It is morning at the Hospital of the University of Pennsylvania (HUP). A palliative care nurse arrives to start her work day. Her team plays a vital role; she and the doctors and nurse practitioners she works with are called in to have some of the most important and powerful conversations in the lives of some patients and their families.
“It involves having a discussion about things that are important to the patient, what they are worried about, what gives them strength, and how much to go through if they get sicker,” said Nina O’Connor, MD, chief of Palliative Care for Penn Medicine. “We document this information in the electronic health record where everyone in care team can see, and that information is shared with the patient’s outpatient provider.”
Palliative care is helpful at the end of life, but these programs are helpful at any stage of disease to provide assistance in treating complex symptoms, extra emotional and psychosocial support, and the opportunity to discuss and establish the goals of care. These conversations are crucial—but too rare. A number of efforts underway at Penn Medicine are working to get more patients to have these conversations before they are too sick to express their wishes. One of these is an educational program to train outpatient physicians in how and when to have such conversations with cancer patients—and is part of the focus of a feature story in the forthcoming July issue of System News. Another, a new program called Palliative Connect, uses cutting-edge predictive data technology to facilitate coordination between palliative care specialists and a patient’s primary clinical care team, and ultimately ensure that patients with the highest need are offered a timely opportunity to have these conversations soon after their arrival at the hospital.
“It’s cutting-edge in that we’re thinking about how to best identify patients for this kind of service,” said O’Connor, who teamed up with Penn Medicine specialists in data science and human factors to develop this new tool. It is the latest in a series of projects Penn Medicine clinicians have developed over the last few years in partnership with the team led by Chief Data Scientist Michael Draugelis that uses predictive analytics to improve health care across a variety of specialties, such as identifying heart failure. Data scientist Corey Chivers, PhD, and human factors scientist Susan Harkness Regli, PhD, were among the key team members on the project.
The Palliative Connect tool works invisibly behind the scenes—but its impact is evident in the daily tasks of the palliative care nurse when she arrives at work. On a day like this a few weeks ago, her tasks would have centered on responding to requests. Clinicians from around the hospital would reach out to her department to ask for a palliative consultation for patients who they felt could benefit from one. A general rule of thumb physicians are encouraged to follow is to ask themselves, based on their finely honed clinical intuition, “Would I be surprised if this patient passed away in the next year?” If the answer is no, they should ensure the patient and family are having appropriate conversations with a qualified expert, and that the patient’s wishes are documented and shared with the care team. But, too often, front-line clinicians who work with high-risk patients don’t initiate these requests soon enough to affect the type of care patients receive in the late stages of their disease.
“I think the challenge is, front-line clinicians are so close to the situation, thinking about the patient’s lab values, medications, testing, and other specific issues during that hospital stay, that they are not always able to step back and picture that this patient has a reasonably high chance of dying in the next six months or year,” O’Connor said.
Now, with the Palliative Connect tool, in addition to receiving consultation requests that come in from physicians, a palliative care nurse is able to log in and see a list of recently admitted patients whose clinical data in their electronic health record indicate they would benefit from a palliative care consultation; they’ve exceeded a designated threshold of risk that was built into the system. It’s a kind of data-driven, digitized version of the physician’s intuition. Armed with the list, the nurse can call the front-line clinicians caring for these patients and suggest a palliative care consultation. If the front-line clinician agrees, a palliative care physician or nurse practitioner meets with the patient that same day. If they decline, the nurse documents the reason why so that the system can continue to improve. And she calls up another front-line clinician to ask about the next patient on the list who passed the risk threshold.
Palliative Connect was initially run as a pilot project at HUP over a two-month period starting in December 2017. During the pilot phase, the multidisciplinary team wanted to assess whether the program was acceptable to patients and clinicians, as well as whether it aided in getting more high-risk patients to receive palliative care consultations, to document their wishes in their electronic health record, and to move on to hospice care if that was the right choice based on the patient’s goals. To find out if they achieved these goals, the team called patients and families, surveyed providers, and reviewed and analyzed data from the electronic health record to quantify the type of care the patients received. By every measure, the pilot was a success.
Patients and families rated the program favorably as a satisfying experience, and helpful for understanding their illness. Front-line clinicians said they felt it was an acceptable intervention and it improved the care of their patients. “They also said it helps with their feeling of moral distress taking care of these patients,” O’Connor said. “Some of these issues weren’t being expressed before.”
Though the pilot sample size was not large enough to detect statistical significance on these measures, there were early indicators that the program may have also had an impact on clinical outcomes such as inpatient mortality and readmissions.
After the success of the pilot program, on June 4, Palliative Connect expanded across HUP and Pennsylvania Hospital.
“I’m excited we’re doing it at two different hospitals with different consultation cultures and different patient populations,” O’Connor said. “We will run it long enough to really measure outcomes, which is important to make decisions about the program and future steps.”
Coming up in in a future Penn Medicine News Blog post: Take a peek behind the curtain of how the Palliative Connect tool was developed. Penn Medicine’s innovative multidisciplinary predictive analytics teams, in which data scientists, human factors scientists, and clinical experts have partnered together, have been learning not just how to improve health care through the use of data—but also how to run such data-driven improvement projects better.