How design methods uncover value in unexpected places
Originally published in 2015 on Medium
In 1974, Lawrence Tesler and his colleagues at Xerox Corporation Palo Alto Research Center developed the first text editors to use copy-and-paste commands to move/copy text. In 2015, a small team at Penn Medicine discovered copy-paste could increase patient access to life-saving care while saving the health system hundreds of thousands of dollars.
Though it may seem as though I buried the lede, this story isn’t about copy-paste. Let’s face it, you probably wouldn’t read that, anyway. What made this project special is the way in which a small group of people used simple design tools to quickly and cheaply boil a complex problem down to actionable change.
In 2014, leaders at The Penn Medicine Center for Health Care Innovation conducted a listening tour in which they collected insights and grievances from employees in all corners of the health system. One strong theme which emerged is widespread clinician dissatisfaction with the outpatient Electronic Health Record (EHR). Clinicians consistently expressed that their job felt like it was 80% clerical and 20% clinical. In other words, they were trained to care for patients, but they found themselves spending an inordinate amount of time in front of the computer.
In the 1960’s, Horst Rittel, a German design theorist, coined the phrase ‘wicked problem’ to describe a challenging problem whose solution requires a great number of people to change their mindsets and behavior. While this phrase is typically used to describe large problems like climate change, it could equally be applied to physician dissatisfaction with EHR. After all, Penn Medicine’s outpatient EHR is used by thousands of employees in dozens of departments, and any change to the EHR would have a ripple effect throughout this ecosystem. So, when my colleague Mike Serpa and I were tasked with alleviating provider dissatisfaction with the EHR, we knew we were facing a challenge.
Secondary Research
While I was eager to leave my desk and observe how Penn physicians use the EHR, I took time to do an informal literature review. As someone who was relatively new to EHRs, I needed at least a baseline understanding to be an effective observer and interviewer.
As health care institutions all over the world shift from pen-and-paper to EHRs, researchers are examining the impacts of the transition. As a result, there is ample secondary research available to better understand how EHRs impact health care.
The pain points clinicians experience in using EHRs seem to fall into three major categories: usability, patient-doctor communication (PDC), and time and expense. Epic, Penn Medicine’s outpatient EHR is frequently identified as being the most user-friendly of the major systems. That said, none of the major EHRs are nearly as user-friendly as we’d like them to be. Next, depending on how they are used, EHRs can damage PDC, reducing patient satisfaction and clinical outcomes. Not surprisingly, communication suffers when the clinician is focused on the computer, rather than on the patient. Finally, EHRs can be very expensive insofar as they take up a significant portion of clinicians’ valuable time.
While these pain points are very much inter-connected, our team was encouraged to focus primarily on time and expense. Our goal was to, wherever possible, empower clinicians, primarily physicians, to spend less time using the EHR and more time caring for patients. With a basic understanding of how EHR impacts clinicians, and with an overarching goal in mind, our team set out to observe clinicians in action to see how EHR fits into their workflows using simple contextual inquiry.
Contextual Inquiry
As we began our contextual inquiry, our team was far from transforming this wicked problem into actionable change. That said, we were optimistic that by observing clinicians using the EHR in the context of their everyday workflows, we would come to better understand the problem and find valuable areas on which to focus.
Initially, we cast a very large net, arranging to observe a range of clinicians and settings, from nurses to surgeons, and from geriatrics to neurology. We quickly learned that not only was Epic a sprawling piece of software, it was being used in a myriad of ways. We soon found that the volume of information we were taking in and processing was just too great to effectively plan and carry out focused tests of change.
Our findings were not focused enough to start tests of change
We shifted the goal of our contextual inquiry. Our team began looking for a partner rather than a project. We hoped that by identifying a smaller group of stakeholders, we could do the kind of deep dive we needed to begin tests of change. Before long, we’d established a close relationship with the medical oncology group within the Abramson Cancer Center (ACC).
Medical oncology was in many ways the ideal partner for our project. First, time given back to a medical oncologists is time than can be spent on very sick patients. Second, medical oncologists are heavy EHR users and strongly feel the associated pain points. Third, medical oncology is a high-revenue setting, so clinician time is particularly valuable. Finally, we had a motivated cohort of clinicians to work with as well as strong support from the executive team at the ACC.
Time Study
In preparing for the next phase of our contextual inquiry, our team chose to be more rigorous in collecting data. Given that our focus was time and expense, we prepared an informal time study to see where medical oncologists, the highest-paid members of the care team, were spending their time in the EHR. Armed with a simple spreadsheet, stopwatch, and pen, we spent long hours observing medical oncologists as they went about their day, taking notes on how the EHR was being used.
It wasn’t pretty, but it didn’t have to be. We had the time data we needed.
Keeping up with clinicians as they navigated Epic’s thousands of screens turned out to be trickier than we’d thought. We didn’t end up with publishable time data, but we did have a clear sense of proportionality — which sections of Epic seemed to take up the largest chunks of the clinicians’ time. In fact, there were some obvious time-sucks that emerged, namely the new patient progress note, the place where physicians record information like a new patient’s medical history and the results of new patient encounters.
When we abstracted our time data, it became clear that NPV progress notes are proportionally the largest time-suck
Most medical oncologists see a few new patients each clinic day. They spend slightly more time with these new patients in the exam room than they do with returning patients, but they spend much more time preparing for new patient visits and wrapping things up after the encounter. In fact, they spend significantly more time working before and after seeing the patient than they do working with the patient directly.
To an extent, this wasn’t surprising. The Abramson Cancer Center is known worldwide for treating very sick patients, many of whom come to Penn Medicine with long histories of illness and treatment. Our physicians spend long hours preparing for these encounters by poring over outside medical records and producing a detailed progress note.
Still, this process of preparing for new patient visits was taking up a significant chunk of our physicians’ time, so we did our best to map out the process and look for areas of improvement.
Typically, a physician will begin by picking up dozens of pages of outside records. She will then carefully read over these records, many of which are of very poor quality due to repeated printing and scanning, possibly highlighting and taking notes. Next, she will re-type sections of the outside records into new note in the patient’s chart, often entire paragraphs. Finally, she may further edit the patient’s note, including perhaps a plan for the upcoming encounter.
Repeated faxing and printing render outside records like this one unintelligible
Our team soon began to question aspects of the existing process. Most importantly, we wondered why health systems with EHRs were sending us faxes, only to have physicians read and re-type some of the information verbatim into another EHR. Why couldn’t outside records be transferred digitally, automatically, and as editable text?
Rapid-Cycle Testing
By now our focus was much more narrow. We had a small, motivated group of stakeholders to work with, as well as very specific pain point to focus on. Our next step was to carry out rapid tests of change to try to move the needle. We set out to test a new process for handling outside records that would enable physicians to more efficiently prepare for the new patient visits.
For a few grueling weeks our team began each morning by picking up outside records from the fax machine, retyping every single letter into Word documents, and sending those documents to a small group of clinicians at the ACC. We asked these clinicians to try preparing for their new patient visits using these Word documents to see what if any value this new process would provide.
Our methods were simple, but the results were promising. Clinicians appreciated the increased legibility of Word documents over beaten-up paper records, enjoyed having the records all in one place rather than having to search the EHR, and reported that being able to copy and paste text into the patient’s note was saving them huge chunks of time. In fact, the most effecient clinicians reported time savings of up to 300%.
“I think this is a giant step forward. My team and I have seen a significant reduction in time spent on patient notes in this pilot. If this continues to progress, I can see this increasing patient throughput.” - Medical Oncologist, ACC
While our stakeholders were happy with the change, this process was very tedious for us, not to mention untenable in the long term. In the interest of efficiency, we began scanning the outside records, enhancing them digitally, then using text recognition to produce PDFs with selectable text. Finally, we manually uploaded these PDFs into the EHR. This process was much quicker for us, and our stakeholders still had access to editable text, the records were still more legible, and they were seeing similar time savings.
With another successful iteration behind us, we took a third pass at the process, this time with scalability in mind. We set up a fax line that automatically converted incoming documents into PDF’s with editable text and saved them onto a shared drive. We then manually uploaded these PDF’s into the EHR. The benefits remained the same, but more streamlined process allowed us to handle a much larger volume of outside records and scale up our test.
We are currently in the process of scaling up our work even further by implementing this new process, with a few tweaks, to handle all the outside records at the Abramson Cancer Center. We expect that the staff and clinicians will soon be independently using the new process. In the meantime, news of our work has spread, and we now demand for a step-by-step guide for setting up this new process in other areas of the health system.
Truth be told, when I began this project, I didn’t expect to be excitedly giving out high-fives because of a simple function like copy-paste. After all, I work at a center for health care innovation, not a center for keyboard shortcuts. What excites me though, is how a team of two used simple design methods to quickly dissect a wicked problem and capture new value (at low cost) using existing tools.
Projected Impact
Every improvement made to health care delivery, no matter how small, has the potential, when scaled up, to both reduce cost and increase quality of care. Our project is no exception. The following are some conservative projected impacts of scaling up our new process at the ACC.
Medical oncologists spend about 45 minutes preparing for new patient visits. Our new process for handling outside records reduces that time by about 50% to 22.5 minutes. Given that many oncologists see about 4 new patients a day, this could result in 90 minutes of savings per day. These 90 minutes could in turn be used to see an additional new patient on each clinic day. Seeing 25% more new patients each day would reduce wait times for sick patients who urgently need access to the word-class care the ACC provides.
Next, given that the average oncologist makes $133 an hour in the US, 90 minutes saved could result in $200 saved per oncologist per clinic day, or hundreds of thousands of dollars each year at the ACC. Finally, the vast majority of new patients receive further treatment after their first encounter, so increasing patient access also has the potential to create new downstream revenue.
Copy-Paste is Actually Pretty Cool
As we scale up our work, we see promising changes on the horizon. First, and most importantly, we see patient access to care increasing. Second, we see physician job satisfaction increasing. Finally, and not insignificantly, we see costs decreasing.
While copy-paste is mundane in and of itself, when it improves health care delivery, it becomes exciting technology. Even more exciting is how our small team used simple design methods like secondary research, contextual inquiry, time studies, and rapid cycle testing to quickly and cheaply uncover value where you’d least expect it.