Josef Bartels spent a year at the CTSI as a medical year-out student, where he studied the role of silence in doctor-patient interactions and completed his MPH. He recently spoke with CTSI Stories about his year.
Thanks for taking the time to chat, Josef. How did you come to apply for the Year-Out program?
It was early third year of medical school and I had an elective rotation in palliative care with Dr. Ronald Epstein. I really enjoyed what I did there because it struck a chord with my humanities background. In palliative care, you’re often at the point where there’s not much more you can do to cure. Instead, you’re dealing in issues that people have written about and physicians have been witness to for centuries – like the meaning of life, the meaning of suffering, and how to die with dignity. The lack of solid answers and the human messiness of it all really resonated within me.
I also realized there was something going on in those conversations that I wasn’t hearing as often in other medical settings. I have a musical background, and realized that there was a lot going on with tone and silence. It all reminded me of a Radiolab podcast I heard called The Musical Language, where they were talking about how mothers all over the world talk to their babies in the same notes – that if you take out the words and play it on a piano, it sounds similar even when from widely different linguistic lineages.
I wondered if there was also some musicality within medical conversations. Are there common patterns? Is silence meaningful? I told Dr. Epstein I wanted to study this, and he said he didn’t think I could pull off what I was considering in my spare time. So that’s when I applied for the Year Out. Dr. Epstein agreed to be my mentor and helped me put together a research team with fellow year out scholar Rachel Rodenbach and Eastman School of Music professor of Voice Katherine Ciesinski.
Tell me more about these conversational moments that you mentioned.
I didn’t set out to study specific moments. I wanted to focus on the rhythm of medical conversations, and specifically the silence in these conversations. There has been a lot written about silence in psychology, linguistics, and neuroscience — so it was a wonderful exploration of a number of different fields in order to focus in on this one phenomenon that happens to be present in medical communication. When I started hearing different types of silences, there was one specific type that was fascinating, and that seemed to occur during these particular moments where there is shared meaning without words. These moments exist in medicine everywhere, but I think they’re more common in situations with mortal consequences and an opportunity for shared decision making, which includes almost every specialty.
So how did you design your study on these silences?
There was another type of silence that popped up a lot, and it was always associated with exam-room computer usage. There has been a lot written on computers and electronic health records, including some work that demonstrates a lot more silence now that there are computers in the exam room. That extra silence, the particular flavor associated with computing, is controversial. Does it give the patient the chance to ask questions? Does it indicate imperfect multitasking?
I set out to measure how long these silences were and how many there are, and I correlated those with counts of specific physician communication tasks. The results are intriguing and suggestive, though the confidence intervals are wide. If I had designed it from the beginning, I would’ve tried hard to find an outcome variable with a normal distribution. Either way it was valuable experience to learn statistics and practice quantitative analyses and manage a large data set, and the results were recently accepted for oral presentation at the International Conference on Communication in Healthcare.
I see. And the other part?
We’re calling these seemingly meaningful pauses in conversation “connectional silences,” because they seem to represent some kind of connection between the physician and the patient – or sometimes between a caregiver and the physician.
How did you define a “connectional silence” for the study? How did you know what to include?
This analysis relied heavily on grounded qualitative methods, which allowed us to develop a rigorous way of identifying a previously unstudied phenomenon. We developed inclusion and exclusion criteria, and then measured interrater reliability between three people identifying these silences to validate these criteria. For example, one of the inclusion criterion is there’s some emotion expressed by the patient which the physician recognized and then acknowledged. Then we ran it by a panel of physicians for additional validation that we weren’t just making these differences up.
For the analysis, I worked with Katherine Ciesinski to define these silences using musical notation, because it is designed to capture the data we were interested in; volume, pitch, rhythm, and tone. When you listen to these moments, they just sound special – they were qualitatively different, and we wanted to know why. We had to define what, exactly, was it that we were hearing in these silences? What is the sonic environment within which these pauses exist? We looked at pitch, volume, intensity, and then we looked at rhythm.
Rhythm was where patterns leapt out and connections became clear. Once we visualized the audio environment of the conversation, we could see that as the “connectional” moments approached, the length of time that each person spoke for began to merge. One person would talk for 5 seconds, then the other would talk for 5 seconds, then 2 and 2, then just a word each. The doctor and patient would match the length of their sentences. Other literature which focused on successful business deals and even successful courting showed similar rhythm in successful conversation patterns – they called it mirroring. Contrast that with a moment where the physician is providing a lot of info, giving a monologue for a minute or two, and then the patient just replies “OK, got it” – and not to put a higher value on one or the other, but they are qualitatively different, and we’ve been able to show that through a new window called dialog rhythm.
We think these patterns might be linked to mirror neurons, which are a hot topic these days. These neurons might light up when you really listen to someone, so a part of your motor cortex actually models what you are listening to, as if you’re moving your mouth and your own tongue. When you begin to respond, you match pitch, volume, and rhythm, because your motor neurons have just practiced it. It that sense, we might be getting close to another channel that can measure engagement, in addition to body position and eye contact.
Where do you go from here? Can you make a career studying silence in medicine?
The short answer is yes. Patient centered communication and a therapeutic relationship are gaining empiric support daily. From issues of medication adherence to physician burnout, we are recognizing that communication and the connection between people is the foundation of medicine, and leads to measurable and meaningful gains in health outcomes. Federal funding through NIH, PCORI, AHRQ and others are interested in exploring how the human experience affects and interacts with more “objective” data like BMI and blood pressure. It’s a relief to me that after a century of trying to control for the placebo effect, the mechanisms and pathways of this and other psychosocial effects of giving care and attention to a fellow human being are getting some attention. Medicine is going to be relationship-centered, because quality communication and caring relationships promote health and wellbeing, and have the potential to act before, during, and after a patient develops disease. Our research took us one step closer to measuring the quality of engagement between patient and doctor, a connection that can then be linked to any number of health outcomes.
In terms of next steps, I’d really like to see if a computer can pick out the silences that are associated with computer use, in order to give a non-invasive and simple measure of computer-centered vs patient-centered care. Then we could measure patient outcomes that might lead to computer redesign or the realization that it might be more effective to pay the physician (or a data expert) to manage the patient’s data separately from interacting with the patient. Electronic medical records have gone far beyond the charting of diagnostic rationale and are really entering the realm of big data. The task of forming a trusting relationship to facilitate preventive, diagnostic, and curative procedures is complex, and managing big data simultaneously may be dangerous in as many ways as it’s helpful. We’re really struggling to measure things like this that affect nearly every medical interaction every day. This is what motivates me to continue research in medical communication.