10 Innovation Pathways: Care Delivery

10 Paths to Innovation
in Health Care Delivery
A collection of original content from NEJM Catalyst
October 2019

October 2019
Dear Colleague,
Innovation is a constant goal in health care. Health care executives, clinical leaders, and clinicians all
want to deliver care better, with improved outcomes, at lower cost, and with better experiences for
patients and staff alike. But what is innovation, exactly? Is it a process or a technology? Does it have
to come from outside health care organizations, or can they generate innovation from inside? Who
is responsible for innovation?
This collection of original articles and reports previously published in NEJM Catalyst shows how health
care leaders create innovation in the real world. New ideas and methods come from all directions,
including in-house innovation groups, physicians and physicians-in-training, and — not to be
overlooked — patients. Innovation in these cases rarely requires technology; rather, it is based on
a recognition of the need to do better and clear thinking on how to get there.
The following articles detail how academic medical centers applied innovation methods from other
industries to launch a telehealth program and used the principles of design thinking to do away with
one of the banes of health care visits, the waiting room. Learn from a detailed case study of how
clinicians used a simple technology to improve maternal outcomes, understand the principles of
patient-driven innovation from a patient who invented her own pancreas device, and see the results
of a hospital-government-business-research partnership in China. Read insights from the innovation
leader of one of the largest health systems in the United States, and a warning against conflating
“bright, shiny, new toys” with true innovation. See how community physicians can contribute to
innovation and how physicians-in-training are hungry to do so. Finally, read the results of a survey
of the NEJM Catalyst Insights Council — a peer group of clinicians and leaders directly involved in
care delivery — on the capacity of design thinking to transform health care.
We invite you to enjoy this collection, read new content each week at catalyst.nejm.org, and register
for our newsletters and events.
The Editors,
NEJM Catalyst
781.893.3800 | catalyst@nejm.org
860 winter street, waltham, ma 02451-1413 usa | catalyst.nejm.org
NEJM Catalyst is a product of NEJM Group, a division of the Massachusetts Medical Society.

Downloaded from catalyst.nejm.org. For personal use only. No other uses without permission.
Copyright © Massachusetts Medical Society. All rights reserved.
10 Paths to Innovation
in Health Care Delivery
1 Nobody Wants a Waiting Room
20 The Challenges of Health Care Innovation’s “Bright, Shiny New Toys”
24 How NYP Used Its Innovation Stack to Launch a Telehealth Program
28 Can We Achieve Scale in Innovation?
31 Opening Up to Patient Innovation
36 Heart Safe Motherhood: Applying Innovation Methodology for Improved
Maternal Outcomes
46 Care Redesign Survey: How Design Thinking Can Transform Health Care
61 Engaging Community Physicians in Innovation
65 Physicians-in-Training: An Untapped Resource for Health Care Innovation
71 Accelerating Medical Device Innovation in China
October 2019

10 Paths to Innovation in Health Care Delivery 1
Return to TOCReturn to TOCcatalyst.nejm.org Return to TOCReturn to TOC
Nobody Wants a Waiting Room
Article
Stacey Chang, MS
Design Institute for Health, Dell Medical School,
University of Texas at Austin
If it’s true that nobody wants a waiting room, then why does everyone still seem to have one?
This initial conversation would be the spark for a series of struggles over the next 2 years as
we attempted to establish a new model of care, with a service blueprint and an environment
to enable it.
In the summer of 2015, as the new Dell Medical School at The University of Texas at Austin
began planning for the launch of its specialty clinics, the school’s embedded Design Institute
for Health was asked to assist in the design of both the service model and the physical layout
of the clinic.
With his nose buried in a binder full of specifications for standardized outpatient
clinics, the architect asked, “Where do you want your waiting rooms?”
“I don’t think anyone wants a waiting room,” came the earnest reply.
Chuckling, the architect persisted, “Yeah, you’re probably right, but let’s figure out
where they’re going to go in this building.”
“Actually, we’re not going to have waiting rooms. Nobody wants a waiting room.”
The room grew quiet, and a few people shifted uncomfortably. The architect,
sensing no jest in the exchange, looked up and gently closed his binder. “But
everyone has waiting rooms . . .”

catalyst.nejm.org
Return to TOCReturn to TOC
The school had been established only a few years earlier, funded in part by a property tax
increase that was passed by the local county taxpayers as an investment in a future model of
care to better serve their needs and priorities. The school’s commitment to challenging
existing norms began with disavowing the fee-for-service model in favor of a value-based1
model of care. This model was a centerpiece for the fledgling medical school — a clean-sheet
opportunity to adopt a payment structure better aligned with society’s needs, to improve the
patient experience, and to focus on deploying Integrated Practice Units (IPUs). IPUs are a
structure for delivering care through highly coordinated multidisciplinary teams, organized
around a specific medical condition (or set of conditions). They are responsible for the full
cycle of care for that condition, including education, engagement, and follow-up, and
incorporate inpatient, outpatient, rehabilitative, and support services (e.g., nutrition and
social work). Teams are co-located, measure processes and outcomes as a team (instead of
individually), and most importantly, accept joint accountability for outcomes and costs.2
The possibility of eliminating waiting rooms represented one
opportunity to transform the patient experience. For patients
and family, the feeling of wasted time, alongside others who
were ill, was a source of frustration and anxiety, and
reinforced the prioritization of the system’s needs over the
patient’s. We felt it was impossible to address these
drawbacks, no matter how many amenities were included,
and waiting rooms wasted valuable floor space that could be
better utilized for other purposes. In agreement with the dean
of the medical school, we decided to make the elimination of
waiting rooms an anchor tenet of the new clinics.
We started with an empty shell. The building had been
designed before most of the leadership of the medical school
had arrived, and certainly before we had committed to a
value-based model of care. The existing plan reflected a more
traditional fee-for-service model with rows of small exam
rooms meant to shuttle patients through at maximum efficiency: an interior layout that
reflected current “best practices.” Fortunately, we took over before any of the interior had
been finalized and took the opportunity to revise the plan for our new care and service
models.
The architects were initially reluctant to depart from known practices. If our experiment
were to go awry, the plans we were developing would require a complete renovation in order
to revert to a more traditional layout, at a cost we couldn’t afford. The architects even offered
The waiting room is
nothing more than a
temporary stock room, or
intermediate warehouse
for patients with billable
conditions that feed exam
rooms every 10–15 minutes
…. No health care provider
I know actually views
patients as a packaged
revenue opportunity, but
the fee-for-service system
has incentivized this
warehousing behavior.”

catalyst.nejm.org
to do a computer simulation of staff and patient flow to show the advantages of having a
waiting room, but we found there was no preexisting data that could be used for a
meaningful simulation.
Instead, we built small-scale prototypes, then full-scale prototypes in which we did mock
operational run-throughs to assess the feasibility of our proposed model. Could we assign
patients one room for the duration of their stay and get providers to circulate effectively
between them? Could patients self-navigate to their rooms? Could we standardize room
layouts across specialties? Each prototype provided valuable feedback about how the launch
model would behave and allayed our concerns.
Despite their initial reservations, the architects eventually became our allies, recognizing that
traditional layouts wouldn’t serve our new care model and that this project was an
opportunity to explore concepts that they didn’t have permission to try in other venues. The
eventual buildout included movable walls and reconfigurable furnishings to produce a
flexible, evolvable space that accommodated not only our launch model, but also the
inevitable changes that would happen as we made new discoveries in the clinics we call our
“learning lab.”
“But I need a waiting room. Where else will I get my patients from?”
A clinical leader at the medical school, when told that we wouldn’t be building waiting
rooms, insisted quite sincerely that the clinic wouldn’t be able to operate efficiently without a
ready supply of patients at hand. The concern reflected a “factory” metaphor that is almost
universal in traditional clinics paid on a fee-for-service basis.
Source: Authors, Design Institute for Health, Dell Medical School
NEJM Catalyst (catalyst.nejm.org) © Massachusetts Medical Society
Full-scale prototypes allowed us to explore concepts and test assumptions
about the behavior of providers and patients.

catalyst.nejm.org
The waiting room is nothing more than a temporary stock room, or intermediate warehouse
for patients with billable conditions that feed exam rooms every 10–15 minutes, ensuring the
unbroken stream of billable encounters demanded by RVU targets or other measures of
productivity. No health care provider I know actually views patients as a packaged revenue
opportunity, but the fee-for-service system has incentivized this warehousing behavior.
Patients waiting in a clinic are akin to materials being held in staging in a factory
production line. They are paused, then moved, only when the next stage of the
production line is ready to process them. A traditional clinic prioritizes the
factory – an integrated practice unit returns focus to the patient.

catalyst.nejm.org
In a value-based model, the pressure to push patients quickly through appointments is not
the primary driver, since each encounter doesn’t represent a distinct billing event. Instead,
getting to productive outcomes is more important, so we incorporate as much as we can into
a single visit, including interactions with multiple providers, imaging and diagnostics,
nutrition counseling, physical therapy, social services and care coordination, and other
associated services. Because single appointments can last 1.5 hours, a small gap in the
provider’s schedule is less consequential. Importantly, patients can get through several stages
of their care journey in a single visit. One long appointment may mean a half-day of missed
work, but three short visits effectively meant three missed half-days of work anyway, when
accounting for travel time. Especially for patients who struggle with transportation access,
this was a welcome improvement.
Even so, providers still wanted to minimize their downtime in
the clinic. To address this, we shifted “ownership” of the exam
rooms from the provider to the patients. Rather than
delivering patients serially to one exam room after another,
each owned by a different provider, we made the patients the
owner of their own rooms, and instead, circulated the
providers to the patients. In our model, patients and family are
shown directly to their rooms when they arrive (now renamed
“care rooms”) and reside there for the duration of their visit.
Providers then walk to patient care rooms as soon as they want
to see the next patient. We no longer need to deliver patients
from a waiting room to provider exam rooms with perfect efficiency. Our model does require
more operational coordination, but the care teams are tightly integrated to help facilitate
this. When one care team member exits, they can let the next team member know. There are
still small gaps of time, but the patients are told about them, and they stay in their care room
instead of being shuttled off to an interstitial waiting area. Care rooms are also arranged
around the central provider team space to minimize traverse times.
In the old model where the providers “owned” the care rooms, they knew exactly where all
their supplies were; in the new model, we have standardized supplies and placement so that
providers are comfortable no matter which room they are in.
Rather than delivering
patients serially to one
exam room after another,
each owned by a different
provider, we made the
patients the owner of their
own rooms, and instead,
circulated the providers to
the patients.”

catalyst.nejm.org
The layout for each Integrated Practice Unit is identical, regardless of specialty.
The clinic is separated into “front-of- house” (blue) and “back-of-house” (white)
zones. The” front-of-house” is for patient ﬂow and the “back-of-house” is for
care team members.

catalyst.nejm.org
“Are we really going to give patients control over their experience?”
We launched this project with in-depth design research to understand what each of the
stakeholders wanted out of the new model and environment. For both the patients and
providers, we established design principles that would guide our work throughout the
project.
The back-of-house is designed to encourage care team members to intersect
and overlap. More informal contact leads to better communication and
coordination, and the design of the space enables this. Physicians, nurses,
researchers, and the rest of the staﬀ all share the same working space.

catalyst.nejm.org
Most of our clinical colleagues embraced the ideas as worthy aspirations, but one provider in
every audience would eventually inquire whether we were really committing to “grant the
patient control over some aspect of their journey?”
Providers didn’t question whether a patient deserved to have some control over their
situation, but some perceived the patient-provider relationship as a zero-sum game — if the
patient gained control, then there was the risk that the provider might lost some.
We sought to give patients control over their experience and
environment without diminishing the provider’s influence,
and eliminating the waiting room was part of that effort.
Anxiety is a known issue for patients in medical contexts.
Granting control to an individual, even if it’s incomplete
control, helps to mitigate that anxiety, which improves patient
engagement and decision-making. Our model also asks
patients to engage meaningfully in their own care, whether
through goal-setting, patient-reported outcomes3, or shared
decision-making. In our model, these are introduced prior to
the first visit, revised at the clinic appointment, and revisited in follow-up care.
Providers didn’t question
whether a patient deserved
to have some control over
their situation, but some
perceived the patient-
provider relationship as a
zero-sum game.”

catalyst.nejm.org
Unallocated Time
Depending on the shift, if a patient and family arrive within 10–15 minutes of their
appointment, they can reasonably expect their care room to be available. However, if they
show up an hour early, we are likely to still be utilizing that room during high-capacity shifts.
Likewise, if they show up an hour or more late, we will have shuffled the room schedule to
accommodate other appointments, and they may have to wait for a room to be available or be
rescheduled.
Traditional clinics stick those patients in a waiting room, to watch TV or the fish in the tank
or their fellow patients. In our clinics, the patients can choose how to spend their unallocated
time — in the atrium, the café, or the learning library where they can get educated on their
condition or therapy. Our concourses have alcoves designed to accommodate different types
of activity (quieter secluded seating for one, chairs facing each other for couples, larger
lounges with couches for families, or workspaces for those who want to be productive). Wi-Fi
is free and pervasive, and plug-ins for electronics are everywhere. The shift in intent is subtle,
but the effect is powerful. Patients note how the experience feels completely different from
any clinic they’ve been to before, how much less stressful it is, and how much happier they
are for it.
The concourse is the ﬁrst thing patients see upon entering the clinics. This is a
strictly “front-of-house” space for patients and family only, and is designed to be
quiet and calming. Alcoves along the concourse accommodate diﬀerent group
sizes and interactions.

catalyst.nejm.org
In the care rooms themselves, we incorporated vital sign devices into each room so that
there’s no need to create inefficiencies or bottlenecks by pausing at another station
beforehand. The care room doors can be closed and locked by the patient (the providers have
a separate entrance). We encourage the patients to adjust the lighting to suit their needs.
Available furniture accommodates companions, and there’s even a closet for the patient to
store their belongings.
At the end of the concourse, windows overlook the cityscape and additional
spaces are provided for focused work, or for larger families.

catalyst.nejm.org
After walking providers through prototypes of the experience, we returned to their original
concerns. Did they think the control we gave to the patient would be meaningful? Would it
threaten a provider’s influence on the course of care? They answered, “Yes, and no.”
“My clinic operates differently from the others.”
At some point across the span of the project, the clinical lead of every specialty clinic
expressed the same sentiment, in roughly the same words. This claim to exceptionalism
wasn’t an issue of arrogance — it was really a testament to the siloed nature of specialty care.
Because specialty practices typically run as their own business unit, they function based on
circumstance and individual preference of the managing providers.
In our clinics, we needed to institute a level of standardization across the different specialty
clinics in order for a no-waiting-room model to work consistently across the entire building.
Furthermore, many of the conditions we’d treat would require collaboration between
multiple specialties as part of a combined IPU. Developing a common core approach to the
service model was as important for the experience of the patients as it was for the outcomes
of the clinics.
The care rooms are designed to be owned by the patients for the duration of
their stay and to equalize the power balance between care team members
and patients.

catalyst.nejm.org
It took 6 months to define the new practice model, reconcile
differences between specialties, and build a universal service
blueprint that would serve as a foundation for every specialty. By
normalizing what we believed would work best, we could elevate
the baseline for everyone. We still accommodated any necessary
variation justified by a specific specialty’s needs, but the vast
majority of the blueprint remained the same between specialties.
In the end, we found much more commonality than difference
between the specialties and preserved our ability to deploy common services to support the
elimination of waiting rooms. Figure 9 shows the 21 stages of the care process that are
common to virtually all the specialties. (The only optional component is diagnostic imaging,
because not every specialty requires it.) There were some minor variations in the rooms
themselves, primarily around specific supplies that were required, the positioning and
selection of furniture, and how the treatment room was utilized. Depending on the clinic,
the treatment room could be used for outpatient procedures (many of which were previously
done in a hospital) or in other specialties, like psychiatry, as group therapy rooms. Otherwise,
the clinics remained the same. Each of our floors (at about 45,000 square feet) accommodates
four and a half clinics, and each clinic accommodates 10 patient rooms.
This claim to
exceptionalism wasn’t an
issue of arrogance — it was
really a testament to the
siloed nature of specialty
care.”

catalyst.nejm.org
“What if we lose the patient?”
We had always planned to greet patients as soon as they arrived in the building (at the garage
or pedestrian entrance), do a lightweight check-in (name and appointment), and then
immediately assign them to a room. Not unlike in a hotel or other hospitality experience, we
expected they would have no issue taking the elevator to the right floor, and then finding
their way to their room.
As we developed the check-in protocol, members of the operations staff raised a concern:
What if patients got lost on the way to their room? We couldn’t track them in between check-
in and arrival, so there was a chance they might fall off our radar. Wouldn’t a waiting room
model work better, because we’d always know where they were?
We believed that the options under consideration represented
a false choice between “enforce sequestration” or “leave them
to unguided autonomy.” A third option, guided autonomy,
could get them where they needed to go through the use of
thoughtful design.
The building was set up for real-time location services using
RFID technology. Eventually, we planned to give each patient a
tag to track their progress through the building, but we had to
delay the deployment of that capability in order to focus on
other priorities more critical to the clinic launch.
In the interim, we developed a paper-based protocol, with
human backup. We had already developed a straightforward
wayfinding scheme: Floors were labeled with numbers,
hallways with letters, and then rooms with a second number. At arrival, each patient was
given a “boarding pass” with their room assignment.
We believed that
the options under
consideration represented
a false choice between
‘enforce sequestration’
or ‘leave [patients] to
unguided autonomy.’
A third option, guided
autonomy, could get
them where they needed
to go through the use of
thoughtful design.”

catalyst.nejm.org
Care rooms are labeled with the hallway letter and the room number.

catalyst.nejm.org
We stationed a concierge on each floor in front of the elevator to direct patients to the
correct hallway if needed. Once they arrived at their room, we developed a number of
prototypes to allow patients to indicate to care team staff that they had arrived and were
ready to be seen.
Patients are provided with a boarding pass upon arrival. It includes simple
directions and a light orientation to the experience. A tear-oﬀ at the bottom
of each boarding pass can be given to accompanying family or friends for
reference in case they need to park their car or explore other parts of
the building.

catalyst.nejm.org
Mock operational run-throughs suggested that this approach would work with some
tweaking, but learnings at launch showed that these were ultimately unnecessary. To further
reduce risk, we moved check-in to each individual floor, just to take the elevator ride out of
the equation, with a plan to move back to the original model later to ensure scalability.
“You have to have a waiting room for the EHR to work.”
While setting up the electronic health record (EHR) for the clinic, members of the operation
staff had discovered that the software was organized around “stages” of a standardized clinic
journey. In the first stage (called “Check-in”), there were a significant number of patient
information fields that had to be completed before the software would allow us to assign a
room to the patient. Our staff, in configuring the software, realized that a number of those
information fields wouldn’t be completed until the patient was already in the care room. Our
model clearly required room assignment much earlier in the process.
To avoid having to institute a waiting room solely to accommodate our EHR, we found a way
to manipulate the software by leaving the first-stage information fields incomplete,
overriding the default rules in order to assign the room, and then coming back to complete
the fields later. This process taught the operational staff the important lesson that the model
could, and should, dictate how the tools are used, rather than allowing the limitations of the
tools to dictate the model.
Two of several approaches were developed to signal to the care team that the
patient had arrived and was ready to be seen.

catalyst.nejm.org
These efforts, and the challenges we faced, weren’t about just
eliminating waiting rooms. Many clinics don’t have waiting
rooms, especially in models like concierge medicine that don’t
emphasize high throughput. Also, some of the best traditional
clinics have waiting rooms that are usually empty.
For us, eliminating the waiting room catalyzed a broader effort
to challenge norms and change perspectives. The obvious
benefits included improved patient experience and control,
better integrated provider teams, and streamlined operations.
Less obviously, the change reminded our staff that we could do things differently and that a
focus on the patient didn’t have to come at the expense of the staff’s priorities. Alongside
team-based care, patient-reported outcomes, and a focus on comprehensive care, the
elimination of waiting rooms served as a visible signal to our community that we were
committed to honoring their investment by doing things differently.
None of the practical solutions were groundbreakingly difficult to achieve. We had to
disentangle people from long-held beliefs, help them understand the system’s true needs,
equip them with a process to discover new solutions and evidence, develop the institutional
courage to persevere to a solution, and understand the role that technology, the
environment, and the human players all have in standing up a new model of care.
Lessons Learned
We had thought that a design of an environment that empowered and enabled patients
would be quickly appreciated and intuitively adopted. But just like providers, patients had
been trained into behaving in certain ways in medical environments, and we would discover
them perched on the edge of the exam chair, afraid to touch anything. To address this issue,
we began utilizing “rooming” concierges who, under the guise of helping patients find the
way to their rooms, explained how this model was different along the way, showed them the
room, and encouraged them to get comfortable, before making a warm handoff to care team
staff. These scripted encounters eliminated the need for the patient to signal to the care team
that they had arrived. We also utilized vinyl lettering on the walls to allow the room to
explain itself to its occupants, revealing the intent of the room’s design and giving the room
its own voice and personality.
The change reminded
our staff that we could
do things differently and
that a focus on the patient
didn’t have to come at
the expense of the staff’s
priorities.”

catalyst.nejm.org
As part of the room buildout, we specified large monitors, with a video camera that could be
deployed to conduct a video checkout directly in the room (instead of a separate checkout
desk). It allowed for patients, providers, and the phone center staff to collaboratively set
follow-up appointments as part of a video call. The available technology still isn’t seamless
enough, so instead the medical assistants check the patients out directly from the room,
setting follow-up appointments as necessary, which removed the need to coordinate with the
third-party phone center staff. For most patients, this method has helped us preserve the goal
of eliminating another stopover on the way out of the clinic. However, for complex follow-
ups that require highly coordinated or stacked visits, an additional conversation with a
dedicated scheduler still is necessary and will require further work to streamline. We are
continuing to pursue technology solutions, but only when they improve the patient
experience and remove staff burden.
The large monitor is also intended to mirror the EHR record
in which the provider is working. The goal is to provide
transparency to the patient about what is being recorded and
to allow the patient to correct any immediate errors, in case
the provider mishears or mischaracterizes something the
patient said. It is not consistently used, as some providers find
it very uncomfortable for patients to be watching what they
type into the record, or because they don’t typically document in the room at all. This issue
will also need further work.
The room speaks directly to the patient, signaling a very diﬀerent experience.
If it’s true that nobody
wants a waiting room,
then why does everyone still
seem to have one?”

catalyst.nejm.org
We launched with three clinics in October of 2017: Musculoskeletal (Orthopedics), Women’s
Health (Complex Gynecology), and WorkLife (Urgent Care). Since then, we have opened
clinics for Multiple Sclerosis, Cognitive Impairment, Bipolar Disorder, Pain Management,
Worker’s Comp, Primary Care, Bipolar Disorders, Psychiatry, Cancer, Rheumatology, and
Heartburn and Esophageal Disorders. Interestingly, in a model that anchors on longer
duration visits (up to 1.5 hours), we manage six room turns a day at capacity, better than the
industry average of three to five room turns per 8-hour shift for fee-for-service specialty
clinics. Our team of multiple physicians, advanced care practitioners, and other providers
makes efficient turnover possible, because the rooms don’t have to wait for the attention of a
single physician.
Whether this model can scale at full capacity across other venues is still to be proven.
Whether it can work in an environment where most appointments aren’t prescheduled is
also not yet known. What role technology will play in the evolving model is yet to be
resolved. We fully expect that changes in disease patterns, new developments in diagnosis
and treatment, and evolving payment mechanisms will require us to continuously modify the
model over time.
In hindsight, one thing is certain, though — I would have revised my reply to the architect’s
initial query. It’s not true that “Nobody wants a waiting room.” Patients certainly don’t, but
plenty of other stakeholders think they do. I should have said that in this model, nobody
actually needs a waiting room.
References
1. Porter M, Teisberg E. Redefining health care: creating value-based competition based on results. Boston, MA: Harvard
Business Review Press, 2006.
2. Porter, M, Lee TH. The strategy that will fix health care. Harvard Business Review October 2013; 91, no. 10: 50–70.
3. Weldring T, Smith SM. Patient-reported outcomes (PROs) and patient-reported outcome measures (PROMs). Health
Serv Insights. 2013;6:61–8.
Stacey Chang, MS
Executive Director, Design Institute for Health, The University of Texas at Austin

Return to TOCReturn to TOCReturn to TOCReturn to TOCcatalyst.nejm.org
The Challenges of Health Care Innovation’s
“Bright, Shiny New Toys”
Interview
Molly Joel Coye, MD, MPH & Amy Compton-Phillips, MD
AVIA
Providence St. Joseph Health
Amy Compton-Phillips and Molly Coye discuss if worrying that embracing new innovations will
lead to the same problems that came with EHRs.
Listen to audio interview.
Amy Compton-Phillips: Hi, this is Amy Compton-Phillips, Chief Clinical Officer with
Providence St. Joseph Health, here with NEJM Catalyst, and I’m speaking today with
Molly Coye, Executive in Residence for AVIA and former Chief Innovation Officer at
UCLA Health. Morning, Molly.
Molly Coye: Good morning. Thank you.
Compton-Phillips: What do you think is the proper role for innovation in health care?
Coye: There are a lot of different ideas about this. There is no single animal out there. My
opinion is that it should be closely linked to the strategy of the organization, that it’s not
about the academic exercise of just inventing new things, which is very valuable on its own.

catalyst.nejm.org
That has a value, and it shouldn’t be principally about commercialization, that it really is a
question of understanding the deep problems for patients and for systems that are trying to
work for those patients and finding innovations that will help in those tasks.
Compton-Phillips: That’s really helpful, because I think one of the challenges we get from
the health care side is we have a lot of people coming to us with bright, shiny, new toys
that they want to bring into our system, and sometimes it feels like it’s simply going to
add cost and add complexity rather than change. How do you help us distinguish what is
a real innovation and what’s a new shiny object?
Coye: First of all, I think in most cases you’re talking about innovations that actually have
been fielded somewhere. If they’re coming to you from the outside, you do have the
opportunity to do due diligence, to actually go out and see what the impact has been, but
don’t waste resources. If they’re not tackling something that’s important to you, no matter
how much traction they get, it’s really not going to benefit you to adopt them. That’s why I
think you start with a problem or the opportunity.
The other problem frankly for a lot of health systems is
internally there’re a lot of people who think they can invent
something better than what’s out there. That may be true, but
they tend to forget how much capital, marketing, organization
building, et cetera is required to get from an idea to something
that’s going to really have an impact.
Compton-Phillips: That makes total sense. I think part of the
challenge with wholeheartedly embracing new innovations
in health care comes from our recent embracing of EHRs, and so many physicians I
speak with, and nurses and caregivers of all stripes, find that with the adoption of EHRs,
we have enhanced communication, but we’ve so increased the burden of care that it’s
blamed for contributing to burnout. So, adopting things becomes a challenge. How do we
adopt things that are right and that help minimize the burden of care and enhance the
joy of care? Do you see innovations that get there?
Coye: Definitely, and I think there’s a whole category of innovations, which we call health
information technology and services. It’s using lighter, simpler in many cases, technologies to
make providing services and enabling patients, engaging them and helping them much easier.
So,we have to have a test. We talk about, is there patient satisfaction? We also need to have
the providers within 12 or 24 months of being involved in an innovation be delighted with it.
Compton-Phillips: Do you have some examples, like some breakthrough kinds of things
that have done that?
If they’re not tackling
something that’s important
to you, no matter how
much traction they get, it’s
really not going to benefit
you to adopt them.”

catalyst.nejm.org
Coye: I’ll give a couple of examples. One is Augmedix, which is — and I want to be clear I
have no financial relationships — Augmedix is the Google Glass used for remote scribing. You
know how much time doctors and nurses spend in documentation. In this case, using the
Google Glass with an offsite scribe, the EHR is completed at the time the patient leaves the
office. That has relieved physicians of so much work. They are delighted with that
innovation.
Another quick example is at UCLA Health, we put care
coordinators in the clinic to focus on the 20% highest-risk
patients. And at first the physicians thought, “Another body
around that I’ve got to explain everything to?” Within 6
months they were killing to get them in their clinics, because it
really helped. So, we should really be focusing on delighting
the patients and helping to relieve the burden on the clinicians.
Compton-Phillips: I really like that, because it’s not only
technology but it’s workflow and people and thinking about
innovations much more holistically. It really isn’t just a new
toy. When you started as Chief Innovation Officer at UCLA, how did you message the role
of innovation for everybody you worked with, whether it was physicians or staff or the
C-suite?
Coye: Well, first of all, we tied it back to the goals of the organization. We had very clear
pillars and strategies, et cetera. I have to tell you, it doesn’t always work. It’s really hard, and
we tried very hard to get the remote management of ICUs in, because we thought UCLA
could be a good hub for that, and that model existed in other places around the country, and
we were completely unsuccessful.
Sometimes it doesn’t work, but when it does what you’re doing is explaining to people why,
and again, I think you’ve got to have pretty short time frames on this, 12 to 24 months in
most cases. Life is going to be better for them and for the patients, and with that you really
have some traction.
Compton-Phillips: Brilliant. One last question. What do you think personally are your top
takeaways from your time at UCLA Health as a leader in really trying to bring this kind of
uphill change you’ve been talking about into an organization?
It really is a question of
understanding the deep
problems for patients and
for systems that are trying
to work for those patients
and finding innovations
that will help in those
tasks.”

catalyst.nejm.org
Coye: Unfortunately, it’s a kind of tough message, but we really need the economic incentives
to buttress what we know patients need and want, that operating in a PPO environment
makes it very difficult to meet the needs of the patients because we should be investing. And
some of these innovations were innovations 30 or 40 years ago, and community health
workers, social workers, mental health workers enabled by technology now as they weren’t 30
years ago. That’s what our patients need and want, and we just don’t have the wherewithal
unless we’re in a system that ties reimbursement to improvement in health.
Compton-Phillips: Aligning incentives always makes a difference. Well, thank you so
much, really appreciate your words of wisdom. It was great talking to you today.
Coye: Thank you.
Molly Joel Coye, MD, MPH
Executive in Residence, AVIA; former Chief Innovation Officer, UCLA Health
Amy Compton-Phillips, MD
Executive Vice President and Chief Clinical Officer, Providence St. Joseph Health

How NYP Used Its Innovation Stack
to Launch a Telehealth Program
Article
Jonathan Gordon, MBA, David Tsay, MD, PhD, Shauna Coyne, Daniel Barchi, MS, Peter Fleischut,
MD & Emme Deland, MBA
NewYork-Presbyterian Hospital
Health care is unlike other industries in a number of ways. Notable among them are the
constraints imposed by regulation, the dependence on a third-party fee-for-service payment
system, and the reliance upon human expertise (clinicians) to deliver services. As a result,
many interesting ideas — both from inside health care and from other industries — are
stymied by the realities of attempting change in such a complex system. Innovation done for
its own sake, or innovation attempted without regard to the context of health care, is
doomed to failure.
At NewYork-Presbyterian, we have learned from our own
experience and that of our peers that pilot projects in health
care, particularly those involving technology innovation,
require 90% of the work of a full implementation but often
result in a dead end. For that reason, we prefer to focus on
large-scale challenges that can be addressed with technology
that can be quickly scaled across the enterprise.
It was in this context that innovation groups at NYP began
looking at the opportunities and challenges posed by
telehealth. We saw a need to develop and apply a structured
approach — what we came to call our innovation stack — to
our innovation processes as a way to achieve sustainable
change across our entire organization. Adopting this construct
enabled us, in under a year, to build NYP OnDemand from a
concept into one of the most comprehensive telehealth
programs in the country.
We saw a need to develop
and apply a structured
approach — what we came
to call our innovation
stack — to our innovation
processes as a way to
achieve sustainable
change across our entire
organization. Adopting this
construct enabled us, in
under a year, to build NYP
OnDemand from a concept
into one of the most
comprehensive telehealth
programs in the country.”

catalyst.nejm.org
In the world of computer science, the term “stack” refers to the full range of hardware and
software necessary to create a complete platform capable of running applications. At NYP, we
have borrowed that term to describe the range of activities that we believe are necessary to
ideate, design, implement, and manage innovation at a major academic health center. Our
innovation stack consists of a set of six interrelated activities intended to achieve the
leadership, front-line buy-in, and engagement necessary for successful innovation.
Beginning in early 2015, our Innovation Center and NYP Ventures groups oversaw the
creation of NYP OnDemand along the six dimensions of our innovation stack.
►► Scope the challenge. Our first step was to identify at a high level the problem we wanted
to solve. NYP continually seeks to optimize the deployment of clinical resources while
also improving patient access, and we thought telehealth might offer a significant
opportunity to achieve these goals. (We had some experience in telehealth, having
launched a successful telestroke program to earn stroke certification at one of our
community hospitals in 2013.) Consequently, we met with stakeholders across NYP, as
well as at Weill Cornell Medical College and Columbia University College of Physicians
and Surgeons (NYP’s affiliated medical schools), to identify opportunities and challenges
associated with implementing a telehealth program. They helped us develop and refine
use cases, such as peer-to-peer emergency department specialist consults for stroke,
psychiatry, and pediatrics. Through this process, some of these stakeholders became our
early internal champions.
►► Research solutions. Our research efforts occurred in tandem with the first step, as we
supplemented our internal conversations with external analysis of opportunities,
challenges, best practices, and possible solutions. This included researching the current
state of state and federal legislation and regulation on telehealth, and speaking with
several leaders in using telehealth, ranging from other academic health centers to the
Veterans Health Administration. The result was an internal whitepaper that was shared
across the organization and was used to set expectations among the members of our
Steering Committee.
►► Identify best-of-breed vendors. Our Innovation Center led a rapid-cycle evaluation
process in which we reviewed 35 telehealth vendors that had been identified in our
research process, in order to produce a small number of finalists. This process also enabled
us to understand what internal capabilities we would need to develop ourselves, such as
creating new clinical workflows and integrating telehealth into our patient-facing app.
Ultimately, we chose three vendors based on their relative strengths: American Well,
Grand Rounds, and Avizia.

catalyst.nejm.org
►► Implement solutions. Our Steering Committee chose three initial use cases — the peer-
to-peer ED specialist consults, remote second opinions, and remote follow-up visits — by
balancing the greatest need against speed to execution. Working with the selected
vendors, the Innovation Center led the technical implementation, while the Steering
Committee created workgroups to identify clinical champions, develop workflows,
address legal, regulatory, and billing issues, and create performance metrics. We phased in
each use case, starting with a single campus and then rolling out across the enterprise (six
campuses in all) after resolving challenges.
►► Scale through partnerships. After gaining experience implementing and managing NYP
OnDemand across the enterprise, we began establishing partnerships with organizations
with broad consumer reach that can help us achieve additional scale. We are speaking
with local and national consumer-facing companies, such as cell phone manufacturers
and pharmacies, about leveraging their distribution platforms to broaden the reach of our
NYP OnDemand programs, and offering to deploy NYP OnDemand to help large
employers manage their employees’ health needs.
►► Make venture investments. Our internal and external research and vendor evaluation
efforts enabled us to develop a robust market map for the telehealth space. As we selected
the vendors to roll out NYP OnDemand, we saw an opportunity to invest in one of them —
Avizia. Given the stage of the company (an established hardware company rolling out an
integrated software platform) and our 3-year experience working with their technology
(we had used their hardware to launch our telestroke program), we saw a compelling
opportunity to add value to both our organizations. Our goal is to foster a virtuous circle,
where we are able to help Avizia refine their product offerings and thereby improve NYP
OnDemand, while also capturing some of the upside through the increased value of our
investment.
Since the launch of NYP OnDemand in July 2016, it has already become one of NewYork-
Presbyterian’s core service offerings — so much so that we are committed to having 20% of
our patient encounters occur virtually by the end of 2018.
Jonathan Gordon, MBA
Director, NYP Ventures, NewYork-Presbyterian Hospital
David Tsay, MD, PhD
Associate Chief Innovation Officer, NYP Innovation Center, NewYork-Presbyterian Hospital
Shauna Coyne
Director, Telehealth, NewYork-Presbyterian Hospital

catalyst.nejm.org
Daniel Barchi, MS
Chief Information Officer, NewYork-Presbyterian Hospital
Peter M. Fleischut, MD
Senior Vice President and Chief Transformation Officer, NewYork-Presbyterian
Emme L. Deland, MBA
Senior Vice President and Chief Strategy Officer, NewYork-Presbyterian Hospital

Can We Achieve Scale in Innovation?
Article
Aaron Martin, MBA & Sara Vaezy, MHA, MPH
The future success and sustainability of health care providers are inextricably tied to
innovation and scale. In an increasingly uncertain future for the health care industry’s
regulatory, policy, and payment environment, providers will likely experience downward
pressure on top-line revenue and an increasing drive toward value, regardless of its form.
Disruption of our care delivery and business models, achieved in part through promoting
digital innovation at scale, will be one of the critical components to succeeding in
challenging times.
How can health care organizations develop and scale innovations? In the Digital & Innovation
group at Providence St. Joseph Health, we apply approaches from consumer industries and
follow a deliberate process to scale up innovations.
Start with a Focus on Solving Big Problems
Achieving scale in innovation begins with focusing on solving the biggest problems you face.
The process starts with answering the following question: “What’s really going to move the
needle for our consumers and for ourselves as providers?” At Amazon, where I managed
Kindle’s North American publishing business before coming to Providence St. Joseph in 2014,
we focused on three goals for online retail: lower prices, better convenience, and broader
selection. No project was approved unless it would move the needle at scale in one of these
areas. In health care, solving big problems should help to achieve the Quadruple Aim:
lowering costs, achieving better clinical outcomes, and improving the patient and clinician
experience.

catalyst.nejm.org
A common organizational trap is funding technology before
internal processes are ready to accept it. In a lot of situations,
we’ve identified big problems that could be helped with
technology, but we’ve recognized that we need to work on core
processes further before we deploy technology. Many
organizations also make the mistake of funding a technology
rather than solving a problem. This leads people to focus on
the vendor ultimately selected, instead of the problem the
organization is trying to solve. The solution set then becomes
constrained to the selected technology vendor’s capabilities.
Build Solutions That Address the Big Problems
Once the focus on the big problems is established and the problem statements are created,
the next step is to develop product ideas, build small pilots called Minimum Viable Products
(MVPs), test and learn around those MVPs, gather data and take measurements, and then
repeat the process until you can determine what will deliver demonstrable value when
extended at scale. Along the way, it’s critical that we incorporate consumer feedback — the
consumers being both our patients and our providers. This approach ensures that we build
the right things that satisfy the needs of our customers and our system.
Use Lean Innovation to Test Experiments Rapidly, and Fail Fast (and Cheaply)
Lean innovation is a product development process used in many industries, particularly
technology, to quickly test new products for market viability. We’re using this technique at
Providence St. Joseph Health to develop new services and products. Lean innovation uses a
series of small experiments (the MVPs) that prove value to patients, clinicians, or both. By
incorporating customer and provider feedback continuously as the product is being
developed, this approach spreads the risk of the product development out over time, which
makes product development less expensive.
Turn Solutions into Strategy
The digital teams at Providence St. Joseph Health use the process outlined here in building
out our digital capabilities. One product we’ve built is CircleTM
, a women’s health mobile app
that delivers relevant content, products, and services to a family’s Chief Medical Officer —
Mom. Circle began as a tool for expectant and new mothers to manage their own and their
new baby’s health with a personalized experience, clinically approved answers to FAQs,
timely and local to-dos, and convenient access to relevant resources. The personalization
platform for Circle has now been extended to pediatrics up to age 18, and will soon be
spanning across the spectrum of women’s health.
The process starts with
answering the following
question: ‘What’s really
going to move the needle
for our consumers and for
ourselves as providers?’”

catalyst.nejm.org
There are many benefits to engaging our patients and
clinicians digitally: more convenient access to health care
at lower cost, more effective population health, new revenue
streams, and an improved clinician experience. As such, our
strategy is to entice new and existing consumers to get online
with us by building an online experience that is an order of
magnitude more convenient than what they experience offline.
We then build relationships and engage with them on an
ongoing basis between their care episodes by delivering
personalized health and health care experiences.
Make Thoughtful Investments in Organizational Culture
Certainly, building an innovative product or service is not sufficient on its own. In addition to
building the right products and services, innovation at scale requires thoughtful investments
in culture and change management, as well as training for caregivers who will ultimately be
responsible for deploying and adopting new tools and technologies, organization wide. We
must be self-critical and accept that in the past the industry hasn’t done the most effective
job at giving our customers and providers the most effective tools, information, resources,
convenience, access, etc., that they need.
For this reason, there is a well-earned amount of skepticism when any innovation is
introduced in health care. A Lean product approach can help. Customers and providers are
highly involved in product development and become evangelists when successful MVPs are
eventually released as finished products.
Lean innovation in health care, and the development of digital tools, are essential to
achieving the full benefits of reducing total cost of care, enhancing quality and access, and
improving our ability to effectively deliver on our population health goals. Innovation also
mitigates the potential for disintermediation by new entrants into the health care industry.
Ultimately, digital innovation brings us closer to our patients, helping health systems move
from having transactions with patients and consumers to developing enduring relationships
with them.
Aaron Martin, MBA
Executive Vice President and Chief Digital Officer, Providence St. Joseph Health
Sara Vaezy, MHA, MPH
Chief Strategy Officer for Digital & Innovation, Providence Health & Services
We must be self-critical
and accept that in the
past the industry hasn’t
done the most effective job
at giving our customers
and providers the most
effective tools, information,
resources, convenience,
access, etc., that they need.”

Opening Up to Patient Innovation
Article
Dana M. Lewis
Open Source Artificial Pancreas System Movement (OpenAPS)
If you were on a plane about to make an emergency water landing, and you had the
knowledge to MacGyver life jackets for everyone else with materials readily available, you
might feel obligated to do everything you could to quickly share your insights with everyone
in need. It would feel immoral not to.
That might seem like a straightforward scenario: You have information and the ability to help
teach other people how to help themselves. What could be the problem with that?
In the current health care landscape, there are quite a few
objections to that, particularly when you begin thinking of
specific examples of do-it-yourself health tools and
technologies. I know these objections well, hearing them
weekly when I share publicly how I helped create the world’s
first open source, do-it-yourself automated insulin delivery
system (commonly known as an artificial pancreas). This self-
built system uses off-the-shelf hardware interfaces with an
existing insulin pump and continuous glucose monitor (CGM),
allowing a computer to read and process the data and determine
what changes in insulin delivery are needed. And it’s all made by people with diabetes and
their loved ones.
Regulation: The Means vs. the Ends
To understand all the objections that arise, you also have to be aware of the way medical
devices and systems are typically made available to the public. For new technologies to reach
the market, devices or new tools are developed, put through clinical trials, and then brought
to the regulatory bodies for review and approval.
Patients like me now
have the ability to create
devices, tools, or systems
for ourselves that outpace
the abilities of what is in
the traditional pathway for
development.”

catalyst.nejm.org
Most people assume that patients who are building (or DIYing) their own devices are getting
away with ignoring or flouting the laws and regulations around medical devices. But if you
look carefully at the jurisdiction of the regulatory authorities, you’ll find that many such laws
and regulations don’t apply to individuals, even if they’re also freely sharing knowledge about
what they’re doing so that others can do the same.
The U.S. FDA’s regulatory authority comes from the medical device amendments to the Food
and Drug Act, which authorizes the FDA to regulate the marketing and sale of medical devices.
That authority in turn derives from the Commerce Clause of the Constitution, which gives the
federal government the ability to regulate interstate commerce. Simply building a medical device
for personal use does not fall within the FDA’s jurisdiction, so the structures in place for
regulating commercial activities do not (currently) apply. And, when patients freely share
information with one another on how to use commodity hardware, they are engaging in non-
commercial speech, which is strongly protected under the First Amendment.
Legality aside, many take the default assumption that these
activities should be regulated: If the companies have to be
regulated, so too should everyone else. But this confuses the
means with the ends: The goal of medical device regulation is to
ensure safety and efficacy. The means by which safety and
efficacy have traditionally been demonstrated involve years (if
not decades) for a company to adopt an idea, test it, trial it, and
get it all the way through regulatory approval without it getting
dropped along the way.
Some health care providers (HCPs) object because patients
self-building or self-selecting technology is different than the
way things have been done in the past. Some may have specific
and valid concerns about individual patients and their ability to
manage such technology. But others have treated regulatory
approval as a rubber stamp, and refuse to discuss anything that is not approved (conveniently
ignoring other times when they have discussed things that would be considered off-label
usage). Instead of outright disavowing new technology that they are less or unfamiliar with,
HCPs could instead see the relatively small population of DIYers as an opportunity to get ahead
of the learning curve of what technology will be coming for the broader patient population.
Instead of outright
disavowing new technology
that they are less or
unfamiliar with, HCPs
could instead see the
relatively small population
of DIYers as an opportunity
to get ahead of the
learning curve of what
technology will be coming
for the broader patient
population.”

catalyst.nejm.org
OpenAPS and Other Efforts Have Emerged to Meet Real-World Patient Needs
In some aspects of drug development and device development, the traditional pathway through
traditional regulatory bodies may still make sense. But as incredible technology, computing
power, and knowledge have become more readily available to the average person, there is now a
new frontier. Patients like me now have the ability to create devices, tools, or systems for
ourselves that outpace the abilities of what is in the traditional pathway for development. In
some cases, self-built technology is half a decade ahead of what may be commercially available.
In automated insulin delivery, the technology involved is fairly straightforward, and the next
steps have been obvious for more than a decade to those in commercial, academic, and non-
profit organizations, and to patients as well. More recently, the availability of better consumer-
grade technology, and the ability to collaborate via social media, has allowed patients to switch
from being passive consumers and having to wait, to having the ability to actively engage and
develop solutions that are personalized and work well for each individual. That’s how things
like the open source artificial pancreas came into existence.
Like someone who has the ability to teach someone else to create their own personal floatation
device in a time of need, we all felt obligated to share our work in DIY automated insulin
delivery technology with as many others as possible. In our case, this meant making it open
source. We knew, given the impact this technology had on improving life with diabetes, that
this would help many other people with type 1 diabetes. We also knew, given the world we live
in, that we cannot (nor would we want to) “distribute” medical devices.
And so, OpenAPS is currently a movement around making
artificial pancreas technology more widely available than it is
currently. Despite the fact that in the U.S. there is now (3 years
after OpenAPS was created) one commercial system available,
there remains strong demand for open-source DIY solutions.
First-generation commercial “hybrid closed loop” technology is
limited, by design, testing, and regulatory constraints, to being
a one-size-fits-all device. Because it cannot be sufficiently
customized, some users find themselves frustrated by their
inability to get the commercial system to work the way they
want it to. There are also questions of affordability and access,
particularly given that the commercial version is not yet
approved outside the U.S. and available to people with diabetes
around the world.
Even as multiple
commercial systems
become available, they
are not likely to make
patient-driven innovation
irrelevant; rather, they
continue to make this
technology more widely
available, while DIY
patient innovators
continue to innovate at
the leading edge of what is
possible.”

catalyst.nejm.org
And perhaps most important, open-source solutions are an additional research path
complementing traditional research labs and medical device companies. The open source
solutions have already demonstrated capabilities and real-world outcomes that most people
thought were impossible with insulin-only (single-hormone) systems. So even as multiple
commercial systems become available, they are not likely to make patient-driven innovation
irrelevant; rather, they continue to make this technology more widely available, while DIY
patient innovators continue to innovate at the leading edge of what is possible.
All Roles in the Health Care Ecosystem Must Adapt and Evolve
Patient-driven innovation is here to stay, and it will spread. OpenAPS may be a commonly
cited example, but it’s not the only one out there.
And while some patient-designed innovation will end up inside a company and traditional
regulatory frameworks, it doesn’t necessarily make sense to follow that path in all cases.
Instead, regulatory frameworks need to be updated to deal with new communities of
innovators. Not all innovation happens inside an organization. Not all innovation will be
brought to market and distributed by a company. And, therefore, it is time to determine
whether regulatory bodies should continue to exercise “enforcement discretion” as more
communities share tools and technologies open source via free speech, or how new regulatory
frameworks can promote a symbiotic relationship to better accelerate innovation and support
a new class of innovators who are working to help those around them live better lives.
Clinicians will also need to learn to adapt to and evaluate
technologies that patients bring into their clinics, rather than
relying on regulatory stamps of approvals. We all need to learn
to better evaluate the risk of harm from inaction versus the
potential risks of taking actions to help people. When our
bureaucracies and institutions focus too much on the risk of
action without sufficiently considering the risk of inaction,
such innovative approaches are left to patients who are able to
assess the risk of using the technology, and who can choose to use those solutions if they
assess a net risk reduction overall.
We all need to learn to
better evaluate the risk of
harm from inaction versus
the potential risks of taking
actions to help people.”

catalyst.nejm.org
Patients continue to innovate in this way because their true needs are not being met by the
way things work right now. Part of that is due to slow speeds of manufacturing and
companies’ innovation processes. Part of that is due to the lack of patient-centered design.
But even as companies learn to build faster and better technology, there will always be unmet
needs in the patient community. There will always be people who deeply understand the
problems we face as patients, and who are willing to spend time and energy to build and
share solutions. While not all of us can build these solutions, we can all do our part to further
the efforts of those working to improve the health of themselves, their families, and all of our
communities.
Dana M. Lewis
Creator & Founder, Open Source Artificial Pancreas System Movement (#OpenAPS)
After building her own DIY “artificial pancreas,” Dana Lewis helped found the open source artificial pancreas movement (known
as “OpenAPS”), making safe and effective artificial pancreas technology available (sooner) for people with diabetes around the
world. She is also a Principal Investigator for a Robert Wood Johnson Foundation funded grant project to work to scale pa-
tient-led innovation and scientific discovery in more patient communities.

Heart Safe Motherhood: Applying
Innovation Methodology for Improved
Maternal Outcomes
Case Study
Adi Hirshberg, MD, Katy Mahraj, MSI & Sindhu K. Srinivas, MD, MSCE
Hospital of the University of Pennsylvania
Postpartum preeclampsia contributes to a substantial portion of maternal morbidity and
mortality in the United States. We used innovation methodology to develop a text-based
surveillance system known as Heart Safe Motherhood to target this clinical problem and
define a new approach to care delivery.
KEY TAKEAWAYS
1
Collaborate with patients to identify the problem and develop the most effective strategy.
Partnering with a small number of patients allowed us to receive feedback and make
changes in response. Patient interviews led us to identify the various barriers to in-person
office visits and develop workable solutions.
2
Utilize a cost-effective up-front strategy to get the intervention right prior to scale. An
agile, phased approach to problem-solving enabled us to move quickly and make data-
driven, targeted investments in what worked and what was necessary for the next phase of
growth.
3
Simpler, more accessible technology means a more scalable, more effective solution.
4
Completing a clinical trial can increase the likelihood of adoption for innovative
interventions that dramatically change the standard of care for high-risk populations.
5
Relying on purely human elements can have challenges; this consideration has led us to
think about automation at every step, which also may increase adoption due to ease of use
and implementation.

catalyst.nejm.org
The Challenge
Pregnancy-related hypertension, including preeclampsia, affects up to 10% of pregnancies
and is a major contributor to maternal mortality in the United States.1 Although delivery
starts to reverse the disease process and associated hypertension, blood pressure may take 3
months to normalize and is likely to spike again in the first 10 days after delivery, leaving
mothers at risk of stroke, seizures, and organ failure.2
In 2013, the American College of Obstetricians and
Gynecologists (ACOG) issued guidelines to monitor blood
pressure in postpartum women with pregnancy-related
hypertension at 2 time points — once at 72 hours and again
at 7–10 days after delivery — given the risk of persistent or
progressive hypertension after hospital discharge.3 Many
providers ask patients to return to the office for a blood
pressure check. However, because of the various obstacles
to attending these visits, including delivery recovery, erratic
newborn feeding and sleeping schedules, and transportation
barriers, most women find it hard to come to the office during
this critical time period. Additionally, home nurse visits are also used by some providers;
however, not all insurance policies cover these visits and not all women are willing or able
to accept them.
At the Hospital of the University of Pennsylvania, a tertiary care center in West Philadelphia
where approximately 4,100 deliveries are performed annually, we were unable to meet ACOG
guidelines for blood pressure monitoring in our practices; at best, we obtained outpatient
blood pressure readings for 30% to 40% of patients. We were unable to improve on these
results despite successive trials of conventional approaches, including alternative staffing
models, increased appointment availability, and phone call appointment reminders.
The Goal
Our goal was to develop a remote monitoring program that would capture blood pressure
data in order to facilitate timely clinical intervention and improve our quality of care.
Our initial ‘aha!’ moment
occurred when we
observed pregnant patients
frequently communicating
via text messaging in our
clinics and asked ourselves,
‘Why not see how this
can be used to improve
outcomes?’”

catalyst.nejm.org
Execution and Outcomes
Approach to Innovation
We applied an agile, phased approach to problem-solving, based in design thinking, that
focused on getting an intervention right, even at small scale, and then scaling only what
worked. In our first phase of work, we tested our idea rapidly and iteratively at low cost to
(1) optimize design based on performance and user feedback and (2) determine whether our
solution could work. In the second phase, we expanded our testing to include more patients
and a more scalable infrastructure to determine whether our solution does work. In the third
phase, we established our solution as the standard of care at our hospital and scaled the
solution within our health system.
Learning Through Rapid-Cycle Testing
Our initial “aha!” moment occurred when we observed pregnant patients frequently
communicating via text messaging in our clinics and asked ourselves, “Why not see how this
can be used to improve outcomes?”
We began in 2014 with a series of small, rapid pilots, funded through a Penn Medicine
Innovation Accelerator grant, to test our hypothesis that sending patients home with a digital
blood pressure monitor and asking them to text us their blood pressures would improve
care.4 With 5 to 7 patients per pilot cycle, we enrolled 32 patients in a daily texting program in
which they checked and reported their blood pressure over a span of 4 months.
Early pilots used a “fake back end” in which a physician, acting
as the automated system that would later be developed,
manually sent twice-daily reminders and feedback to the
patient. This approach was non-scalable but allowed us to
pivot quickly. We used a Plan-Do-Study-Act cycle approach
and tried various interventions, including improved nurse-to-
patient communication (e.g., through the distribution of
preeclampsia education pamphlets at discharge), sending
personalized phone messages that mentioned the names of the
patient and baby, sending reminders at different times
throughout the morning and afternoon, “snoozing” the
reminder and sending a repeat alert later in the time period, providing motivational
countdowns of the number of days left in surveillance, and sending a reminder to a support
person (usually a spouse or mother) if the participant did not respond to a morning message.
Following 3 months of
rapid-cycle testing, we
identified a final texting
strategy that worked best
for the majority of patients
and was the most scalable
approach: simple reminders
sent in the morning and
afternoon.”

catalyst.nejm.org
We tested the impact of these iterations on compliance by tracking the frequency and timing
of blood pressure readings texted in by the patients. Patient interviews (conducted with
about 40% of the pilot participants) drove changes at each iteration and provided feedback
regarding the frequency and timing of reminders, the convenience of the program, the
impact of the messages, and recommendations for improvement.
Over 80% of enrolled patients reported at least 1 blood pressure reading during the first week
postpartum, and our ability to meet ACOG guidelines rose to 65%. Following 3 months of
rapid-cycle testing, we identified a final texting strategy that worked best for the majority of
patients and was the most scalable approach: simple reminders sent in the morning and
afternoon. From this small cohort, we determined that this approach could have a positive
impact on patient care. Specifically, we were able to initiate medication remotely for 2
patients using data gathered via text message, likely potentially preventing morbidity, and we
noted that none of the 32 patients were readmitted because of hypertension.
Preparing for Scale
Given the promising pilot results, we transitioned to an automated, more scalable process
with support and funding from two institutional grants, including the Patient-Centered
Outcomes Research Pilot Studies Program and Penn Presbyterian Harrison Fund. We
partnered with Way to Health, a HIPAA-compliant technology platform developed at the
University of Pennsylvania to enable research, development, and deployment of evidence-
based patient engagement strategies. The platform — which facilitates communication
through integrated wireless devices, two-way texting, and interactive online homes — has
been used successfully in projects focused on improving healthy behavior and supporting
new care models in clinical settings.
After discharge, Way to Health starts patients on a provider-
derived monitoring protocol, with twice-daily reminders to send
blood pressure readings over a 10-day surveillance period. Way
to Health automates responses to patients on the basis of our
three-tiered, provider-developed algorithm for normal, slightly
elevated, and severely elevated blood pressures. We use an
exception-handling model for providers, alerting them only when
blood pressures are high enough to necessitate intervention. This
model has allowed us to increase efficiency such that one
provider can manage our hospital’s entire caseload of postpartum
women with hypertensive disorders — representing an average
of 40 patients delivered within a 10-day period — in a few
hours per week.
Way to Health automates
responses to patients on
the basis of our three-
tiered, provider-developed
algorithm for normal,
slightly elevated, and
severely elevated blood
pressures. We use an
exception-handling model
for providers, alerting them
only when blood pressures
are high enough to
necessitate intervention.”

catalyst.nejm.org
Gathering the Evidence
While many innovations can scale without trial-quality evaluation, we decided that an IRB-
approved randomized controlled trial comparing the Heart Safe Motherhood approach
with the standard care protocol (consisting of onetime, office-based blood pressure checks)
would meaningfully increase the likelihood of adoption of our intervention in a high-risk
population. Two hundred and six eligible women with hypertensive disorders of pregnancy
were randomized to either text-based surveillance or standard office care, and the ability
to obtain a single outpatient blood pressure reading in the first 10 days postpartum was
compared between the two arms.5 Baseline characteristics (including age, insurance,
medical comorbidities, gestational age at the time of diagnosis of pregnancy-related
hypertension, and severity of disease) were similar between the two groups, as described
in the randomized trial.

catalyst.nejm.org
From this randomized trial, we again found that text messaging resulted in an increase in our
ability to obtain at least 1 blood pressure reading within 10 days after discharge as well as in
our ability to meet ACOG guidelines, with >90% of patients in the Heart Safe Motherhood
intervention group texting in at least 1 blood pressure reading and 84% of patients in that
group texting in blood pressure readings at the 2 ACOG-recommended time points.
Heart Safe Motherhood not only enables providers to receive real-time notification of
severely elevated blood pressures, but it also allows them to request additional blood pressure
readings following any medication changes. By adapting a similar algorithm already in place
for office visits, which specifies the blood pressure reading at which medication should be
started as well as the specific agent and dosage to be used, we were able to safely start
antihypertensive medications remotely. The readmission rate was 0% among patients who
were enrolled in the texting program. In addition, patients in the texting group were
significantly more likely to attend their postpartum visit, which represents a valuable touch
point for the evaluation of physical and mental well-being after delivery, given the
importance of interpregnancy care and contraception.

catalyst.nejm.org
Changing the Standard of Care
On the basis of the results of our trials, we implemented Heart Safe Motherhood as the
standard of care at two of the five hospitals in the Penn Medicine system that offer obstetrical
services. The Hospital of the University of Pennsylvania and Pennsylvania Hospital together
perform over 9,000 deliveries a year. With funding through our department, we have
monitored the blood pressure for nearly 1,300 women with hypertensive disorders associated
with pregnancy (from a total of about 4,100 deliveries) in the first year at scale at the Hospital
of the University of Pennsylvania and have enrolled over 300 women in 5 months at
Pennsylvania Hospital.
Overall, we have been able to obtain blood pressure readings
remotely at both time points recommended by ACOG for
nearly 85% of our patients. We have been able to intervene and
start medications remotely and have continued to observe
strong patient and provider satisfaction at both sites (as
indicated by patient surveys and provider discussions at
meetings).
Specifically, patients have found the program to be convenient
and effective, and providers have reported increased
reassurance in association with closer, more frequent
hypertension surveillance of their high-risk patients. Although we have not been able to
completely eliminate postpartum hypertension readmissions after the introduction of the
program at scale, the postpartum hypertension readmission rate at the Hospital of the
University of Pennsylvania has dropped from 5% prior to Heart Safe Motherhood to 1% and
hypertension is no longer the leading cause of 7-day obstetrical readmissions. Similar results
are projected for Pennsylvania Hospital.
Success Factors
Technology choices played an important role in our success. We weighed the benefit of
Bluetooth devices against the additional cost, added complexity, and limited patient access
to wireless service in our population.6 Simpler, more accessible technology (in our case, text
messaging sent via cell phones) meant a more scalable, more effective solution.
The readmission rate was
0% among patients who
were enrolled in the texting
program. In addition,
patients in the texting
group were significantly
more likely to attend their
postpartum visit.”

catalyst.nejm.org
Four main factors helped us to achieve buy-in for our solution. First, we had dramatic, real-
world results. We chose a problem with significant room for improvement, and testing our
idea quickly and cheaply gave us a persuasive advantage. Second, we had an evidence-based,
accessible solution to a nationally recognized problem. With poor maternal outcomes on the
rise in Philadelphia and across the country and new ACOG recommendations, there was a
clear call to action and an audience eager for a solution. Third, we had external validation.
We proactively elevated the visibility of our work through publications, conferences, and
innovation competitions, demonstrating to stakeholders that respected organizations were
taking note. Finally, we had a business model that led to improved quality, outcomes, and
patient experience while simultaneously reducing costs.
When taking into account the number of readmissions prior to
the introduction of Heart Safe Motherhood, the average cost of
a readmission, payer reimbursements, and the cost savings
resulting from annual avoided hypertension readmissions
following the introduction of the program, we grossly estimate
the program can save the hospital over $20,000 a year. While we
are in the process of completing a more formal business analysis,
this contribution margin likely underestimates the program
savings given that (1) office blood pressure checks potentially can
be replaced by other revenue-generating visits and (2) inpatient
postpartum lengths of stay potentially can be decreased as
providers become more comfortable with earlier discharge and
the use of remote blood pressure monitoring for their patients.
When we compare the program cost of about $50 per patient (including blood pressure
monitors, educational materials, platform set up and maintenance, and the cost for a
managing provider) with the average direct variable cost of $4,450 for a hypertension
readmission (excluding overhead and other indirect costs), we estimate that the cost of the
program for every 90 patients is similar to the cost of a single readmission. As this estimate is
specific to our patient population and implementation model (in which a single attending-
level physician manages all blood pressures), we hope to obtain more critical cost-
effectiveness data that will be generalizable for different populations, payer practices, and
practice models as we implement the program in different hospital systems and perform
more formal return-on-investment calculations.
Overall, we have been able
to obtain blood pressure
readings remotely at both
time points recommended
by ACOG for nearly 85% of
our patients. We have been
able to intervene and start
medications remotely and
have continued to observe
strong patient and provider
satisfaction at both sites.”

catalyst.nejm.org
Hurdles
Technology was one major hurdle. In our institution, text messages were not considered
sufficiently secure for personal health information. However, approved communication
channels such as the telephone and secure e-mail were ineffective. We were able to persuade
the organization that ineffective use of authorized communication channels resulted in
increased maternal morbidity in this vulnerable population and therefore the privacy policy
should be adjusted. The compromise was that we could use text messaging if patients
consented in advance.
Although patients were eager to receive care in the comfort of
their homes, providers were less comfortable using telehealth
and telemedicine in a population at risk of stroke and seizure.
Education regarding our positive results at every step of
development, including the randomized trial, ultimately led
to widespread acceptance and adoption among >25 attending
physicians and >40 residents involved in obstetrical care.
Improvements to workflow were appreciated by nurses, physicians, and midwives, and
outpatient providers received positive feedback from their patients, further increasing
support.
Finally, even with a highly automated program, there is an element of manual work. The
program relies on patient identification, provider experience, and quick response to patient
health care needs. Simple human errors have led to missed patient enrollments and missed
opportunities for intervention. We hope to develop a data-analytics approach that will enable
us to (1) identify and automatically enroll eligible patients via the electronic health record and
(2) use our blood pressure data to predict those who are at highest risk for intervention in
order to lower intervention costs and increase impact.
Next Steps
We are thinking strategically about growth and are purposefully testing delivery channels
with payers who may reimburse for program costs, providers, home health nurses, research
partners, and technology platforms to make this solution available to as many patients as
possible. Through expansion to hospitals beyond our own, we are learning about replicability
and about customizing the program to adapt to different contexts and populations, including
different non-centralized models of obstetrical care at non-academic centers. Payment model
innovations such as readmission penalties and quality incentives, as well as new CPT codes
that allow for reimbursement for remote data review and management, will accelerate
widespread adoption of this type of valuable patient-centered program.
We grossly estimate the
program can save the
hospital over $20,000
a year.”

catalyst.nejm.org
Within our own population, we are refining our approach to
increase patient engagement and to assess broader
applicability. We are studying differences between patients
who text their blood pressure readings and those who do not
in order to provide better care to more patients, and we are
working to simplify the inpatient enrollment process through
connections to the electronic medical record and program-
specific patient education videos. The program not only opens
doors to new research opportunities, but it also provides a
model for text-based communication that can be used to target other critical elements of
postpartum care, such as depression and breastfeeding support.
Disclosures: The authors have no conflict of interest to disclose, including no financial conflict of
interest with the Way to Health.
References
1. Creanga AA, Syverson C, Seed K, Callaghan WM. Pregnancy-related mortality in the United States, 2011-2013. Obstet
Gynecol. 2017;130(2):366–73.
2. Too G, Wen T, Boehme AK, Miller EC, Leffert LR, Attenello FJ, Mack WJ, D’Alton ME, Friedman AM. Timing and risk
factors of postpartum stroke. Obstet Gynecol. 2018;131(1):70–8.
3. American College of Obstetricians and Gynecologists, Task Force on Hypertension in Pregnancy. Hypertension
in pregnancy. Report of the American College of Obstetricians and Gynecologists’ Task Force on Hypertension in
Pregnancy. Obstet Gynecol. 2013;122(5):1122–31.
4. Hirshberg A, Bittle MD, Vandertuyn M, Mahraj K, Asch DA, Rosin R, Bennett I, Srinivas SK. Rapid-cycle innovation
testing of text-based monitoring for management of postpartum hypertension. J of Clin Outcomes Manage.
2017;24(2):77–85.
5. Hirshberg A, Downes K, Srinivas S. Comparing standard office-based follow-up with text-based remote monitoring in
the management of postpartum hypertension: a randomised clinical trial. BMJ Qual Saf. 2018;27:871–7.
6. Gilbert J, Schnoll R, Morrison MF, Srinivas SK, Pond T, Curtis B, Henry J, Kranzler HR. Smoking and cellular telephone
use among pregnant women seeking prenatal care: opportunities for intervention. Addict Disord Their Treat.
2015;14(4):203–10.
Adi Hirshberg, MD
Assistant Professor of Obstetrics and Gynecology and Obstetrician-Gynecologist, Maternal Child and Health
Research Program, Perelman School of Medicine and Hospital of the University of Pennsylvania
Katy Mahraj, MSI
Director of Operations, Acceleration Lab, Penn Medicine Center for Healthcare Innovation, Perelman Center for
Advanced Medicine, University of Pennsylvania
Sindhu K. Srinivas, MD, MSCE
Associate Professor of Obstetrics and Gynecology, Obstetrician-Gynecologist, Maternal Child and Health Research
Program, and Director, Obstetrical Services, Perelman School of Medicine and Hospital of the University of
Pennsylvania
Simpler, more accessible
technology (in our case,
text messaging sent via
cell phones) meant a more
scalable, more effective
solution.”

Care Redesign Survey: How Design Thinking
Can Transform Health Care
Insights Report
Amy Compton-Phillips, MD & Namita Seth Mohta, MD
NEJM Catalyst
Analysis of the NEJM Catalyst Insights Council Survey on Care Redesign: How Design Thinking
Can Transform Health Care. Qualified executives, clinical leaders, and clinicians may join the
Insights Council and share their perspectives on health care delivery transformation.
Advisor Analysis
BY AMY COMPTON-PHILLIPS AND NAMITA SETH MOHTA
Health care in the United States has evolved to be a complex and disconnected system. We
are unable to deliver the excellent care that our patients deserve, that is fiscally responsible
to citizens, and that clinicians sought to provide when they took on this work.
How oen does your organization actively employ principles/techniques of design thinking?
Base = 625
Always Mostly Occasionally Seldom Never Not sure
4%
20%
39%
21%
7% 9%

catalyst.nejm.org
If we could start over, how would we redesign health care delivery?
A recent survey of NEJM Catalyst Insights Council members shows them ready to embrace
design thinking to transform care. They view design thinking as valuable for a range of health
care issues — yet its application has been impeded by limited organizational buy-in and
limited understanding of the concept.
Design thinking is defined as a discipline that uses designers’ sensibility and methods, such as
collective idea generation, rapid prototyping, and continuous testing, to match people’s needs
with what is technologically feasible, provides customer value, and is a viable organizational
strategy.
The great majority of Insights Council members — who are clinical leaders, clinicians, and
executives directly involved in health care delivery — believe design thinking can be
extremely useful, very useful, or useful in the health care industry (say 95% of respondents)
and their own organizations (91%).
The survey identifies a wide range of issues that would benefit
from design thinking: staff and provider flow collaboration,
patient scheduling, care coordination, and social determinants
of health, to name a few. In a written comment, an executive
respondent highlights the need to overhaul office and clinic
checkout processes: “A patient has to go to a checkout desk,
wait for assistance, wait for scheduling, have their information
overheard, be uncomfortable, etc.” Stacey Chang, MS,
Executive Director at the Design Institute for Health at the
University of Texas at Austin, puts it well: “The dysfunction
of our modern health care system isn’t about failure of intention, but rather pursuit of siloed
and sometimes conflicting priorities.”
Design for Patients
Patient input is often the best ingredient for true transformation. Take, for instance, the
dialysis clinic in Jönköping, Sweden, which redesigned its facility to enable self-dialysis.
While health care leaders on the project thought state-of-the-art beds and beautiful artwork
would promote healing, patients encouraged them to forgo all that, buy cheap dialysis chairs,
and allocate resources to exercise equipment so they could work out while they underwent
their treatment. No health care professional thought of that — the patients did.
The dysfunction of our
modern health care
system isn’t about failure
of intention, but rather
pursuit of siloed and
sometimes conflicting
priorities.”

catalyst.nejm.org
The family of one of the authors (ACP) recently experienced the need for patient-centric design
firsthand. My brother, who is deaf, was diagnosed with tongue cancer by his dentist. He spent
six months suffering from not only his condition, but also a lack of coordinated care. Nothing
seamlessly connected his extended care team, which comprised the dentist; an ear, nose, and
throat specialist; a medical oncologist; a radiation oncologist; and others. The knowledge
transfer between these providers was abhorrent, and it left him with intense anxiety.
Design thinking would enable that health system to reimagine these handoffs in a more
elegant and productive manner. The care team members would be encouraged to walk in the
patient’s shoes, really observe and experience the system from his perspective, and develop
practical solutions.
With so many health care professionals considering design
thinking a useful endeavor, one would expect it to be widely
utilized. Yet two-thirds of survey respondents say their
organization employs design thinking occasionally, seldom,
or never.
So how can health care professionals progress from merely
understanding the value of design thinking to putting it into
everyday practice? The burden, and opportunity, lies at the feet
of the survey respondents, the NEJM Insights Council. Clinical leaders (45%), executives (37%),
and clinicians (33%) are ranked as the three most appropriate stakeholders to champion the
application of design thinking. What is a bit surprising in this result is that patients fall closer to
the bottom of the list, at 14%.
As health care leaders and frontline clinicians begin to implement design thinking to
accelerate needed changes, they should expect to encounter significant barriers, including
these top three from the survey: limited buy-in from decision makers (chosen by 52% of
respondents), limited understanding of design (47%), and insufficient training in design (32%).
To us, the fourth barrier, return on investment (28%), is somewhat more interesting. Good
design tends to be efficient and therefore cost-effective.
In this new era of patient-driven care delivery, it’s not enough to adapt existing systems; we
need to create something better. Design thinking can get us there.
In this new era of patient-
driven care delivery, it’s not
enough to adapt existing
systems; we need to create
something better. Design
thinking can get us there.”

10 Innovation Pathways: Care Delivery

10 Innovation Pathways: Care Delivery

Recommended

Recommended

More Related Content

What's hot

What's hot (16)

Similar to 10 Innovation Pathways: Care Delivery

Similar to 10 Innovation Pathways: Care Delivery (20)

More from TrustRobin

More from TrustRobin (20)

Recently uploaded

Recently uploaded (20)

10 Innovation Pathways: Care Delivery