Health under award number UL1TR001878, and supported in part by the Institute for Translational Medicine and Therapeutics of the Perelman School of Medicine at the University of Pennsylvania.

Disclosures

Dr Fanaroff reports research funding to the institution from Abbott and Boston Scientific. Dr Damrauer reports research funding to the institution from RenalytixAI and in‐kind research support from Novo Nordisk. Dr Patel reported receiving personal fees as the owner of Catalyst Health LLC. Dr Giri reports research funds to the institution and serving on advisory boards for Boston Scientific and Inari Medical. All other authors report no relevant conflicts of interest.

Supporting information

Data S1. Trial Protocol

Tables S1–S5