5 Lessons from Penn Medicine’s Crisis Response

by Katherine Choi , Srinath Adusumalli , Kathleen Lee , Roy Rosin and David A. Asch - June 22, 2020

Almost overnight, traditional hospital or office-based care became inadequate for serving the community in the midst of the Covid-19 pandemic. By using an iterative sprint process, we were able to quickly design, validate, and scale much-needed services during our health system’s early response to the crisis. We illustrate what we learned using two of those new care delivery innovations: satellite drive-thru care sites and a platform for virtual health care visits.

Use Makeshift Approaches Designed to Flex

In a Friday in early March, as the dangers of a novel infection were dawning on the nation, the urgent request came in: establish a site to test the public by Monday. Our health system’s pre-Covid testing locations were not built to minimize the exposure risk or accommodate the high patient volumes the outbreak brought. We drew on health fair and drive-thru concepts as inspiration, and in under 24 hours converted a vacant parking lot into a testing site using simple rope, tape, traffic cones, and large A-frame whiteboard signs as directional guideposts. To minimize physical contact, cars became makeshift exam rooms, and duct-tape sidewalk indicators kept our walk-ins socially distanced. We registered patients using photographs of their IDs, and developed a text-based contactless discharge and test result follow-up process.

We were scrappy. The public lot’s WiFi hotspots, an unexpected blessing, became essential for placing on-site orders for walk-ins. Redeployed healthcare staff and campus security workers helped us direct flow. Self-proclaimed “delight rabbits,” colleagues who volunteered to run errands, delivered just-in-time coffee, donuts, hand warmers, ponchos, and paper towels that boosted morale and made support visible to the front-lines. The bare-bones site brought out the creativity and dedication of the team behind it.

Make Teams into Squads

Our small squad - three of us leading just under 10 clinical staff - was empowered to make quick changes and make things work. We added more patient privacy for our walk-in assessments by simply moving five cones to a more secluded spot. When walk-ins declined in the afternoons, medical assistants shifted roles to manage the drive-in queue. We used erasable whiteboard drawings instead of printouts of protocols because our script and screening questions had to evolve alongside the changing testing criteria.

We also found ways to stay in sync and adapt as quickly as the plans changed. We kept a rolling count and burn rate of test inventory as useful indicators for planning and setting expectations for the day. Ad-hoc check-ins helped the team resolve safety concerns or inefficiencies like redundant symptom screens that led to avoidable delays and frustrated patients. Great ideas were shared and praised at “kudos” rounds, and these changes - from creating float staff roles to writing on windshields with dry erase markers to communicate down the line - became the way we ran the next day.

This drive-walk-thru model was low-cost and reproducible. It was adapted a week later to organize care tents to manage a projected surge of Covid-19 infected patients. It was repurposed again two months later as physical clinics began to reopen but needed a text-based check-in process for patients as they waited in parking garages. Later iterations eventually moved into covered brick-and-mortar locations to withstand variable weather, but the low-contact workflows and ability to triage patients quickly “at-or-before-the-door” will be as relevant and valuable for the post-Covid era as it is useful today.

Go the Extra Mile for Your Early Adopters

As our clinicians converted in-person patient visits to telemedicine, the bandwidth of our legacy video platform could no longer meet the demand. The decision to switch enterprise vendors was made within two days.

Acknowledging limitations of the new software during town halls allowed concerned clinicians to define the needs to be met, particularly the worry that multiple patients might join the same virtual “exam room,” and that errors could be introduced by manually sending patients links. What we needed was an automated “switchboard” that generated and sent unique links to patients, and created a meeting schedule for providers to conduct their clinic days. Leveraging existing custom data feeds from our electronic medical record and the vendor’s meeting software expedited our first working prototype, a solution developed over the weekend that embedded the video service within the context of key patient data from the chart, and soon after, automatically sent patients appointment reminders, links, and setup instructions by text and email.

Addressing the most pressing need - ensuring patient privacy - attracted our earliest clinician adopters. But it was the features that delighted them and kept them engaged: Live status indicators to show when a patient had joined their virtual exam room, auto-calculated meeting lengths for easier documentation, daily schedule download into email calendars, and rotating public messages of gratitude sent in from the community. Our immediate and personal response to feedback took the form of visible, daily improvements to the platform (or thoughtful explanations of what couldn’t be done) which invited further engagement. These early physicians’ praise won over their more skeptical colleagues, and ultimately drove adoption across the health system.

Embed into the Front-line Experience

Our app development squad of four gathered insights by immersing itself into the patient and clinician experience. Not only did we conduct clinic visits through our platform, we attended our own virtual visits as patients and helped family members attend theirs. We personally helped call over 300 patients to set them up for their virtual visits. Downloading the vendor’s app to join the meeting was the single biggest point of failure. Keeping track of the appointment meeting link was close behind. Knowing where patients got stuck helped us build an effective text message-based bot that, if started, got over 95% of patients to a successful setup on their own.

Our texting system auto-replied to patient questions but also forwarded inbound texts to us. We saw caregivers wanting to share instructions on how best to reach their family members, and patients simply venting: “App is not working. Totally frustrated. Please call my cell.” In response, we built two-way texting so physicians and patients could pivot how they contacted each other in real-time. Because we faced so many language barriers, we enabled a live text translate feature across more than 65 languages.

Solving these technical hurdles returned everyone’s focus back to health care. One expression of success came when one of our physicians noted to her patient “we are going to get through this together” - and she was referring to the patient’s health concern, and not the frustrations of a virtual visit.

Change Favors the Prepared

Thousands of patients were swabbed the first week in a single parking lot, and soon more sites opened. In less than a month, zero virtual visits grew to over 24,000 visits a week being conducted through our new platform.

The crisis added unprecedented urgency and alignment of priorities, but we also owed our speed to our preparedness. In the years before, we had been investing in innovation infrastructure, including both technical and design capacity. In-house Application Programming Interfaces (APIs) and human-centered designers experienced with the clinical enterprise enabled rapid implementation. Text-based platforms were already in place that automated and protocolized communication with patients and escalated to human backup when needed. All this decreased our time to deployment from months to hours or days.

It’s impossible to outsource change and innovation. We could not have met our needs in time looking externally for a vendor or prebuilt solution. Change leaders are found within an organization and defined by the mindset that standing up a new service overnight is often the best way to test it. That kind of bias towards action, paired with an enabling infrastructure, is essential for responding to a crisis, but also in leading organizations through change in any era.

The authors would like to thank Lauren Hahn, Christina O’Malley, Nida Al-Ramahi, Damien Leri, Yevgeniy Gitelman, Erik Lang, Aaron Leitner, Liz Deleener, Catherine Shi, Neha Patel, Jake Moore, Timothy Jones, Mohan Balachandran, Christianne Sevinc, and Stephanie Brown for their contributions to these efforts.

Katherine Choi, MD, is the Director of Practice Transformation at the Department of Medicine and the Center for Health Care Innovation at the University of Pennsylvania.

Srinath Adusumalli, MD, MSHP, is Assistant Professor of Clinical Medicine in the Division of Cardiovascular Medicine and a Clinical Innovation Manager at the Center for Health Care Innovation at the University of Pennsylvania

Kathleen Lee, MD, is an Assistant Professor of Clinical Emergency Medicine and the Director of Clinical Implementation at the Center for Health Care Innovation at the University of Pennsylvania.

Roy Rosin, MBA, is the Chief Innovation Officer of Penn Medicine at the University of Pennsylvania.

David A. Asch, MD, MBA is the John Morgan Professor at the Perelman School of Medicine and the Wharton School and Director of the Center for Health Care Innovation at the University of Pennsylvania.