To analyse the average number of concepts in a typical question, we can break down each sentence into its fundamental conceptual components, similar to sentence such as “the cat sat on the mat”, by considering subjects, verbs, objects, prepositions and qualifiers as distinct cognitive units.

Step 1: Define Conceptual Elements in a Sentence

Each sentence contains core linguistic and cognitive components, such as:

·       Nouns (e.g. “cat” = a distinct entity)

·       Verbs (e.g. « sat » = an action or state)

·       Prepositions (e.g. “on” = a spatial relationship)

·       Definite / Indefinite Articles (e.g. “the” = specificity)

·       Modifiers (adjectives, adverbs, e.g. “quickly” adds another layer of meaning.)

Step 2: Extract Concept from Common Clinical Outcome Questions

Let us take anonymised real-world COA/PRO questions and break them down into conceptual units.

Example 1: Depression Assessment

“Over the past two weeks, how often have you felt down, depressed or hopeless?”

Conceptual Breakdown:

1.       Over the past two weeks = Time frame: 1 concept.

2.       How often = Frequency scale: 1 concept.

3.       You = Subject, patient as reference: 1 concept.

4.       Felt = Cognitive state: 1 concept.

5.       Down = Emotion: 1 concept.

6.       Depressed = Emotion: 1 concept.

7.       Hopeless = Emotion: 1 concept.

Total concepts = 7.

Example 2: Fatigue

“In the past 7 days, how often did you feel tired, weak or exhausted?”

Conceptual breakdown:

1.       In the past 7 days = time frame: 1 concept.

2.       How often = frequency: 1 concept.

3.       You = subject, patient as reference: 1 concept.

4.       Felt = Subjective experience: 1 concept.

5.       Tired = Physical symptom: 1 concept.

6.       Weak = Physical symptom: 1 concept.

7.       Exhausted = Physical symptom: 1 concept.

Total concepts = 7.

Example 3: Pain question

“On a scale of 0-10, how would you rate your worst pain in the past 24 hours?”

Conceptual Breakdown:

1.        On a scale of 0-10 = measurement method: 1 concept (although arguably 2, if we separate ‘scale’ from ‘0-10’)

2.        How = Question structure: 1 concept.

3.        Would you rate = cognitive evaluation: 1 concept.

4.        Your = personal preference: 1 concept.

5.        Worst pain = symptom severity: 1 concept.

6.         In the past 24 hours = time frame: 1 concept.

Total concepts = 6.

Example 4: Functionality question

“During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)?”

Conceptual breakdown:

1.       During the past 4 weeks = time frame: 2 concepts.

2.       How much = degree of impact: 1 concept.

3.       Pain = symptom reference: 1 concept.

4.       Interfere = impact on function: 1 concept.

5.       With your normal work = daily activity: 1 concept.

6.       Including both work outside the home and housework = clarification of work types: 2 concepts.

Total concepts = 8.

Step 3: Compute the Average Number of Concepts in a Typical PRO Question

Based on these four common examples:

·       Depression item: 7 concepts.

·       Fatigue question: 7 concepts.

·       Pain inventory: 6 concepts.

·       Health survey item: 8 concepts.

Average number of concepts per question = 7.

On average, a typical PRO or COA question contains around 7 concepts. This already presses against Miller’s “magical” 7 ± 2, but more importantly, it exceeds Cowan’s more realistic 4 ± 1 limit. In other words, even the “average” COA item is designed at or beyond the natural capacity of working memory, particularly for patients under stress, pain or fatigue. The implications for cognitive load and PRO design include:

·       Questions with more than 7 concepts may exceed short-term memory capacity (if we take Miller’s Magic Number 7±2 as a standard).

·       Even 6–7 concepts exceed Cowan’s 4 ± 1 realistic working memory span, meaning many patients are already overloaded by design.

·       Multi-concept questions increase cognitive burden, leading to fatigue, confusion, and response bias.

·       Regulatory guidance (FDA, EMA) emphasizes reducing complexity, supporting single-concept, simple-structure questions.

Best Practices for Designing Low-Cognitive Load COA Questions

·       Aim for fewer than 7 concepts per question. Better still, design toward Cowan’s 4 ± 1.

·       Separate symptoms or experiences into distinct items.

·       Use clear, concise language (avoid unnecessary qualifiers).

·       Limit recall periods to reduce memory strain.

·       Reduce response options (5-7 choices) to prevent decision fatigue.

Thank you for reading,

Mark Gibson, Giggleswick, United Kingdom

Nur Ferrante Morales, Ávila, Spain

August 2025