As important as it is to sustain and continue building engagement with your patients, the most effective method — in terms of the relationship and the cost to your organization — may not involve a living person on your side of the conversation.
The role of conversational artificial intelligence (AI) in healthcare is a frequent topic as the industry continues to find ways to meet growing consumer-driven demands in a cost-conscious fashion.
During the 2021 Medical Practice Excellence: Pathways Conference, Jeff Haggard, director of sales, healthcare and strategic accounts, Orbita, noted in his session that chatbots, virtual assistants and other tools powered by conversational AI are doing much more than passive response to simple inquiries.
Newer uses are calling patients to action and “removing friction” for patients who may only be returning to care for the first time after deferring it during the COVID-19 pandemic, Haggard said.
Haggard’s co-presenter, Chelsea Biel, BSN, RN, senior director, customer success and clinical informatics, Orbita, noted that even as overall vaccination efforts across the United States gained steam throughout the spring months, there are many medical groups and health systems still working to reach “populations of patients that may have barriers to care or associated risk factors that make it challenging to connect” with them.
Especially as the country reaches the tipping point for vaccine availability and supply outstrips demand, there remains a broad need from the public health perspective to build “operational processes to support patient communication, patient scheduling and patient inquiries,” Biel said.
Reaching patient populations changed by the pandemic
The expansion of telehealth during the pandemic is just one small piece of how patients have adapted their expectations about care delivery in the past year. A Feb. 16 MGMA Stat poll found that 61% of medical group leaders whose organizations measure patient satisfaction with telehealth say that those patient satisfaction scores improved in the past year.1
As the telehealth boom continues — about 65% of healthcare leaders in a Feb. 9 MGMA Stat poll said they expect utilization to increase or stay the same throughout 20212 — the growing comfort level with virtual care delivery is also prompting those same patients to adjust their comfort level with making choices in searching for healthcare and choosing providers online via new technologies, Haggard said.
Even nearly a year into the pandemic, most medical practices reported their primary communication channel for patient appointments was phone calls.3 With more patients returning for in-person care and provider schedules filling up,4 there’s opportunity for growing phone call volumes to be reduced by automation in the form of conversational AI.
Optimizing patient access also is vital to address the potential for total costs to rise in the coming years because of delayed care during the pandemic.
“Once that immediate and direct impact of COVID passes, we will have those lingering effects of the deferred care that has taken place over the last year or longer,” Biel noted, pointing to studies that estimate between $30 billion and $65 billion in increased costs to treat conditions that worsened due to care deferrals, including cancer, COPD, diabetes, hypertension and more.5
Recognizing shifts in patient behavior in setting up appointments also is crucial in understanding where conversational AI might have a role in transitioning to new patient access and engagement strategies. Haggard notes that it’s not surprising that millennials and Gen. X patients are comfortable using digital sources to book appointments and interact with providers, but recent studies show those trends increasing and extending into the cohort of older patients in the Baby Boomer generation (see Figure 1).
“We’re seeing a bit of a decrease in reliance on traditional phone to do those things … we’re seeing increased smart speaker usage, we’re seeing increased use of apps and mobile-enabled services,” Haggard said. “If patients can’t avail themselves of digital modalities to do that, they’re making choices to move to different providers with whom they can.”
What conversational AI looks like going forward
As healthcare provider organizations look beyond the pandemic, the traditional web chatbot has been a likely first step to build new ways for patients to interact with something that can authenticate the patient via a connection to a CRM, EHR or other system, which opens the door to learning about online bill-pay options.
But Haggard, noting the deferred care phenomenon, said that it’s more important for providers to use these systems to create a call to action across multiple channels and modalities of interaction. “Patients that engage well with a web chatbot may not engage well with SMS and vice versa, so we can give them a similar experience via text,” or embedded in an app, via a voice interaction on the web or a smart speaker via natural language processing (NLP) technology, Haggard said.
Deciding on which channels to use and how begins with a shift in mindset away from the understanding of a traditional virtual assistant or chatbot, located on the bottom-right corner of a website, “being something that is passively waiting for a patient to come engage with it,” Haggard said.
The next generation of virtual assistant strategy will focus more on the system’s “ability to reach out to populations of patients either based on a particular trigger or based on a planned campaign based on an order set,” Haggard said.
One example is a communication cascade, in which a patient who doesn’t respond to an SMS message would then receive an email at a later time or date. Further lack of response could then prompt an automated phone call. Failing that, it might be time to use a human to reach out.
The cascade approach is especially helpful in that it helps reveal patient preferences for communication on an individual level as well as demographic-level preferences based on age, what type of care the patient receives and more.
That campaign-style approach might lend itself to a patient discharged from an acute stay or emergency department visit with medication to pick up, Haggard noted. Use of virtual assistants to check on the patient’s ability to fill the prescription and be adherent to the medication plan gives a care team some insight into what to expect from that patient — and potentially take action if lack of adherence might lead to a readmission or return to the ED.
“It gives us the opportunity to reach out to that patient to proactively gain information from them that may lead to better treatment [and] better outcomes from that patient,” Haggard said. For patients who are staying on track with their care plan, the increased engagement helps “bring that patient into the experience,” he added.
Biel said this is especially important as remote patient monitoring (RPM) grows as an area of virtual care delivery. “One of the challenges with remote monitoring and some of the disease management programs to date is enrolling those patients in those programs … and that enrollment process is oftentimes very manual within those population health or care management teams,” Biel said, adding that it frequently requires clinical team members to reach out to patients during typical business hours, which can be unsuccessful if the patient is working or cannot respond.
“Being able to automate and scale that enrollment process to those remote monitoring programs and other disease management programs,” Biel said, allows providers such as nurses and other staff to shift away from work that could otherwise be centered on conversational AI programs.
Haggard suggested automating the front end of a program to address patients who might stop sending back vitals or other information in an RPM program. A human-like virtual assistant can be a proactive way to help the patient reconnect with the care team to ensure they are OK. “Reinforcing those messages [and] doing that in an automated way … allows that care team to continue to practice at top of license” with care delivery, Haggard said, rather than spending significant amounts of time on the phone.
Another benefit of this multichannel approach to patient outreach is the ability to designate a caregiver or trusted family member who is assisting a patient to receive the same communications about any call to action the patient needs given his or her care situation — which would be difficult to do if the patient and caregiver were not simultaneously available for a live phone call with the provider organization, Haggard said.
Biel shared one real-world example of an automated patient outreach system that focused on patients with negative COVID-19 rest results. During the pandemic, about 17 dedicated full-time-equivalent (FTE) staff were handling patient outreach by phone.
The organization switched to a bot-powered outreach system that automatically cascaded test results to more than 100,000 patients across SMS, email and automated phone calls. This automation increased response rates from patients from less than 30% via manual phone calls to more than 70% aggregated across the cascaded communication channels.
The other benefit is that results could be sent in real time once a negative COVID-19 test result was in the organization’s system, which increased patient satisfaction. “When you’re waiting for a COVID test result, life is kind of on hold until you hear back on that result,” Biel said.
Overall, the number of FTEs needed to manage these notifications after the automated campaign dropped to four, resulting in a savings of 520 work hours per week.
Learning more to communicate more effectively and build value
Using virtual assistants, chatbots and other conversational AI tools holds significant potential for workflow efficiencies and boosting patient engagement and satisfaction, but Biel noted that healthcare providers need to ensure that these systems are capturing the data that patients provide and then analyze that data to help guide future initiatives.
“We’re able to take a look at how patients have actually perceived care” and how they’ve shifted behaviors during the pandemic, Biel added. “It’s really important to be able to have that data from an organizational perspective to actually implement some of those changes.”
Overall, patient engagement strategies such as conversational AI need to reflect what an organization knows about the value of patient retention and the imperative to build long-lasting relationships and restore some of the connections that were interrupted in 2020.
“We think about what happens when a patient misses an appointment and then doesn’t reschedule; we think about where that patient goes,” Haggard said, noting that providers should be able to quantify the “long-term value of that patient.”
To build on that value, reducing operational costs to ensure that long-term patient relationship is a successful one requires the types of key performance indicators (KPIs) that can be easily influenced by a robust conversational AI program.
- “Almost one year into COVID-19, patient satisfaction with telehealth growing.” MGMA. Feb. 18, 2021. Available from: mgma.com/stat-telesatisfaction.
- “Is the telehealth boom set to continue in 2021? Congress and consumers will help decide.” MGMA. Feb. 11, 2021. Available from: mgma.com/stat-021121.
- “Patients and medical practices staying in touch remains vital as COVID-19 pandemic continues.” Feb. 25, 2021. Available from: mgma.com/stat-022321.
- “Nearly half of America’s doctors are busier than ever as COVID-19 recovery continues.” MGMA. June 2, 2021. Available from: mgma.com/stat-060221.
- McKinsey & Company. “Understanding the hidden costs of COVID-19’s potential impact on US healthcare.” Sept. 4, 2020. Available from: mck.co/3pwqafb.
- Kyruus. 2020 Patient Access Journey Report. October 2020. Available from: https://bit.ly/3jJwvkW.