The Three B’s That Complicate UX Research In Hospitals

We’ve been thinking much about user research in hospital settings recently. Our new company, smart Helios, a digital health development firm, is a spin-off of Helios Kliniken, Europe’s largest private hospital chain. To inform our lean software development, we thoroughly embrace iterative empathetic observation and end-user interviews at each step of a development cycle (i.e., ideation, prototyping, and testing).

We learned that qualitative research in a hospital setting brings its challenges. We think it is worth considering them, especially those we discuss here, called the three B’s: Biases, Barriers, and Background.

Here’s an overview:

1. Biases
   Psychological mechanisms can affect our patients’ thinking and hence diminish the results of our findings.
2. Barriers to trust
   Patients rarely share intimate needs with a non-medical interviewer easily. This can create blind spots in our research.
3. Background
   Internal and external factors such as wealth, socio-economic status, sanitation, education, and access to healthcare can influence our health as well as the care patients receive. The quality of care also depends on the hospital’s infrastructure and staff experience with a particular disease or procedure. These differences make it challenging to generalize findings.

We also discuss remedies that can help overcome these challenges and distill more valuable insights. They include:

• Conducting a thorough and well-prepared interview.
• Including healthcare providers in the process.
• Drawing on quantitative data to guide some of the qualitative user research.

Biases: We Are Fantastic At Lying To Ourselves, Whether Or Not We’re Patients

Psychologists call them cognitive biases: they negatively affect how accurately we perceive and remember events or feelings, and we possess impressive amounts of them.

For example, people remember information better if it is more recent or salient. If you ask a patient about the initial appointment with her oncologist (this is rarely good news), don’t be surprised if she remembers only a quarter of it. Hospitalized patients are typically overwhelmed by the unfamiliar situation they’re in and often under stress and that influences their memory.

Hence it makes a difference when you ask them in their patient journey. Even in one day, patients face different problems that affect what’s top of mind at the moment you observe or interview them:

• In the morning, when the painkillers have worn off during sleep, regaining control over physical pain is all that matters.
• During the day, worries about upcoming procedures might dominate their thoughts or they could be focused on their hunger while they’re not allowed to eat or drink ahead of a diagnostic test.
• In the evening, they might be afraid that they won’t be able to update their relatives appropriately.
• At night, some struggle to sleep because of the hospital noise or their worries.

Takeaways:

• Try interviewing users at different times of the day and different moments of their journey and take note of how findings vary.
• Always get an orientation about where your users stand within their patient journey. Explore what happened in the previous hours or days and what diagnostics or treatments lie ahead of them.
• Beware that our psyches possess a plethora of mechanisms to limit rationality and prevent past events from entering our conscious mind. You cannot control them all, but it improves your research if you notice them.

Barriers To Trust: Who Is Asking Matters, Too

It’s not just about how and when; it also matters who is asking. Even if a patient agrees to speak with us, what she shares will highly depend on how much she trusts us. One of us observed as a clinician how often his patients need to build up trust, sometimes over days until they ‘confess’ certain concerns. That’s especially true when problems have a psychological component (e.g., sleeping disorders) or are stigmatized (e.g., certain infectious diseases).

The more knowledgeable you are about health problems, the better you can steer interviews towards relevant issues. If you do this empathetically, your interviewees might find it easier to speak about them. Don’t get frustrated, however, if they don’t. Some people need a lot of trust, and there’s rarely a shortcut to earning it. In these cases, there’s something else you can do: include subjects who are not your target users but still have crucial insights in your interviews.

Take the nurse, for example. She might know from her previous night shift how many patients had trouble sleeping and who might agree to talk about it. The doctors will know which crucial questions they get asked frequently. And the housekeeping staff can share stories about the patients’ hygiene concerns. Listen closely to them: many of the staffs’ pain points likely hint to patients’ pain points, too. The more observers you allow to shed light on one subject, the better your chances to understand your patients’ experiences and the broader your perception of the system will be.

Takeaways:

• Try to interview people involved in your users’ care.
• Ask the clinical staff for guidance on which patients to ask about specific problems.
• Even if patients are your primary users, make sure to ask health care providers, such as doctors, nurses, or therapists about common patient needs.
• Schedule repeated interviews with patients if possible to build the trust necessary for sharing critical concerns.

Background: Mind The Worlds In And Around The Patients

Different patients have different needs. This is obvious, and part of the reason why UX researchers develop personas, conduct semi-structured interviews, focus group discussions, and observe subjects to explore the multiple realities of our participants. But there’s still a problem we can’t dismiss. Even within our hospitals inside Germany, people’s realities can differ greatly depending on their income, education, insurance, place of residence, etc.

What you discover in one hospital or region might not apply elsewhere. This can be a problem if you want to deploy your products at scale. If you have the opportunity, you should go the extra mile to conduct UX research in different hospitals to understand the needs and what drives adoption.

If some regions show poor sales, conduct field research in these regions to revisit your personas. In fact, don’t just challenge the personas, also explore the environment.

Here’s an example of why that matters: We developed a tool that relies on data from a hospital information system. This worked well in one clinic were staff was spread over different parts of the building, turning digital communication into an effective medium. In another hospital, however, the relevant people sat in the same room, making face-to-face communication significantly better than typing information into electronic records.

Takeaways:

• Go beyond personas or archetypes and seek to understand the different realities between hospitals, wards, and regions.
• Develop and constantly improve a rollout-playbook that lists local challenges and how you solved them to inform future expansions.

Sorry, Ignorance Is Not Bliss

Some UX researchers seem to believe that it is beneficial to enter an interview unprepared to avoid bias. Some shy away from acquiring relevant medical knowledge, thinking that (since they are not trained medical professionals) they won’t grasp the concepts anyway. Some feel that understanding the medical context of their interviewees’ situation will not add value to their research since they solely focus on the subjective experience of the disease.

But in evidence-based healthcare, conducting research on patients without previous peer-reviewed literature and guideline research is not only unprofessional but often even considered unethical. We should not fall prey to the illusion that ignorance frees us from bias.

The good news is that in healthcare, we are privileged to have a large body of well-conducted studies and systematic reviews available online, many of them free to access. PubMed is an excellent and open source, tutorials on how to use are abundantly available online (we think this is a good primer). Or if you have the budget, paid sites like UpToDate provide comprehensive disease reviews written both for professionals and for laypeople.

We know that UX researchers who are rightfully focused on ‘getting out of the building’ might not enjoy spending many hours on literature research but we are convinced that this will help you form better hypotheses and questions.

Moreover, if you start with clearly predefined research questions and seek answers in the scientific medical literature, you might save time and discover questions that you wouldn’t have thought of. For example, it is advised that individuals undergoing hip surgery should practice using crutches before the operation because it is already difficult enough even without the postoperative pain and swelling. This knowledge, obtained from literature research, could help move from more open questions, such as:

• “How do patients prepare for a hip replacement surgery?” or
• “What perceived needs to patients have ahead of hip surgery?”

To more concrete questions such as:

• “What are the most important preparatory measures that many patients are currently unaware of?”

Takeaway:

• Do a systematic literature review to inform your research.

Let The Data Guide You And You Guide The Data

To take this further, we’re also developing methods to use quantitative analytics and Deep Learning to guide our qualitative research. Our machine learning engineer just deployed AI to crawl the web for colon cancer blogs to identify hot topics that remained unmentioned in qualitative reviews. We defined “hot” as having many views, many comments, and many likes.

Or you can uncover semantic structures (see picture). These findings can then guide the UX researchers. Similarly, qualitative research can yield hypotheses that we can try to validate with passively collected data. For example, if you think that sleeping problems are common, you could (user consent provided) use your app to measure phone use at night as a proxy for sleeplessness.

Conclusion

Many of our suggestions are not new to well-trained UX researchers. We are aware of that. But in our experience, it is worth stressing the importance of mindfulness towards the three Bs: Biases, Barriers to trust, and Background. Here’s a summary of some of the recommendations to overcome the 3 Bs:

• Prepare interviews with literature research on the topic (e.g., on Pubmed.gov).
• Ask doctors which patients are suitable for interview.
• Include those who care for your users, including nurses, therapists, and relatives.
• Cooperate with the data scientists or web analysts in your team, if you have them.
• Understand that it takes users time to build trust to tell you about some needs.
• Explore how realities differ not only by personas, but also by regions and hospitals.
• Stay aware that, no matter how much you try, the influence of the 3Bs can only be reduced, and not entirely removed.

We wish you well and thank you for making the world a healthier place.

The authors would like to thank their former colleague, Tim Leinert, for his thoughtful input to this piece.

(cc, ra, yk, il)