Many of the articles we have written recently about Clinical Outcome Assessments (COAs) have focused on how fragile and biased working memory is and how this potentially limits the accuracy of patient-reported outcomes. But COAs are only part of the broader ecosystem of patient-facing documents. Patients also encounter informed consent forms, labels, patient information sheets and leaflets, Instructions for Use (IFUs) and, increasingly, digital interfaces, such as eDiaries, eCOAs, apps and wearables.

All of these materials rely on patients being able to read, recall, evaluate and act on information. And all of them are vulnerable to the same set of working memory biases that shape COA responses.

This article explores how working memory biases appear in these other contexts, showing that issues like recency bias, salience bias and satisficing are not just problems for questionnaires, they also distort understanding of risk, device safety and digital adherence as well.

Consent Forms: Legal Density Meets Memory Bias

Consent forms are meant to enable informed choice but in practice they are often dense, lengthy and legalistic. They present multiple risks, rights and procedural details and expect patients to absorb and retain them.

Working memory biases systematically distort how these forms are processed:

• Recency bias means patients remember only the final clauses –often just administrative details– and forget crucial sections buried in the middle.

• Primacy bias ensures that the very first risks, such as very mild side effects, are recalled, while later, more serious but less vivid risks are neglected.

• Salience bias gives disproportionate weight to rare but dramatic outcomes (“death”, “paralysis”) overshadowing common but manageable risks like nausea or fatigue.

• Mood-congruent recall matters too: a patient already anxious about treatment tends to fixate on negative risks and forget protective safeguards.

The consequence is predictable: patients may sign forms with an overinflated sense of catastrophic risk and underappreciation of common but important issues.

Patient Information Sheets: Lists That Overwhelm

Information sheets, especially drug leaflets, are notorious for overwhelming patients. They often present long lists of side effect written in regulatory jargon.

Here, biases interact with volume:

• Interference and overlap: Similar descriptors (tiredness, fatigue, lethargy) blur together, making it hard for patients to know what is distinct.

• Average bias: Patients compress long lists into a vague sense of “this drug has lots of side effects”, rather than distinguishing between common and rare.

• Anchoring: Patients anchor on one striking side effect (hair loss, seizures), which defines their impression of safety.

• Omission bias: Routine but less salient effects (dry mouth, mild nausea) are forgotten, even if they are common.

The outcome is not just confusion but often non-adherence. Patients may stop taking medication altogether, convinced by salience and anchoring that the treatment is too risky.

IFUs: Memory Under Pressure

Instructions for use are operational documents. They tell patients how to set up and use devices: injectors, inhalers, monitors, often in stressful moments.

Biases emerge in predictable ways:

• Recency and primacy shape execution: the first and last steps are remembered but middle steps (often safety-related) are omitted.

• Telescoping distorts when errors are remembered, a mishap may be misattributed to the wrong occasion.

• Satisficing leads to patients skipping small-print warning to get the device working quickly.

• Interference arises when multiple warnings (“do not shake” versus “do not drop”) seem redundant, so patient ignore them altogether.

Because IFUs are often used under stress, current state bias compounds everything. Anxiety shrinks capacity further, increasing error rates.

Apps and Wearables: Alerts as Memory Tests

Digital health tools bring new patterns of bias. Apps and wearables generate alerts, reminders and metrics that demand patient attention.

• Recency bias makes the most recent alert seem most important, even if an earlier one was clinically critical.

• Salience means alerts that vibrate or beep feel urgent, regardless of true severity.

• Omission occurs as alert fatigue sets in; frequent but low-level notifications are ignored.

• Confirmation bias colour interpretation: a patient who believes their health is declining interprets neutral metrics as further evidence.

These biases explain why apps sometimes backfire: instead of supporting adherence, they become sources of noise, stress and misinterpretation.

The Shared Thread

Across all these document types, the pattern in clear. Working memory biases lead to:

• Distorted perception of risk (consent forms, information sheets).

• Omitted safety steps (IFUs).

• Misprioritised alerts (apps/wearables).

The same cognitive shortcuts that affect COA responses ripple outwards, undermining comprehension, safety and adherence.

Conclusion

Working memory biases are not just methodological nuisances for COAs. They are fundamental constraints on how patients engage with all clinical documentation. Consent forms are shared by primacy, salience and recency. Information sheets are undermined by interference, anchoring and averaging. IFUs collapse under omission and stress. Apps magnify recency, salience and confirmation biases.

The implication is clear: if we want patient-facing materials to be effective, we must design them with working memory in mind. That means brevity, chunking, clear grouping, reduced redundancy and awareness of how biases shape recall.

Thank you for reading,

Mark Gibson, Leeds, United Kingdom

Nur Ferrante Morales, Ávila, Spain

September 2025