Researchers are aware the inner workings of a patient participant’s mind can impact the placebo response in clinical trials. What they tend to be less cognizant of, however, is exactly how that important relationship works in study settings—and that lack of understanding can have a negative effect on drug development efforts.
Clinical insights firm Tools4Patient has come up with its Placebell method, offering researchers a way to evaluate components of patient psychology that are critical to the placebo response, using machine learning to reduce the impact. Erica Smith, vice president of business development for Tools4Patient, spoke to Outsourcing-Pharma about the finer points of the placebo-personality relationship, and how the Placebell technology stands to help bridge the gap in understanding.
OSP: Could you please explain how psychological factors get overlooked in clinical research?
ES: Clinical research has historically focused on the measurement of efficacy endpoints, which can be either objective or subjective. Objective endpoints are easily measured and quantified – vital signs, serum biomarkers, imaging, etc.
As objective endpoints are not readily available in all diseases, the industry has developed validated scales to assess subjective endpoints, such as pain scores, the MADRS scale for depression, or quality of life indices. Except in psychiatric indications (in which psychological factors are important), these have not historically been considered as relevant to efficacy evaluation – in part, because they are subjective and thus more complex to measure.
This scientific relationship between health and personality is only recently emerging and is dedicated to the evaluation of the influence of the personality on long-term disease progression and survival. Until now, this was not considered a critically important patient characteristic when conducting clinical trials.
OSP: Could you share how a patient’s psychological/emotional response might be impacted by their health, and vice versa?
ES: There is a strong connection between an individual’s personality/emotional status and biology, disease, and mortality that is intuitive for neuropsychiatric diseases like depression, but it also exists for many other indications. First, it is important to make the distinction between personality traits (a person’s psychological makeup, which generally remains stable over time) and emotional status (how one feels in the moment, which can be highly variable daily, weekly, monthly, etc.). As personality traits are both easier to measure and easier to associate with long-term health status, these have been the most widely studied.
The well-known relationship between personality and biology is best exemplified by the connection between neuroticism and stress. Individuals with high neuroticism (e.g., propensity to experience negative mood states) have been strongly correlated with stress, which, in turn, is well understood to cause increased cortisol levels (among other biological consequences). It is not surprising that neuroticism, along with conscientiousness and extraversion, are all related to risk of death from heart attack or stroke. This relationship can vary between diseases as, for example, no relationship was found between personality traits and cancer incidence or cancer mortality in a large cohort of more than 42,000 men and women.
The importance of individual personality extends even to mortality risk, as meaningful correlations have been noted with several personality domains. For example, studies have associated high conscientiousness with a lower risk of mortality), while high neuroticism has been associated with a risk of increased mortality. In fewer studies, both openness and agreeableness) have been associated with decreased risk, and extraversion has been associated with increased risk of death.
In addition to the relationship between personality and health, disease, and mortality, it is important to understand that personality is also associated with health behaviors, including things like seeking medical attention when symptoms arise, following doctor’s instructions, and taking medication as prescribed. As one very important and contemporary example, an individual’s personality traits have been demonstrated to influence the likelihood of receiving a COVID-19 vaccine. A recent study of more than 3,000 individuals found distinct relationships between personality and vaccine acceptance, hesitancy, or resistance in the UK and Ireland.
Of course, health and disease characteristics may in turn also influence both emotional status and personality traits, although emotional status is more malleable and thus more easily influenced. Disease status impacts the patient’s emotions, but emotional status also affects the disease. This, in fact, makes studying the relationship between emotional status and disease complex, as it is a classic case of “which comes first, the chicken or the egg?”
OSP: Can you give some examples of how a psychological response by a patient might significantly impact trial results?
ES: There are several ways in which a participant’s psychology can influence clinical trial results. First, a patient’s decision to participate in the trial in the first place may likely be influenced by personality, and in turn, specific motivations for participating will influence both behavior during the trial and the resulting data. It is easy to imagine that data derived from patients motivated primarily by financial compensation may differ from that derived from patients motivated by access to better healthcare or new, investigational therapies.
Next, patient psychology will influence their compliance throughout the trial – including whether a patient will complete the entire trial, will comply with all the trial procedures (e.g., site visits, recording of symptoms in e-diaries, etc.), and whether they will be adherent to the prescribed medical regimen.
In addition to these factors, patient psychology will further impact both their response to placebo and response to the active drug. The relationship between placebo response and psychology has been studied for decades, and personality traits like optimism along with other factors such as expectation for improvement are known to relate to the placebo response of an individual patient.
The influence of patient personality on placebo response is far more complex than these simple associations, however. Clinical trials are complex and demanding, and patients experience many procedures and interactions, either on a physical clinical site or virtually while their measured outcomes are further influenced by patient perceptions of all these cues throughout the trial experience. These complex relationships are again influenced by individual psychological traits.
OSP: What are some of the dangers of leaving out psychological or personality-related data in studies?
ES: As mentioned above, clinical trials are complex – and drug efficacy and safety are measured in very diverse patient populations comprised of individuals with unique biological and psychological makeups. The heterogeneity of clinical trial patient populations can lead to dramatic variability – or noise – in the data, which makes it more difficult to detect the efficacy or safety signal.
Clinical trial data analysis, however, has traditionally focused only on measuring and accounting for biological differences between patients. A simple example of this is age – patients of varying ages may participate in clinical trials, and trial statisticians can adjust for age when conducting analyses, thus minimizing the impact of this specific type of patient heterogeneity on the ability to detect efficacy.
As stated above, patient heterogeneity in personality has not historically been considered, allowing this source of noise to potentially obscure important differences in data. Put simply, the better the clinical trial team understands, quantifies, and accounts for all differences in their patient population – including psychological differences – the better data analysis they can conduct.
Failing to consider patient psychology when analyzing trial data leaves a major source of data variability unchecked and can increase the risk of clinical trial failure even if the experimental drug is truly efficacious.
OSP: How can trial teams better take such patient responses/perceptions into account?
ES: The best tool to date to measure an individual’s personality or psychological traits is the questionnaire, and several questionnaires have been developed to evaluate specific components of personality, psychology, or emotion for a variety of purposes.
For information on psychology, perceptions or beliefs to be used in clinical trials, two criteria must be met:
- the relevance of the information being collected on the outcome data from the trial must be demonstrated, ideally as a causative and not just associative relationship
- the information must be mathematically integrated into the data analysis in some way.
This may require the development of new, specially designed questionnaires as well as advanced data analytical methods to best process and utilize these data. Finally, this information has to be taken into account in the (statistical) data analysis without increasing the risk of a type 1 error, or false positives (e.g., inflation of efficacy).
OSP: Could you please tell us about the “Big Five” approach? How can that be put to use in clinical research?
ES: The “Big Five” Model – or Five Factor Model – is the most well-known and scientifically accepted theory of personality, which states that each individual can be quantified in five key dimensions: openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism.
These personality traits are measured by a standard battery in which participants rate the accuracy of simple statements relative to their personality. The Big Five have been demonstrated to be predictive of factors ranging from financial behavior to career success to overall health and wellness.
While the Big Five approach provides a reasonable tool to better understand patient personality in clinical trials and drug development, it is only a starting point. First, personality is necessary but not sufficient to understand and predict patient response to treatment and behavior in clinical trials. Additional information on expectation of improvement in the trial, perceptions of the drug, trial and experience, and beliefs also need to be considered.
Second, some components of the Big Five are more important than others. Ideally, all components of the Big Five battery would be included in any evaluation, but this would simply require too much time.
The Big Five components (openness, conscientiousness, extraversion, agreeableness, and neuroticism) – also called domains – are themselves broken down furthers into facets. For example, neuroticism (sometimes referred to as “emotional stability”) can be further divided into the following facets: anxiety, angry hostility, depression, self-consciousness, impulsiveness, and vulnerability.
Just as some Big Five domains may be more strongly related to a specific outcome or behavior than others, some facets may be more meaningful than others. Focusing on the most impactful domains and facets is critical strategy to minimize patient burden and evaluation time in clinical trials.
OSP: A patient’s feelings and perceptions might be difficult to capture accurately, since they typically are considered subjective (vs. a pulse, blood pressure, thyroid level, which are a definite number). Is it possible to make such recorded feedback more accurate?
ES: Any tool or questionnaire used to evaluate personality in clinical trial patients needs to meet several criteria, including being robust, stable and sensitive – above being shown to be highly relevant to the response or behavior that it is being used to understand or predict (as described above).
As personality traits are relatively stable over an individual’s lifetime, the measurement of personality should also reflect this stability with high reproducibility. The tool also needs to be adequately sensitive to detect subtle differences between individuals. Both objectives can be met by including several questions in a questionnaire for each domain and each facet, noting again that it is a delicate balance between questionnaire robustness and evaluation time/patient burden.
Just as it is important to focus on the most important domains and most important facets, it is also prudent to select the most informative questions and disregard those that bring less information or contribute less to the overall questionnaire stability.
OSP: Please share how the placebo response and a patient’s personality are interrelated, and how that relationship might impact the quality/accuracy of data captured in research.
ES: The placebo response is a complex phenomenon that has been investigated for decades, with tremendous consequences on drug development. Recent research has demonstrated that the placebo response can account for more than 50% of the measured treatment response across diseases). High placebo response rates – or even high variability in placebo response between clinical trial participants – have caused significant numbers of Phase II and Phase III clinical trial failures.
There have been many studies that establish the relationship between patient personality and placebo response, including research focusing on biological changes in patients’ brains after placebo administration. For example, in pain, agreeableness and neuroticism facets have been associated with the endogenous opioid system activation.
Imaging techniques like fMRI have further demonstrated the connection between psychological factors and biological component of the placebo response. Furthermore, patient expectation for improvement contributes significantly to placebo response, and it has been demonstrated that this may coexist with emotional feeling about treatments.
The placebo response is a significant issue that leads to increased trial enrollment and increased risk of trial failure, thus increasing drug development cost and timeline. While there are several methods that have been used to manage the placebo response (e.g., training clinical sites or patients to improve symptom resorting), these have had a modest impact and, in 2021, the placebo response remains a significant challenge. New approaches are sorely needed, and methods that incorporate an understanding of patient psychology offer the best opportunity to better understand and manage this challenge.
OSP: Could you please tell us a bit about Placebell—how it works, and how its use can help researchers get a better handle on patient personality impacts?
ES: Placebell provides drug developers with a proven method to evaluate key components of patient personality and, when combined with a predictive algorithm powered by machine learning, integrate this information into data analyses to increase clinical trial power and reduce the risk of trial failure. The method uses a well-defined tool, the Multi-dimensional Psychological Questionnaire (MPsQ), to quantify patient psychology, expectations, perceptions, and beliefs. These data are used as inputs to a machine learning algorithm that calculates a single score for each patient related to their predicted placebo responsiveness.
When included in statistical analyses, this score reduces data variability related to the inherent differences between clinical trial patients, increases clinical trial power, improves p-value, and improves clinical trial success rates – all with no operational or mathematical risk and with a defined regulatory path with the US Food and Drug Administration (FDA) and European Medicines Agency (EMA).
OSP: Anything to add?
ES: Placebell represents a major step forward as it offers, for the first time, drug developers and KOLs an easy, no-risk way to integrate an understanding of patient psychology into data analysis from small to large, multi-center, industry-sponsored, and academic clinical trials – but it truly is only the beginning. Placebell provides proof-of-principle that patient personality and surrounding environmental influences can be measured with adequate sensitivity and robustness and used effectively when analyzing data to improve clinical trial assay sensitivity.
The ability to reduce the “noise” in the data that arises from the patient-to-patient differences in placebo response allows better discrimination of the efficacy signal. This approach is now being applied to address other critical sources of noise to further improve the evaluation of safety and efficacy in clinical trials.