How can behavioral science help make the world a healthier place? Last month, University College London held their 7th annual Centre for Behavior Change conference to shed light and open conversations around this question. The conference theme, “Enabling Behavior Change to Build Back Better—for health and sustainability,” could not have been timelier, considering how much of an impact human behavior has had on individual and population health during the pandemic.
Behavioral science—the study of human action—continues growing as a valuable piece of the puzzle in improving health outcomes. From COVID-19 vaccination to perinatal health, research presented throughout the conference showed how the behavioral science lens can be applied to improve problem-solving in healthcare. Models such as the Theoretical Domains Framework and COM-B (the most frequently mentioned) can be utilized to ensure audience research captures holistic insight into health behaviors and their drivers. The frameworks help ideate and map the most relevant, evidence-backed solutions to insights gained. Evaluating the impact of these solutions can go beyond standard A/B testing to focus more specifically and rigorously on actual behaviors changed.
The conference presentations showed how a growing, diverse array of theories and methodologies can be practically applied across healthcare to change behavior and ultimately improve outcomes. Read on for our biggest takeaways from the health-focused presentations across the 3-day conference.
Behavioral science is not the only piece of the puzzle—it is also taking pages from the human-centered design playbook, and vice versa.
Behavioral science is not a silver bullet. Behavior change is hard, and rarely is this the only lens applied to a problem. One synergistic lens that appeared across numerous presentations was human-centered design (HCD)—a co-creative, iterative approach for designing solutions in a manner that focuses relentlessly on users. The empathetic, agile principles of HCD are utilized in behavioral science projects in such ways as:
- Co-creation, where the target audience is involved in ideation and development of behavioral interventions—not in research, but truly as a part of non-hierarchical decision making, with the appropriate training to participate to their fullest potential
- Utilizing personas and journeys to ensure solutions are based on the customer’s experience, and using journey structures in research to enable holistic insight
- Incorporating design elements such as ease of use in measurement, utilizing frameworks including the Technology Acceptance Model and (u)MARS (User Version of the Mobile Application Rating Scale)
- Utilizing ethnographic research approaches to understand the audience’s experiences and contexts more deeply
Evidence-based behavioral science principles are also utilized across the HCD process, for example:
- Conducting literature reviews around design challenges to understand behavioral insights and behavioral interventions that have previously been studied
- Utilizing quantitative research methodologies during discovery and user testing
- Coding results of discovery research based on behavioral models
- Infusing behavioral models in co-design sessions, for example defining key barriers/drivers and ideating solutions within each element of the COM-B framework.
To enable the above, the behavioral science community is focused on making behavioral theory more accessible to practitioners who are not experts in the subject, such as designers. For example, the Behavioral Lenses tool was created at Hogeschool Utrecht to equip designers to seamlessly leverage behavioral theory.
Our take: Those developing impactful interventions are increasingly leveraging a multi-disciplinary, collaborative approach across all phases of planning and implementation. This is an exciting direction for behavioral science, as interventions should be more likely to succeed from the start due to stakeholder input and the built-in advocates this creates. We feel that behavioral science and HCD are two sides of the same coin, both enabling a deeper understanding of audiences and thus more relevant solutions—and these “puzzle pieces” can be complemented by other disciplines such as marketing and data science.
Like most industries, behavioral science interventions and research methodologies are increasingly digital.
Health technologies are exploding, and behavioral scientists are working to improve the impact of digital tools by applying behavioral theory and measurement. For example, Gro Health utilized the COM-B model to enhance their chronic disease management offering, resulting in features such as cultural norm support, goal-based tailoring, and integration into platforms such as Alexa. Personalization is critical, and increasingly feasible, in the digital space. For example, content can be tailored by a user’s preferred learning modality and health goals. However, this requires investment in infrastructure, such as a content tagging system.
Digital engagement offers a wide range of measures that can be used to understand a solution’s impact—cognitive, affective, behavioral, and other measures. Health outcome measures can also be included, whether through self-report (eg mental health questions) or objective measurement (eg vitals), and then correlated with engagement metrics to understand which digital behaviors most impact health. The appropriate measures to use can vary based on where a solution is in its development—for example, eye tracking at the pilot stage, vs retention as a program is up and running for the long term.
The abundance of health data, at least partially driven by the growth of digital solutions, has spurred the need for more advanced analytics approaches such as machine learning. Behavioral scientists are working withmachines to identify trends in audience insights and utilizing tools such as natural language processing.
Our take: It is natural that this industry-wide digital transformation is reflected in behavioral science—it offers opportunities to enhance the impact of interventions in the real world, from personalization to advanced analytics. It is important, however, to maintain the “human” element—for example, digital interventions can be more impactful if complemented by human interaction, and humans (such as pharmaceutical reps) often have unique customer insight to leverage.
Partially enabled by digital, leaders in the field are beginning to advocate for a greater focus on behavioral processes and adaption over discrete, singular changes.
In his keynote, UC San Diego professor Eric Hekler proposed that, “To save ourselves and our earth, let’s move from thinking in terms of things (behavior change) to processes (tuning).” This demands that we think about not only a desired future behavior, but how that behavior adjusts, how a person responds, what they then sense and predict, and then how they again adjust their behavior—and so on. It requires a focus on “states” of being (eg vulnerability), and a mindset of helping individuals in a specific context achieve more good states and less bad states, over time. We should think about not only how behavioral adaptation is “outsourced,” but also how it can be “insourced” by the individual building knowledge, skills, and practices—and what feedback loops are and should be created in this process.
Studying the behavior change journey well requires a shift in the way researchers predominantly approach their trials. A range of approaches were raised, such as micro-randomized control trials (RCTs) where an individual’s decision points are randomized rather than individuals themselves, and experiments in a box—short, self-guided behavioral simulations. Advancements in data science will enable more adaptive and include techniques such as control systems engineering, predictive modeling, adaptation algorithms, and other approaches in studies that are focused on this idea of “tuning”.
Our take: This focus on behavioral processes is not yet commonplace and increases technical requirements. However, it is a compelling approach that enables researchers to get closer to the way behaviors work in the real world, especially with digital health tools. Those of us applying behavioral science can take learnings and begin to dip our toes in, such as setting behavioral goals that are process-focused and using research methodologies such as micro-RCTs.
As interest in and application of behavioral science grows, a need for common language and shared resources is top of mind.
Good behavioral research provides insight into what works (intervention, MOA), compared to what, for changing different behaviors (outcome, value), who it works for, how well it works and for how long, in what setting (context). A major challenge, however, is the inconsistency in the way researchers describe and report on these elements, making it difficult for humans and, increasingly machines, to easily find, analyze, and keep up with the vast information on a chosen topic.
The behavioral science community has recognized this and is investing in developing a standardized lexicon and guidelines. The Human Behavior Change Project, for example, is building on previous efforts such as the Theory and Techniques tool by continually collecting behavior change intervention reports, creating an ontology for common definitions and relationships, and providing other resources such as a paper authoring tool to ensure publications include all necessary information. These initiatives are ongoing, require significant time and skills, from behavioral to computer science, and are often open source.
Researchers are also a focus on the future of behavioral research—the International Behavioral Trials Network has prioritized topics to focus on—specifying research components, developing novel research designs (eg just-in-time adaptive intervention designs), disseminating trial findings, and ensuring interventions are tailored and implementable.
Our take: The focus on standardization and consolidation is much needed to enhance learning from past efforts, avoiding duplication and equipping broader stakeholders to leverage research. In doing this, it is important that developers of these tools consider not only scientific robustness on the back end, but also user experience on the front-end. HCD processes, and even use of behavioral science models, are valuable to enable this.
Behavioral science has a growing place outside the lab, which can require tradeoffs between academic rigor and real-world practicality.
Health technology developers are increasingly leveraging behavioral science to achieve clinical outcomes, taking insights from the evidence base around human-delivered interventions and understanding how they apply digitally. With time and resource pressures, and the realities of the real-world context, some differences from the lab setting are noteworthy:
- You may not have a captive RCT audience, so a user may not be as willing to engage with the tool
- The product might change during a study as other functions may not be able to wait to release new features
- Users are unlikely to complete long questionnaires—shorter, plain-language sets of questions may be required to get the necessary insight
- Technology engagement is cyclical—in many cases, the goal may not be to have someone engaged consistently over a long period of time; rather, “good” might be getting people to come back to a tool when they begin to struggle again
- Decisions may need to be based on meaningful signals on a minimum viable product rather than waiting for robust evidence around clinical outcomes
- Commercial populations are broad and diverse—this can be seen as an opportunity to understand how a solution works for various groups and tailor it accordingly
- Different sets of stakeholders, such as such as designers and developers, need to be engaged in effectively implementing behavioral science principles in health tech solutions. This requires clear communication around the desired user experience, and concrete, example-filled education around behavioral science principles.
In applying behavioral science in the real-world, a focus on diversity, equity, and inclusion is increasingly top-of-mind. This includes everything from ensuring multi-lingual resources to enabling allyship, uncovering where behavioral drivers and barriers are causing disparities, and prioritizing work to address these.
Our take: Healthcare companies, from pharma to health tech to health systems, are investing in behavioral science initiatives and resources, prioritizing the development of dedicated teams and/or individuals to make a stronger impact on health outcomes. Trade-offs are unavoidable—a continued partnership between academics and industry will help manage these compromises, deliver the required education and skill-building around behavioral theories, and implement behavioral science in a tailored way that meets organizational goals.
There is a massive focus on how to move from smaller-scale pilots to scalable implementation, often at a global level.
Numerous presentations gave advice on expanding existing solutions to new geographies. For example, Baby Buddy Forward discussed expanding a successful prenatal health education program into Cyprus through country-specific research and analysis to account for unique cultural dynamics.
Another example was the Quality and Acute Stroke Care project—this started as a successful intervention in Australia but is scaling internationally, including across Europe. The team saw the strong clinical impact of getting care teams, especially nurses, to “do the simple things well,” which looked different across countries. They developed country and site-specific champions, provided education and support remotely, utilizing reminders, and incorporated protocol behaviors in clinical audits. Partnership between academia and local industry from the outset was crucial, as was equipping stakeholders in this case nurses, as leaders. By taking this approach, they could consider the strong variance across countries in project perceptions, roles, and communication needs.
Our take: Continuing to publish learnings on scaling initiatives will be critical for creating behavioral interventions that have a strong impact on health outcomes. Local stakeholder engagement from the start is the most foundational requirement for scaling. The digital environment enables us to continue learning and optimizing even after an intervention is launched.
As Professor Susan Michie noted in her opening remarks, “More than ever, there is a recognition that human behavior lies at the heart of the threats to our health and global sustainability … and at the heart of the solutions to these threats.” Anyone working in healthcare, from R&D teams looking to engage patients more effectively in trials, to product developers looking to create tools that drive productive changes in everyday health behaviors, can improve their impact through behavioral science. The UCL conference was an inspiring view into the potential for behavioral science to transform health outcomes, and we are excited to continue partnering across the industry to make this a reality.