Advisory Board tasked us with completing a 5-day design sprint to create a medication-tracking solution for inpatient hospital nurses. In doing so, they presented us with this scenario:
A hospital nurse has just visited with a patient and administered a medication. The patient has not taken the medication before, but needs to take the medication 4 more times in the next 48 hours. The patient is currently taking 3 other medications. One medication is for a chronic condition, and the other two medications were first administered when the patient arrived to the facility.
Researching the Problem Space
A key part of this project was reframing the problem. Although we were provided with a scenario that outlined a typical interaction between a hospital nurse and a patient, we had to dig deeper to understand why this could be an opportunity for design. By interviewing nurses and other individuals involved in clinical work, we discovered pain points within their current use of systems for medication tracking.
To get a better understanding of the problem space, we interviewed 2 practicing nurses, 4 nursing students at Indiana University’s School of Nursing and a behavior health technician, all of whom had familiarity with the process of administering and tracking medication in a clinical environment.
We gained the following insights from our interviews:
The process of tracking medication administration is slow and inefficient.
Nurses appreciate safeguards provided by their EHR systems, including the “wrong medication” warning that is displayed when the incorrect medication is scanned for their current patient.
Inefficiency leads to nurses looking for shortcuts (e.g., getting two patients’ medications at once), increasing the potential for mistakes in administration and tracking despite these safeguards.
Nurses can become distracted as they’re pulled away from their duties by doctors, social workers, other nurses, etc.
Data entry can be tedious and leaves less time for their primary responsibiltiy: caring for their patients.
Receiving too many notifications leads to “alert fatigue,” and nurses fail to notice or even disregard notifications on their EHR systems.
We organized data from our research into a journey map that represents the current processes nurses follow in giving their patients medication.
How Might We…?
Following our interviews, we parsed out some key problems and insights and wrote them out on post-it notes in the form of “How might we…?” questions. Some of these included:
design with nurses’ busy schedules and other responsibilities in mind
streamline the process of giving medications
integrate our system into existing habits/workflow
create something that nurses enjoy using
We then clustered these into more general problem areas to narrow the focus of our design.
Based on the problem presented by Advisory Board, we recognized that medication tracking and patient/provider identification should be included as part of the core design. We also determined that finding ways to increase efficiency and integrate existing workflows into the design would allow us to address critical pain points identified in our research.
What does efficiency and integration mean? From talking to nurses, we realized that many of the tasks involved in medication tracking require them to divert their attention from providing patient-centered care. For example, nurses have to carry around a large and cumbersome computer on wheels (“COW”), slowing them down and requiring them to view and confirm data on a screen. We decided to focus on finding ways to integrate medication tracking into the flow of patient care, while still maintaining safeguards that provide nurses with peace of mind.
Constraining to our Core Problem
As they exist, EHR systems remove nurses’ attention away from their patient and onto the computer screen. Traditional approaches to incorporating technology into hospitals require trade-offs to patient-centered care. Our challenge was to see if we could leverage technology to ensure safeguards while allowing the nurse to focus on the patient’s care.
We can determine the weather in the morning while we’re getting ready by asking Alexa, we can enter our homes without fumbling for our keys, we can pay for toll roads without stopping…
...so why can’t nurses care for their patients while data is inputted automatically?
We looked at exemplars of ubiquitous computing (the integration of computers throughout our surrounding environments) and technology that the user can engage with without disrupting their current routine. This exemplar collection helped us to look at various uses of technology to determine what we may be able to leverage and incorporate in our design.
Looking at these examples, we thought of ways these technologies could be incorporated into a nurse’s daily routine while facilitating more personal interactions with patients rather than the nurse’s focus being split between caring for the patient and data entry.
From here, we were able to create concept sketches that incorporated various uses of ubiquitous computing, all of which attempted to increase the face time between nurses and patients.
Some of these concepts included:
- using voice recognition to record administration of medication as nurses explained them to their patients
- using sensors to track nurses as they entered patients’ rooms
- redesigning existing EHR systems to make them less obtrusive
With concept sketches in hand, we presented our ideas to nursing students at Indiana University to get feedback on our design direction. We learned that a daily worry was providing a patient the wrong medication. We received positive feedback on our concepts that provided additional safeguards to ensure correct medications were being given.
To guide our design process, we utilized personas representing characteristics of the nurses with whom we spoke.
After concept testing, we created a paper prototype to get feedback on our design direction.
Testing the Prototype
We recruited 5 participants to try our prototype. We asked them to imagine that they were a nurse administering medication to their patient and presented them with our band. We presented three possible scenarios:
- an already prescribed medication is being given
- a wrong medication is being given
- a new medication is being given
After briefing them, we simulated the indicator light showing one of three colors along and playing the corresponding audio tone. We asked them which of the three scenarios they thought was indicated by each light/tone combination.
5/5 subjects recognized the error tone and color
5/5 subjects recognized green as an already-prescribed medication that is OK to give to the patient
Only 1/5 subjects recognized the difference between the new medication tone and the already-prescribed medication tone
When asked, 4/5 subjects recommended using yellow as the visual indicator for a new medication, as they felt it better signaled caution
Participants had a difficult time distinguishing between the meanings of the green and blue lights. When they were shown the blue light and exposed to the “new medication” tone prior to the green light and the “correct medication” tone, it was unclear to them which represented which.
At first we were hesitant about the possible associations that would result from the pairing of green, yellow, and red. We wanted a color to represent a new situation. However, we ended up swapping blue for yellow to better conform to people’s expectations. After all, this fits more with existing models: yellow is associated with caution, and when a nurse administers a new medication for the first time they must take extra precautions.
5/5 subjects recognized that yellow signified a new medication
Our final design, MedWatch, can be implemented with existing EHR systems by connecting wirelessly. As the nurse interacts with MedWatch, information is automatically updated in a patient’s eMAR, eliminating the need for the nurse to manually input this data.
How it works
Michael has just retrieved the medication Billy is due to receive. Among these three medications is a new antibiotic ordered by the doctor.
On his way to Billy’s room, Michael is stopped by Jane, a social worker, who is inquiring about another patient’s discharge. He talks to Jane for a few minutes before continuing on to Billy.
Michael enters Billy’s room. He verifies his identity on Billy’s MedWatch by placing his finger on the fingerprint scanner. He then confirms Billy’s identity by asking for Billy’s name and date of birth, matching it against what’s written on the band’s display.
Next, Michael scans each of the medications by holding them up to the MedWatch one by one. For the first two packets, he hears a tone and sees a green light, confirming these two medications are due for Billy at this time and that they have been verified in the system.
For the third medication, the MedWatch emits a different tone and displays a yellow light. Michael, who had almost forgotten about the new antibiotic during his conversation with Jane, is reminded to provide Billy with a more detailed explanation of what the medication is, including its potential side effects.
How it addresses the problem
MedWatch automates many of the clerical duties performed by a nurse in administering and tracking medication. Additionally, it takes the task of scanning and confirming medication from their computer on wheels and incorporates it into the interaction between the patient and nurse by means of the MedWatch. Meanwhile, the same safeguards that are present in the current EHR system, including patient and provider authentication and verification the correct medication is being given, are still preserved. MedWatch accomplishes our goal of using technology to facilitate patient-centered care and reducing the amount of attention a nurse must devote to their computer screen.
For this problem, we constrained our design to how ubiquitous computing could improve the way nurses track the administration of medication. We realize that our solution alone does not eliminate the nurse’s reliance on their computer in all segments of the process of administering medication, including verifying the accuracy of the dosage instructions (especially for IV routes) and recording additional notes. However, we imagine MedWatch being part of a larger system in which ubiquitous computing technologies are deeply integrated into nurses’ workflows with the intention of facilitating more face-to-face, natural interactions between nurses and their patients throughout— all without compromising a focus on their patients’ safety.
Trusting the process
Of all the design projects I have completed so far, this is my favorite in terms of process. Uncertainty is scary. Especially when you are working within such a short time constraint, it is easy to jump straight to solutions in your head without a clear rationale.
Here, we let our insights from research guide us to our concept. Once we arrived at a general concept for solution (ubiquitous computing), we thought about a variety of ways that could be implemented in medication tracking before we settled on one. And, when we did make assumptions in our design, we validated them through testing. All of this resulted in a more user-centered design.
The importance of validating and testing in context
There were definitely a few unclosed loops in our design that we considered but did not have the chance to address due to time constraints. For example, were there any sanitary concerns with the bands? Would nurses have difficulty accessing the band for certain patients? Would we need a different way of authentication on the device if nurses are wearing gloves?
When we validated our concepts with nursing students, these problems didn’t really come up. I imagine a big part of this was that it was difficult to envision how the solution would work out of context. Moving forward, it would be important to try a prototype out with experienced nurses within a hospital to identify issues with our design that might not arise during an interview.