Anatomy of a Failure: A Retrospective Evaluation of a Cognitive Bias Modification Intervention to Promote Physical Activity in Cardiac Rehabilitation

This retrospective evaluation of the IMPACT trial reveals that a cognitive bias modification intervention for promoting physical activity in cardiac rehabilitation patients suffered from low enrollment and engagement due to perceived task irrelevance, monotony, and boredom, highlighting the critical need for patient-centered design and pre-implementation qualitative research to ensure clinical feasibility.

The Gist

The Big Picture: A Well-Intentioned Experiment That Stalled

Imagine a group of researchers trying to help heart patients get back in shape. They had a clever idea: instead of just telling patients to "exercise more," they wanted to rewire their brains automatically. They used a computer game to train patients to instinctively approach images of exercise and avoid images of sitting on the couch.

They called this the IMPACT trial. The goal was to see if this "brain training" could make patients move more after leaving the hospital.

The Result? The experiment hit a wall. Very few people signed up, and even fewer finished the game. The researchers had to stop the study early because they couldn't get enough data to prove if it worked. This paper is their "autopsy" of the failure—looking at why it didn't work so future experiments can do better.

---

The Setup: The "Brain Gym" Game

Think of the intervention like a video game.

• The Players: 68 heart patients (mostly men, average age 58).
• The Game: A tablet app called VAAST. When a picture of someone running appeared, the player had to push a button to "zoom in" (approach). When a picture of someone watching TV appeared, they had to push a button to "zoom out" (avoid).
• The Plan: Patients were supposed to play this game 12 times over 6 weeks.
• The Goal: To change their automatic habits so they naturally wanted to exercise.

What Went Wrong? (The Three Leaks in the Bucket)

The researchers looked at three main reasons why the project failed, using a framework called the "Theoretical Framework of Acceptability."

1. The "Empty Seat" Problem (Behavioral Issues)

Imagine a concert where 352 tickets were printed, but only 68 people showed up.

• The Reality: Of the 352 patients the researchers hoped to recruit, only 68 actually signed up. Of those 68, only 43 finished the minimum amount of training. By the end, only 17 people provided the final data.
• The Takeaway: The "no-show" rate was massive. The study was like a restaurant that ordered food for 350 guests but only had 17 diners. They couldn't cook a proper meal (get scientific results) with so few people.

2. The "Is This Even Worth It?" Problem (Cognitive Issues)

Imagine a student being asked to do extra homework that doesn't seem to help them pass the test.

• The Reality: Patients were skeptical. They asked, "How does pushing buttons on a screen make me want to run?" They felt the game was irrelevant to their actual recovery.
• The Burden: Patients were already tired from their real medical rehab (physical therapy, doctor visits). They saw this computer game as an annoying "extra chore" rather than a helpful part of their healing.
• The Staff Gap: The researchers (who ran the game) weren't fully part of the medical team. Patients and doctors saw them as "outsiders" bringing extra work, not as helpers integrated into the care plan.

3. The "Boredom" Problem (Emotional Issues)

Imagine being asked to play a very simple, repetitive card game for an hour every day.

• The Reality: Patients found the game boring. They said the pictures were dull, the task felt long, and it wasn't fun.
• The Visuals: The images on the screen were described as "unappealing" or even "depressing." Patients wanted something colorful and engaging, not a gray, repetitive task.
• The Result: The "novelty" wore off quickly. Once the game stopped being new, the boredom set in, and people quit.

Other Hurdles

• Tech Glitches: There were only 8 tablets for a whole hospital. If a patient was ready to play but the tablet was in use, they had to wait or skip a session.
• No Family Support: Usually, family members help keep patients motivated. In this study, only 10% of patients had family with them during sessions, leaving many to face the challenge alone.

What Did the Researchers Learn? (The "Fix-It" List)

The paper doesn't just list problems; it suggests how to fix them for the next attempt:

1. Make it Feel Like Part of the Team: Don't treat the computer game as a separate "research project." Integrate it so it feels like a normal, helpful part of the doctor's advice, not an extra burden.
2. Make it Fun (Gamify It): If the game is boring, people won't play. Future versions should look more like a video game with rewards, animations, and "vitality gauges" (like a health bar in a video game) rather than a boring test.
3. Ask the Patients First: Before building the game, ask the patients what they want. If they say "we don't like these pictures," change them. This is called "Patient and Public Involvement."
4. Get the Family Involved: Bring family members into the process to help support the patient.
5. Check the Tech: Make sure there are enough devices and that the staff running the game are fully welcomed by the hospital team.

The Bottom Line

The researchers admit: "This specific version of the brain-training game failed because patients found it boring, irrelevant, and too much work."

They aren't saying the idea of changing brain habits is bad. They are saying that to make it work, the "delivery" needs to be much better. It needs to be less like a boring homework assignment and more like an engaging, helpful, and seamless part of the patient's recovery journey.

Important Note: The paper explicitly states that because the study failed so badly (with so few participants), they cannot say whether the game would have worked if everyone had finished it. They are only reporting on why people didn't want to play it.

Technical Summary

Technical Summary: Anatomy of a Failure in Cognitive Bias Modification for Cardiac Rehabilitation

Problem Statement
Promoting regular physical activity (PA) is critical for patients with cardiovascular diseases to reduce mortality and improve quality of life. However, many patients exhibit low PA levels, and previous interventions have struggled to produce sustained behavioral changes. While recent research suggests targeting automatic approach-avoidance tendencies via Cognitive Bias Modification (CBM) could enhance intervention efficacy, no randomized controlled trials (RCTs) had previously evaluated this approach specifically in cardiac rehabilitation populations. The Improving Physical Activity (IMPACT) trial was designed to address this gap by using a CBM intervention to modify automatic tendencies toward exercise-related stimuli. This paper presents a retrospective evaluation of the IMPACT trial, which was terminated early due to significant implementation failures, specifically focusing on the low acceptance and engagement of the intervention.

Methodology
The study was a retrospective evaluation of a pre-registered RCT conducted at the University Hospital of Geneva, Switzerland.

• Participants: The target sample was 352 cardiac rehabilitation patients. Ultimately, 68 patients were enrolled (Mage = 57.76 years, 87% male).
• Intervention: The intervention group received 12 CBM sessions over 6 weeks using the Visual Approach/Avoidance by the Self Task (VAAST). In this task, patients used a touchscreen tablet to approach images of physical activity (presented in portrait format) and avoid images of sedentary behavior (presented in landscape format), with 90% of trials reinforcing the approach tendency. The control group received a sham intervention with a 50/50 distribution of approach/avoidance formats.
• Outcome Measures: The primary outcome was moderate-to-vigorous PA measured via accelerometers post-discharge. However, this paper focuses on acceptance, assessed through three dimensions based on the Theoretical Framework of Acceptability (TFA):
  1. Behavioral: Enrollment rates, engagement in CBM sessions (completing ≥3 sessions), and provision of valid accelerometer data.
  2. Cognitive: Perceived effectiveness and task relevance, derived from informal verbal feedback.
  3. Emotional: Affective reactions (boredom, disinterest), also derived from informal verbal feedback.
• Data Collection Note: The cognitive and emotional data were not collected via standardized questionnaires or formal qualitative methods. Instead, they were compiled from unsystematic, informal observations and verbal reports by research assistants (RAs) during the intervention. The authors explicitly acknowledge this as a methodological limitation, noting the risk of recall bias.

Key Results
The trial demonstrated extremely low acceptance and high attrition, leading to its termination:

• Enrollment: Only 68 of the 262 patients screened (26%) agreed to participate, representing just 19% of the originally required sample size (352).
• Engagement: Of the 68 enrolled, only 43 (63%) completed the minimum required three CBM sessions.
• Data Completion: Only 17 patients (25% of those enrolled, 6% of the target) provided valid accelerometer data for the primary outcome.
• Attrition: The overall attrition rate was 75% among enrolled participants. The final sample size yielded a statistical power of only 10%, far below the planned 90%.
• Barriers Identified:
  • Cognitive: Patients expressed skepticism regarding the task's relevance to their physical recovery and perceived the intervention as an unnecessary burden on top of their standard rehabilitation.
  • Emotional: Patients reported significant boredom, finding the task monotonous, long, and unengaging. The visual design (images) was criticized as unappealing.
  • Implementation: Research assistants were not fully integrated into the clinical team, leading to a perception of the study as an "extra" task rather than part of care. Resource constraints, including a shortage of tablets (8 devices for all patients) and limited RA availability, further hindered recruitment and consistency.

Key Contributions
This paper contributes to the field by:

1. Documenting a Failure: Providing a transparent, retrospective analysis of why a theoretically sound CBM intervention failed in a real-world cardiac rehabilitation setting, a type of reporting often lacking in literature.
2. Multidimensional Acceptance Assessment: Utilizing a framework that examines behavioral, cognitive, and emotional indicators of acceptance, highlighting that low engagement stems from a complex interplay of perceived irrelevance, boredom, and logistical friction.
3. Identifying Specific Barriers: Pinpointing that the "digital" nature of the intervention, when not seamlessly integrated into clinical workflows, can be perceived as a burden rather than a support, particularly for a population already managing a demanding rehabilitation protocol.
4. Proposing Implementation Strategies: Offering concrete recommendations for future trials, including the necessity of Patient and Public Involvement (PPI) in the design phase, the integration of research staff into clinical teams, the use of more engaging/gamified task designs (e.g., ABC training or VR), and the importance of family involvement.

Significance and Claims
The authors modestly claim that the study's primary significance lies in its ability to "accelerate the development of more effective digital interventions" by openly sharing implementation lessons. They argue that the low acceptance observed reveals critical barriers—specifically perceived task irrelevance, monotony, and implementation burden—that undermine the feasibility of digital CBM in cardiac care.

The paper concludes that for such interventions to succeed, they must be:

• Patient-centered: Designed with input from patients and stakeholders to ensure relevance and engagement.
• Seamlessly Integrated: Embedded into existing clinical routines so they are not viewed as additional burdens.
• Methodologically Rigorous: Developed through iterative phases (including qualitative research and pilot studies) as outlined in the Medical Research Council (MRC) framework, rather than moving directly to full-scale RCTs.

The authors emphasize that while the IMPACT trial failed to demonstrate efficacy due to these implementation issues, the insights gained are vital for preventing similar failures in future research and for ensuring that digital health tools are acceptable and feasible for the target population.