Personality traits are at work behind a patient’s behavior in clinical trials, leading to data variability and more difficult decisions. Here’s how Cognivia’s validated questionnaire – when paired with an ML model – gives clinical trial teams the information they need to understand those traits and make more confident decisions about the future of a compound. 

In clinical trials, data variability makes it more difficult to demonstrate if a compound is truly safe and efficacious. However, variability is what makes patients human. Differences in ages, gender, medical history, and psychology create noise in clinical trial results. In other words, the true difference between treatment groups gets muddied by all the other differences between each patient. 

Clinical trial teams have methods for accounting for differences in measurable factors like age, gender, and medical history. Traits and contextual factors, however, are large influencers of patient response and behavior but historically impossible to quantify. 

Until now. 

Turning patient personality traits into data

The most notable patient response with roots in traits is the placebo response – a challenge that has impacted drug development for years. 

To help clinical trial teams demonstrate statistically significant differences in outcomes between treated and placebo groups, Cognivia developed Placebell®: placebo response ML that predicts each patient’s placebo responsiveness. Clinical trial statisticians can use this predicted score as a covariate in the analysis, the same way they would with a covariate representing age, gender, and other relevant factors.

For the Placebell® ML to work, Cognivia needed a way to identify and characterize traits related to the placebo response. The team leveraged decades of experience in clinical development and an existing wealth of literature describing the connection between patient psychology, expectations, and the placebo response to develop the Multi-Dimensional Participant Questionnaire (MPsQ). 

Development & Validation 

The MPsQ included (but was not limited to) questions that evaluated specific traits that were suggested to be relevant to the placebo response. Then, it was administered to patients in early clinical studies investigating the placebo response in peripheral neuropathic pain.  

Cognivia’s proprietary machine learning-based algorithm then evaluated the relationship between each patient’s questionnaire and their reported placebo response to determine the relevant features and relative weights. From there, Cognivia revised the MPsQ to include only the most pertinent placebo-related psychological traits and characteristics. 

As part of the validation process, the MPsQ was administered to a subset of N=180 patients repeatedly. Some patients even completed the questionnaire multiple times for two years after the initial administration. 

Based on this exercise, the MPsQ’s Cronbach’s alpha is greater than 80%, indicating that this battery is highly repeatable over time. This suggests that the MPsQ measures stable facets of personality, not emotional statuses that can vary from day-to-day. Patient conduct in several pain clinical trials also confirmed a commonality in individual personality features that predict placebo response.

Results

The MPsQ has been administered to thousands of patients across the world in clinical trials in a variety of indications, including: 

• Pain
• Osteoarthritis
• Parkinson’s disease
• Dry eyes
• Allergy & immunology
• Alzheimer disease 
• …

The MPsQ has been translated into more than 40 languages with full cultural and linguistic validation. 

By the end of 2024, we anticipate that the MPsQ will have been administered to nearly 15,000 patients. 

Conclusion

Too often, good compounds are discontinued due to insufficient efficacy in the treated group over the placebo group. Placebell® is a powerful method that combines baseline patient data – collected via the MPsQ – with a disease-specific AI model, producing a score that can be used as a baseline covariate in the statistical analysis, in alignment with regulatory requirements. With this score, clinical trial teams can “see through” the noise placebo response causes and make more confident decisions about the future of a compound. 

The MPsQ has proven reliable for over a decade, effectively characterizing and quantifying patient traits and characteristics related to the placebo response. As a result, Placebell® has proven to reduce data variability by as much as 30% across many indications and reduce false negatives (type II error) by 60%.  

Contact us today to tell us about your study needs.